Cell culture and differentiation

The cells were cultured at 37°C in a humidified incubator with 5% CO₂. The hES cells (SA121, Takara Bio Europe AB) were cultured in the DEF-CS™ 500 system (Takara Bio Europe AB). For differentiation, the cells were dissociated into single cell suspension with TrypLE Select (Invitrogen) and seeded at 40.000cells/cm² in DEF-CS medium according to the supplier’s instructions [23]. The cells were kept undifferentiated in DEF medium for 3 days. Differentiation to DE cells was done in basal medium containing, RPMI1640 + Glutamax (RPMI) medium (Gibco) with 2% B27 without insulin (Gibco) and 0.1% Pen strep (Invitrogen). On day 4 after cell seeding, the cells were rinsed with basal media and basal media with 100ng/ml recombinant mouse Wnta3 (R&D Systems) was added. On day 5 after cell seeding, the cells were rinsed with basal media and basal media with 10ng/ml Activin A was added (Peprotech). Fresh basal media with 10ng/ml Activin A was added on day 6 and 7. Cells from different passages were used as biological replicates. Viability and number of cells before plating were measured using a NucleoCounter® NC-3000TM (Chemometec, DK) according to manufacturer’s instructions.

Preparation of extracellular matrix coated substrates

The ECMP microarray slides were prepared with a non-contact micro-dispensing robot (Nanoplotter^TM (GeSiM, Germany)). The Nanoplotter had cooling of the microtitre plate and the humidity was kept to 60% during the spotting to avoid rapid evaporation. Solutions of ECMP and PBS (Gibco) were mixed in 384 microtitre plates. 487 combinations of ECMPs were made from 14 ECM proteins and peptides, which were selected from literature studies and represent the major ECM families. The following ECMPs were used: Fibronectin (Sigma) (Fn), laminin 511 (Biolamina) (Ln511), laminin 111 (Biolamina) (Ln111), laminin 521 (Biolamina) (Ln521), laminin from human fibroblast (Sigma) (LnH), vitronectin (Sigma) (Vn), fibronectin adhesions promoting peptide (Sigma) (FnAd), collagen I from rat tail (Gibco) (Col1), collagen IV (R&D Systems) (Col4), collagen II (Millipore) (Col2), collagen III (Millipore) (Col3), collagen V (Millipore) (Col5), heperan sulphate proteoglycan (Sigma) (Hep), nidogen 1 (R&D Systems) (Nid) and netrin 1 (R&D Systems) (Ne). The concentrations of the individual ECMP reflect the recommendations from of the suppliers. Further details about the ECMPs are described in S1 Table. 10% (w/v) bovine serum albumin (BSA, Sigma) was used as a negative control because it is an unrelated protein to ECMP. Each protein combination (S1 Table) was spotted in four replicates per array using a Nano-tipA (GeSiM) with the following tip settings: 150 droplets per spot, 250μs delay, 50μs pulse, 75 V, 100hz and 1.1mm distance between each spot. The resulting spot diameter was about 400±50μm. The 487 protein combinations and replication required two microscope slides. The slides were made of polystyrene (Nunc) the same material as cell culture dishes. The ECMP microarray slides were stored at -18°C until use, with a maximum storage time of 3 weeks. The slides were transferred to cell culture dishes fitted to microscope slides (Nunc, Denmark) and they were incubated at 37°C for two hours to ensure complete absorption of ECMPs. Subsequently, the slides were sterilized with 70% ethanol for 30 minutes and washed 3 times with PBS and were incubated at 37°C until seeding of the cells. Microtitre well plates (Cellbind, Corning) were coated by adding ECMPs diluted in PBS and incubated at 37°C for 1–2 hours. The ECMP dilution was removed immediately before seeding of cells, leaving the well plate constantly wet.

