Model for tooth development in vitro
vorgelegt von
Diplom-Ingenieurin
Jennifer Rosowski
geboren in Berlin
von der Fakultät III Prozesswissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften
(Dr.-Ing.)
genemigte Dissertation
Promotionsausschuss
Vositzende: Prof. Dr. Vera Meyer
Gutachter: Prof. Dr. Roland Lauster
Gutachterin: Prof. Dr. Claudia Fleck
Gutachter: Prof. Dr. Juri Rappsilber
Gutachter: PD Dr. Frank P. Strietzel
Tag der wissenschaftlichen Aussprache: 25. Januar 2019
Berlin 2019

„Wer am Ende ist, kann von vorn anfangen,
denn das Ende ist der Anfang von der anderen Seite.“
- Karl Valentin -

Abstract
The ultimate goal of Tissue Engineering is the regeneration of whole organs or func-
tional organ units. Recent approaches rely on the usage of adult tissue-specific cells in
three-dimensional culture. It is now accepted that "Developmental Engineering", the
recapitulation of critical initial steps of organogenesis in vitro, is more expedient. Identi-
fication of key players and basic mechanisms behind organ formation is one of the major
challenges for tissue engineering. Developmental biology addresses these questions and
much knowledge has been accumulated on the mechanisms behind organogenesis by study-
ing animal models. However, animal models evince species-related limitations and arise
ethical issues. Functional in vitro models emulating the physiological processes during
human organ formation and homeostasis are requested and invaluable for future research.
Here, a developmentally inspired approach is pursued to reproduce fundamental steps of
human tooth organogenesis in vitro. The presented model comprises the usage of human
dental pulp cells from adult donors that were expanded in monolayer before to culture under
non-adherent conditions to accomplish a 3D self-organized mesenchymal condensation. The
cells reproducibly formed uniform organoids of 500 µm in diameter, resembling the size of
a human mesenchymal tooth germ. Comprehensive gene expression analysis by qPCR and
Next Generation Sequencing revealed a dramatic remodelling of the signalosome in the
DPCs. The cells instantly responded to the changed culture conditions and subsequent
cell-cell interaction with significantly modulated gene expression regarding cytoskeletal
rearrangement. The suppression of RhoA activity was exhibited, a central molecule in
actin-mediated mechanotransduction leading to cell shape change and subsequent induction
of tissue-specific cell fate. The central role of actin-mediated signalling was evidenced by
a functional assay, in which the actin polymerization and thereby cytoskeletal signalling
was inhibited. No cell-cell contacts or condensation was observed in the DPCs upon
administration of the inhibitor. Furthermore, central signalling pathways of mesenchymal
condensation have been shown to be upregulated within the first six hours of the culture
model. Among them are the TGFß- and Activin, the FGF, the Wnt, and the Notch
pathway. The bivalent results from transcriptional analysis, regarding the activation status
of the MAPK pathway, were unraveled by a MAPK reporter assay, which evidenced an
immediate-early MAPK activity within minutes of condensation. A Collagen-binding
eGFP-fused protein (CNA-eGFP) was applied to the DPC condensations to track Collagen
protein secretion, which was visible after 4 hours. From protein analyses by immunohisto-
logical staining, after 4 days of condensation, the ECM components Tenascin, Fibronectin,
Collagen type I and type IV were present in the DPC condensates. Long-term cultures of
DPC condensates (4 weeks) showed odontogenic differentiation by upregulated expression
of Dentin sialophosphoprotein, TGFß1, Activin, and BMP7. Inductive abilities of the
self-organized dental mesenchyme were evidenced in co-culture with epithelial cells, that
exhibited invagination and cytodifferentiation.
The presented model of tooth development fulfills all demands on a functional human
organogenesis model. It recapitulates the initial step of mesenchymal condensation and
subsequent odontogenic differentiation and is capable of directing reciprocal differentiation
induction of combined epithelium. Therefore, it represents a valuable model to study
basic developmental mechanisms and is in the pipeline to find application in regenerative
therapies.

Zusammenfassung
Das Ziel des Tissue Engineering ist es, vollständige Organe oder funktionelle Organeinheiten
in vitro herzustellen. Aktuelle Forschungsansätze beruhen auf der Verwendung adulter,
gewebsspezifischer Zellen in drei-dimensionaler Zellkultur. Dabei wird die Rekapitulation
initialer Schritte der Organogenese als vielversprechendste Herangehensweise betrachtet.
Ein wesentlicher Aspekt des Tissue Engineering befasst sich daher mit der Identifizierung
von zentralen Molekülen und den grundlegenden Mechanismen der Organentwicklung. Das
Forschungsfeld der Entwicklungsbiologie hat unter Einsatz von Tiermodellen dahingehend
viel zum aktuellen Wissensstand beigetragen. Diese bergen allerdings spezies-geschuldete
Limitierungen und ethische Probleme. Funktionelle in vitro-Modelle, die die physiologischen
Prozesse der humanen Organentwicklung und -aufrechterhaltung nachbilden, sind daher
von unschätzbarem Wert für die zukünftige Wissenschaft.
In der vorliegenden Arbeit wurde die Nachahmung der natürlichen Prozesse bei der ini-
tialen Zahnentwicklung im Menschen unter Kulturbedingungen erreicht. Das Kulturmodell
basiert auf der Verwendung adulter in vitro-expandierter Zellen aus der Zahnpulpa (DPCs),
die unter nicht-adhärenten Bedingungen mesenchymale Kondensation autonom vollziehen.
Dabei formierten sich einheitliche Organoide von 500 µm Durchmesser, was der Größe
eines humanen fetalen Zahnkeims entspricht. Umfassende Genexpressionsanalysen mit-
tels qPCR und Next Generation Sequencing, offenbarten drastische Veränderungen im
Signalosom der DPCs. Die Zellen antworteten auf die veränderten Kulturbedingungen
und zahlreichen Zell-Zell-Kontakte mit einer signifikanten Regulation der Gene, die eine
Rolle bei Remodellierung des Zytoskeletts spielen. Die RhoA-Aktivität wurde supprimiert,
einem zentralen Molekül der Aktin-vermittelten Mechanotransduktion, das Veränderung
der Zellform und die Induktion der gewebsspezifischen Zellidentität ermöglicht. Durch
Inhibition der Aktinfilamentbildung, konnte die zentrale Rolle des Aktin-vermittelten
Signalwegs in einem funktionellen Assay nachgewiesen werden. Die Ausbildung von Zell-
Zell-Kontakten und die mesenchymale Kondensation wurde dabei komplett verhindert.
Zentrale Signaltransduktionswege der mesenchymalen Kondensation waren nach sechs
Stunden der in vitro-Kondensation hochreguliert, unter anderem der TGFß-, der Activin-,
der FGF-, der Wnt- und der Notch-Signalweg. Der aufgrund der RNA Seq Ergebnisse
uneindeutige Aktivierungsstatus des MAPK-Signalwegs konnte in einem weiteren funktion-
ellen Assay mithilfe eines Reportervektors geklärt werden. Innerhalb von Minuten wurde
die MAPK-Aktivität unmittelbar als Reaktion auf die in vitro-Kondensation getriggert.
Nach vierstündiger Kultur im Kondensationsmodell konnte die Kollagensekretion mittels
eines Kollagen-bindenden-eGFP-Fusionsproteins in Echtzeit beobachtet werden. Weitere
Proteinanalysen per Immunhistologie zeigten nach viertägiger Kultur die Anwesenheit
der extrazellulären Moleküle TNC, FN, Collagen Typ I und IV. Die Genexpression von
DSPP, TGFß1, Activin und BMP7 wiesen nach vierwöchiger Langzeitkultur auf eine
odontogene Differenzierung innerhalb der Kondensate hin. Die Funktionalität im Hinblick
auf Induktivität und reziproker Interaktion mit Epithelzellen, konnte in Co-Kulturansätzen
gezeigt werden, wobei die epithelialen Zellen in das Mesenchym invaginiert sind und dort
differenziert sind. Zusammengefasst kann in dieser Arbeit gezeigt werden, dass das an-
gewandte Zellkulturmodell einem funktionellen humanen Organogenesemodell entspricht.
Der initiale Schritt der mesenchymalen Kondensation und anschließende odontogene Dif-
ferenzierung führen zur Bildung eines induktiven Zahnprimordiums, das mit epithelialen
Zellen in Interaktion treten kann. An diesem Modell können zukünftige weitere Studien zur
Aufklärung grundlegender Mechanismen der Organentwicklung stattfinden. Die produzier-
ten Zahnkeime sollen in naher Zukunft Anwendung in Studien für regenerative Therapien
finden.
Abbreviations
7-AAD 7-aminoactinomycin D
AKT protein kinase B
bmMSC bone marrow-derived mesenchymal stromalcell
BMP bone morphogenic protein
CD cluster of differentiation
cDNA complementary DNA
CDS coding sequence
CFSE carboxyfluorescein succinimidyl ester
CNA collagen adhesin
COL collagen
CytD Cytochalasin D
DAPI 4,6-diamidin-2-phenylindol
DEJ dentin-enamel-junction
DMEM Dulbecco’s modified Eagle medium
DNA deoxyribonucleic acid
DPC dental pulp cell
DPSC dental pulp stem cell
DSPP dentin sialophophoprotein
ECM extracellular matrix
eGFP enhanced green fluorescent protein
EMT epithelial-mesenchymal transition
FACS fluorescence-activated cell sorting
FC fold change
FCS fetal calf serum
FDR false discovery rate
FGF fibroblast growth factor
FN fibronectin
FOXO forkhead box protein O
FPKM fragments per kilobase of transcript permillion mapped reads
FSC forward scatter
GAG glycosaminoglycan
GAPDH glyceraldehyde 3-phosphate dehydrogenase
GO gene ontology
GSEA Gene Set Enrichment Analysis
HEK293T human embryonic kidney cell line
HGF hepatocyte growth factor
INHBA inhibin ß A subunit
iPSC induced pluripotent stem cell
LiCl Lithiumchlorid
MAPK mitogen-activated protein kinase
mRNA messenger RNA
MSX1 msh homeobox
NCAM neural cell adhesion molecule
NES nestin
OD optical densitiy
PAX paired box
PBS phosphate buffered saline
PCA principal component analysis
PDL periodontal ligament
PEI polyethylenimine
Pi3K phosphatidylinositol-3-kinase
PS penicillin/streptomycin
PTX pertussis toxin
qRT-PCR quantitative reverse transcriptionpolymerase chain reaction
RE response element
RNA ribonucleic acid
RPM reads per million
SDC syndecan
SEM standard error of mean
SMAD mothers against decapentaplegic homolog
SSC side scatter
TF transcription factor
TGF transforming growth factor
TNC tenascin
ULA ultra-low attachment
WNT Wingless-type MMTV integration site
Contents
1 Introduction 1
1.1 Tooth anatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Dental Pulp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Dentin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.3 Enamel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.4 Periodontium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Tooth development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 Thickening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Bud stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.3 Cap stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.4 Bell stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.5 Amelogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.6 Dentinogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.7 Tooth eruption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.8 Permanent dentition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.9 Signaling pathways in tooth development . . . . . . . . . . . . . . . . 17
Transforming growth factor receptor pathway . . . . . . . . . . . . . 17
1.2.10 Mesenchymal Condensation . . . . . . . . . . . . . . . . . . . . . . . . 21
1.3 State of the Art - Tissue Engineering . . . . . . . . . . . . . . . . . . . . . . 23
1.4 Aims of this work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2 Materials & Methods 27
2.1 Cell isolation and cell culture . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.1 Isolation of Human Dental Pulp Cells from adult third molars . . . . 29
2.1.2 Isolation of gingival keratinocytes . . . . . . . . . . . . . . . . . . . . 29
2.1.3 Standard cell culture . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.4 Condensation process for formation of artificial tooth primordium . . 29
2.1.5 Co-Culture assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.1.6 In vitro differentiation assays . . . . . . . . . . . . . . . . . . . . . . . 30
2.1.7 Freezing and thawing of cells . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Flow Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.1 Surface marker staining . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.2 CFSE assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.3 Viability assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3 Cell Tracker Red labeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.4 Collagen tagging via CNA35-eGFP . . . . . . . . . . . . . . . . . . . . . . . 32
i
ii Contents
2.5 Histology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.1 Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.2 Hematoxylin/Eosin staining . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.3 Immunohistochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.6 Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7 Reporter assay strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7.1 Cloning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7.2 Transfection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.7.3 Transduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.7.4 Reporter functionality test . . . . . . . . . . . . . . . . . . . . . . . . 36
2.8 Treatment with inhibitory agents . . . . . . . . . . . . . . . . . . . . . . . . 36
2.9 Transcriptome analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.9.1 Reverse transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.9.2 Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) . . . 37
2.9.3 Whole Genome Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.10 Statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.11 Bioinformatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3 Results 41
3.1 Characterisation of Human Dental Pulp Cells . . . . . . . . . . . . . . . . . 41
3.1.1 Isolation and in vitro growth characteristics . . . . . . . . . . . . . . . 41
3.1.2 Surface Marker Expression . . . . . . . . . . . . . . . . . . . . . . . . 42
3.1.3 Expression profile of relevant genes . . . . . . . . . . . . . . . . . . . 44
3.1.4 Differentation capacity . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.2 Mechanisms of ectomesenchymal condensation . . . . . . . . . . . . . . . . . 49
3.2.1 Establishment of 3D cell culture . . . . . . . . . . . . . . . . . . . . . 49
3.2.2 Expression kinetics of relevant genes . . . . . . . . . . . . . . . . . . . 50
3.2.3 Transcriptome of early phase of in vitro condensation . . . . . . . . . 53
3.2.4 In vitro live imaging of collagen deposition . . . . . . . . . . . . . . . 65
3.3 Functional analyses of the in vitro model for mesenchymal condensation . . 66
3.3.1 Analysis of relevant pathways . . . . . . . . . . . . . . . . . . . . . . 66
3.3.2 Inhibition of condensation with small molecules . . . . . . . . . . . . 70
3.3.3 Interaction of inductive condensates with epithelial cells . . . . . . . . 71
4 Discussion 79
4.1 Characteristics of dental pulp cells . . . . . . . . . . . . . . . . . . . . . . . . 79
4.2 Culture model for tooth development . . . . . . . . . . . . . . . . . . . . . . 84
4.3 Early phase of condensation . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.4 Co-culture with epithelial cells . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.5 Conclusion and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Bibliography 97
Contents iii
5 Appendix A 115
5.1 Supplemental figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.2 Supplemental videos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.3 List of differentially expressed genes . . . . . . . . . . . . . . . . . . . . . . . 130
Publications 145
iv Contents
CHAPTER 1
Introduction
Mammalian teeth represent the mineralized organ appendage located in the oral cavity
as part of the gastrointestinal tract. Their functions are diverse, including mastication,
phonetic articulation, aesthetic perception and they comprise an immunological niche.
Oral health status is associated with overall health, and meta-studies indicate correlation
between poor oral health and cardiovascular diseases, stroke and negative social impacts
(Humphrey et al., 2008; Montebugnoli et al., 2004; Yoshida et al., 2011). Especially
on account to the aesthetic issue, tooth replacement is in the focus of regenerative therapy
research. Meaningful regenerative therapy research necessitates concise knowledge about
the cellular and molecular processes which occur during embryonic organogenesis and the
anatomy of the fully developed organ.
1.1 Tooth anatomy
The mature permanent tooth organ
is divided into two parts: the crown
and the root. The crown is covered
by the acellular enamel, the hardest
biological tissue in the mammalian or-
ganism. Beneath the enamel, dentin
forms the bulk of the tooth. dentin is
a mineralized rigid but elastic collagen
matrix.
On the root side, the dentin is covered
by a thin layer of cementum. Together
with the periodontal ligament, the alve-
olar bone and the gingiva, it belongs to
the supportive tissues to anchor teeth
in the jaw (Figure 1.1). Dentin is pro-
duced by the innermost cellular part
of the tooth, the odontoblasts in the
dental pulp. The dental pulp is a con- Fig. 1.1: Anatomy of a human mandibular
nective tissue containing the dental molar tooth (Lacruz et al., 2017)).
pulp stem cells, odontoblasts, fibro-
blasts, and it is home for the tooth’s vasculature and innervation.
1
2 Chapter 1 Introduction
1.1.1 Dental Pulp
The dental pulp resides within the rigid pulp chamber and root canals. As a consequence
of its function to produce dentin, a unique tissue organization of this specialized connective
tissue is determined.
The dentin-producing cells (Odontoblasts) represent the effector part of the pulp and
line the pulpal-dentin junction. The mature odontoblast is a polarized columnar cell,
attached to the secondary dentin, and projects into dentinal tubules with long processes
(Fig. 1.2). Odontoblasts deposit secondary dentin throughout life and tertiary dentin upon
injuries (see Chapter 1.2.6 Dentinogenesis). Besides this prime function, the odontoblast
layer functions as selective barrier. The integrity of the barrier is maintained by cell-cell-
junctions between the odontoblast cell bodies, including desmosomes, tight junctions, and
gap junctions. This property preserves the "well-being" of the pulp: a dentinal fluid is
constantly pressed outwards through the dentinal tubules to dilute potential bacterial or
environmental toxins. The release of proinflammatory cytokines by odontoblasts is further
supporting their protective role in the pulp. By regulating the concentration of potassium
and sodium ions in the tubules, the excitability of nerve endings, neighboring the long
odontoblast processes in the tubules, is adjusted. Thereby, the odontoblasts have a role
in sensory reception of heat and coldness. Gap junctions are formed to synchronize the
odontoblasts in the layer.
Figure 1.2: Schematic representation of odontoblasts lining the pulpal-dentin complex and
their projection into the dentin (modified from Nanci, 2012).
All other elements of the dental pulp have supportive functions by protecting, feeding,
replacing and mechanically stabilizing the odontoblast layer.
The cellular composition and tissue arrangement is depicted in figure 1.3. Odontoblasts are
post-mitotic cells and instantly die upon irreversible injury. They are replaced by dividing
1.1 Tooth anatomy 3
ectomesenchymal undifferentiated cells originating from dental pulp stem cells. Under
pathological conditions, their regenerative function is induced by activation signals (About,
2013) to produce tertiary dentin at the injured site to seal the pulp. Dental pulp stem
cells (DPSCs) are multipotent cells. In vivo, they are tissue-specific (Berkovitz et al.,
2009). Thus, these cells of ectomesenchymal origin represent precursor cells and regulate
the dentin-pulp tissue homeostasis. In vitro, these DPSCs are defined by similarities in
gene expression to mesenchymal stromal cells from other tissue sources and odontoblasts.
Immunosurveillance of the pulp is achieved by inflammatory cells that are located through-
out the pulp center. T cells increase in number and infiltrate the odontoblast layer upon
stimulation by dendritic cells in the pulp. Macrophages are involved in pulp turnover by
elimination of dead cells.
Figure 1.3: Histologic overview of the dentin-pulp complex in lower (left) and higher magni-
fication (right)(Nanci, 2012).
The fibroblast is a ubiquitously occuring cell type in connective tissues. In the dental
pulp, it represents the cellular majority. Interestingly, fibroblasts in the dental pulp appear
stellate-like and show direct cell-cell-contacts among each other. Their function is to main-
tain the pulp matrix, consisting of collagens and glycosaminoglycans. Collagens, principally
type I and type III, are the predominant extracellular matrix component (25-32% dry
weight) and form fibrils of around 50 nm in diameter (Amerongen et al., 1983). Collagen
type V and VI are also abundant to a smaller extent, functioning as linking proteins
between the main collagens. Collagen type IV can be detected at the basal lamina between
the pulp and blood vessels, and in the dentino-enamel junction (DEJ) (McGuire et al.,
2014). Other glycoproteins in the pulp are integrins and tenascin. The "ground substance"
of the pulp is composed of water, glycosaminoglycans, and proteoglycans. All four types of
glycosaminoglycans (GAGs) are present in the dental pulp: chondroitin sulfate, dermatan
sulfate, heparan sulfate, and hyaluronic acid. These hydrophilic molecules swell and form
gels in their hydrated form. Also, they allow the movement of ions, water, and nutrients
within the ECM and facilitate cell movement (Berkovitz et al., 2009). In the developing
pulp, chondroitin sulfate is the major GAG, whereas in the mature pulp 60 % of the GAGs
4 Chapter 1 Introduction
are hyaluronic acid. The GAGs are attached to protein cores forming proteoglycans. In
the pulp, these include versican, syndecan, decorin, biglycan, fibronectin, and osteoadherin
(Berkovitz et al., 2009). Molecules exclusively expressed by odontoblast in the periphery
of the pulp to form (pre-)dentin, are dentin sialoprotein and dentin phosphoprotein (see
chapter 1.2.6).
Innervation and blood supply of the dental pulp
Innervation and vasculature enter and exit the tooth
pulp through the apical foramen, the opening in the
root tip. One or two artery vessels of 150 µm enter
the pulp in conjunction with sensory and sympathe-
tic nerve bundles. The vasculature branches in the
coronal area and an extensive vascular capillary net-
work is formed (see figure 1.4). The capillaries supply
the productive odontoblast layer with oxygen and
nutrients. The capillaries do not enter the dentinal
tubules. The pulp blood flow is under nervous control
and it is not surprising, that the dental pulp is richly
innervated. The innervation pattern follows the vas-
culature pattern, beginning with larger bundles in
the center, which extensively branch in coronary
direction. The nerve bundles consist primarily of
sensory afferent fibers and sympathetic branches. A
small number of nerves pass the odontoblast layer
and enter the dentinal tubules to reach the dentinoe-
Fig. 1.4: Vascular cast of a dog namel junction. Together with the odontoblasts that
tooth pulp. Arterioles occupy regulate the interstitial composition, they account
a central position. Beneath the
dentin a high degree of branch- for of the "dentin sensitivity". Dentin sensitivity is
ing is visible (Takahashi et al., responsible for the well-known thermal and tactile
1982). sensation which causes pain and discomfort.
1.1.2 Dentin
Dentin is a yellowish mineralized, yet elastic avascular tissue that encloses the dental
pulp. It is a remarkable tissue since it compensates for mastication forces that act on the
brittle enamel. dentin has a unique tubular architecture to withstand great compressive
and tensile strength. It is a sensitive tissue and capable of repair. Concerning volume,
dentin is composed of 50 % inorganic, 30 % organic compounds and 20 % water. The
mineral component is formed as hydroxyapatite crystals with around 27 % calcium, 13
% phosphorus and carbonate, sodium and magnesium. The crystals are embedded in an
organic matrix which is synthesized by the odontoblasts. This organic matrix is similar to
bone and is composed of 90 % collagens (Type I, Type II, Type V). The collagens form
1.1 Tooth anatomy 5
large fibrils of 100 nm and run parallel to the pulpal surface. Less extant, yet essential are
the other organic matrix components in the dental pulp, such as dentin phosphoproteins,
proteoglycans, osteonectin, osteopontin and growth factors. The dentin phosphoproteins
are the most acidic proteins due to their high phosphate content. Their calcium binding
activity is involved in mineralization of the dentin matrix. Proteoglycans allow collagen
assembly and have a role in cell signal transduction in many processes, such as proliferation,
differentiation, adhesion and migration. Several growth factors, e.g., bone morphogenic
protein (BMP)-2, transforming growth factor (TGF)-ß1 and insulin growth factor (IGF)-II
could be isolated from the dentin matrix. They are suggested to play a role in carious
destruction and induction of repair.
The dentin is traversed by dentin tubules. They contain extracellular fluid and cellular
processes of odontoblasts as well as of neurons. They have a diameter of 1 to 2 µm and
play a major role in tooth sensitivity. Histologically a distinct layer can be observed at the
innermost border to the pulp. This matrix is called pre-dentin and contains the deposited
dentin matrix protein before mineralization by the odontoblast processes to form mature
dentin (see chapter Dentinogenesis 1.2.6).
1.1.3 Enamel
Enamel covers the tooth crown and is the hardest biological tissue. It has an enormous
abrasion resistance and withstands shearing and impact forces. Enamel cannot be restored
biologically since the producing cells, the ameloblasts no longer exist in the mature tooth.
Along with the hardness and stiffness comes the brittleness. Luckily, this feature is
compensated by the underlying elastic dentin. The principal component of enamel is
calcium hydroxyapatite which is architectured in the form of crystallites forming prisms.
They represent 90 % of the volume of enamel.
1.1.4 Periodontium
The tooth is held in the jaw by three supporting tissues
which together are termed as periodontium. The period-
ontium comprises the periodontal ligament (PDL), the
alveolar bone and the cementum. Cementum covers the
dentin at the root side. The key role of the cementum
is to maintain the integrity of the root, including rege-
neration and positional correction. On the inner surface,
cementum adheres to dentin and on the outer side it at-
taches to collagen fibres of the periodontal ligament. The
fibres of the PDL reach into the organic matrix of the
mineralizing cementum and are embedded into the mature
cementum. The periodontal ligament is a fibrous connec-
tive tissue between the root and the alveolar bone. The
periodontal ligament has important dental functions. Dur-
ing organogenesis it has a specific role in eruption while in Fig. 1.5: Schematic repres-
mature teeth it supports the tooth in the jaw. During mas- entation of the periodontium
ticory loading the PDL tissue acts as suspensory tissue and constisting of the alveolar bone,the periodontal ligament, the ce-
comprising mechanoreceptors are involved in neurological mentum and the gingiva (modi-
control of mastication. In figure 1.5 the innervation of the fied from Nanci, 2012).
PDL is schematically depicted. In contrast to the fibrous
6 Chapter 1 Introduction
dental pulp tissue, the collagens of the PDL are arranged in stong bundles of 5 µm, which
reflects the mechanical demands on that tissue. The predominant cell type in the PDL is the
fibroblast, but since this tissue is regenerative, a proportion of undifferentiated PDL stem
cells can be found. Furthermore, the tissue is home for defence cells, cementoblasts at the
cementum border, osteoblasts lining the tooth socket as well as osteo- and cementoclasts.
Intriguingly, an epithelial remain of the developmental root sheet is located in the PDL,
referred to as epithelial rest of Malassez. Several groups postulate that these remnant cells
represent an epithelial stem cell source for tissue engineering and experimental models to
investigate tooth growth (Rincon et al., 2006; Tsunematsu et al., 2016). The alveolar
process represents the compact bone which lines the alveolus (tooth socket). It is consisting
of a buccal and a lingual cortical plate both meeting the cortical jaw bone 2 mm below
the cementoenamel junction. The alveolar bone is perforated with traversing vessels and
nerves. Collagen fiber bundles from the PDL are embedded into the bone matrix at the
PDL-alveolar border. Thereby, a close interaction between these two supporting tissues
is guaranteed. The alveolar bone shows high plasticity. As other bone tissues in the
body, the alveolar process continously remodels to adapt to functional situations. This
remodeling is accomplished by a balanced interaction of osteoblast mechanotransduction,
bone metabolism, and osteoclast-mediated bone catabolism.
Figure 1.6: Longitudinal cut through the mandible showing the alveolar bone lining the
tooth socket (Nanci, 2012).
1.2 Tooth development 7
1.2 Tooth development
In mammalian embryogenesis, the blastocyst has formed at day 5 to 6 after gestation. It is
divided into the trophoblast which gives rise to extraembryonic tissue and the embryoblast
which gives rise to the embryo itself. This embryoblast undergoes differentiation and during
week three of development, gastrulation starts. After the first two germ layers have formed
(ectoderm dorsally and endoderm ventrally), now the third germ layer, the mesoderm
forms between the two other layers through the development of the primitive streak. From
this three-layered embryo, all tissues and organs develop in the following weeks. The head
and face, including teeth, depend on differentiation of the neural crest from the ectoderm,
mesodermal differentiation, embryonic folding in the lateral and rostrocaudal axis, and
neural tube development. When the neural tube reaches the head region, it massively
expands to form the brain. The folding tissue at the head gives rise to the face with the oral
cavity, which is formed by the ectoderm and the endoderm. Thereby a future connection
between the oral cavity and the gut will form. Cells from the neuroectoderm separate,
undergo epithelial-mesenchymal transition which then allows them to migrate laterally
through the embryo (see figure 1.7, green cells). This process is marked by the expression of
the transcription factors Snail, and furthermore, morphogens (bone morphogenic proteins,
Wnt and fibroblast growth factors) play a critical role. Interestingly, additionally to the
sensory system, the neural crest cells form all connective tissues. This is special since, in
the rest of the embryo, connective tissue is formed from mesoderm. The next steps include
the development of the skull, the face with lips, the palate, the tongue, and the mandible,
and maxilla with joints and teeth. In the following paragraphs, only the process of tooth
formation will be delineated.
Figure 1.7: Neural tube formation and migration of neural crest cells. Upon signaling of the
notochord, the neural plate folds and forms the neural tube. Neural crests cells (green) develop
at the edges of the closing plate, migrate laterally through the embryo to form craniofacial
structures. Image modified from Pearson Education Inc.
In the dental context, it is important to understand that teeth form from an interplay
between the ectodermal epithelium and the ectomesenychme. Odontogenesis is divided
into three major phases: initiation, morphogenesis, and histogenesis. During initiation
phase, the sites of the future tooth organs are established, and early tooth germs form.
The tooth shape (morphogenesis) is determined by cell proliferation, differentiation and
movement in response to morphogen gradients in the oral cavity. The progression of
8 Chapter 1 Introduction
differentiation leads to histogenesis of the different dental tissues, both mineralized and
unmineralized. Histologically, the developing tooth germs are classified into the following
stages: thickening, bud stage, cap stage, and bell stage.
1.2.1 Thickening
During week six after gestation, a thickened epithelial band forms around the future jaw.
At the site of the presumptive tooth formation, this band forms a placode (see figure 1.8)
and invaginates into the mesenchyme that has formed a mesenchymal condensate. It is
generally accepted that the epithelial invagination induces the mesenchymal condensation,
but recent studies suggest that mechanical strains directly generated by mesenchymal
condensations produce local curvature, namely evagination and invagination, at tissue
interfaces (Hughes et al., 2018). The signaling events involved in placode formation and
mesenchymal condensation interaction are conserved among species, and the underlying
mechanisms apply to all ectodermal appendages, such as teeth, hair, feathers, nails and
glands. The invaginated epithelial band is referred to as dental lamina.
Most molecular analyses have been conducted in mouse
models. Therefore, the majority of the proposed signaling
cascade is based on murine research for understanding develop-
mental pathways during odontogenesis. An exact translation
to human signaling events, especially on the involved (homo-
logous) genes, is not possible. In murine models, an interplay
between stimulatory (FGFs, Wnts, Shh) and inhibitory signals
(BMPs) determines the site of the tooth initiation. PITX2
expression marks the oral epithelium, and SHH marks the
site where the teeth develop. As also shown in other tissue
developmental processes (neural tube patterning, hair, gut,
bladder) epithelial SHH cross-talks with mesenchymal Wnt
(WNT7B) to form the organ boundaries (Ahn et al., 2010; Fig. 1.8: Histologicimage of the dental
Lei et al., 2006; Rishikaysh et al., 2014; Shin et al., 2011). placode (Nanci, 2012).
The earliest mesenchymal marker for dental capability are
the LIM-homeobox (Lhx) domain genes LHX6 and LHX7 whose expression is induced by
FGF8 in the first branchial arch. PAX9 (paired-box transcription factor 9) expression in
the mesenchyme marks the site of tooth formation. The PAX9 expression results from
morphogen gradients between FGF8, BMP2, and BMP4 which pattern the oral-aboral sites.
Together with Activin A expression, in the ectomesenchyme PAX9 expression is associated
with the condensation and epithelial budding, respectively. More than 90 genes have been
identified to play important roles in tooth initiation, all enmashed in a complex regulatory
network. Interestingly, once the ectomesenchyme has acquired the program to form teeth,
it will not lose it. Ex vivo cultured dedifferentiated tooth germ ectomesenchymal cells
will re-form teeth when recombined with dental epithelial cells. Even more intriguing is
the observation that murine dental ectomesenchyme induces tooth formation even when
recombined with chicken epithelium (E. Kollar et al., 1980; Thimios A Mitsiadis,
Chéraud et al., 2003), although the avian taxon lost the dental competence 116 million
years ago (Meredith et al., 2014).
1.2 Tooth development 9
1.2.2 Bud stage
The following transitory developmental stage is referred to as bud stage. During that
stage, the two adjacent components (epithelium and mesenchyme) are separated by a basal
membrane. The developing tooth organ at this stage appears as a compact ovoid with
densely packed ectomesenchymal cells around the epithelial bud. In the following step,
the tooth type (incisor, premolar, molar) and tooth shape is determined. This stage is
referred to as bud-to-cap transition. The tip of the bud invaginates (see figure 1.9, center
(arrowhead)) and will form the primary enamel knot. Formation of secondary knots will
form at the site of future tooth cusps.
Figure 1.9: Bud-to-cap transition in murine tooth development (E13.5 to E14.5). Green
signal indicates cell membranes and red represents collagen type IV. Arrowhead indicates tip
invagination (Takigawa-Imamura et al., 2015).
It is suggested that tooth type specification is related to the odontogenic homeobox code
(see figure 1.10). This "field model hypothesis" is based upon a patterned expression of
various homeobox genes (MSX1, MSX2, BARX1, DLX1, DLX2, ALX3). This patterned
expression of transcription factors is established by overlapping morphogen gradients of
FGFs and BMPs (McCollum et al., 2001). FGF8 and FGF9 are expressed laterally
over the molar field, and BMP4 is expressed medially over the incisor field. BMP4 is in
a regulatory feedback loop with MSX1 and is expressed in the epithelium as well as in
the mesenchyme. Contrarily, in human dental development, the BMP4 expression is not
restricted to the incisor region (Lin et al., 2007).
1.2.3 Cap stage
By the 11th week of human development, the morphogenesis of the tooth organ has
progressed: the epithelium, or enamel organ, invaginates deeper and forms a cap on the
dental mesenchyme (see figure 1.9, right). The condensed mesenchymal tissue is called
dental papilla and will form the pulp and dentin. In histological sections also a mesenchymal
"sac" surrounds the organ, referred to as dental follicle which gives rise to the periodontium.
The enamel organ cells produce extracellular matrix, and the cells are distantly separated.
Still, they keep direct cell-cell-contact via desmosomes and through the pushing forces,
the cell become star-shaped (stellate reticulum). The complete structure is called tooth
germ at this stage. At that stage, the dental follicle becomes innervated. Interestingly,
due to nerve-repelling factors, the innervation does not enter the pulp until amelo- and
dentinogenesis take place.
10 Chapter 1 Introduction
Figure 1.10: The odontogenic
homeobox code in the developing
dental mesenchyme, putatively medi-
ated by epithelial FGF8 and BMP4
morphogens. Domains of BARX1
and DLX1/2 expression overlap in
the molar region, while MSX1/2 and
ALX3 overlap in the incisor region
(McCollum et al., 2001).
1.2.4 Bell stage
Fourteen weeks after gestation morphodifferentiation has further developed and a bell
shape forms in the tooth germ. Up to that stage, the germ was connected to the oral
epithelium through the dental lamina. Now, this dental lamina degenerates. During that
stage, the crown shape is finally determined. The inner enamel epithelium folds inwards
and encompassed the mesenchyme (see figure 1.11). By taking a closer look at the tips
of the enamel organ, which are deepest in the mesenchyme, a zone where the outer and
inner epithelial cells meet, is visible. This region is called the cervical loop. The outer
enamel epithelium, which borders the periphery, appear cuboidal whereas the inner enamel
epithelial cells, which line the dental papilla, assume a columnar shape (see figure 1.12).
Figure 1.11: Histologic image of
the bell stage of tooth development
(Nanci, 2012). The dental lamina
degenerates and the inner enamel
epithelium folds inwards and finally
shapes the tooth.
1.2 Tooth development 11
The cells of the loop have a
high mitotic activity and prolif-
erate until the crown attains its
full size. As indicated before,
within the inner enamel epithe-
lium a delicate structure forms,
the enamel knot. Enamel knots
are locally well-defined clusters of
non-dividing cells. The enamel
knot shares similarities with ec-
todermal ridges in limb develop-
ment. These act as major signal-
ing centers to ensure proper de-
velopment of limbs. Both express
FGFs, BMPs, and MSX2. Early in
bell stage, vascularization of the
pulp starts with small branches
from the vascular trunks, entering
the papilla base. They form a bed
of capillary network in the odon- Fig. 1.12: Histologic image of the cervical loop with
toblast layer. Occasionally small detailed structure of the inner and outer epithelium
loops enter the dentin (see figure (modified from Poja Collection (3.5 POJA-L142 and
1.4 and chapter 1.1.1). Spatiotem- L145), L.G. Poels PhD and P.H.K. Jap PhD)
porally with the blood vessels, also the sensory nerve fibers enter the pulp, whereas
sympathetic innervation follows later.
The late bell stage commences at week 18 after gestation and is associated with the
deposition of hard tissues. Detailed information on the enamel and dentin formation are
given in the subsequent chapters. Briefly, the adjacent cells from the pulp and the enamel
organ begin to differentiate to pre-secretory cells. Both cell types acquire a columnar
phenotype, and by deposition of matrix, they form the dentinoenamel junction (DEJ) (see
Figure 1.13). The whole process starts with the deposition of a layer of dentin which acts
as signaling inducer for further cytodifferentiation of the ameloblasts and odontoblasts.
12 Chapter 1 Introduction
.
Figure 1.13: Histologic images from human late bell stage with Azan staining. (A) Complete
tooth germ in late bell stage. (B) Higher magnification of the cuspal tip. (C) Detailed image
of the adjacent cell layers of ameloblasts (top) and odontoblasts (bottom). In between, a layer
of dentin has been deposited (blue). (D) Later in development, more dentin (blue) and initial
enamel containing amelogenin (red) has been deposited (from Poja Collection (3.5 POJA-146,
L154 and L164), L.G. Poels PhD and P.H.K. Jap PhD)
1.2.5 Amelogenesis
The epithelial-mesenchymal interactions in the dentinoenamel junction constitute the
initiation of the enamel protein secretion by the inner enamel epithelium. During presecret-
ory stage, the inner enamel epithelial cells undergo a morphogenetic phase throughout
the bell stage of odontogenesis. These cells divide mitotically and are cuboidal or low
columnar. A basal lamina separates the inner enamel epithelium from the underlying
mesenchyme. At the future cusp tips of the tooth, differentiation to preameloblasts begins.
The cells become columnar and polarized. Substantial reciprocal interactions between
the preameloblasts and the preodontoblast allow cytodifferentiation on both sides (see
figure 1.14). Preameloblasts deposit TGFß ligands in the basal lamina which promote
odontoblast differentiation. The basal lamina becomes degraded resulting in an intimate
contact between the future ameloblasts and odontoblasts. This contact is lost when the first
dentin is laid down by the odontoblast, and also preameloblast secrete dentin sialoprotein
transitory. Dentin proteins secreted by odontoblasts (non-collagenous proteins and dentin
sialophosphoprotein (DSPP)) provide signals for the ameloblast to also begin secretion. At
the beginning of the secretory phase, the distal ends of the ameloblasts (facing towards
the odontoblasts), develop an extension, called Tomes’ process (see figure 1.14). The
shape of the Tomes’ processes is responsible for the prismatic pattern of enamel. In these
cellular processes, enamel proteins are translated and packaged into secretory granules
being constitutively secreted. Mineralization and subsequent crystallization immediately
take place interdigitately with the first dentin layer. With growing matrix deposition, the
1.2 Tooth development 13
ameloblasts are displaced from the DEJ, and the Tomes’ processes deeply penetrate the
forming enamel. The mineralizing front of the enamel appears as a honeycomb structure,
consisting of interrod structures, as a result from the proximal part of the Tomes’ processes,
and of enamel crystal rods, formed by the distal proportion of the processes (see Figure
1.15). The immature enamel has a high content of water and proteins, a low content of
carbonated hydroxyapatite including HPO2–4 , CO2–3 , Na+, F-, and is porous. Once the
full thickness of enamel matrix has been laid down, the Tomes’ processes degenerate and
become flattened and ruffle-ended. Before the tooth erupts, the enamel is matured by
mineralization. During a transitional phase, the ameloblasts shorten and around 50 %
undergo apoptosis.
Figure 1.14: Schematic illustration of differentiation processes involved in enamel development.
(a) Pre-secretory: At the interface of the inner enamel epithelium (IEE) and the dental
mesenchyme (DM), matrix deposition is initiated. The two differentiating cell layers of pre-
odontoblasts (Pre-Od) and pre-ameleoblasts (Pre-Am) are separated by a basal membrane
(BM) (b) Secretory: Beginning at the cusp tips, the epithelium will differentiate to ameloblast
(Am) and will produce Enamel (En) after dentin (De) has been deposited by the odontoblast
(Od). Note the special form of the secretory ameloblasts at the proximal end (Tomes’ process;
TP). The deposited enamel has two alternating configurations: interrod (IR) and rod (R)
enamel (either as result of the proximal or distal border of the Tomes’ process). (c) Maturation:
Enamel proteins will be degraded and replaced by hydroxyapatite crystals. During this stage,
the ameloblast become shorter and exhibit transitory phenotypes with ruffle-ends (resorptive)
or smooth ends (secretory). BV - blood vessels, PL - periodontal ligament, SI - stratum
intermedium (B Kulkarni et al., 2017).
During enamel maturation, the bulk of water and organic material is removed by the
ameloblasts and the space is filled with additional inorganic material (calcium, phosphate
and carbonate ions). Thereby, the crystal seeds from the secretory stage harden and
enlarge in diameter. The enamel proteins that were laid down during secretory phase now
play a role in the maturation. 90 to 95 % of the proteins are amelogenins, and 5-10 %
are non-amelogenins such as enamelins, ameloblastin, tuftelin, and proteolytic enzymes.
Amelogenins are hydrophobic low-molecular proteins that undergo extensive extracellular
processes by proteolytic enzymes (MMP20 and kallikrein). The resulting pieces all have
different functions in crystal growth. As described above, mineralization runs almost
parallel to secretion. That means that the ameloblasts modulate their phenotype from
14 Chapter 1 Introduction
secretory (smooth ended) to resorptive (ruffle-ended) about five to seven times in their
lifespan.
Figure 1.15: Left: Schematic representation of the organization of secretory stage ameloblasts.
Top right: Enamel crystal alignment within the enamel rods. Bottom right: a transversal view
on enamel exhibits a honeycomb structure with the hexagonal rod crystallites surrounded by
inter-rod enamel (Gasse et al., 2013; Nanci, 2012). Bar represents 20 µm.
1.2.6 Dentinogenesis
In parallel with the differentiation of the inner enamel epithelium during bell stage, also the
differentiation of the adjacent cells of the dental papilla occurs. The ectomesenchymal cells
lining the basal membrane elongate rapidly, polarize and turn on their protein-synthesizing
machinery. This odontoblast differentiation is thought to be induced by the reciprocal
epithelial-mesenchymal interaction, namely from signaling factors (BMPs, SHH, Wnt,
FGFs) expressed from the enamel knot and deposited in the basal lamina. In preparation
to matrix secretion, the surface of the odontoblasts is firstly enlarged by the formation of
1.2 Tooth development 15
numerous small cell processes towards the basal membrane. This border is later reduced
to single, thin and long processes, which occupy the dentin tubuli (Tomes’ fiber). After
differentiation, the organic matrix produced by the odontoblasts, is the basis for dentin.
Large collagen type III fibrils associated with fibronectin are the first product of secretion
(Korff’s fibers) (Abrahão et al., 2006). Afterwards, also smaller fibrils of collagen type I
are laid down parallel (in mantle dentin) and later orthogonal (in circumpulpal dentin) to
the later DEJ. Through constitutive deposition of matrix the thickness of the predentin
enlarges. The plasma membrane of the odontoblast extends, and the Tomes’ fiber (long
single process) is left behind in the matrix and thereby elongates (see figure 1.2). The
odontoblast secretes dentin sialophosphoprotein (DSPP). By protein cleavage, this precursor
results in dentin phosphoprotein (DPP) or phosphophorin and dentin sialoprotein (DSP).
