Design of a novel serum-free monolayer differentiation ... · embryonic stem cells (ESCs) can differentiate into articular cartilage by forming aggregates (embryoid body (EB), pellet,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Design of a novel serum-free monolayer differentiation system for murine embryonic stem cell-
derived chondrocytes for potential high-content imaging applications
by
Yan Ling Elaine Waese
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Department of Chemical Engineering and Applied Chemistry
Institute of Biomedical and Biomaterials Engineering
(PDGFRα) (Nishikawa, Nishikawa et al. 1998; Sakurai, Era et al. 2006). Regardless of whether
differentiation was initiated via the EB or monolayer culture method, the up-regulation of early
primitive streak/mesoderm marker genes such as Brachyury or Flk1 was observed between days
3 to 5 of differentiation, with peak expression levels detected on day 4. Using this knowledge, I
decided to design a one-step differentiation system that can generate mesodermal derivatives.
41
Specifically, my overall goal is to devise a system that can potentially be used in the discovery of
inducers that can enhance the generation of mesoderm-derived chondrocytes via high-throughput
screening (HTS) and high-content imaging (HCI) technologies. As such, I have devised the
following design criteria for the differentiation system:
1. Develop a system with minimal manual manipulation (i.e., without the need for cell
dissociation, cell sorting, subculturing, etc.) required to establish differentiation cultures
using known mesoderm and chondrogenic inducers.
2. Establish a versatile system that can be adapted to different ESC-derived cell types of
interest.
3. Develop a culture system that will be used as a baseline tool to study the effects of
modulators.
Since the differentiation system is intended to be used as a screening tool, it will be
beneficial to generate a mixed population of cells instead of using pure cell populations (e.g.
cells isolated from EBs using fluorescence activated cell sorting (FACS)) such that one can
detect the effects of modulators on the propagation of cell populations of interest. In addition,
one should be able to establish these test cultures quickly and easily in a cost-effective manner to
facilitate the screening of multiple test molecules in a high-throughpout manner. Furthermore, it
is essential to equip such a screening platform with a direct readout that allows for the easy
identification of candidate molecules. As such, I have identified three major project objectives:
Objective 1: Develop a 2D ESC differentiation system in defined conditions using exogenous
growth factors such as BMP4, Activin A or Wnt3a.
42
Objective 2: By comparing the mRNA and protein expression patterns of my cultures with
published data, I can validate whether my culture system supports the differentiation of ESCs
into A) mesoderm followed by B) chondrocytes as well as whether the known signaling
pathways that occur in vivo and in 3D cultures play similar roles in my 2D cultures.
Objective 3: Develop a transgenic reporter system that can be used in lieu of wild-type ESCs in
my differentiation system to facilitate the easy identification of ESC-derived chondrocytes in
live-cell imaging applications.
In terms of Objectives 1 and 2A, based on published results on the derivation ESC-
derived mesodermal cells, I hypothesize that I can recapitulate conventional EB culture
technique to generate ESC-derived mesoderm cells by establishing monolayer
differentiation cultures in a defined condition using exogenous growth factors. I have
identified the following project aims to test my hypothesis:
Aim 1: Identify the suitable basal conditions necessary to establish serum-free (SF) monolayer
ESC cultures.
Aim 2: Examine the feasibility of inducing mesoderm differentiation in SF monolayer ESC
cultures by adding BMP4, Activin A or Wnt3a at the onset of differentiation.
Aim 3: Determine whether the addition of BMP4, Activin A or Wnt3a at the onset of
differentiation can generate different mesoderm subsets.
For the establishment of the basal SF culture conditions (Aim 1), I am focusing on
examining the effects of SF medium formulation, ECM and seeding density on my cultures.
Specifically, I am looking for conditions that will not lead to dramatic spontaneous
43
differentiation without the addition of inductive factors because that may skew the results of my
assessment. To do so, I will culture the cells for two days in various combinations of SF
medium, ECM and seeding density in the presence of LIF, and I will choose the basal conditions
based on the extent of the maintenance of OCT4 expression in my cultures, as quantified by the
Cellomics ArrayScan® high-content screening system from Thermo Scientific (Fig. 1.6).
Fig. 1.6 – Schematic of the screening assay to be conducted to assess the basal conditions to be used in my monolayer differentiation culture system. ESC cultures established in every three consecutive wells
(representing three technical replicates) of the 96-well plate will be subjected to a specific combination of test
conditions, namely, SF medium, ECM and seeding density. After two days of culture, the expression of OCT4
protein in each well will be assessed via immunostaining and the fluorescence intensity will be quantified using the Cellomics ArrayScan® high-content screening system. Histograms representing the distribution of OCT4
expression for each combination of test condition can then be generated.
To induce mesoderm differentiation (Aim 2), I will add BMP4, Activin A or Wnt3a to
my SF monolayer cultures at the time of plating in the absence of LIF and measure the
expression of the primitive streak marker BRACHYURY on days 2, 4 and 6 of differentiation
using the Cellomics ArrayScan® high-content screening system. Additional analyses will be
performed on four-day differentiation cultures, concurrent with the time of peak expression of
primitive streak/early mesoderm markers, to determine the marker gene expression levels for
posterior and anterior mesoderm as well as mesendoderm (Aim 3) (Fig. 1.7). I want to determine
if the addition BMP4, Activin A or Wnt3a at the onset of differentiation to my SF culture system
44
can recapitulate the results of other in vitro ESC mesoderm differentiation systems as well as in
vivo studies of early embryonic development.
Fig. 1.7 – Schematic of the analyses to be conducted to verify mesoderm induction in my ESC SF monolayer differentiation system. Mesoderm inducers will be added to cultures at the onset of differentiation (Day 0) and
cells will be differentiated for 6 days during which BRACHYURY protein expression will be assessed via
Cellomics ArrayScan® high-content screening system (represented by image of the instrument) every two days.
Transcript level analyses of early mesoderm markers as wells mesoderm subset-specific marker genes (represented
by bar graph) will be conducted on day 4 of differentiation to confirm the recapitulation of published results regarding effects of growth factors on the induction of specific mesoderm populations.
As will be discussed in detail in Chapter 2, my culture system behaved similarly to other
published ESC mesoderm differentiation systems in that BMP4 could induce the up-regulation of
posterior mesoderm marker genes while Activin A addition enhanced the generation of anterior
mesoderm populations. The ability to control the type of mesoderm intermediates being formed
in monolayer differentiation cultures based on the addition of specific mesoderm inducers
enhanced the versatility of the system. To address Objective 2B, I suspect that I can use my
differentiation system to generate various mesoderm-derived cell types with prolonged culture
knowing that posterior mesoderm intermediates can give rise to hematopoietic or endothelial-
type cells, while more anterior populations can further differentiate into cell types such as those
45
of the skeletal and cardiac lineages. I am particularly interested in the generation of
chondrocytes from ESCs because of the challenges associated with cartilage repair (outlined in
Section 1.3.2), and as mentioned before, I want to generate an ESC-derived source of
chondrocytes that can potentially be used in the screening for novel therapeutics for cartilage
repair. I hypothesize that I can also bypass the conventional 3D EB, pellet or micromass
culture techniques to generate ESC-derived chondrocytes from ESCs by establishing
monolayer differentiation cultures in a defined condition using exogenous Activin A,
BMP4, Wnt3a and/or known chondrogenic inducers such as TGFββββ3 and FGF8. The
following aims are identified in order to test this particular hypothesis:
Aim 4: Determine the effect of prolonged exposure to Activin A, BMP4 or Wnt3a on
chondrogenic induction in my differentiation system.
Aim 5: Upon verifying that mesoderm inducers can also induce chondrocyte formation in my
differentiation culture system, examine whether known chondrogenic inducers such as TGFβ3
and FGF8 can further enhance chondrogenic induction from ESCs.
Aim 6: Determine the duration of growth factor supplementation necessary to induce
chondrocyte formation in my ESC differentiation cultures.
I plan to assess the efficacy of my culture system based on the expression of major
chondrogenic marker genes such as Col2a1, Sox9, Aggrecan on days 7 and/or 15 of
differentiation, and I will also examine the transcript levels of hypertrophic chondrocyte markers
Col10a and Runx2 to determine whether or not the cells are undergoing terminal differentiation.
Protein expression of chondrogenic markers such as COL2A1 and SOX9 will be assessed via
46
immunofluorescence (IF) analysis and proteoglycan production would be verified with Alcian
blue staining (Fig. 1.8).
Fig. 1.8 – Schematic of the experimental strategy to be used in the derivation of ESC-derived chondrocytes in a defined condition. Growth factors will be added at the onset of differentiation (Day 0) and cells will be
cultured for five days, at which point BMP4, Activin A or Wnt3a can be removed and be replaced by other growth
factors like TGFβ3 and FGF8. Cultures will be continued until day 7 or day 15 of differentiation, and transcript analyses, immunostaining and Alcian blue staining will be conducted to assess the extent of chondrocyte
formation.
When I assessed the extent of chondrocyte formation in my culture system, I observed
that Co2a1 was dramatically up-regulated and there was robust formation of COL2A1 networks
in my cultures, indicating that COL2A1 expression was a suitable readout for my analyses.
However, I could not use IF-based strategy to quantify the percentage of chondrocytes in my
culture because I was visualizing the protein network secreted by all the COL2A1-producing
cells. As such, the third objective (Objective 3) of this project is to generate a transgenic ESC
line that will allow me to identify the transient COL2A1+ population during chondrogenic
differentiation. A construct will be assembled such that Col2a1 promoter will drive the
transcription of a fluorescent protein as the reporter. To facilitate the isolation of transfected
cells, the construct will also be equipped with a selectable marker in the form of a cassette
47
consisting of an antibiotic resistance gene under the transcriptional control of a ubiquitous
promoter, and the cassette can be removed via site-specific recombination from the genome of
the transgenic cell line to minimize the amount of genetic manipulation (Fig. 1.9).
Fig. 1.9 – Design schematic of the reporter construct T2A to be used in the identification of COL2A1+ ESC-
derived chondrocytes generated in the SF monolayer differentiation system.
I plan to assemble my reporter construct using the Gateway® cloning technology which
incorporates fragments of insert into a cloning vector in a modular manner. By separating the
promoter, the fluorescent protein and the antibiotic resistance gene cassette into separate
modules, this reporter system that can be used in a multitude of cell tracking analyses by
substituting in different promoters and reporter genes. Although it is not necessary for my task
at hand, which is to identify COL2A1+ cells in my chondrongenic differentiation cultures, I have
further modified the construct design such that it can be implemented in a two-step reporter
system that will utilize my reporter construct to activate a Cre-inducible cell line (EST2B) that
was previously generated in our laboratory (Handy 2005). To do so, I will insert the sequence
encoding Cre recombinase into my reporter construct such that the activation of the tissue-
specfic promoter will lead to the expression of both the fluoresecent protein and the Cre
recombinase. I will then insert this modified tissue-specific reporter construct into EST2B cells
to create the EST2 transgenic ESC line (Fig. 1.10).
48
Fig. 1.10 – Diagrammatic representation of the derivation of EST2 transgenic ESC line. Linearized T2A
construct will be inserted into EST2B cells via electroporation or lipofection, and the culture will be subjected to
antibiotic selection using both puromycin and G418 24-48hrs. post-transfection. About a week after the start of the
selection process, puromycin/neomycin-resistant ESC clones will arise, at which point each clone will be
individually transferred to a 96-well plate line with MEFs. Clones of interest will be further expanded for further
analyses.
Using my chondrogenic differentiation system as a proof-of-principle assay, the
differentiation of EST2 cells into chondrocytes will activate the Col2a1 promoter and lead to the
transcription of Cre. Cre recombinase will then excise the floxed PuroR that is part of the
construct targetd to the ubiquitous Rosa26 locus (Soriano 1999) of the mouse genome in the Cre-
inducible cell line. The recombination will facilitate the constitutive expression of DsRed T3
RFP in the blasticidin-resistant cells (Fig. 1.11). This two-step system can be used to identify
any ESC-derived transient populations of interest. Furthermore, since the transgenic ESCs will
constitutively express RFP upon Cre excision, all the progeny of the transient cell population can
be identified even after the tissue-specific promoter was no longer active, allowing this versatile
system to be used in various fate-mapping studies.
49
Fig. 1.11 – Schematic of the interaction between tissue-specific promoter-driven reporter construct (T2A) and
Rosa26-targeted Cre-inducible reporter construct (T2B) when the Col2a1 promoter was activated upon
chondrogenic differentiation of the transgenic EST2 cells.
The coupling of the SF monolayer differentiation system with the use of EST2 cells can
have many potential downstream uses. By establishing a minimal system that acts as a baseline
for chondrogenic induction from ESCs, one can use it in the discovery of novel regulators by
screening siRNA, shRNA, over-expression or small molecules libraries. In addition, the
versatility associated with both the differentiation system and construct T2A was greatly
increased due to their modular designs. As such, one can use my design as a blueprint for
establishing screening platforms for a plethora of ESC-derived cell types.
50
Chapter 2 Serum-free derivation of ESC-derived mesoderm and
chondrocytes from monolayer cultures
This chapter is a modified version of the work published in Stem Cell Research titled “One-step
generation of murine embryonic stem cell-derived mesoderm progenitors and chondrocytes in a
serum-free monolayer differentiation system.”
Waese, E. Y. and W. L. Stanford. "One-step generation of murine embryonic stem cell-derived
mesoderm progenitors and chondrocytes in a serum-free monolayer differentiation system."
Stem Cell Research. 2011 Jan; 6(1): 34-49.
51
2.1 Overview
As mentioned in Section 1.7, published ESC models of primitive streak formation and
lineage commitment have validated the possibility of generating functional mesodermal cell
types in defined conditions by adding BMP4, Activin A or Wnt3a on either day 0 or day 2 of
differentiation. I was interested in desigining a culture system for ESC-derived chondrocytes
that could be use as a baseline tool to test the chondrogenic enhancing effects of molecules in the
discovery of novel therapeutics for cartilage repair. Currently, the majority of in vitro
chondrogenic differentiation strategies rely on the establishment of dense pellet or micromass
cultures in serum-containing or conditioned media to mimic mesenchymal condensation.