Immunofluorescence staining and microscopy

The cells were fixed with 4% formaldehyde (Lilly’s fixative, Mallinsckridt Baker) for 20 minutes at room temperature. The samples were permeabilized with 0.5% Triton-X-100 (Sigma Aldrich) in PBS for 10 minutes at room temperature and subsequently blocked for 30 minutes using TNB blocking buffer (0.1M tris-HCL pH 7.5, 0.15M NaCl and 0.5% blocking reagent from Perkin Elmer TSA kit). Primary antibodies were diluted in 0.1% TritronX-100 in PBS and applied to the samples followed by incubation overnight at 4°C. The following primary antibodies were used: goat polyclonal anti Sox17 (1:1000) (R&D Systems, AF1924), mouse polyclonal anti Oct3/4 (1:500) (Santa Cruz, sc5279). The samples were washed three times with PBS and the secondary antibodies Alexa Fluor 594 conjugated donkey-anti-mouse IgG (Invitrogen) and Alexa Fluor 488 donkey-anti-goat IgG (Invitrogen) were added in a 1:1000 dilution together with DAPI (1:2000) (Sigma- Aldrich) and occasionally Alexa Fluor 488 Phalloidin (1:20) (Invitrogen) in 1% TritronX-100 in PBS for 45 minutes at room temperature. The samples were rinsed three times with PBS for five minutes.

Microscopy images were obtained on the InCell analyzer 2000 (GE Healthcare). A 10X/0.45 Plan Apo objective was used with laser auto focus. For the microarray screen one image per spot was taken, whereas in well plate formats 9–16 images per well were taken. Images were analysed and quantified using the InCell Developer Toolbox 1.9.2 (GE healthcare). Representative images were processed in ImageJ or InCell Developer Toolbox. Example images are given in S2 Fig.

Quantitative Polymerase Chain Reaction (qPCR)

RNA was isolated and purified from cell cultures using Nucleospin RNA Kit (Macherey-Nagel) and RNA concentration and purity were measured with the NanoDrop ND-100 spectrophotometer (Thermo Scientific). cDNA was prepared from 500ng RNA per sample using the iSCript cDNA Synthesis kit (Bio-rad) and the cDNA was subsequently diluted 10 times. qPCR was performed using 1/100 of the cDNA, 500nM each of the respective forward and reverse primers and 2x Brilliant III Ultra-FastSYBR Green master mix (Agilent) in 10μl reactions. Primer sequences are listed in S2 Table The reactions were performed on a MX3005P qPCR system (Agilent) using the following program; 95°C for 3 minutes, 40 cycles of 95°C for 15 seconds 60°C for 20 seconds, followed by a melting curve cycle of 95°C for 20 seconds and 55°C for 30 seconds. Three independent biological replicates were obtained with hES cells from different cell passages. For each biological replicate, technical triplicates were analysed. The two housekeeping genes TBP and GUSB, were used to normalize the data [5].

Statistical Analysis

For the ECMP micro array screen four biological replicates with hES cells from different cell passage where used and each biological replicate contained four technical replicas for each ECMP combination. Data from the ECMP microarray screen was normalized by calculating a z-score for each spot:

R_iSox17 was the ratio between the number of Sox17 positive cells and total cells (DAPI stained) for the given spot. μ_Sox17 was the average of Sox17/DAPI ratio of all the spots in the given experiment (N1-N4), whereas σ_Sox17 was the standard deviation of Sox17/DAPI ratio of all the spots in the given experiment. In addition a z-score for number of cells (DAPI) was calculated in a similar way: where DAPI was the number of DAPI positive cells. Z-scores from replication spots (four per ECMP combination) were averaged, and the given values were an average from the four independent experiments (N1-N4).

For the experiments in the microtitre plates, paired–test was used for statistical analysis and values in graphs were presented as mean ±S.E.M of at 3–5 independent biological replicas with hES cells from different cell passage.