DPP represents the most acidic protein known with an isoelectric point of 1, due to the
very high phosphate content. Its high calcium ion binding properties have been implicated
in the role of the mineralization process of the dentin. The mineralization process occurs in
two patterns: globular and linear calcification. In mantle dentin, calcification is initiated by
deposition of several crystal seeds, which grow in size, capturing the collagen matrix, and
then fuse (globular calcification). In circumpulpal dentin, also mineralization takes place,
which is under control of the protein DPP. Calcium ions are attracted to the predentin,
allowing the growth of crystals or calcospherites. The calcospherites are located in between
the collagen triple helices. This marks the mineralization front from where hydroxyapatite
formation nucleates with a growth parallel to the collagen fibers (see figure 1.16).
The formation of secondary dentin is carried out throughout life-, but is performed at
a much slower rate. The composition is indistinguishable from primary dentin, only the
direction of the dentinal tubules is different. Secondary dentin formation begins after
completion of root formation as the tooth occludes. Tertiary dentin is only produced upon
severe injury of the dentin (e.g., through caries) and depending on the severity of injury it
is described to be slightly irregular regarding tubule patterning or even completely atubular
and bone-like.
Figure 1.16: Section showing the region of min-
eralized dentin. (A) predentin, (B) mineralized
dentin, (C) calcospherites (Berkovitz et al.,
2009)
16 Chapter 1 Introduction
1.2.7 Tooth eruption
Tooth eruption starts once the roots begin to develop. The process is remarkable, nowhere
else in the body, an organ leaves an intrabony location. In summary, tooth eruption
describes the axial movement of the tooth from its alveolar crypt into the oral cavity.
This includes root formation, bone remodeling, and formation of the periodontal ligament.
During root development, the roots’ hard matrix, the cementum has to be produced.
Cementoblasts derive from the dental follicle cells, and similarly to the odontoblasts, they
produce collagens and other organic material that will be mineralized. The cells of the
dental follicle also give rise to the fibroblasts of the periodontal ligament. They produce
a strong connective tissue that connects to bone and cementum and thereby ensures
attachment of the tooth organ. It is still not understood what initiates tooth eruption.
Undoubtedly, the complex process requires a well-regulated sequence of events. The tooth
germ expresses parathyroid-hormone-related protein and interleukin 1 as well as TGFß1
and epidermal growth factor to attract monocytes and regulate osteoclastogenesis. For
bone resorption of the overlying alveolar bone, mononuclear cells enter the eruption site to
form osteoclasts. They are recruited by macrophage-colony stimulating factor (MCSF) and
monocyte chemotactic protein (MCP)-1, released from the bone. As the tooth reaches the
proximity of the oral cavity, the outer cells of the tooth germs, namely the reduced enamel
epithelial cells, secrete enzymes to degrade the overlying connective tissue of gingiva. After
eruption, these cells unite with the gingiva epithelium to form the dentinogingival junction.
1.2.8 Permanent dentition
In humans and most mammals, there is a limited succession of teeth. To account for the
growth of the face and jaw, the deciduous teeth are replaced by permanent teeth during
physical development in childhood. The described dental lamina that gives rise to the
primary, deciduous dentition, also generates the permanent dentition (see figure 1.17). At
the incisor, canine and premolar site, the dental lamina exhibits a further proliferative
activity so that a second tooth bud forms at the lingual aspect of the deciduous tooth germ
(see figure 1.11). The permanent molars have no predecessor. They arise from the dental
lamina that grows posteriorly under the oral mucosa once the jaw has grown long enough.
The subsequent odontogenesis is virtually identical between deciduous and permanent
teeth although induced at different times of organism development.
Figure 1.17: Schematic representation of tooth formation depicting the developed tooth buds
from the primary dentition and the small bud at the dental lamina (darker blue) that give rise
to permanent teeth (Nanci, 2012).
1.2 Tooth development 17
1.2.9 Signaling pathways in tooth development
Reciprocal and sequential tissue interactions have been identified as the key inducers and
regulators of embryonic development. All inductive interactions in organisms follow a
conserved principle with conserved key players. The nature of the interaction is a chain
reaction of ligand 1 expression by tissue A, reception on tissue B that induces a signaling
cascade to activate gene transcription of ligand 2 in tissue B and subsequent reception of
ligand 2 on tissue A, which induces the expression of a further signaling molecule in tissue A,
etc. As discussed earlier, the reciprocal tissue interactions also apply to tooth development.
The Wnt, BMP, FGF and SHH pathways as well as their respective cross-talks are central
pathways in the reciprocal tissue interaction and will be described in detail here (see figure
1.18). The majority of current understanding of cellular and molecular mechanisms are
based on genetic analyses in the mouse as well as tissue recombinations.
Figure 1.18: Regulation of tooth formation by reciprocal interactions of epithelial and
mesenchymal tissue. The molecular signaling chain reaction is mediated by BMP/TGFß, FGF,
SHH, and Wnt signaling molecules. The expression sites of transcription factors are listed in
bold boxes (modified from Nanci, 2012; Thesleff, 2018). For detailed information see text.
Transforming growth factor receptor pathway
The TGFß superfamily includes more than 30 ligand members which are grouped in the
TGFß, the BMP/GDF, the Activin/Inhibin and the GDNF subfamily. They all share the
characteristic of binding to a hetero-tetrameric receptor complex with a serine/threonine
kinase domain and thereby activate an intracellular SMAD-signaling. The activated
receptor complex always consists of two type I and two type II subunits. Once the ligand
binds the type II receptor dimer, it recruits the type I receptor dimer to form the fully
functional receptor complex. The serine/threonine kinase domain of type II receptor
subsequently phosphorylates a glycine-serine-rich domain on the intracellular domain of
18 Chapter 1 Introduction
type I receptor. This leads to a conformational change to expose the catalytic site of type
I receptor and the prior phosphorylation acts as a binding site for the downstream effector
molecule SMAD. The SMAD, in turn, is phosphorylated by the type I receptor and is
thereby activated to dimerize with a co-activating SMAD resulting in the release from
the inhibitor. This enables the dimer to enter the nucleus where it binds transcriptional
promoters and cofactors to regulate transcription of target genes (Massagué et al., 2000).
There exist several levels of regulation and fine-tuning of the TGFß-signaling pathways.
For example, co-receptors can act as "storage" for ligands (e.g., betaglycan). Since the
TGFß superfamily ligands represent the epitome of morphogens, they act in morphogenic
gradients in the extracellular matrix (ECM) between the signaling and the receiving cell in
developing tissues. This gradient is realized by the attachment of the ligands in the ECM,
the expression of inhibitors/antagonists of the ligands or receptor subunits, by expression of
proteases to restrict their sphere, or by the release of proteases by the target cell to liberate
the ligand from the ECM. Expression of these proteases is again subject to complex levels
of regulation, and often an autoregulatory feedback loop with the ligand itself is involved.
On the intracellular level, the signaling is regulated by the balance of availability of receptor
molecules, of SMADs and inhibitors of the intracellular signalosome. This includes the
presence of inhibitory SMADs that compete with Co-SMADs and guide the SMAD dimer
to the proteasome. It becomes evident that the fine-tuned signaling cascade induced by
TGFß superfamily ligands is highly complex and there are many more inhibitory events
that could not be discussed here.
In embryonic tissue development, the TGFß and the BMP pathway seize opposing roles.
They bind to different type I receptor complex types: TGFß and Activin/Inhibin ligands
bind to ALK4/5/7 and subsequently SMAD2 or SMAD3 dimerized with the Co-SMAD 4,
whereas the bone morphogenic proteins bind to ALK1/2/3/6 and thereby activated SMAD1,
SMAD5 or SMAD8 to dimerize with SMAD4. By that they compete on the SMAD4
abundance and once the balance shifts to one side (e.g., through elevated stimulation), the
SMAD4 abundance is no longer sufficient for the other side and indirectly this pathway is
inhibited. These opposing roles come into play during epithelial-mesenchymal-transition of
embryonic development. The transition from epithelial cells to mesenchymal cells and vice
versa is a central process in tissue differentiation and morphogenesis. Epithelial cells are
not capable of migration, and therefore, whenever they have to leave a specific location in
the body, beginning at gastrulation, they have to acquire a mesenchymal phenotype which
is established by TGFß signaling. When cells lose their migratory phenotype and polarize
(e.g., during somitogenesis), they underwent mesenchymal-epithelial-transition induced by
BMPs. A hallmark publication of Zeisberg et al. shows the opposing roles of BMP7 and
TGFß1 in renal fibrosis and regeneration (Zeisberg et al., 2003).
During early murine tooth development, a vibrant activity of BMP2 and BMP4 in the
dental epithelium and mesenchyme is described. The BMPs not only determine the tooth-
forming site and tooth type. Interestingly, they are also implicated in bud-to-cap stage
transition, whereby the odontogenic potential shifts from the epithelium to the dental
mesenchyme. Tissue recombinations have shown that the prime potential to instruct tooth
formation lies in the dental epithelium. However, after the initiation stage at murine E12
(day 12 after gestation), the dental epithelium loses its odontogenic potential, which is
1.2 Tooth development 19
then shifted to the ectomesenchyme (Mina et al., 1987). Epithelial BMP4 induces the
expression of MSX1 in the dental ectomesenchyme which in turn, induces mesenchymal
BMP4 expression (Vainio et al., 1993a). PAX9 and MSX1 act as synergistic transcription
factors on the Bmp4 promoter. In bud-to-cap-stage transition, it induces activation of p21
and MSX2, inducing the epithelium to proliferate and form the enamel knot. Activin A is a
homodimer of two ßA subunits (gene symbol INHBA) and is expressed in the presumptive
tooth germ mesenchyme. It regulates a central receptor, EDAR. INHBA mutants fail
to develop mandibular teeth beyond bud stage (Ferguson et al., 1998). The inhibitor
follistatin has been shown to be required in enamel knots for normal crown and cusp
formation by blocking ameloblast differentiation (X.-P. Wang, Suomalainen, Jorgez
et al., 2004).
TGFß1 also plays a crucial role in ameloblast differentiation and enamel mineralization by
upregulation of MMP20 and KLK4 which degrade the enamel matrix to be replaced by
inorganic material (Cho et al., 2013).
Fibroblast growth factor pathway
The TGFß/BMP signaling during tooth development is in a tight cross-talk with the
fibroblast growth factor (FGF) signaling pathway. FGFs are expressed in all tissues in
adult life and during embryogenesis. The canonical FGFs function auto- or paracrine to
control proliferation, differentiation, and survival. FGFs are tightly bound to heparan
sulfate proteoglycans in the ECM to limit their sphere and to increase their specificity. FGF
receptors (FGFR) are tyrosine-kinase linked with an extracellular immunoglobulin-like
domain for ligand binding. Upon ligand binding, the intracellular domains of the receptor
units cross-phosphorylate and thereby activate docking sites for downstream messengers
or effectors. Different classes of pathways can be induced intracellularly, depending on
the receptor adapter that is recruited: Ras-MAPK, Pi3K-Akt, phospholipase C or STAT
pathway (Ornitz et al., 2015).
In murine tooth development, FGF signaling plays a critical role in invagination of the
epithelium and mesenchymal condensation (T. Mammoto, A. Mammoto, Torisawa
et al., 2011). FGF8 and semaphorin 3f, expressed in the early dental epithelium (E11)
attract and repulse ectomesenchymal cells. The resulting mechanical compaction induces
expression of key odontogenic transcription factors PAX9 and MSX1, as well as BMP4
in the dental mesenchyme. Fgfr2b-knockout mice fail to progress tooth formation beyond
bud stage, demonstrating the role of FGFs in the bud-to-cap-stage (Kettunen et al.,
2007). In tooth development, FGF3 and FGF10 have shown to be functionally overlapping.
Double-mutant mice, but not single-mutant, show an arrest at bud stage. FGF3 regulates
epithelial stem cell proliferation in the cervical loop, whereas BMP4 inhibits proliferation
and induces ameloblast differentiation (X.-P. Wang, Suomalainen, Felszeghy et al.,
2007). Knock-out mice of FGF inhibitors, the Sprouty genes, exhibit supernumerary teeth
and ectopic tooth formation (Klein et al., 2008) while Fgf8 knockout is lethal in early
embryonic development. Migration and with it gastrulation is impaired (Sun et al., 1999).
In tooth development, FGF8 of the epithelium induces DLX2 of the mesenchyme (Thomas
et al., 2000).
20 Chapter 1 Introduction
Sonic hedgehog pathway
In murine tooth development, Sonic hedgehog (SHH) ligand expression is restricted to
the presumptive dental ectoderm at E11. Mutations in the human gene lead to severe
holoprosencephaly, so the role of the ligand in odontogenesis cannot be identified from
this phenotype. Conditional knockout of Shh in mice led to the disruption of tooth
morphogenesis from bud stage onwards (Dassule et al., 2000). Upon binding of SHH
to the receptor Patched (PTCH1/2), a signaling cascade with the involvement of the
co-receptor Smoothend (Smo), a repressor complex is inhibited, and the transcription
factors GLI1/2/3 are stabilized to induce target gene transcription. It is known, that SHH
is required for the maintenance of the epithelial stem cell niche and is a key morphogen in
organogenesis. The spatiotemporal expression of SHH gradients in the ventral neural tube
is fate determining for the floor plate cells and motoneurons. Furthermore, it induces the
neural tube formation, explaining the severe phenotypes of SHH mutants.
SHH is not only needed in the dental epithelial stem cell niche, but also in odontoblast
differentiation and the dental stem cell population. Interestingly, alveolar nerves secrete
SHH and the loss of innervation leads to reduced GLI1 expression, first in the mesenchyme,
later also in the epithelium. The loss of the SHH led to a decrease in the stem cell
populations (Hu Zhao et al., 2014). Gli-mutant mice have a complete lack in tooth
development which supports the early role of SHH in tooth development (Hardcastle
et al., 1998).
Wnt pathway
Wnt ligands are short range morphogens that regulate tissue development and homeostasis
by supporting the stem cell niche. Especially permanently regenerated tissues, such as
the intestine, skin, stomach, and blood, are dependent on active stem cell niches whose
proliferative activity is animated by Wnt signaling. Wnts bind to receptors of the 7-
transmembrane frizzled family (FRZ). Upon ligand-receptor interaction, the co-receptor
LRP5/6 is recruited to FRZ and a degradation complex, comprising the APC (adenomatous
polyposis coli) protein, axin, dishevelled (DVLl) and GSK3ß (glycogen synthase kinase 3),
is inhibited. Thereby, ß-catenin is stabilized, enters the nucleus and activates TCF/LEF-
mediated Wnt target gene transcription. Multiple Wnt genes are expressed throughout
dental developmental stages. WNT4, WNT6 and WNT10 are expressed in the early dental
epithelium (Sarkar et al., 1999). Mutations of the WNT10A gene in humans is associated
with tooth agenesis and hypodontia. Introduction of stabilized ß-catenin, and thereby
constitutively active Wnt pathway, resulted in massive development of supernumerary teeth
(Järvinen et al., 2006). Inactivation of ß-catenin in the epithelium results in arrest at
bud stage, whereas inactivation in the mesenchyme leads to arrest at bud-to-cap transition
(J. Chen et al., 2009). A reported Wnt-Bmp feedback loop includes the activation of
LEF1 by BMP4 and its target MSX1, respectively. Additionally BMP4 suppresses the
expression of dickkopf 2 (DKK22) and odd-skipped-related 1 (OSR1), both inhibitors of
Wnt (Jia et al., 2013). In turn, LEF1 activates FGF3 signaling, which is followed by
Shh transcription in the epithelium (Kratochwil et al., 2002), demonstrating the tight
network between the signaling pathways.
1.2 Tooth development 21
Lessons from human congenital tooth malformations
Developmental biology excellently describes the morphologic stages of odontogenesis in
humans and mice. Nevertheless, the knowledge on molecular pathways involved in human
tooth development is necessarily limited, since ethics prohibit genetic engineering in human
(development) for basic research. Yet, experiments of nature, hereditary developmental
malformations, give insight to the molecular control of human tooth development. Tooth
agenesis describes conditions were the dentition, primary, permanent or both, has failed.
The term includes hypodontia (maximum of six teeth), oligodontia (some missing teeth)
and anodontia (no teeth). Mostly, tooth agenesis is non-syndromic. However, cleft lip
and cleft palate are often associated with missing teeth. Genetic defects in non-syndromic
tooth agenesis are described to be affected by MSX1, PAX9 and AXIN2. In mice, all
these genes have been implicated in early stages, particularly bud-to-cap stage. Contrarily
to what is expected from mouse (where only incisor region was impaired), MSX1, as
well as PAX9, mutations affect premolars and molars in humans. AXIN2 phenotypes
exclusively affect permanent dentition. Canines are rarely affected universally. It is of
particular importance that the common model for organogenesis, the mouse only has one
generation and only two classes of teeth (incisors and molars). Syndromic implications of
MSX1 result in Witkop syndrome where also nail development is defected. In ectodermal
dysplasia, where all skin appendages are affected, genetic defects in EDA (Ectodysplasin A)
pathway, which includes the ligand EDA, the receptor EDAR, and downstream molecules
such as EDARADD, IKKgamma, NEMO or p63 are reported. Defects in PITX2 cause
absence of maxillary incisors. Other significant roles in tooth agenesis play TGFA, IRF6,
FGFR1, MMP1 and MMP20. All the described hereditary developmental malformations
are detailed in the online compendium OMIM (Online Mendelian Inheritance in Man,
https://www.omim.org/) and are summarized in T. Mitsiadis et al., 2011.
1.2.10 Mesenchymal Condensation
A focus of this thesis has been laid on the process of mesenchymal condensation. This
process is conserved concept in embryonic formation of organs, such as skin appendages
(teeth, salivary gland, hair, mammary gland, nail, feather, eye), the skeleton (bone, cartil-
age, tendons), kidney and lung (Paus, 2003). At the site of presumptive organ formation,
scattered connective tissue mesenchymal cells condense or aggregate to a compact cell mass.
This compaction is mechanotransduced via an extracellular and intramembrane sensor
network into the cell signalosome. At the end of an intracellular signaling cascade, the
stimulated cell then changes its enzymatic activity and the cytoskeletal arrangement and,
more importantly, tissue-specific transcription factor expression is induced, committing
the cells’ fate. Mediated by the expression of the transcription factors, tissue-specific
morphogens are produced that drive organ formation. These morphogens regulate organ
size, comparted tissue and cell differentiation, and all required spatiotemporal activities.
The concept of mesenchymal condensation has been extensively studied in skeletal develop-
ment (Hall et al., 2000). Different hypotheses have been proposed, and various molecules
are described for their role in mesenchymal condensation, but still the precise molecular
22 Chapter 1 Introduction
mechanism remains unknown. Thus, it is accepted that the player may differ among the
different tissue types but the mechanism is transferable.
It is now common sense, that mechanobiology is the main driver of organ formation (T.
Mammoto, A. Mammoto and Ingber, 2013). Through different physical properties (e.g.,
stiffness), of two interacting and via a basal lamina connected tissue types folding, stretch-
ing, compaction or other mechanical cues are exerted on the cells. Cells sense physical
force changes and transduce these signals intracellularly to produce a response, which is
either a change in enzymatic activity and/or transcription factor expression. Of particular
interest are the ECM molecules (collagens, fibronectin, proteoglycans, hyaluronan) that
are connected to cell surface molecules such as N-CAM, integrin, and syndecan. Also,
direct cell-cell contact via cadherins plays a pivotal role. Inside the cell, conformation of
focal adhesion proteins (FAK) is changed, small GTPases are activated and kinases such
as MAPK, ERK and JNK are regulated.
In tooth development, the dental epithelium expresses two chemokines to induce mesen-
chymal cell migration. FGF8 has a wide action range and attracts underlying ectomesen-
chymal cells while the second chemokine Semaphorin 3f (SEMA3f) is expressed at the exact
same site, repulses mesenchymal cells. Thereby, the cells are stacked and squeezed, which
has an enormous impact on their cytoskeleton and on the extracellular environmental cues.
A cascade of kinases is regulated and, finally, RhoA kinase is downregulated, resulting
in rearrangement of the cytoskeletal actin filaments and gene transcription of PAX9 and
MSX1 is induced (T. Mammoto, A. Mammoto, Torisawa et al., 2011) (see figure 1.19).
Recent findings suggest that the actin-p38MAPK-SP1 pathway is activated as a response
to the mechanical stress, which in turn regulates the expression of interconnecting collagen
type VI. Collagen VI in turn physically stabilizes the compressed mesenchymal cell shapes
to ensure efficient cell fate switching (T. Mammoto, A. Mammoto, Jiang et al., 2015).
1.3 State of the Art - Tissue Engineering 23
Figure 1.19: Mechanochemical control of mesenchymal condensation during embryonic tooth
formation. FGF8 and SEMA3F expressed from dental epithelium, induce attraction and
repulsion of mesenchymal cells, which undergo compaction. The induced actin remodeling
upregulates organ-fate specific transcription factors, and the extracellular matrix is rebuilt and
stabilized (T. Mammoto, A. Mammoto and Ingber, 2013).
Taken together, cellular mechanobiology is a prime example of biologic beauty and
complexity. The molecular mechanism remains to be unraveled. One reason for that was
the lack of a functional model systems in vitro. In recent times reasonable efforts have
been made to create those models (T. Mammoto, A. Mammoto, Torisawa et al., 2011;
S. Zhang et al., 2018).
1.3 State of the Art - Tissue Engineering
As extensively illustrated above, tooth organogenesis comprises complex temporal-spatial
organization of tissue interaction. The mature organ has well defined, intricates structures
with unique properties of the mineralized and non-mineralized structures. The traditional
approach to restoring missing or lost teeth are dental implants. However, this requires a
significant surgical procedure and, eventually, laborious follow-up care. Synthetic implants
do not meet all demands on a tooth, like proper integration into the jaw bone by the
natural periodontal ligament. New approaches in the field of tissue engineering attempt
to regenerate dental tissues (e.g., pulp, enamel or dentin) or entire teeth (Volponi et al.,
2018). For whole-tooth-bioengineering, two main hurdles exist: cell sources capable of
tooth-formation, and the appropriate "setting", including scaffolds, signaling molecules,
mechanical cues and time.
One of the first approaches to regenerate teeth was performed in 1966 when embryonic
mouse tooth germs (epithelium and mesenchyme) were explanted, dedifferentiated in vitro,
24 Chapter 1 Introduction
and then isologously transplanted in vivo. The cultures "developed into incisor teeth that
were almost perfect in shape and structure." (Main, 1966). In a variety of following studies,
embryonic and postnatal dental cells were presented to be the ideal cell source to regenerate
teeth for research purpose (Oshima et al., 2011; E. E. Smith et al., 2017; Young et al.,
2005). In all attempts, epithelial and mesenchymal dental cells were recombined, sometimes
from different developmental stages. Only early dental epithelium (before E11) has been
proven to be inductive, whereas this odontogenic potential shifts to mesenchyme after E12
(Mina et al., 1987). Additionally, embryonic stem cells (ESCs) have been proven to be
capable of differentiating to odontogenic epithelia and odontogenic mesenchyme (Kawai
et al., 2014; Ning et al., 2010). The approach to use embryonic or postnatal (dental) cells
is successful in terms of tooth formation, but when it comes to patient therapy, it is not
applicable.
Therefore, researchers seek for tooth inducing stem cells from adult dental or nondental
tissue. Adult mesenchymal stromal cells are characterized by a panel of surface markers
(i.e., CD44, CD73, CD90, CD105, CD106, CD13) and their plastic adherence in cell culture.
They can be obtained from various tissue sources such as the bone marrow, adipose tissue,
umbilical cord, and the tooth. In extracted adult teeth, stem cells reside in the dental pulp
(DPSC), in deciduous teeth named human exfoliated deciduous cells (SHED), in the dental
follicle (DFC), in the apical papilla (SCAP) and in the periodontal ligament (PDLSC).
A benefit of DPSCs is that they derive from discarded teeth, mainly from third molar
extractions. This cell type has been extensively studied (Gronthos et al., 2000; Sloan
et al., 2009). Their odontogenic capacity has been shown in vitro and in vivo (Hu et al.,
2018). Interestingly, even isolated bone marrow MSCs have been capable to contribute to
tooth regeneration in vivo, when cultured under appropriate conditions (Ohazama et al.,
2004). So far, the presence of an epithelial reciprocal interaction partner is indispensable
to induce tooth formation from adult stem cells in vitro and in vivo, which is a drawback
since inductive epithelium can only be found in embryonic dental epithelium.
One idea to circumvent embryonic epithelium is the usage of induced pluripotent stem
cells (iPSCs). So far, this was not achieved, yet Arakaki et al., 2012 established a system
to create ameloblasts from murine iPSCs by usage of an enamel-secreting feeder layer of
ameloblasts from mouse embryos.
Either epithelial or mesenchymal cell populations are induced to be odontogenic (capable
of initiation of de novo odontogenesis) and were subsequently recombined with the respons-
ive cell population counterpart. An early-stage tooth primordium can be generated from
an epithelial-mesenchymal-cell recombination, which can be either directly transplanted at
the location of the missing tooth or cultured ex vivo to form a whole tooth to replace the
missing tooth.
In addition to the used cell type, an appropriate environment (signaling molecules,
scaffolds, culture time) is crucial for whole tooth engineering. Growth factors that found
application in regenerative dentistry and whole-tooth bioengineering included BMPs,
TGFß1 and TGFß3, insulin-like growth factor (IGF) and platelet-derived growth factor
(PDGF) (Hosseini et al., 2018). Natural and synthetic materials have been used as
replacements for missing teeth. Decellularized tooth buds or organic polymers (gelatine
1.3 State of the Art - Tissue Engineering 25
gels, poly-lactic acid, polyglycolic acid, PLGA, hydroxyapatite and more) were fabricated
with a biometic porosity, stiffness-elasticity, texture and shape to support cell adhesion
and differentiation. These scaffolds were then seeded with cells from different sources and
were implanted in vivo (Oshima et al., 2011; W. Zhang et al., 2017). Occasionally, these
construct produced organized, bioengineered teeth. Outstanding research was published in
2017, where the group of Takashi Tsuji showed whole-tooth restoration by bioengineered
tooth germ in a large-animal model, the beagle dog. Permanent molar tooth germs were
extracted from postnatal individuals, epithelium and mesenchyme were dissected and then
reconstituted in a collagen gel. After two days of organ culture, the organoids were then
transplanted into the mandible and 180 days later, erupted, and functional bioengineered
teeth were grown (Ono et al., 2017).
Figure 1.20: Vision of autologous tooth regeneration. Suitable adult cells (epithelial and
mesenchymal) are collected from dental patients and expanded in vitro. By an induction step
(eventually chemical or mechanical cues), these cells are converted into odontogenic cells, that
will be recombined and either transplanted as tooth germ or cultured to form a whole tooth ex
vivo (Volponi et al., 2018).
Taken together, human odontogenesis needs to be understood in more detail, and an
appropriate human organogenesis models are required to elucidate molecular networks and
events. Furthermore, different cell sources and configurations have to be evaluated further
for dental engineering, to achieve the ultimate goal of autologous tooth bioengineering in
human patients (see figure 1.20).
26 Chapter 1 Introduction
1.4 Aims of this work
In the course of one’s life, everyone is highly likely to be confronted with tooth loss,
depending on individual health status and life expectancy. The gold standard of tooth
replacement techniques are dental implants from composite materials, such as ceramics
with titan screws. However, synthetic materials neither converge nor communicate with the
body at the site of implantation, therefore, an appropriate response to changing demands
cannot be assured.
Numerous studies for regenerative therapies, i.e., partial tissue replacement and whole-tooth
bioengineering, have been published, but to date, no regenerative therapy approach was
successful in humans. Two key elements need to be improved. Firstly, the knowledge on
molecular events of human organogenesis with a special focus on the initiation step for tooth
development is very limited. Most of the available data was conducted in murine models,
but for at least two aspects the mouse dentition is not comparable with human. Mice only
have on set of teeth, and they do not develop a second set as humans do during their youth.
Furthermore, mice only have two types of teeth: molars and incisors. Therefore, much more
effort has to be made to obtain reliable human data on organogenesis by providing suitable
human in vitro models. The second key element of human whole-tooth bioengineering is
the cell source. It is just not possible to use embryonic or postnatal dental cells in the
clinics. For the moment also induced pluripotent stem cells do not represent an alternative
due to potential risk of genetic aberrations and teratoma formation. The ideal disposable
cell source for transplantation would be adult autologous cells.
One aim of this work is to show that adult human dental pulp stem cells can be programmed
to inductive dental mesenchymal tooth germs in vitro when cultured under appropriate
conditions. Thereby, cellular self-organization to induce the cascade of molecular event
sequences that lead to differentiation is allowed and propagated. Their nature of inductivity
shall be demonstrated by expression of dental marker genes as well as by interaction with
a receptive interaction or partner cell, which are, in the dental context, epithelial cells.
The steps to achieve an inductive tooth germ have to be evaluated in detail, focussing on
the process of mesenchymal condensation. Mesenchymal condensation is considered to be
the key to organ development initiation. It has to be validated that the presented in vitro
model of condensation of dental pulp stem cells recapitulates the steps of mesenchymal
condensation by comparison to published in vivo data in human or mouse. Finally, by
usage of this condensation model, new putative targets of organogenesis research shall be
defined.
CHAPTER 2
Materials & Methods
Table 2.1: Materials for cell culture applications
Product Distributor
PBS Applichem
DMEM (High Glucose + stable glutamin) Corning
FCS Corning
Trypsin EDTA Corning
Penicillin/Streptomycin Corning
Collagenase NB4 Serva
Dispase II Sigma Aldrich
DermaLife Basal medium Lifeline Cell Technology
DermaLife K Life Factors Lifeline Cell Technology
Collagen A Biochrom
Neurobasal Plus Life Technogies
B27 Plus Life Technogies
gelatin Sigma Aldrich
L-ascorbic acid phosphate Sigma Aldrich
Isobutylmethylxanthine Applichem
Hydrocortisone Sigma Aldrich
Indomethacin Sigma Aldrich
Insulin Sigma Aldrich
L-Glutamin Sigma Aldrich
rh Epidermal growth factor Promokine
rh basic Fibroblast growth factor Promokine
Dexamethasone Applichem
ß-glycerophosphate Sigma Aldrich
DMSO Sigma Aldrich
NaCl Chemsolute Th.Geyer
Polyethylenimine 9002-98-6 Polysciences
Polybrene VWR
LY294002 Cell Signaling
LiCl VWR
rh Transforming Growth Factor ß 1 Promokine
27
28 Chapter 2 Materials & Methods
Table 2.1: Materials for cell culture applications
Cytochalasin D VWR
Pertussis Toxin Sigma Aldrich
CellTracker™ Red CMTPX Dye Thermo Fisher
standard tissue culture flasks Greiner Bio-One
tissue culture dishes Sarstedt
Ultra-low attachment 96 well plate Corning
Table 2.2: Materials for molecular biology applications
Product Distributor
Formaldehyde Roth
Ethanol VWR
Acetone VWR
Oil Red O Alfa Aesar
Alizarin Red Roth
Hematoxylin VWR
Eosin Roth
Fast Digest Restriction Enzymes Fermentas
T4 Ligase Fermentas
NucleoSpin-Gel and PCR Clean up Macherey-Nagel
NucleoBond-Xtra Midi Plus Macherey-Nagel
Agarose Serva
Tris-HCl Roth
Imidazole Roche
DNase Macherey-Nagel
PMSF Sigma
Lysozyme Roth
Cell Trace CFSE cell Proliferation Kit Invitrogen
7-Aminoactinomycin R&D Systems
DAPI Sigma
Tissue-Tek O.C.T. Compound Sakura
APODirect TUNEL Assay Kit Phoenix Flow Systems
Imsol Mount VWR
TaqMan® Reverse Transcription Kit ThermoFisher
SensiFast SYBR No-ROX Kit Bioline
TruSeq® Stranded mRNA LT Kit Illumina
Mag-Bind® RXNPure Plus beads Omega Bio-Tec
SnakeSkin ThermoFisher
Ni-NTA agarose qiagen
2.1 Cell isolation and cell culture 29
2.1 Cell isolation and cell culture
2.1.1 Isolation of Human Dental Pulp Cells from adult third molars
Extracted third molars were collected, with patient’s informed consent and approval by
Ethics Committee Charité, from 18 to 30-year old adults from Charité Department of
Cranio Maxillofacial Surgery (Berlin, Germany) and from dental practice Amler, Mönch,
Knebel (Berlin, Germany) as well as from specialist for oral and maxillofacial surgery
Gülseren Köksal (Berlin, Germany). Teeth were kept in DMEM high glucose containing
10% FCS and 1 % antibiotic-antimycotic solution and stored at 4 °C for not longer than
24 hours. Human dental pulp cells are isolated under sterile conditions according to a
modified protocol (Stan Gronthos et al., 2011). Briefly, extracted specimens were wiped
with ethanol to reduce contamination risk. To open the pulpal cavity, teeth were cracked
mechanically. Pulp tissue was removed with forceps and placed into a phosphate buffered
saline (PBS) containing petri dish. The tissue was minced and washed twice with PBS to
remove debris and blood. Enzymatic tissue digestion was performed with a collagenase (3
mg/ml)/dispase II (4 mg/ml) enzyme mix in PBS for 1 h at 37 °C. After filtering through
a 70 µm cell strainer and washing with PBS by centrifugation (400 x g for 5 min), the cell
pellet is resuspended in DMEM with 10% FCS and antibiotics.
2.1.2 Isolation of gingival keratinocytes
Residual gingival tissue from molar extractions (see above) was enzymatically digested
with dispase II (4 mg/ml) for 30 min at 37 °C in PBS. After filtering through a 70 µm
cell strainer and washing with PBS by centrifugation (300 x g for 5 min), the cell pellet
was resuspended in complete DermaLife Basal medium supplemented with DermaLife K
LifeFactors. Cells were grown on collagen A-coated (0.1mg/mL Collagen A) tissue culture
vessels under standard culture conditions (see below).
2.1.3 Standard cell culture
Isolated dental pulp cells were cultivated and expanded in a humidified atmosphere at 37 °
C, 5 % vol/vol CO2 in standard tissue culture flasks in DMEM high glucose containing
10% FCS and 1 % antibiotic-antimycotic solution.
2.1.4 Condensation process for formation of artificial tooth primordium
For culture under non-adherent conditions, human dental pulp cells of between passage
two and eight were passaged two days before use. They were harvested and resuspended in
DMEM with 10% FCS to yield up in a suspension of 106 cells per ml. The cell suspension
(0.2 ml per well) was given to 96 well low attachment plate (Ultra Low Cluster Plate,
Corning, Germany). In contrast to the negatively charged, hydrophilic surface of standard
tissue culture dishes, the ultra-low attachment plates possess a neutral, hydrophilic hydrogel
coated surface that greatly minimizes the binding of attachment proteins. Condensation
process starts shortly after seeding and is observed macroscopically by cells forming
aggregates. To ensure constant culture conditions, the medium was changed regularly
every three days. Cultures were kept under standard cell culture conditions.
30 Chapter 2 Materials & Methods
2.1.5 Co-Culture assays
For co-culture of human dental pulp-derived cells and cells of epithelial origin (gingiva or
skin-derived) the condensates, produced by the method described above, are transferred
between day 1 to 2 in a composite medium appropriate for both cell types (DMEM high
glucose (with FCS) and Dermal Life; 1:1). A single cell suspension of epithelial cells, in
a ratio of 1:4 related to the initial cell number used for mesenchymal condensation, was
added and the resulting mixture was cultured under non-adherent conditions. To ensure
constant culture conditions, the medium was changed regularly every 1 to 3 days.
2.1.6 In vitro differentiation assays
Adipogenic differentiation
Cultured DPCs were grown until confluency before the differentiation was induced by
exchanging the media to adipogenic media (standard DMEM supplemented with 10 % FCS
and 1 % antibiotic-antimycotics, 0.1 mM L-ascorbic acid phosphate, 0.5 mM isobutylmethyl-
xanthine, 0.5 µM hydrocortisone, 60 µM indomethacin and 10 µg/ml insulin). Media was
changed three times per week for three weeks. To confirm adipogenic differentiation, cell
cultures were stained for lipid vacuoles by Oil Red O, and gene expression of lipid marker
genes was assessed by qRT-PCR. Before staining, cells were fixed in 10 % formaldehyde
for 30 minutes at room temperature and washed with PBS. Lipid droplets were stained
with freshly filtered Oil Red O working solution (0.7 % in propylene glycol). Excessive Oil
Red O was carefully rinsed off with tap water. The extent of intracellular lipid vesicles was
documented microscopically.
Neurogenic differentiation
Neurogenic differentiation was performed according to the protocol of Miura et al., 2003.
Briefly, culture plates were coated with 0.1 % gelatin for 30 minutes and allowed to dry
for at least two hours. Cultured DPCs were then seeded at a low density of 10 000 cells
per cm2 in Neurobasal Plus Medium (Life Technologies) supplemented with 2 % B27
Plus (Life Technologies), 250 µM L-Glutamine, 20 ng/ml EGF, 40 ng/ml FGF2 and 1 %
antibiotic-antimycotics. The medium was changed every three days. Cells were cultured
for 14 days and were then analyzed via qRT-PCR and immunocytological staining (see
Chapter 2.5.3).
Odontogenic/Osteogenic differentiation
DPCs were grown until confluency before osteogenic/odontogenic differentiation was
induced. For this purpose media was exchanged to "mineralisation" media containing
10 % FCS and 1 % antibiotic-antimycotics, 0.1 mM L-ascorbic acid phosphate, 100 nM
dexamethasone, 0.5 µM hydrocortisone, and 5 mM ß-glycerophosphate. Media was changed
regularly two to three times per week for three weeks. Gene expression of osteogenic
marker genes was assessed by qRT-PCR. For detection of mineralization, the cell cultures
were stained with Alizarin Red. Therefore, the fragile cell layer was rinsed gently with
PBS without Calcium or Magnesium, fixed in ice-cold 70 % ethanol for one hour and
washed twice with distilled water. Afterwards filtered Alizarin Red (2 g in 100 ml dH2O,
pH 4.1 to 4.3 (HCl)) was incubated for 45 Minutes at room temperature. After four
2.2 Flow Cytometry 31
washing steps with distilled water, the red staining of the mineralization was documented
photographically.
2.1.7 Freezing and thawing of cells
Prior to freezing, 106 cells were centrifuged (400 x g for 5 min) and the cell pellet was gently
resuspended in ice-cold FCS containing 10 % DMSO. With a cooling rate of 1 °C/min
cooling rate (in Mr. Frosty container, Nalgene), cells were frozen down to -80 °C. After
one day they were transferred to liquid nitrogen.
For thawing, cell vials were removed from the liquid nitrogen, transported on ice to a 37
°C water bath, where they were thawed for 30 seconds under gentle agitation. Pre-warmed
culture medium was then added dropwise to the cells that were subsequently transferred to
a culture vessel with pre-warmed medium. After approximately 12 hours of regeneration,
the medium was exchanged for fresh culture medium.
2.2 Flow Cytometry
Flow cytometry allows quantification and characterization of the cellular morphology,
vitality and protein expression. Hydrodynamic focusing inside the flow cell of the cytometer
ensures single-cell analysis via laser excitation. Specific photomultipliers detect the emitted
cell signals.
Human dental pulp cells were analyzed by flow cytometry for the surface marker expression
by immunofluorescent staining, for viability, proliferation and eGFP expression. All flow
cytometric analyses were conducted on MACSQuant analyzer (Miltenyi Biotec, Germany).
2.2.1 Surface marker staining
After cell harvest by trypsinization, the cells were resuspended in PBS/BSA (0.5 %).
Incubation of the respective antibodies was conducted on ice for 30 minutes. For each isotype
and fluorescent label, an isotype control was included. The following antibodies were used:
anti-human-CD105-APC (Biolegend), anti-human-CD106-APC (Biolegend), anti-human-
CD90-FITC (Miltenyi Biotec), anti-human-CD146-FITC (Miltenyi Biotec), anti-human-
CD13-APC-Cy7 (Biolegend), anti-human-CD44-PacificBlue (Miltenyi Biotec), anti-human-
CD45-VioBlue (Miltenyi Biotec), anti-human-CD14-FITC (eBioscience), anti-human-CD34-
APC (Miltenyi Biotec) and anti-human-CD31-Alexa405 (DRFZ) After incubation, the cells
were washed to remove unbound antibodies and resuspended in PBS/BSA prior to analysis
on flow cytometer.
2.2.2 CFSE assay
To track proliferation time, a CFSE assay was conducted using the CellTrace™ CFSE
Cell Proliferation Kit. Carboxyfluorescein succinimidyl ester (CFSE) is a cell permeable
dye that is converted inside the cell by internal esterases, whereby it is covalently bound
to intracellular proteins and acquires a fluorescence with an excitation maximum of 492
nm and an absorption maximum of 517 nm. With each cell division, the proteins, and
subsequently the bound CFSE molecules are equally distributed among the daughter cells,
resulting in a halved fluorescence intensity compared to the parent cell.
Dental pulp cells from passage 3 were trypsinized and labeled with CFSE according to
the manufacturer’s protocol. 200 000 cells were directly fixed with 10 % formalin and stored
32 Chapter 2 Materials & Methods
at 4 °C for later analysis. The remaining positively labeled cells were seeded into six-well
dishes. After 24, 48 and 120 hours one well each was trypsinized and fixed in formalin.
Afterwards, the cells were analyzed by flow cytometry to assess the CFSE fluorescence
intensity. Control of fixed, non-labeled cells was included.
2.2.3 Viability assay
To discriminate viable from non-viable cells, the molecule 7-aminoactinomycin D (7-AAD)
was used. 7-AAD is a DNA intercalating dye that emits at 635 to 675 nm wavelength upon
DNA binding. Intact cell membranes of viable cell exclude 7-AAD, whereas compromised
membranes of dying/dead cells allow the dye to pass. Cells were analyzed by flow cytometry
after 10 minutes of incubation with 7-AAD in the channel B3 of the MACSquant.
2.3 Cell Tracker Red labeling
Keratinocytes were labeled fluorescently to be visualized by fluorescence microscopy.
Therefore, a labeling dye, CellTracker™ Red CMTPX, was used. Briefly, keratinocytes
were trypsinized and resuspended in pre-warmed working solution, containing serum-free
DermaLife medium and 5 µM of the dye (solubilized in DMSO). Cells were incubated at 37
°C for 45 minutes under gentle agitation. After washing and resuspension in fresh medium,
the cells convert the permeable dye to a membrane impermanent label via glutathione-S-
transferase. CellTracker Red is excitable at 577 nm and emits at 602 nm, which corresponds
to the red channel in the fluorescence microscope.
2.4 Collagen tagging via CNA35-eGFP
For real-time live imaging of collagen production, a recombinant fusion protein CNA35-
eGFP was used. The CNA represents a 35 kDa collagen-binding protein domain of bacterial
adhesins. An overexpression vector pET28a-EGFP-CNA35 was purchased from addgene (#
61603) for recombinant prokaryotic expression of the CNA35-eGFP fusion protein. Briefly,
a starter culture of transformed E.coli BL21 DE3 was prepared with LB media containing
kanamycin. After 5 hours, the starter culture was transferred to the main culture, that
was incubated with agitation at 37 °C until OD 0.55. Protein expression was induced by
addition of isopropyl ß-D-1-thiogalactopyranoside (IPTG) at a concentration of 0.8 mM.
