We tested whether defined human ECM proteins could replace Matrigel in the matrix sandwich cardiac differentiation protocol. The matrix sandwich protocol uses Matrigel coating for hPSC adhesion and expansion followed by overlaying the proliferating hPSCs with Matrigel at day –2 and day 0 of differentiation with the addition of growth factors Activin A at day 0–1, followed by addition of BMP4 and bFGF at day 1–5 (Figure 1A). Therefore, we first tested the ability of defined ECM proteins to support hPSC adhesion and expansion. DF19-19-11T iPSCs or H1 ESCs were seeded on human LN111, LN521, COL4, and FN coated surfaces and cultured in mTeSR1 medium. The hPSCs grew as a monolayer on LN111, LN521, and FN and exhibited similar morphology and expression of the pluripotency markers OCT4 and SSEA4 as hPSCs grown on Matrigel (Figure 1B, Figure 1—figure supplement 1). However, the hPSCs seeded on COL4 did not grow as a confluent monolayer (Figure 1, Figure 1—figure supplement 2), so this matrix coating was not tested further.

For those matrix substrates that supported monolayer growth of hPSCs, including LN111, LN521 and FN, we tested overlay of defined ECM proteins in the matrix sandwich protocol using DF19-9-11T iPSCs and H1 ESCs. Cardiac differentiation was measured by flow cytometry for cTnT⁺ cells at 15 days of differentiation (Figure 1C, Figure 1—figure supplement 3). hPSCs seeded on Matrigel showed poor cardiac differentiation in response to the growth factors without matrix overlay, but with Matrigel overlays the percentage of cTnT⁺ cells was significantly increased as we previously reported (Zhang et al., 2012). Interestingly, FN overlays were as effective as Matrigel overlays in promoting cardiac differentiation of the hPSCs growing on Matrigel. However, if cells were seeded on LN111, LN521, or FN coated surfaces, the overlay of Matrigel or FN did not further increase the efficiency of hPSC-CM generation, and overlays of LN111, LN521, and COL4 strongly inhibited cardiac differentiation. These results demonstrated that the defined ECM proteins of LN111, FN and to a lesser extent LN521 support hPSC adhesion, growth and cardiac differentiation in monolayer hPSC culture and do not require a matrix overlay for efficient cardiac differentiation using the Activin A/BMP4/bFGF growth-factor-directed protocol.

Since FN promoted cardiac differentiation as both a culture matrix and an overlay, and also because FN plays key roles in EMT during gastrulation and cardiogenesis (Boucaut et al., 1996; Boucaut and Darribere, 1983; Boucaut et al., 1984; Darribère et al., 1988; Johnson et al., 1993; Lee et al., 1984; Lim and Thiery, 2012; Linask and Lash, 1986; Linask and Lash, 1988a, Linask and Lash, 1988b; Nieto et al., 2016; Suzuki et al., 1995; Thiery and Sleeman, 2006), we next examined for the presence of FN ECM in hPSC culture. DF19-9-11T iPSCs plated on Matrigel-coated surfaces and cultured in mTeSR1 per the matrix sandwich protocol were immunolabeled with a FN antibody on days −3,–2, –1 and 0 without permeabilizing the cells to examine extracellular FN protein (Figure 2A). On days –3 and –2, minimal immunolabeled FN in ECM was observed. However, after 4 days of culture (day 0) immunolabeled fibrillar FN ECM was abundant in the matrix sandwich culture (Figure 2A and B). In contrast, the monolayer culture control in the absence of Matrigel overlay had significantly less FN ECM present by day 0 (Figure 2A and B). This suggests that the Matrigel overlay promotes the production of endogenous FN or remodeling of FN ECM relative to the monolayer culture control without Matrigel overlay. Because the hPSCs cultured on LN111 coated surface without matrix overlay enabled efficient cardiac differentiation (Figure 1C, Figure 1—figure supplement 3), we examined the endogenous FN production in the hPSC culture on LN111 coated surface in which no exogenous FN was added. The hPSCs grown on LN111 coated surface without any matrix overlay were immunolabeled with the FN antibody without permeabilizing the cells. Confocal z-scan of the cell culture showed no detectable FN ECM at days –3 and –2, similar to the Matrigel/Matrigel sandwich culture; however, by day 0, dense fibrillary FN ECM was present in the cell culture on LN111 coated surface (Figure 2C, DF19-9-11T iPSCs; Figure 2—figure supplement 1, H1 ESCs). Similar to our previous study of the Matrigel/Matrigel sandwich culture (Zhang et al., 2012), the hPSCs growing on the LN111 coated surface without any matrix overlay formed multilayer cultures as shown in the side view of the confocal z-scan as did FN/FN matrix sandwich culture (Figure 2D). To determine if the hPSC culture results in accumulation of endogenously produced laminin ECM as well, DF19-9-11T iPSCs cultured on FN, LN111, and LN521 coated surfaces were immunolabeled with an antibody detecting laminins without permeabilizing the cells, and did not show measurable laminin ECM after 4 days of growth on these defined matrices (Figure 2—figure supplement 1). These results together with the above cardiac differentiation results supported the potential role of FN in promoting ActivinA/BMP4/bFGF-directed hPSC cardiac differentiation.