Gene expression analysis using DNA microarrays

RNA was isolated as described above and the quality of RNA samples was measured using the 2100 Bioanalyzer (Agilent) and only samples with RNA integrity number (RIN) above 7 were used. Microarray analysis was performed by the array facility DMAC at Technical University of Denmark. RNA samples were prepared according to the protocol One-Color Microarray-Based Gene Expression Analysis, Low Input Quick Amp Labelling, Version 6.6, September 2012 (Agilent Technologies, USA). 100ng total RNA per samples was used. Samples were subsequently hybridized to Agilent SurePrint Whole Human Genome (4x44K) Oligo Microarrays and scanned as described by the manufacturer (Agilent Technologies, Santa Clara, CA, US) using the DNA microarray scanner from Agilent Technologies. The raw fluorescent data was extracted using the Agilent Feature extraction software.

Batch effects were removed using the ComBat function of the sva R package for modelling and intensity values were normalized between expression arrays by applying quantile normalization [24]. t-test using the eBayes function of the limma R package was used to test significance of differential expression. Bonferroni correction and Benjamini-Hochberg correction were used to correct for multiple testing. The gene expression from cells cultured on the respective ECMP was compared to the gene expression of cells cultured on Fn. Pathway analysis was done using WikiPathways [25–28] using differentially expressed genes at the P<0.01 after Benjamini-Hochberg correction.

Identification of extracellular matrix proteins combinations that promote definitive endoderm differentiation

The aim of this work was to identify ECMPs that promote the differentiation of hES cells into DE cells. hES cell to DE differentiation reach >90% on Fn using optimal protocols [29]. It is difficult to quantify possible positive effects of ECMPs with such high differentiation efficiency. Therefore a suboptimal DE differentiation protocol was evaluated. This protocol includes 100ng/mL Wnt3a and 10ng/mL Activin A and generates around 60% Sox17 (DE marker) positive cells when differentiated on Fn substrates (S1 Fig). Increasing the Activin A concentration to 100ng/mL resulted in >80% DE cells (S1 Fig) indicating that adding cues to the suboptimal protocol increased the DE differentiation. The suboptimal protocol was used to screen for ECMPs improving DE differentiation. The ECMP microarray screening method was developed and optimized, including selection of microscope slide material as well as selection of the size of the individual protein spots and the distance between them. Polystyrene (same material as tissue culture flasks) was superior to all other tested slide types. Differentiation of hES cells on Fn spotted polystyrene slides showed that 400±50μm spots, spaced 700 μm apart gave about 60% differentiation and thus similar as in microtitre plates.

A microarray of 487 different combinations of ECMPs (S1 Table) was used to screen for ECMP combinations that promote DE differentiation. DAPI staining of cells at day one after plating on the array showed that some ECMPs immobilized hES cells homogeneously and efficiently (e.g. Fn) while other ECMPs did not capture hES cells efficiently (e.g. pure Col1, Fig 1E). The hES cells did not attach to the negative control (BSA) spots. The differentiation was evaluated by quantitative immunofluorescence microscopy of Oct3/4 (pluripotent marker) and Sox17 (DE marker) followed by microscopy and image analysis. Examples of immunostainings of spots after differentiation are given in Fig 1B. Morphologically, cells on spots after differentiation often appeared in distinct clusters of Oct3/4 positive and Sox17 positive cells respectively. Large regions with 100% Sox17 positive cells were observed on the spots with Fn+Hep+LnH, Col1+Fn+FnAD and Fn+Ne. Also Oct3/4 positive cells occasionally displayed a sub-spot organization. Oct3/4 positive cells were organized in the periphery of the spots on some ECMP combinations (e.g. Fn+Ne+Vn and Vn alone, Fig 1B) or organized in scattered islands (e.g. on Fn+Ne). Some ECMPs spots, including pure Col1, contained too few or no cells to evaluate DE induction after differentiation (Fig 1E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The microarray screen with combinations of ECMPs demonstrated that the DE differentiation was affected by the ECMPs.