The culture was induced overnight at room temperature. The culture was centrifuged at
3500 x g for 20 minutes and the pellet was resuspended in lysis buffer (50 mM Tris-HCl,
300 mM NaCl, 20 mM Imidazole). Lysozyme (0.2 mg/ml) to ensure bacterial cell wall lysis,
DNase with 10 mM MgCl2 to break released DNA polymers that bring up a high viscosity,
and 1 mM phenylmethylsulfonyl fluoride (PMSF), a serine protease inhibitor that inhibits
proteases that would destroy the recombinant protein, were added. After 10 minutes of
incubation, the suspension-containing tubes were transferred to liquid nitrogen to enhance
the lysis. The tubes were thawed and shock-frosted for two more cycles. Disrupted cells
were centrifuged at 10000 x g for 30 minutes. The 6x-histidine-tagged CNA35-eGFP in
the supernatant was now purified by affinity chromatography with Ni-NTA Agarose. The
histidine residues of the His tag bind to immobilized nickel ions in the agarose matrix with a
high affinity and specificity. After equilibration of the matrix with lysis buffer, the bacterial
supernatant was incubated with the Ni-NTA agarose. Upon binding of the protein, the
2.5 Histology 33
matrix color changes from blue to white. Afterwards the loaded matrix was transferred
to chromatography columns. After intense washing with lysis buffer, the CNA-eGFP was
eluted from the matrix with increasing concentrations of imidazole in the elution buffer (70
mM, 150 mM, and 250 mM). Imidazole acts as a competitor for the nickel ion binding. In
between the elution steps, the column was washed. For each used imidazole concentration
1 ml of the eluted protein in buffer was collected. Since high imidazole concentrations
are harmful to living cells, the eluted protein was dialyzed against PBS for 5 hours in
SnakeSkin dialysis membrane.
The CNA-eGFP was directly added to the DPC cultures at a concentration of 0.4 µg/µl
during to condensation. The fluorescence was imaged by fluorescence microscopy in the
green channel.
2.5 Histology
2.5.1 Sample preparation
Samples for histological analyses were embedded in cryomolds in O.C.T. Compound (Tissue-
TEK) and frozen at -80 °C. Prior to sectioning by the cryotome (Leica), the frozen blocks
were equilibrated at -20 ° C for 30 minutes. At a cutting temperature of -18 °C 8 µm
sections were transferred on glass slides (Histobond, Marienfeld, Germany). Slides were
stored at -20 °C after two hours of drying at room temperature.
2.5.2 Hematoxylin/Eosin staining
Slides were thawed at room temperature and fixed in 10 % formalin for 10 minutes at
room temperature. After washing twice with distilled water, they were transferred to
Hematoxylin solution and incubated for 5 minutes. Slides were then washed in running
tap water for 10 minutes and cleared in distilled water. The slides were transferred into
Eosin solution, incubated for 2 minutes and after washing with distilled water, they were
dehydrated in a graded ethanol series with terminal xylene treatment. Prior to microscopic
analysis, they were mounted in a resinous medium.
2.5.3 Immunohistochemistry
After thawing, the slides were fixed in acetone for 20 minutes at -20 °C. Slides were washed
three times in PBS. Unspecific binding was blocked with 10 % serum from host animal of
the secondary antibody for 20 minutes. Afterwards, primary antibody dilution was added
to the sections and was incubated overnight at 4 °C. After washing with PBS, secondary
antibody dilution was pipetted on the sections and was incubated for 45 minutes at room
temperature in the dark. During the last 10 minutes of this incubation, DAPI (1:500) was
added to the solution. After the last washing step, the sections were covered with Imsol
Mount and a coverslip and were analyzed by fluorescence microscopy. To exclude unspecific
background from the secondary antibody, a no primary antibody control was included.
34 Chapter 2 Materials & Methods
Table 2.3: Antibodies for immunhistochemistry
Antibody Distributor Dilution
mouse anti-human Collagen type I Sigma C2456 1:100
mouse anti-human Collagen type IV Sigma C1926 1:100
mouse anti-human Cytokeratin 14 Abcam ab7800 1:100
mouse anti-human Cytokeratin8/18 Invitrogen MA5-14088 1:100
mouse anti-human Ki67 eBioscience 14-5699-82 1:100
rabbit anti-human Cytokeratin 15 Abcam ab52816 1:100
rabbit anti-human Cytokeratin 19 DB-biotech DB103-T 1:100
rabbit anti-human Tenascin C Santa Cruz sc-20932 1:100
rabbit anti-human Vimentin Invitrogen PA5-27231 1:100
rabbit anti-human Fibronectin 1 Abcam ab2413 1:100
goat anti-mouse CF488 Biotium 20010 1:200
goat anti-rabbit CF594 Biotium 20112 1:200
goat anti-mouse CF594 Biotium 20119 1:200
2.6 Imaging
Keyence BZ-9000 Microscope was used for all fluorescence imaging with the BZ-II-Viewer
software and images were analyzed via BZ-II-Analyzer software. Cell cultures were imaged
either with Keyence microscope, Carl Zeiss AxioVert A1 or Carl Zeiss AxioVert 40C. Phase
contrast and fluorescent images were adjusted for tone levels, brightness and contrast.
For long-term video, a stage top incubator system (TOKAI-Hit) was installed on the
microscope, that allows standard culture conditions with a humidified atmosphere at 37
°C, 5 % vol/vol CO2.
2.7 Reporter assay strategies
2.7.1 Cloning
Backbone for all constructed reporter vectors was the lentiviral pLL3.7. The sequences of
the response element including a minimal promoter was amplified by PCR from the donor
vectors by usage of specific up- and downstream primers. The donor vectors which had the
MAPK response element (RE), the TGFß response element and the FoxO response element
were based on pGL4.26 (Promega). The primer sequences for fragment amplification for
these three elements were: pGl4.26up 5’ AGTCATGCATAATAGGCTGTCCCCAGTGC
and pGL4.26do 5’CTGAACCGGTTTGTCATGGCCTAACGGTTC. The underlined se-
quences mark the restriction site for NsiI and AgeI for cloning into the target vector pLL3.7.
After enyzmatic digestion of the pLL3.7 and the amplified PCR fragments with AgeI and
NsiI (Fast Digest, Thermo Scientific), they were ligated with T4 ligase (Fermentas) at 14
°C overnight in front of the eGFP CDS.
The control vector, which had only the minimal promoter in front of the eGFP coding
sequence (CDS) was constructed by insertion of an oligonucleotide (minPup 5’ TTAGAGG-
GTATATAATGGAAGCTCGACTTCCAGCTTGGCAATCCGGTACTGTTA, minPdo
5’ CCGGTATCTCCATATATTACCTTCGAGCTGAAGGTCGAACCGTTAGGCCAT-
2.7 Reporter assay strategies 35
GACAAATGCA). For hybridization, 60 pmol of each oligonucleotide were mixed in 60 µl
H2O and placed at 95 °C for 5 minutes. Optimized hybridization conditions were provided
by a slow continuous incubation temperature decrease (0,5 °C/min) to 65 °C followed by
accelerated cooling rate of 2 °C/min to 40 °C. The obtained fragment was then inserted
into the pLL3.7 as described above by restriction with AgeI and NsiI.
The Wnt-RE fragment was amplified from the M50 Super 8x Topflash vector with the
following primer pair: Wnt-RE-up 5’ AGGTCTCGAGCAAAATAGGCTGTCCCCAGT
and WNT-RE do 5’ GCACATGCATCGCGCCCCTTTGATCT. The amplified fragment
was inserted into the minimal-promoter-pLL3.7 using the restriction enzymes NsiI and
XhoI (Fast Digest, Fermentas) and subsequent ligation.
Ligated vector constructs were transformed into E.coli TOP10 strain by heat shock. Colon-
ies were picked, and cells were amplified for plasmid isolation (Macherey-Nagel). After
restriction analysis with the above-mentioned restriction enzymes, positive clones were
sequenced. It was confirmed by sequencing (GATC Biotech) with the sequencing primer
pLLseq 5’ AAGCTCGCTTCACGAGATTC that the respective response elements with
the minimal promoter (Wnt-, MAPK-, TGFß- or FoxO-RE) were then located in front
of the eGFP-CDS on the lentiviral plasmid pLL3. All vector maps can be found in the
Appendix. To obtain sufficient amount of vector DNA, a maxi preparation of the plasmid
were accomplished by Nucleobond Xtra maxi plus kit (Macherey-Nagel).
2.7.2 Transfection
Virus was produced by transient transfection of HEK293T packaging cells. Therefore,
4.5x106 cells were seeded on a 10 cm culture dish one day prior to transfection in 10 ml
DMEM with 10 % FCS. To assure biosafety, a three plasmid system with incapacity to
viral replication was used. Viral gag-pol region, encoding reverse transcriptase, protease,
integrase, and capsid are located on the used pAX2 plasmid, viral env region for envelope
protein on the pVSVG plasmid, and the eGFP-marker under response-element control
was on the expression plasmid pLL3.7. Cell transfection with the three plasmids was
carried out by polyethyleneimine (PEI) transfection. DNA was premixed with the following
components:
Transfection mix:
10 µg shuttle plasmid (pLL3.7)
6.6 µg psPAX2
3.3 µg pVSVG
1mL NaCl
A second mix from 20 µl PEI and 1 ml NaCl wass prepared, and both mixtures were
incubated for 5 minutes at room temperature. Subsequently, they were mixed and dropwise
added to the HEK293T cell layer. After 12 hours of incubation, the medium was exchanged.
Virus harvest began additional 12 hours later by collecting the cell culture medium and
subsequent sterile filtration to prevent HEK293T-transfer to the primary cell cultures.
36 Chapter 2 Materials & Methods
2.7.3 Transduction
The dental pulp cells were transduced in suspension. Therefore, they were harvested by
trypsinization and two million cells were resuspended in 10 ml of the viral supernatant, 1
ml FCS, additional 5 ml standard culture medium and 1.5 µl polybrene, and were seeded
in a culture dish. 24 hours later, the medium was replaced for standard culture medium.
Transduction efficiency was measured by flow cytometry 72 hours later.
2.7.4 Reporter functionality test
The functionality of the reporter vectors was tested in transduced DPCs. The cells were
seeded in 6-well-culture dishes and after confluency was reached, they were treated with
molecules that activate the eGFP expression by activation of the respective response
elements.
The TGFß1-response element was activated by stimulation of the TGFß-SMAD2/3 pathway
using the TGFß1 ligand.
The response element for the MAPK/ERK pathway is receptive for high serum concentra-
tions and is activated by the transcription factors Elk1/SRF (serum-response factor). The
respective signaling cascade was induced here by usage of the epidermal growth factor EGF.
The FoxO response element is a binding site of the FoxO transcription factor. Phosphoinositol-
3-kinase (PI3K) is constantly phosphorylating FoxO to mark it for proteasomal degradation.
By inhibition of PI3K with the small molecule LY294002, FoxO is stabilized and translo-
cates to the nucleus to activate the FoxO gene transcription as well as the reporter.
The Wnt response element consists of eight repeats of the TCF/Lef binding sites. Those
two transcription factors are activated through ß-catenin. Without stimulation of the Wnt
pathway, ß-catenin is constantly degraded by the proteasome. The stimulation of cells with
Lithium chloride (LiCl), inhibits the degrading complex of ß-catenin, thereby stabilizing it,
and allows ß-catenin activated gene transcription via TCF/Lef1.
The concentrations of the stimuli can be found in table 3.3.1.
2.8 Treatment with inhibitory agents
To get insight into the molecular processes during condensation, DPCs were treated with
different inhibitory molecules in ultra-low attachment culture. Pertussis toxin inhibits G
protein-coupled signaling by inhibition of Gi subunit, LY249002 inhibits the phosphoinositol-
3-kinase, and Cytochalasin D inhibits actin filament polymerization. The substances were
reconstituted according to supplier’s recommendations and administered to the culture
wells with the DPCs in the designated concentrations (see chapter 3.3.2). The control
group without inhibitor was supplemented with an equal amount of DMSO. After five and
24 hours, microscopic pictures were taken.
2.9 Transcriptome analysis
2.9.1 Reverse transcription
In preparation to qRT-PCR, reverse transcription of mRNA was performed according to
manufacturer’s protocol using 100-200 ng of total RNA, oligo d(T)16, random hexamers
and reverse transcriptase (TaqMan® Reverse Transcription Kit, ThermoFisher).
2.9 Transcriptome analysis 37
2.9.2 Real-Time Quantitative Reverse Transcription PCR (qRT-PCR)
Quantitative analysis of RNA expression by Real Time PCR was performed in the Strata-
gene MX3005P QPCR System (Agilent Technologies, Germany) using the SYBR Green®
format. SYBR® Green is a double-strand DNA specific dye that binds in the minor groove
and emits light upon excitation. After each PCR cycle, the amount of double-stranded
DNA is measured. Thus, as the PCR product accumulates the fluorescence increases. To
exclude non-specificity of the PCR reaction due to contamination or primer dimers, melting
curve analysis was performed after each PCR run.
Gene expression levels were normalized to glyceraldehyde 3-phosphate dehydrogenase
(GAPDH). All primers were obtained from TIB MOLBIOL (Berlin, Germany). The qPCR
primer sequences are listed in table 1.4.
Measurements were conducted in 20 µl assays with the following components:
1 µl cDNA
0.5 µl gene specific upstream and downstreamprimer (10 µM each)
∑︀ 10 µl 2xSYBR NoROX Mix8 µl dH2O
20 µl
Temperature profile:
4 min 95 °C ⎫
12 sec 95 °C ⎬
11 sec 64 °C ⎭ 40 x12 sec 72 °C
Melting point analysis:
Samples were heated to 95 °C for 3 min, cooled to 62 °C for 30 sec, and subsequently, the
temperature was ramped from 62 °C to 95 °C in steps of 1 °C every 30 s, while fluorescence
intensity was measured.
Table 2.4: Primer pairs used for qPCR
Primer name Sequence
GAPDH up 5’TGTTGCCATCAATGACCCCTT
GAPDH do 5’CTCCACGACGTACTCAGCG
BMP4 up 5’CGGGATCTTTACCGGCTTC
BMP4 do 5’TCTGCTGGGGGCTTCATAAC
continued on next page . . .
38 Chapter 2 Materials & Methods
Primer name Sequence
BMP7 up 5’ACTGTGAGGGGGAGTGTGC
BMP7 do 5’CGAAGTAGAGGACGGAGATGG
INHBA up 5’GGGAGAACGGGTATGTGGAG
INHBA do 5’CTGGAAGAGGCGGATGGT
TGFß1 up 5’CCTGGCGATACCTCAGCAAC
TGFß1 do 5’GCCATGAGAAGCAGGAAAGG
FGF2 up 5’AGCGACCCTCACATCAAGC
FGF2 do 5’GCCCAGTTCGTTTCAGTGC
HGF up 5’GCCATGAATTTGACCTCTATGAAAAC
HGF do 5’TTTACCCCGATAGCTCGAAGG
PAX9 up 5’CCTACCACAgCCCCAAGGT
PAX9 do 5’AGCAACATAACCAGAAGGAGCAG
MSX1 up 5’GAGAGGACCCCGTGGATG
MSX1 do 5’CGATGGACAGGTACTGCTTCTG
DSPP up 5’GGCCATTCCAGTTCCTCAAA
DSPP do 5’TGGGTATTCTCTTGCCTTCCTC
COL1A1 up 5’GCCGTGACCTCAAGATGTG
COL1A1 do 5’GCCGAACCAGACATGCCTC
OPN up 5’CACTGATTTTCCCACGGACCT
OPN do 5’CCATTCAACTCCTCGCTTTCC
OCN up 5’CTCACACTCCTCGCCCTATTG
OCN do 5’CTCCCAGCCATTGATACAGGTAG
LPL up 5’GGCTGAAACTGGGCGAATCTAC
LPL do 5’CGTTGGAGGATGTGCTATTTGG
FABP4 up 5’GCAGAAATGGGATGGAAAATCA
FABP4 do 5’CGTCCCTTGGCTTATGCTCTC
Nestin up 5’AGGCTGAGGGACATCTTGAGG
Nestin do 5’AGCGTTGGAACAGAGGTTGG
TUJ1 up 5’CAGCAAGGTGCGTGAGGAG
TUJ1 do 5’TGCGGAAGCAGATGTCGTAG
SDC up 5’GGGGAGCAGGACTTCACCT
continued on next page . . .
2.9 Transcriptome analysis 39
Primer name Sequence
SDC do 5’TCCCAGCACCTCTTTCCTGT
TNC up 5’GGGACAGCAGGTGACTCCAT
TNC do 5’TGTCCCCATATCTCCCCATC
NCAM up 5’GTTCAAGCAACACCCCCTCTT
NCAM do 5’TTCTTCACCAACTGCTCTCCAC
FN1 up 5’CAGACCTATCCAAGCTCAAGTGG
FN1 do 5’TGGGTGGGATACTCACAGGTC
For calculation of expression ratios, the following formula was used, whereas the efficiency
(E) of amplification was set to 1.95 and the crossing point (Cp) indicates the number of
cycles required for the fluorescent signal to cross the predefined background level:
𝑐
𝐸 𝑝𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑔𝑒𝑛𝑒
𝑐
𝐸 𝑝𝑔𝑒𝑛𝑒 𝑜𝑓 𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡
2.9.3 Whole Genome Analysis
Transcriptome analysis was performed by using next-generation sequencing (NGS) from
Illumina. First, a cDNA library was generated employing the TruSeq® Stranded mRNA
LT Sample Prep Kit (Illumina), according to the TruSeq® Stranded mRNA Sample
Preparation Protocol LS (Illumina). The initial quantity of total RNA was 1 µg per
sample. The cDNA was enriched via PCR. Diverse purification steps to separate the nucleic
acid from reaction mix between the steps were performed with Mag-Bind® RXNPure
Plus magnetic beads (Omega Bio-Tec Inc.). As a quality control and to determine
the concentration and size of the fragments, the cDNA was analyzed with a UV-Vis
spectrophotometer (NanoDrop2000, Thermo Scientific) and gel electrophoresis (2% agarose).
Sequencing was carried out by the Illumina MiSeq® System. Raw data, generated by
the Illumina MiSeq® platform, was processed on the Galaxy Project Platform (Afgan
et al., 2018), using the tools FASTQ Groomer, to convert the output data into Sanger
sequencing data. Fragments were then mapped against the human genome (hg38) to
detect splice junctions between exons by HISAT2. Mapped reads were processed to the
FeatureCount tool on the Galaxy Project Platform. Differential expression analysis was
performed by the DESEQ2 algorithm (nominal p-value < 0,05; FD>2) provided by DE
analysis (https://yanli.shinyapps.io/DEApp/ - App bioinformatics core, Center for Research
Informatics (CRI), Biological Science Division (BSD), University of Chicago). Differentially
regulated genes were analyzed for overrepresented gene sets by GSEA (Gene Set Enrichment
Analysis – http://software.broadinstitute.org/gsea/index.jsp; broad institute, Cambridge)
(Subramanian et al., 2005). Heatmap generation was conducted by ClustVis analysis
platform (Metsalu et al., 2015) and the Morpheus tool from broad institute Cambridge
(https://software.broadinstitute.org/morpheus).
40 Chapter 2 Materials & Methods
2.10 Statistical analysis
Statistical analyses were conducted using GraphPad Prism software version 6.04 (GraphPad
Software Inc., USA). Paired t-test was applied to paired data series and Mann-Whitney
U test was applied to unpaired series, both for not normally distributed data as verified
by Shapiro-Wilk test. P values smaller than or equal 0.05 were considered as statistically
significant.
2.11 Bioinformatics
Primers were designed with Primer3 version 0.4.0 (http://bioinfo.ut.ee/primer3-0.4.0/). Re-
striction analyses were performed with NEB Cutter V2.0 (http://nc2.neb.com/NEBcutter2/)
and sequence alignments were conducted using CLUSTALW (https://www.genome.jp/tools-
bin/clustalw). Vector sequences were obtained from addgene (https://www.addgene.org/)
and vector maps were designed with SnapGene Viewer 4.2.6.
Raw data from qPCR runs conducted on Stratagene MX3005P qPCR thermal cycler were
analyzed using MxPro QPCR software (Agilent Technologies, Germany) and MS Excel
software (Microsoft Corporation).
Flow cytometric data obtained from MacsQuant analyzer were evaluated by usage of the
FlowJo, LLC software V10 (FlowJo, Oregon, United States) including proliferation kinetic
calculation.
Time-lapse videos were assembled from individual frames with Photolapse software.
Single-cell fluorescence intensity measurements were conducted using Fiji ImageJ software
(Schindelin et al., 2012).
CHAPTER 3
Results
3.1 Characterisation of Human Dental Pulp Cells
3.1.1 Isolation and in vitro growth characteristics
Human dental pulp cells (DPCs) reside in the pulp tissue of teeth. The pulp tissue was
obtained from wisdom teeth of donors after informed consent. The Hematoxylin/Eosin
staining (see figure 3.1) showed, that the obtained pulp tissue consisted of the fibrous
extracellular matrix, with the fibroblasts dispersed in it, a cell-rich layer of odontoblasts
which was located at the pulpal-dentin-junction (top) and blood vessels lined by cells
(pericytes and endothelial cells) (bottom).
Figure 3.1: Hematoxylin/Eosin staining of pulpal specimen obtained from human donor
patients with detailed structure of the remaining odontoblast layer (top) and vasculature
(bottom).
After cracking up the hard tissue of the teeth, the pulp was enzymatically dissociated,
and single cells were isolated. Since they are of mesenchymal stromal cell type (Shi et al.,
2005), DPCs can be isolated by plastic adherence on tissue culture dishes. They exhibited
a fibroblastic phenotype and expanded quickly with the ability to generate clonogenic
cell clusters (see figure 3.2 b, c). In comparison to mesenchymal stromal cells from bone
marrow, dental pulp cells were much smaller in in vitro culture.
41
42 Chapter 3 Results
(a) (b) (c)
Figure 3.2: (a) Removal of dental pulp from tooth specimen, (b) colony of DPCs 72 hours
after isolation, (c) confluent layer of DPCs in standard cell culture condition and visible
fibroblastic morphology of expanded cells.
For analysis of cell growth char-
acteristics, a CFSE assay was per-
formed. Cells were covalently
labeled with an intracellular fluor-
escent dye which is evenly distrib-
uted among daughter cells after
cell division, resulting in a fluores-
cence intensity half of the parental
generation. This assay was used to
quantify population doublings per
time by FACS analysis. The CFSE
assay resulted in a generation time
of 25 hours, calculated by FlowJo
Proliferation Platform for stand-
ard monolayer culture (see figure Fig. 3.3: Histogram plots of CFSE intensity of
3.3) which is in conformity with lit- DPCs after 1, 2 and 5 days in monolayer culture
erature (Huang et al., 2006; Pis- compared to initial CFSE-labeled population (0 h).
ciotta et al., 2015; Suchánek,
Soukup, Ivančaková et al., 2007).
3.1.2 Surface Marker Expression
Cultured human dental pulp stem cells displayed the typical surface marker expression
that are described for mesenchymal stromal cells. They were positive for CD90 (Thy-1),
CD105 (SH2), CD106 (VCAM-1), CD13, CD44 (hyaluronic acid receptor) and CD146
(MCAM), and were negative for the hematopoietic markers CD45, CD14, CD34 as well
as the endothelial cell marker CD31 (PECAM-1) as shown in figures 3.4 and 3.5. For
comparison, the expression of the markers was also measured in bmMSCs (see Appendix
5.1 A).
3.1 Characterisation of Human Dental Pulp Cells 43
Figure 3.4: Representative plots of surface marker expression of DPSCs analyzed by Flow
Cytometry. Open histograms (red) are shown for the respective stained surface molecule, and
tinted histograms represent the corresponding isotype controls (grey). The upper left plot
shows the "live" gate in FSC/SSC.
Figure 3.5: Summary of expression of typical MSC surface markers assessed by flow cytometric
analysis. Positive markers are shown in red and negative in green (N=4).
44 Chapter 3 Results
3.1.3 Expression profile of relevant genes
Upon standard cell culture, when dental pulp cells adhered to plastic, they changed their
morphology and exhibited a fibroblast-like phenotype. Since the process of dedifferentiation
upon ex vivo culture, especially of mesenchymal cells, is described, a literature review
was performed to identify marker genes of odontoblast or mature dental pulp cell type.
The expression of these genes was measured of freshly isolated cells from the dental pulp
(DPCs), as well as from cultured DPCs over six passages. Additionally, dedifferentiated
bone marrow mesenchymal stromal cells (MSCs) from passage six (MP6) were characterized
for the respective gene expression.
Figure 3.6: Gene expression profile of freshly isolated human dental pulp cells, in standard
culture over six passages (P0-P6) and in comparison to MSCs from bone marrow in passage
six (MP6). Quantitative RT-qPCR data is shown as mean ± SEM compared to GAPDH
expression. Statistical analysis was performed using Mann-Whitney U test. Differences were
significant for P < 0.05 (*), P < 0.01 (**), P < 0.005 (***). 𝑁 ≥ 5 biological replicates. n.d. -
not detected
3.1 Characterisation of Human Dental Pulp Cells 45
Figure 3.6 shows the relative expression of selected genes which are described to play
central roles in dental homeostasis, cytodifferentiation or tooth development. Among
them are morphogens belonging to the TGFß superfamily: BMP7, TGFß1 and INHBA
(Activin).
Freshly isolated dental pulp cells expressed BMP7 at a moderate level and the expression
entirely dropped in two-dimensional cell culture. Over the culture period of six passages,
the expression further decreased and was barely detectable. In cultured MSCs from late
passage six also no transcripts of BMP7 were detected. INHBA was expressed at a
comparable level in freshly isolated pulp cells and decreased about 3-fold upon monolayer
culture. The INHBA gene encodes a protein subunit whose homodimers form ACTIVIN A.
The TGFß1 expression was higher, compared to the other genes, and decreased more than
2.5-fold in cultured DPCs compared to directly isolated. It did not change over time of
monolayer culture. For none of these three markers of the TGFß superfamily, a significant
difference between dedifferentiated DPCs and MSCs was detected.
Furthermore, transcript expression of two other morphogens, fibroblast growth factor
2 (FGF2 ) and hepatocyte growth factor (HGF), was analyzed. No difference in FGF2
expression was observed between primary isolated and monolayer cultured DPCs. Although
not statistically significant, a tendency of a higher expression of FGF2 in MSCs from bone
marrow was observed. No statistically significant difference in HGF expression between
isolated DPCs and cultured DPCs as well as MSCs was detected.
Two main dental transcription factors were included in this analysis: the homeobox
transcription factor Msh homeobox 1 (MSX1 ) and paired box gene 9 (PAX9 ). Interestingly,
both transcription factors tended to be specific for the dental pulp cells here, since they
were not or barely measured in MSCs from bone marrow. It is known that in general, the
expression height of transcripts from transcription factors is lower than that of secreted
molecules, such as growth factors or extracellular matrix proteins. Nevertheless, PAX9
was expressed at a considerably high level in DPCs and was not diminishing upon 2D
culture. MSX1 was expressed about 2.5-fold lower than PAX9 in freshly isolated DPCs
and significantly decreased upon monolayer culture.
Collagen type I (COL1A1 ) is the predominant extracellular matrix molecule in the dental
pulp. Therefore, it is not surprising that a remarkable high expression in the isolated
cells from the dental pulp was detected. After monolayer culture, the cells exhibited a
diminished expression of COL1A1, though it was still high (one- to two-fold higher than
GAPDH ). The expression of COL1A1 in MSCs in culture was at a comparable level.
The DSPP gene encodes for a preproprotein secreted by odontoblasts. This preprotein is
proteolytically processed to generate dentin sialoprotein and dentin phosphoprotein, both
principal mineralization proteins of the dentin extracellular matrix. Here, the transcripts
of DSPP were exclusively detected in primary isolated cells from the dental pulp. Neither
the de-differentiated DPCs nor the MSCs showed any expression of this marker gene.
Compared to COL1A1 the expression of DSPP was more than 25-fold lower.
46 Chapter 3 Results
3.1.4 Differentation capacity
On the basis of marker gene expression of cultured DPCs in monolayer (see Chapter 3.1.3),
dedifferentiation of the cells was assumed. The cells represented a precursor phenotype,
and a mandatory feature of mesenchymal precursor cells is the ability to differentiated
to multiple cell types under appropriate conditions. This attribute is referred to as
multipotency. In order to assess the multipotent capacity of the isolated DPCs, standard
protocols for adipogenic, osteogenic and neurogenic differentiation were applied.
Adipogenic differentiation
As shown in figure 3.7 the expression of lipoprotein lipase
(LPL) and fatty acid binding protein 4 (FABP4), marker genes
for adipocytes, tended to be upregulated upon adipogenic
differentiation. Nevertheless, in the control group as well as in
the differentiated cells, expression of these two marker genes
was extremely low (10 000 to 2000-fold lower than GAPDH ).
The poor differentiation capacity to adipogenic lineage was not
only observed at molecular level. The Oil Red O staining for
detection of lipid vacuoles inside the cytoplasm revealed only
faint staining (see figure 3.8, right). The vacuoles tended to be
numerous, yet small in size. However, a microscopic difference
between the two groups (undifferentiated vs. differentiated
DPCs) was observed. The left picture in figure 3.8 depicts
a representative image the control group of undifferentiated
cells.
Fig. 3.7: Marker gene
expression of adipogenic
differentiation. Quantit-
ative RT-qPCR data is
shown as mean ± SEM Figure 3.8: Oil Red O staining for lipid droplets upon
compared to GAPDH adipogenic differentiation (representative image), N=6
expression. Statistical
analysis was performed
using paired t test. N=5
biological replicates.
3.1 Characterisation of Human Dental Pulp Cells 47
Osteogenic differentiation
To assess differentiation capacity to osteogenic lineage upon
culture in specific differentiation media, gene expression ana-
lysis of marker genes of osteoblasts was performed 28 days
after induction. Figure 3.9 indicates an upregulation of Os-
teopontin (OPN ) and Osteocalcin (OCN ). Osteopontin is a
major component of the bone matrix and binds hydroxylapat-
ite. Osteocalcin represents a peptide hormone that regulates
bone mineralization and calcium ion homeostasis. During the
osteogenic culture a remarkable matrix deposition was noticed
on the cell layer. Macroscopically, a strong positivity of Aliz-
arin Red staining was observed (see figure 3.10, right picture).
Alizarin Red stains calcified matrix deep red. The left picture
in 3.10 depicts a representative image of non-differentiated
DPCs in monolayer stained with Alizarin red.
Fig. 3.9: Marker gene
expression of osteogenic
differentiation. Quantit-
ative RT-qPCR data is
shown as mean ± SEM
compared to GAPDH
Figure 3.10: Alizarin Red Staining for deposition of expression. Statistical
calcified matrix upon osteogenic differentiation analysis was performed
(representative image), N=6 using paired t test. N=5
biological replicates.
48 Chapter 3 Results
Neurogenic differentiation
Human dental pulp cells are of ectomesenchymal origin. There-
fore, ontologically they have a close relationship to ectodermal
cells. Several groups have shown that they are capable to dif-
ferentiate to neurogenic lineage upon suitable differentiation
induction. To show that the used cells are also capable of
neurogenic differentiation, they were cultured in neurogenic
media. Figure 3.11 displays an approximately 3-fold upregula-
tion of the measured neurogenic markers genes Nestin (NES)
and TUJ1 after two weeks of differentiation induction. The
TUJ1 gene encodes the neuron-specific class III ß-tubulin, a
structural protein that contributes to microtubule stability
in neuronal cell bodies and axons. Nestin is an intermediate
filament expressed mostly in nerve cells with an implication
in the radial growth axons. Morphologically the differentiated
cells were distinct from the non-differentiated DPCs. They
were elongated, bigger and had larger cell processes. On pro-
tein level, a high accumulation of the NESTIN protein was
detected in the cytoplasm as assessed by immunocytological
staining (see figure 3.12). Nuclei were counterstained with
DAPI.
Fig. 3.11: Marker gene
expression of neurogenic
differentiation. Quantit-
ative RT-qPCR data is
shown as mean ± SEM
compared to GAPDH
expression. Statistical
analysis was performed
using paired t test. Dif-
ferences were significant
for P < 0.05 (*), N=5
biological replicates.
Figure 3.12: Immunocytological staining for Nestin (green)
and nuclear counterstaining with DAPI (blue).Left: Dental pulp
cells without neuronal differentiation induction (Control); Right:
Dental pulp cells after 14 days of neuronal differentiation
induction. (representative image), N=5
3.2 Mechanisms of ectomesenchymal condensation 49
3.2 Mechanisms of ectomesenchymal condensation
In the course of embryogenesis, the development of numerous organs such as hair, teeth,
feathers, salivary glands, mammary glands, bone, cartilage, kidney, lung and many more
is induced by the so-called mesenchymal condensation. Organ fate-specific mesenchymal
cells form a tightly packed signaling center and thereby orchestrate the organ development.
For long, this step of organogenesis was underestimated, and to date, only little is known
about the molecular events during human mesenchymal condensation especially in the
tooth development. Therefore, the focus of this thesis was to generate a suitable model to
study human dental mesenchymal condensation. Based on my experiences of mesenchymal
condensation in cartilage development, the model was established as a non-forced aggrega-
tion of cells. In contrast to previous publications from other groups, that actively pack the
cells by micromass, pellet, hanging drop or fibronectin island culture, here the cells were
seeded into ultra-low attachment plates. In this setup, the cells have no opportunity to
interact with a matrix or the culture dish.
3.2.1 Establishment of 3D cell culture
Human tooth initiation begins in the fifth week after gestation when the ectomesenchymal
cells from the neural crest have already migrated into the developing jaws and crowd
underneath the oral epithelium at the site of tooth development. The complete mesenchymal
condensate in late cap stage of tooth formation has an approximate size of 500 µm. Only
one condensate per tooth is formed. Therefore, the requirement for the tooth condensate
model was to generate one aggregate of 500 µm diameter per well from single cell DPCs.
The ideal cell concentration to obtain a single aggregate turned out to be approximately 1
million DPCs per ml medium seeded into ultra-low attachment (ULA) wells. The number
of used cells was decisive for the condensate size. It turned out that 200 000 DPCs seeded
in 96-well of ULA resulted in a condensate with 500 µm in diameter. A videomicroscropic
analysis was performed to assess the kinetics of the condensation process. Depending on
the primary human donor, the time until a dense aggregate has formed, took between 20
to 36 hours. In figure 3.13 representative images from the microscopic video records are
shown. After 12 hours all cells were integrated into a cellular network, and afterwards,
the cell-cell-attachments led to an intense self-orchestrated aggregation. The upper right
picture, shows the seeded DPCs at the start of the experiment. In the lower right picture
a higher magnification, of the condensing cells at 6 hours after induction in ULA-culture is
imaged. Notably, strong cell-cell contacts were identified (arrows).
50 Chapter 3 Results
Figure 3.13: Time-lapse videomicroscopic analysis of in vitro condensation in 96-well format
over a course of 24 hours. At higher magnification the suspended cells at the beginning (top
right) and after six hours of condensation (bottom right) are depicted. The cell-cell contacts
are visible (arrows).
3.2.2 Expression kinetics of relevant genes
To gain insight into the molecular events triggered by the initiation of the in vitro con-
densation, a panel of described dental developmental genes was chosen to be analyzed
by qRT-PCR. Furthermore, whole transcriptome analysis was performed by using a Next
Generation Sequencing method (Illumina). First, gene expression in long-term range was
investigated. The dental pulp cells were seeded into the ultra-low attachment wells, and
RNA was extracted after 24 hours, four days and four weeks for subsequent qRT-PCR.
These expressions were compared to the expression of cells in standard monolayer culture,
as well as to freshly isolated dental pulp cells. The panel of genes included four members
of the TGFß superfamily ligands (TGFß1, BMP7, BMP4 and INHBA), FGF2, the ECM
molecules COL1A1 and DSPP, as well as the transcription factors MSX1 and PAX9.
As described before, the molecules of the TGFß superfamily play a significant role in
early and late tooth development. Three of the four investigated members (BMP7, TGFß1
and INHBA) were subject to the dedifferentiation, exhibiting a lowered abundance in
cultured monolayer compared to directly isolated cells. BMP4 expression was not changed
upon standard culture. 24 hours after condensation induction, the expression level of
BMP4 dramatically dropped and also after four days barely any BMP4 transcripts were
detected. Only after four weeks, higher expression was observed again. The expression
3.2 Mechanisms of ectomesenchymal condensation 51
BMP4 BMP7 TGFß1
ns
10 -1 10 -1 ** 100 ****
**** ** ns
**** ns ns
10 -2
10 -2
10 -1
10 -3
10 -3
10 -4
10 -2
10 -4 ns -5 *** *10
ns ns ns
10 -5 10 -6 10 -3
te
d
-P
6 4h2 4
d 4w ed L 4h 4d 4w ed P6 4h da 4 at Ml l 2 lat - 2
4 4w
iso P iso so P
4
y Lhl M hly hly
i L
M
es esfr fr fre
s
Comparison P value Comparison P value Comparison P value
f.i. vs ML 0.9980 ns f.i. vs ML 0.0003 *** f.i. vs ML 0.0244 *
ML vs 24h < 0.0001 **** ML vs 24h 0.7267 ns ML vs 24h 0.3672 ns
ML vs 4d < 0.0001 **** ML vs 4d 0.0072 ** ML vs 4d 0.6837 ns
ML vs 4w 0.0946 ns ML vs 4w 0.0013 ** ML vs 4w < 0.0001 ****
f.i. vs 4w 0.0635 ns f.i. vs 4w 0.2222 ns f.i. vs 4w 0.9004 ns
INHBA FGF2 MSX1
10 -1 ** 100 **
ns
-1
* ns
10
****
ns ns
-2 ****10
10 -1
10 -3
10 -2
10 -4
10 -2
** ns -5 ****10
ns * **
10 -3 10 -3 10 -6
ed P6 4h 4d 4w ed P6 4h 4d 4w edt - 2 t - 2 t -P
6 h d w
24 4 4
iso
la aP4 ol P4 l
a 4
y L i
s L is
o P
L
hl M hly M hly Ms s s
fre fre fre
Comparison P value Comparison P value Comparison P value
f.i. vs ML 0.0084 ** f.i. vs ML 0.6827 ns f.i. vs ML < 0.0001 ****
ML vs 24h 0.3680 ns ML vs 24h 0.1542 ns ML vs 24h < 0.0001 ****
ML vs 4d 0.0170 * ML vs 4d 0.0656 ns ML vs 4d < 0.0001 ****
ML vs 4w 0.0053 ** ML vs 4w 0.0017 ** ML vs 4w 0.5635 ns
f.i. vs 4w 0.6753 ns f.i. vs 4w 0.0173 * f.i. vs 4w 0.0051 **
PAX9 DSPP COL1A1
ns
10 -1 100ns ns 10
2 *
****
** 10 -1 ns **
101
10 -2 10 -2
10 -3 100
10 -3 10 -4
ns 10 -1 ns
10 -5
ns ns
10 -4 10 -6 * 10 -2
ed P6 4h 4d 4w ed P6 4h 4d 4w ed P6 4h 4dt - 2 t - 2 t - 2 4
w
iso
la a aP4 ol P4 ol P4
L is L is L
hly M ly M ly Ms sh sh
fre fre fre
Comparison P value Comparison P value Comparison P value
f.i. vs ML 0.7037 ns f.i. vs ML 0.0238 * f.i. vs ML 0.3983 ns
ML vs 24h 0.0050 ** ML vs 24h 0.9000 ns ML vs 24h 0.0011 **
ML vs 4d 0.2005 ns ML vs 4d 0.8000 ns ML vs 4d < 0.0001 ****
ML vs 4w 0.2318 ns ML vs 4w ML vs 4w 0.0484 *
f.i. vs 4w 0.3095 ns f.i. vs 4w f.i. vs 4w 0.0823 ns
Figure 3.14: Gene transcription level of chosen marker genes during the long-term condens-
ation process. Depicted are relative expression values (mean ± SEM; relative to GAPDH )
of the DPC condensates at 24 hours (24h), 4 days (4d) and 4 weeks (4w) after induction of
condensation culture. For comparison also expression values of freshly isolated DPCs (f.i.) and
dedifferentiated monolayer DPCs in different passages (ML P2-P6) are presented. Significance
was calculated by Mann-Whitney U-Test: ns - not significant; * - P < 0.05 ; ** - P < 0.01; ***
- P < 0.001; **** - P < 0.0001. 𝑁 ≥ 5 biological replicates.
Ratio GAPDH Ratio GAPDH Ratio GAPDH
Ratio GAPDH Ratio GAPDH Ratio GAPDH
Ratio GAPDH Ratio GAPDH Ratio GAPDH
52 Chapter 3 Results
profiles of BMP7, TGFß1 and INHBA were concordant: no significant change was detected
after 24 hours in low attachment culture. After four days, the expression was elevated
(TGFß1 only slightly). Notably, after four weeks the transcript level of TGFß1, BMP7,
and INHBA was similar to that of the primary isolated dental pulp cells.
The fibroblast growth factor 2 (FGF2) is described to promote the survival and proliferation
of multiple cell types and to enhance angiogenesis (El Agha et al., 2016). Also it is
described to promote mesenchymal condensation in vivo (Perantoni et al., 1995; Song
et al., 2004). Since the signaling cascade, inducing the condensation in the presented ultra-
low attachment culture model was unknown, it was hypothesized that it is FGF2-based
by autocrine stimulation. Nevertheless, FGF2 expression was not elevated in the earlier
phases of condensations (24 hours and four days). Only after four weeks, a significant
increase in expression was noticed.
Two important transcription factors were included in the panel: MSX1 and PAX9. Surpris-
ingly, MSX1 expression dramatically decreased upon 3D culture and only slightly increased
at four weeks. PAX9 transcript level was not changed upon dedifferentiation in 2D culture.
After induction of in vitro condensation, the expression slightly dropped after 24 hours and
then reconstituted to the initial level. Taking into account that PAX9 is a transcription
factor, the overall expression was comparably high.
The initial expression of dentin sialophosphoprotein (DSPP) in the freshly isolated DPCs
was completely lost upon standard monolayer culture. In some donor populations, an
increase in expression was observed upon long-term cultivation of the condensates (4 weeks).
Here, statistical analysis was not possible due to the small number of expression values. As
described before, the collagen type I expression (COL1A1 ) slightly, but not significantly,
decreased in monolayer culture and diminished further after induction of condensation
and after four days. In the 4 week-condensates, an elevation of transcript expression was
detected.
Odontogenic differentiation requires initial cytoskeletal rearrangement induced by mechan-
ical compaction and traction forces. Therefore, the very first signals, that are induced upon
and in the course of the contraction during the in vitro condensation were of interest in the
next experiment. A short-term kinetic was performed, and condensation and odontogenic
marker genes were measured. The goal was to determine the time point at which the cells
reacted with the most prominent change in gene expression on the condensation induction.
Figure 3.15 depicts the analysis of the condensations at two to six hours, each in comparison
to the monolayer. Genes that are described to be physically involved in cell-cell-adhesion
show the same tendency. Transcripts of the extracellular matrix glycoproteins fibronectin 1
(FN1) and tenascin C (TNC), and the membrane receptors neural cell adhesion molecule
(N-CAM) and syndecan 1 (SDC1) decreased slightly with time. Tenascin C was expressed
10-fold higher than fibronectin. After four and six hours, transcript level of fibronectin was
significantly reduced. Also, in expression of BMP4, INHBA and FGF2 striking differences
were observed over time. The dramatic drop of BMP4 expression, which has been observed
in the long-term analysis (see figure 3.14), occurred already after four hours. The expression
of INHBA and FGF2 rose with time around 3-fold. No tendency of change in PAX9
expression was detected at the depicted early time points. A significant reduction was only
3.2 Mechanisms of ectomesenchymal condensation 53
observed after 24 hours (see figure 3.14).
Figure 3.15: Gene expression (mean ± SEM) of condensation marker genes during early
phase of three-dimensional culture relative to GAPDH expression. Significance was calculated
by paired t-test: ns - not significant; * - P < 0.05. N=4 biological replicates.