However, the 3D clustering of heterogeneous cell populations creates an unknown culture
environment that obscures the effects of exogenous factors due to more severe fate-determining
paracrine interactions among various cell populations compared to 2D cultures, while the
presence of serum components masks the effects of growth factors. As such, I decided to
establish my differentiation cultures in monolayers by adding BMP4, Activin A or Wnt3a at the
onset of differentiation to induce mesoderm formation for five days, after which I examined the
effect of the three mesoderm inducers plus known chondrogenic inducers TGFβ3 and FGF8 on
chondrocyte formation by analyzing the expression of the major chondrocyte markers Col2a1,
Sox9, Aggrecan, Col10a and Runx2. I also studied the protein expression of COL2A1 and SOX9
via IF as well as proteoglycan production through Alcian blue staining (Fig. 1.8). I found that
my 2D SF culture system recapitulated the cellular behaviour both in vitro and in vivo in that I
was able to generate poseterior and anterior mesoderm by day 4 of differentiation in cultures
supplemented with BMP4 (10ng/ml) or Activin A (30ng/ml), respectively. I also discovered that
short-term exposures (five days) of ESCs to Activin A (30ng/ml) or BMP4 (25ng/ml) was
52
sufficient to induce chondrocyte formation, while Wnt3a (100ng/ml) only promoted
chondrogenic differentiation as a late inducer after mesoderm specification, consistent with
results obtained in limb bud studies.
2.2 Materials and Methods
2.2.1 Maintenance of ESCs
R1 ESCs (Nagy, Rossant et al. 1993) were thawed onto irradiated MEFs feeder layer in
ES medium containing high glucose Dulbecco’s Modified Eagle Medium (D-MEM, Gibco)
supplemented with GlutaMAX™-1 (2mM) (100x, Gibco), β-mercaptoethanol (0.1%) (1000x,
Factor 1 Day 0-15 Factor 2 Day 5-15 Factor 1 Day 0-5 Factor 2 Day 5-15
Activin
(30ng/ml)
BMP4 (10ng/ml)
Activin
(30ng/ml)
BMP4 (10ng/ml)
Wnt3a (100ng/ml) Wnt3a (100ng/ml)
TGFβ3 (10ng/ml) TGFβ3 (10ng/ml)
FGF8 (50ng/ml) FGF8 (50ng/ml)
BMP4
(10ng/ml)
Activin A (30ng/ml)
BMP4
(10ng/ml)
Activin A (30ng/ml)
Wnt3a Wnt3a
TGFβ3 TGFβ3
FGF8 FGF8
BMP4
(25ng/ml)
Activin A
BMP4
(25ng/ml)
Activin A
Wnt3a Wnt3a
TGFβ3 TGFβ3
FGF8 FGF8
Wnt3a
(100ng/ml)
BMP4
Wnt3a
(100ng/ml)
BMP4
Activin A Activin A
TGFβ3 TGFβ3
FGF8 FGF8
TGFβ3 (10ng/ml)
FGF8 (50ng/ml)
Serum (15%)
Table 2.1– List of test conditions used to examine the effects of BMP4, Activin A, Wnt3a, TGFβ3, FGF8 and serum on monolayer chondrogenic differentiation from R1 ESCs in chemically defined conditions. Factor 1
was added on day 0 of differentiation and was withdrawn after 15 days or 5 days of culture while Factor 2
supplementation commenced on day 5 of differentiation until the end of the time course.
58
2.3 Results
2.3.1 N2B27 supported ESC adhesion and proliferation on collagen IV
I examined the combined effects of LIF-supplemented SF medium formulations, ECMs
and seeding densities on OCT4 expression in undifferentiated ESCs. LIF- and BMP4-
supplemented N2B27 medium has been previously shown to maintain undifferentiated ESCs in
culture (Ying, Nichols et al. 2003). N2B27 has also been used in various ESC mesoderm
differentiation studies (Gadue, Huber et al. 2006; Nostro, Cheng et al. 2008; Purpura, Morin et
al. 2008). CDM was formulated to study the roles of Activin A and BMP4 in mesoderm and
hematopoietic development (Johansson and Wiles 1995); while X-Vivo™10 was developed for
human hematopoietic cells and ESC cultures. For ECM selection, gelatin has been widely used
in ESC cultures, gelatin+fibronectin has been commonly used in HCI assays (Davey and
Zandstra 2006; Walker, Ohishi et al. 2007) and collagen IV has been used in serum monolayer
differentiation cultures (Nishikawa, Nishikawa et al. 1998; Tada, Era et al. 2005; Sakurai, Era et
al. 2006; Fujiwara, Hayashi et al. 2007). Seeding densities were set at 6x104; 3x10
4; and
1.5x104cells/cm
2.
Histograms representing OCT4 expression profiles were generated from the fluorescence
intensity data quantified by HCI analysis. Regardless of seeding density or medium formulation,
two-day cultures established on gelatin+fibronectin demonstrated significant loss in OCT4
expression with a clear separation between OCT4+ and OCT4
- populations (Figs. 2.1A and
A.1A), while OCT4 levels remained high in gelatin or collagen IV cultures (Figs. 2.1B and
A.1B, respectively). OCT4 expression decreased in gelatin+fibronectin cultures as cell density
decreased (Fig. 2.1Ci) while it varied in collagen IV cultures depending on SF medium
formulation and remained steady in gelatin cultures (Fig. A.1B). OCT4 levels decreased and
59
varied greatly in collagen IV cultures seeded at 1.5x104cells/cm
2 in X-Vivo™10 medium due to
the scarcity of colonies, while the opposite trend was observed in CDM cultures due to over-
confluency (Fig. 2.1Cii). Therefore, ESC cultures established on collagen IV and maintained in
LIF-supplemented N2B27 medium were most tolerant to varying seeding densities.
A
B
C
Fig. 2.1 – Two-day ESC cultures on collagen IV in N2B27 medium with LIF maintained high OCT4 expression. (A) N2B27 cultures on (i) gelatin+fibronectin exhibited a biomodal distribution of OCT4 levels
compared to those on (ii) gelatin and (iii) collagen IV, leading to a lower percentage of OCT4+ population (B). (C) (i) OCT4 levels of gelatin+fibronectin cultures reduced with decreasing seeding density regardless of SF medium
formulation. (ii) OCT4 expression varied dramatically at a high seeding density in X-Vivo cultures on collagen IV,
while the opposite trend was observed in CDM cultures. OCT4 expression remained relatively stable at all seeding
densities in N2B27 cultures on collagen IV. Plotted values represented means±SEM (n=2).
factor-supplemented cultures displayed stronger cell-matrix adhesion than untreated SF cultures
consisting of tightly packed colonies that detached easily from culture surface (Fig. 2.2v).
I determined whether differences in cell adhesion capabilities in growth factor-
supplemented cultures were indicative of a compromise in cell survival, specifically that of the
nascent mesodermal cells. As such, I examined the expression of the early mesoderm marker
FLK1 in my SF monolayer differentiation cultures. Flow cytometry analysis of FLK1:eGFP
ESCs (Ema, Takahashi et al. 2006) differentiated for four days in SF monolayer cultures showed
that FLK1:eGFP expression was significantly higher in Activin A-supplemented cultures
compared to untreated cultures (Fig. 2.3A). Interestingly, greater Flk1 transcript expression was
observed in BMP4- or Wnt3a-supplemented cultures than that in Activin A-treated cultures (Fig.
2.3B). Expression of the apoptotic marker ANNEXIN V was significantly higher in BMP4-
supplemented cultures than in Activin A-treated cultures (Fig. 2.3C). Dot plots showed that <1%
61
of the population co-expressed FLK1:eGFP and ANNEXIN V (Fig. 2.3D), indicating that
monolayer ESC cultures established on collagen IV in growth factor-supplemented SF medium
supported mesoderm differentiation.
Fig. 2.2 –Morphologies of four-day SF, growth factor-supplemented ESC monolayer differentiation cultures established on collagen IV. Brightfield images (100x) taken with the camera-mounted Leica DM IL inverted
microscope illustrated that serum cultures (i) exhibited more pronounced cell adhesion and spreading than SF
Activin A (ii) and Wnt3a (iii) cultures. Raised colonies (arrows) were present in BMP4 cultures (iv), while untreated
cultures (“No GF”) (v) consisted of tightly packed cell populations that adhered poorly.
62
A Flow cytometry B qPCR
C Flow cytometry
D Flow cytometry
Fig. 2.3 – Characteristics of four-day SF, growth factor-supplemented ESC monolayer differentiation cultures established on collagen IV. (A) Flow cytometric analyses and (B) qPCR indicated that although Activin
A, BMP4 and Wnt3a induced FLK1:eGFP expression and Flk1 transcript in SF monolayer differentiation cultures,
respectively, expression of the apoptotic marker Annexin V was distinctly higher in BMP4 cultures (C). (D) Flow
cytometric dot plots of Annexin V vs. FLK1:eGFP showed that nascent mesodermal cells generated in the SF
monolayer cultures were not apoptotic, as <1% of the population expressed both markers (highlighted in red). Transcript levels were compared to those in undifferentiated ESCs. Plotted values from flow cytometric analyses
and qPCR represent means±SEM (n≥2).
2.3.3 Endogenous Wnt3a was up-regulated in serum cultures as well as BMP4-supplemented and untreated SF differentiation cultures
Endogenous Bmp4, Nodal and Wnt3a expression in SF differentiation cultures was
quantified by qPCR to determine if they were specifically up-regulated by their respective
exogenous ligands. Activin A-supplemented cultures had significantly higher Nodal transcript
% F
LK
1-p
ositiv
e c
ells
%
AN
NE
XIN
V-p
ositiv
e c
ells
63
levels than BMP4-supplemented and untreated four-day SF cultures (Fig. 2.4A). Interestingly,
Bmp4 expression was up-regulated by either BMP4 or Wnt3a ligand (Fig. 2.4B). The expression
of Nodal, Bmp4 and Wnt3a increased competitively in serum cultures. Surprisingly, Wnt3a
expression was markedly up-regulated to comparable levels in both untreated and Wnt3a-
supplemented cultures (Fig. 2.4C).
A Nodal
B Bmp4
C Wnt3a
Fig. 2.4 – Potential synergistic effects of Activin A, BMP4 and Wnt3a in four-day SF, growth factor-supplemented ESC monolayer differentiation cultures. Quantitative PCR analyses of endogenous expression of
(A) Nodal; (B) Bmp4 and (C) Wnt3a mRNA showed that Wnt3a was significantly up-regulated in BMP4-
supplemented cultures. Expression levels were compared to those in undifferentiated ESCs. Plotted values
represent means±SEM (n≥2).
Rela
tive
Expre
ssio
n o
f B
mp
4
64
Fig. 2.5 – Addition of exogenous Activin A and Wnt3a led to robust induction of BRACHYURY protein expression in four-day monolayer differentiation cultures. IF images (200x) showed that (i) Activin A, (ii)
Wnt3a and (iii) serum cultures displayed comparable BRACHYURY protein levels, while (iv) BMP4 cultures
appeared to have less BRACHYURY+ cells. (v) BRACHYURY was not detected in untreated cultures.
2.3.4 BMP4, Activin A or Wnt3a induced BRACHYURY+ primitive streak-like populations in monolayer differentiation cultures
To corroborate with the FLK1 results (Fig. 2.3A), four-day monolayer cultures were
immunostained for expression of the primitive streak/early mesoderm marker BRACHYURY.
Activin A-, Wnt3a- or serum-supplemented cultures showed comparable levels of
BRACHYURY expression (Fig. 2.5i-iii). BMP4 did not induce BRACHYURY expression as
robustly as Activin A or Wnt3a (Fig. 2.5iv); however, qPCR analysis results demonstrated
similar Brachyury transcript levels in all growth factor-supplemented cultures (Fig. 2.6A). HCI
and IF analyses showed that BRACHYURY was induced in a dose-dependent manner (Figs.
2.6B, A.2A). Addition of Activin A or Wnt3a to BMP4-supplemented cultures increased
BRACHYURY expression, suggesting that Activin A and Wnt3a were synergistic inducers of
BRACHYURY
DAPI/Hoechst
65
primitive streak-like cells at the tested concentrations (Fig. A.2Bi-ii) but their simultaneous
presence did not further increase BRACHYURY expression (Fig. A.2Biii). SF media alone did
not induce noticeable BRACHYURY expression (Fig. 2.5v). As expected, BRACHYURY
expression was significantly higher in serum-containing cultures than in SF conditions, while
serum components masked the inductive effect of Activin A on BRACHYURY expression (Figs.
2.6C and A.2Biv).
A qPCR
B HCI analysis
C HCI analysis
Fig. 2.6 – Early mesoderm specification in four-day growth factor-supplemented SF monolayer differentiation cultures. (A) qPCR results suggested that Activin A, BMP4 and Wnt3a exerted similar inductive
effects on Brachyury transcription while serum effect was the most potent. Transcript levels were compared to
those in undifferentiated ESCs. Plotted values represent means±SEM (n≥3). (B) HCI analysis demonstrated that
BRACHYURY protein level was directly correlated with growth factor concentration (e.g. Activin A), while this
effect was masked in serum cultures (C). Plotted values represent means±SEM (n=2).