To characterize the early stages of cardiac differentiation of hPSCs cultured on LN111 and treated with Activin A/BMP4/bFGF growth factors, we examined markers of EMT, mesoderm and cardiac mesoderm. Gene expression was assessed by quantitative RT-PCR, and upon the addition of Activin A, BMP4 and bFGF, there was significant upregulation of transcription factors associated with EMT including SNAI1 (Leptin and Grunewald, 1990), SNAI2 (Nieto et al., 1994), and TWIST (Thiery et al., 2009; Figure 3A). The mesenchymal cell markers of vimentin (VIM), FN (FN1) and N-cadherin (CDH2) were also greatly upregulated by day 3 (Figure 3A). In contrast, E-cadherin expression (CDH1), an epithelial cadherin, was greatly downregulated by day 3 of differentiation. The mesendoderm/mesoderm transcription factors GSC, MIXL1, SOX17 and TBXT were transiently upregulated followed by expression of cardiac transcription factors of MESP1, ISL1, NKX2-5, and GATA4 at days 3–5 (Figure 3B).

To determine if FN ECM persisted or remodeled during the early stages of cardiac differentiation on LN111 substrate, immunolabeling of the early differentiated DF19-9-11T iPSCs on days 0–3 for FN and Brachyury was performed. Confocal z-scan imaging showed abundant FN ECM at each day, and Brachyury⁺ cells were associated with the dense network of FN ECM (Figure 3C), suggesting that the Brachyury⁺ cells interact with FN. Similar Brachyury⁺ cells and the FN ECM network were also observed in the Matrigel/Matrigel and FN/FN matrix sandwich cultures (Figure 3—figure supplement 1). Together, these results show that hPSCs grown on LN111, undergo the early stages of cardiac differentiation with transitions to mesoderm and cardiac mesoderm progenitors occurring in an endogenously generated FN-rich ECM, similarly as in the Matrigel/Matrigel and FN/FN matrix sandwich cultures.

To determine if FN is essential for mesoderm formation in our protocol and to investigate the stage-specific roles of FN during cardiac differentiation of hPSCs, we generated a doxycycline (dox) inducible FN knockdown system using FN1 shRNA (Figure 4—figure supplement 1). The two vectors shown in Figure 4A were incorporated into lentivirus and transduced into hPSCs. Clones were selected by neomycin resistance from both hESC line H1 and hiPSC line DF19-9-11T. To confirm the dox inducibility of the FN1 shRNA in the cell lines, we first assessed dox-induced bicistronic mCherry expression (Figure 4—figure supplement 2A). Inducible FN knockdown was demonstrated by immunolabeling with FN antibody (Figure 4—figure supplement 2B) and quantitative western blot for FN expression (Figure 4—figure supplement 2C).