3.2.3 Transcriptome of early phase of in vitro condensation
Transcriptome analysis was performed at a very early time point, at six hours after
condensation induction. This point in time was chosen for two reasons. Firstly, the goal
was to identify gene expression changes that are directly associated with the change of
cellular morphology. Secondly, the short-term kinetic (see figure 3.15) revealed that changes
regarding the expression of the chosen marker genes was minor and only at six hours
after condensation induction, three of the eight chosen genes were significantly up- or
down-regulated. For two individual donor cell populations, the total RNA of monolayer
cells, and of cells, that were cultured in ultra-low attachment for six hours, was prepared
for RNA sequencing. This work was done within the supervised Diploma thesis of Julia
Bräunig.
Data quality
Figure 3.16 (a) shows that the obtained data from the RNA Seq significantly correlated
with qRT-PCR results. Real-Time PCR was performed by using the cDNA from the
54 Chapter 3 Results
samples, that were used for RNA Seq. The Pearson coefficient is calculated to be R2=0.829
(P value<0.0001). A "Differential Gene Expression Analysis" by the DESEQ2 method
was performed to identify the genes which are significantly differentially expressed upon
the culture condition in both donor cell populations. The analysis with a cutoff in fold
change of 1.5 and a significant nominal P value resulted in 521 differentially expressed
genes (303 upregulated, 218 downregulated). The complete list of differentially expressed
genes containing the "Reads per Million" for each gene and each sample is attached in the
Appendix.
Figure 3.16: Quality of the RNA Seq data. (a) Significant correlations between expressions
of quantitative real-time PCR and RNA-Seq. Depicted are the log2 fold changes of seven genes,
obtained by RNA Seq (FPKM values) or by qRT-PCR (𝛥 Ct to GAPDH values). FPKM-
fragments per kilobase of transcript per million mapped reads.
Analysis of the whole transcriptome
For significant determination of cell type-specific and condensation dependent expression
pattern an additional data set of cell condensation (24h) conducted with bone marrow-
derived stromal cells was included. Heatmap analysis produced two main groups represented
by the DPC and MSC subpopulations. Within the MSC group, monolayer cultivated cells
and condensed cells after 24h of low attachment cultivation formed two separate clusters.
Interestingly, regarding the DPC subpopulations, clustered comprehensive heatmap analysis
of all expressed genes revealed the transcriptome properties of individual primary cells
with donor intrinsic expression intensities, characterized by donor-specific clustering (see
figure 3.17).
Figure 3.17: A heatmap over relative expression levels of all expressed genes in monolayer
cultured (ML) or condensed (cond) DPCs or MSCs respectively. The color code represents the
range between minimal (blue) and maximal (red) expression based on RPM values.
3.2 Mechanisms of ectomesenchymal condensation 55
Assessment of differentially expressed genes and affected molecular and biological functions
To assign differentially expressed genes upon condensation induction the DESEQ2 al-
gorithm from the Center for Research Informatics (CRI), University of Chicago (ht-
tps://yanli.shinyapps.io/DEApp/) was applied. Comparative cluster analysis of the defined
521 regulated genes demonstrate an almost even distribution of 58.2 % up- and 41.8 %
downregulated genes.
Figure 3.18: A heatmap over relative expression levels of all differentially expressed genes
between monolayer cultured (ML) or condensed (cond) DPCs according to DESEQ2 algorithm
(nominal P value < 0.05). 521 differentially expressed genes display even distribution of
upregulated and downregulated genes The color code represents the range between minimal
(blue) and maximal (red) relative expression based on RPM values.
To allocate affected biological processes, molecule classes and signal transduction path-
ways, a Gene Set Enrichment Analysis (GSEA) with focus on particular terms like biological
function or signal transduction pathway was performed. Induction of DPC condensation
produced mainly GO terms connected to cell differentiation, developmental processes and
cell communication in the “biological process” set of genes. According to the intensified
cell-cell contact in condensation culture, GO terms associated to anchoring junction, focal
adhesion, and gap junction formation were defined to be overrepresented within the set of
differentially expressed genes.
Several more specialized gene sets for signal transduction pathways emerged repeatedly
in the particular sub-categories and are applicable to describe cellular changes in detail.
For evaluation of the selected pathways not only the regulated genes were taken into
consideration, but also the general expression strengths was included to estimate the
activity of the pathway. The detailed results can be obtained from the appendix (see
figures 5.7 and 5.8).
56 Chapter 3 Results
The TGFß signal transduction pathway represents a key element in regulation of
developmental processes and differentiation. The selected gene set KEGG – TGFß signaling
pathway comprises 86 genes. Cluster heatmap analysis of the expressed genes of the gene
set overruled the cluster relationship caused by the primary cell character and resulted
in a distinct ML and condensation sample cluster. The heatmap diagram reduced to the
significantly regulated genes showed a higher proportion of upregulated genes.
Figure 3.19: Clustered heatmap for KEGG – TGFß signaling pathway. (A) Cluster analysis
of all expressed genes of the gene panel and (B) with focus on significantly regulated genes.
The color code represents the range between minimal (blue) and maximal (red) expression
based on RPM values.
The three growth factors TGFß1, TGFß and TGFß3 were significantly upregulated
along with the TGFß specific signal transducer SMAD3. Together with the downregulation
of BMP4 and the TGFß signaling specific inhibitor SMURF2 the data suggest a shift from
BMP to TGFß signaling. On the other hand, the BMP pathway specific inhibitor Noggin
(NOG) was downregulated and the member of the BMP subfamily BMP8A was upregulated.
To gain a more significant statement a recently published panel of TGFß signaling inhibitor
and enhancer molecules (Chang, 2016) was included in the evaluation (see Appendix figure
5.9). Furthermore, the absolute expression strength of the pathway-associated genes was
taken into account, despite their regulation (see figure 3.20). From the 33 in this context
described regulatory molecules, 29 were found to be expressed in condensed DPCs on RNA
level with conditions to have moderate gene expression level in the DPC condensation
samples or at least a two-fold difference in the RPM values of both samples. Three further
TGFß regulatory molecules were significantly regulated according to the DESEQ2 analysis.
The TGFß signaling supporting molecule connective tissue growth factor (CTGF) displayed
down-modulated expression levels, and TGFß enhancer syndecan2 (SDC2 ) and the BMP
antagonist gremlin 2 (GREM2 ) were upregulated. Furthermore, several genes like the BMP
inhibitors biglycan (BGN ), follistatin (FST ) and the cysteine rich transmembrane BMP
regulator 1 (CRIM1 ), the TGFß inhibitor EMILIN1 as well as the bivalent acting molecules
Follistatin-like 1 (FSTL1 ), Glypican (GPC ) and Twisted gastrulation 1 (TWSG1 ) exhibit
high expression levels, independent from the cultivation condition (see Appendix figure
5.9).
3.2 Mechanisms of ectomesenchymal condensation 57
Figure 3.20: RPM values of TGFß signal transduction pathway-associated genes are depicted.
The KEGG gene subset comprises 86 genes. The diagram is reduced to genes with at least
moderate gene expression level in the DPC condensation samples and/or genes with an at
least two-fold difference in the RPM values of both samples, resulting in 64 genes. Dashed
line marks the level of highly expressed genes (RPM>100); dotted line marks the level of
moderately expressed genes (RPM>5). Statistical significance was calculated from DESEQ2
analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
The comprehensive analysis of the expression levels of the KEGG – TGFß signaling
pathway reveals an extensive expression of molecules involved in all hierarchical levels of
the TGFß signal transduction. Figure 3.20 shows the expression of TGFß pathway-assigned
genes, subcategorized to high, moderate, and low gene expression level, while figure 3.21
pictures the genes in the context of the pathway. Both, the BMP as well the TGFß pathway
are active in DPC. However, the regulated genes and the expression levels of the regulative
gene panel drives the assumption of a BMP to TGFß signaling shift.
58 Chapter 3 Results
Figure 3.21: The TGFß signaling pathway scheme of the KEGG database relate the particular
genes to their position and function within the signaling cascade. The pathway diagram displays
the genes with at least moderate gene expression level in the DPC condensation samples and
the genes with at least a two fold difference in the RPM values of both samples, marked with
red stars (64 genes). +p/-p: phosphorylation/dephosphorylation, +u: ubiquitinylation.
The binding of TGFß ligands to their receptor system elicits, besides the canonical
Smad2/3 pathway, two further signal transduction pathways: the MAPK pathway and
signaling by the small GTPase RhoA (see figure 3.21). MAPK kinase signaling is central
to multiple signal transduction pathways and activates or inactivates target molecules
by phosphorylation. According to the GSEA results, the MAPK signaling pathway
represents the most significant overrepresented pathway in both the KEGG panel and the
BIOCARTA gene sets. For further analysis, the KEGG gene set MAPK Signaling Pathway
comprising 267 genes was selected. Clustered heatmap analysis of the gene set resulted in
two distinct clusters for monolayer culture and DPC condensation (see figure 3.22).
3.2 Mechanisms of ectomesenchymal condensation 59
Figure 3.22: Clustered heatmap for KEGG – MAPK signaling pathway. (A) Cluster analysis
of all expressed genes of the gene panel and (B) with focus on significantly regulated genes.
The colour code represents the range between minimal (blue) and maximal (red) expression
based on RPM values.
Figure 3.23: The MAPK signaling pathway scheme of the KEGG database relate the partic-
ular genes to their position and function within the signaling cascade. The pathway diagram
displays the genes with at least moderate gene expression level in the DPC condensation samples
and the genes with at least a two-fold difference in the RPM values of both samples, marked
with red stars (166 genes). +p/-p: phosphorylation/dephosphorylation, +u: ubiquitinylation.
60 Chapter 3 Results
Although 22 genes of the panel are significantly regulated, the impact on the pathway
activity is hardly predictable. To determine if the condensation resulted in decreased or
elevated activity, a specific reporter construct was generated and applied in condensation
experiments (see chapter 3.3.1). However, comparable to the TGFß signaling pathway, the
vast majority of the gene group was shown to be expressed, providing the requirements of
active MAPK signal transduction. The signaling cascade is composed of hierarchical kinase
levels, where each kinase activates the subjacent level. The expression strength analysis
evidenced the presence of members of each kinase level (see Appendix figure 5.11). The
putative activity is emphasized by the plot analysis where the particular members of the
signaling pathways are allocated to their positions within the pathway map (see figure 3.23).
The second non-canonical pathway elicited by TGFß ligand binding is mediated by the
small GTPase RhoA. RhoA signaling is highly associated with cytoskeletal actin reor-
ganization, which is among the enriched gene sets by GSEA. Although the RhoA signal
transduction pathway itself is not targeted by the enrichment analysis, the general presence
and thus, the principal pathway activity was analyzed. The applied gene set was generated
by merging the gene panels PID RHOA REG PATHWAY and PID RHOA PATHWAY
from the Pathway Interaction Database (PID, http://pid.nci.nih.gov) and the addition
of RhoA activating GEF (guanidine exchange factor) and inactivating GAP (GTPase
activating factor) molecules (141 genes).
Figure 3.24: Clustered heatmap for KEGG – RHOA signaling molecules. (A) Cluster
analysis of all expressed genes of the selected gene panel and (B) with focus on significantly
regulated genes. The color code represents the range between minimal (blue) and maximal
(red) expression based on RPM values.
-0.3cm Comparable to the TGFß signaling the vast majority of the gene panel was detected
on RNA level. The 119 verified transcripts prevailed against the donor-specific differences in
expression and generated distinct clusters for monolayer cultivated cells and condensed cells.
Remarkably, the set of differentially expressed genes is characterized by downregulation
of two Rho GTPases (RND3 and RHOBTB3 ) as well as the transcriptional reduction of
GTPase activating guanidine exchange factor molecules ARHGEF25 and ARHGEF28 (see
figure 3.24). The Lim kinase 2 (LIMK2 ) phosphorylates downstream targets in the Rho
signal transduction pathway and is activated by the Rho/Rock. Significantly decreased
RNA transcript levels for LIMK2 enzyme and the downregulation of the direct Rho
axis target gene CYR61 emphasize the hypothesis of mitigated RhoA signaling activity.
3.2 Mechanisms of ectomesenchymal condensation 61
On the other hand, encompassing analysis independent of differentially expressed genes
demonstrates a broad expression of genes associated with the Rho pathway (see figure
3.25).
Figure 3.25: RPM values of RHOA signal transduction pathway-associated genes are depicted.
The KEGG gene subset comprises 150 genes. The diagram is reduced to genes with at least
moderate gene expression level in the DPC condensation samples and/or genes with an at
least two-fold difference in the RPM values of both samples, resulting in 95 genes. Dashed
line marks the level of highly expressed genes (RPM>100); dotted line marks the level of
moderately expressed genes (RPM>5). Statistical significance was calculated from DESEQ2
analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
62 Chapter 3 Results
The Wnt signaling pathway is known to be involved in early phases of tooth devel-
opment and emerges in the gene set enrichment analysis in the HALLMARK panel from
broad institute (http://software.broadinstitute.org/gsea/msigdb) of signal transduction
pathways. However, for optimal analysis of the WNT signaling pathway in the context of
DPC condensation, the KEGG associated panel of 151 genes KEGG WNT SIGNALING
PATHWAY was selected because of the more comprehensive gene panel.
Figure 3.26: Clustered heatmap for KEGG – WNT signaling pathway. (A) Cluster analysis
of all expressed genes of the gene panel and (B) with focus on significantly regulated genes.
The color code represents the range between minimal (blue) and maximal (red) expression
based on RPM values.
Clustered heatmap analysis demonstrates the assembling of a monolayer and a con-
densation cluster formed by the 121 expressed genes from the WNT gene panel (see
figure 3.26). However, to draw general conclusions about the impact on pathway activity
from the regulated gene set remains challenging. Almost all of the regulated genes show
higher transcriptional activity. Among the differentially expressed genes, the transcription
factors TCF7 and NFATC2 show elevated RNA levels providing the foundation of higher
pathway activity at the downstream level of the pathway. Two extracellular suppressor
molecules from WNT signaling are found to be expressed reciprocally. Both the upregulated
secreted frizzled related peptide 4 (SFRP4 ), as well as the downmodulated Dickkopf 2
(DKK2 ), interfere with the receptor WNT ligand binding and attenuate the triggering
signal. A comprehensive analysis by taking the general expression and the expression
intensity into account validates the principal requirements of an active WNT signaling.
Molecules covering all hierarchical pathway levels are defined to be expressed (see figure
3.27), with exceptionally high abundance of RNA levels for the ligands WNT5A and
WNT5B, the receptor molecules frizzled 1 and frizzled 7 (FZD 1 and FZD7 ), as well as
the membrane-bound co-receptor for the canonical WNT pathway LPR5 (see Appendix
figure 5.12).
3.2 Mechanisms of ectomesenchymal condensation 63
Figure 3.27: The WNT signaling pathway scheme of the KEGG database relate the particular
genes to their position and function within the signaling cascade. The pathway diagram displays
the genes with at least moderate gene expression level in the DPC condensation samples and
the genes with at least a two-fold difference in the RPM values of both samples, marked with
red stars ( genes). +p/-p: phosphorylation/dephosphorylation, +u: ubiquitinylation.
64 Chapter 3 Results
TheNotch signaling pathway was deduced repeatedly by gene set enrichment analysis,
especially in the REACTOME database context. For in-depth analysis of the regulation of
Notch signal transduction, a set of genes was generated by merging the hallmark gene set
HALLMARK NOTCH SIGNALING from Broad Institute, and the gene panel derived from
Reactome database (http://www.reactome.org) REACTOME SIGNALING BY NOTCH
resulting in an assembled group of 126 genes.
Figure 3.28: Clustered heatmap for KEGG – NOTCH signaling pathway. (A) Cluster analysis
of all expressed genes of the gene panel and (B) with focus on significantly regulated genes.
The color code represents the range between minimal (blue) and maximal (red) expression
based on RPM values.
Clustered heatmap analysis revealed the formation of a monolayer cluster and a cell
condensation cluster of the 110 expressed genes repealing the donor intrinsic clustering
(see figure 3.28). A set of 13 genes are differentially expressed according to the DESEQ2
analysis. Remarkably, both, the ligand molecules JAG1 as well as the receptor NOTCH3
were defined to be significantly higher expressed upon condensation induction and suggest
an autocrine activation of the signaling. This idea is emphasized by the downregulation
of inhibitory co-repressors Histone deacetylases (HDAC) 7 and 11 as well as the elevated
levels of transcriptional targets like HES1 and HEYL and the receptor processing molecule
FURIN. Expression analysis independent of transcriptional regulation substantiates the
Notch pathway activity. Ligand binding to the receptor activates ADAM proteases like
ADAM10 and the gamma-secretase complex (consisting of four subunit proteins). All
molecules exhibit high RNA levels (see figure 5.13). Furthermore, all molecules required to
elicit and transduce a signal upon ligand binding from the cell membrane to the nucleus
are expressed (see figure 3.29).
3.2 Mechanisms of ectomesenchymal condensation 65
Figure 3.29: The NOTCH signaling pathway scheme of the KEGG database relate the
particular genes to their position and function within the signaling cascade. The pathway
diagram displays the genes with at least moderate gene expression level in the DPC condensation
samples and the genes with at least a two-fold difference in the RPM values of both samples,
marked with red stars (36 genes from 47).
3.2.4 In vitro live imaging of collagen deposition
Collagen transcripts have been shown to be among the most abundant transcripts in the
DPCs. All collagen isoforms together summed up to 5 % or 7.5%, respectively, of all mapped
transcripts in the RNA Seq of both donor populations. Collagen type I represented the
vast majority. T. Mammoto, A. Mammoto, Jiang et al., 2015 describe that the collagen
deposition in a mesenchymal condensate mechanically supports the cellular compaction and
thereby ensures a sustained and stabilized cytoskeletal arrangement that induces cell fate
switch by upregulation of specific transcription factors. Here, a fusion protein composed of
the collagen-binding adhesion protein 35 (CNA35, 35 kDa) and enhanced green fluorescent
protein (eGFP) was used to track collagen production of DPCs in low-attachment culture
in real time. The collagen-binding adhesion proteins represent a collagen-binding domain
of the bacterial collagen adhesin, originating from Staphylococcus aureus. S. aureus is
the most common pathogen in bacterial arthritis and acute osteomyelitis (Patti et al.,
1994). The cell surface molecule adhesin mediates the attachment of the bacteria to the
cartilage or bone tissue by specific binding to collagen motifs. The approach to use the
collagen-binding domain for in situ and in vitro experiments was first described by Krahn
et al., 2006 and has the advantage to track collagen deposition or rearrangement in living
culture systems.
Here, the recombinant CNA35-eGFP was produced in the prokaryotic system, isolated by
affinity chromatography, dialyzed and was then administered to the culture medium of
the DPCs, that were transferred into ultra-low attachment wells. Fluorescent microscopic
pictures were taken every five minutes for 12 hours. In figure 3.30, frames from different
time points are depicted. The condensation process was not perturbed by the presence
of the CNA-eGFP protein. After four hours small spots on the surface of the cells in the
condensing aggregate appeared. With time, the fluorescence signal on the surface and
pericellularily increased. After 24 hours, at higher magnification, delicate meshes or fibers
66 Chapter 3 Results
were detected (bottom left picture), although most of the fluorescence, tagging the collagen
deposition, emerged pericellularly (bottom right).
Figure 3.30: Live in vitro-imaging of collagen synthesis of DPCs during condensation. The
two upper rows represent a kinetic of fluorescent microscopy over 12 hours detecting the
fluorescence of eGFP coupled to CNA35 (bars represent 500 µm). The bottom pictures depict
the fluorescently tagged collagen arrangement after 24 hours of low-attachment culture at two
different magnifications.
3.3 Functional analyses of the in vitro model for mesenchymal condensation
3.3.1 Analysis of relevant pathways
Establishment of reporter constructs
Four different reporter constructs were cloned to analyze real-time induction of relevant
pathways during in vitro condensation. In these constructs, the open reading frame for
enhanced green fluorescent protein (eGFP) was cloned behind a minimal promoter and a
response element (RE), comprising repetitive transcription factor binding sites for a specific
pathway. Here, the activation of the pathways leads to an activation of the promoter and
subsequent eGFP expression was detected by flow cytometry and fluorescence microscopy.
To test the functionality of the constructs, first, they were transduced in human dental
pulp cells in monolayer cultures and were then stimulated with known inducers of the
3.3 Functional analyses of the in vitro model for mesenchymal condensation 67
respective pathways. The concentrations were the following:
pathway inducing agent low dose high dose
TGFß1 TGFß1 2.5 µg/ml 25 µg/ml
MAPK EGF 100 µg/ml 200 µg/ml
FoxO LY294002 20 µM 40 µM
Wnt LiCl 5 mM 50 mM
FACS analysis after transduction shows that transduction with the reporter constructs was
successful (see Appendix 5.2). The minimal promoter control exhibited a low background
eGFP signal (compared to untransduced DPCs). Without exogenous stimulation, approx-
imately 19 % of the Wnt-RE transduced cells were eGFP positive, indicating endogenous
Wnt signaling in the cultured ML. Upon stimulation with LiCl, the percentage rose to
25 % at low dose and 52 % at high dose, respectively. Furthermore, in the high dose
sample, the geometric mean fluorescence of the positively transduced cells was increased
2.5-fold (see figure 3.31). It can be concluded that the Wnt-RE-reporter vector was
functional. The DPCs in monolayer exhibited strong activation of endogenous MAPK
signaling, evidenced by the high geometric mean of fluorescence of non-stimulated cells.
By administration of EGF, the percentage of positive cells was not changed significantly.
Nevertheless, the geometric mean of the transduced cells rose to 110-120 % compared to
non-stimulated DPCs transduced with the MAPK-RE-eGFP construct (see figure 3.31).
The administration of the PI3K inhibitor LY29002 simulates starvation of growth factors
and survival signals in the cells. If no inhibitor is present, specific growth factors bind
to a PI3K-associated receptor, leading to a signaling cascade, that induces proteasomal
degradation of FoxO. Without ligand-induced receptor activation (starving signal), FoxO
is stabilized, translocates into the nucleus and promotes inhibition of cell growth and/or
apoptosis signaling. Therefore, it is not surprising that the cell count of the LY294002
treated samples for the FoxO-RE-reporter testing, dramatically diminished. Nevertheless,
the stimulated, FoxO-RE transduced cells responded to the stimulation with an increase in
eGFP expression (approximately 1.5-fold geometric mean fluorescence). The percentage of
eGFP-positive cells slightly dropped. The transduced TGFß-RE cells exhibited, unstimu-
lated, only a low endogenous activation (5 %). Upon low and high dose stimulation with
TGFß1, an increase to 44 and 40 % in eGFP-positive cells was detected. Additionally, the
stimulation enhanced the eGFP expression in the positive fraction by approximately 2-fold
(see figure 3.31).
The functionality of three reporter vectors was shown (TGFß-RE-eGFP, MAPK-RE-eGFP,
and Wnt-RE-eGFP). Thereafter, the reporter vectors were used in 3D in vitro condensa-
tion. The vectors were transduced into monolayer cultured DPCs of passage 6, and after
seven days of recovery and phenotypic expression of the transgene, cells were put into
ultra-low attachment plates (96-well). The condensation process was videomicroscopically
documented beginning with the start of the condensation (0 min). No exogenous stimuli
were added to the cells.
68 Chapter 3 Results
Figure 3.31: Reporter construct functionality test: DPCs were transduced with the respective
reporter vector and stimulated with inducing stimuli for 24 hours. Afterwards, cells were
harvested and their reporter activity was assessed by flow cytometric measurement of the
geometric mean fluorescence (eGFP) of positive cells. This work was done within the supervised
Diploma thesis of Julia Bräunig.
MAPK-reporter vector during in vitro condensation
The goal was to identify or verify enriched pathways from the GSEA, that are involved in
the condensation process. Therefore, the transduced dental pulp cells, that were described
above (section3.3.1) were cultured in ultra-low attachment plates for several hours and
pictures were taken every 5 minutes. Thereby, the green fluorescence intensity resulting
from eGFP expression was analyzed. Since the eGFP accumulated over time, a second
control vector was included. DPCs transduced with the backbone vector pLL3.7 in which
the eGFP expression is under control of the constitutive CMV promoter, was used, to
acquire the accumulation rate of eGFP, since here, the expression cannot be influenced by
external stimuli. The mean fluorescence under the strong CMV promoter was higher so
that the exposure time was adjusted in each sample. Hence, for quantification, only the
change of mean fluorescence is of importance, but not the absolute intensity.
In figures 3.32 and 3.33 the pictures of the first 30 minutes of the kinetic from control
CMV-eGFP-vector and MAPK-responsive reporter vector are shown (1 frame per 5
minutes). Both exhibited a background eGFP-signal. The images were analyzed with
ImageJ to calculate geometric mean fluorescence of every single GFP-positive cell in the
capture. 12 000 to 13 000 cellular mean fluorescence intensities per picture resulted.
The mean of all cells of each time point in the respective culture was calculated and
plotted against time (figure 3.34). A slight increase in fluorescence intensity in the control
cells was observed. This represents the rate of accumulation of expressed eGFP under
control of a constitutive promotor. A conditionally activatable promoter, as it is in the
MAPK-responsible reporter vector, should exhibit the same accumulation rate when it is
not stimulated. In the MAPK-RE-eGFP-transduced cells, a steep and highly significant
increase of fluorescence activity was observed after five minutes of in vitro condensation.
After ten minutes the fluorescence further increased and was significantly higher than after
five minutes. Afterwards, only a slight increase was observed, transitioning after 20 minutes
3.3 Functional analyses of the in vitro model for mesenchymal condensation 69
Figure 3.32: Kinetik of condensation with constitutively eGFP-expressing DPCs (CMV
control vector).
Figure 3.33: Kinetic of condensation with MAPK-responsive eGFP-expressing DPCs
70 Chapter 3 Results
into a constant fluorescence level. Here, no external stimuli were given to the culture at
any time point, and the increased eGFP activity resulted from upregulated MAPK activity
induced by the condensation itself.
Figure 3.34: Graph of calculated mean fluorescence intensities (arbitray units a.u.) of the
constitutive CMV-eGFP (right)- or MAPK-RE-eGFP-reporter (left) in transduced DPCs
during in vitro condensation over a time course of 30 minutes (geometric mean of each cell per
time point in the condensation well over time). Significance was calculated by unpaired t-test:
** - P<0.01 and **** - P<0.001 from more than 12000 cells per frame.
The other reporter vectors with Wnt-RE, TGFß-RE, and FoxO-RE were also used in
the same manner. None of these vectors showed an increased activity in the early time
points. After more than one hour, the method is not applicable for measurements based
on imaging techniques. The crowding and "tissue folding" of the condensing cells in ULA
prohibited accurate measurements of fluorescence intensity per cell.
3.3.2 Inhibition of condensation with small molecules
The comprehensive analysis of the transcriptome of the in vitro condensations resulted
in specific pathways, considered for further investigations. Of particular interest was the
regulation of the cytoskeletal rearrangement. Mesenchymal condensation has been described
to be mediated by actin-p38-MAPK-SP1 pathway (T. Mammoto, A. Mammoto, Jiang
et al., 2015). This was evidenced in vivo and in vitro in a fibronectin-island model. To
show that our model of condensation resembles this process, we tested inhibitors of different
pathways. Besides an inhibitor of actin filament polymerization, cytochalasin D (2.5 µM or
0.25 µM), also an inhibitor of PI3K-Akt pathway (LY294002 at 10 µM), and an inhibitor
of G𝛼i-protein-coupled receptors (25 ng/ml), pertussis toxin (PTX), were given into the
culture medium prior to initiation of condensation in the ultra-low attachment wells. After
5 and 24 hours, respectively, cell condensation was inspected microscopically. In the control
wells (containing the solvent DMSO, that was used to reconstitute the inhibitors), the cells
condensed "normally" as described above.
3.3 Functional analyses of the in vitro model for mesenchymal condensation 71
After five hours, prominent cell-cell contacts
were visible, and after 24 hours all cells were
included into a single aggregate. The inhibit-
ors LY294002 and PTX had no visible effect
on this process. Also here, cells condensed as
in the control group. The inhibitor of actin
filament polymerization cytochalasin D at a
concentration of 2.5 µM completely disrup-
ted the process of condensation. No cell-cell
contacts were visible between the single cells
and after 24 hours still no clustering occurred
(see figure Appendix 5.3, middle). This effect
was dose-dependent. In the culture with 0.25
µM cytochalasin D, after 5 hours, smaller and
lesser cell-cell contacts were visible in compar-
ison to control (see figure Appendix 5.3, left
and right). After 24 hours, cells with low con-
centration of cytochalasin D condensed, but
to a much smaller extent than the control cul-
tures. From microscopic analyses it could not
be excluded that the cells did not condensate
due to reduced viability upon the stimulation
with cytochalasin D. Therefore, after condens-
ation of 24 hours, cells from all conditions were
Fig. 3.35: Inhibition of condensation by
addressing different pathways via small mo- harvested, stained with the cell viability dye
lecules. Pictures were taken 5 and 24 hours 7-aminoactinomycin D (7-AAD) and measured
after condensation induction under influence in the flow cytometer. The dye cannot enter
of denoted molecules. Representative images; intact cell membranes so that viable cells ap-
N=2 peared 7-AAD negative and dying or dead cells
appear positive. The cells were all viable (see
figure Appendix 5.3) except in the positive "kill control" showing functionality of the dye.
3.3.3 Interaction of inductive condensates with epithelial cells
It is generally accepted that an inductive mesenchyme induces odontogenic cell fate in
dental and non-dental epithelium (Duailibi et al., 2004; Edward J. Kollar et al.,
1970a; Mina et al., 1987; Ohazama et al., 2004). Mammoto et al. show that mechanical
compaction of mesenchymal cells induces their cell fate determination and equip them
with inductive abilities (T. Mammoto, A. Mammoto, Torisawa et al., 2011). In the in
vitro condensations, described here, a self-organized mechanical compaction was induced
in the cells, with implications of the cytoskeleton rearrangement (see chapter 3.3.2), an
activation of MAPK signaling pathway and a striking activation of transcription factor
activity (see chapters 3.3.1 and 3.2.3). To show that the produced condensate from adult
dental pulp cells is indeed capable to induce interaction and differentiation in epithelial
cells, a co-culture was performed with keratinocytes from different tissue sources (skin or
gingiva).
72 Chapter 3 Results
Co-culture
After 24 hours of condensation, the dense DPC aggregates were co-cultured in ultra-low
attachment wells with suspended adult oral keratinocytes. After additional 18 hours they
showed to have been attracted by the mesenchyme (3.36 b), enwrapped it (3.36 b) and
formed densely packed aggregates consisting of the two cell types (3.36 d and e). The
keratinocytes themselves did not condense (3.36 a).
(a) (b) (c)
(d) (e)
Figure 3.36: Interaction of gingival keratinocytes and DPC condensates in vitro in ultra-low
attachment culture over time. (a) Oral keratinocytes only. (b) Co-culture of oral keratinocytes
and DPC condensate (18h). (c) Co-culture of oral keratinocytes and DPC condensate (24h).
(d) Co-culture of oral keratinocytes and DPC condensate (4 weeks). (e) Co-culture of oral
keratinocytes and DPC condensate (4 weeks) at higher magnification.
To distinguish between the two cell types in live microscopy, the dental pulp cells
were transduced with the pLL3.7 and constitutively expressed eGFP. The skin-derived
keratinocytes were labeled with CellTracker™ Red which retains in the cytoplasm after
labeling. Figure 3.37 shows the well distinguishable cell types under fluorescence microscope.
On the left side (a), the cells were cultured for three days in monolayer. A strong dye
intensity in the green (eGFP) and in the red (CellTrackerRed) channel was observed. Also,
the different cellular morphologies of the dental pulp cells and the gingival keratinocytes
were obvious. Three days after addition of suspended and red-labeled keratinocytes to
a mesenchymal condensate from green-fluorescent dental pulp cells, the keratinocytes
attached to the surface of the aggregate (b). Four weeks later, the two cell types were
tightly grown together with a proportion of red keratinocyte inside the condensate, or, at
least underneath the green DPCs (c).
3.3 Functional analyses of the in vitro model for mesenchymal condensation 73
(a) (b) (c)
Figure 3.37: Co-culture of red and green labeled cells. (a) eGFP expressing DPCs in
monolayer co-culture with red-labeled gingival keratinocytes after three days. (b) 3D co-culture
after three days of red (keratinocytes) and green (DPCs in condensate) cells. (c) 3D co-culture
after four weeks of red (keratinocytes) and green (DPCs in condensate) cells.
Histological appearance of condensates
To get an insight into the cellular arrangement inside the cultured organoids, DPC con-
densates were cryosectioned, and their histological properties were analyzed. At day
four, the DPCs in the condensate showed a homogenous distribution with high cellular-
ity. Vimentin is a cytoskeletal intermediate filament of mesenchymal cells. The whole
condensate appeared Vimentin positive, evidencing the high cellular density. The two
extracellular glycoproteins Tenascin C and Fibronectin also exhibit a high fluorescence
intensity with a homogenous distribution. Collagen type I and Collagen type IV showed to
be expressed in distinct zones. The outer border showed no collagen expression. Especially
from the collagen type I pattern, the process of condensation and tissue folding or rolling
is traceable. A marker of direct mechanical stress, transmitted through cell-ECM and
cell-cell-junctions, is the localization of ß-catenin. Interestingly, here the ß-catenin signal
completely overlapped with the nuclear DAPI signal (see figure 3.38).
Figure 3.38: Histology of 4-day condensates of DPCs alone: Depicted are an overview stain
with Hematoxylin/Eosin (H&E), and immunohistological stainings against Vimentin (VIM),
Tenascin C (TNC), Fibronectin (FN), Collagen type I (COL1), Collagen type 4 (COL4) and
ß-catenin (CTNNB1). The ß-catenin staining was counterstained with DAPI to visualize
nuclear localization.
74 Chapter 3 Results
With the addition of keratinocytes, 24 hours after condensation of the DPCs (see figure
3.39), the distribution of nuclei was still homogeneous, but there were eosin-stained pink
areas which resemble either acidophilic proteins of the cytoplasm or collagens. Sparsely
distributed pink dots in the aggregate (H&E) co-localized with the cytokeratin-positive
cells. The pink area in the center co-localized with the collagen type I staining. The
collagen type IV signal was predominantly seen at the outer border of the condensate.
The structural proteins Vimentin, Fibronectin, and Tenascin C gave a homogenous bright
signal. As a result of the preparation of the condensate for analysis, the keratinocyte outer
layer detached from the condensate. The cytokeratin antibodies against CK18, CK8/18,
CK14, and CK15 clearly mark the keratinocytes. They are much bigger in size than
the mesenchymal cells. Some epithelial cells found their way inside the mesenchymal
condensate, and it appears as if they migrate collectively. Interestingly, the cytokeratin
14 signal was more prominent than the signals of the other cytokeratins and it partially
overlapped with the mesenchyme.
Figure 3.39: Histology of 4 day co-culture condensates. DPCs were condensated for 24
hours, skin keratinocytes were added and co-cultured for three days. Depicted is an overview
stain with Hematoxylin/Eosin (H& E), and immunohistological stainings against Fibronectin
(FN), Cytokeratin 19 (CK19), Collagen type 4 (COL4), Vimentin (VIM), Cytokeratin 8 or 18
(CK8/18), Tenascin C (TNC), Cytokeratin 14 (CK14), Collagen type I (COL1) and Cytokeratin
15 (CK15). Blue color represents nuclear counterstaining with DAPI.
When the DPCs and skin keratinocytes were co-cultured for two weeks in spheroids,
clusters of attached keratinocytes at the surface of the condensate became visible. Concord-
ant with the mRNA data, the collagen type I expression lowered and was only evident in the
center. The amount of Cytokeratin 8/18/14/15/19 positive cells inside of the condensate
increased. No obvious change in Tenascin C, Fibronectin or Vimentin expression was
observed (see figure 3.40).
3.3 Functional analyses of the in vitro model for mesenchymal condensation 75
Figure 3.40: Histology of 2 week co-culture condensates. DPCs were condensed for 24 hours,
skin keratinocytes were added and co-cultured for thirteen days. Depicted is an overview
stain with Hematoxylin/Eosin (H& E), and immunohistological stainings against Fibronectin
(FN), Cytokeratin 19 (CK19), Collagen type 4 (COL4), Vimentin (VIM), Cytokeratin 8 or 18
(CK8/18), Tenascin C (TNC), Cytokeratin 14 (CK14), Collagen type I (COL1) and Cytokeratin
15 (CK15). Blue color represents nuclear counterstaining with DAPI.
For the next long-term experiments, the mesenchymal DPC condensates were co-cultured
with gingival keratinocytes. The keratinocytes from the oral epithelium interacted much
more with the condensates than the skin keratinocytes, observed during the cell culture
handling (data not shown). Here, they formed a sheath around the mesenchymal cells, as
indicated in figure 3.36 (c) and (d). This sheath also became visible in the histological
section (see figure 3.41). The epithelial sheath grew inwards into the mesenchyme. From
the fluorescence stainings with chosen antibodies against either mesenchymal proportion
(Collagen type I, Collagen type IV, Vimentin) or epithelial tissue (Cytokeratin 15, Cy-
tokeratin 8/18) can be concluded that the epithelium not only lined the outer border,
but also formed a cluster of cells inside the mesenchymal condensate. The center of this
cluster was acellular, but rich in extracellular matrix, which was not composed of collagen
type I or type IV as seen in the H&E overview and corresponding stainings. The inner
epithelial cells, which lined the border to the mesenchyme, arranged columnar, indicating
the formation of an epithelial basal layer (see arrowhead in bottom right high magnification
of VIM staining).
76 Chapter 3 Results
Figure 3.41: Histology of 4 week co-culture condensates. Gingival keratinocytes were added
to 24 hour-condensate prior to 4 weeks of co-culture. Hematoxylin/Eosin (H&E) and immun-
ohistological stainings against Collagen type I (COL1), Collagen type 4 (COL4), Vimentin
(VIM) as well as against epithelial cytokeratins Cytokeratin 15 (CK15) and Cytokeratin 8/18
(CK8/18) are depicted. Inner epithelial cells arrange columnar (see arrowhead).Blue color
represents nuclear counterstaining with DAPI.
Figure 3.42 shows that after ten weeks of co-culture of gingival keratinocytes with the DPC
condensates, the organoids formed calcified "cysts" as indicated by Alizarin Red staining.
Still, the cells were alive, evidenced by the sharp nuclear signal in Hematoxylin/Eosin
staining and from the DAPI signal (bottom right). The mesenchymal cells were marked by
3.3 Functional analyses of the in vitro model for mesenchymal condensation 77
the Vimentin signal. They were located between the mostly acellular round structures and
deposited collagen type IV. Only very few CK8/18 positive cells were detected, and the
blue smear inside of the round structures indicated dead cell remnants, presumably from
the epithelial proportion, which formed these structures.
Figure 3.42: Co-cultured aggregates were long-term cultures for 7 (upper left) or 10 weeks.
The sections were stained with Hematoxylin/Eosin (upper row), Alizarin Red for calcification
(lower left) or Rhodamine/Hoechst (lower right).
In conclusion, the histological analysis reveals interaction of the produced mesenchymal
DPC condensates with epithelial cells from skin epidermis or gingiva. Invagination-like
structures, cytodifferentiation, and matrix deposition are exhibited.

CHAPTER 4
Discussion
4.1 Characteristics of dental pulp cells
In recent years, many successful attempts for whole-tooth bioengineering have been made
in animal models. The majority of them include usage of embryonic or postnatal dental
cells. For human patients, this cell source is not applicable, and it is common sense
that only autologous cells of the patient are acceptable in ethical terms as well as for
immunological issues. For dental regenerative therapies, there are two conceivable cell
sources: mesenchymal stromal cells (MSCs) from the dental pulp or from other adult
tissues of a patient (bone marrow, skin, fat, hair). Although it has been shown that dental
pulp (stem) cells (DPCs) meet the criteria defined for mesenchymal stromal cells by the
International Society for Cellular Therapy, there are differences. Like MSCs, DPCs grow
plastic adherent in vitro and express the surface markers CD105, CD90, CD106 and lack
CD45, CD34, CD19 and CD14. Additionally, as also shown for MSCs from different tissue
sources (Krampera et al., 2013; Mafi et al., 2011; Rahman et al., 2014), they express
CD44 (hyaluronic acid receptor), CD13, and CD146.
The isolation procedure for the cells that were used in this thesis did not include a specific
sorting step. Therefore, it is likely that the obtained population is, to a certain extent,
a mixed population. The presence of endothelial cells is excluded by CD31 negativity,
as well as the presence of hematopoietic cells (CD34-, CD45-). Nevertheless, it is known,
that besides dental pulp (stem) cells, also fibroblasts and pericytes grow plastic adherent.
In recent times, the identity of the MSC was questioned (Hematti, 2012). The bona
fide MSC marker has not been identified, a panel of markers and characteristics rather
identifies the cells. Interestingly, all these characteristics are also fulfilled by fibroblasts and
pericytes. In the dental pulp, fibroblasts are dispersed in the fibrous tissue and maintain
the extracellular matrix by constant turnover. Pericytes are located on the surrounding of
vessels and regulate blood flow through contractility. Both are marked by the expression
of CD146 and STRO-1 and as stated by Caplan et al., 2011, pericytes dedifferentiate
to MSCs upon injury or inflammation, proliferate, and become immunomodulatory as
marked by the expression of various secreted factors. Immunomagnetic sorting of CD146+
cells from the dental pulp that were localized in situ at the outer vessel walls beforehand,
resulted in a population of dental pulp stem cells that were capable to produce dentin
and dental pulp fibrous tissue in vivo. In the exact same manner CD146+ cells from
bone marrow, produced ectopic bone after transplantation in vivo (Songtao Shi et al.,
2003). Taken together, these experiments indicate, that perivascular MSCs or DPCs turn
into fibroblasts and specialized differentiated cells in vivo, to maintain or regenerate the
79
80 Chapter 4 Discussion
respective tissue and that they can switch between the cell fate.
Nonetheless, this experiment also emphasizes the differences between stromal cells from
different tissue source. The MSCs from bone marrow formed ectopic bone whereas the
DPCs formed dentin. Even after 20-30 population doublings of ex vivo monolayer culture,
the cells formed bone or dentin, respectively, upon implantantation (Gronthos et al.,
2000). That existence of a cellular memory is based on methylation patterns of promoter
regions of cell-type specific genes and has been shown by Kadler, 2017.
When discussing the roles and differences of MSCs from various tissues, their ontogenic
origin should be taken into consideration. Dental pulp cells arise from the neuroectodermal
lineage. Unlike MSCs from e.g., bone marrow, they do not derive from the mesodermal
layer of the gastrulated embryo. Other tissues, deriving from neuroectoderm, comprise
the craniofacial structures such as facial cartilage, the retina, melanocytes, ganglia, para-
follicular cells of the thyroid, the adrenal medulla, the aorticopulmonary septum of the
prenatal heart and lung, and last but not least, the brain. Therefore, it is not surprising,
that during this work, differences between DPCs and MSCs could be shown. Firstly, the
DPCs in tissue culture are much smaller, which becomes evident under the microscope. A
confluent layer of DPCs gave about three times more cells in yield than a confluent layer of
bmMSCs. The population doubling time was about 25 hours for the DPCs. Bone marrow
MSCs have doubling time of approximately 70 hours in early passages and more than 100
hours in late passages (Jin et al., 2013; X. Li et al., 2014). Interestingly, FACS analysis
of the marker panel revealed that the bone marrow MSCs contain a significantly lower
percentage of CD146+ cells than the isolated dental pulp cells (see Appendix figure 5.1 A),
indicating that many perivascular-derived mesenchymal stromal cells are abundant in the
isolated DPC populations.