% B
RA
CH
YU
RY
-positiv
e c
ells
% B
RA
CH
YU
RY
-positiv
e c
ells
66
2.3.5 Mesoderm marker genes expression patterns correlated with those in EB cultures and in murine embryos studies
Previous reports showed that BMP4 has a posteriorizing effect on differentiating ESCs,
while Activin A promotes the formation of increasingly more anterior populations in a
concentration-dependent manner (reviewed in (Murry and Keller 2008)). Based on marker
expression, I observed similar growth factor-dependent enrichment of mesoderm subsets in my
four-day monolayer cultures. Expression of the posterior mesoderm markers Even-skipped
homeobox 1 (Evx1), Homeobox B1 (HoxB1), T-cell acute leukemia 1 (Tal1) and GATA2 was
dramatically up-regulated in BMP4-supplemented cultures (Fig. 2.7i-iv). LIM homeobox 1
(Lhx1) expression was induced by Activin A, BMP4 and Wnt3a (Fig. 2.7v) while BMP4 and
Wnt3a supplementation led to higher transcript levels of the paraxial mesoderm marker Pdgfrα
compared to Activin A (Fig. 2.7vi). Activin A and Wnt3a, but not BMP4, induced the
expression of the anterior marker Mesenchyme homeobox 2 (Meox2) (Fig. 2.7vii), while all three
growth factors exerted similar inductive effects on Follistatin (Fst) and Mesoderm posterior 2
(Mesp2) expression (Fig. 2.7viii-ix). As expected, expression of the mesendoderm markers
Goosecoid (Gsc) and Forkhead box a2 (Foxa2) was up-regulated by Activin A (Fig. 2.7x-xi).
Wnt3a appeared to have a pan-mesodermal inductive effect based on transcript level analysis.
Therefore, differential growth factor supplementation at the onset of differentiation facilitated
enrichment of mesoderm subsets in monolayer cultures without cell sorting.
67
i) Evx1
ii) HoxB1
iii) Tal1
iv) GATA2
v) Lhx1 vi) Pdgfrα
vii) Meox2
viii) Fst
ix) Mesp2
x) Gsc
xi) Foxa2
Fig. 2.7 – Quantitative PCR analysis showed that BMP4, Activin A and Wnt3a induced the expression of marker genes of various mesoderm subsets. BMP4 was more inductive in the up-regulation of the posterior
primitive streak markers (i) Evx1, (ii) HoxB1, (iii) Tal1 and (iv) GATA2 while Activin A appeared to be equally
inductive in (v) Lhx1. (vi) BMP4 was more effective in inducing the paraxial mesoderm marker Pdgfrα than
Activin A, which up-regulated the anterior mesoderm marker (vii) Meox2, (viii) Fst and (ix) Mesp2 to a lesser extent. Activin A effectively up-regulated the mesendoderm markers (x) Gsc and (xi) Foxa2, and Wnt3a appeared
to have a pan-mesodermal inductive effect. Transcript levels were compared to those in undifferentiated ESCs.
Plotted values represent means±SEM (n≥2).
Re
lativ
e E
xp
ressi
on
of P
dg
frα
68
2.3.6 Activin A facilitated chondrogenic differentiation in SF monolayer cultures
Fifteen-day growth factor-supplemented monolayer cultures were established to
determine if prolonged exposure to mesoderm inducers could trigger chondrogenic induction.
Activin A-supplemented (30ng/ml) cultures (Fig. 2.8Ai) showed more intense Alcian blue
staining than those treated with BMP4 (10ng/ml), Wnt3a (100ng/ml) or serum (Fig. 2.8Aii-iv);
also, robust COL2A1 networks were only present in Activin A-supplemented cultures (Fig.
2.8C). SOX9-positive cells were also present in Activin A-supplemented cultures (Fig. 2.8D).
Similar to four-day cultures, 15-day untreated SF differentiation cultures demonstrated poor cell-
matrix adhesion with the formation of EB-like structures that were loosely anchored via
filamentous protrusions (Fig. 2.8B) and were easily dislodged during media replenishment.
Real-time qPCR analysis showed that Activin A-supplemented cultures showed marked
up-regulation of the chondrogenic markers Col2a1, Sox9 and Aggrecan while cultures with
BMP4, Wnt3a and serum showed minimal changes in gene expression (Fig. 2.9A-C). Activin A
did not strongly enhance the expression of Col10a and Runx2 compared to non-inductive
conditions (Fig. 2.9D), suggesting the maintenance of non-hypertrophic chondrocytes after 15
Activin A culture (i) was more intensely stained with Alcian blue than BMP4 (ii), Wnt3a (iiii) and serum (iv)
cultures. (B) Untreated SF cultures adhered poorly and formed aggregates loosely anchored on the culture surface
(400x). (C) COL2A1 networks were formed in (i) Activin A but not (ii) BMP4, (iii) Wnt3a or (iv) serum cultures
(200x). (D) SOX9 was expressed in 11-day (i) Activin A cultures but not in (ii) BMP4-, (iii) Wnt3a- or (iv) serum-
treated cultures (200x).
COL2A1
DAPI/Hoechst
SOX9
DAPI/Hoechst
70
A
B
C
D
Fig. 2.9 – Real-time qPCR results confirmed the up-regulation of (A) Col2a1, (B) Sox9 and (C) Aggrecan in
day 7 and day 15 of Activin A-supplemented SF monolayer differentiation cultures, while the levels of hypertrophic markers Col10a and Runx2 were similar to non-inductive conditions (D). Expression levels were
compared to those in undifferentiated ESCs. Transcript levels were compared to those in undifferentiated ESCs.
Plotted values represent means±SEM (n≥3).
Comparison of the expression levels of chondrogenic marker genes between monolayer
and micromass cultures suggested that both systems behaved similarly. Interestingly, I was
unable to generate micromasses in cultures supplemented with chondrogenic media (Woods,
Wang et al. 2007) due to poor adhesive properties of the droplets. Micromass cultures
established in SF media supplemented with Activin A (30ng/ml) exhibited similar levels of
71
Col2a1 and Col10a as my monolayer cultures, with slightly enhanced Sox9 levels and lower
Aggrecan expression (Fig. 2.10A compared to Fig. 2.9A-D). However, IF analysis of COL2A1
expression highlighted the shortcoming of the 3D micromass culture in that clear network
formation could only be visualized at the periphery of the micromass (Fig. 2.10B compared to
Fig. 2.8Ci), while the resolution of the image deteriorated towards the centre of the micromass.
As such, it would be difficult to quantify protein expression in these 3D cultures using HCI
strategies.
A
B
Fig. 2.10 – Formation of micromasses after 15 days of differentiation in SF medium supplemented with Activin A (30ng/ml). (A) Real-time qPCR analysis of chondrogenic marker gene expression in 15-day micromass
cultures established in SF media supplemented with Activin A (30ng/ml). Compared to the Activin A-treated SF
monolayer cultures, Col2a1 and Col10a expression levels were similar in both culture systems; however, Sox9
expression appeared to be higher in micromass cultures while Aggrecan expression was slightly inferior to that in
monolayer cultures. Transcript levels were compared to those in undifferentiated ESCs. Plotted values represent
means±SEM (n=2). (B) Bright field (i) and IF (ii) images of COL2A1 protein expression in micromass established
in SF media supplemented with Activin A (30ng/ml). COL2A1 network was visible at the periphery of the
micromass while the staining became blurred and out of focus towards the denser part of the structure, hence
highlighting the disadvantage of using 3D cultures in imaging applications.
COL2A1
DAPI/Hoechst
Col2a1 Sox9 Aggrecan Col10a
72
2.3.7 TGFβ3 induced chondrocyte formation when added at the onset of differentiation
Treatment of SF monolayer cultures with TGFβ3 (10ng/ml) (Fig. 2.11Ai-ii) or FGF8
(50ng/ml) (Fig. 2.11Aiii-iv) induced chondrogenic differentiation with evident COL2A1 network
formation beginning on day 7 of differentiation. However, cells cultured in FGF8 alone
exhibited poor cell-matrix adhesion similar to untreated SF cultures. Aside from COL2A1, both
FGF8- and TGFβ3-treated cultures also possessed SOX9-expressing populations (Fig. 2.11B).
TGFβ3 and FGF8 were not superior to Activin A in their chondrogenic inductive abilities.
Compared to Activin A-supplemented cultures, Sox9 and Col10a transcript expression only
increased minimally in TGFβ3- or FGF8-supplemented cultures, respectively (Fig. 2.11C).
COL2A1 protein was undetectable in BMP4-supplemented cultures treated with TGFβ3
or FGF8 beginning on day 5 of differentiation (Fig. 2.12Ai-ii). Despite its confirmed role as a
chondrogenic inducer, Activin A addition to BMP4-supplemented cultures failed to induce
COL2A1 or proteoglycans production (Fig. A.3A-B, respectively). These data suggest that
either BMP4 exerted a dominant chondrogenic inhibitory effect on my SF monolayer cultures, or
Activin A, TGFβ3 and FGF8 functioned early on during chondrogenic induction. Although
strong COL2A1 networks were formed in Activin A-supplemented cultures containing TGFβ3 or
FGF8 (Fig. 2.12Aiii-iv) and this result was corroborated by Alcian blue staining (Fig. 2.12B),
qPCR results indicate that TGFβ3 or FGF8 addition to Activin A-supplemented cultures did not
further enhance chondrogenic markers gene expression. Interestingly, the presence of TGFβ3 or
FGF8 in BMP4-supplemented cultures increased the gene expression of Col2a1, Sox9 and
Aggrecan (in the case of TGFβ3) compared to BMP4 alone (Fig. 2.12C).
73
A COL2A1
B SOX9
C
Fig. 2.11 – Supplementation of SF differentiating cultures with FGF8 or TGFβ3, beginning on day 0 of differentiation, was able to induce chondrogenic differentiation. (A) FGF8 (i-ii) and TGFβ3 (iii-iv) were also
found to induce COL2A1 expression when added alone to SF monolayer differentiation cultures. However, FGF8
cultures exhibited similar morphology as shown in Fig. 2.8D. (B) 11-day (i) FGF8- and (ii) TGFβ3-treated cultures
also consisted of SOX9-positive populations. (C) Compared to Activin A, TGFβ3 appeared to be more potent in
inducing Sox9 expression, while FGF8 further up-regulated Col10a transcript level. Transcript levels were compared to those in undifferentiated ESCs. Plotted values represent ratios of means±relative errors (n≥3).
COL2A1
DAPI/Hoechst
SOX9
DAPI/Hoechst
74
A
B
C
Fig. 2.12 – Supplementation of SF BMP4- or Activin A-treated differentiating cultures with FGF8 or TGFβ3 did not dramatically enhance chondrocyte formation. (A) When added as a potential enhancer to BMP4 (i-ii) or
Activin A (iii-iv), TGFβ3 and FGF8 did not have a noticeable effect on COL2A1 expression, as supported by Alcian blue staining (B). Interestingly, the addition of those two factors to BMP4-supplemented cultures markedly
improved Col2a1, Sox9 and Aggrecan expression from BMP4 addition alone. However, their effects were not as
pronounced in Activin A-supplemented cultures (C). Transcript levels were compared to those in undifferentiated
ESCs. Plotted values represent ratios of means±relative errors (n≥3).
COL2A1
DAPI/Hoechst
75
2.3.8 Five-day Activin A treatment achieved competitive chondrogenic differentiation in SF monolayer cultures
Since my data suggest that Activin A acts as an early inducer of chondrogenic
differentiation, I examined the feasibility of shortening the duration of supplementation. Robust
COL2A1 networks were present in 15-day SF monolayer cultures supplemented with Activin A
for the first five days of differentiation but not in BMP4 or Wnt3a cultures (Fig. 2.13Ai-iii).
Similarly, SOX9 protein expression was detected in Activin A-supplemented cultures (Fig.
2.13Bi); however, there appeared to be very weak SOX9 expression in Wnt3a-treated cultures
also (Fig. 2.13Biii). In terms of transcript levels, five-day Activin A supplementation led to
increased expression of Col2a1, Sox9, Aggrecan and Col10a compared to BMP4- and Wnt3a-
treated cultures (Fig. 2.13C).
I examined whether the non-inductive BMP4 (10ng/ml) and Wnt3a (100ng/ml) would
inhibit the progression of chondrogenic induction initiated by five-day Activin A
supplementation. IF data indicated that BMP4 or Wnt3a addition from day 5-15 of
differentiation to Activin A-supplemented cultures did not hinder COL2A1 network formation
(Fig. 2.14Ai-ii). The replacement of BMP4 or Wnt3a with TGFβ3 or FGF8 did not further
enhance COL2A1 protein expression in differentiation cultures initiated by Activin A (Fig.
2.14Aiii-iv). However, the presence of FGF8 or TGFβ3 in Activin A-supplemented cultures
increased Aggrecan expression, while those with TGFβ3 showed up-regulation in Sox9 and
Runx2 to a very small extent. FGF8 treatment also led to a marginal increase in Col10a
expression. Interestingly, addition of Wnt3a to Activin A-supplemented differentiation cultures
also marginally enhanced Sox9, Aggrecan and Runx2 expression, suggesting that Wnt3a may
play a minor role as a chondrogenic inducer (Fig. 2.14B). Similar to the results obtained from
76
cultures with continuous BMP4 supplementation, Activin A, TGFβ3 or FGF8 addition to
cultures with five-day BMP4 treatment did not induce COL2A1 expression (Fig. A.4).
A
B
C
Fig. 2.13 – Chondrogenic differentiation was achieved in SF monolayer cultures supplemented with Activin A from day 0-5 of differentiation. (A) IF images (200x) showed that (i) Activin A addition on the first five days of
differentiation was sufficient to generate COL2A1 networks in 15-day cultures, while (ii) BMP4 and (iii) Wnt3a
treatment failed to do so. (B) IF analyses (400x) also confirmed the expression of SOX9 in 11-day cultures
subjected to five-day Activin A supplementation while (ii) BMP4- and (iii) Wnt3a had faint to no positive staining.
(C) Real-time qPCR analyses of five-day supplementation cultures reflected similar up-regulation patterns in Col2a1, Sox9 and Aggrecan as 15-day supplementation cultures. Col10a transcript level was slightly increased in
cultures with five-day Activin A treatment, but Runx2 expression was unaffected. Transcript levels were compared
to those in undifferentiated ESCs. Plotted values represent means±SEM (n≥3).