(a) Schematic overview of the ECMP microarray screen. The ECMP microarray slides were prepared by inkjet spotting and undifferentiated hES cells were subsequently seeded and differentiated towards DE. To assess the differentiation, the samples for immunofluorescence were stained for the markers Oct3/4 (undifferentiated hES cells), Sox17 (DE) and DNA (DAPI), and microscope images were captured using an automated microscope, InCell Analyzer. The images were quantified and the data was analysed. (b) Fluorescence microscope images of selected spots stained for Oct3/4 (red) and Sox17 (green) (scale bar = 200μm). (c) Sox17 z-scores plotted against cell number z-scores for each respective ECMP combination (n = 4). The green square indicate the ECMP combinations giving highest fraction of DE cells, whereas the blue square represents the ECMP combinations giving highest cell number. (d) Tables of the ECMP combinations giving to the 10 highest fractions of Sox17 positive cells and top 10 highest cell count respectively. (e) Immunofluorescence staining of spots from the protein microarray screen. Images were captured at day 1 (one day after seeding) and day 8 (at the end of the DE differentiation protocol) (Scale bar = 200μm).

https://doi.org/10.1371/journal.pone.0145389.g001

The numbers of Oct3/4 positive, Sox17 positive and DAPI positive cells on each spot were quantified and z-scores were calculated (Fig 1C). It was possible to identify ECMP combinations that produced a higher percentage of DE cells (Fig 1C, green box and Fig 1D) or the same percentage of DE cells but with higher cell counts (Fig 1C, blue box and Fig 1D) compared to cells on Fn. However, it was not possible to identify ECMPs that both resulted in high cell number and high DE differentiation fraction using the microarray screen. 9 of the top 10 ECMP combinations giving high Sox17 fraction were a mixture of ECMPs: Fn, Ne, Col1, LnH, Col2, Col3 and Hep in combinations seem to promote DE induction in the screen. The single most dominant ECMP resulting in high cell counts was Vn.

Extracellular matrix proteins component analysis

The top 10 combinations based on z-scores (Fig 1C and 1D) from the ECMP microarray screen and new ECMPs combinations based on the top 10 combinations were further tested in corresponding microtitre plate assays, where the ECMPs were immobilized by 1-2h incubation in a buffer. Overall, the majority of the tested combinations supported good cell growth and/or cell attachment well as the majority of the ECMP combinations resulted in cell numbers >18000 (Fig 2A). All the ECMP combinations with high cell number during the differentiation in the ECMP microarray screen were also verified to have high cell number in the well plate format (S3 Table). Furthermore, these ECMP combinations only induced intermediate percentage of Sox17 positive cells (43–61%, Fig 2A, S3 Table), which confirm the data from the ECMP microarray screen. It appeared that Vn had a positive effect on the cell number in the well plate format (Fig 2A blue squares and S3 Table) but suppressed DE induction.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. (a-d) DE differentiation of hES cells in microtitre plates on selected ECMP combinations.

(a-d) Quantification of Sox17 positive cells and total cell number based on immunofluorescence image analysis of cells differentiated on different ECMP combinations in microtitre plates. n = 3–6, mean ± S.E.M.). * and ** indicate statistical significant differences to Fn, P<0.05 and P<0.005 respectively. Note that not all statistical significant differences are displayed in the graphs. Full data set is given in S3 Table. Abbreviations are explained in S1 Table. (a) Scatter plot of percentage of Sox17 positive cells and the total number of cells for each respective ECMP combination tested (see S3 Table for descriptions), (b) Scatterplot of percentage Sox17 positive cells and the total yield of Sox17 positive cells (calculated as percentage of Sox17 positive cells * total cell number /100). Linear regression of data excluding six outliers (out of 65) not considered following the linear correlation (see results). r² = >0.95. (c) Percentage of Sox17 positive cells of a selected ECMP combinations. (d) The total cell number of a selected ECMP combinations.