The defining criteria for MSCs include multipotency. Therefore, in vitro differentiation
assays were performed to adipogenic, osteogenic and neurogenic lineage. The DPCs
were poorly able to differentiate into adipocytes. Adipogenic differentiation is marked
by deposition of large lipid vesicles in the cytoplasm. The functionality of the protocol
was evidenced by bmMSC differentiation (see Appendix figure 5.1 B). The dental pulp
cells deposited only small vesicles and gene expression analysis of adipogenic marker genes
reveals extremely low abundance of LPL and FABP4 mRNA. This result is in concordance
with literature (Struys et al., 2011). One reason could be that DPCs do not undergo
adipogenesis on any account in their lifetime in vivo. Unlike bone marrow MSCs, that
differentiate into adipocytes in the aging marrow, the aged DPC rather deposits increased
amounts of dentin in the aging pulp. By size measurement of the remaining pulpal tissue,
which narrows with age due to increased deposition of dentin, an age-estimation of an
individuum is possible. The osteogenic differentiation of DPCs (and bmMSCs) is successful.
Alizarin Red staining indicates a strong deposition of calcified matrix and the marker
genes Osteopontin (OPN ) and Osteocalcin (OPN ) are upregulated. Published protocols
for osteogenic and odontogenic in vitro differentiation, are overlapping or identical. In
addition, OPN and OCN are also expressed by odontoblasts and are a composite of dentin
(Papagerakis et al., 2002). Therefore, it cannot be determined here, if the observed
4.1 Characteristics of dental pulp cells 81
differentiation corresponds to the osteogenic or odontogenic lineage.
Since dental pulp cells originate from the neuroectoderm, they have a close ontogenic
relationship to neurons. Therefore, it was tested whether the isolated DPCs could differ-
entiate into neuronal cells under appropriate culture conditions. In neuronal induction
media, the cells elongate and enlarge. After two weeks of culture, they show an increased
expression of neuron-specific class III ß-Tubulin (TUJ1 ) as well as of Nestin (NES), which
was detected by immunocytochemistry and qRT-PCR. The DPCs’ tendency to differentiate
into neuronal-like cells has been published before, and their potential use in regenerative
therapies for neuronal disease is evaluated and passionately discussed among experts (Mead
et al., 2016). A clinical study phase I has been started in 2016 in Australia that evaluates
the use of DPCs in stroke patients (Nagpal et al., 2016).
Especially for tissue engineering purposes, a sufficient amount of cells is needed to build
an organoid. Expansion of cell number is achieved by monolayer culture when dedifferen-
tiated cells proliferate. To evaluate if dedifferentiated DPCs can be "reprogrammed" to
be odontogenic, the DPCs were expanded before use in the experiment. A set of marker
genes was selected, that are expressed during tooth initiation, tooth development, ECM
regulation, stem cell phenotype, and regeneration in the dental pulp at different stages
of development. By qRT-PCR, their basal expression in the used DPCs was monitored
during the monolayer culture and was compared to bmMSCs. The gene expression was of
exceptional interest for following induction of in vitro condensation and odontogenesis.
In literature, BMP7 is described to be mainly expressed in the dental epithelium
throughout tooth organogenesis and is only detected in the mesenchymal part during the
secretory stage in mature odontoblasts (http://bite-it.helsinki.fi/; Zouvelou et al.,
2009).When isolating the cells from the pulpal specimen, a mixture of cells is obtained,
containing stem cells, odontoblasts, fibroblasts, and pericytes (see figure 3.1). Since only
the odontoblasts are supposed to express BMP7 in the adult tissue, the relative expression
level of the cell mixture is rather low. Furthermore, in culture, these odontoblasts lose their
specific phenotype, by withdrawal from their natural environment. It is conceivable that
the proportion of odontoblasts in the culture dedifferentiates to mesenchymal progenitors
and acquire the characteristics of the DPCs and fibroblasts in means of morphology and
gene expression.
The TGFß1 expression is comparably higher. It is described to be expressed by secretory
odontoblasts as well as fibroblasts of the dental pulp (Vaahtokari et al., 1991). TGFß1
is known to be a regulator of growth, cytodifferentiation, and inducer of matrix deposition.
TGFß1 and TGFß2 regulate DSPP and DMP1 expression by odontoblasts as well as collagen
expression by fibroblasts in the dental pulp (Chan et al., 2005; Niwa et al., 2018). In
numerous studies, TGFß1 has been shown to maintain the mesenchymal phenotype in order
to prevent inappropriate epithelialization during tissue healing and homeostasis (Mincione
et al., 2008; Zeisberg et al., 2003; Zhu et al., 2010). Here, the TGFß1 expression declines
upon DPC monolayer culture, but compared to the BMP7 expression, it is still high. This
observation suggests that the cells still actively maintain the mesenchymal phenotype, and
presumably, the decline is a result of the dedifferentiation of odontoblasts again, that lose
82 Chapter 4 Discussion
their specific secretory phenotype in culture. Notably, the BMP family members, especially
BMP7, shift the balance of epithelial-mesenchymal transition to the epithelial phenotype
(Zeisberg et al., 2003). Therefore, the complete loss of BMP7 in the cultures is in line
with the suggested active maintenance of the mesenchymal phenotype by the DPCs in
monolayer culture.
The Activin ligands exist as dimers of ßA-ßA subunits (Activin A), of ßB-ßB subunits
(Activin B) and ßA-ßB subunits (Activin AB). Furthermore, the ßA or ßB subunit can
dimerize the 𝛼-subunit to form Inhibin A or Inhibin B, respectively. Here, the transcript
of the ßA-subunit is measured (INHBA). As seen in the whole transcriptome analysis, no
𝛼-subunit is expressed in the used DPCs. Thus, the ßA subunit, measured here, will only
form Activins, not Inhibins. It is known that Activin is expressed in pulpal cells, but their
role in the mature tissue has not been elucidated yet (Goldberg et al., 2004). Here, the
INHBA expression of the freshly isolated DPCs is comparably low and declines over the
culture period. Activins also play a role during early odontogenesis. Ferguson et al.,
1998 describe Activin A as "an early mesenchymal signal" expressed in the condensing
mesenchyme of the tooth germ, and mice lacking the Inhba gene exhibit a molar arrest at
bud stage.
Fibroblast growth factor (FGF2) is described to induce mesenchymal condensation in
feather, in renal, and limb development(Perantoni et al., 1995; Song et al., 2004; Su
et al., 2014). In these tissue types, the protein FGF2 can be found in the epithelium as
well as the condensing mesenchyme. Whether both tissue types express FGF2, or if it is
only the epithelial signal diffusing to the mesenchyme, is not clear. Since in this study
mesenchymal condensation was in focus, this molecule was selected in the marker panel.
The expression of FGF2 in the dental pulp cells is rather low and does not change over
the monolayer culture period.
The dentin matrix contains numerous growth factors that are deposited by the odontoblasts
during dentine formation. They function in activation of odontoblasts upon injury. When
they are released from the dentin matrix, the respective receptors on the odontoblasts are
activated, and the regenerative matrix will be produced. One of the deposited factors is
hepatocyte growth factor (HGF). Just as FGF2, HGF binds to a tyrosine-kinase receptor.
Upon ligand binding, different pathways can be activated intracellularly, most prominently
the Ras-MAPK pathway. During development, HGF and its corresponding receptor c-MET,
regulate important mesenchymal-epithelial interactions. It serves as an important growth
factor for epithelial cells and is expressed in the mesenchyme. This role becomes evident
when HGF-antisense-treated tooth germs exhibited abnormal tooth morphogenesis, an
"inside-out"-phenotype with dentin around enamel (M. Tabata et al., 1996). In adult
tissues, HGF synergizes with nerve growth factor (NGF) and acts as a survival signal for
neurons (Birchmeier et al., 1998). Putatively, the expression of HGF in the isolated
DPCs is explained by their neuron-supporting function, since the dental pulp is highly
innervated. It is not surprising that the cells down-modulate the HGF expression as a
result of the missing tissue interaction.
Collagen type I is the major organic component of the intertubular dentin (90 %), whereas
the peritubular dentin is non-collagenous. The odontoblasts secrete collagen type I, as well
as proteoglycans, that are implicated in collagen fibril formation (e.g., decorin, biglycan,
4.1 Characteristics of dental pulp cells 83
fibromodulin). The mineralization process is initiated by the deposition of DPP, a product
of DSPP, which attracts calcium ions. In the dental pulp, the fibrous tissue of the tooth,
collagen type I is also the predominant ECM molecule. With the high content of glyc-
osaminoglycans, this tissue is hydrated and not mineralized. The isolated DPCs exhibit
enormously high expression of collagen type I, that drops in monolayer culture. Still,
the expression is, compared to other analyzed genes, very high. The DSPP expression
of the freshly isolated cells, which is supposed to come from the odontoblast fraction, is
completely lost due to the dedifferentiation. These results are concordant with the decline
of TGFß1 expression since TGFß1 is described to be a positive regulator of DSPP and
COL1A1 (Niwa et al., 2018).
Two mesenchymal transcription factors were chosen for the panel, MSX1, and PAX9.
During early tooth development, they act synergistically to induce tooth initiation. As
depicted in figure 1.18 they play a role in every developmental stage. The prime odon-
togenic potential lies in the dental epithelium. With the expression of FGFs and BMPs
by the epithelium and subsequent induction of PAX9 and MSX1 in the mesenchyme, the
odontogenic potential then shifts to the dental mesenchyme. In homozygous PAX9- as
well as in MSX1-mutant mice the mesenchyme fails to condense normally (Peters et al.,
1999). Interestingly, in murine molar development, PAX9 is constantly expressed during all
developmental stages of the dental mesenchyme. Before any morphological manifestation
of tooth development, PAX9 is expressed at the prospective sites of tooth formation (but
not in between the individual site for each tooth) (Neubüser et al., 1997). This finding,
together with tooth agenesis upon Pax9 loss, concludes that PAX9 is a marker for dental
mesenchyme which is needed to transduce the incoming epithelial signals to achieve tooth
formation. It is therefore of particular interest that the PAX9 expression is not lost here
during monolayer culture, suggesting that the used cells, even if dedifferentiated, are
capable to respond to appropriate signals with dental differentiation.
The expression of MSX1 has been described to be unique for human dental pulp cells when
compared to other cell types from various tissues (Fujii et al., 2015). Therefore, this marker
has been proposed to identify dental pulp stem cells. Here, MSX1 expression is lower
than PAX9 expression and upon monolayer culture, the expression significantly declines.
Interestingly, PAX9 is not and MSX1 is barely detectable in bmMSCs. Therefore, it can
be concluded, that these markers are specific and that dedifferentiated DPCs are different
from bmMSCs. The PAX9 expression, which does not change upon dedifferentiation may
be a result of the epigenetic memory of that cell type. To verifiy that, methylation analysis
of the promoter should be conducted. The expression of both transcription factors is
described, but their role in postnatal teeth is unknown (Fujii et al., 2015).
In this context, it should be mentioned that in hair follicle morphogenesis the inductive
signal is supposed to come from the mesenchyme and the signaling between the mesen-
chymal condensate and the epithelium dictates the morphogenesis of both cell populations
and orchestrates the formation of the hair follicle (Schmidt-Ullrich et al., 2005). Both
skin appendages have the same evolutionary origin, share similar developmental processes,
and are the result of epithelial-mesenchymal tissue interactions. It is reasonable that also
in tooth development the mesenchyme bears the inductive signal to orchestrate tooth
development.
84 Chapter 4 Discussion
To summarize these findings, it can be concluded that the isolated human DPCs rep-
resent a mixture of dental cells of ectomesenchymal origin, i.e., odontoblasts, fibroblasts,
and perivascular cells. The cell population, as a whole, dedifferentiates to a mesenchymal
progenitor phenotype, regarding gene expression and morphology. The dedifferentiated
DPCs are different from cultured bone marrow MSCs, as evidenced by the expression of
PAX9 and MSX1, their differentiation capacity and their proliferative behavior. Addi-
tionally, the elevated PAX9 and MSX1 expression, compared to bmMSCs, suggests an
odontogenic memory of the cells.
4.2 Culture model for tooth development
The observation that DPCs seem to be more directed towards odontogenic lineage, com-
pared to bmMSCs, is concordant with previous studies (Hosseini et al., 2018; Tatullo
et al., 2015). Tissue-engineered structures, generated from DPCs and synthetic scaffolds
(e.g., PGLA, HA, TCP), resemble organized tubuli-traversed dentin or reparative dentin
(Hosseini et al., 2018). DPCs are also used to regenerate lost dental pulp tissue, following
root canal treatments. A recent attempt has been published by Athirasala et al., 2017,
where pre-vascularized hydrogel pulp-like tissue constructs were engineered ex vivo and
dispensed in root canals. A pilot clinical study in humans was completed, which shows
safety and efficacy of usage of autologously transplanted dental pulp cells for total pulp
regeneration (Nakashima et al., 2017). Thus, in those numerous investigations, adult
dental pulp cells have been shown to be an excellent cell source for dentin reconstruction
and regeneration.
Nevertheless, until now, no study has been published, showing that adult human dental pulp
cells give rise to whole bioengineered teeth. Moreover, for adult mouse DPCs it has been
shown, that after 24 hours of in vitro culture, they lost their tooth forming capability when
recombined with embryonic epithelium and subsequent in vivo transplantation (Zheng
et al., 2016).
The underlying hypothesis is that the inductive ability of the DPCs is locked upon mono-
layer culture due to dedifferentiation. A "liberation" of this inductivity is thought to be
achieved by introducing mechanical cues on the cell, that are sensed by the cytoskeleton
and drive cell fate switching and subsequent differentiation as proposed by T. Mammoto,
A. Mammoto and Ingber, 2013. The setup to achieve this aspect of this thesis was to
recapitulate mesenchymal condensation in vitro since it represents a crucial step in tooth
development.
Mesenchymal condensation is well-studied in the context of skeletal development. Never-
theless, in many other organs, such as skin appendages, kidney, and lung, mesenchymal
condensation also constitutes a central role as being one of the first developmental steps.
Since these organs are thought be "epithelium-derived", past research was mainly focussed
on epithelial dental stem cells and their developmental stages. At the site of presumptive
tooth formation at around week 5 of human fetal development, the ectomesenchymal
cells undergo mesenchymal condensation, forming the tooth bud. It is now accepted
that after that process, the odontogenic fate lies in the mesenchymal condensate. Long
4.2 Culture model for tooth development 85
forgotten, Kollar and Baird already published between 1969 and 1970 three consecutive
papers, in which they evidenced this: When an embryonic murine dental mesenchymal
condensate was recombined with non-dental epithelium, i.e., from foot or snout pad, teeth
formed, whose morphologic appearance was specific to the tooth germ from which the
mesenchyme was harvested (Edward J. Kollar et al., 1969; Edward J. Kollar et al.,
1970a; Edward J Kollar et al., 1970b). Only later, the underlying mechanisms were
elucidated: The acquired inductive property is marked by the expression of morphogens
and inhibitors downstream of tissue-specific transcription factors. These transcription
factors are activated by cytoskeletal changes as a result of mechanical signals of the con-
densation. The cytoskeleton ensures resilience to the increasing pressure. Simultaneously,
the rearrangement of the cytoskeleton, which is directly coupled to intracellular domains
of ECM receptors and nuclear lamina structures that transduce the mechanical signal into
the nucleus, induces a change in transcriptional and enzymatic activity. The process of
cellular attraction and motility which is the motor of condensation is resembled in the
spheroid culture, here. Under standard culture conditions, the cells adhere to the plastic
and flatten. During ultra-low attachment culture, the cells seek for contact, since that is
their nature as mesenchymal cells in three-dimensional connective tissues. Thereby, they
express chemokines, show high motility, and adhere to each other via cell-cell junctions.
Traction forces are applied by and to the connected cells and long F-actin bridges form
between them in vivo (see figure 4.1)(Ray et al., 2015).
Long cellular processes between the DPCs in ultra-
low attachment culture can be observed (see fig-
ure 3.13). The data from the RNA Seq show, that
gamma-actin as well as beta-actin are highly abund-
ant in the cells. Furthermore, from the CNA labeling
experiment, it can be excluded, that the long pro-
cesses are composed of extracellular collagen fibers
(see figure 3.30). Therefore, it is highly likely, that
the "arms" consist of actin-bridges in this condensa- Fig. 4.1: Mesenchymal condensa-
tion model. To address, whether the condensation is tion during chick columella develop-
indeed actin-mediated, Cytochalasin D (CytD) was ment exhibits strong F-actin bridges
administered to the in vitro condensing cells. CytD between the condensing cells (arrow-
is a mycotoxin that inhibits F-actin polymerization heads) (Ray et al., 2015).
and induces depolymerization of F-actin (Casella et al., 1981) and thereby block the
formation of cortical actin fibers. Indeed, CytD entirely blocks cell-cell-attachment and
condensation at a concentration at 2.5 µM. At a 10-fold lower concentration, the inhibiting
effect was still visible, resulting in a loose aggregation of cells, that did not form actin
bridges. The other inbititors pertussis toxin (PTX), an inhibitor G protein-coupled sig-
naling, and LY294002, an inhibitor of phosphoinositide 3-kinase (PI3K)–AKT signaling
pathway, did not affect the condensation process of DPCs. To my knowledge, it is the
first time, that the stage of actin bridge-forming dental pulp cells upon condensation is
resembled in vitro.
All experiments were conducted with cells from individual donors. The primary state of
86 Chapter 4 Discussion
the cells from different donors is not negligible and had an influence on the kinetics of
condensation regarding the morphology as well as gene expression. As outlined above,
a specific marker panel of gene expression was chosen to monitor the condensation and
putative differentiation. The two key transcription factors, PAX9 and MSX1, exhibited
a surprising expression profile. Within the first 24 hours of condensation, they are sig-
nificantly downregulated. In concordance with that observation, BMP4 expression also
decreases upon the first 24 hours of condensation compared to monolayer cells, since MSX1
and PAX9 are known to be synergistic regulators of the human BMP4 promoter (Liang
et al., 2012). Expression of this trio rises in the course of later culture time and their
expression level is re-established to the initial level of freshly isolated dental pulp cells.
Three more members of the TGFß superfamily, TGFß1, INHBA, and BMP7, are upregu-
lated after four days.
The loss of BMP7 expression upon 2D monolayer culture is thought to be the reason for
the loss of odontogenic potential of the dedifferentiated adult dental pulp cells. It has
been shown that human fetal dental pulp cells (FDPC) express BMP7 approximately
50-fold higher than adult dental pulp cells (ADPC). When FDPC were co-cultured with
oral keratinocytes, an amelogenic differentiation could be observed. Contrarily, ADPCs
were not able to induce the amelogenic differentiation. With the addition of BMP7 to
the culture, also the adult cells were able to transform the oral keratinocytes (Gao et al.,
2015). This effect was even more enhanced with the combined administration of BMP7
and epiregulin, a member of epidermal growth factors, that activates MAPK signaling.
Thus, the upregulation of BMP7 in the mesenchymal dental condensates can be seen here
as the acquisition of inductive abilities.
As described above, TGFß1 is considered to be a marker for differentiation, positively
regulating DSPP and COL1A1 expression (Niwa et al., 2018). Since the expression of all
these three genes increases during long-term culture (four weeks) in ultra-low attachment,
it can be assumed here that the DPCs undergo odontoblastic differentiation. Collagen
type I and Dentin Sialophophoprotein expression runs in parallel with TGFß ligands,
emphasizing their regulatory relationship. Interestingly, Activin expression exhibits a steep
rise between 24 hours and four days, and stagnates afterwards. This is in concordance
with the above-described role of Activins during the early phase of tooth development.
Taken together, it can be concluded, that a suitable culture setup was established to
allow production of singular dental mesenchymal condensates of 500 µm in diameter in
vitro to resemble the human in vivo equivalent during organogenesis. In this culture model,
a self-orchestrated condensation of the DPCs was allowed to induce the mechanotrans-
ductional forces on the cells which take place during condensational processes in vivo.
During the condensation in vitro, the marker genes BMP7, TGFß1, INHBA, COL1A1 and
DSPP display a congruent transcript expression profile. Long-term culture of the condens-
ates induces the re-induction of genes that were subject to dedifferentiation, indicating a
differentiation within the condensate.
4.3 Early phase of condensation 87
4.3 Early phase of condensation
Although, the condensation seems to be the central step in organogenesis, so far no data
on the molecular events that occur during human odontogenesis have been published.
Much information about skeletal mesenchymal condensation is available, and it is valuable
knowledge that arose from those numerous studies to understand the basic principles and
pathways. Nevertheless, the exact molecules that are part of the condensation signalosome
and their spatiotemporal expression differ from organ to organ. Though, when discussing
induction of replacement teeth as regenerative therapies, these factors should be known,
and their actions understood. For odontogenesis, one approach to resemble mesenchymal
condensation in vitro, is promising, but until now, experiments were only conducted with
mouse DPCs (T. Mammoto, A. Mammoto, Jiang et al., 2015). The method is based
on high-density cell cultures of murine DPCs on fibronectin islands. Fibronectin is the
initiating and guiding ECM molecule during mesenchymal condensation (Hall et al., 2000)
and was therefore well-chosen. Similarities to the in vivo condensation are evidenced by
gene expression analysis. Nevertheless, when it comes to the handling of the condensed
cells, e.g., for transplantational purposes, this model seems to be disadvantageous.
One goal of this thesis was to validate that the presented method to induce condensation
resembles the in vivo process, and can also be considered as a model to study dental organo-
genesis. A comprehensive analysis by Next Generation Sequencing was performed to identify
the expression profile of condensing DPCs in vitro after six hours in low-attachment culture.
By analysis of the expression of all transcripts, it becomes evident that the dediffer-
entiated DPCs are substantially different from bmMSCs. In the heat map, the monolayer
population did not form a cluster (see figure 3.17). Furthermore, after six hours of DPC
condensation, the donor-specific expression intensities overruled the culture condition-
specific expression, characterized by the donor-clustering. In contrast, the bmMSCs did
not cluster donor-specific but condition-specific(ML and 24 hours condensation). It is
reasonable that this difference is explained by the longer period of condensation (6h vs. 24
h) and raises the question whether a transcriptome analysis of DPCs condensates after a
later time point gives rise to the same effect.
Based on knowledge about precartilage condensation, the mechanochemical principles
of condensation can be easily explained (Chimal-Monroy et al., 2003; Hall et al.,
2000). Fibronectin (FN) is highly abundant in mesenchymal tissues. Upon initiation of
mesenchymal condensation by TGFß1, TGFß2 and TGFß3 ligands, the mesenchymal cells
express integrins on their surface, which bind to FN. They use the FN fibers as "guiding
ropes" and migrate along them towards the TGFß morphogen gradient. The cells react on
the FN binding on the integrin receptors with cytoskeletal contraction via actin bundles
and myosin II. Thereby, they generate a tensile strain on the ECM as well as on further
cells connected to the ECM. Along with that, the activation of the integrin signaling
induces upregulation of FN itself, of collagens (COL), tenascin (TNC), and syndecans
(SDC) as well as of the cell-cell adhesion molecules N-CAM and N-Cadherin. The cells
form tight cell-cell contacts, which in turn is in a positive feedback loop with the ECM
88 Chapter 4 Discussion
proteins. The secreted ECM builds a "cage" around the condensed cells. This step is
referred to as "boundary setting" of the mesenchymal condensate. The entrapment of the
cells is of importance: The gathered cells stay in spatial proximity and unitely form the
organ by producing morphogen gradients that drive organogenesis. An important aspect
in developmental processes is proper fine-tuning. Here, a tensile strain exerted by and
on the cells is needed to generate the condensate. RhoA is a small GTPase regulating
actin assembly and stress fiber formation. It is reported, that suppression of RhoA is
indispensable for the expression of organ-fate specific transcription factors and functional
tissue properties. By release of the tensile strain via chemical suppression of RhoA, in
chondrogenic condensations in vitro, the produced cartilage had improved characterist-
ics compared to untreated pellet cultures. It was of less compact structure, but with
increased GAG and collagen content, as well as an enhanced diffusity for growth factors
was shown, which was important for the subsequent differentiation (K.-C. Wang et al.,
2018). The inhibition of RhoA activity is also seen in vivo during mesenchymal condens-
ation of murine tooth development (T. Mammoto, A. Mammoto, Torisawa et al., 2011).
The RNA Seq data suggest a RhoA inactivation in my DPC condensations. In both
conditions (ML and 6h cond) RhoA is among the highly abundant transcripts, and it
is known that stress fibers, induced by active RhoA pathway, are very prominent in
cells grown on rigid substrates, such as plastic dishes (Burridge et al., 2016). Never-
theless, RhoA functions as a molecular switch, cycling between an active GTP- and an
inactive GDP-bound state, independent from the total RhoA abundance. This cycling
is mediated by activating guanine nucleotide exchange factors (GEFs) and deactivating
GTPase-activating proteins (GAPs). Among the differentially expressed genes, are two
RhoA-specific GEFs (ARHGEF25 and ARHGEF28), both significantly downregulated.
Furthermore, two downstream targets of the RhoA pathway (CYR61 and LIMK2), which
are positively regulated by RhoA, are significantly downregulated, hence, enabling organ
fate specific transcript expression. Interestingly, FGF2 is significantly upregulated in the
DPCs in condensations. FGF2 has been reported to antagonize Rho activation, leading to
loss of stress fibers and regulating the reorganization of cortical actin (Lee et al., 2003).
Thus, the observation from the in vitro model regarding the RhoA pathway is in con-
cordance with the described mechanism of cytoskeletal transduction of mechanical cues
during mesenchymal condensation and the requirements on the cytoskeleton to allow
differentiation and ECM deposition are fulfilled. The immunohistologic analysis of the
DPC condensates at a later time point (4 days, see figure 3.38), shows a high abundance
of the condensation-specific ECM molecules fibronectin, collagen type I and collagen type
IV, suggesting that the early mechanical cue of condensation (6h) induced this specific
ECM production. The collagens exhibit a zonal expression, highest at the sites where
condensation began.
Among the morphogens, expressed by the condensing cells, and incorporated into the
ECM for availability in the subsequent steps of cytodifferentiation, i.e., odontoblast and
ameloblast differentiation, are molecules of the TGFß superfamily (Begue-Kirn et al.,
1994; Vaahtokari et al., 1991). The canonical TGFß- and BMP signaling pathway
4.3 Early phase of condensation 89
involves the action of SMADs. SMAD1/5/8 activation is specific for BMP-induced pathway
and SMAD2/3 for TGFß- or Activin-induced pathway. They both share the Co-SMAD4 as
dimerization partner, thereby interconnecting the TGFß and BMP pathway (see figure 3.21).
The balancing between the two pathways during tissue regeneration and embryogenesis and
their different outcomes are described for many organs (Oshimori et al., 2012; Xu et al.,
2009; Zeisberg et al., 2003). During tooth development, both pathways are regulated
in all steps of organogenesis. The complex sequences of spatiotemporal expression and
fine-tuning of activity, mediated by expression of inhibitory or supporting molecules, is not
elucidated yet.
The RNA Seq data suggest a strong shift of the balance to the TGFß/Activin signaling
pathway. All three TGFß ligands 1, 2, and 3 are significantly upregulated after 6 hours
of mesenchymal condensation, as well as INHBA transcripts. Being the only expressed
isoform from the Inhibin/Activin dimers, it can be concluded that only Activin, not Inhibin,
is translated. Additionally, the Activin receptor type I is significantly upregulated. The
mere existence of the transcripts is no evidence for functional ligands. The high abundance
of the TGFß1/2/3 mRNA is accompanied by a high occurrence of latent TGFß-binding
protein 1 (LTBP1 ). TGFß ligands are only secreted as large latent complex (LLC), com-
prising the TGFß homodimer, the latency associated peptide (LAP) as well as the LTBP.
This mechanisms allows proper mediation of the TGFß signaling, especially for paracrine
signaling. The activation of latent TGFß is mediated by conformational rearrangement of
the LTBPs to release TGFß through thrombospondins, by proteases MMPs, or by reactive
oxygen species. Matrix metalloproteinase 2 (MMP2), which is described to liberate TGFß,
is among the most abundant transcripts. Thrombospondin 1, 2, and 3 are also in the group
of highly expressed transcripts, pointing to active TGFß signaling.
Also, indications for an active BMP signaling can be drawn from the expression data
from RNA Seq. Interestingly, two main BMP antagonists, Noggin (NOG) and Gremlin
2 (GREM2 ) are significantly downregulated. Connective-tissue growth factor (CTGF),
a supporting factor for TGFß and antagonist of BMP4 is downregulated. BMP8A is
significantly upregulated, whereas BMP4 is significantly downregulated. The downreg-
ulation of BMP4 was unexpected at first glance. In murine tooth development, as a
transcriptional product of the PAX9/MSX1 duo, BMP4 is described to be a key molecule
for tooth initiation. It has been detected in the murine condensing mesenchyme directly
after epithelial induction of mesenchymal condensation. Here, in our model, no epithelium
is present. The condensation is induced cell-autonomously. It is reasonable, that in vivo
the BMP4 expression of the condensed mesenchyme is not a result of the mechanical cues of
condensation, but rather of the epithelial signal itself. In concordance with that hypothesis
is the observation of Ray et al., 2015 that during avian columella development "BMP
signaling is irrelevant for the process of condensation, it is required for dorsal morphology,
indicating that condensation and dorsal cell shape can be uncoupled". It is thought that
the BMP morphogen gradients rather define the axis patterning of organs and tissues,
than being involved in the promotion of the sequence of development (Bier et al., 2015).
In long-term culture, the BMP4 expression slightly increases after four weeks. Possibly,
despite skipping the step of epithelial attraction to induce mesenchymal condensation,
the self-orchestrated mesenchymal condensation and subsequent mechanical stimulation
90 Chapter 4 Discussion
was sufficient to induce odontoblast cytodifferentiation. All other measured genes of the
chosen marker panel were upregulated after long-term culture, supporting the preceding
assumption (see figure 3.14).
In general, the overall high abundance of Activin/BMP and TGFß antagonists, supports
the hypothesis, that the DPCs in my model produce the ligands, but not only for autocrine
signalling but at a large proportion for deposition in the extracellular matrix that is
produced around the condensed cells. The deposition of ligands, agonists, and antagonists
equates the establishment of morphogenic gradients. In vivo, these morphogenic gradients
are needed to produce the respective cell layers of differentiated cell types, such as secreting
odontoblasts and ameloblasts (Vaahtokari et al., 1991).
It can be assumed here, that the TGFß ligands, which are significantly upregulated in the
condensates after six hours in ultra-low attachment culture, are active and act auto- and
paracrine. As described above, TGFß expression is assigned to the mesenchymal phenotype
and induces expression of fibronectin and tenascin in murine precartilage condensation
(Chimal-Monroy et al., 2003), and activate collagen synthesis and DSPP expression,
thereby switching the cellular machinery to ECM production.
An interesting indication for active TGFß pathway is the observation that two direct
TGFß-target genes, SNAI1 and SNAI2, are both significantly upregulated during condens-
ation. In literature, the Snail genes are described to be the bona fide marker genes for
epithelial-mesenchymal transition and are implicated in metastasizing tumors. The Snail
transcription factors downregulate epithelial markers, such as E-cadherin, cytokeratins,
and desmoplakins, and upregulate the mesenchymal marker vimentin, fibronectin, and
induce RhoB activation to enhance migration. The loss of epithelial adhesion molecules and
elevated migratory capacity is associated with invasive tumors. Very few publications show
the expression of SNAI1/SNAI2 in mesenchymal condensates, e.g., in dermal condensates
of developing murine hair follicles and during palatal development of the chick (Martinez-
Alvarez et al., 2004; Sennett et al., 2015). When the murine snai gene was discovered in
1992, expression was "observed in condensing cartilage and in the mesenchymal component
of several tissues (lung, kidney, teeth, and vibrissae) that undergo epithelial-mesenchymal
inductive interactions during development." (Nogai et al., 2008; D. E. Smith et al., 1992).
Gain- and loss-of-function studies for bone development indicate a role of snai in controlling
chondrocyte proliferation and differentiation (Y. Chen et al., 2013). A literature search
did not reveal any articles on the role of Snai genes neither in tooth development nor
for other skin appendage organs. Thus, the Snail transcription factors may represent a
valuable target for further investigations.
Taken together, the data suggest induction of cytoskeletal rearrangement through regulation
of actin assembly via tuned RhoA activity. Putatively, as a consequence, the TGFß signal-
ing activity is elevated, with emphasis on TGFß1/2/3 signaling. These results show, that
the model of self-orchestrated condensation of the DPCs resembles the mechanochemical
signals that are described in vivo.
In the condensation, the most significantly overrepresented pathway resulting from Gene
Set Enrichment Analysis (GSEA), was the MAPK signaling pathway (see Appendix figure
5.8). Therefore, a detailed analysis was performed for this pathway (see figure 3.23).
4.3 Early phase of condensation 91
Multiple pathways can induce the MAPK pathway. The most prominent is the induction
via receptor tyrosine kinases (RTKs), that are activated by various growth factors. Among
those growth factors are fibroblast growth factors (FGFs), epidermal growth factors (EGFs),
platelet-derived growth factors (PDGFs), and many more. There are three main pathways,
that are elicited upon stimulation of the respective receptor: the ERK1/2 pathway (activ-
ated through classical growth factors), the p38/JNK pathway (activated through TGFß
ligands, inflammatory signals, and stress), and the ERK5 pathway (induced by growth
factors, and cytokines).
Several groups describe the involvement of the MAPK signaling in tissue morphogenesis
and condensation (T. Mammoto, A. Mammoto, Jiang et al., 2015). However, from
the RNA Seq data presented here, a prediction which MAPK module is activated or
deactivated is challenging. When taking a closer look at the superior level of the pathway,
three ligands (FGF2, TGFß1/2/3 ) are upregulated, as well as the EGF receptor, indicating
an activation of the ERK1/2 pathway. Regarding, FGF2, the qPCR and RNA Seq data
appear contradictory. After six hours, FGF2 is upregulated. However, no significant
change in expression is observed after 24 hours of mesenchymal condensation compared to
monolayer DPCs. Only after four weeks, the DPCs in the condensate exhibit a significant
increase again. This observation can be explained by complex fine-tuning of the molecule
expression. As described above, FGF2 plays a role in RhoA activity modulation. It is
likely that the downregulation after 24 hours is a result of feedback inhibition, and in later
phases, when FGF2 is upregulated, it putatively has a role in cytodifferentiation.
Almost all involved molecules of the MAPK signaling are either expressed and/or differen-
tially regulated. The ratio between up- and downregulated genes is evenly balanced after
six hours of condensation (see heatmap figure 3.22). Furthermore, the activation of almost
each molecule in the pathway depends on phosphorylation (+p in Appendix figure 3.23)
by activity of upstream kinases, that were activated by upstream kinases beforehand. To
determine, if the MAPK signaling is indeed activated during the in vitro condensation, a
MAPK-responsive reporter construct, was transduced into DPCs transferred into ultra-low
attachment culture for condensation. In the MAPK reporter vector, the TF binding
site for serum response factor (SRF) and ETS domain-containing protein (ELK1) was
cloned as 6x repeats in front of the eGFP gene. Thereby, a eGFP expression is directly
correlated with MAPK activity, through the ERK1/2, or p38 pathway. The measured
reporter activity, assessed by geometric mean fluorescence of each cell in the well on single
cell level, impressively confirms that MAPK activated genes belong to the immediate early
response genes (IEGs). IEGs are reported to be activated and transcribed within minutes
after stimulation (Bahrami et al., 2016). The majority of these genes are expressed
upon activation of MAPK signaling or RhoA activation and their expression usually peaks
between 15 to 30 minutes. The expression is described to be transient, since the gene
products are in a negative feedback loop with the receptor or associated adapter molecules.
Here, the eGFP intensity, and therefore the MAPK activity, was significantly elevated after
five minutes and was at maximum after 20 minutes. Thereafter, the expression level was
sustained, indicating an attenuation of the immediate early signaling. Due to the long
half-life of eGFP, only the activation but the the signalling decline could be analyzed.
92 Chapter 4 Discussion
Another important signaling pathway during embryonic tooth development is the Wnt
signaling pathway. The role of Wnt in dental epithelium has been extensively studied,
and a loss of epithelial LEF1 transcription factor induces arrest of development in early
bud stage. Only few studies were performed to investigate the role in the early dental
mesenchyme. Mesenchyme-specific loss of ß-catenin leads to failure of bud-to-cap-stage
transition. Therefore, the role of ß-catenin in the dental mesenchyme is more attributed
to differentiation than to mesenchymal condensation. Interestingly, the loss of nuclear
ß-catenin leads to absence of mesenchymal BMP4, independent of the presence of Msx1
which was thought to be the crucial inducer of BMP4 before. The mesenchymal BMP4 is
in turn needed to activate Shh expression in the epithelium, providing the next step of
reciprocal signaling to progress the development of the tooth (J. Chen et al., 2009). It is
likely, that the missing BMP4 expression in the early mesenchymal condensates, generated
from the ultra-low attachment culture is a result of the missing epithelium described to be
the main supplier of Wnt ligands. During in vitro condensation, it can be assumed from
the RNASeq data, that the Wnt pathway is activated or, at least "primed" after six hours
of mesenchymal condensation. The Frizzled receptor 7 (FRZ7 ) is highly abundant, and two
important inhibitors of Wnt ligands, DKK1 and DKK2, are downregulated. The two main
transcription factors acting synergistically as effectors of Wnt pathway, Lymphoid Enhancer
Binding Factor 1 (LEF1 ) and Transcription Factor 7 (TCF7 ), are upregulated. Peroxisome
Proliferator Activated Receptor Delta (PPARD), an enhancer of the TCF/LEF duo, as
well as FRAT2, a regulator of the ß-catenin degrading complex, are both significantly
upregulated. Whether the Wnt pathway is indeed active or the mesenchymal condensate is
awaiting an epithelial signal, is hard to predict from the RNA data. WNT5A and WNT5B
are highly abundant in the DPCs, which is in concordance with literature describing WNT5
ligands to be exclusively expressed in the dental mesenchyme (Sarkar et al., 1999). A
comparison of published Wnt target genes (https://web.stanford.edu/group/nusselab/cgi-
bin/wnt/target_genes) revealed no clear result if the pathway is active or not. Moreover
most described target genes are also activated upon TGFß or MAPK signaling, which
emphasizes the complex network of cross-talk between the pathways. After four days,
the DPC condensates display a complete nuclear localization of ß-catenin (see figure
3.38), suggesting an activation of the Wnt pathway at this time point. Nevertheless, ß-
catenin can also be stabilized and transferred to the nucleus via PI3K/Akt phosphorylation.
An intriguing result from the GSEA of the differentially expressed genes was the emer-
gence of "signaling by notch" with the highest significance of being enriched (False Discovery
Rate (FDR)=10-5) (see figure Appendix 5.8). The Notch signaling pathway is triggered by
direct cell-cell contact. The receptor Notch on the receiver cell and the ligand Delta-like
(DLL) or Jagged (JAG; Serrate) of the signaling cell are both transmembrane molecules.
When the two cells come in close proximity, DLL binds to Notch, and through the mechan-
ical stimulus on the receptive cell, proteolytic cleavage is induced. ADAM metallopeptidase
domain 17 (ADAM17)/TACE cleaves the external protein domain, and the ligand-receptor
complex is endocytosed by the ligand producing cell. The remaining intracellular receptor
domain in the target cell becomes substrate for a gamma-secretase-complex that releases
the Notch intracellular domain (NCID). NCID subsequently translocates to the nucleus to
4.3 Early phase of condensation 93
replace the repressing complex (consisting of NCoR, HDAC, and TBL) on the transcrip-
tional activator RBPJ/CSL to induce transcription of Notch target genes of the HES/HEY
family (hairy and enhancer of split), hypoxia-inducible factor 1 alpha (HIF1A), and/or
Estrogen Receptor 1 (ESR1) (Berridge, 2014). The Notch pathway is highly conserved
and functions in cell-fate decision. Upon asymmetrical cell division of stem cells, one
daughter cell retains the stem cell phenotype, whereas the other adopts a new fate. Notch
is used by the later cell to signal back to the neighbor cell to suppress it from adopting a
similar cell fate. The same principle is applied by Notch during establishment of lateral
inhibition of neighboring neuronal cells. From developmental biology, it is known that
Notch plays a role in mesenchymal condensates of skeletogenesis, where Notch maintains
the mesenchymal precursors in an undifferentiated state and through lateral signaling to
the neighboring cells; the distinct developmental zones of long bone growth are established
(Hilton et al., 2008). The function of Notch is also described for mesenchymal condensates
in feather bud development and kidney (A. Li et al., 2013; Sirin et al., 2012).
Although Delta and Notch have been shown to be expressed in murine mesenchyme of
developing molar teeth (Thimios A Mitsiadis, Hirsinger et al., 1998), so far, no function
has been assigned of this pathway during tooth development, neither in the animal model,
nor in humans. Since TGFß1 and BMP4 induced the expression of DLL1 and NOTCH3 in
explant cultures of murine dental mesenchyme, a role in odontoblast differentiation was
suggested (Thimios A Mitsiadis, Hirsinger et al., 1998). The DPCs in my condensation
model clearly show upregulation of the Notch pathway on the transcriptional level. The
receptor NOTCH3 and the ligand JAG1 are significantly upregulated. FURIN, a protease
that functions in processing the pre-protein of the Notch receptor is also upregulated.
Furthermore, at least five validated target genes of the Notch signaling pathway are upregu-
lated, including HES1, HEY1, HEY2, HEYL and HIF1A1. The co-repressors of target gene
expression, HDAC11 and HDAC7 are significantly downregulated, whereas the co-activator
TLE3 is significantly upregulated indicating induction of Notch signaling.
To summarize the molecular events of the early phase of in vitro condensation, it can
be concluded, that the cytoskeleton of the condensing mesenchymal is remodeled via
regulated activity of RhoA with the involvement of reduction of stress fibers and formation
of cortical actin that support direct cell-cell-adhesion. Immediate early response on that
mechanical cues is the modulation of the MAPK activity. Furthermore, in concordance
with literature describing the key molecular pathways that drive tooth organogenesis, there
is striking evidence that the TGFß pathway is activated. Additionally, the FGF pathway
(see Appendix figure 5.10) as well as the Wnt pathway are active or primed to react on
induction (e.g., by epithelial interaction partner).
Therefore, the presented easy and practicable method to culture human dedifferenti-
ated DPCs in ultra-low attachment culture represents a suitable in vitro model to resemble
embryonic mesenchymal condensation and therefore, and is a valuable tool to elucidate the
mechanisms of molecular events without the use of animal models and embryonic cells.
94 Chapter 4 Discussion
4.4 Co-culture with epithelial cells
The next step would include a comprehensive analysis of the behavior and molecular
signature of the organoid in combination with epithelial cells. Since the dental epithelial
stem cell niche giving rise to ameloblast or successional teeth, is degenerated in the adult
human, another source of epithelial cells that interact with the DPC condensate has to be
identified. First co-culture experiments with skin or gingiva-derived keratinocytes proved
that the DPC condensate is indeed instructing on the keratinocytes. The keratinocytes
gather around the DPC aggregate and, in the case of gingiva keratinocytes, they form a
surrounding sheath. Due to the limited availabilty of gingiva as well as their short culture
period, it was not feasible to perform all co-cultures with gingiva-derived keratinocytes.
In contrast to the gingiva-derived keratinocytes, the skin keratinocytes partially detached
from the condensate and only a few cells intensely interacted with the mesenchymal cells
(see figures 3.39 and 3.40). After four days of co-culture the majority of the residual
skin keratinocytes were found at the border to the DPC condensate, nonetheless, a few
cytokeratin-positive cells were detected inside of the condensate. After two weeks, only
very few keratinocytes were detected at the surrounding. Those were tightly bound to
the cellular aggregate. At that time point, also cytokeratin-positive cells were localized
inside the mesenchymal condensate. Remarkably, in native skin section, epidermal ker-
atinocytes do not show cytopkeratin 8/18- and cytokeratin 19-positivity. Only the sweat
and sebaceous gland epithelia stain positive. Cytokeratin 19 is regarded as a marker of
ameloblast differentiation which was shown in several publications (Apellaniz et al.,
2015; Domingues et al., 2000; Heikinheimo et al., 1989; Kasper et al., 1989. The
acidic cytokeratin 19 pairs, in filament formation, with the basic cytokeratin 8. Also
cytokeratin 8, is described to be expressed in all odontoblast epithelia including the dental
lamina and the enamel organ (Heikinheimo et al., 1989; Kero et al., 2014). Cytoker-
atin 14 has been shown to physically interact with amelogenin (Ravindranath et al.,
2001) during amelogenesis and has also been proposed as an ameloblast-lineage marker
in rat cells (M. J. Tabata et al., 1996), but skin immunohistological stainings show cy-
tokeratin 14 expression by epidermal keratinocytes of human skin (see Appendix figure 5.5).