COL2A1
DAPI/Hoechst
SOX9
DAPI/Hoechst
CO
L2
A1
S
OX
9
77
A
B
Fig. 2.14 – Sequential addition of growth factors did not lead to dramatically enhanced chondrogenic induction. (A) IF images (200x) of day 15 differentiation cultures supplemented with Activin A for the first five
days followed by 10-day addition of (i) BMP4, (ii) Wnt3a, (iii) FGF8 and (iv) TGFβ3 all showed similar extent of
COL2A1 network formation. (B) Real-time qPCR results suggested that culture supplemented with Activin A for
five days followed by BMP4 for 10 days did not have notable effect on chondrogenic induction. Replacement of
BMP4 with FGF8 caused a marginal increase in Aggrecan and Col10a expression, while addition of Wnt3a or
TGFβ3 facilitated slight increases in Sox9, Aggrecan and Runx2. Transcript levels were compared to those in
undifferentiated ESCs. Plotted values represent ratios of means±relative errors (n≥3).
COL2A1
DAPI/Hoechst
78
Fifteen-day cultures with continuous Activin A supplementation maintained higher
expression of Prg4 than those subjected to a five-day treatment regime (Fig. 2.15). Prg4 is
specifically expressed in chondrocytes located at the surface of articular cartilage. Replacement
of Activin A with TGFβ3 on day 5 of differentiation induced similar or higher Prg4 transcript
levels than Activin A alone, while the presence of FGF8 did not further promote Prg4 expression
regardless of the length of Activin A supplementation. Therefore, sustained Activin A
supplementation or the combination of Activin A and TGFβ3 appeared to facilitate articular
chondrocyte formation in my SF monolayer culture system.
Fig. 2.15 – Real-time qPCR analysis of Prg4 expression suggested that sustained
Activin A supplementation (“A”) or the
sequential addition of Activin A followed
by TGFβ3 (“T”) appeared to promote
articular chondrocyte formation. Transcript levels were compared to those in
undifferentiated ESCs. Plotted values
represent means±SEM (n≥3). Note:
“F”=FGF8.
2.3.9 High BMP4 concentration induced chondrogenic differentiation, while Wnt3a acted as a late chondrogenic inducer
To further investigate the role of BMP4 and Wnt3a in my SF monolayer chondrogenic
cultures, I differentiated ESCs in the presence of Wnt3a (100ng/ml) (for 15 days or five days)
followed by BMP4 (10ng/ml) (added from day 5-15 of culture) and in cultures supplemented by
the two growth factors in the reversed order. IF analyses of COL2A1 deposition confirmed that
Rela
tive E
xpre
ssio
n o
f P
rg4
79
Wnt3a was ineffective in chondrogenic induction when added on day 0 of differentiation, while
further addition of BMP4 exerted minimal effect on chondrogenic differentiation (Fig. 2.16Ai-
ii). Addition of Activin A, TGFβ3 or FGF8 to Wnt3a-supplemented cultures did not facilitate
COL2A1 network formation or proteoglycans production (Fig. A.5A-B), reinforcing their
described roles as early chondrogenic inducers. When the order of Wnt3a and BMP4
administration was reversed, however, COL2A1 networks were successfully formed (Fig.
2.16Aiii-iv). Compared to cultures supplemented with Wnt3a at the onset of differentiation,
treatment with BMP4 followed by Wnt3a led to enhanced expression of Col2a1 and Sox9 as well
as the late chondrogenic marker Col10a and the osteogenic transcription factor Runx2 (Fig.
2.16B), suggesting that Wnt3a acted as a late inducer of chondrogenesis in place of BMP4
(10ng/ml). Alcian blue staining was only present in cultures supplemented with BMP4 followed
by Wnt3a (Fig. 2.16Ciii-iv) albeit with less intensity than those present in chondrogenic cultures
induced by Activin A. The decreased proteoglycan deposition was also reflected in Aggrecan
transcript expression levels (Fig. 2.16B).
Contrary to published data establishing BMP4 as an inducer of ESC chondrogenesis
(Heng, Cao et al. 2004; van Osch, Brittberg et al. 2009; Vinatier, Mrugala et al. 2009), BMP4
(10ng/ml) did not have an appreciable effect on chondrogenic differentiation in my culture
system. To explain this disparity, I examined the concentration effect of BMP4 on my
monolayer differentiation cultures. With 25ng/ml of BMP4, robust COL2A1 networks were
evident regardless of the duration of supplementation (Fig. 2.17Ai-ii). Similar to Activin A,
addition of TGFβ3 or FGF8 to BMP4-supplemented cultures showed positive COL2A1 staining
as BMP4 alone (Fig. 2.17Aiii-iv, v-vi, respectively). Although overall transcript levels were
lower than those in Activin A-treated cultures, IF data were corroborated by qPCR (Fig. 2.17B).
Specifically, Aggrecan gene expression was similar to data obtained from cultures treated with
80
BMP4 (10ng/ml) followed by Wnt3a and was similarly reflected by Alcian blue staining (Fig.
2.17C). These findings confirm that Activin A (30ng/ml) and TGFβ3 (10ng/ml) are more
effective than BMP4 (25ng/ml) as early inducers of chondrogenesis.
A
B
C
Fig. 2.16 –Wnt3a acted as a late inducer to amplify the effect of BMP4 (10ng/ml). (A) Wnt3a addition at the
onset of differentiation did not lead to COL2A1 expression even with further addition of BMP4 (i-ii), but robust
networks were formed in BMP4 (10ng/ml) cultures supplemented with Wnt3a as a secondary factor (iii-iv) (200x).
(B) Compared to the treatment regime of adding Wnt3a (“W”) at the onset of differentiation followed by BMP4
(“B”), Wnt3a treatment of BMP4-treated cultures beginning on day 5 of differentiation led to increases in Col2a1,
Sox9, Col10a and Runx2 transcripts levels but not Aggrecan, regardless of the duration of BMP4 supplementation.
“F1”=factor added at the onset of differentiation. Transcript levels were compared to those in undifferentiated
ESCs. Plotted values represent ratios of means±SEM (n≥3). (C) Alcian blue staining reinforced the observation that Wnt3a acted as a late inducer of chondrogenesis.
COL2A1
DAPI/Hoechst
81
A
B
C
Fig. 2.17 –BMP4 (25ng/ml) acted as an early inducer of chondrogenesis. (A) BMP4 acted as an early
chondrogenic inducer with robust COL2A1 network formation regardless of a 15-day (i, iii, v) or five-day (ii, iv, vi)
supplementation regime. Further addition of (iii-iv) TGFβ3 or (v-vi) FGF8 did not have noticeable enhancing effect
on COL2A1 network formation (200x). (B) qPCR analysis showed slight increases in Col2a1 and Sox9 in five-day
BMP4 (“25B”) cultures compared to a 15-day treatment schedule. Additional treatment with TGFβ3 or FGF8 did not have significant enhancing effects on chondrogenic markers expression. Transcript levels were compared to
those in undifferentiated ESCs. Plotted values represent means±SEM (n≥3). (C) Alcian blue staining confirmed the production of proteoglycans in cultures with BMP4 under both (i-iii) 15-day and (iv-vi) five-day supplementation
schedules. Presence of TGFβ3 (ii, v) and FGF8 (iii, vi) did not lead to increased proteoglycan production. Staining
intensity in BMP4 (25ng/ml) cultures was weaker than those supplemented with Activin A.
COL2A1
DAPI/Hoechst
82
2.4 Discussion
In this study, I examined if long-term treatment of mESC SF monolayer cultures with
BMP4, Activin A, Wnt3a, TGFβ3 and FGF8 at the onset of differentiation could direct
chondrogenic differentiation with minimal culture manipulation.
ECM selection was crucial as two-day LIF-supplemented SF ESC cultures established on
gelatin+fibronectin exhibited accelerated differentiation with 5-10% lower OCT4 levels
compared to cultures on gelatin or collagen IV (Fig. 2.1A- B). The bimodal OCT4 expression
profile was commonly observed in other HCI assays established on gelatin+fibronectin (Davey
and Zandstra 2006; Walker, Ohishi et al. 2007). Although fibronectin is endogenously expressed
by differentiating ESCs (Hayashi, Furue et al. 2007) and promotes cell adhesion and spreading
(Dufour, Duband et al. 1986), I rejected gelatin+fibronectin for my system to minimize
spontaneous ESC differentiation towards undesired lineages in the absence of inductive factors.
Gelatin, being a mixture of collagens, was less defined than collagen IV, and collagen IV has
been shown to facilitate ESC differentiation towards the mesodermal lineages (Nishikawa,
Nishikawa et al. 1998; Tada, Era et al. 2005; Sakurai, Era et al. 2006). Despite the successful
establishment of adherent chondrocyte cultures on fibronectin or collagen (Ho, Yang et al. 2009;
Khan, Bishop et al. 2009), gelatin+fibronectin cultures did not show enhanced BRACHYURY or
chondrogenic markers expression compared to collagen IV cultures (data not shown).
Similar to published EB studies (Nostro, Cheng et al. 2008), BMP4, Activin A and
Wnt3a all induced FLK1 (Fig. 2.3A-B) and BRACHYURY expression (Figs. 2.5 and 2.6A) in
my four-day monolayer differentiation cultures. Activin A- or Wnt3a-treated cultures consisted
of flattened colonies with stronger adhesion and spreading on collagen IV than with BMP4 (Fig.
83
2.2). Conversely, there was ~6-8% more of ANNEXIN V+ apoptotic cells present in BMP4
cultures than in Wnt3a and Activin A cultures (Fig. 2.3C). Activation of the TGFβ signaling
pathway induces EMT, leading to the up-regulation of Neural cell adhesion molecule (NCAM)
(Thiery and Sleeman 2006). NCAM promotes the phosphorylation of Focal adhesion kinase
(FAK) and integrin-dependent cell spreading (Frame and Inman 2008). FAK phosphorylation
alters its downstream target Growth factor receptor-bound protein 2 (GRB2) and facilitates its
interaction with the Ras/Mitogen-activated protein kinase (MAPK) pathway, which modulates
cell survival and proliferation (Schlaepfer, Hanks et al. 1994; Harburger and Calderwood 2009).
Members of the canonical Wnt and integrin signaling pathways (specifically the collagen-
binding integrins α1β1 and α2β1) have been shown to act synergistically via GRB2, (Crampton,
Wu et al. 2009), possibly contributing to the satisfactory cell spreading and survival observed in
Wnt3a-supplemented cultures. The non-uniform morphology of differentiating colonies in BMP4
culture (Fig. 2.2Aiv) could be due to the potency of BMP4 at 10ng/ml as the cultures also
showed weaker BRACHYURY protein expression (Fig. 2.5iv); consequently, serum-deprivation
promoted apoptosis in the slowly differentiating cells. Alternatively, although BMP4 (10ng/ml)
was less potent than Activin A or Wnt3a in EMT initiation and mesoderm induction, its presence
may be sufficient to prevent neuroectoderm differentiation in my culture system by inducing
apoptosis in early precursors of neural cells (Gambaro, Aberdam et al. 2006).
Up-regulation of endogenous Wnt3a in both four-day untreated and BMP4-supplemented
monolayer differentiation cultures (Fig. 2.4C) suggested possible crosstalk between BMP4 and
Wnt3a. Recently, exogenous BMP4 was found to cause increases in Wnt3a levels in a similar
monolayer differentiation system and both signaling pathways functioned synergistically to
induce different mesoderm populations (Tanaka, Jokubaitis et al. 2009). Components in the
N2B27 medium may also induce Wnt3a expression that contributed to the seemingly pan-
84
mesodermal inductive ability of the Wnt3a ligand (Fig. 2.7). However, qPCR data from 15-day
cultures showed that exogenous Wnt3a was less inductive than Activin A in the expression of the
anterior cardiac markers α Myosin heavy chain (αMHC) and NK2 transcription factor related,
locus 5 (Drosophila) (Nkx2.5) (Fig. A.6i-ii and A.7i-ii) while it promoted the expression of the
posterior hematopoietic marker GATA1 (Fig. A.6iii and A.7iii). The data was consistent with
reports showing that WNT3a inhibited cardiomyocyte differentiation upon mesoderm induction
(Naito, Shiojima et al. 2006; Ueno, Weidinger et al. 2007). My SF monolayer differentiation
system did not appear to support the formation of definitive endoderm (Fig. A.6iv and A.7iv).
Posterior and anterior mesodermal populations were enriched without cell sorting in
growth factor-supplemented SF monolayer cultures. Similar to EB studies (Gadue, Huber et al.
2006), there were clear increases in posterior mesoderm marker genes expression (Evx1, HoxB1,
Tal1 and GATA2) in BMP4-supplemented cultures compared to Activin A-supplemented
cultures (Fig. 2.7i-iv), while the mesendoderm markers Gsc and Foxa2 exhibited the opposite
expression patterns (Fig. 2.7x-xi). This distinction was not as definitive in the expression of
paraxial mesoderm marker genes like Fst and Mesp2 (Fig. 2.7viii-ix). This phenomenon was
expected because lateral plate and paraxial mesoderm form adjacent to each other in
development with some overlapping of gene expression patterns. Similarly, although Lhx1 has
been identified as a marker for lateral plate mesoderm (Tam and Loebel 2007) and I anticipated
higher Lhx1 expression in BMP4-treated cultures than in Activin A cultures, I observed
comparable Lhx1 up-regulation in Activin A-, BMP4- and Wnt3a-supplemented cultures (Fig.
2.7v). This was logical because Lhx1 is a known target of the Nodal signaling pathway (Shen
2007) and is also expressed in lateral-intermediate mesoderm, anterior mesendoderm and
visceral endoderm (Shawlot, Wakamiya et al. 1999; Tsang, Shawlot et al. 2000; Tam, Khoo et al.
2004). In contrast, despite the use of Pdgfrα to characterize ESC-derived paraxial mesoderm
85
(Sakurai, Era et al. 2006; Sakurai, Okawa et al. 2008) and reports showing that Pdgfrα
expression was induced by NODAL activation (Takenaga, Fukumoto et al. 2007) in ESC
cultures cultured established on collagen IV (Sakurai, Era et al. 2006), Pdgfrα up-regulation was
more responsive to exogenous BMP4 and Wnt3a than Activin A in my culture system (Fig.
2.7vi). Indeed, BMP4 has also been shown to induce Pdgfrα in ESCs (Sakurai, Inami et al.