https://doi.org/10.1371/journal.pone.0145389.g002

Seven of the top 10 ECMP combinations improving DE induction in the ECMP microarray screen were confirmed to improve DE induction in the microtitre plate format (Fig 1D, S3 Table). In general there was a strong correlation of the presence of Col1 in mixtures and high DE induction (Fig 2A green rhombs). The highest percentage of Sox17 positive cells (84%) was obtained with pure Col1 (Fig 2A and 2C). When Vn was present in ECMP combinations containing Col1, low DE induction was observed (47–56% Sox17 positive cells), (Fig 2A). This indicates the Vn has a suppressive effect on DE induction that dominates the positive effect of Col1 (Fig 2A and 2D). Vn had by contrast a positive effect on Col1’s ability to yield high cell numbers (Fig 2D). The presence of Col2 also correlated with high DE induction but less so than Col1 (Fig 2A). However, Col2 needed Fn as cofactor in order to also yield high cell numbers (Fig 2C and 2D). Ne contributed to DE induction, but only in combination with Fn (Fn+Ne, Fig 2C and 2D). Otherwise Ne together with other ECMP did not seem to have an additive effect on DE induction. Ne was associated with lower cell number yields after differentiation; Ne alone gave no cells and Col1+Ne had significantly reduced cell number compared to Col1 alone. By contrast, Fn seems to rescue poor cell yield of Ne and the combination resulted in high Sox17 induction and high cell yield (Fig 2C and 2D).

The total Sox17 positive cell yield ((percentage Sox17 positive * cell number)/100) was plotted (Fig 2B). The majority (59 out of 65) of the combinations followed a straight line (r²>0.95) suggesting that the Sox17 positive cell yield is a function of DE induction. I.e., the higher the DE induction frequency, the higher the Sox17 positive cell yields. However, this correlation was not seen in ECMP combinations containing Col1 and Col2, which had low cell yields while still giving high percentage of Sox17 positive cells (Fig 2B). The highest Sox17 positive cell yields were obtained using Col1+Fn but this combination resulted in a slightly lower Sox17 positive percentage (Fig 2C and 2D). The higher Sox17 positive yield was mainly due to higher total cell number of Col1+Fn compared to Col1 alone (Fig 2C and 2D).

Kinetics of differentiation on different extracellular matrix proteins

In order to pinpoint the time when the effect of different ECMPs affects differentiation, a time course study was performed. Five ECMP coatings were selected for the time course studies of DE differentiation. Col1, Col2+Fn and Ne+Fn were selected because their ability to induce DE and that they represent different types of ECMPs. Fn was used as a cofactor to Col2 and Ne respectively in these experiments. Vn was selected based on its poor ability to induce DE and to contrast the effects of the high performing ECMPs. Pure Fn was used as a reference. The cells were analysed at each time point by immunofluorescence (Figs 3 and 4) and quantitative PCR (Fig 5). hES cells initially attached (day 1) in tight clusters on Col1 and Col2+Fn coated wells. By contrast, the hES cells seeded on Fn and Vn were evenly distributed (Fig 3). The number of hES cells attached to Col1 coated wells one day after seeding (day 1) was significant lower than the number of cells on Fn coated wells (Fig 4A), corroborating the findings in the ECMP microarray screen (Fig 1E). The other investigated combinations resulted in similar number of cells attached as on Fn (Fig 4A). Three days in pluripotent culture conditions expanded the number of cells about 8–10 fold from initial attachment cell numbers (Figs 3, 4A and 4B respectively). However, the Col1 coating resulted in about 16-fold increase in cell number (Fig 4A and 4B). At this stage all the cell cultures were confluent. The total number of cells decreased throughout differentiation for all ECMPs except for Col1 where the total number of cells was stable (Fig 4B). At day 8, the total number of cells was at similar level for all ECMPs except for Vn, which had higher number of cells (Fig 4B). After one day with Wnt3a treatment (day 5), the overall culture morphology differed greatly in Col1 cultures compared to the other ECMPs. Cells adhered tightly together on Col1 and holes in the cell layer appeared (Fig 3C). It should be noted that the morphology of individual cells did not differ on the respective substrate (S3 Fig). Sox17 positive cells appeared on day 6 in all cultures (one day with Activin A). The Col1 coating had about 45% Sox17 positive cells on day 6 while the Fn coating had about 12% Sox17 positive cells. The cell culture morphology indicated that the holes formed at day 5 were repopulated within a day. It appeared that both Sox17 positive cells and Oct3/4 positive cells repopulated the holes in the cell layer observed at day 5 on Col1. The percentage of Sox17 positive cells increased for all the ECMPs during differentiation with Activin A (Figs 3 and 4). The Col1 coating had a higher percentage of Sox17 positive cells compared to Fn coating throughout the entire differentiation process (Fig 4C). Ne+Fn and Col2+Fn followed the differentiation pattern of Fn until the last day, day 8, where these ECMP combinations resulted in higher percentage of Sox17 positive cells compared to Fn (Fig 4C). Vn coated cultures followed Fn coated cultures in the percentage of Sox17 positive cells throughout the differentiation (Fig 4C). Sox17 positive cells had similar morphology on all substrate (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The dynamic differentiation of hES cells to DE is influenced by ECMPs.