When gingival keratinocytes are co-cultured with the DPC condensates, they tightly
bind to the surface as a sheath. This sheath is maintained for the whole culture time.
After five weeks of co-culture, the organoids were histologically analyzed (see figure 3.41).
Remarkably, the keratinocytes form a multilayered sheath and are detectable in the inner
core of the condensate. Luckily, the cutting angle was chosen well, so that the H&E
staining impressively depicts the invagination that was performed by the epithelial cells
into the mesenchyme. The epithelial cells in the inside of the condensate form a columnar
epithelium around an acellular matrix deposit. The columnar arrangement can be distinctly
seen in the higher magnification of the VIM staining. This arrangement is similar-looking
to basal epithelia. The augmented collagen type IV expression between the mesenchyme
and the epithelia is a strong indication for the formation of a basal lamina. The cytokeratin
expression revealed a strong expression of CK15 in all layers of the epithelial cells, whereas
cytokeratin 8/18 is only detected in the lining epithelium of the sheath.
4.5 Conclusion and Perspectives 95
After ten weeks of co-culture, the aggregates exhibit a
puzzling morphology (see figure 3.42). The cellularity is low,
and several mineralized nodules have formed inside of the
organoids. These nodules are connected by collagen type
IV/vimentin-positive cells. With very few exceptions, the
cytokeratin 8/18 expression is disapparent. The nodules
resemble keratinized cysts that occasionally emerge on the
alveolar ridge of newborns (2 weeks to 5 months after birth)
(see figure 4.2). This occurrence is common (prevalence >50%)
and harmless. The cysts develop from fragments of the dental
lamina, that degenerates during embryonic tooth formation.
Sporadically, remnants of this lamina reside in the mesen-
chyme and form cysts that are lined by odontogenic epithelium
with a keratin-filled lumen (Moda, 2011). There are reports,
that upon trauma or fracture these epithelial remnants can
give rise to supernumerary teeth in adults (Anthonappa et Fig. 4.2: Gingival cysts
al., 2013), which demonstrates that adult mesenchymal cells in an infant (Shear et al.,2008).
are capable of tooth formation in vivo when being instructed
with appropriate signals. Although, from the histological sections, it is hard to conclude
which epithelial phenotype the keratinocytes adopted by co-culture with the inductive
mesenchyme. Nonetheless, it can be summarized that a conversion of the cell type was
achieved, as marked by the cytokeratin 19 and cytokeratin 8/18 staining and that the cells
differentiated and arranged columnar. Moreover, they produced an unneglectable amount
of extracellular matrix.
The resulting structures from the co-cultures are promising, and the missing features
of proper tooth development are simply attributed to the limited conditions of a in
vitro culture. It is of interest, which morphology these organoids would arise in an in
vivo environment, as the jaw providing positional information as well as vasculature and
innervation. Moreover, the results raise the question, whether the produced mesenchymal
condensates that are instructing and most probably inductive, would act as signalling
center for tooth formation when transplanted in the adult jaw bone.
4.5 Conclusion and Perspectives
To examine the function of genes and their causative roles in crucial processes of cell
differentiation, and in the emergence of diseases, model systems are indispensable. Current
models range from living organisms, explant cultures, standard cell cultures, organoids,
and in silico models. Combining the knowledge deriving from all those models aims to the
development of meaningful pharmaceuticals and regenerative therapies.
The demands on an ideal culture model combine availability of cells (ideally adult and/or
patient-derived), practicability, and close correlation to the physiological situation.
The presented model of a three-dimensional culture of human dental pulp cells represents
a valuable tool, mimicking mesenchymal condensation on the one hand and differentiation
towards odontogenic lineage on the other. After expansion and subsequent dedifferentiation
96 Chapter 4 Discussion
of the dental pulp cells, they are cultured under non-adherent conditions to allow self-
orchestrated cell-cell-interaction. As evidenced by comprehensive gene expression analysis
and cellular behavior, this in vitro step highly correlates to the events described for mesen-
chymal condensation in vivo. Furthermore, the mechanochemical signals, cell-autonomously
induced by and, in turn, on DPCs, provoke odontogenic differentiation in the condensates
upon long-term culture. Inductive abilities are distinctive for mesenchymal condensations
in embryogenesis. The produced condensates prove to be instructing and inductive when
combined with epithelial cells. Therefore, the culture model represents the first human
model to study mechanisms of dental mesenchymal condensation and subsequent odon-
togenic differentiation, to find novel involved pathways and examine cross-talks between
pathways.
The next steps will include in-depth analyses of signalling pathways to unravel mechanisms
of human organogenesis. The reporter system for specific pathways will be enhanced. Novel
vector constructs with destabilized eGFP were already cloned and tested in monolayer
DPCs. With a half-life of one hour, the spatiotemporal resolution is enhanced and putat-
ively declining signals from feedback inhibition can be assessed. Furthermore, the vector
portfolio will be extended by a Notch signalling-responsive vector.
The whole transcriptome analysis will be repeated with additional donor samples and a
second time point, 24 hours of condensation, will be included. A comprehensive analysis of
protein level by high-throughput proteomics, comprising posttranslational modifications, is
of particular interest to evaluate the actual activation state of the signalling in DPCs upon
condensation and during differentiation.
Due to the promising results of this thesis, the Technische Universität Berlin, together
with the inventors, we decided to file a patent entitled "Method of preparing an artificial
tooth primordium in vitro and artificial tooth primordium derived therefrom". This patent
is granted, and the method is en route to a pre-clinical pilot study. A corresponding
research proposal is already prepared and will be submitted soon. The method aims to
produce tooth primordia for transplantation purposes to cure tooth loss. However, the
proof of safety and efficacy is required in advance to Advanced Therapy Medicinal Products
(ATMP) approval.
Bibliography
About, Imad (2013): ‘Dentin–pulp regeneration: the primordial role of the microenviron-
ment and its modification by traumatic injuries and bioactive materials’. Endodontic
Topics, vol. 28(1): pp. 61–89 (cit. on p. 3).
Abrahão, Ivete Jorge, Manoela Domingues Martins, Emilio Katayama, João
Humberto Antoniazzi, Angelo Segmentilli and Márcia Martins Marques
(2006): ‘Collagen analysis in human tooth germ papillae’. Brazilian dental journal, vol.
17(3): pp. 208–212 (cit. on p. 15).
Afgan, Enis, Dannon Baker, Bérénice Batut, Marius Van Den Beek, Dave
Bouvier, Martin Čech, John Chilton, Dave Clements, Nate Coraor, Björn A
Grüning et al. (2018): ‘The Galaxy platform for accessible, reproducible and collaborat-
ive biomedical analyses: 2018 update’. Nucleic acids research, vol. 46(W1): W537–W544
(cit. on p. 39).
Ahn, Youngwook, Brian W Sanderson, Ophir D Klein and Robb Krumlauf
(2010): ‘Inhibition of Wnt signaling by Wise (Sostdc1) and negative feedback from Shh
controls tooth number and patterning’. Development, vol. 137: pp. 3221–3231 (cit. on
p. 8).
Amerongen, J.P. van, Inez G. Lemmens andG.J.M. Tonino (1983): ‘The concentration,
extractability and characterization of collagen in human dental pulp’. Archives of Oral
Biology, vol. 28(4): pp. 339–345 (cit. on p. 3).
Anthonappa, R. P., N. M. King and A. B. M. Rabie (2013): ‘Aetiology of supernu-
merary teeth: a literature review’. European Archives of Paediatric Dentistry, vol. 14(5):
pp. 279–288 (cit. on p. 95).
Apellaniz, Delmira, Sabrina Nieves, Gabriel Tapia, Alvaro Maglia, Adalberto
Mosqueda-Taylor and Ronell Bologna-Molina (2015): ‘Immunohistochemical
analysis of CK14 and CK19 in tooth germ and ameloblastoma’. ODONTOESTOMATO-
LOGIA, vol. 17(25): pp. 4–10 (cit. on p. 94).
Arakaki, Makiko, Masaki Ishikawa, Takashi Nakamura, Tsutomu Iwamoto, Aya
Yamada, Emiko Fukumoto, Masahiro Saito, Keishi Otsu, Hidemitsu Harada,
Yoshihiko Yamada and Satoshi Fukumoto (2012): ‘Role of epithelial-stem cell
interactions during dental cell differentiation’. Journal of Biological Chemistry, vol.
287(13): pp. 10590–10601 (cit. on p. 24).
Athirasala, Avathamsa, Fernanda Lins, Anthony Tahayeri, Monica Hinds,
Anthony J Smith, Christine Sedgley, Jack Ferracane and Luiz E Bertassoni
(2017): ‘A novel strategy to engineer pre-vascularized full-length dental pulp-like tissue
constructs’. Scientific reports, vol. 7(1): p. 3323 (cit. on p. 84).
97
98 Bibliography
B Kulkarni, Ashok, Derk Joester, John Wright, James P Simmer, Janet
Moradian-Oldak, John D Bartlett, Megan Pugach, Olivier Duverger, Ophir
D Klein, Pamela DenBesten, Rodrigo Lacruz, Sarah Millar, Thomas GH
Diekwisch and Wendy Shaw (2017): ‘Meeting report: a hard look at the state of
enamel research’. International journal of oral science, vol. 9(11): e3 (cit. on p. 13).
Bahrami, Shahram and Finn Drablos (2016): ‘Gene regulation in the immediate-early
response process’. Advances in Biological Regulation, vol. 62: pp. 37–49 (cit. on p. 91).
Begue-Kirn, C, AJ Smith, M Loriot, C Kupferle, JV Ruch and H Lesot (1994):
‘Comparative analysis of TGFßs, BMPs, IGF1, msxs, fibronectin, osteonectin and bone
sialoprotein gene expression during normal and in vitro-induced odontoblast differen-
tiation’. The International journal of developmental biology, vol. 38(3): pp. 405–420
(cit. on p. 88).
Berkovitz, Barry KB, Graham Rex Holland and Bernard J Moxham (2009):
Oral Anatomy, Histology and Embryology. 4th ed. Elsevier Mosby (cit. on pp. 3, 4, 15).
Berridge, Michael J. (2014): ‘Module 2: Cell Signalling Pathways’. Cell Signalling
Biology, vol. 6 (cit. on p. 93).
Bier, Ethan and Edward M. De Robertis (2015): ‘BMP gradients: A paradigm for
morphogen-mediated developmental patterning’. Science, vol. 348(6242) (cit. on p. 89).
Birchmeier, Carmen and Ermanno Gherardi (1998): ‘Developmental roles of HGF/SF
and its receptor, the c-Met tyrosine kinase’. Trends in cell biology, vol. 8(10): pp. 404–410
(cit. on p. 82).
Burridge, Keith and Christophe Guilluy (2016): ‘Focal adhesions, stress fibers and
mechanical tension’. Experimental Cell Research, vol. 343(1): pp. 14–20 (cit. on p. 88).
Caplan, Arnold I and Diego Correa (2011): ‘The MSC: an injury drugstore’. Cell
stem cell, vol. 9(1): pp. 11–15 (cit. on p. 79).
Casella, James F, Michael D. Flanagan and Shin Lin (1981): ‘Cytochalasin D
inhibits actin polymerization and induces depolymerization of actin filaments formed
during platelet shape change’. Nature, vol. 293(5830): pp. 302–305 (cit. on p. 85).
Chan, CP, WH Lan, MC Chang, YJ Chen, WC Lan, HH Chang and JH Jeng (2005):
‘Effects of TGF-ßs on the growth, collagen synthesis and collagen lattice contraction of
human dental pulp fibroblasts in vitro’. Archives of oral biology, vol. 50(5): pp. 469–479
(cit. on p. 81).
Chang, Chenbei (2016): ‘Agonists and Antagonists of TGF-ß Family Ligands’. Cold
Spring Harbor Perspectives in Biology, vol. 8: a021923 (cit. on p. 56).
Chen, Jianquan, Yu Lan, Jin-A Baek, Yang Gao and Rulang Jiang (2009):
‘Wnt/beta-catenin signaling plays an essential role in activation of odontogenic mesen-
chyme during early tooth development’. Developmental biology, vol. 334(1): pp. 174–185
(cit. on pp. 20, 92).
Chen, Ying and Thomas Gridley (2013): ‘Compensatory regulation of the Snai1 and
Snai2 genes during chondrogenesis’. Journal of Bone and Mineral Research, vol. 28(6):
pp. 1412–1421 (cit. on p. 90).
Chimal-Monroy, Jesus and L Diaz De Leon (2003): ‘Expression of N-cadherin, N-CAM,
fibronectin and tenascin is stimulated by TGF-ß1, ß2, ß3 and ß5 during the formation of
99
precartilage condensations.’ International Journal of Developmental Biology, vol. 43(1):
pp. 59–67 (cit. on pp. 87, 90).
Cho, Andrew, Naoto Haruyama, Bradford Hall, Mary Jo S. Danton, Lu Zhang,
Praveen Arany, David J. Mooney, Yassine Harichane, Michel Goldberg,
Carolyn W. Gibson and Ashok B. Kulkarni (2013): ‘TGF-ß Regulates Enamel
Mineralization and Maturation through KLK4 Expression’. PLOS ONE, vol. 8(11) (cit.
on p. 19).
Dassule, Hélène R, Paula Lewis, Marianna Bei, Richard Maas and Andrew P
McMahon (2000): ‘Sonic hedgehog regulates growth and morphogenesis of the tooth’.
Development, vol. 127(22): pp. 4775–4785 (cit. on p. 20).
Domingues, Manoela G, Márcia MM Jaege, Vera C Araújo and Ney S Araújo
(2000): ‘Expression of cytokeratins in human enamel organ’. European journal of oral
sciences, vol. 108(1): pp. 43–47 (cit. on p. 94).
Duailibi, Monica T, Silvio E Duailibi, Conan S Young, John D Bartlett, Joseph
P Vacanti and Pamela C Yelick (2004): ‘Bioengineered teeth from cultured rat tooth
bud cells’. Journal of dental research, vol. 83(7): pp. 523–528 (cit. on p. 71).
El Agha, Elie, Djuro Kosanovic, Ralph T Schermuly and Saverio Bellusci
(2016): ‘Role of fibroblast growth factors in organ regeneration and repair’. Vol. 53:
pp. 76–84 (cit. on p. 52).
Ferguson, Christine A, Abigail S Tucker, Lars Christensen, Anthony L Lau,
Martin M Matzuk and Paul T Sharpe (1998): ‘Activin is an essential early mes-
enchymal signal in tooth development that is required for patterning of the murine
dentition’. Genes & development, vol. 12(16): pp. 2636–2649 (cit. on pp. 19, 82).
Fujii, Sakiko, Katsumi Fujimoto, Noriko Goto, Masami Kanawa, Takeshi Kawa-
moto, Haiou Pan, Petcharin Srivatanakul, Waralak Rakdang, Juthamas
Pornprasitwech, Tania Saskianti, Ketut Suardita, Fusanori Nishimura and
Yukio Kato (2015): ‘Characteristic expression of MSX1, MSX2, TBX2 and ENTPD1
in dental pulp cells’. Biomedical Reports, vol. 3(4): pp. 566–572 (cit. on p. 83).
Gao, Bo, Xin Zhou, Xuedong Zhou, Caixia Pi, Ruoshi Xu, Mian Wan, Jing
Yang, Yue Zhou, Chengcheng Liu, Jianxun Sun, Yan Zhang and Liwei Zheng
(2015): ‘BMP7 and EREG Contribute to the Inductive Potential of Dental Mesenchyme’.
Scientific reports, vol. 5: p. 9903 (cit. on p. 86).
Gasse, B., E. Karayigit, E. Mathieu, S. Jung, A. GARRET, M. Huckert, S. Mork-
mued, C. Schneider, L. Vidal, J. Hemmerle, J. -Y. Sire and A. Bloch-Zupan
(2013): ‘Homozygous and compound heterozygous MMP20 mutations in amelogenesis
imperfecta’. Journal of dental research, vol. 92(7): pp. 598–603 (cit. on p. 14).
Goldberg, Michel and Anthony J Smith (2004): ‘Cells and extracellular matrices of
dentin and pulp: a biological basis for repair and tissue engineering’. Critical Reviews in
Oral Biology & Medicine, vol. 15(1): pp. 13–27 (cit. on p. 82).
Gronthos, S, M Mankani, J Brahim, P G Robey and S Shi (2000): ‘Postnatal human
dental pulp stem cells (DPSCs) in vitro and in vivo’. Proceedings of the National Academy
of Sciences of the United States of America, vol. 97(25): pp. 13625–30 (cit. on pp. 24,
80).
100 Bibliography
Gronthos, Stan, Agnieszka Arthur, P Mark Bartold and Songtao Shi (2011):
‘A Method to Isolate and Culture Expand Human Dental Pulp Stem Cells’. Methods in
molecular biology (Clifton, N.J.) Vol. 698: pp. 107–121 (cit. on p. 29).
Gurdon, JB (1971): ‘Gene activity during embryogenesis’. Triangle; the Sandoz journal
of medical science, vol. 10(1): pp. 23–28.
Hall, Brian K and Tsutomo Miyake (2000): ‘All for one and one for all: condensations
and the initiation of skeletal development’. Bioessays, vol. 22(2): pp. 138–147 (cit. on
pp. 21, 87).
Hardcastle, Zoe, Rong Mo, CC Hui and Paul T Sharpe (1998): ‘The Shh signalling
pathway in tooth development: defects in Gli2 and Gli3 mutants’. Development, vol.
125(15): pp. 2803–2811 (cit. on p. 20).
Heikinheimo, Kristiina, Marketta Hormia, Göran Stenman, Ismo Virtanen and
Risto-Pekka Happonen (1989): ‘Patterns of expression of intermediate filaments in
ameloblastoma and human fetal tooth germ’. Journal of Oral Pathology & Medicine, vol.
18(5): pp. 264–273 (cit. on p. 94).
Hematti, Peiman (2012): ‘Mesenchymal stromal cells and fibroblasts: a case of mistaken
identity?’ Cytotherapy, vol. 14(5): pp. 516–521 (cit. on p. 79).
Hilton, Matthew J, Xiaolin Tu, Ximei Wu, Shuting Bai, Haibo Zhao, Tatsuya
Kobayashi, Henry M Kronenberg, Steven L Teitelbaum, F Patrick Ross,
Raphael Kopan et al. (2008): ‘Notch signaling maintains bone marrow mesenchymal
progenitors by suppressing osteoblast differentiation’. Nature medicine, vol. 14(3): p. 306
(cit. on p. 93).
Horiguchi, Kana, Takuya Shirakihara, Ayako Nakano, Takeshi Imamura, Kohei
Miyazono and Masao Saitoh (2009): ‘Role of Ras Signaling in the Induction of
Snail by Transforming Growth Factor-ß’. Journal of Biological Chemistry, vol. 284(1):
pp. 245–253.
Hosseini, Samaneh, Shahrbanoo Jahangir and Mohamadreza Baghaban Eslam-
inejad (2018): ‘Tooth tissue engineering’. Biomaterials for Oral and Dental Tissue
Engineering. Elsevier: pp. 467–501 (cit. on pp. 24, 84).
Hu, L, Y Liu and S Wang (2018): ‘Stem cell-based tooth and periodontal regeneration’.
Oral diseases, vol. 24(5): pp. 696–705 (cit. on p. 24).
Huang, George T.-J., Kristina Shagramanova and Selina W. Chan (2006): ‘Form-
ation of Odontoblast-Like Cells from Cultured Human Dental Pulp Cells on Dentin In
Vitro’. Journal of Endodontics, vol. 32(11): pp. 1066–1073 (cit. on p. 42).
Hughes, Alex J, Hikaru Miyazaki, Maxwell C Coyle, Jesse Zhang, Matthew T
Laurie, Daniel Chu, Zuzana Vavrušová, Richard A Schneider, Ophir D Klein
and Zev J Gartner (2018): ‘Engineered tissue folding by mechanical compaction of
the mesenchyme’. Developmental cell, vol. 44(2): pp. 165–178 (cit. on p. 8).
Humphrey, Linda L., Rongwei Fu, David I. Buckley, Michele Freeman and
Mark Helfand (2008): ‘Periodontal disease and coronary heart disease incidence: A
systematic review and meta-analysis’. Journal of General Internal Medicine, vol. 23(12):
pp. 2079–2086 (cit. on p. 1).
Järvinen, Elina, Isaac Salazar-Ciudad, Walter Birchmeier, Makoto M. Taketo,
Jukka Jernvall and Irma Thesleff (2006): ‘Continuous tooth generation in mouse
101
is induced by activated epithelial Wnt/ß-catenin signaling’. Proceedings of the National
Academy of Sciences, vol. 103(49): pp. 18627–18632 (cit. on p. 20).
Jia, Shihai, Jing Zhou, Yang Gao, Jin-A Baek, James F Martin, Yu Lan and
Rulang Jiang (2013): ‘Roles of Bmp4 during tooth morphogenesis and sequential tooth
formation’. Development, vol. 140(2): pp. 423–432 (cit. on p. 20).
Jin, Hye Jin, Yun Kyung Bae, Miyeon Kim, Soon-Jae Kwon, Hong Bae Jeon,
Soo Jin Choi, Seong Who Kim, Yoon Sun Yang, Wonil Oh and Jong Wook
Chang (2013): ‘Comparative analysis of human mesenchymal stem cells from bone
marrow, adipose tissue, and umbilical cord blood as sources of cell therapy’. International
journal of molecular sciences, vol. 14(9): pp. 17986–18001 (cit. on p. 80).
Kadler, Shirin (2017): ‘Tissue engineered cartilage: how hard can it be?’ PhD thesis.
Technische Universität Berlin (cit. on p. 80).
Kasper, Michael, Uwe Karsten, Peter Stoslek and Roland Moll (1989): ‘Distri-
bution of intermediate-filament proteins in the human enamel organ: unusually complex
pattern of coexpression of cytokeratin polypeptides and vimentin’. Differentiation, vol.
40(3): pp. 207–214 (cit. on p. 94).
Kawai, R, N Ozeki, H Yamaguchi, T Tanaka, K Nakata, M Mogi and H Nakamura
(2014): ‘Mouse ES cells have a potential to differentiate into odontoblast-like cells using
hanging drop method’. Oral diseases, vol. 20(4): pp. 395–403 (cit. on p. 24).
Kero, Darko, D Kalibovic Govorko, K Vukojevic, M Cubela, V Soljic and
M Saraga-Babic (2014): ‘Expression of cytokeratin 8, vimentin, syndecan-1 and Ki-67
during human tooth development’. Journal of molecular histology, vol. 45(6): pp. 627–640
(cit. on p. 94).
Kettunen, Päivi, Bradley Spencer-Dene, Tomasz Furmanek, Inger Hals Kvinns-
land, Clive Dickson, Irma Thesleff and Keijo Luukko (2007): ‘Fgfr2b mediated
epithelial–mesenchymal interactions coordinate tooth morphogenesis and dental trigem-
inal axon patterning’. Mechanisms of development, vol. 124(11): pp. 868–883 (cit. on
p. 19).
Klein, Ophir D, David B Lyons, Guive Balooch, Grayson W Marshall, M
Albert Basson, Miroslav Peterka, Tomas Boran, Renata Peterkova and
Gail R Martin (2008): ‘An FGF signaling loop sustains the generation of differentiated
progeny from stem cells in mouse incisors’. Development, vol. 135(2): pp. 377–385 (cit. on
p. 19).
Kollar, Edward J. and Grace R. Baird (1969): ‘The influence of the dental papilla
on the development of tooth shape in embryonic mouse tooth germs’. Development, vol.
21(1): pp. 131–148 (cit. on p. 85).
– (1970a): ‘Tissue interactions in embryonic mouse tooth germs: II. The inductive role of
the dental papilla’. Development, vol. 24(1): pp. 173–186 (cit. on pp. 71, 85).
– (1970b): ‘Tissue interactions in embryonic mouse tooth germs: I. Reorganization of the
dental epithelium during tooth-germ reconstruction’. Development, vol. 24(1): pp. 159–
171 (cit. on p. 85).
Kollar, EJ and C Fisher (1980): ‘Tooth induction in chick epithelium: expression of
quiescent genes for enamel synthesis’. Science, vol. 207(4434): pp. 993–995 (cit. on p. 8).
102 Bibliography
Krahn, Katy Nash, Carlijn V.C. Bouten, Sjoerd van Tuijl, Marc A.M.J. van
Zandvoort and Maarten Merkx (2006): ‘Fluorescently labeled collagen binding
proteins allow specific visualization of collagen in tissues and live cell culture’. Analytical
Biochemistry, vol. 350(2): pp. 177–185 (cit. on p. 65).
Krampera, Mauro, Jacques Galipeau, Yufang Shi, Karin Tarte and Luc Sensebe
(2013): ‘Immunological characterization of multipotent mesenchymal stromal cells—The
International Society for Cellular Therapy (ISCT) working proposal’. Cytotherapy, vol.
15(9): pp. 1054–1061 (cit. on p. 79).
Kratochwil, Klaus, Juan Galceran, Sabine Tontsch, Wera Roth and Rudolf
Grosschedl (2002): ‘FGF4, a direct target of LEF1 and Wnt signaling, can rescue
the arrest of tooth organogenesis in Lef1-/- mice’. Genes & development, vol. 16(24):
pp. 3173–3185 (cit. on p. 20).
Lacruz, Rodrigo S, Stefan Habelitz, J Timothy Wright and Michael L Paine
(2017): ‘Dental enamel formation and implications for oral health and disease’. Physiolo-
gical reviews, vol. 97(3): pp. 939–993 (cit. on p. 1).
Lee, Hyung Taek and EP Kay (2003): ‘FGF-2 induced reorganization and disruption
of actin cytoskeleton through PI 3-kinase, Rho, and Cdc42 in corneal endothelial cells’.
Mol Vis, vol. 9(76-78): pp. 624–634 (cit. on p. 88).
Lei, Qiubo, Yongsu Jeong, Kamana Misra, Shike Li, Alice K Zelman, Douglas J
Epstein and Michael P Matise (2006): ‘Wnt signaling inhibitors regulate the transcrip-
tional response to morphogenetic Shh-Gli signaling in the neural tube’. Developmental
cell, vol. 11(3): pp. 325–337 (cit. on p. 8).
Li, Ang, Meng Chen, Ting-Xin Jiang, Ping Wu, Qing Nie, Randall Widelitz
and Cheng-Ming Chuong (2013): ‘Shaping organs by a wingless-int/Notch/nonmuscle
myosin module which orients feather bud elongation’. Proceedings of the National
Academy of Sciences, vol. 110(16): E1452–E1461 (cit. on p. 93).
Li, Xiuying, Jinping Bai, Xiaofeng Ji, Ronggui Li, Yali Xuan and Yimin Wang
(2014): ‘Comprehensive characterization of four different populations of human mesen-
chymal stem cells as regards their immune properties, proliferation and differentiation’.
International Journal of Molecular Medicine, vol. 34(3): pp. 695–704 (cit. on p. 80).
Liang, Jia, Guangtai Song, Qing Li and Zhuan Bian (2012): ‘Novel missense mutations
in PAX9 causing oligodontia’. Archives of oral biology, vol. 57(6): pp. 784–789 (cit. on
p. 86).
Lin, Dahe, Yide Huang, Fenglei He, Shuping Gu, Guozhong Zhang, YiPing Chen
and Yanding Zhang (2007): ‘Expression survey of genes critical for tooth development
in the human embryonic tooth germ’. Developmental dynamics: an official publication of
the American Association of Anatomists, vol. 236(5): pp. 1307–1312 (cit. on p. 9).
Mafi, P, S Hindocha, R Mafi, M Griffin and WS Khan (2011): ‘Adult mesenchymal
stem cells and cell surface characterization - a systematic review of the literature’. Open
Orthop J, vol. 5(Suppl 2): pp. 253–260 (cit. on p. 79).
Main, James HP (1966): ‘Retention of potential to differentiate in long-term cultures of
tooth germs’. Science, vol. 152(3723): pp. 778–780 (cit. on p. 24).
103
Mammoto, Tadanori, Akiko Mammoto and Donald E Ingber (2013): ‘Mechanobio-
logy and developmental control’. Annual review of cell and developmental biology, vol.
29: pp. 27–61 (cit. on pp. 22, 23, 84).
Mammoto, Tadanori, Akiko Mammoto, Amanda Jiang, Elisabeth Jiang, Basma
Hashmi and Donald E Ingber (2015): ‘Mesenchymal condensation-dependent accumu-
lation of collagen VI stabilizes organ-specific cell fates during embryonic tooth formation’.
Developmental Dynamics, vol. 244(6): pp. 713–723 (cit. on pp. 22, 65, 70, 87, 91).
Mammoto, Tadanori, Akiko Mammoto, Yu-suke Torisawa, Tracy Tat, Ashley
Gibbs, Ratmir Derda, Robert Mannix, Marlieke de Bruijn, Chong Wing
Yung, Dongeun Huh and Donald E Ingber (2011): ‘Mechanochemical control of
mesenchymal condensation and embryonic tooth organ formation’. Developmental cell,
vol. 21(4): pp. 758–769 (cit. on pp. 19, 22, 23, 71, 88).
Martinez-Alvarez, Concepcion, Maria J Blanco, Raquel Perez, M.Angeles
Rabadan, Marta Aparicio, Eva Resel, Tamara Martinez and M.Angela Nieto
(2004): ‘Snail family members and cell survival in physiological and pathological cleft
palates’. Developmental Biology, vol. 265(1): pp. 207–218 (cit. on p. 90).
Massagué, Joan, Stacy W Blain and Roger S Lo (2000): ‘TGFß signaling in growth
control, cancer, and heritable disorders’. Cell, vol. 103(2): pp. 295–309 (cit. on p. 18).
McCollum, Melanie A and Paul T Sharpe (2001): ‘Developmental genetics and early
hominid craniodental evolution’. Bioessays, vol. 23(6): pp. 481–493 (cit. on pp. 9, 10).
McGuire, J.D., J.P. Gorski, V. Dusevich, Y. Wang and M.P. Walker (2014): ‘Type
IV Collagen is a Novel DEJ Biomarker that is Reduced by Radiotherapy’. Journal of
Dental Research, vol. 93(10): pp. 1028–1034 (cit. on p. 3).
Mead, Ben, Ann Logan, Martin Berry, Wendy Leadbeater and Ben Scheven
(2016): ‘Dental Pulp Stem Cells: A Novel Cell Therapy for Retinal and Central Nervous
System Repair: DPSC therapy for neural and retinal repair’. STEM CELLS, vol. 35(1):
pp. 61–67 (cit. on p. 81).
Meredith, Robert W, Guojie Zhang, M Thomas P Gilbert, Erich D Jarvis
and Mark S Springer (2014): ‘Evidence for a single loss of mineralized teeth in the
common avian ancestor’. Science, vol. 346(6215): p. 1254390 (cit. on p. 8).
Metsalu, Tauno and Jaak Vilo (2015): ‘ClustVis: a web tool for visualizing clustering
of multivariate data using Principal Component Analysis and heatmap’. Nucleic acids
research, vol. 43(W1): W566–W570 (cit. on p. 39).
Mina, M and EJ Kollar (1987): ‘The induction of odontogenesis in non-dental mes-
enchyme combined with early murine mandibular arch epithelium’. Archives of Oral
Biology, vol. 32(2): pp. 123–127 (cit. on pp. 19, 24, 71).
Mincione,Gabriella,Maria Carmela Di Marcantonio, Luciano Artese,Giovina
Vianale, Alessandro Piccirelli, Marcello Piccirilli, Vittoria Perrotti,
Corrado Rubini, Adriano Piattelli and Raffaella Muraro (2008): ‘Loss of
expression of TGF-ß1, TßRI, and TßRII correlates with differentiation in human oral
squamous cell carcinomas’. International journal of oncology, vol. 32(2): pp. 323–331
(cit. on p. 81).
Mitsiadis, TA and HU Luder (2011): ‘Genetic basis for tooth malformations: from mice
to men and back again’. Clinical genetics, vol. 80(4): pp. 319–329 (cit. on p. 21).
104 Bibliography
Mitsiadis, Thimios A, Yvonnick Chéraud, Paul Sharpe and Josiane Fontaine-
Pérus (2003): ‘Development of teeth in chick embryos after mouse neural crest trans-
plantations’. Proceedings of the National Academy of Sciences, vol. 100(11): pp. 6541–
6545 (cit. on p. 8).
Mitsiadis, Thimios A, Estelle Hirsinger, Urban Lendahl and Christo Goridis
(1998): ‘Delta–Notch Signaling in Odontogenesis: Correlation with Cytodifferentiation
and Evidence for Feedback Regulation’. Developmental Biology, vol. 204(2): pp. 420–431
(cit. on p. 93).
Miura, Masako, Stan Gronthos, Mingrui Zhao, Bai Lu, Larry W Fisher, Pamela
Gehron Robey and Songtao Shi (2003): ‘SHED: stem cells from human exfoliated
deciduous teeth’. Proceedings of the National Academy of Sciences, vol. 100(10): pp. 5807–
5812 (cit. on p. 30).
Moda, Aman (2011): ‘Gingival cyst of newborn’. International journal of clinical pediatric
dentistry, vol. 4(1): p. 83 (cit. on p. 95).
Montebugnoli, L, D Servidio, R A Miaton, C Prati, P Tricoci and C Melloni
(2004): ‘Poor oral health is associated with coronary heart disease and elevated systemic
inflammatory and haemostatic factors.’ Journal of clinical periodontology, vol. 31(1):
pp. 25–9 (cit. on p. 1).
Nagpal, Anjali, Karlea L Kremer, Monica A Hamilton-Bruce, Xenia Kaidonis,
Austin G Milton, Christopher Levi, Songtao Shi, Leeanne Carey, Susan
Hillier, Miranda Rose, Andrew Zacest, Parabjit Takhar and Simon A Koblar
(2016): ‘TOOTH (The Open study Of dental pulp stem cell Therapy in Humans): Study
protocol for evaluating safety and feasibility of autologous human adult dental pulp stem
cell therapy in patients with chronic disability after stroke’. International Journal of
Stroke, vol. 11(5): pp. 575–585 (cit. on p. 81).
Nakashima, Misako and Koichiro Iohara (2017): ‘Recent Progress in Translation from
Bench to a Pilot Clinical Study on Total Pulp Regeneration’. Journal of Endodontics,
vol. 43(9, Supplement): S82–S86 (cit. on p. 84).
Nanci, Antonio (2012): Ten Cate’s oral histology: Development, Structure, and Function.
8th ed. Elsevier Mosby (cit. on pp. 2, 3, 5, 6, 8, 10, 14, 16, 17).
Neubüser, Annette, Heiko Peters, Rudi Balling and Gail R Martin (1997):
‘Antagonistic Interactions between FGF and BMP Signaling Pathways: A Mechanism for
Positioning the Sites of Tooth Formation’. Cell, vol. 90(2): pp. 247–255 (cit. on p. 83).
Ning, Fang, Yunshan Guo, Juan Tang, Jing Zhou, Hongmei Zhang, Wei Lu, Yuan
Gao, Lei Wang, Duanqing Pei, Yinzhong Duan and Yan Jin (2010): ‘Differentiation
of mouse embryonic stem cells into dental epithelial-like cells induced by ameloblasts
serum-free conditioned medium’. Biochemical and biophysical research communications,
vol. 394(2): pp. 342–347 (cit. on p. 24).
Niwa, Takahiko, Yasuo Yamakoshi, Hajime Yamazaki, Takeo Karakida, Risako
Chiba, Jan C.-C. Hu, Takatoshi Nagano, Ryuji Yamamoto, James P. Simmer,
Henry C. Margolis and Kazuhiro Gomi (2018): ‘The dynamics of TGF-ß in dental
pulp, odontoblasts and dentin’. Scientific reports, vol. 8(1): p. 4450 (cit. on pp. 81, 83,
86).
105
Nogai, Hendrik, Mark Rosowski, Joachim Grün, Anika Rietz, Nils Debus, Gül
Schmidt, Carola Lauster, Michal Janitz, Andrea Vortkamp and Roland
Lauster (2008): ‘Follistatin antagonizes transforming growth factorß3-induced epithelial–
mesenchymal transition in vitro: implications for murine palatal development supported
by microarray analysis’. Differentiation, vol. 76(4): pp. 404–416 (cit. on p. 90).
Ohazama, A, SAC Modino, I Miletich and PT Sharpe (2004): ‘Stem-cell-based tissue
engineering of murine teeth’. Journal of dental research, vol. 83(7): pp. 518–522 (cit. on
pp. 24, 71).
Ono, Mitsuaki, Masamitsu Oshima, Miho Ogawa, Wataru Sonoyama, Emilio
Satoshi Hara, Yasutaka Oida, Shigehiko Shinkawa, Ryu Nakajima, Atsushi
Mine, Satoru Hayano, Satoshi Fukumoto, Shohei Kasugai, Akira Yamaguchi,
Takashi Tsuji and Takuo Kuboki (2017): ‘Practical whole-tooth restoration utilizing
autologous bioengineered tooth germ transplantation in a postnatal canine model’.
Scientific reports, vol. 7: p. 44522 (cit. on p. 25).
Ornitz, David M and Nobuyuki Itoh (2015): ‘The fibroblast growth factor signaling
pathway’. Wiley Interdisciplinary Reviews: Developmental Biology, vol. 4(3): pp. 215–266
(cit. on p. 19).
Oshima, Masamitsu, Mitsumasa Mizuno, Aya Imamura, Miho Ogawa, Masato
Yasukawa, Hiromichi Yamazaki, Ritsuko Morita, Etsuko Ikeda, Kazuhisa
Nakao, Teruko Takano-Yamamoto, Shohei Kasugai, Masahiro Saito and
Takashi Tsuji (2011): ‘Functional Tooth Regeneration Using a Bioengineered Tooth
Unit as a Mature Organ Replacement Regenerative Therapy’. PloS one, vol. 6(7): e21531
(cit. on pp. 24, 25).
Oshimori, Naoki and Elaine Fuchs (2012): ‘Paracrine TGF-ß Signaling Counterbalances
BMP-Mediated Repression in Hair Follicle Stem Cell Activation’. Cell Stem Cell, vol.
10(1): pp. 63–75 (cit. on p. 89).
Papagerakis, P, A Berdal, M Mesbah, M Peuchmaur, L Malaval, J Nydegger,
J Simmer and M Macdougall (2002): ‘Investigation of osteocalcin, osteonectin, and
dentin sialophosphoprotein in developing human teeth’. Bone, vol. 30(2): pp. 377–385
(cit. on p. 80).
Patti, Joseph M, Tomas Bremell, Danuta Krajewska-Pietrasik, Arturo Abdel-
nour, Andrzej Tarkowski, Cecilia Rydén and M Höök (1994): ‘The Staphylococ-
cus aureus collagen adhesin is a virulence determinant in experimental septic arthritis.’
Infection and immunity, vol. 62(1): pp. 152–161 (cit. on p. 65).
Paus, R (2003): ‘Biology of hair and nail’. Dermatology, vol.: pp. 1007–1032 (cit. on p. 21).
Perantoni, Alan O, Lee F Dove and Irina Karavanova (1995): ‘Basic fibroblast
growth factor can mediate the early inductive events in renal development’. Proceedings
of the National Academy of Sciences, vol. 92(10): pp. 4696–4700 (cit. on pp. 52, 82).
Peters, Heiko and Rudi Balling (1999): ‘Teeth: where and how to make them’. Trends
in Genetics, vol. 15(2): pp. 59–65 (cit. on p. 83).
Pisciotta, Alessandra, Gianluca Carnevale, Simona Meloni, Massimo Riccio,
Sara De Biasi, Lara Gibellini, Adriano Ferrari, Giacomo Bruzzesi and Anto
De Pol (2015): ‘Human Dental pulp stem cells (hDPSCs): isolation, enrichment and
106 Bibliography
comparative differentiation of two sub-populations’. BMC Developmental Biology, vol.
15(1): p. 14 (cit. on p. 42).
Rahman, M Mamunur, Jaganathan Subramani, Mallika Ghosh, Jiyeon Kim
Denninger, Kotaro Takeda, Guo-Hua Fong, Morgan E Carlson and Linda
H Shapiro (2014): ‘CD13 promotes mesenchymal stem cell-mediated regeneration of
ischemic muscle’. Frontiers in physiology, vol. 4: p. 402 (cit. on p. 79).
Ravindranath, Rajeswari MH, Wai-Yin Tam, Pablo Bringas, Valentino Santos
and Alan G Fincham (2001): ‘Amelogenin-cytokeratin 14 interaction in ameloblasts
during enamel formation’. Journal of Biological Chemistry, vol. 276(39): pp. 36586–36597
(cit. on p. 94).
Ray, Poulomi and Susan C. Chapman (2015): ‘Cytoskeletal Reorganization Drives
Mesenchymal Condensation and Regulates Downstream Molecular Signaling’. PLOS
ONE, vol. 10(8): pp. 1–24 (cit. on pp. 85, 89).
Rincon, J. C., W. G. Young and P. M. Bartold (2006): ‘The epithelial cell rests of
Malassez – a role in periodontal regeneration?’ Journal of Periodontal Research, vol.
41(4): pp. 245–252 (cit. on p. 6).
Rishikaysh, Pisal, Kapil Dev, Daniel Diaz, Wasay Mohiuddin Shaikh Qureshi,
Stanislav Filip and Jaroslav Mokry (2014): ‘Signaling Involved in Hair Follicle
Morphogenesis and Development’. International Journal of Molecular Sciences, vol. 15(1):
pp. 1647–1670 (cit. on p. 8).
Sarkar, Lena and Paul T Sharpe (1999): ‘Expression of Wnt signalling pathway genes
during tooth development’. Mechanisms of development, vol. 85(1-2): pp. 197–200 (cit. on
pp. 20, 92).
Schindelin, Johannes et al. (2012): ‘Fiji: an open-source platform for biological-image
analysis’. Nature methods, vol. 9(7): p. 676 (cit. on p. 40).
Schmidt-Ullrich, Ruth and Ralf Paus (2005): ‘Molecular principles of hair follicle
induction and morphogenesis’. Bioessays, vol. 27(3): pp. 247–261 (cit. on p. 83).
Sennett, Rachel, Zichen Wang, Amélie Rezza, Laura Grisanti, Nataly Roiter-
shtein, Cristina Sicchio, Ka Wai Mok, Nicholas J Heitman, Carlos Clavel,
Avi Ma’ayan et al. (2015): ‘An integrated transcriptome atlas of embryonic hair follicle
progenitors, their niche, and the developing skin’. Developmental cell, vol. 34(5): pp. 577–
591 (cit. on p. 90).
Shear, M. and P. Speight (2008): ‘Gingival Cyst and Midpalatal Raphe Cyst of Infants’.
Cysts of the Oral and Maxillofacial Regions. John Wiley and Sons, Ltd. Chap. 2: pp. 3–5
(cit. on p. 95).
Shi, S, PM Bartold, M Miura, BM Seo, PG Robey and S Gronthos (2005):
‘The efficacy of mesenchymal stem cells to regenerate and repair dental structures’.
Orthodontics and Craniofacial Research, vol. 8(3): pp. 191–199 (cit. on p. 41).
Shi, Songtao and Stan Gronthos (2003): ‘Perivascular niche of postnatal mesenchymal
stem cells in human bone marrow and dental pulp’. Journal of bone and mineral research,
vol. 18(4): pp. 696–704 (cit. on p. 79).