Cardiac differentiation was performed using the monolayer based protocol with the H1 inducible FN knockdown clones growing on LN111 coated surface and treated with the growth factors Activin A, BMP4 and bFGF as shown in Figure 4B. To probe the stage-specific effect of FN knockdown during cardiac differentiation, dox was added at different time points: day 0–1, day 1–5, and day 5–7, and cardiac differentiation was measured by flow cytometry for cTnT⁺ cells at 15 days of differentiation. Cardiac differentiation was significantly inhibited when FN was knocked down at day 0–1, whereas FN knockdown at days 1–5, or days 5–7, did not have significant impact on the percentage of cTnT⁺ cells when compared to the no dox control (Figure 4C).

Because FN knockdown at day 0–1 significantly inhibited cardiac differentiation, we next tested if exogenous FN at this stage can rescue cardiac differentiation. Using the same protocol as shown in Figure 4B with dox induction of FN knockdown at day 0–1, we added soluble human FN (3 µg/cm²) in the cell culture on day 0–1. Cells were differentiated for 15 days, and cardiac differentiation was measured by flow cytometry for cTnT⁺ cells as above. Adding exogenous FN fully rescued the hPSC-CM differentiation, giving rise to a similar percentage of cTnT⁺ cells compared to the no FN knockdown control (Figure 4C). The effect of dox-induced FN knockdown on day 0–1 was concentration-dependent (Figure 4—figure supplement 3A), but at the highest concentration of dox tested (8 µg/ml), there was evidence for dox toxicity based on the loss of viability of nontransgenic H1 cells undergoing the differentiation protocol (Figure 4—figure supplement 3B). Adding exogenous FN (3 µg/cm²) along with dox at days 1–5 or days 5–7 did not significantly increase the percentage of cTnT⁺ cells when compared to the no dox control at days 1–5 and days 5–7, respectively (Figure 4C).

As FN knockdown at day 0–1 dramatically inhibited cardiac differentiation which could be rescued by the addition of exogenous FN, we first evaluated the expression of EMT genes that mark the initial transition of hPSCs to mesoderm over the first two days of differentiation. Quantitative RT-PCR was performed on the same H1 inducible FN knockdown clones at 0, 24, 36, and 48 hr after hPSCs differentiation was initiated (Figure 5A). The effect of dox-induced knockdown of FN1 transcripts was confirmed by the greater than 50% reduction in mRNA levels in the FN knockdown and FN knockdown +exogenous FN cell samples at 24 hr relative to the no dox control (Figure 5A). By 36 hr, FN1 transcripts recovered to the control level for FN knockdown condition or were significantly increased as in the FN knockdown +exogenous FN samples after dox was removed at 24 hr. By 48 hr, there were similar levels of FN1 transcripts in both the control and the FN rescue samples, but no cells survived in the FN knockdown group (Figure 5A). The key EMT transcription factors upregulated during gastrulation (Barrallo-Gimeno and Nieto, 2005; Nieto, 2002; Nieto et al., 2016; Thiery and Sleeman, 2006), SNAI1 and SNAI2, were examined. Quantitative RT-PCR showed SNAI1 expression increased significantly in both the FN knockdown and FN knockdown +exogenous FN samples compared to the control at 24 hr, and its expression continuously upregulated in the FN knockdown sample at 36 hr. By 48 hr, there were similar level of SNAI1 expression in both the control and the FN rescue samples; whereas, SNAI2 expression was not significantly different between the groups at 24 and 36 hr but a general increase in expression over this time window was observed (Figure 5A). VIM expression, similar to SNAI1 expression, was increased significantly in the FN knockdown cells by 36 hr compared to the control (Figure 5A), which is consistent with this mesenchymal marker and known target of SNAI1.