2009; Tanaka, Jokubaitis et al. 2009) and such a reversed expression pattern has been observed
in monolayer cultures of differentiating hESCs (Lee, Peerani et al. 2009).
R1 ESCs have been shown to have poor chondrogenic differentiation capabilities in EB
studies (Kramer, Hegert et al. 2005). Although Activin A has been shown to be both an inducer
(Jiang, Yi et al. 1993) and an inhibitor (Chen, Yu et al. 1993) of chondrogenic differentiation in
limb bud mesodermal cells, I showed that Activin A induced chondrocyte formation in my R1
SF monolayer cultures with intense Alcian blue staining (Fig. 2.8A), robust COL2A1 network
formation (Fig. 2.8C), detection of SOX9 protein expression (Fig. 2.8D) and marked up-
regulation of Col2a1, Sox9 and Aggrecan expression (Fig. 2.9A-C). TGFβ3-supplemented
cultures achieved comparable chondrogenic differentiation as Activin A (Fig. 2.11Aiii-iv, Bii-C),
validating that TGFβ is required at the initial stages of chondrogenesis (Nakayama, Duryea et al.
2003; Kawaguchi, Mee et al. 2005; Diekman, Rowland et al. 2010). Although FGF8 was shown
to induce chondrogenesis (Abzhanov and Tabin 2004; Bobick, Thornhill et al. 2007; Yu and
Ornitz 2008), it could not be used alone in my culture system because of poor cell-matrix
attachment. This finding is consistent with the role of FGF8 in anchorage-independent cell
growth and survival through interaction with the adaptor protein called crk-like protein (Seo,
Suenaga et al. 2009). Neither TGFβ3 nor FGF8 compensated for the non-inductive effect of
BMP4 (10ng/ml) or further enhanced the progress of chondrocyte formation initiated by Activin
86
A when they were added from day 5-15 of differentiation (Fig. 2.12A-C), reinforcing the stage-
specific nature of TGFβ- and FGF-modulated chondrogenic induction.
I demonstrated that five-day supplementation of Activin A was sufficient to induce
chondrogenesis in 15-day monolayer cultures (Fig. 2.13A-C), suggesting that ESCs acquired the
capacity to become chondrocytes within the first five days of differentiation in Activin A-
supplemented cultures. BMP4 has been shown to induce ESC chondrogenic differentiation
(Kramer, Hegert et al. 2000), and developing chondrocytes appear to undergo a BMP-dependent
stage after initiation by TGFβ in vivo (Nakayama, Duryea et al. 2003). I achieved BMP4-
induced chondrogenic differentiation only when BMP4 concentration increased from 10ng/ml to
25ng/ml, with robust COL2A1 network formation (Fig. 2.17Ai-ii) and Sox9 up-regulation (Fig.
2.17B), indicating that BMP4 also acted as an early inducer. However, BMP4 failed to up-
regulate Aggrecan expression significantly when compared to Activin A-supplemented cultures.
Also, prolonged exposure to BMP4 (25ng/ml) led to marginally lower transcript levels of Col2a1
and Sox9 but slightly higher Runx2 expression compared to cultures with five-day BMP4
treatment (Fig. 2.17B). Activation of canonical Wnt signaling in nascent chondrocytes has been
shown to block downstream chondrocyte development (Akiyama, Lyons et al. 2004). Limb-bud
and ESCs studies have concluded that WNT3a is required during late-stage chondrocyte
maturation, hypertrophy and mineralization (Enomoto-Iwamoto, Kitagaki et al. 2002; Kitagaki,
Iwamoto et al. 2003; Tamamura, Otani et al. 2005; Davis and Zur Nieden 2008). Similarly, early
Wnt3a treatment of my monolayer differentiation system generated mesoderm progenitors but
did not promote chondrocyte formation. However, delayed Wnt3a supplementation of
differentiation cultures initialized by Activin A slightly enhanced Sox9, Aggrecan and Runx2
expression (Fig. 2.14B), while Wnt3a treatment of BMP4 (10ng/ml)-supplemented cultures
resulted in robust COL2A1 network formation, superior expression of Col2a1, Sox9, Col10a and
87
Runx2, and more intense Alcian blue staining than in BMP4 alone (Fig. 2.16A-C). Since Wnt3a
treatment of BMP4- and Activin A-supplemented cultures appeared to up-regulate early and late
chondrogenic markers, respectively, Wnt3a may have a compensatory pro-chondrogenic role in
BMP4-containing cultures, while it promoted chondrocyte maturation in cells already induced by
Activin A.
In short, I have established a differentiation protocol for the SF monolayer derivation of
ESC-derived chondrocytes. In my culture system, Activin A, BMP4 and TGFβ3 acted as early
inducers of chondrogenesis while Wnt3a exerted its pro-chondrogenic effect only after
mesoderm specification (Fig. 2.18).
Fig. 2.18 – Schematic of my SF monolayer chondrogenic differentiation strategy. Supplementation of ESC
cultures with BMP4 (10ng/ml), Activin A (30ng/ml) or Wnt3a (100ng/ml) on day 0 of differentiation (dark blue
thunderbolt) successfully induced mesoderm progenitors. Expression levels of mesoderm markers in SF monolayer
cultures were in agreement with the notion that BMP4 and Wnt3a induced more posterior populations of mesoderm
(PM) while Activin A induced anterior mesoderm subsets (AM). Prolonged supplementation with Activin A,
TGFβ3 (10ng/ml) or an increased concentration of BMP4 (25ng/ml) (yellow thunderbolts) could induce chondrogenic differentiation after 15 days of culture. However, chondrogenic induction was not compromised when
the duration of supplementation was shortened to five days (green arrows). Wnt3a was found to be a late inducer of
chondrogenesis, and TGFβ3 could replace Activin A on day 5 of differentiation to promote the formation of
articular cartilage (light blue thunderbolts).
88
2.5 Potential uses of 2D culture system in HTS/HCI applications
I developed my SF monolayer chondrogenic differentiation protocol such that it could be
adopted in the establishment of HTS assays to be used in examining the effect of candidate
molecules on chondrocyte formation from ESCs. By defining the seeding density based on
culture area, one could easily scale-down my culture system (e.g. from 24-well format to 96-well
format) to facilitate screening in a high-throughput manner. To set up the differentiation assay as
a baseline tool for a screen, one can induce chondrogenic differentiation by adding Activin A
(30ng/ml) at the onset of differentiation and withdraw it after five days of differentiation. To test
the efficacy of candidate molecules in a primary drug screen, they can be added at varying time
points and concentrations during the 15 days of culture to determine the optimal conditions for
chondrogenic induction. The order in which combinations of molecules will be added can also
be a test parameter. A suitable readout for HCI-based HTS will be fluorescence intensity based
on IF analyses of protein expression of chondrogenic markers such as COL2A1 and/or SOX9
(Fig. 2.19). To avoid the formation of terminally differentiated ESC-derived chondrocytes,
similar IF-based analyses can also be conducted for hypertrophic chondrocyte markers such as
COL10A and RUNX2.
To move towards the identification of an actual therapeutic that can be used clinically,
one has to translate the results obtained from a system that is based on embryonic development
into the context of an adult. Indeed, MSCs that participate in embryonic development (e.g. limb
formation) may not behave the same way as MSCs that play a role such as wound healing in the
adult body. Amputated limbs of neonatal mice could be partially regenerated when limb buds of
mouse embryos were grafted to the limb stump (Masaki and Ide 2007). However,
transplantation of adult MSCs into damaged cartilage tends to lead to the formation of scar
89
tissues such as fibrocartilage. In addition, it has been determined that the earliest transient
population of MSCs that arose during embryogenesis was actually derived from neuroepithelium
and neural crest, only to be replaced by MSCs derived from other sources such as the bone
marrow (Takashima, Era et al. 2007). As such, it is possible that the effect of exogenous
reagents on ESC-derived chondrocytes differ from that on adult chondrocytes; however, one can
test for the expression of marker genes that may play similar roles in both embryonic
chondrocytes and in adult articular chondrocytes. For example, one can establish a secondary
screen to identify agents that will induce the expression of GDF5 and ERG. GDF5 has been
shown to promote cell adhesion during mesenchymal condensation and proliferation of
chondrocytes in growth plate cartilage (Francis-West, Abdelfattah et al. 1999; Buxton, Edwards
et al. 2001). GDF5 is also found to be a potent inducer of Erg, which is a transcription factor
expressed in articular chondrocytes of the developing synovial joint. In fact, the over-expression
of ERG was found to maintain chondrocytes in an immature, articular-like state both in vivo
(Iwamoto, Tamamura et al. 2007) and in vitro (Iwamoto, Koyama et al. 2005). Therefore, the
use of therapeutics to induce GDF5 and ERG expression in damaged articular cartilage may
contribute to its regeneration and restoration of joint function.
As shown by the results presented in this chapter, the most robust readout from my
chondrocyte differentiation assay was COL2A1 expression. However, the quantification of
COL2A1 protein network using screening platforms such as the Cellomics ArrayScan® would
not be as straight-forward as the quantification of cells that express COL2A1. The need to
identify COL2A1+ cells prompted me to design a reporter construct that could be incorporated
into ESCs to generate a transgenic cell line to be used in my differentiation assay. The assembly
of this reporter system is discussed in Chapter 3.
90
Fig. 2.19 – Schematic diagram depicting the set up of a molecule screen by establishing the ESC-derived
chondrocyte cultures using my SF monolayer differentiation system.
91
Chapter 3 Generation of a bi-colour fluorescent reporter mESC line for
potential chondrocyte-specific fate mapping and drug screen applications
3.1 Overview
The derivation of media formulations and inductive culture environments that drive
controlled differentiation of ESCs is often coupled with genetic manipulation of the cells to
allow for the identification and isolation of desired cell types. As mentioned in Section 1.6,
faithful expression of reporters, such as fluorescent proteins, β-galactosidase or luciferase, under
the transcriptional control of genes of interest in these transgenic ESCs facilitates the
visualization of gene expression patterns; therefore, they are used regularly in chimeric studies
for lineage mapping and mutagenesis. For example, the BRACHYURY:GFP ESC line (Fehling,
Lacaud et al. 2003) is widely used in in vitro and in vivo differentiation studies involving
mesodermal derivatives. This cell line has also been modified with the addition of human CD4
knocked into the Foxa2 locus, and it was used in the development of an in vitro primitive streak
model where distinct populations of cells were identified based on the expression levels of GFP
and CD4 (Gadue, Huber et al. 2006).
To have further control on the timing of reporter markers expression, conditional
mutagenesis using SSR strategy such as the Cre/loxP system is frequently employed in fate
mapping studies. The targeting of reporters to specific loci of interest is an efficient method of
tracking gene expression in vitro and in vivo; however, the abolishment of one of the alleles, as is
the case for the BRACHYURY:GFP line, causes haploinsufficiency and hence does not
92
recapitulate wild-type expression of the genes of interest and can affect ESC cell fate. As a
result, it may be beneficial to devise another reporter system that does not disrupt endogenous
protein expression.
As outlined in Section 1.7, I designed and created components of a versatile Cre/loxP
recombinase system-based transgenic cell line that can be used for any maker gene of interest
with minimal modifications required. A Rosa26 locus-targeted, Cre-inducible R1 ESC line
(EST2B) that changes from being puromycin resistant to expressing DsRedT3 and conferring
blasticidin resistance upon Cre excision was previously created in our laboratory (Fig. 3.1)
(Handy 2005). The Cre-inducible construct was targeted to the Rosa26 locus because gene
trapping studies have confirmed that reporter gene targeted to the Rosa26 locus exhibited
ubiquitious expression during embryonic development (Friedrich and Soriano 1991;
Zambrowicz, Imamoto et al. 1997). Therefore, the EST2B cells would be ideal for fate mapping
studies because upon transgene activation, all the progeny of the trasgene-expressing cells would
express the same reporter, allowing them to be tracked both in vitro and in vivo. The MultiSite
Gateway® cloning platform (Invitrogen) was used to assemble the Cre-expressing construct T2A
under the transcriptional control of a Col2a1 promoter. The incorporation of T2A into EST2B
cells results in the creation of the EST2 ESC line that could be used to monitor the progression of
ESC chondrogenic differentiation. While it was unnecessary for me to design such an elaborate
two-step system for the identification of COL2A1+ cells in my differentiation cultures, the
derivation of the EST2 line could be used to test the feasibility of using such a tissue-specific,
Cre-inducible construct system to fate map any transient population of choice. In the case of
chondrogenic differentiation, the emergence of Venus+DsRed T3
+ cells would signify the
expression of COL2A1. As the cells continue to differentiate, the loss of the double-positive
population and the generation of Venus- DsRed T3
+ (i.e., YFP
-RFP
+) cells would suggest that the
93
cells may have stopped the production of COL2A1 and have begun the terminal differentiation
program whereby they would undergo hypertrophy and eventually become calcified.
Fig. 3.1 – Schematic diagram of the targeted insertion of Cre-inducible vector T2B into the Rosa26 locus. T2B
consists of a floxed PuroR and polyadenylation (pA) signal, followed by the DsRedT3 RFP and BlastR. Upon Cre
excision, the ubiquitous CAGGS promoter (a chicken β-actin promoter coupled with a cytomegalovirus enhancer element (Niwa, Yamamura et al. 1991)) drives the constitutive expression of DsRedT3 and BlastR, allowing the
treated transgenic ESCs to remain RFP+ and blasticidin resistant.
3.2 Materials and Methods
3.2.1 Differentiation of EST2B cells
EST2B cells were subcultured twice on gelatinized tissue culture plastic to deplete MEFs.