Shin, Kunyoo, John Lee, Nini Guo, James Kim, Agnes Lim, Lishu Qu, Indira
U Mysorekar and Philip A Beachy (2011): ‘Hedgehog/Wnt feedback supports
107
regenerative proliferation of epithelial stem cells in bladder’. Nature, vol. 472(7341):
pp. 110–114 (cit. on p. 8).
Sirin, Yasemin and Katalin Susztak (2012): ‘Notch in the kidney: development and
disease’. The Journal of pathology, vol. 226(2): pp. 394–403 (cit. on p. 93).
Sloan, Alastair J and Rachel J Waddington (2009): ‘Dental pulp stem cells: what,
where, how?’ International Journal of Paediatric Dentistry, vol. 19(1): pp. 61–70 (cit. on
p. 24).
Smith, David E, F Franco Del Amo and THOMAS Gridley (1992): ‘Isolation of
Sna, a mouse gene homologous to the Drosophila genes snail and escargot: its expression
pattern suggests multiple roles during postimplantation development’. Development, vol.
116(4): pp. 1033–1039 (cit. on p. 90).
Smith, Elizabeth E, Weibo Zhang, Nathan R Schiele, Ali Khademhosseini,
Catherine K Kuo and Pamela C Yelick (2017): ‘Developing a biomimetic tooth bud
model’. Journal of tissue engineering and regenerative medicine, vol. 11(12): pp. 3326–
3336 (cit. on p. 24).
Song, Hee-Kyung, So-Hyung Lee and Paul F Goetinck (2004): ‘FGF-2 signaling is
sufficient to induce dermal condensations during feather development’. Developmental
dynamics: an official publication of the American Association of Anatomists, vol. 231(4):
pp. 741–749 (cit. on pp. 52, 82).
Sonoyama, Wataru, Takayoshi Yamaza, Stan Gronthos and Songtao Shi (2007):
‘Multipotent Stem Cells in Dental Pulp’. Culture of Human Stem Cells. John Wiley and
Sons, Inc.: pp. 187–206.
Struys, Tom, Marjan Moreels, Wendy Martens, Raf Donders, Esther Wolfs
and Ivo Lambrichts (2011): ‘Ultrastructural and immunocytochemical analysis of
multilineage differentiated human dental pulp-and umbilical cord-derived mesenchymal
stem cells’. Cells Tissues Organs, vol. 193(6): pp. 366–378 (cit. on p. 80).
Su, Nan, Min Jin and Lin Chen (2014): ‘Role of FGF/FGFR signaling in skeletal
development and homeostasis: Learning from mouse models’. Bone Research, vol. 2:
p. 14003 (cit. on p. 82).
Subramanian, Aravind, Pablo Tamayo, Vamsi K. Mootha, Sayan Mukherjee, Ben-
jamin L. Ebert, Michael A. Gillette, Amanda Paulovich, Scott L. Pomeroy,
Todd R. Golub, Eric S. Lander and Jill P. Mesirov (2005): ‘Gene set enrichment
analysis: A knowledge-based approach for interpreting genome-wide expression profiles’.
Proceedings of the National Academy of Sciences, vol. 102(43): pp. 15545–15550 (cit. on
p. 39).
Suchánek, Jakub, Tomáš Soukup, Romana Ivančaková, Jana Karbanová, Věra
Hubková, Robert Pytlik and Lenka Kučerová (2007): ‘Human Dental Pulp Stem
Cells – Isolation and Long Term Cultivation’. Acta Medica (Hradec Kralove, Czech
Republic), vol. 50(3): pp. 195–201 (cit. on p. 42).
Suchánek, Jakub, Tomáš Soukup, Benjamin Visek, Romana Ivančaková, Lenka
Kučerová and Jaroslav Mokry (2009): ‘Dental pulp stem cells and their characteriz-
ation’. Biomedical Papers, vol. 153(1): pp. 31–35.
108 Bibliography
Sun, Xin, Erik N Meyers, Mark Lewandoski and Gail R Martin (1999): ‘Targeted
disruption of Fgf8 causes failure of cell migration in the gastrulating mouse embryo’.
Genes & development, vol. 13(14): pp. 1834–1846 (cit. on p. 19).
Tabata, Makoto J, Tatsushi Matsumura, Ji-Guang Liu, Satoshi Wakisaka and
Kojiro Kurisu (1996): ‘Expression of cytokeratin 14 in ameloblast-lineage cells of the
developing tooth of rat, both in vivo and in vitro’. Archives of oral biology, vol. 41(11):
pp. 1019–1027 (cit. on p. 94).
Tabata, Makoto, KKim, J G Liu, K Yamashita, TMatsumura, J Kato, Mitsumasa
Iwamoto, Satoshi Wakisaka, Kunio Matsumoto, T Nakamura, M Kumegawa
and K Kurisu (1996): ‘Hepatocyte growth factor is involved in the morphogenesis of
tooth germ in murine molars’. Development, vol. 122(4): pp. 1243–1251 (cit. on p. 82).
Takahashi, Kazuto, Yoshiaki Kishi and Syngcuk Kim (1982): ‘A scanning electron
microscope study of the blood vessels of dog pulp using corrosion resin casts’. Journal of
endodontics, vol. 8(3): pp. 131–135 (cit. on p. 4).
Takigawa-Imamura, Hisako, Ritsuko Morita, Takafumi Iwaki, Takashi Tsuji
and Kenichi Yoshikawa (2015): ‘Tooth germ invagination from cell–cell interaction:
Working hypothesis on mechanical instability’. Journal of theoretical biology, vol. 382:
pp. 284–291 (cit. on p. 9).
Tatullo, M, G Falisi, M Amantea, C Rastelli, F Paduano and M Marrelli (2015):
‘Dental pulp stem cells and human periapical cyst mesenchymal stem cells in bone tissue
regeneration: comparison of basal and osteogenic differentiated gene expression of a
newly discovered mesenchymal stem cell lineage’. J Biol Regul Homeost Agents, vol.
29(3): pp. 713–8 (cit. on p. 84).
Thesleff, Irma (2018): ‘From understanding tooth development to bioengineering of
teeth’. European journal of oral sciences, vol. 126(Suppl.1): pp. 67–71 (cit. on p. 17).
Thomas, Bethan L, Jen Kuei Liu, JL Rubenstein and Paul T Sharpe (2000):
‘Independent regulation of Dlx2 expression in the epithelium and mesenchyme of the
first branchial arch’. Development, vol. 127(2): pp. 217–224 (cit. on p. 19).
Tsunematsu, Takaaki et al. (2016): ‘Human odontogenic epithelial cells derived from
epithelial rests of Malassez possess stem cell properties’. Laboratory Investigation, vol.
96(10): pp. 1063–1075 (cit. on p. 6).
Vaahtokari, A., S. Vainio and I. Thesleff (1991): ‘Associations between transforming
growth factor beta 1 RNA expression and epithelial-mesenchymal interactions during
tooth morphogenesis’. Development, vol. 113(3): pp. 985–994 (cit. on pp. 81, 88, 90).
Vainio, Seppo, Irina Karavanova, Adrian Jowett and Irma Thesleff (1993a):
‘Identification of BMP-4 as a signal mediating secondary induction between epithelial
and mesenchymal tissues during early tooth development’. Cell, vol. 75(1): pp. 45–58
(cit. on p. 19).
– (1993b): ‘Identification of BMP-4 as a signal mediating secondary induction between
epithelial and mesenchymal tissues during early tooth development’. Cell, vol. 75(1):
pp. 45–58.
Volponi, Ana Angelova, Lucia K Zaugg, Vitor Neves, Yang Liu and Paul T
Sharpe (2018): ‘Tooth Repair and Regeneration’. Current Oral Health Reports, vol. 5(4):
pp. 295–303 (cit. on pp. 23, 25).
109
Wang, Kuo-Chen, Thomas T. Egelhoff, Arnold I. Caplan, Jean F. Welter
and Harihara Baskaran (2018): ‘ROCK Inhibition Promotes the Development of
Chondrogenic Tissue by Improved Mass Transport’. Tissue Engineering Part A, vol.
24(15-16): pp. 1218–1227 (cit. on p. 88).
Wang, Xiu-Ping, Marika Suomalainen, Szabolcs Felszeghy, Laura C Zelarayan,
Maria T Alonso, Maksim V Plikus, Richard L Maas, Cheng-Ming Chuong,
Thomas Schimmang and Irma Thesleff (2007): ‘An integrated gene regulatory
network controls stem cell proliferation in teeth’. PLoS biology, vol. 5(6): e159 (cit. on
p. 19).
Wang, Xiu-Ping, Marika Suomalainen, Carolina J Jorgez, Martin M Matzuk,
Sabine Werner and Irma Thesleff (2004): ‘Follistatin regulates enamel patterning
in mouse incisors by asymmetrically inhibiting BMP signaling and ameloblast differenti-
ation’. Developmental cell, vol. 7(5): pp. 719–730 (cit. on p. 19).
Xu, Jian, Samy Lamouille and Rik Derynck (2009): ‘TGF-ß-induced epithelial to
mesenchymal transition’. Cell research, vol. 19(2): p. 156 (cit. on p. 89).
Yoshida, Mitsuyoshi and Yasumasa Akagawa (2011): ‘The relationship between tooth
loss and cerebral stroke’. Japanese Dental Science Review, vol. 47(2): pp. 157–160 (cit. on
p. 1).
Young, Conan S, Harutsugi Abukawa, Rose Asrican, Michael Ravens, Maria J
Troulis, Leonard B Kaban, Joseph P Vacanti and Pamela C Yelick (2005):
‘Tissue-engineered hybrid tooth and bone’. Tissue engineering, vol. 11(9-10): pp. 1599–
1610 (cit. on p. 24).
Zeisberg, Michael, Jun-ichi Hanai, Hikaru Sugimoto, Tadanori Mammoto, David
Charytan, Frank Strutz and Raghu Kalluri (2003): ‘BMP-7 counteracts TGF-ß1–
induced epithelial-to-mesenchymal transition and reverses chronic renal injury’. Nature
medicine, vol. 9(7): p. 964 (cit. on pp. 18, 81, 82, 89).
Zhang, Siyuan, Patricia Buttler-Buecher, Bernd Denecke, Victor E Arana-
Chavez and Christian Apel (2018): ‘A comprehensive analysis of human dental pulp
cell spheroids in a three-dimensional pellet culture system’. Archives of oral biology, vol.
91: pp. 1–8 (cit. on p. 23).
Zhang, Weibo, B Vazquez, D Oreadi and PC Yelick (2017): ‘Decellularized tooth
bud scaffolds for tooth regeneration’. Journal of dental research, vol. 96(5): pp. 516–523
(cit. on p. 25).
Zhao, Hu, Jifan Feng, Kerstin Seidel, Songtao Shi, Ophir Klein, Paul Sharpe
and Yang Chai (2014): ‘Secretion of shh by a neurovascular bundle niche supports
mesenchymal stem cell homeostasis in the adult mouse incisor’. Cell stem cell, vol. 14(2):
pp. 160–173 (cit. on p. 20).
Zheng, Yunfei, Jinglei Cai, Andrew Paul Hutchins, Lingfei Jia, Pengfei Liu,
Dandan Yang, Shubin Chen, Lihong Ge, Duanqing Pei and Shicheng Wei (2016):
‘Remission for Loss of Odontogenic Potential in a New Micromilieu In Vitro’. PLOS
ONE, vol. 11(4): pp. 1–16 (cit. on p. 84).
Zhu, Yihong, Mikael Nilsson and Karin Sundfeldt (2010): ‘Phenotypic plasticity of
the ovarian surface epithelium: TGF-ß1 induction of epithelial to mesenchymal transition
(EMT) in vitro’. Endocrinology, vol. 151(11): pp. 5497–5505 (cit. on p. 81).
110 Bibliography
Zouvelou, Vasiliki, Hans-Ulrich Luder, Thimios A. Mitsiadis and Daniel Graf
(2009): ‘Deletion of BMP7 affects the development of bones, teeth, and other ectodermal
appendages of the orofacial complex’. Journal of Experimental Zoology Part B: Molecular
and Developmental Evolution, vol. 312B(4): pp. 361–374 (cit. on p. 81).
List of Figures
1.1 Tooth anatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Schematic representation of odontoblasts . . . . . . . . . . . . . . . . . . . . 2
1.3 Histologic overview of pulp tissue . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Capillary network in the pulp . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Schematic of periodontium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Alveolar bone lining the tooth socket . . . . . . . . . . . . . . . . . . . . . . 6
1.8 Histologic image of the dental placode . . . . . . . . . . . . . . . . . . . . . 8
1.9 Bud-to-cap transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.10 Odontogenic homeobox code . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.11 Histologic image of the dental bell stage . . . . . . . . . . . . . . . . . . . . 10
1.12 Histologic image of the cervical loop . . . . . . . . . . . . . . . . . . . . . . 11
1.13 Histologic images from human late bell stage . . . . . . . . . . . . . . . . . 12
1.14 Schematic illustration of enamel development . . . . . . . . . . . . . . . . . 13
1.15 Ultrastructural features of human enamel . . . . . . . . . . . . . . . . . . . 14
1.16 Section showing the region of mineralized dentin . . . . . . . . . . . . . . . 15
1.17 Schematic representation of tooth formation in the jaw . . . . . . . . . . . . 16
1.18 Regulation of tooth formation by reciprocal interactions . . . . . . . . . . . 17
1.19 Mechanochemical control of condensation . . . . . . . . . . . . . . . . . . . 23
1.20 Vision of autologous tooth regeneration . . . . . . . . . . . . . . . . . . . . 25
3.1 Histological overview of used pulp tissue . . . . . . . . . . . . . . . . . . . . 41
3.2 Isolation of DPCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3 CFSE profile of DPCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.4 Surface Marker Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.5 Surface Marker Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.6 Gene expression profile of DPCs over passages . . . . . . . . . . . . . . . . . 44
3.7 Marker gene expression of adipogenic differentiation . . . . . . . . . . . . . 46
3.8 Oil Red O staining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.9 Marker gene expression of osteogenic differentiation . . . . . . . . . . . . . . 47
3.10 Alizarin Red staining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.11 Marker gene expression of neurogenic differentiation . . . . . . . . . . . . . 48
3.12 Staining for NESTIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.13 Time-lapse video of in vitro condensation . . . . . . . . . . . . . . . . . . . 50
3.14 Transcript levels of marker genes during in vitro condensation . . . . . . . . 51
3.15 Gene expression of marker genes during early phase of condensation . . . . 53
3.16 Quality of the RNA Seq data . . . . . . . . . . . . . . . . . . . . . . . . . . 54
111
112 List of Figures
3.17 Heatmap of all expressed genes . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.18 Heatmap of all differentially expressed genes . . . . . . . . . . . . . . . . . . 55
3.19 Clustered heatmap for KEGG – TGFß signaling pathway. . . . . . . . . . 56
3.20 KEGG – TGFß signaling pathway gene panel . . . . . . . . . . . . . . . . . 57
3.21 KEGG – TGFß signaling pathway map . . . . . . . . . . . . . . . . . . . . 58
3.22 Clustered heatmap for KEGG – MAPK signaling pathway . . . . . . . . . . 59
3.23 KEGG – MAPK signaling pathway map . . . . . . . . . . . . . . . . . . . . 59
3.24 Clustered heatmap for RHOA signaling molecules . . . . . . . . . . . . . . . 60
3.25 RHOA signaling pathway gene panel . . . . . . . . . . . . . . . . . . . . . . 61
3.26 Clustered heatmap for KEGG –WNT signaling pathway. . . . . . . . . . . 62
3.27 KEGG – WNT signaling pathway map . . . . . . . . . . . . . . . . . . . . . 63
3.28 Clustered heatmap for KEGG –NOTCH signaling pathway. . . . . . . . . . 64
3.29 KEGG – NOTCH signaling pathway map . . . . . . . . . . . . . . . . . . . 65
3.30 Live-imaging of collagen production . . . . . . . . . . . . . . . . . . . . . . 66
3.31 Reporter construct functionality test . . . . . . . . . . . . . . . . . . . . . . 68
3.32 Kinetik of condensation with constitutively eGFP-expressing DPCs (CMV
control vector) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.33 Kinetic of condensation with MAPK-responsive eGFP-expressing DPCs . . 69
3.34 Graphs of fluorescence intensities of MAPK-reporter vector and control vector 70
3.35 Inhibition of condensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.36 Co-culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.37 Co-culture of red and green labeled cells . . . . . . . . . . . . . . . . . . . . 73
3.38 Histology of 4-day condensates . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.39 Histology of 4 day co-culture condensates . . . . . . . . . . . . . . . . . . . 74
3.40 Histology of 2 week co-culture condensates . . . . . . . . . . . . . . . . . . . 75
3.41 Histology of 4 week co-culture condensates . . . . . . . . . . . . . . . . . . . 76
3.42 Long-term culture of co-culture aggregates . . . . . . . . . . . . . . . . . . . 77
4.1 F-actin bridges in mesenchymal condensation . . . . . . . . . . . . . . . . . 85
4.2 Gingival cysts in an infants . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.1 Comparison of DPCs with MSCs . . . . . . . . . . . . . . . . . . . . . . . . 115
5.2 FACS analysis of reporter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.3 Higher magnification of condensation with CytD administration . . . . . . . 117
5.4 Viability of condensated cells in inhibition experiment . . . . . . . . . . . . 118
5.5 Expression of cytokeratins in human skin . . . . . . . . . . . . . . . . . . . 118
5.6 Vector maps of reporter assay constructs . . . . . . . . . . . . . . . . . . . . 123
5.7 GSEA on the classical GO terms and hallmark gene panel . . . . . . . . . . 124
5.8 GSEA on specific pathway gene panels . . . . . . . . . . . . . . . . . . . . . 125
5.9 TGF-ß signaling agonists and antagonists . . . . . . . . . . . . . . . . . . . 126
5.10 KEGG – FGF signaling pathway gene panel . . . . . . . . . . . . . . . . . . 126
5.11 KEGG – MAPK signaling pathway gene panel . . . . . . . . . . . . . . . . 127
5.12 KEGG – WNT signaling pathway gene panel . . . . . . . . . . . . . . . . . 128
5.13 KEGG – NOTCH signaling pathway gene panel . . . . . . . . . . . . . . . . 129
List of Tables
2.1 Materials for cell culture applications . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Materials for molecular biology applications . . . . . . . . . . . . . . . . . . . 28
2.3 Antibodies for immunhistochemistry . . . . . . . . . . . . . . . . . . . . . . . 34
2.4 Primer pairs used for qPCR . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.1 Differentially expressed genes . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
113

CHAPTER 5
Appendix A
5.1 Supplemental figures
Figure 5.1: Comparison of MSC characteristics between bone marrow mesenchymal cells
(bmMSCs) and dental pulp cells (DPCs). (A) Expression profile of MSC surface markers
revealed that both cell population exhibited the same profile except for the expression of CD146,
the perivascular marker, which was higher in DPCs (N=4). (B) Representative images of
adipogenic and osteogenic differentiation of bmMSC (negative control in bottom right corner),
assessed by Oil Red O and Alizarin Red staining, respectively (N=4).
115
116 Chapter 5 Appendix A
Figure 5.2: FACS analysis on reporter functionality. DPCs in monolayer were transduced
with the respective reporter constructs and stimulated with the denoted stimuli at two different
concentrations. Percentage of positive cells, as well as their geometric mean fluorescence of the
eGFP signal was measured.
5.1 Supplemental figures 117
Figure 5.3: Microscopic analysis at higher magnifcation after 5 hours of condensation with
administration of Cytochalasin D at high concentration (2.5 µM; middle) or low concentration
(0.25 µM; right) compared to control culture (left). Representative images, N=2
118 Chapter 5 Appendix A
Figure 5.4: Viability analysis of cell after 24 hours of condensation. Depicted are the FSC-
A/SSC-A gate as well as the 7-AAD (PerCP-Vio700-A) gate to measure viabiltiy. Viable cells
do not take up the 7-AAD label and appear negative in the FACS plot. In this experiment,
cells were treated with three different small molecular inhibitors (Cytochalasin D, LY294002,
PTX), no inhibitor as control or with ethanol directly before FACS measurement to show
functionality of the 7-AAD dye. Representative plots, N=2
Figure 5.5: Human skin biopses were immunohistologically stained for CK8/18, CK19, CK15
and CK14. Blue colour represents DAPI nuclear counterstaining. These pictures were kindly
provided by Anna-Klara Amler
5.1 Supplemental figures 119
120 Chapter 5 Appendix A
5.1 Supplemental figures 121
122 Chapter 5 Appendix A
5.1 Supplemental figures 123
Figure 5.6: Vector maps of reporter assay constructs
124 Chapter 5 Appendix A
Figure 5.7: Differentially expressed genes were analyzed by Gene Set enrichment analysis.
The top 20 over-represented gene ontology (GO) terms for each top level term as well as for
the hall mark gene set is depicted. The diagrams plot the total number of genes overlapping
in the differentially expressed genes and the ontology list (upper x-axis, # genes, black bars).
Furthermore, the percentage of overlap (related to the group size to the GO term) is shown
(lower x-axis, red dots) is depicted.
5.1 Supplemental figures 125
Figure 5.8: Differentially expressed genes were analyzed by Gene Set enrichment analysis. The
top 20 over-represented pathway terms acoording to KEGG, Reactome, Biocarta or canonical
pathway gene panel is depicted. The diagrams plot the total number of genes overlapping
in the differentially expressed genes and the gene panel (upper x-axis, # genes, black bars).
Furthermore, the percentage of overlap (related to the respective group size) is shown (lower
x-axis, red dots) is depicted.
126 Chapter 5 Appendix A
Figure 5.9: TGF-ß signaling agonists and antagonists. RPM values of inhibiting and
supporting molecules of the TGF-ß signal transduction pathway are depicted. The gene set
is reduced to genes with at least moderate gene expression level in the DPC condensation
samples and the by genes with an at least two fold difference in the RPM values of both
samples. Dashed line marks the level of highly expressed genes (RPM>100); dotted line marks
the level of moderately expressed genes (RPM>5). Statistical significance was calculated from
DESEQ2 analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
Figure 5.10: RPM values of key molecules of FGF signal transduction pathway are depicted.
Gene set was reduced to lignads, receptors and adaptor molecules that elicit multiple signaling
pathways . . Dashed line marks the level of highly expressed genes (RPM>100); dotted line
marks the level of moderately expressed genes (RPM>5). Statistical significance was calculated
from DESEQ2 analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
5.1 Supplemental figures 127
Figure 5.11: RPM values of MAPK signal transduction pathway-associated genes are depicted.
The KEGG gene subset comprises 267 genes. The diagram is reduced to genes with at least
moderate gene expression level in the DPC condensation samples and/or genes with an at
least two-fold difference in the RPM values of both samples, resulting in 166 genes. Dashed
line marks the level of highly expressed genes (RPM>100); dotted line marks the level of
moderately expressed genes (RPM>5). Statistical significance was calculated from DESEQ2
analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
128 Chapter 5 Appendix A
Figure 5.12: RPM values of WNT signal transduction pathway-associated genes are depicted.
The KEGG gene subset comprises 151 genes. The diagram is reduced to genes with at least
moderate gene expression level in the DPC condensation samples and/or genes with an at
least two-fold difference in the RPM values of both samples, resulting in 100 genes. Dashed
line marks the level of highly expressed genes (RPM>100); dotted line marks the level of
moderately expressed genes (RPM>5). Statistical significance was calculated from DESEQ2
analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
5.1 Supplemental figures 129
Figure 5.13: RPM values of NOTCH signal transduction pathway-associated genes are
depicted. The diagram is reduced to genes with at least moderate gene expression level in
the DPC condensation samples and/or genes with an at least two-fold difference in the RPM
values of both samples. Dashed line marks the level of highly expressed genes (RPM>100);
dotted line marks the level of moderately expressed genes (RPM>5). Statistical significance
was calculated from DESEQ2 analysis: *** - P < 0.001; ** - P < 0.01; * - P < 0.05.
130 Chapter 5 Appendix A
5.2 Supplemental videos
Supplemental videos are available on the attached CD-ROM.
Z28P4_30h_low:
Video of ultra-low attachment condensation of DPCs (time-lapse of 30 hours)
CNA_15h:
Condensing DPCs with CNA-eGFP administered to the medium (time-lapse of 15 hours)
pll3.7_Z38P4_16h:
Condensing DPCs transduced with pLL3.7 (time-lapse of 16 hours)
WNT_Z38P7_24h:
Condensing DPCs transduced with pLL3.7-WNT-RE (time-lapse of 24 hours)
TGF_Z38P6_22h:
Condensing DPCs transduced with pLL3.7-TGFß-RE (time-lapse of 22 hours)
MAPK_Z38P6_24h:
Condensing DPCs transduced with pLL3.7-MAPK-RE (time-lapse of 24 hours)
5.3 List of differentially expressed genes
Table 5.1: Differentially expressed genes between human DPCs in monolayer (ML) or six
hours in condensation (6h) according to DESeq2 algorithm (|𝐹𝐶| ≥ 1.5, nominal P value<
0.05). Two biological replicates were performed. Upregulated genes are highlighted in red and
downregulated in blue.
Reads per Million (RPM)
GeneID Gene Symbol ZZ29ML ZZ29 6h ZZ31ML ZZ31 6h log2FC Pvalue adj.Pvalue
7043 TGFB3 12.578 397.953 0.000 330.028 3.891 9.00E-17 3.08E-13
2847 MCHR1 0.000 80.365 0.000 69.663 3.354 1.46E-11 3.00E-08
1909 EDNRA 4.193 110.381 0.000 82.725 3.241 2.41E-11 4.13E-08
50509 COL5A3 129.977 1572.446 97.270 1565.674 3.184 1.03E-17 5.28E-14
4071 TM4SF1 28.301 506.397 38.709 459.775 3.183 2.12E-16 5.45E-13
10221 TRIB1 2.096 71.651 5.955 102.753 2.909 1.32E-09 1.69E-06
57801 HES4 2.096 50.349 2.978 82.725 2.836 9.45E-09 9.71E-06
54498 SMOX 102.723 1222.906 45.657 560.787 2.670 2.56E-09 2.92E-06
4773 NFATC2 2.096 54.222 0.000 40.927 2.647 1.94E-07 9.80E-05
3400 ID4 29.350 336.953 34.739 235.112 2.541 4.71E-10 6.92E-07
3696 ITGB8 9.434 63.905 7.940 146.292 2.480 1.37E-07 7.80E-05
84632 AFAP1L2 8.386 87.143 9.925 103.624 2.462 5.59E-08 4.10E-05
4741 NEFM 11.530 152.016 3.970 62.697 2.453 5.48E-07 2.45E-04
3720 JARID2 18.868 151.048 13.896 141.938 2.427 1.69E-08 1.45E-05
57561 ARRDC3 16.771 172.349 34.739 271.685 2.417 1.46E-08 1.36E-05
8877 SPHK1 44.024 444.429 27.791 203.764 2.393 6.83E-08 4.68E-05
112399 EGLN3 0.000 20.333 1.985 69.663 2.374 4.66E-06 1.84E-03
5139 PDE3A 2.096 61.000 0.000 21.770 2.313 8.23E-06 2.68E-03
84959 UBASH3B 15.723 134.587 14.888 108.848 2.273 2.00E-07 9.80E-05
51129 ANGPTL4 46.121 509.302 8.933 148.034 2.240 7.20E-06 2.55E-03
6542 SLC7A2 10.482 45.508 11.911 191.573 2.211 6.05E-06 2.22E-03
5.3 List of differentially expressed genes 131
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
56895 AGPAT4 25.157 119.095 12.903 168.933 2.174 9.14E-07 3.91E-04
340075 ARSI 0.000 48.413 3.970 29.607 2.164 2.49E-05 6.40E-03
57471 ERMN 2.096 26.143 0.000 37.444 2.162 3.01E-05 7.53E-03
6303 SAT1 71.277 344.699 68.486 502.444 2.144 2.19E-08 1.73E-05
100499467 LINC00673 8.386 82.302 18.858 122.781 2.127 3.06E-06 1.26E-03
84935 MEDAG 2.096 118.127 0.000 6.966 2.119 5.23E-05 1.12E-02
2247 FGF2 85.952 574.175 87.344 402.303 2.063 1.06E-07 6.71E-05
8835 SOCS2 31.446 168.476 10.918 136.713 2.053 7.52E-06 2.58E-03
9077 DIRAS3 17.819 114.254 1.985 80.983 2.040 4.43E-05 9.90E-03
23551 RASD2 0.000 6.778 1.985 94.916 2.039 9.95E-05 1.89E-02
26207 PITPNC1 63.940 316.619 58.560 350.927 2.025 1.11E-07 6.71E-05
9945 GFPT2 13.627 138.460 22.829 90.562 1.991 8.55E-06 2.68E-03
196740 VSTM4 67.085 367.937 33.747 233.371 1.969 5.71E-06 2.17E-03
643837 LINC01128 12.578 61.000 11.911 100.140 1.951 2.20E-05 5.81E-03
55326 AGPAT5 6.289 34.857 10.918 104.494 1.941 8.12E-05 1.66E-02
3754 KCNF1 4.193 29.048 3.970 47.022 1.910 1.58E-04 2.80E-02
84627 ZNF469 54.506 272.080 25.806 177.640 1.871 2.20E-05 5.81E-03
6615 SNAI1 10.482 66.810 8.933 54.860 1.845 9.32E-05 1.84E-02
392617 ELFN1 11.530 104.572 3.970 29.607 1.845 2.87E-04 4.46E-02
26508 HEYL 0.000 14.524 0.000 28.736 1.819 5.17E-04 6.72E-02
5376 PMP22 25.157 170.413 27.791 97.528 1.787 4.66E-05 1.02E-02
25861 WHRN 19.916 95.857 4.963 67.051 1.779 2.68E-04 4.38E-02
29923 HILPDA 7.337 52.286 10.918 55.730 1.768 2.06E-04 3.41E-02
1407 CRY1 6.289 48.413 7.940 41.798 1.755 3.30E-04 4.91E-02
3280 HES1 4.193 50.349 1.985 15.674 1.749 7.83E-04 9.46E-02
19 ABCA1 50.314 196.556 49.627 236.854 1.724 1.12E-05 3.19E-03
79870 BAALC 30.398 158.794 53.598 207.247 1.695 3.92E-05 9.15E-03
353500 BMP8A 2.096 21.302 4.963 38.315 1.688 1.07E-03 1.17E-01
10171 RCL1 6.289 16.460 2.978 64.438 1.684 1.12E-03 1.20E-01
83716 CRISPLD2 46.121 296.286 13.896 72.275 1.663 8.54E-04 9.74E-02
26353 HSPB8 2.096 12.587 3.970 47.893 1.653 1.54E-03 1.52E-01
50700 RDH8 0.000 17.429 0.000 17.416 1.636 1.78E-03 1.66E-01
3624 INHBA 51.362 234.318 25.806 123.652 1.632 3.06E-04 4.69E-02
220323 OAF 111.109 293.381 51.613 419.719 1.630 1.26E-04 2.27E-02
54541 DDIT4 201.254 888.858 72.456 409.270 1.621 4.27E-04 5.93E-02
182 JAG1 34.591 100.698 9.925 120.169 1.593 7.37E-04 9.12E-02
3554 IL1R1 274.628 1098.001 386.102 1311.405 1.592 8.87E-06 2.68E-03
5054 SERPINE1 801.872 2897.987 704.710 2586.236 1.583 9.80E-06 2.88E-03
9419 CRIPT 0.000 19.365 1.985 17.416 1.577 2.61E-03 2.12E-01
169792 GLIS3 50.314 175.254 8.933 137.584 1.574 1.25E-03 1.31E-01
3638 INSIG1 17.819 139.429 17.866 40.056 1.571 1.28E-03 1.33E-01
132 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
387763 C11ORF96 76.518 235.286 54.590 269.073 1.564 8.26E-05 1.66E-02
1942 EFNA1 0.000 7.746 1.985 35.702 1.559 2.89E-03 2.23E-01
145581 LRFN5 0.000 7.746 1.985 35.702 1.559 2.89E-03 2.23E-01
6932 TCF7 37.735 93.921 30.769 194.185 1.558 3.58E-04 5.18E-02
8553 BHLHE40 57.651 301.127 63.523 168.933 1.555 1.97E-04 3.33E-02
166647 adgra3 50.314 161.699 48.635 215.084 1.553 9.85E-05 1.89E-02
1846 DUSP4 9.434 68.746 3.970 19.157 1.540 2.78E-03 2.22E-01
284018 C17orf58 5.241 38.730 9.925 40.056 1.539 1.84E-03 1.66E-01
4216 MAP3K4 31.446 125.873 23.821 96.657 1.536 4.03E-04 5.75E-02
3775 KCNK1 5.241 25.175 3.970 32.219 1.526 2.83E-03 2.22E-01
2012 EMP1 462.255 2153.399 515.133 1311.405 1.523 7.24E-05 1.52E-02
7223 TRPC4 0.000 21.302 5.955 30.478 1.514 3.62E-03 2.62E-01
7779 SLC30A1 44.024 164.603 27.791 118.427 1.513 4.74E-04 6.40E-02
5163 PDK1 18.868 53.254 24.814 134.972 1.503 1.16E-03 1.23E-01
8013 NR4A3 13.627 74.556 1.985 26.994 1.498 3.73E-03 2.64E-01
83786 FRMD8 68.133 337.921 64.516 162.837 1.496 4.82E-04 6.40E-02
26035 GLCE 19.916 61.000 38.709 195.927 1.495 1.49E-03 1.49E-01
7378 UPP1 11.530 62.937 3.970 25.253 1.483 3.61E-03 2.62E-01
9052 GPRC5A 0.000 11.619 0.000 19.157 1.483 4.48E-03 2.90E-01
114088 TRIM9 0.000 13.556 0.000 16.545 1.479 4.60E-03 2.95E-01
136 ADORA2B 3.145 16.460 0.000 21.770 1.464 5.19E-03 3.15E-01
467 ATF3 9.434 45.508 3.970 26.994 1.460 4.01E-03 2.75E-01
56937 PMEPA1 49.265 218.826 69.478 187.219 1.448 2.77E-04 4.44E-02
6424 SFRP4 11.530 50.349 15.881 60.955 1.440 2.00E-03 1.76E-01
5142 PDE4B 35.639 107.476 9.925 87.949 1.427 2.82E-03 2.22E-01
10184 LHFPL2 47.169 152.016 68.486 244.691 1.420 4.86E-04 6.40E-02
6383 SDC2 138.362 312.746 167.741 761.067 1.419 5.99E-04 7.65E-02
85463 ZC3H12C 28.301 105.540 17.866 70.534 1.414 1.85E-03 1.66E-01
10579 TACC2 13.627 61.000 7.940 34.831 1.413 3.99E-03 2.75E-01
54852 PAQR5 10.482 38.730 22.829 102.753 1.410 3.50E-03 2.59E-01
23313 KIAA0930 109.013 259.492 77.419 366.601 1.399 4.12E-04 5.79E-02
6095 RORA 24.109 90.048 19.851 70.534 1.394 1.78E-03 1.66E-01
10560 SLC19A2 2.096 16.460 2.978 20.899 1.392 7.80E-03 4.20E-01
29116 MYLIP 8.386 46.476 17.866 57.472 1.379 3.79E-03 2.67E-01
94031 HTRA3 95.386 194.619 98.262 470.225 1.376 1.04E-03 1.16E-01
23401 FRAT2 3.145 28.079 0.000 9.579 1.369 8.91E-03 4.53E-01
23413 NCS1 85.952 295.318 110.173 313.483 1.364 2.85E-04 4.46E-02
4291 MLF1 0.000 21.302 5.955 22.640 1.351 9.69E-03 4.76E-01
7803 PTP4A1 55.555 195.587 61.538 174.157 1.351 6.29E-04 7.88E-02
23254 KAZN 16.771 40.667 10.918 67.921 1.346 4.71E-03 2.97E-01
6446 SGK1 36.687 133.619 23.821 79.242 1.345 2.77E-03 2.22E-01
5.3 List of differentially expressed genes 133
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
406 ARNTL 4.193 39.698 9.925 25.253 1.344 8.06E-03 4.27E-01
80018 NAA25 4.193 43.571 16.873 43.539 1.323 7.87E-03 4.21E-01
2903 GRIN2A 0.000 24.206 0.000 5.225 1.323 1.04E-02 4.93E-01
687 KLF9 41.928 115.222 26.799 111.461 1.314 2.27E-03 1.93E-01
9586 CREB5 4.193 22.270 1.985 16.545 1.312 1.21E-02 5.31E-01
10570 DPYSL4 16.771 50.349 27.791 106.236 1.301 4.83E-03 2.99E-01
23704 KCNE4 6.289 57.127 0.000 5.225 1.300 1.28E-02 5.39E-01
2114 ETS2 41.928 143.302 6.948 67.051 1.299 9.76E-03 4.77E-01
153478 PLEKHG4B 0.000 22.270 7.940 26.994 1.298 1.28E-02 5.39E-01
51312 SLC25A37 81.759 241.095 45.657 168.933 1.295 2.46E-03 2.07E-01
55 ACPP 2.096 13.556 0.000 14.803 1.293 1.33E-02 5.49E-01
92104 TTC30A 2.096 13.556 5.955 32.219 1.288 1.36E-02 5.59E-01
7090 TLE3 26.205 99.730 42.680 114.944 1.280 2.58E-03 2.12E-01
57464 STRIP2 4.193 21.302 3.970 20.899 1.278 1.38E-02 5.64E-01
89797 NAV2 14.675 85.206 46.650 120.169 1.277 5.51E-03 3.26E-01
11138 TBC1D8 40.880 117.159 15.881 80.983 1.271 6.41E-03 3.62E-01
10808 HSPH1 81.759 288.540 76.426 184.607 1.270 1.79E-03 1.66E-01
85450 ITPRIP 88.049 336.953 67.493 152.388 1.267 3.70E-03 2.64E-01
6591 SNAI2 67.085 154.921 63.523 236.854 1.265 1.56E-03 1.52E-01
23462 HEY1 0.000 5.810 0.000 20.899 1.260 1.45E-02 5.73E-01
64131 XYLT1 95.386 297.254 112.158 290.843 1.253 8.47E-04 9.74E-02
84870 RSPO3 2.096 30.984 0.000 3.483 1.246 1.58E-02 6.00E-01
5159 PDGFRB 569.172 1950.065 515.133 1146.826 1.245 1.43E-03 1.45E-01
10758 TRAF3IP2 25.157 94.889 9.925 37.444 1.239 1.22E-02 5.33E-01
51010 EXOSC3 6.289 21.302 0.000 20.899 1.237 1.81E-02 6.44E-01
7042 TGFB2 100.627 308.873 109.180 278.652 1.235 1.08E-03 1.17E-01
4783 NFIL3 37.735 104.572 11.911 70.534 1.234 9.95E-03 4.84E-01
1956 EGFR 78.615 245.937 45.657 138.455 1.233 4.88E-03 3.00E-01
24145 PANX1 124.736 377.619 101.240 269.944 1.232 1.86E-03 1.66E-01
10397 NDRG1 198.109 595.477 121.091 350.927 1.228 3.33E-03 2.51E-01
2774 GNAL 2.096 5.810 0.000 24.382 1.226 1.82E-02 6.44E-01
23186 RCOR1 25.157 62.937 27.791 101.882 1.225 5.49E-03 3.26E-01