We next examined the mesendodermal/mesodermal progenitors generated in the initial differentiation stage of the H1 inducible FN knockdown clones by flow cytometry. The cell counts of the attached cells on days 0–5 of differentiation showed a great reduction of cell number at day 1 in all three groups (Figure 5B); however, the cells in the control and the FN rescue groups rapidly proliferated after day 1. In contrast, no cells survived in the FN knockdown group after day 2 (Figure 5B). Because Brachyury and Sox17 are both expressed in mesendodermal progenitors, we co-labeled the cells with Brachyury and Sox17 antibodies on days 0–5 and analyzed by flow cytometry. Brachyury⁺ cells started to emerge at day 1 in all three groups. By day 2, 97–98% of the cells were Brachyury⁺ in both the control and FN rescue groups, but there were no surviving cells in the FN knockdown samples (Figure 5C). The fraction of cells that were Brachyury⁺ rapidly decreased after day 2 in both the control and FN rescue groups, and by day 5 there were minimal Brachyury⁺ or Sox17⁺ cells present in both groups (Figure 5C). We performed the same experiment using the DF19-9-11T inducible FN knockdown clones and observed similar results (Figure 5—figure supplement 1). In contrast to the day 0–1 treatment with dox, a later timed dox pulse from days 1 to 5 did not alter the abundance of Brachyury⁺ cells in all three groups over the same time course of differentiation (Figure 5—figure supplement 2). Together these results suggest that knockdown of FN at day 0–1 does not stop the initiation of EMT upon addition of Activin A at d0 (Nieto et al., 2016; Thiery and Sleeman, 2006), but it prevents the generation and/or survival of Brachyury⁺ mesodermal progenitors. In contrast, shRNA knockdown of FN at later time points in the protocol did not impact the fate of the differentiating cells.

To investigate the mechanisms underlying FN’s essential role in the formation of Brachyury⁺ mesoderm, we evaluated integrins expressed in undifferentiated hPSCs and known to bind FN. Integrins are a family of heterodimeric transmembrane proteins composed of α and β subunits that interact with ECM. Of the 24 known heterodimeric integrin receptors, 13 have been shown to bind FN (Bachmann et al., 2019; Bharadwaj et al., 2017; Hynes, 2002; Ruoslahti, 1991; Wu et al., 1995). Review of RNA-seq data from undifferentiated hiPSCs shows expression of integrin subunits associated with FN binding including integrin α3, α4, α5, αV, β1, β5, and β8 (Zhang et al., 2019). Of these integrin subunits, knockout studies have implicated only α4, α5, αV, and β1 with various developmental defects impacting the heart (Hynes, 2002), so we focused our studies on these integrins. Integrin β1 shows the highest level of expression of all integrin subunits in hPSCs, and so we first tested blocking integrin β1 with a monoclonal antibody, P5D2, during differentiation. We added P5D2 at day –2 through day –1 (pluripotent stage) or day 0 through day 1 (mesoderm formation) in the matrix sandwich protocol when Matrigel overlays were applied (Figure 6A). Adding P5D2 at day –2 did not block cardiac differentiation; however, adding P5D2 at day 0 significantly inhibited cardiac differentiation as measured by flow cytometry of the cTnT⁺ cells using DF19-9-11T iPSCs and H1 ESCs (Figure 6B, Figure 6—figure supplement 1A). Furthermore, the P5D2 antibody showed concentration-dependent inhibition of hPSC-CM generation in DF19-9-11T iPSCs and H1 ESCs (Figure 6C, Figure 6—figure supplement 1B). We next tested antibody blocking (3 µg/ml) of relevant integrin α subunits including α5, αV, and α4. Adding the monoclonal antibodies P1D6 (anti-integrin α5), P3G8 (anti-integrin αV), or P4G9 (anti-integrin α4) at day –2 as shown in Figure 6A did not significantly impact hPSC-CMs differentiation as measured by flow cytometry of the cTnT⁺ cells, similar to the results blocking integrin β1 at day –2 (Figure 6D, Figure 6—figure supplement 1C). When the integrin α blocking antibodies were added on day 0, block of integrin α5 did not show inhibition of hPSC-CMs differentiation for both DF19-9-11T and H1 lines. Whereas block of integrin αV showed borderline inhibition and block of integrin α4 showed significant inhibition of hPSC cardiac differentiation in both DF19-9-11T and H1 lines (Figure 6D, Figure 6—figure supplement 1C). To probe the impact of blocking integrin α4 and integrin αV further, we tested a range of blocking antibody concentrations added at day 0 and found concentration-dependent inhibition of cardiac differentiation by P4G9 (anti-integrin α4), as well as significant inhibition by the highest concentration (5 µg/ml) of P3G8 (anti-integrin αV) (Figure 6E and F). Taken together, these results showed antibody blocking FN integrin receptors of β1, α5, αV, or α4 at the pluripotent stage of hPSCs (day –2) in the matrix sandwich protocol did not inhibit cardiac differentiation; however, blocking integrin β1, α4, or αV at day 0–1, when mesodermal progenitors are generated resulted in significant inhibition of hPSC-CM differentiation compared to the control for multiple hPSC lines. Thus, integrin α4β1 and αVβ1 heterodimers are likely key mediators of the FN effect on early differentiation stages.