(a) Schematic overview of the semi-optimal protocol for DE differentiation. hES cells were seeded and kept undifferentiated for 4 days to allow expansion. Subsequently the cells were treated with Wnt3a to direct the cells from the pluripotent stage into the mesendoderm stage, followed by 3 days with Activin A to direct the cells from the mesendoderm stage to the DE stage. (b) Representative immunofluorescence images from three independent experiments of cells cultured on the five different ECMP combinations at different time points during the DE differentiation protocol. The tested ECMP combinations were fibronectin (Fn), collagen II+fibronectin (Col2+Fn), collagen I (Col1), netrin 1+fibronectin (Ne+Fn) and vitronectin (Vn) (scale bar = 200μm). (c) Magnification of cultures on Col1 at day 5 (scale bar = 200μm).

https://doi.org/10.1371/journal.pone.0145389.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Quantitative immunofluorescence of cell cultures at different time points during the DE differentiation protocol on the different ECMP substrates.

(a) Quantification of number of cells, which were attached 24h after seeding on the respective ECMP substrates (day 1). (b) Quantification of the number of cells at the different days during DE differentiation. (c) Quantification of the fraction Sox17 positive cells different days during DE differentiation. (d) Quantification of the fraction Oct3/4 positive cells at different days during DE differentiation. (n = 3, mean ± S.E.M. and * indicate statistical significant differences, P<0.05). ECMP abbreviations are explained in legend in Fig 3.

https://doi.org/10.1371/journal.pone.0145389.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Gene expression analysis for markers at different time points during DE differentiation.

Following genes were analysed; the pluripotent markers POU5F1 (a) and NANOG (b); mesendoderm/early DE marker BRACHUYRY (c); the DE markers GOOSECIOD (d), SOX17 (e), FOXA2 (f); the WNT downstream target AXIN2 (g); and the WNT inhibitor DKK1. (n = 3, mean ± S.E.M.). Abbreviations are explained in legend Fig 3.