At the onset of differentiation, cells were trypsinized and seeded onto 6-well low cluster plates
(Costar) for EB induction. EBs were cultured in ES media without LIF (ES differentiation
medium) for 10 days, with medium addition (50% of total volume) and passaging (1:2 ratio)
taking place on alternating days. EST2B cells were also differentiated into cells of the three
germ layers to verify their pluripotency. For seven-day ectodermal differentiation cultures,
104cells/cm
2 were seeded onto gelatinized tissue culture plastic and induced to form neural cells
in N2B27 medium containing retinoic acid (Ying, Stavridis et al. 2003). For cardiomyocyte
differentiation, day 3 EBs were plated onto gelatin-coated tissue culture plastic and cultured for
94
another six to nine days in ES differentiation medium. Endoderm differentiation was achieved
by initiating EB formation with 4x104cells/cm
2 in SF medium developed in (Gouon-Evans,
Boussemart et al. 2006), consisting of I-MDM (75%), Ham’s F12 medium (25%), N2 and B27
Fig. 3.8 – Restriction digest analyses of the expression clones suggested the successful generation of the T2A construct to be integrated into EST2B cells. (A) Linearization of plasmid clones using the restriction enzyme
XhoI generated a single fragment >10kb, which was slightly bigger than the expected size of the expression vector.
(B) Restriction digests carried out with EcoRV+NotI, PmeI+SpeI and XhoI+NotI generated DNA fragments of
sizes: 1633bp, 1526bp and 1171bp, respectively, which corroborated with the theoretical fragment sizes (see Fig.
3.7).
109
NLS-Cre-SV40pA attR3 attL3
110
Fig. 3.9 – Sequencing results confirmed the proper integration of cloning fragments into destination vector
using the Multisite Gateway® Pro 4.0 system.
To determine the functional efficacy of the expression clone, it was transiently
transfected into HEK 293T cells in the presence of human Sox9 cDNA, which activated the
Col2a1 promoter via the five-time repeat of the 48bp SOX9 enhancer element. After 48hrs.,
Venus expression was visible in cultures transfected with the expression clones (Fig. 3.10Aii,
Bii) and the control plasmid (Fig. 3.10Cii). Although the distribution of Venus expression was
similar between the two cultures, cells transfected with the control plasmid had more intense
Venus expression. As expected, cells lacking the Sox9 cDNA did not express Venus (Fig.
3.10D-E); however, there were a few positive cells in the culture containing the control plasmid
only (Fig. 3.10Fii).
Aside from Venus YFP, expression of Cre recombinase was also detected 48hrs. after
HEK 293T cells were transiently co-transfected with the T2A expression clone and Sox9 cDNA.
Cells that co-expressed both Venus YFP and Cre recombinase were identified (Fig. 3.11).
Conversely, cells transfected with the control plasmid Col2a1-eYFP and Sox9 cDNA did not
exhibit any Cre expression, while cells transfected with pCAGGS-NLS-Cre showed a lack of
Venus YFP expression.
111
A D
B E
C F
Fig. 3.10 – Bright field (i) and fluorescence (ii) images (100x) documenting transgene expression of plasmid
T2A in live HEK 293T cells 48hrs. after transient co-transfection with human Sox9 cDNA via lipofection. Plasmid clone 1 (A, D) and clone 2 (B, E) behaved similarly in terms of the distribution and level of Venus YFP
expression in the presence (A, B) and absence (D, E) of Sox9 cDNA. (C) Cells transfected with the Col2a1-eYFP
control plasmid and Sox9 cDNA expressed eYFP at a significantly higher intensity than those containing T2A;
however, a small fraction of the cells containing control plasmid alone also showed eYFP expression (D).
Fluorescence images were taken using the same amount of exposure time.
Venus YFP
eYFP
112
Fig. 3.11 – IF analysis of the expression of Cre recombinase in HEK 293T cells transiently transfected with construct T2A and Sox9 cDNA. Images (200x) displayed the co-expression of Venus YFP and Cre recombinase
in cells transfected with both construct T2A and Sox9 cDNA (see insets), while cells transfected with the control
plasmids only expressed Venus (in the case of Col2a1-eYFP+Sox9 cDNA) or Cre recombinase (in the case of
pCAGGS-NLS-Cre).
3.3.3 Validation of the transgenic EST2 line
Linearized expression plasmid was electroporated into EST2B cells and ESC clones
resistant to both G418 and puromycin were isolated after antibiotic selection. To test the
functionality of the resulting EST2 cells, they were first transiently transfected with Sox9 cDNA
to determine if transgene expression could be detected. Only faint Venus expression was seen in
a few cells. To further characterize the cells, they were differentiated into chondrocytes in
Venus YFP
Cre
Hoechst
113
monolayer cultures, as outlined in Chapter 2. However, neither Venus or DSRedT3 expression
was detected when cultures were examined at multiple time points, leading to my speculation
that there were insufficient copy numbers of the plasmid present in the EST2 cells. The cells
were re-electroporated in an attempt to increase the copy numbers; however, a four-fold increase
in G418 concentration failed to isolate new clones after an extended period of selection. As
such, the lines would need to be re-established.
The re-derivation of the EST2 transgenic cell line was achieved via electroporation using
the Neon™ Transfection System from Invitrogen. We have previously established optimized
transfection conditions for hESCs, mESCs and human fibroblasts using this transfection system.
Expression plasmid was linearized with PmeI, and it was purified as well as concentrated via
ethanol precipitation. EST2B cells were trypsinized, washed with PBS and resuspended in
Resuspension Buffer R at a concentration of 5x106cells/ml. Linearized T2A plasmid was mixed
with cells at a concentration of 1µg of plasmid per 100µl of cells. Electroporation was
performed in 100µl reactions under the conditions outlined in Table 3.1 and each reaction was
plated onto gelatinized 6-well tissue culture plate (i.e., one reaction condition per well). Clones
were isolated via antibiotic selection with G418 and puromycin.
It was discovered that transfection carried out in three pulses at 1400V and a pulse width
of 10ms generated the most colonies after antibiotic selection. Therefore, electroporation was
performed on a larger scale at this reaction condition where every 106cells were transfected with
1.5µg of plasmid. A total of 96 clones were isolated after antibiotic selection and the clones
were cryopreserved. Functionality of the cells will be validated.
114
Well Pulse Voltage (V) Pulse Width (ms) Pulse Number
1 1400 10 3
2 1200 20 2
3 1500 20 1
4 1500 20 2
5 1000 20 1
6 1000 20 2
Table 3.1 – Test conditions for the transfection of EST2B cells with T2A expression plasmid using Neon™
Transfection System.
3.4 Current work
To examine the efficacy of the EST2 cell line, cells will be transfected with Sox9 cDNA
and will undergo chondrogenic differentiation to activate the 5x48 Col2a1 promoter in order to
induce transgene expression. Temporal Venus YFP pattern will be compared with endogenous
COL2A1 expression to determine the faithfulness of transgene expression. The co-expression of
DsRedT3 RFP and Venus YFP will signify the activation of the Col2a1 promoter, which leads to
the expression of Cre recombinase, and the subsequent constitutive DsRedT3 RFP expression
can be used to track the cells that transiently expressed COL2A1.
3.5 Future work
Since construct T2A was assembled using Multisite Gateway® technology where the
promoter, fluorescent protein, Cre recombinase and selectable marker were separated into
different entry vectors, the modular nature of this construct system provides great versatility as
different elements of the vector can be substituted to be used in various fate mapping and HCI
115
applications. For example, the 5x48 Col2a1 promoter can be replaced with other chondrogenic
markers such as the Sox9 promoter, and parallel screens for novel therapeutics can be conducted
to discover agents that promote the expression of all the markers or selectively activate some of
the promoters. Candidate agents that fail to maintain the expression of any early chondrogenic
marker of choice, as indicated by the sole expression of DsRedT3 due to the loss of Venus YFP,
will be easily identified.
Upon verifying the function of the two-step reporter system, the same experimental steps
can be applied to generate various transgenic cell lines using promoters specific to other cell
types of interest by re-assembling construct T2A using the Multisite Gateway® cloning method.
In addition to using the resulting reporter cell lines to identify the formation of ESC-derived cell
types of interest in vitro, they can be used to generate chimeric mice to facilitate in vivo fate
mapping, which allows one to visualize and track the emergence of specific transient cell types
and the localization of their progeny through development. For example, it will be interesting to
track cells that express the paraxial mesoderm-specific markers such as Mesp2 or the neural crest
marker Paired box 3 (Pax3), both of which can give rise to chondrocytes as well as other cell
types such as bone and muscles during development. In addition to tissue-specific markers,
signaling pathway-specific genes such as Lef/Tcf, which are downstream targets of β-CATENIN,
and members of the Notch signaling pathway are also interesting candidates because of the
oscillatory expression patterns of Wnt and Notch signaling during somitogenesis. In addition,
tracking the expression of the BMP inhibitor Noggin will also be very informative due to the
temporal and spatial specificity of Noggin expression during embryonic development. It will be
interesting to visualize the development of somites-derived tissues in vitro and in vivo as well as
observe the effects caused by perturbations of the expression of these genes.
116
Chapter 4 Discussion and conclusion
In Chapters 2 and 3, I have demonstrated my progress towards the establishment of a
screening platform for novel therapeutics that promote endogenous cartilage repair. It involves
the development of a growth-factor mediated, SF monolayer differentiation system for ESC-
derived chondrocytes. I have also assembled a Col2a1 promoter-driven reporter construct (T2A)
that can be used to identify and quantify chondrocytes generated using the monolayer
differentiation system. Although it was not necessary for my project, I have modified T2A to
include the gene for encoding Cre recombinase such that it can be incorporated into an existing
Cre-inducible reporter ESC line (EST2B). This two-step reporter system can identify cells that
express COL2A1; in addition, due to the ubiquitous expression of a DsRed T3 RFP upon Cre
excision, all the progeny of COL2A1+ cells will also be identified. The expression pattern of
Venus YFP and DsRed T3 will allow one to track the progression of chondrocyte differentiation
both in vitro and in vivo.
In Chapter 2, I have shown that I was able to generate ESC-derived chondrocytes in
monolayer cultures in a defined chemical condition. I used IF, qPCR and Alcian blue staining to
confirm the presence of chondrogenic cells in my cultures. Real-time qPCR analyses provided
transcript level information to verify the faithful up-regulation of chondrogenic marker genes. IF
analyses confirmed marker gene expression at the protein level, while positive Alcian blue
staining indicated proteoglycan production. Transcript and protein expression provided
sufficient evidence of chondrogenic differentiation using my ESC SF monolayer differentiation
system. However, to obtain additional phenotypic and functional data that can specify the type
117
of chondrocytes being produced, one can measure GAG production, which increases as cells
differentiate into chondrocytes as well as perform additional histochemistry analyses with stains
such as Masson’s trichrome stain for collagen detection and Toluidine blue for cartilage
detection. Functional assays can be performed on the ESC-derived chondrogenic cultures if one
wants to be truly rigorous with the characterization process. For example, ESC-derived
chondrocytes can be implanted into SCID mice and the tissue at the point of injection will be
retrieved at a later time point to assess the proliferation and maturation of the injected cells.
Alternatively, the cells can be incorporated into scaffolds or engineered constructs to be
implanted into an injury site to assess the extent of cartilage repair. However, for the purpose of
generating a cell source for drug/small molecule screens, the establishment of a cartilage repair
model using the ESC-derived chondrocytes may be too exhaustive and time consuming.
Heng et al. suggested that a defined SF culture milieu for directed chondrogenic
differentiation should include the incorporation of cytokines/growth factors, chemicals and ECM
in conjunction with biophysical parameters such as oxygen tension, temperature and cell density
(which mediates the amount of cell-cell contact) (Heng, Cao et al. 2004). In addition to ECMs,
media formulation, seeding density and exogenous growth factors, one can examine the
enhancing effects of chemical additives on the chondrogenic differentiation cultures. Although I
have briefly examined the effect of dexamethosone and ascorbic acid addition on my cultures
and did not observe any additional benefit in terms of chondrogenic induction (data not shown),
one can test the efficacy of these factors and others such as thyroid hormones more rigorously as
they were shown to promote chondrogenesis in other culture systems. Dexamethosone is a
glucocorticoid that has been shown to induce chondrogenic differentiation in human MSCs
(Johnstone, Hering et al. 1998; Mackay, Beck et al. 1998). Ascorbic acid has been shown to
stimulate cartilage matrix production (Farquharson, Berry et al. 1998), while thyroid hormones
118
are steroid derivatives of cholesterol metabolism that have been shown to play a role in
chondrogenic differentiation (Wakita, Izumi et al. 1998). In terms of oxygen tension, hypoxia
(1-2% O2) has been demonstrated to increase the chondrogenic potential of cells differentiated
from MSCs (Robins, Akeno et al. 2005), ESCs (Koay and Athanasiou 2008) and primary
articular chondrocytes (Egli, Bastian et al. 2008) by promoting the increased expression of
chondrogenic marker genes and the production of GAG. Specifically, it was discovered that the
inductive effects of low oxygen tension was more potent in cells undergoing early differentiation
or expansion, while hypoxic conditions had minimal effect on the cells during late-stage
differentiation compared to normoxic conditions (Egli, Bastian et al. 2008; Koay and Athanasiou
2008).
Despite the successful derivation of ESC-derived chondrocytes, I discovered that
differentiating ESCs cultured in SF media had inferior cell spreading, adhesive properties and
possibly a slower rate of proliferation compared to those established in serum media, consistent
with reported studies on SF ESC culture (Ying, Nichols et al. 2003; Chaudhry, Vitalis et al.
2008). To ensure even cell spreading and strong cell-ECM adhesion, the undifferentiated cells
can be plated in serum-containing medium for a few hours to establish cell-ECM adhesion before
changing to SF differentiation medium. However, this strategy may cause delays in the up-
regulation of differentiation marker genes (especially early differentiation markers) due to
residue serum effect present in the culture microenvironment.