Integrin-linked kinase (ILK) interacts with the cytoplasmic domain of β1 and β3 integrins and plays crucial roles in transducing signals from ECM components and growth factors to downstream signaling components (Delcommenne et al., 1998; Hannigan et al., 1996; Oloumi et al., 2004). The kinase activity of ILK is stimulated rapidly and transiently by the engagement of integrins to FN ECM, as well as by insulin via receptor tyrosine kinases (RTKs), in a PI3K dependent manner (Delcommenne et al., 1998). To determine if ILK signaling is involved in the initiation of cardiac differentiation, we tested the specific ILK inhibitor, cpd22 (Lee et al., 2011), in the matrix sandwich protocol. Cpd22 was added at day 0 at concentrations of 0, 1, 3, and 10 µM, and cells were treated for 24 hr in the matrix sandwich protocol as shown in Figure 7A. Cardiac differentiation was measured by flow cytometry of the cTnT⁺ cells at 15 days of differentiation. Cpd22 showed significant, concentration-dependent inhibition of cardiac differentiation in both DF19-9-11T iPSCs and H1 ESCs (Figure 7B, Figure 7—figure supplement 1).

We sought to confirm the role of ILK in the initiation of cardiac differentiation in a distinct hPSC-CM differentiation protocol based on biphasic modulation of Wnt signaling by small molecules (Lian et al., 2012; Lian et al., 2013). This protocol involves the activation of canonical Wnt signaling by inhibition of GSK3β to promote Brachyury⁺ mesoderm formation, followed by inhibition of Wnt to promote hPSC-CM differentiation (GiWi protocol). Cpd22 was added on day 0 at concentrations of 0, 1, 3, and 10 µM and cells were treated for 24 hr in the GiWi protocol as shown in Figure 7C. Cardiac differentiation was measured by flow cytometry of the cTnT⁺ cells at 15 days of differentiation. Cpd22 showed significant inhibition of cardiac differentiation in a concentration-dependent manner in the GiWi protocol (Figure 7D), similar to the matrix sandwich protocol (Figure 7B). Collectively these results show that ILK plays an important role in mesoderm formation and cardiac differentiation of hPSCs potentially acting downstream of integrin β1 and FN.