3717 JAK2 8.386 40.667 23.821 71.405 1.224 1.13E-02 5.17E-01
2300 FOXL1 6.289 24.206 5.955 26.994 1.221 1.66E-02 6.13E-01
6515 SLC2A3 25.157 119.095 44.665 89.691 1.221 5.06E-03 3.09E-01
57214 CEMIP 271.483 908.223 98.262 324.803 1.219 9.64E-03 4.76E-01
84803 GPAT3 0.000 14.524 0.000 8.708 1.218 1.82E-02 6.44E-01
55647 RAB20 0.000 13.556 1.985 12.191 1.218 1.94E-02 6.63E-01
1832 DSP 3.145 25.175 11.911 36.573 1.214 1.75E-02 6.28E-01
64651 CSRNP1 12.578 65.841 18.858 38.315 1.207 1.14E-02 5.19E-01
25799 ZNF324 4.193 19.365 0.993 13.933 1.204 2.15E-02 7.03E-01
134 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
23446 SLC44A1 123.687 259.492 116.128 411.882 1.203 1.86E-03 1.66E-01
114880 OSBPL6 2.096 22.270 7.940 23.511 1.198 2.10E-02 6.96E-01
85027 SMIM3 34.591 138.460 44.665 91.433 1.197 5.23E-03 3.16E-01
55893 ZNF395 85.952 177.191 101.240 357.893 1.194 3.40E-03 2.53E-01
6533 SLC6A6 107.964 329.207 47.642 155.000 1.189 9.19E-03 4.61E-01
3570 IL6R 11.530 53.254 2.978 14.803 1.188 2.24E-02 7.06E-01
23433 RHOQ 37.735 141.365 61.538 136.713 1.188 3.91E-03 2.73E-01
4163 MCC 49.265 129.746 38.709 121.910 1.188 4.25E-03 2.82E-01
57509 MTUS1 0.000 12.587 0.000 9.579 1.185 2.14E-02 NA
9700 ESPL1 12.578 36.794 10.918 42.669 1.184 1.42E-02 5.70E-01
7539 ZFP37 2.096 13.556 1.985 13.933 1.179 2.41E-02 7.43E-01
6197 RPS6KA3 51.362 142.333 87.344 246.433 1.171 4.77E-03 2.97E-01
864 RUNX3 56.603 169.445 48.635 124.522 1.170 4.77E-03 2.97E-01
201501 ZBTB7C 0.000 15.492 0.000 6.966 1.169 2.28E-02 NA
57596 BEGAIN 0.000 10.651 1.985 13.933 1.168 2.46E-02 7.44E-01
481 ATP1B1 8.386 23.238 8.933 39.185 1.154 2.11E-02 6.96E-01
2043 EPHA4 10.482 29.048 25.806 95.787 1.154 1.95E-02 6.64E-01
8609 KLF7 2.096 23.238 8.933 23.511 1.151 2.66E-02 7.80E-01
9695 EDEM1 63.940 124.905 66.501 236.854 1.147 5.53E-03 3.26E-01
285958 SNHG15 0.000 11.619 0.000 9.579 1.146 2.56E-02 NA
80853 KDM7A 4.193 23.238 11.911 36.573 1.143 2.46E-02 7.44E-01
8531 YBX3 94.338 263.365 101.240 246.433 1.143 2.59E-03 2.12E-01
9394 HS6ST1 104.820 287.572 61.538 177.640 1.140 7.68E-03 4.15E-01
6376 CX3CL1 2.096 29.048 1.985 3.483 1.139 2.77E-02 7.93E-01
9261 MAPKAPK2 187.627 461.858 159.800 438.006 1.138 2.24E-03 1.92E-01
9781 RNF144A 14.675 68.746 62.531 175.899 1.134 1.92E-02 6.63E-01
114299 PALM2 2.096 16.460 3.970 14.803 1.128 3.13E-02 8.38E-01
196527 ANO6 76.518 267.238 129.031 261.236 1.122 4.10E-03 2.76E-01
57132 CHMP1B 169.808 457.016 74.441 242.079 1.121 1.21E-02 5.31E-01
55818 KDM3A 18.868 72.619 52.605 134.101 1.116 1.53E-02 5.89E-01
55197 RPRD1A 19.916 52.286 14.888 51.376 1.116 1.65E-02 6.10E-01
285203 EOGT 29.350 49.381 28.784 121.910 1.112 1.53E-02 5.89E-01
221830 TWISTNB 20.964 62.937 22.829 60.955 1.111 1.33E-02 5.49E-01
171546 SPTSSA 39.832 86.175 29.776 105.365 1.111 9.62E-03 4.76E-01
160760 PPTC7 28.301 122.000 41.687 70.534 1.109 1.26E-02 5.39E-01
6792 CDKL5 0.000 4.841 0.993 17.416 1.105 3.02E-02 NA
283209 PGM2L1 14.675 26.143 11.911 59.213 1.093 2.58E-02 7.65E-01
81792 ADAMTS12 8.386 38.730 13.896 31.348 1.092 2.63E-02 7.74E-01
10123 ARL4C 81.759 297.254 22.829 59.213 1.092 3.06E-02 8.28E-01
54910 SEMA4C 63.940 239.159 57.568 94.045 1.091 1.52E-02 5.89E-01
6659 SOX4 100.627 209.143 231.264 736.685 1.091 1.75E-02 6.28E-01
5.3 List of differentially expressed genes 135
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
51582 AZIN1 134.169 372.778 226.301 529.438 1.090 4.69E-03 2.97E-01
6482 ST3GAL1 88.049 254.651 72.456 161.096 1.088 8.38E-03 4.35E-01
3589 IL11 33.542 476.381 13.896 60.955 1.086 3.05E-02 8.28E-01
8808 IL1RL2 0.000 9.683 3.970 19.157 1.086 3.78E-02 9.18E-01
56925 LXN 12.578 23.238 26.799 107.978 1.085 3.03E-02 8.28E-01
84365 NIFK 2.096 28.079 10.918 21.770 1.085 3.65E-02 9.10E-01
157506 RDH10 109.013 289.508 295.780 829.860 1.080 1.71E-02 6.26E-01
10788 IQGAP2 0.000 6.778 1.985 16.545 1.079 3.66E-02 9.10E-01
50640 PNPLA8 16.771 62.937 24.814 52.247 1.076 1.88E-02 6.56E-01
83641 FAM107B 41.928 161.699 63.523 108.848 1.074 1.10E-02 5.09E-01
3566 IL4R 102.723 259.492 44.665 148.034 1.073 1.72E-02 6.27E-01
4007 PRICKLE3 5.241 24.206 4.963 15.674 1.071 3.93E-02 9.33E-01
57181 SLC39A10 2.096 19.365 13.896 40.927 1.068 3.93E-02 9.33E-01
1195 CLK1 20.964 67.778 40.695 97.528 1.067 1.65E-02 6.10E-01
2150 F2RL1 4.193 23.238 35.732 125.393 1.059 4.11E-02 9.45E-01
90627 STARD13 73.374 196.556 104.218 237.725 1.057 6.35E-03 3.62E-01
2013 EMP2 59.747 130.714 63.523 177.640 1.057 8.15E-03 4.27E-01
79656 BEND5 0.000 5.810 1.985 17.416 1.056 4.02E-02 9.38E-01
3755 KCNG1 14.675 39.698 21.836 62.697 1.051 2.47E-02 7.46E-01
57647 DHX37 30.398 69.714 18.858 64.438 1.048 2.10E-02 6.96E-01
93145 OLFM2 57.651 97.794 18.858 120.169 1.047 2.45E-02 7.44E-01
7074 TIAM1 4.193 16.460 13.896 47.022 1.044 4.25E-02 9.52E-01
121642 ALKBH2 4.193 11.619 1.985 18.287 1.043 4.66E-02 1.00E+00
5641 LGMN 30.398 61.968 25.806 86.208 1.040 1.84E-02 6.51E-01
6781 STC1 270.435 1834.843 38.709 792.416 1.039 4.08E-02 9.43E-01
10194 TSHZ1 24.109 89.079 28.784 51.376 1.039 2.27E-02 7.09E-01
116987 AGAP1 63.940 161.699 47.642 119.298 1.038 1.33E-02 5.49E-01
100133941 CD24 4.193 1.937 8.933 69.663 1.037 4.70E-02 1.00E+00
10435 CDC42EP2 12.578 51.318 1.985 11.320 1.036 4.78E-02 1.00E+00
6018 RLF 12.578 27.111 13.896 51.376 1.036 3.34E-02 8.66E-01
254170 FBXO33 15.723 48.413 12.903 33.961 1.036 3.16E-02 8.42E-01
30811 HUNK 6.289 37.762 9.925 16.545 1.035 4.34E-02 9.65E-01
55422 ZNF331 13.627 44.540 12.903 32.219 1.034 3.27E-02 8.58E-01
81831 NETO2 13.627 11.619 11.911 80.112 1.033 4.50E-02 9.85E-01
54793 KCTD9 36.687 129.746 54.590 96.657 1.030 1.50E-02 5.86E-01
56180 MOSPD1 2.096 8.714 3.970 20.899 1.028 4.95E-02 1.00E+00
23268 DNMBP 41.928 122.000 31.762 68.792 1.022 2.23E-02 7.06E-01
6461 SHB 44.024 101.667 33.747 94.045 1.022 1.65E-02 6.10E-01
3269 HRH1 14.675 34.857 3.970 27.865 1.021 4.56E-02 9.93E-01
153443 SRFBP1 2.096 13.556 0.000 7.837 1.020 4.77E-02 NA
3099 HK2 63.940 225.603 68.486 105.365 1.020 1.91E-02 6.63E-01
136 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
164284 APCDD1L 56.603 112.318 15.881 93.174 1.018 3.23E-02 8.50E-01
10116 FEM1B 51.362 108.445 58.560 160.225 1.017 1.27E-02 5.39E-01
83930 STARD3NL 20.964 64.873 31.762 69.663 1.015 2.23E-02 7.06E-01
5128 CDK17 16.771 65.841 30.769 57.472 1.014 2.63E-02 7.74E-01
10746 MAP3K2 26.205 59.064 24.814 72.275 1.013 2.25E-02 7.06E-01
665 BNIP3L 150.941 395.048 241.189 518.118 1.012 7.42E-03 4.04E-01
8819 SAP30 9.434 38.730 15.881 33.090 1.009 3.86E-02 9.28E-01
2034 EPAS1 420.327 1016.668 188.584 534.663 1.007 2.20E-02 7.06E-01
57597 BAHCC1 98.531 267.238 108.188 215.955 1.007 9.07E-03 4.57E-01
8497 PPFIA4 12.578 17.429 12.903 62.697 1.007 4.56E-02 9.93E-01
22856 CHSY1 66.036 188.810 85.359 167.191 1.006 1.06E-02 4.98E-01
23132 RAD54L2 16.771 67.778 30.769 54.860 1.005 2.81E-02 7.95E-01
79720 VPS37B 44.024 116.191 48.635 106.236 0.999 1.53E-02 5.89E-01
5045 FURIN 227.459 445.397 167.741 479.803 0.999 8.28E-03 4.32E-01
1130 LYST 17.819 60.032 31.762 66.180 0.998 2.71E-02 7.84E-01
4088 SMAD3 146.748 389.238 180.644 357.893 0.998 6.96E-03 3.86E-01
254531 LPCAT4 16.771 45.508 32.754 84.466 0.995 3.21E-02 8.46E-01
6566 SLC16A1 47.169 104.572 45.657 117.556 0.989 1.61E-02 6.04E-01
23235 SIK2 68.133 127.810 42.680 142.809 0.989 1.85E-02 6.51E-01
64764 CREB3L2 97.482 329.207 145.905 231.629 0.988 1.28E-02 5.39E-01
9587 MAD2L1BP 22.012 35.825 6.948 50.506 0.986 4.72E-02 1.00E+00
23387 SIK3 40.880 128.778 52.605 94.045 0.985 1.93E-02 6.63E-01
7422 VEGFA 759.944 1595.684 474.438 1289.635 0.985 1.16E-02 5.20E-01
59277 NTN4 29.350 76.492 26.799 62.697 0.980 2.69E-02 7.84E-01
55790 CSGALNACT1 100.627 260.461 61.538 136.713 0.977 2.54E-02 7.59E-01
50999 TMED5 42.976 122.000 43.672 85.337 0.977 2.18E-02 7.06E-01
4854 NOTCH3 128.928 297.254 132.009 293.455 0.977 8.79E-03 4.52E-01
23516 SLC39A14 263.098 602.254 105.210 325.674 0.975 2.97E-02 8.28E-01
6513 SLC2A1 214.881 674.874 335.482 552.079 0.971 9.98E-03 4.84E-01
25963 TMEM87A 60.795 129.746 55.583 141.067 0.968 1.60E-02 6.04E-01
6535 SLC6A8 30.398 42.603 24.814 101.011 0.964 3.77E-02 9.18E-01
3732 CD82 190.772 382.461 122.084 339.607 0.960 1.54E-02 5.89E-01
63893 UBE2O 45.073 111.349 44.665 98.399 0.960 2.10E-02 6.96E-01
29969 MDFIC 87.000 128.778 51.613 206.376 0.959 2.39E-02 7.39E-01
51186 TCEAL9 68.133 183.968 95.285 183.736 0.956 1.44E-02 5.73E-01
8728 ADAM19 26.205 91.984 81.389 172.416 0.951 3.80E-02 9.20E-01
23336 SYNM 24.109 45.508 38.709 113.202 0.950 4.00E-02 9.38E-01
121260 SLC15A4 36.687 90.048 33.747 74.017 0.931 3.12E-02 8.38E-01
286343 LURAP1L 35.639 68.746 19.851 67.051 0.928 4.15E-02 9.45E-01
5069 PAPPA 114.254 184.937 29.776 181.124 0.919 4.98E-02 1.00E+00
7461 CLIP2 123.687 326.302 205.458 383.146 0.917 1.62E-02 6.06E-01
5.3 List of differentially expressed genes 137
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
256987 SERINC5 52.410 87.143 85.359 238.596 0.914 4.03E-02 9.38E-01
90 ACVR1 81.759 160.730 79.404 192.444 0.912 1.92E-02 6.63E-01
131578 LRRC15 81.759 134.587 77.419 221.180 0.911 2.38E-02 7.38E-01
7040 TGFB1 343.809 583.858 264.018 735.815 0.910 1.46E-02 5.73E-01
490 ATP2B1 47.169 119.095 34.739 70.534 0.904 4.17E-02 9.45E-01
483 ATP1B3 83.856 172.349 76.426 175.899 0.902 2.11E-02 6.96E-01
9188 DDX21 71.277 186.873 108.188 196.798 0.896 2.22E-02 7.06E-01
7035 TFPI 83.856 144.270 41.687 141.938 0.887 4.01E-02 9.38E-01
3572 IL6ST 626.823 883.048 482.379 1507.332 0.881 2.19E-02 7.06E-01
5209 PFKFB3 163.519 416.350 140.942 236.854 0.878 3.12E-02 8.38E-01
10867 TSPAN9 148.844 290.476 156.823 346.573 0.875 1.79E-02 6.40E-01
10458 BAIAP2 37.735 68.746 30.769 84.466 0.871 4.48E-02 9.83E-01
23710 GABARAPL1 87.000 190.746 78.411 156.742 0.859 3.06E-02 8.28E-01
205 AK4 28.301 66.810 38.709 79.242 0.855 4.87E-02 1.00E+00
4814 NINJ1 199.158 453.143 150.867 281.264 0.851 3.54E-02 8.99E-01
84255 SLC37A3 32.494 77.460 41.687 81.854 0.850 4.65E-02 1.00E+00
60485 SAV1 54.506 121.032 51.613 101.882 0.841 4.18E-02 9.45E-01
23541 SEC14L2 38.783 75.524 36.724 86.208 0.839 4.90E-02 1.00E+00
57403 RAB22A 34.591 97.794 50.620 81.854 0.837 4.96E-02 1.00E+00
55603 FAM46A 75.470 148.143 103.225 219.438 0.833 3.54E-02 8.99E-01
9689 BZW1 105.868 156.857 101.240 275.169 0.831 3.79E-02 9.18E-01
5467 PPARD 66.036 141.365 73.449 143.680 0.830 3.58E-02 8.99E-01
9804 TOMM20 67.085 143.302 88.337 170.674 0.819 3.74E-02 9.18E-01
54477 PLEKHA5 66.036 113.286 71.464 168.933 0.816 4.29E-02 9.57E-01
23481 PES1 58.699 127.810 67.493 128.006 0.816 4.15E-02 9.45E-01
57003 CCDC47 103.772 213.984 103.225 200.281 0.816 3.34E-02 8.66E-01
5151 PDE8A 89.097 184.937 64.516 132.360 0.814 4.99E-02 1.00E+00
1910 EDNRB 158.278 234.318 194.540 472.837 0.789 4.84E-02 1.00E+00
3091 HIF1A 231.652 375.683 276.921 621.742 0.788 3.69E-02 9.15E-01
8662 EIF3B 163.519 252.715 126.054 313.483 0.772 4.48E-02 9.83E-01
1983 EIF5 132.073 289.508 153.845 256.011 0.771 4.04E-02 9.38E-01
7291 TWIST1 115.302 224.635 111.165 208.989 0.762 4.61E-02 9.98E-01
23243 ANKRD28 93.290 192.683 133.002 238.596 0.759 4.87E-02 1.00E+00
80381 CD276 171.905 344.699 205.458 351.798 0.736 4.43E-02 9.81E-01
960 CD44 933.945 1610.208 690.814 1421.995 0.733 4.85E-02 1.00E+00
966 CD59 538.774 1139.636 723.568 1141.601 0.723 4.11E-02 9.45E-01
7162 TPBG 263.098 177.191 325.556 147.163 -0.745 4.80E-02 1.00E+00
2316 FLNA 3367.861 2115.637 4362.253 2223.989 -0.750 3.05E-02 8.28E-01
56731 SLC2A4RG 160.374 89.079 166.748 87.949 -0.766 4.83E-02 1.00E+00
57674 RNF213 448.629 243.032 329.526 166.320 -0.793 4.61E-02 9.98E-01
4641 MYO1C 488.460 295.318 571.708 270.815 -0.801 2.64E-02 7.74E-01
138 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
283078 MKX 105.868 48.413 124.069 69.663 -0.811 4.71E-02 1.00E+00
440275 EIF2AK4 114.254 68.746 145.905 64.438 -0.813 4.24E-02 9.51E-01
8573 CASK 99.579 43.571 120.098 67.921 -0.818 4.81E-02 1.00E+00
254102 EHBP1L1 278.821 148.143 282.876 146.292 -0.821 2.73E-02 7.84E-01
80332 ADAM33 132.073 68.746 135.979 68.792 -0.821 3.85E-02 9.28E-01
6938 TCF12 166.664 91.984 212.405 103.624 -0.824 3.18E-02 8.45E-01
311 ANXA11 287.206 162.667 266.003 120.169 -0.832 3.19E-02 8.46E-01
5156 PDGFRA 896.210 523.826 1184.111 559.045 -0.839 1.75E-02 6.28E-01
8476 CDC42BPA 100.627 64.873 162.778 65.309 -0.839 3.98E-02 9.38E-01
142 PARP1 123.687 70.683 141.934 61.826 -0.845 3.48E-02 8.91E-01
122953 JDP2 113.205 51.318 99.255 53.989 -0.850 3.89E-02 9.29E-01
10512 SEMA3C 337.520 187.841 245.160 94.045 -0.851 4.47E-02 9.83E-01
59 ACTA2 67.085 28.079 63.523 33.961 -0.860 4.95E-02 1.00E+00
152503 SH3D19 118.446 61.968 111.165 49.635 -0.865 3.58E-02 8.99E-01
3667 IRS1 244.230 118.127 278.906 145.421 -0.869 2.01E-02 6.74E-01
5865 RAB3B 126.832 75.524 181.636 74.888 -0.872 2.80E-02 7.95E-01
1848 DUSP6 140.459 86.175 313.645 128.876 -0.874 3.71E-02 9.15E-01
57060 PCBP4 89.097 31.952 153.845 79.242 -0.887 4.64E-02 1.00E+00
71 ACTG1 2637.267 1092.191 4142.899 2222.247 -0.888 2.40E-02 7.39E-01
23264 ZC3H7B 515.714 301.127 638.209 271.685 -0.889 1.40E-02 5.68E-01
4141 MARS 174.001 88.111 133.002 56.601 -0.892 3.33E-02 8.66E-01
115207 KCTD12 535.629 275.953 968.728 450.197 -0.895 2.22E-02 7.06E-01
64780 MICAL1 81.759 27.111 113.151 61.826 -0.900 4.15E-02 9.45E-01
23211 ZC3H4 47.169 19.365 63.523 29.607 -0.903 4.69E-02 1.00E+00
81603 TRIM8 268.339 125.873 279.899 141.938 -0.910 1.44E-02 5.73E-01
3275 PRMT2 136.266 77.460 157.815 60.955 -0.911 2.32E-02 7.23E-01
51564 HDAC7 123.687 64.873 181.636 79.242 -0.912 2.24E-02 7.06E-01
678 ZFP36L2 138.362 77.460 146.897 55.730 -0.913 2.48E-02 7.46E-01
23576 DDAH1 53.458 34.857 173.696 60.955 -0.914 4.91E-02 1.00E+00
440 ASNS 33.542 17.429 50.620 17.416 -0.925 5.00E-02 1.00E+00
400550 FENDRR 143.603 84.238 226.301 87.949 -0.928 1.89E-02 6.58E-01
54862 CC2D1A 46.121 14.524 36.724 20.028 -0.930 4.97E-02 1.00E+00
1841 DTYMK 56.603 22.270 42.680 20.028 -0.931 4.43E-02 9.81E-01
221 ALDH3B1 95.386 30.984 35.732 22.640 -0.941 5.00E-02 1.00E+00
16 AARS 210.688 91.984 146.897 72.275 -0.942 2.17E-02 7.06E-01
51115 RMDN1 48.217 23.238 57.568 21.770 -0.942 3.77E-02 9.18E-01
22836 RHOBTB3 84.904 51.318 116.128 37.444 -0.947 2.57E-02 7.65E-01
26227 PHGDH 107.964 40.667 89.329 47.893 -0.957 2.25E-02 7.06E-01
4171 MCM2 85.952 28.079 35.732 20.899 -0.960 4.46E-02 9.83E-01
1969 EPHA2 89.097 55.191 149.875 49.635 -0.962 2.15E-02 7.03E-01
64859 NABP1 50.314 36.794 151.860 45.281 -0.963 3.57E-02 8.99E-01
5.3 List of differentially expressed genes 139
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
7402 UTRN 76.518 40.667 259.055 94.045 -0.964 3.91E-02 9.32E-01
81563 C1ORF21 37.735 26.143 86.352 23.511 -0.966 3.75E-02 9.18E-01
9400 RECQL5 29.350 13.556 36.724 11.320 -0.973 4.61E-02 9.98E-01
8989 TRPA1 101.675 43.571 61.538 23.511 -0.974 3.40E-02 8.77E-01
152573 SHISA3 0.000 0.000 20.844 0.000 -0.976 4.77E-02 NA
5921 RASA1 105.868 56.159 213.398 81.854 -0.978 2.02E-02 6.75E-01
653513 LOC653513 14.675 0.000 2.978 0.000 -0.979 4.94E-02 NA
359845 RFLNB 95.386 54.222 288.832 102.753 -0.979 3.03E-02 8.28E-01
388650 FAM69A 57.651 30.016 67.493 21.770 -0.980 2.89E-02 8.12E-01
114876 OSBPL1A 45.073 21.302 66.501 24.382 -0.981 3.06E-02 8.28E-01
11234 HPS5 37.735 15.492 28.784 8.708 -0.984 4.68E-02 1.00E+00
54858 PGPEP1 27.253 11.619 41.687 13.933 -0.991 4.18E-02 9.45E-01
29028 ATAD2 36.687 23.238 68.486 17.416 -0.995 3.38E-02 8.72E-01
23179 RGL1 79.663 18.397 54.590 36.573 -0.995 3.05E-02 8.28E-01
55619 DOCK10 58.699 24.206 95.285 39.185 -0.999 2.43E-02 7.44E-01
93627 TBCK 16.771 5.810 41.687 13.062 -1.003 4.92E-02 1.00E+00
523 ATP6V1A 71.277 43.571 136.972 42.669 -1.004 1.97E-02 6.64E-01
79924 ADM2 13.627 0.000 3.970 0.000 -1.005 4.49E-02 NA
134957 STXBP5 40.880 22.270 47.642 10.449 -1.005 3.70E-02 9.15E-01
55901 THSD1 4.193 0.000 13.896 0.000 -1.008 4.46E-02 NA
84162 KIAA1109 89.097 42.603 127.046 49.635 -1.008 1.50E-02 5.86E-01
727800 RNF208 14.675 0.968 4.963 1.742 -1.009 4.79E-02 NA
64770 CCDC14 42.976 15.492 58.560 24.382 -1.009 2.92E-02 8.21E-01
6563 SLC14A1 340.664 131.683 117.121 34.831 -1.011 3.88E-02 9.29E-01
10827 FAM114A2 20.964 7.746 24.814 6.966 -1.013 4.69E-02 1.00E+00
4013 VWA5A 75.470 28.079 248.137 92.303 -1.014 3.42E-02 8.79E-01
81493 SYNC 72.326 21.302 25.806 13.933 -1.015 4.02E-02 9.38E-01
79885 HDAC11 28.301 1.937 27.791 14.803 -1.020 4.89E-02 1.00E+00
3516 RBPJ 171.905 63.905 260.048 121.039 -1.020 1.27E-02 5.39E-01
10602 CDC42EP3 59.747 42.603 126.054 30.478 -1.021 2.25E-02 7.06E-01
54532 USP53 27.253 11.619 63.523 20.899 -1.022 3.56E-02 8.99E-01
79690 GAL3ST4 25.157 7.746 28.784 10.449 -1.025 4.07E-02 9.43E-01
79890 RIN3 32.494 7.746 21.836 10.449 -1.027 4.15E-02 9.45E-01
388963 C2ORF81 12.578 1.937 12.903 3.483 -1.028 4.94E-02 1.00E+00
374393 FAM111B 25.157 2.905 11.911 6.966 -1.030 4.88E-02 1.00E+00
2139 EYA2 19.916 5.810 38.709 13.062 -1.032 4.16E-02 9.45E-01
5860 QDPR 42.976 11.619 31.762 15.674 -1.032 3.29E-02 8.60E-01
421 ARVCF 4.193 3.873 32.754 3.483 -1.033 4.83E-02 1.00E+00
64388 GREM2 319.700 150.079 186.599 26.124 -1.033 3.53E-02 8.99E-01
55103 RALGPS2 30.398 10.651 88.337 30.478 -1.033 3.59E-02 9.01E-01
6574 SLC20A1 278.821 114.254 162.778 67.051 -1.037 1.61E-02 6.04E-01
140 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
25802 LMOD1 12.578 3.873 44.665 12.191 -1.038 4.53E-02 9.91E-01
154091 SLC2A12 10.482 0.968 6.948 0.000 -1.042 3.93E-02 NA
5257 PHKB 40.880 11.619 51.613 22.640 -1.045 2.73E-02 7.85E-01
221895 JAZF1 11.530 7.746 54.590 11.320 -1.046 4.23E-02 9.51E-01
6773 STAT2 278.821 99.730 154.838 80.983 -1.046 1.29E-02 5.39E-01
83660 TLN2 58.699 20.333 158.808 59.213 -1.048 2.72E-02 7.84E-01
6935 ZEB1 124.736 57.127 248.137 94.916 -1.048 1.29E-02 5.39E-01
3985 LIMK2 38.783 11.619 32.754 13.933 -1.048 3.04E-02 8.28E-01
57484 RNF150 46.121 20.333 59.553 20.028 -1.050 2.16E-02 7.03E-01
23536 ADAT1 24.109 5.810 20.844 7.837 -1.054 3.95E-02 9.34E-01
6742 SSBP1 18.868 3.873 15.881 5.225 -1.059 4.21E-02 9.49E-01
5905 RANGAP1 286.158 126.841 326.549 137.584 -1.061 4.13E-03 2.76E-01
5063 PAK3 22.012 1.937 11.911 6.096 -1.062 4.26E-02 9.52E-01
79960 JADE1 47.169 12.587 86.352 35.702 -1.064 2.51E-02 7.52E-01
330 BIRC3 94.338 32.921 34.739 5.225 -1.066 3.76E-02 9.18E-01
85379 KIAA1671 25.157 6.778 11.911 1.742 -1.067 4.17E-02 9.45E-01
80169 CTC1 6.289 0.968 20.844 3.483 -1.067 4.01E-02 9.38E-01
390 RND3 469.593 237.222 428.781 129.747 -1.071 8.12E-03 4.27E-01
144363 ETFRF1 4.193 1.937 21.836 1.742 -1.073 3.78E-02 9.18E-01
114796 PSMG3-AS1 20.964 6.778 17.866 3.483 -1.074 3.88E-02 9.29E-01
9569 GTF2IRD1 52.410 12.587 28.784 15.674 -1.075 2.72E-02 7.84E-01
11046 SLC35D2 13.627 7.746 49.627 10.449 -1.077 3.53E-02 8.99E-01
2825 GPR1 23.060 7.746 27.791 7.837 -1.083 3.21E-02 8.46E-01
9577 BRE 16.771 5.810 23.821 5.225 -1.084 3.61E-02 9.01E-01
115557 ARHGEF25 89.097 30.016 75.434 34.831 -1.084 1.19E-02 5.27E-01
862 RUNX1T1 8.386 0.000 9.925 0.000 -1.085 3.28E-02 NA
5783 PTPN13 159.326 81.333 262.033 90.562 -1.085 6.04E-03 3.46E-01
91624 NEXN 23.060 11.619 88.337 23.511 -1.091 2.83E-02 7.98E-01
219285 SAMD9L 78.615 31.952 54.590 15.674 -1.094 1.95E-02 6.64E-01
10669 CGREF1 14.675 1.937 6.948 0.000 -1.094 3.36E-02 NA
84939 MUM1 81.759 22.270 88.337 42.669 -1.105 1.19E-02 5.27E-01
60686 C14ORF93 10.482 2.905 15.881 1.742 -1.112 3.35E-02 8.68E-01
388115 C15ORF52 56.603 22.270 35.732 4.354 -1.117 2.80E-02 7.95E-01
57523 NYNRIN 37.735 15.492 52.605 15.674 -1.117 1.75E-02 6.28E-01
6339 SCNN1D 15.723 0.000 15.881 5.225 -1.124 3.20E-02 8.46E-01
55793 FAM63A 20.964 5.810 18.858 4.354 -1.125 3.00E-02 8.28E-01
84795 PYROXD2 29.350 3.873 34.739 14.803 -1.128 2.61E-02 7.72E-01
55614 KIF16B 16.771 6.778 21.836 1.742 -1.129 3.08E-02 8.30E-01
29128 UHRF1 59.747 12.587 48.635 25.253 -1.131 1.58E-02 6.00E-01
2027 ENO3 12.578 0.000 6.948 0.000 -1.132 2.63E-02 NA
23051 ZHX3 98.531 44.540 114.143 36.573 -1.133 6.68E-03 3.73E-01
5.3 List of differentially expressed genes 141
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
55388 MCM10 19.916 4.841 11.911 0.000 -1.134 3.03E-02 8.28E-01
4061 LY6E 32.494 11.619 23.821 0.871 -1.135 2.99E-02 8.28E-01
7791 ZYX 236.893 98.762 353.347 135.843 -1.139 3.10E-03 2.38E-01
596 BCL2 14.675 1.937 13.896 3.483 -1.139 2.97E-02 8.28E-01
5021 OXTR 19.916 1.937 5.955 1.742 -1.140 2.81E-02 7.95E-01
387758 FIBIN 33.542 9.683 39.702 13.933 -1.140 1.87E-02 6.56E-01
79443 FYCO1 149.892 54.222 195.532 80.983 -1.141 4.28E-03 2.82E-01
10245 TIMM17B 20.964 6.778 30.769 7.837 -1.141 2.44E-02 7.44E-01
291 SLC25A4 27.253 4.841 21.836 8.708 -1.143 2.53E-02 7.57E-01
84171 LOXL4 63.940 24.206 43.672 10.449 -1.150 1.71E-02 6.26E-01
9223 MAGI1 47.169 17.429 68.486 22.640 -1.156 1.13E-02 5.17E-01
5137 PDE1C 32.494 5.810 7.940 1.742 -1.157 2.73E-02 7.84E-01
5828 PEX2 42.976 15.492 66.501 21.770 -1.157 1.23E-02 5.34E-01
54899 PXK 41.928 14.524 65.508 21.770 -1.158 1.27E-02 5.39E-01
23022 PALLD 176.097 95.857 328.534 94.916 -1.162 3.71E-03 2.64E-01
3306 HSPA2 48.217 23.238 78.411 20.028 -1.167 1.02E-02 4.90E-01
23371 TNS2 107.964 28.079 70.471 36.573 -1.168 7.94E-03 4.23E-01
5939 RBMS2 77.567 24.206 74.441 30.478 -1.174 7.04E-03 3.89E-01
2115 ETV1 37.735 13.556 128.039 36.573 -1.179 1.57E-02 6.00E-01
81848 SPRY4 47.169 23.238 70.471 15.674 -1.180 1.06E-02 4.99E-01
9260 PDLIM7 227.459 98.762 423.818 140.197 -1.207 2.48E-03 2.07E-01
2992 GYG1 38.783 7.746 29.776 12.191 -1.208 1.45E-02 5.73E-01
24139 EML2 48.217 10.651 43.672 18.287 -1.210 1.08E-02 5.04E-01
10507 SEMA4D 17.819 1.937 8.933 1.742 -1.217 1.97E-02 6.64E-01
55504 TNFRSF19 202.302 120.064 303.720 49.635 -1.218 5.30E-03 3.19E-01
27147 DENND2A 71.277 15.492 34.739 17.416 -1.218 1.11E-02 5.09E-01
10611 PDLIM5 102.723 56.159 253.100 65.309 -1.219 4.68E-03 2.97E-01
84189 SLITRK6 19.916 2.905 9.925 1.742 -1.221 1.97E-02 6.64E-01
57724 EPG5 73.374 24.206 64.516 21.770 -1.233 5.62E-03 3.28E-01
60 ACTB 2728.460 1067.985 3786.574 1397.612 -1.241 4.68E-04 6.40E-02
171024 SYNPO2 29.350 24.206 111.165 13.933 -1.241 1.16E-02 5.20E-01
1958 EGR1 127.880 45.508 474.438 135.843 -1.241 8.88E-03 4.53E-01
1902 LPAR1 546.111 200.429 530.021 205.506 -1.241 6.04E-04 7.65E-02
64750 SMURF2 165.615 36.794 138.957 70.534 -1.242 3.18E-03 2.42E-01
7846 TUBA1A 312.363 101.667 856.570 276.039 -1.247 5.66E-03 3.28E-01
64221 ROBO3 28.301 9.683 52.605 13.062 -1.252 1.02E-02 4.90E-01
154796 AMOT 32.494 20.333 155.830 31.348 -1.254 1.04E-02 4.93E-01
112574 SNX18 150.941 45.508 199.502 77.500 -1.258 1.98E-03 1.75E-01
147495 APCDD1 78.615 38.730 198.510 49.635 -1.271 3.99E-03 2.75E-01
8507 ENC1 62.892 25.175 75.434 18.287 -1.279 4.36E-03 2.84E-01
23677 SH3BP4 137.314 53.254 157.815 48.764 -1.279 1.42E-03 1.44E-01
142 Chapter 5 Appendix A
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
4329 ALDH6A1 13.627 3.873 25.806 3.483 -1.280 1.41E-02 5.70E-01
54328 GPR173 12.578 3.873 24.814 2.612 -1.282 1.42E-02 5.72E-01
6869 TACR1 30.398 5.810 12.903 1.742 -1.289 1.38E-02 5.64E-01
57761 TRIB3 56.603 9.683 25.806 12.191 -1.293 9.08E-03 4.57E-01
92154 MTSS1L 235.845 93.921 247.145 73.146 -1.298 8.49E-04 9.74E-02
5393 EXOSC9 15.723 2.905 26.799 5.225 -1.305 1.21E-02 5.31E-01
9742 IFT140 63.940 16.460 61.538 21.770 -1.315 3.61E-03 2.62E-01
2491 CENPI 8.386 1.937 19.851 0.000 -1.316 1.15E-02 5.20E-01
4857 NOVA1 20.964 1.937 18.858 5.225 -1.318 1.16E-02 5.20E-01
116832 RPL39L 19.916 3.873 27.791 6.096 -1.322 1.02E-02 4.90E-01
9241 NOG 12.578 0.000 13.896 1.742 -1.328 1.08E-02 5.04E-01
2250 FGF5 32.494 14.524 134.987 27.865 -1.332 6.57E-03 3.69E-01
323 APBB2 31.446 11.619 39.702 5.225 -1.343 7.17E-03 3.92E-01
26230 TIAM2 25.157 3.873 20.844 5.225 -1.353 8.72E-03 4.50E-01
4008 LMO7 109.013 82.302 459.550 74.017 -1.354 3.35E-03 2.51E-01
10641 NPRL2 20.964 0.000 12.903 3.483 -1.359 9.49E-03 4.73E-01
6218 RPS17 32.494 3.873 77.419 20.899 -1.360 7.09E-03 3.90E-01
64900 LPIN3 27.253 9.683 43.672 6.966 -1.369 5.81E-03 3.36E-01
158056 MAMDC4 33.542 7.746 36.724 8.708 -1.405 4.35E-03 2.84E-01
9020 MAP3K14 22.012 5.810 25.806 2.612 -1.411 6.41E-03 3.62E-01
1265 CNN2 272.532 103.603 671.956 174.157 -1.414 8.14E-04 9.61E-02
9659 PDE4DIP 101.675 27.111 78.411 24.382 -1.427 1.10E-03 1.19E-01
5924 RASGRF2 56.603 19.365 76.426 16.545 -1.429 1.58E-03 1.52E-01
9760 TOX 6.289 0.000 23.821 0.000 -1.435 5.66E-03 3.28E-01
22874 PLEKHA6 42.976 10.651 30.769 5.225 -1.435 4.11E-03 2.76E-01
25959 KANK2 361.628 121.032 517.118 155.000 -1.442 1.14E-04 2.10E-02
9079 LDB2 33.542 12.587 59.553 8.708 -1.458 2.63E-03 2.13E-01
347902 AMIGO2 11.530 3.873 72.456 8.708 -1.493 4.05E-03 2.75E-01
84962 AJUBA 42.976 4.841 63.523 17.416 -1.531 1.80E-03 1.66E-01
57608 KIAA1462 281.965 106.508 320.593 54.860 -1.540 1.98E-04 3.33E-02
79633 FAT4 129.977 25.175 93.300 33.090 -1.553 3.15E-04 4.76E-02
3790 KCNS3 12.578 1.937 31.762 2.612 -1.603 2.18E-03 1.88E-01
23566 LPAR3 16.771 0.000 17.866 1.742 -1.609 2.13E-03 1.86E-01
2353 FOS 18.868 13.556 204.465 19.157 -1.621 1.56E-03 1.52E-01
54510 PCDH18 805.016 140.397 871.458 304.775 -1.623 4.21E-05 9.61E-03
22861 NLRP1 27.253 4.841 25.806 1.742 -1.663 1.31E-03 1.35E-01
27123 DKK2 27.253 3.873 39.702 6.096 -1.700 7.60E-04 9.29E-02
10129 FRY 6.289 0.000 38.709 0.000 -1.718 1.00E-03 1.13E-01
79094 CHAC1 27.253 0.968 13.896 1.742 -1.754 8.14E-04 9.61E-02
113146 AHNAK2 32.494 1.937 49.627 8.708 -1.808 3.56E-04 5.18E-02
10290 SPEG 116.350 16.460 150.867 37.444 -1.814 3.30E-05 8.08E-03
5.3 List of differentially expressed genes 143
Table 5.1 continued from previous page
Reads per Million (RPM)
GeneID Gene Symbol ZZ29 ML ZZ29 6h ZZ31 ML ZZ31 6h log2FC P value adj.P value
55784 MCTP2 73.374 20.333 146.897 21.770 -1.825 3.92E-05 9.15E-03
79640 C22ORF46 35.639 5.810 36.724 2.612 -1.888 1.90E-04 3.30E-02
4739 NEDD9 10.482 0.000 46.650 0.871 -2.032 1.06E-04 1.97E-02
89795 NAV3 105.868 8.714 42.680 8.708 -2.110 1.45E-05 4.03E-03
64283 ARHGEF28 34.591 5.810 64.516 1.742 -2.238 8.84E-06 2.68E-03
652 BMP4 63.940 6.778 69.478 5.225 -2.421 3.30E-07 1.54E-04
1490 CTGF 170.856 52.286 1006.445 16.545 -2.599 1.55E-07 8.39E-05
3491 CYR61 684.473 16.460 1222.820 8.708 -4.862 3.86E-31 3.97E-27

Publications
Patent
Technische Universität Berlin, 2018.
Method of preparing an artificial tooth primordium in vitro and artificial tooth primordium
derived therefrom
Inventors: Roland Lauster, Jennifer Binder, Mark Rosowski, Uwe Marx.
Granted 01.08.2018. EP2633870B1
Scientific publications
Rosowski, J., Bräunig, J., Strietzel, F., Lauster, R. & Rosowski, M. (in preparation)
Emulating the early phases of human tooth development in vitro
Drzymala, S. S. *, Binder, J.* , Brodehl, A., Penkert, M., Rosowski, M., Garbe, L. A., &
Koch, M. (2015). (* Co-first authors)
Estrogenicity of novel phase I and phase II metabolites of zearalenone and cis-zearalenone.
Toxicon, 105, 10-12.
Presentations
Dechema Conference on 3D Cell Culture 2014 in Freiburg, Germany
"Tooth formation in vivo and in vitro"
Oral presentation
Gordon Research Seminar on Craniofacial Morphogenesis and Tissue Regeneration 2014
in Lucca, Italy
"Tooth formation in vivo and in vitro"
Oral presentation
Gordon Research Conference on Craniofacial Morphogenesis and Tissue Regeneration 2014
in Lucca, Italy
"Tooth formation in vivo and in vitro"
Poster presentation
4th BSRT Symposium on Regenerative Medicine 2013, Berlin, Germany
"Mesenchymal Condensation of Dental Pulp Stem Cells"
Speed talk & Poster presentation, 2nd poster prize
Conference on Stem Cells in Development ans Disease 2013, MDC Berlin, Germany
"Formation of mesenchymal dental tissue from human adult DPCs in vitro"
Poster presentation
145

Danksagung
An dieser Stelle möchte ich mich bei den vielen Menschen bedanken, die mich bei der
Entstehung dieser Arbeit unterstützt haben.
Zunächst gilt mein Dank Roland Lauster, der seine Visionen und seine Begeister-
ung für das Thema mit mir geteilt hat, für umfangreiche wissenschaftliche Diskussionen,
für die Freiheit der Selbstverwirklichung und für sein großes Herz für Gemeinschaft und
Gemeinsamkeit.
Für die Begutachtung meiner Arbeit möchte ich Prof. Dr.-Ing. Claudia Fleck
und Prof. Dr. Juri Rappsilber danken, sowie Prof. Dr.-Ing. Vera Meyer für den Vorsitz
meines Promotionsverfahrens.
Herrn PD Dr. Frank Strietzel möchte ich ebenfalls für die Begutachtung meiner
Arbeit danken, sowie für die hervorragende Zusammenarbeit und die Versorgung mit
Zahnkeimen. Ebenfalls möchte ich in diesem Sinne Dr. Gülseren Köksal und dem
Praxisteam von Herrn Hans-Uwe Amler danken. Vielen Dank an Anna-Klara Amler für
die Vermittlung und viele spontane Einsätze.
Uwe Marx und Silke Hoffmann möchte ich ganz besonders für die große Unter-
stützung bei Erstellung und Betreuung des Patents und des Antrags danken. Von euch
habe ich sehr viel gelernt!
Meiner Arbeitsgruppe aus der GMA möchte ich ganz herzlich für die außerordent-
lich schöne Atmosphäre danken. Jedes Mitglied mit seinen Eigenheiten und besonderen
Fähigkeiten ist dort Teil eines lebendigen, funktionierenden und glücklichen Organismus.
Erfolge werden ehrlich miteinander geteilt und gefeiert, große oder kleine Niederlagen
aufgefangen. Agnes, die alles zusammenhält und mit ihrer hilfsbereiten Art einfach eine
wunderbar liebenswerte Person ist. Hinter jeder erfolgreichen Doktorarbeit steht eine starke
Luzie. Wenn alles schiefging, musste Luzie ran und hat es gerichtet mit ihrer langjährigen
Erfahrung und ihren Zaubertricks. Während meiner Vorlesung in der Elternzeit hat sie sich
liebevoll um meine Kinder gekümmert und sie hat als Labormama stets mit einem offenen
Ohr für den Seelenfrieden gesorgt. Meinen ehemaligen Studenten Julia und Sandro möchte
ich ganz besonders dafür danken, dass sie mich experimentell, produktiv und konstruktiv bei
vielen Teilen dieser Arbeit unterstützt haben und dass sie so wunderbar liebe Menschen sind.
147
148
Ein großer Dank geht auch an unser Büro. An das osmanische Trio, Özlem, Kübrah und
Zehra für die wunderbare Endphase meiner Promotion inklusive der Vorbereitung des
grandiosen LabQuiz. Ich danke natürlich auch den "alten Hasen", Marielle und Stefan
sowie Karolina. Auch von euch konnte ich viel lernen. Meiner liebsten Kollegin Shirin
möchte ich für so vieles danken. Sie hat mich stets in allem unterstützt, hat mir Mut
gemacht, hat mir meine Stärken aufgezeigt und die richtigen Fragen gestellt. Ohne dich
und deinen Einsatz beim Akkord-Korrekturlesen wäre diese Arbeit nicht möglich gewesen
und ich danke dir von ganzem Herzen dafür!
Meinen Freunden möchte ich dafür danken, dass sie mich stets emotional unter-
stützt haben, für ihr Verständnis und für die Verlässlichkeit, obwohl uns oft viele Kilometer
trennen. Sarah, Steffi und Chrissi möchte ich für die schönste Zeit im (Studenten)-Leben
danken. Franzi, Gwenni und Flo möchte ich danken, dass sie mich täglich motiviert und
mir Energie gegeben haben!
Zu guter Letzt danke ich meiner Familie. Ich kann nicht ausdrücken, wie dank-
bar ich meinen Eltern für ihr Vertrauen in mich, ihr Verständnis für meine Arbeit und
ihre bedingungslose Wärme und Liebe bin. Sie haben mich in jeder Phase mit vollem
Einsatz unterstützt und ich weiß das sehr zu schätzen. Ihr meine großen Vorbilder! Meinen
Kindern danke ich dafür, dass sie mir jahrelang den Schlaf abtrainiert haben, was vor
allem beim Schreiben außerordentlich hilfreich war. Mit euch zusammen kann ich Kraft
tanken und den Moment leben und dafür liebe ich euch! Meinem Mann gilt der größte
Dank. Als hervorragender Wissenschaftler hat er mich mit konstruktiven Gesprächen
motiviert und vorangebracht. Er hat sich mit mir durch die Datensätze gequält und mit
mir alle Interpretationen diskutiert. In der stressigsten Zeit hat er unsere beiden Rebellen
in Schach gehalten. Er hat einen maßgeblichen Anteil an dieser Arbeit, indem er mich
jeden Tag unterstützt und mental getragen hat.
DANKE!
Erklärung der Selbstständigkeit
Hiermit erkläre ich, dass mir die geltende Promotionsordung der TU Berlin vom 23.
Oktober 2006, zuletzt geändert mit der Änderungssatzung vom 15. Januar 2014, bekannt
ist. Hiermit erkläre ich an Eides statt, dass ich die Dissertation selbstständig verfasst habe.
Alle benutzen Quellen und Hilfsmittel sind aufgeführt.
Hiermit erkläre ich, dass diese Dissertation noch keiner anderen Fakultät oder Universität
zur Prüfung vorgelegt wurde. Veröffentlichungen von irgendwelchen Teilen der vorliegenden
Dissertation sind von mir wie folgt vorgenommen worden:
Patent:
Technische Universität Berlin, 2018
„Method of preparing an artificial tooth primordium in vitro and artificial tooth primordium
derived therefrom“ EP2633870B1
Erfinder: Roland Lauster, Jennifer Binder, Mark Rosowski, Uwe Marx
Erteilt am 1.08.2018
Ort, Datum