It has also been shown that mesoderm induction in serum monolayer cultures was inferior
to that in EB cultures (Nishikawa, Nishikawa et al. 1998). There is an increasing preference for
3D cultures because they better recapitulate the cell-cell, cell-ECM and paracrine interactions
that exist in vivo, while cells established in 2D cultures are subjected to unnatural geometrical
119
constraints and therefore lack many of the mechanical and biochemical cues that define cellular
behaviour (Birgersdotter, Sandberg et al. 2005; Sun, Jackson et al. 2006; Maltman and
Przyborski 2010). Interestingly, these advantageous qualities of 3D cultures over monolayer
cultures are also the cause of much criticism against 3D cultures. The microenvironment within
a 3D structure such as an EB is highly variable due to the different kinds of cellular interactions
present within an EB, leading to significant heterogeneity both within any given EB but also
between different EBs during differentiation (Metallo, Mohr et al. 2007; McDevitt, Carpenedo et
al. 2008). In addition, the transport of soluble morphogens and other crucial molecules into EBs
is often hindered by diffusion limitations, which are dependent on factors such as aggregate size
and ECM content (Carpenedo, Seaman et al. 2010). Various strategies have been devised to
attain better control of ESC fate in both undifferentiated colonies and EBs. Studies in hESCs
suggested that the EB size itself and the size of undifferentiated hESC colonies used to generate
these EBs exerted certain biases towards the tendency to generate different germ layers, and
together these parameters influenced the efficiency in the formation of specific cell types such as
cardiomyocytes (Bauwens, Peerani et al. 2008; Niebruegge, Bauwens et al. 2009). In addition,
hESC colony and EB sizes can be precisely and reproducibly controlled via the integration of
technologies such as microcontact printing and the use of microwells (e.g. AggreWell™ from
STEMCELL Technologies), respectively. (Ungrin, Joshi et al. 2008; Lee, Peerani et al. 2009;
Sakai, Yoshiura et al. 2011). Aside from controlling the size of colonies and EBs, strategies
have also been devised to improve the delivery of soluble morphogens into EBs by aggregating
ESCs with morphogen-containing biodegradable microspheres in rotary suspension cultures
(Carpenedo, Bratt-Leal et al. 2009; Carpenedo, Seaman et al. 2010).
In spite of the development of novel strategies to enrich for the cell types of interest when
differentiating ESCs as EBs, it remains impossible to obtain 100% pure cell populations from
120
these 3D cultures. Oftentimes transgenic cell lines need to be created to facilitate the isolation of
cell types of interest via FACS or antibiotic selection, both of which are labour intensive
processes that significantly compromise the health of the cells. In addition, the removal of
specific cell types from their 3D culture environment poses similar concerns as those listed for
2D cultures in that the isolated cells are no longer exposed to the appropriate cellular signals.
Furthermore, it is difficult to observe the cellular behaviour of subsets of cells within an EB
using conventional imaging techniques without the aid of confocal microscopy, which can be
very time-consuming and costly especially when put in the context of the development of HTS
and HCI platforms.
Conversely, although it is argued that 2D cultures cannot provide the full spectra of
cellular signals to recapitulate the proper physiological environment for the generation of ESC-
derived cell types, it remains the easiest method of establishing platforms for first-phase drug
and small molecule screens due to the limited amount of manual manipulation required to
establish these cultures, the low cost associated with culture setup and the speediness at which
testing can be done (Giese, Kaufmann et al. 2002). Although 2D ESC differentiation cultures
may not generate many of the differentiated cell types efficiently (Nishikawa, Nishikawa et al.
1998), it may be advantageous for directed differentiation because one can obtain a higher
percentage of the cell type of interest (i.e. chondrocytes). In addition, it provides more flexibility
in terms of culture manipulation and the ability to do so with ease (Heo, Lee et al. 2005), as
demonstrated by my ECM/media/seeding density screen using HCI as described in Chapter 2.
Another advantage of using 2D cultures in HTS is the ability to identify changes in cellular
behaviour in cell types of interest with the option of not isolating the cells via FACS or selection.
Although it is ideal to perform cellular assays with pure cell populations, it may not be
detrimental to have supportive cell types present in the culture so long as they do not interfere
121
with the results of the assays. Since the cultures are in 2D, cell types of interest can be easily
identified via immunostaining or with any reporter genes as in the case of transgenic cell lines,
and they can be easily imaged and quantified in HTS applications.
One of the important questions that remain to be answered at the current stage of my
screening platform is the quantification of the yield of chondrocytes from my 2D cultures, that is,
the percentage of ESCs that actually differentiate into chondrocytes versus other cell types. I
agree that presently, it is difficult to quantify the percentage of chondrogenic cells in my culture
system based on COL2A1 protein expression using HCI strategies because IF stains for the
collagen networks and not individual cells. However, with the successful derivation of the EST2
line, one can count the number of COL2A1+ cells based on live-cell imaging of Venus YFP and
DsRedT3 RFP co-expression, which can be corroborated with IF analyses of SOX9 or other
chondrogenic markers. Furthermore, one can identify the chondrogenic cells that have
undergone terminal differentiation based on the down-regulation of Venus YFP expression and
the maintenance of DsRedT3 RFP. As mentioned in Chapter 3, because T2A construct was
assembled in a modular manner using Gateway® technology, one can generate different
transgenic lines using various promoters of chondrogenic marker genes, and HCI results
accumulated from data generated from all the different transgenic lines can be compared.
Another critical consideration is the ability to adapt this screening platform to hESC
studies. Unfortunately, major parameter re-testing will probably be involved because the
culturing technique for hESCs varies substantially from that for mESCs. The use of collagen IV
as a potential ECM for hESC adhesion has been examined (Draper, Moore et al. 2004), but
hESCs differentiating on collagen IV appeared to form epithelial-like cells (Ahmad, Stewart et
al. 2007). On the other hand, Matrigel™, a heterogeneous mixture of ECMs, is routinely used in
122
MEF-free culture of hESCs. Human ESCs have been reported to undergo osteogenic
differentiation, along with the formation of chondrocyte-like cells, in monolayers on gelatin-
coated tissue culture plastic. However, monolayer differentiation cultures were not established
from single cell suspensions; hESCs were seeded as clumps that were partially dissociated via
collagenase digestion and mechanical scraping (Karner, Unger et al. 2007). Despite the fact that
undifferentiated hESCs can be dissociated and replated as single cells (Bauwens, Peerani et al.
2008), the efficiency of establishing hESC monolayer mesodermal differentiation cultures from
single cells has not been thoroughly validated. However, with the advancements in microcontact
printing and patterning technologies it is believed that single cell-derived long-term hESC
monolayer differentiation cultures can soon be routinely established. In terms of the generation
of an equivalent transgenic line as EST2 cells using hESCs, one can use a similar approach as
that used to generate EST2B cells to target the T2B construct into the human Rosa26 locus
(Irion, Luche et al. 2007) or the R4 targeting locus (Lieu, Machleidt et al. 2009).
Apart from the differences in culture techniques, one has to be mindful of the influences
of exogenous factors on hESC cell fate decisions and how they may differ from those on mESC
differentiation. Combinations of exogenous BMP2, BMP7, TGFβ1, TGFβ3 and even Insulin-
like growth factor 1 (IGF1) have been reported to promote hESC chondrogenic differentiation as
EBs, micromasses or pellets (Koay, Hoben et al. 2007; Toh, Yang et al. 2007; Nakagawa, Lee et
al. 2009; Gong, Ferrari et al. 2010). However, TGFβ1 has also been shown to inhibit
chondrogenic differentiation, albeit in cultures maintained in a chondrogenic medium containing
10% FBS as opposed to the usual concentration of 1% FBS or no serum (Yang, Sui et al. 2009).
Interestingly, many of these studies used BMP concentrations of 100-300ng/ml, which were
about 5-10 folds more than what would be used in mESC studies. Activin A and Wnt3a have
similar effect on mESCs and hESCs in that they promote the formation primitive streak-like cells
123
(Lee, Peerani et al. 2009; Evseenko, Zhu et al. 2010) as well as definitive endoderm at higher
concentrations (D'Amour, Agulnick et al. 2005). It was also found that the addition of Activin A
helped maintain the undifferentiated state of hESCs (Beattie, Lopez et al. 2005; James, Levine et
al. 2005; Xiao, Yuan et al. 2006), and as mentioned in Section 1.5.2.1.4, it has been shown that
Wnt3a stimulated the proliferation of undifferentiated hESCs (Dravid, Ye et al. 2005). In terms
of FGFs, FGF2 is one of the requisite components in the maintenance of undifferentiated hESC
cultures under SF conditions (Amit, Carpenter et al. 2000; Xu, Rosler et al. 2005). It has been
suggested that FGF2 and NOGGIN work synergistically to maintain hESC pluripotency in the
absence of feeder layers (Wang, Zhang et al. 2005; Xu, Peck et al. 2005). Therefore, although
some of the exogenous growth factors have overlapping functions in both mESCs and hESCs,
others such as Activin A also exert a divergent influence on hESC cell fate. As such, careful
testing of the various growth factors at different concentrations is critical when adapting my
monolayer culture system for hESC differentiation purposes.
In conclusion, I have developed a one-step strategy for generating monolayers of ESC-
derived chondrogenic cells on collagen IV in a chemically defined condition. My system
recapitulated the published expression patterns of a plethora of mesoderm marker genes and
confirmed the stage-specific nature of TGFβ-, BMP- and Wnt-modulated chondrogenesis. The
simplicity of my system facilitates the establishment of test cultures for HCI/HTS with minimal
manipulation. The 2D nature of my system also provides a platform that permits easy
visualization of changes in chondrogenic markers or reporter expression in knock-down/over-
expression studies and in the identification of novel chondrogenic modulators. By combining the
establishment of a monolayer differentiation protocol for ESC-derived chondrocytes with the
generation of a transgenic ESC line under the transcriptional control of the chondrogenic marker
Col2a1, this system has the potential to generate multiple sets of quantitative data, from the
124
expression levels of key chondrogenic marker genes to the percent formation of chondrocytes in
cultures supplemented with different molecules. Using the Gateway® cloning system, one can
re-assemble the Cre-expressing vector T2A with ease using different promoters to create parallel
screens to test the efficacy of candidate molecules. One can also target the Cre-inducible T2B
construct into the Rosa26 locus in hESCs. With optimization, my system can be adapted to carry
out similar screens in hESCs to determine if the candidate molecules identified in the mESC
screens have similar chondrogenic inductive effects in hESCs.
125
References
Aberle, H., A. Bauer, et al. (1997). "beta-catenin is a target for the ubiquitin-proteasome
pathway." Embo J 16(13): 3797-804.
Abzhanov, A. and C. J. Tabin (2004). "Shh and Fgf8 act synergistically to drive cartilage
outgrowth during cranial development." Dev Biol 273(1): 134-48.
Ahmad, S., R. Stewart, et al. (2007). "Differentiation of human embryonic stem cells into corneal
epithelial-like cells by in vitro replication of the corneal epithelial stem cell niche." Stem
Cells 25(5): 1145-55.
Ahmed, N., W. L. Stanford, et al. (2007). "Mesenchymal stem and progenitor cells for cartilage
repair." Skeletal Radiol 36(10): 909-12.
Akiyama, H., M. C. Chaboissier, et al. (2002). "The transcription factor Sox9 has essential roles
in successive steps of the chondrocyte differentiation pathway and is required for
expression of Sox5 and Sox6." Genes Dev 16: 2813-2828.
Akiyama, H., J. P. Lyons, et al. (2004). "Interactions between Sox9 and beta-catenin control
chondrocyte differentiation." Genes Dev 18(9): 1072-87.
Akizuki, S., V. C. Mow, et al. (1986). "Tensile properties of human knee joint cartilage: I.
Influence of ionic conditions, weight bearing, and fibrillation on the tensile modulus." J
Orthop Res 4(4): 379-92.
Alper, J. (2009). "Geron gets green light for human trial of ES cell-derived product." Nat
Biotechnol 27(3): 213-4.
Amit, M., M. K. Carpenter, et al. (2000). "Clonally derived human embryonic stem cell lines
maintain pluripotency and proliferative potential for prolonged periods of culture." Dev
Biol 227(2): 271-8.
Anraku, Y., H. Mizuta, et al. (2008). "The chondrogenic repair response of undifferentiated
mesenchymal cells in rat full-thickness articular cartilage defects." Osteoarthritis
Cartilage 16(8): 961-4.
Arnold, S. J. and E. J. Robertson (2009). "Making a commitment: cell lineage allocation and axis
patterning in the early mouse embryo." Nat Rev Mol Cell Biol 10(2): 91-103.
Arnold, S. J., J. Stappert, et al. (2000). "Brachyury is a target gene of the Wnt/beta-catenin
signaling pathway." Mech Dev 91(1-2): 249-58.
Artelt, P., R. Grannemann, et al. (1991). "The prokaryotic neomycin-resistance-encoding gene
acts as a transcriptional silencer in eukaryotic cells." Gene 99: 249-254.
126
Austin, T. W., S. Salimi, et al. (2000). "Long-term multilineage expression in peripheral blood
from a Moloney murine leukemia virus vector after serial transplantation of transduced
bone marrow cells." Blood 95(3): 829-36.
Avilion, A. A., S. K. Nicolis, et al. (2003). "Multipotent cell lineages in early mouse
development depend on SOX2 function." Genes Dev 17(1): 126-40.
Bafico, A., G. Liu, et al. (2001). "Novel mechanism of Wnt signalling inhibition mediated by
Dickkopf-1 interaction with LRP6/Arrow." Nat Cell Biol 3(7): 683-6.
Baird, G. S., D. A. Zacharias, et al. (2000). "Biochemistry, mutagenesis, and oligomerization of
DsRed, a red fluorescent protein from coral." Proc Natl Acad Sci U S A 97(22): 11984-9.
Baroffio, A., M. Hamann, et al. (1996). "Identification of self-renewing myoblasts in the progeny
of single human muscle satellite cells." Differentiation 60(1): 47-57.
Barrallo-Gimeno, A. and M. A. Nieto (2005). "The Snail genes as inducers of cell movement and
survival: implications in development and cancer." Development 132(14): 3151-61.
Bartek, J. and J. Lukas (2001). "Pathways governing G1/S transition and their response to DNA
damage." FEBS Lett 490(3): 117-22.
Battler, A., J. Leor, et al. (2006). Cartilage. Stem Cell and Gene-Based Therapy, Springer
London: 179-193.