https://doi.org/10.1371/journal.pone.0145389.g005

To characterize the DE differentiation and to identify any potential differences between the ECMPs, the expression of different genes were assessed for the five time points (Fig 5). The expression of the pluripotent gene POU5F1 (OCT3/4) [30] declined in general throughout the differentiation towards DE irrespective of ECMPs used. Cells on Col1 coating had a significant lower expression (P<0.05) of POU5F1 compared to cells on Fn coating from day 6 and onwards (Fig 5A). The expression of the pluripotent marker NANOG [31] increased after Wnt3a priming (day 5) and subsequently declined throughout the differentiation (Fig 5B) which is in accordance with previous results [32]. At day 5, cells on Col1 and Ne+Fn coatings had significant higher (P<0.05) expression level of NANOG compared to cells on Fn coating. At day 8, cells on Col1 had significant lower (P<0.05) NANOG expression level compared to cells on all other ECMP coatings. Cells on Col2+Fn and Ne+Fn coatings followed cells on Fn coating in the relative expression of NANOG after day 4 (Fig 5B). The expression of BRACHUYRY, a transient mesendoderm marker [7,33], increased during the first days of differentiation and subsequently declined, which corroborates previous studies [5,32]. The expression of BRACHUYRY in cells on Col1 coating increased 675 fold on day 5 compared to day 1 (Fig 5C). The expression was significantly higher (P<0.0005) than in cells on Fn coating. The BRACHUYRY expression for cells on Ne+Fn and Vn coatings also peaked on day 5, where the expression was significantly higher than in cells on Fn coating (P<0.05) but not as high as cells on Col1. However, for cells on Fn and Col2+Fn coatings, the expression of BRACHUYRY peaked one day after the other ECMPs (day 6), and only with about a 200-fold increase in expression (Fig 5C). The expression of the anterior mesendoderm marker GOOSECOID [34], appeared 1–2 days later than BRACHUYRY, which is comparable to the results of previous studies [5]. The cells on the different ECMPs displayed comparable gene expression profiles for GOOSECOID and SOX17. These genes were upregulated on day 6 and 7 and decreased on day 8. There was no significant difference in GOOSECOID (Fig 5D) or SOX17 (Fig 5E) expression in cells on the different ECMPs tested. However, there was a tendency that cells on Col1 had higher induction of these genes. FOXA2, a DE marker [35], had similar expression pattern as SOX17 and GOOSECOID. However, the FOXA2 expression for cells on Col1 coating peaked at day 6, whereas the expression of FOXA2 on the other ECMPs peaked at day 7 (Fig 5). The expression of AXIN2, a downstream target of WNT signalling [36], were similar in cells on all the tested ECMPs. The expression increased on day 4 and subsequently declined (Fig 5G). In contrast, the expression of DKK1, a Wnt inhibitor [37], was significant higher on day 7 in cells on Col1 coating compared to cells on Fn coating (P<0.05). The DKK1 expression in cells on the other ECMPs was similar to cells on Fn coating (Fig 5H). Quantitative PCR of CXCR4 showed that all ECMPs induced CXCR4 expression after the 8 day differentiation protocol (S4 Fig) indicating that the cells where differentiated towards definitive endoderm and not visceral endoderm [38]. To summarize, kinetics of the majority of the investigated genes were similar for the ECMPs and concur with previous studies of DE differentiation. However, the data suggests that cells on Col1 entered the differentiation path through earlier and stronger mesendoderm induction compared to cells on other substrates.

Gene expression arrays

Global gene expression analysis was used to investigate the cell populations obtained after 8 days growth and differentiation in order to evaluate how similar the cells are after differentiation on different substrates. The data showed that only cells cultured on Col1 showed significant changes in gene expression compared to cells cultured and differentiated on Fn (the data is summarised in S4 and S5 Tables and completely presented in S6–S10 Tables). About 92 genes were differentially expressed in cells on Col1 compared to cells on Fn using Bonferroni correction (P<0.05). The results also showed that the qPCR data was confirmed by the DNA microarray analysis (Fig 5 and S4 Table). In general, there were only modest differences in gene expression profile of cell differentiated on Col1 compared to cells on Fn (S5 Table) indicating that the cells on the respective substrate were largely similar. The top downregulated gene (about three-fold) was DACT1 which is an inhibitor of Wnt signalling [39,40] (S5 Table). Other downregulated genes were POU5F1 (2.8-fold) and SOX2 (2.5-fold), which are markers for pluripotent cells, indicating a better differentiation on Col1. One interesting gene that was upregulated was Nitric oxide synthase 2 (NOS2) (2.6-fold), which has been shown to activate beta-catenin and EMT in breast cancer cells [41]. It should be noted, that any transient effects of Col1 observed at day 5–6 (Fig 5) would not be expected to be detected by the DNA microarray analysis since it was only conducted for day 8. Pathway analysis of the differentially expressed genes showed that cells cultured on Col1 compared to cells cultured on Fn had differentially expressed genes involved in focal adhesion, neural crest differentiation, spinal cord injury and Notch signalling.