Bauwens, C. L., R. Peerani, et al. (2008). "Control of human embryonic stem cell colony and
Yamanaka, S. (2009). "Elite and stochastic models for induced pluripotent stem cell generation."
Nature 460(7251): 49-52.
Yamashita, H., P. ten Dijke, et al. (1995). "Osteogenic protein-1 binds to activin type II receptors
and induces certain activin-like effects." J Cell Biol 130(1): 217-26.
Yan, H. and C. Yu (2007). "Repair of full-thickness cartilage defects with cells of different
origin in a rabbit model." Arthroscopy 23(2): 178-87.
Yang, J., S. A. Mani, et al. (2004). "Twist, a master regulator of morphogenesis, plays an
essential role in tumor metastasis." Cell 117(7): 927-39.
Yang, Z., L. Sui, et al. (2009). "Stage-dependent effect of TGF-beta1 on chondrogenic
differentiation of human embryonic stem cells." Stem Cells Dev 18(6): 929-40.
Yeom, Y. I., G. Fuhrmann, et al. (1996). "Germline regulatory element of Oct-4 specific for the
totipotent cycle of embryonal cells." Development 122(3): 881-94.
Ying, Q. L., J. Nichols, et al. (2003). "BMP induction of Id proteins suppresses differentiation
and sustains embryonic stem cell self-renewal in collaboration with STAT3." Cell
115(3): 281-92.
Ying, Q. L., M. Stavridis, et al. (2003). "Conversion of embryonic stem cells into
neuroectodermal precursors in adherent monoculture." Nat Biotechnol 21(2): 183-186.
Yu, J., K. Hu, et al. (2009). "Human induced pluripotent stem cells free of vector and transgene
sequences." Science 324(5928): 797-801.
154
Yu, J., M. A. Vodyanik, et al. (2007). "Induced pluripotent stem cell lines derived from human
somatic cells." Science 318(5858): 1917-20.
Yu, K. and D. M. Ornitz (2008). "FGF signaling regulates mesenchymal differentiation and
skeletal patterning along the limb bud proximodistal axis." Development 135(3): 483-91.
Zambrowicz, B. P., A. Imamoto, et al. (1997). "Disruption of overlapping transcripts in the
ROSA beta geo 26 gene trap strain leads to widespread expression of beta-galactosidase
in mouse embryos and hematopoietic cells." Proc Natl Acad Sci U S A 94(8): 3789-94.
Zelzer, E., D. J. Glotzer, et al. (2001). "Tissue specific regulation of VEGF expression during
bone development requires Cbfa1/Runx2." Mech Dev 106(1-2): 97-106.
Zhang, G., V. Gurtu, et al. (1996). "An enhanced green fluorescent protein allows sensitive
detection of gene transfer in mammalian cells." Biochem Biophys Res Commun 227(3):
707-11.
Zhang, J., R. E. Campbell, et al. (2002). "Creating new fluorescent probes for cell biology." Nat
Rev Mol Cell Biol 3(12): 906-18.
Zhou, H., S. Wu, et al. (2009). "Generation of induced pluripotent stem cells using recombinant
proteins." Cell Stem Cell 4(5): 381-4.
Zuk, P. A., M. Zhu, et al. (2001). "Multilineage cells from human adipose tissue: implications
for cell-based therapies." Tissue Eng 7: 211-228.
155
Appendix A
Supplementary Data for Chapter 2
Gene Sequences
Aggrecan Forward:
Reverse:
TGGCTTCTGGAGACAGGACT
TTCTGCTGTCTGGGTCTCCT
αMHC Forward:
Reverse:
GACGCCCAGATGGCTGACTT
GTCACCGTCTTTCCGTTTTC
βIII-tubulin Forward:
Reverse:
TAGACCCCAGCGGCAACTAT
GTTCCAGGTTCCAAGTCCACC
Bmp4 Forward:
Reverse:
AGCCAACACTGTGAGGAGTTTCCA
TGCTGCTGAGGTTGAAGAGGAAACGA
Brachyury Forward:
Reverse:
TCCTCCATGTGCTGAGACTTGT
TGCCACTTTGAGCCTAGAAGATC
Col2a1 Forward:
Reverse:
CCGTCATCGAGTACCGATCA
CAGGTCAGGTCAGCCATTCA
Col10a Forward:
Reverse:
AAGGAGTGCCTGGACACAAT
GTCGTAATGCTGCTGCCTAT
Ef1 Forward:
Reverse:
GGCGATGCTGCCATTGTT
GGAGGGTAGTCAGAGAAGCTCTCA
Evx1 Forward:
Reverse:
CAACCTAGTAGCTCAGACACCGAA
CGGTCTTGAAACGTAGTTCTCCCT
Foxa2 Forward:
Reverse:
GACATACCGACGCAGCTACA
GGCACCTTGAGAAAGCAGTC
Flk1 Forward:
Reverse:
CACCTGGCACTCTCCACCTTC
GATTTCATCCCACTACCGAAAG
156
Fst Forward:
Reverse:
GGGCAGATCCATTGGATTAGC
CCTTGGAATCCCATAGGCATT
GAPDH Forward:
Reverse:
TGAGGACCAGGTTGTCTCCT
CCCTGTTGCTGTAGCCGTAT
GATA1 Forward:
Reverse:
GATGGAATCCAGACGAGGAA
ACCAGCTACCACCATGAAGC
GATA2 Forward:
Reverse:
CGGCCTCTTCTTCTGCAGG
TGGTACTTGACGCCATCCTTG
Gsc Forward:
Reverse:
CGGCACCGCACCATCT
TGGGTACTTCGTCTCCTGGAA
HoxB1 Forward:
Reverse:
CAATGAAACGCAGGTGAAGA
GACTGGTCAGAGGCATCTCC
Lhx1 Forward:
Reverse:
CACCTCAACTGCTTCACCTG
TGTTCTCTTTGGCGACACTG
Meox2 Forward:
Reverse:
GTCTGTGGCAGTGTGGCTTA
AGCCAAAGCAAACATCCATC
Mesp2 Forward:
Reverse:
GGCTCAGATGCTTGGTCCTA
TCCCAAGGTTTTCAGGTGAG
Nestin Forward:
Reverse:
CTCGAGCAGGAAGTGGTAGG
GCCTCTTTTGGTTCCTTTCC
NeuroD Forward:
Reverse:
GCATGCACGGGCTGAACGC
GGGATGCCCGGGAAGGAAG
Nkx2.5 Forward:
Reverse:
AGTGGAGCTGGACAAAGCC
GACAGGTACCGCTGTTGCTT
Nodal Forward:
Reverse:
ACTTTGCTTTGGGAAGCTGA
CCAGCCAATCAGGTTGAAGT
157
Pdgfrα Forward:
Reverse:
TGCGTACATCGGTGTCACTT
GGGGATGATGTAGCCACTGT
PRG4 Forward:
Reverse:
GAACCGCCGGCTGTGGATGA
TGTGGTGACTTTGCTGTGTGGAGT
Runx2 Forward:
Reverse:
ACCATGGTGGAGATCATCG
GGCAGGGTCTTGTTGCAC
Sox9 Forward:
Reverse:
GCTGAACGAGAGCGAGAAGA
GAGGAGGAATGTGGGGAGTC
Sox17 Forward:
Reverse:
CCGAGATGGGTCTTCCCTAC
CGTCAAATGTCGGGGTAGTT
Tal1 Forward:
Reverse:
CCCACCAGACAAGAAACTAAGCA
GGCCAGGAAATTGATGTACTTCA
Wnt3a Forward:
Reverse:
GCTCTGCCATGAACCGTCACAACAAT
ATAGCCCGTGGCATTTGCACTTGA
Table A.1 – Primer sequences for qPCR analysis
158
A
B
Fig. A.1 – (A) HCI analysis of OCT4 expression from two-day CDM (i-iii) and X-Vivo™10 (iv-vi) cultures
showing similar biphasic profiles from cultures established on gelatin+fibronectin. (B) Compiled HCI data
indicated that OCT4 expression remained stable when cultures were established on gelatin, although cultures in N2B27 appeared to have variable OCT4 expression when initiated at a high seeding density. Dark grey bars
stands for N2B27 cultures, light grey stands for X-Vivo cultures and white represents cultures in CDM. Plotted
percentages represent means±SEM (n=2).
159
A
B
Fig. A.2 – (A) Four-day SF differentiation culture supplemented with Activin A (10ng/ml) had less
BRACHYURY+ cells compared to that with Activin A (30ng/ml). (B) Addition of both BMP4 and Activin A
(i) or Wnt3a (ii) on day 0 of differentiation enhanced the proportion of BRACHYURY+ cell population
compared to BMP4 alone. Cultures supplemented with Activin A+Wnt3a (iii) or serum+Activin A (iv) did
not appear to generate more BRACHYURY+ cells than cultures with Activin A, Wnt3a or serum alone.
Images were taken at 200x magnification.
A B
Fig. A.3 – (A) IF image (200x) of COL2A1 antibody staining and (B) Alcian blue staining for 15-day SF
monolayer differentiation culture supplemented with BMP4 (10ng/ml, from day 0 to day 15) and Activin A
(30ng/ml, from day 5 to day 15) confirmed the lack of COL2A1 networks and proteoglycan production, respectively.
BRACHYURY
DAPI/Hoechst
BRACHYURY
DAPI/Hoechst
COL2A1
DAPI/Hoechst
160
Fig. A.4 – Addition of (i) Activin A, (ii) TGFβ3 (10ng/ml) and (iii) FGF8 (50ng/ml) on day 5 of differentiation
to BMP4-treated cultures (from day 0 to 5) did not compensate for the non-inductive nature of BMP4, as
exhibited by the lack of COL2A1 networks (IF images at 200x).
COL2A1
DAPI/Hoechst
161
A
B
Fig. A.5 – (A) As part of the confirmation that Wnt3a acted as a late chondrogenic inducer, IF images (200x)
showed minimal COL2A1 staining in SF monolayer cultures supplemented with Wnt3a for (i-iii) 15 days or (iv-vi) five days. Addition of (i, iv) Activin A, (ii, v) TGFβ3 and (iii, vi) FGF8 to Wnt3a-supplemented
cultures from day 5 to 15 of differentiation did not improve COL2A1 network formation. This observation
was corroborated by the weak Alcian blue staining of the same cultures showing the lack of proteoglycan
production (B).
COL2A1
DAPI/Hoechst
162
αMHC Nkx2.5
GATA1 Sox17
Fig. A.6 – qPCR analysis of αMHC, Nkx2.5, GATA1 and Sox17 transcript levels in 15-day SF monolayer differentiation cultures subjected to 15-day BMP4, Activin A or Wnt3a supplementation. (i) Activin A or
serum treatment induced αMHC expression to similar levels as those in EB cultures. (ii) Serum-treated EB and
monolayer cultures had higher Nkx2.5 expression than Activin A cultures; BMP4 exerted the least enhancing
effect and Wnt3a caused a reduction in gene expression. (iii) Activin A exerted the least inductive effect on
GATA1 expression than other test conditions. (iv) All monolayer culture conditions showed decreased levels of
Sox17. Expression levels were compared to those in undifferentiated ESCs. Plotted values represent
means±relative error (n≥2).
163
αMHC Nkx2.5
GATA1 Sox17
Fig. A.7 – qPCR analysis of αMHC, Nkx2.5, GATA1 and Sox17 transcript levels in 15-day SF monolayer differentiation cultures subjected to five-day BMP4, Activin A or Wnt3a supplementation. Activin A
treatment dramatically enhanced the expression of αMHC (i) and Nkx2.5 (ii) compared to BMP4 and Wnt3a. Wnt3a
induced GATA1 gene expression (iii), while all three growth factors exerted similar effects on Sox17 expression (iv).
Expression levels were compared to those in undifferentiated ESCs. Plotted values represent means±SEM (n≥2).
164
Appendix B
Supplementary Data for Chapter 3
Gene Sequences
β-actin
Forward: GGCCCAGAGCAAGAGAGGTATCC
Reverse: ACGCACGATTTCCCTCTCAGC
Oct4
Forward: GGCGTTCTCTTTGGAAAGGTGTTC
Reverse: CTCGAACCACATCCTTCTCT
Brachyury
Forward: ATGCCAAAGAAAGAAACGAC
Reverse: AGAGGCTGTAGAACATGATT
α-MHC
Forward: GGAAGAGTGAGCGGCGCATCAAGG
Reverse: CTGCTGGAGAGGTTATTCCTCG
Foxa2
Forward: TGGTCACTGGGGACAAGGGAA
Reverse: GCAACAACAGCAATAGAGAAC
Sox17
Forward: GCCAAAGACGAACGCAAGCGGT
Reverse: TCATGCGCTTCACCTGCTTG
Pax6
Forward: GCTTCATCCGAGTCTTCTCCGTTAG
Reverse: CCATCTTGCTTGGGAAATCCG
NeuroD
Forward: CTTGGCCAAGAACTACATCTGG
Reverse: GGAGTAGGGATGCACCGGGAA
Targeting Primer Sequences
Rosa 5’ Forward TCTGTTGGACCCTTACCTTGAC
CAGGS Reverse GCCAAGTAGGAAAGTCCCATAAG
BSD Forward CATAGTGAAGGACAGTGATGGACAGC
Rosa 3’ Reverse AGCAACATTTAACACAGTG
Table B.1 – Primer sequences for RT-PCR and targeting PCR analyses
165
A
B
Fig. B.1 – Schematic of (A) the BP reaction that generates an entry clone from PCR-amplified DNA fragment
and the donor vector and (B) the LR reaction that creates an expression clone from an entry clone and a
destination vector (Invitrogen 2006).
Fig. B.2 – Schematic of the promoterless destination vector used in MultiSite Gateway® cloning (Invitrogen
2006).
166
A
B
C
D
Fig. B.3 – Schematics of the MultiSite Gateway® donor vectors used in a four-fragment cloning reaction (Invitrogen 2006). Fragments of interest were cloned into the donor vectors in a specific order such that the 5’ most fragment was inserted into (A) while the 3’ most fragment was cloned into (D).