Because inhibition of ILK by cpd22 significantly inhibited cardiac differentiation in the matrix sandwich protocol, we next investigated apoptosis after 18 hr of cpd22 addition on day 0 in the matrix sandwich protocol. To identify apoptotic and necrotic cells, annexin V/ propidium iodide (PI) double staining was used in flow cytometry for both DF19-9-11T iPSCs and H1 ESCs. Annexin V has a high affinity for the anionic phospholipid phosphatidylserine (PS) specifically present in the outer leaflet (extracellular side) of the plasma membrane in apoptotic cells. PI is used to detect necrotic or late apoptotic cells, characterized by the loss of the integrity of the plasma and nuclear membranes. Cells were collected for flow cytometry at 18 hr of differentiation in the matrix sandwich protocol in the absence or presence of cpd22 at 1, 3, and 10 µM added on day 0 as shown in Figure 7A. The flow cytometry data plots can be divided in four regions corresponding to: 1. viable cells which are Annexin V^-/PI^- (Q4); 2. early apoptotic cells which are Annexin V⁺/PI^- (Q1); 3. late apoptotic cells which are Annexin V⁺/PI⁺ (Q2); and 4. necrotic cells which are Annexin V^-/PI⁺ (Q3) (Figure 8A, Figure 8—figure supplement 1A). Under control conditions for matrix sandwich protocol, a substantial fraction of cells were undergoing apoptosis by 18 hr for both DF19-9-11T (71.6%, total Annexin V⁺ cells, Figure 8A) and H1 (47.6%, Figure 8—figure supplement 1A) lines. This cell death is typical of monolayer-based cardiac differentiation protocols in response to cell signaling (Activin A) and media conditions that select for survival specifically of mesendodermal fated cells. However, addition of cpd22 resulted in a concentration-dependent increase in the total Annexin V⁺ cells (Figure 8B, Figure 8—figure supplement 1B). Furthermore, 10 µM cpd22 significantly increased the population of late apoptotic cells (Figure 8C, Figure 8—figure supplement 1C).

Because cpd22 also inhibited cardiac differentiation in the GiWi protocol, we tested if inhibition of ILK by cpd22 also promoted apoptosis in the first 24 hr of the GiWi protocol. Using the same apoptosis assay, cells were collected for flow cytometry after 18 hr of differentiation in the GiWi protocol in the absence or presence of cpd22 at 1, 3, and 10 µM as shown in Figure 7C. Similar to the matrix sandwich protocol, the GiWi control cells exhibited significant apoptosis (58.5%, total Annexin V⁺ cells of DF19-9-11T, Figure 8D) at 18 hr of differentiation. Adding cpd22 at 10 µM significantly increased the total Annexin V⁺ cells as well as the late apoptotic cells (Figure 8E and F). This finding of increased apoptosis in response to cpd22 for both the matrix sandwich and GiWi protocols correlates with the inhibition of cardiac differentiation by cpd22 in these protocols, and thus suggests that survival of the critical mesodermal population fated for cardiac differentiation requires ILK signaling.

To investigate how ILK regulates cells survival and differentiation, we examined regulation of AKT and GSK3β, known downstream targets of ILK signaling (Delcommenne et al., 1998; Legate et al., 2006). Activation of ILK can lead to phosphorylation of AKT on serine-473 which inhibits apoptosis and promotes cell survival. ILK activation by the ECM and/or growth factors in PI3K-dependent manner, can increase phosphorylation of GSK3β at serine-9 which inhibits GSK3β activity. Quantitative Western blots for AKT, pAKT (S473), GSK3β, and pGSK3β (S9) were performed on cells undergoing the matrix sandwich protocol in the absence or presence of the ILK inhibitor, cpd22 (10 µM), added at day 0 (Figure 9). The pAKT level (normalized to total protein) at 10, 30, and 120 min showed a significant decrease at 120 min by cpd22, but no significant differences at other time points (Figure 9A and B). At the 24 hr time point, an inadequate number of cells survived in the presence of cpd22 to provide a sample. The pGSK3β level (normalized to total protein) at the same time points in the protocol did not show any significant effect of cpd22 (Figure 9D and E). Treatment with cpd22 did not significantly change the levels of total AKT or total pGSK3 over the time course (Figure 9A, C, D and F). Together, these results are consistent with ILK inhibition reducing pAKT levels and thus leading to increased apoptosis and associated failure of cardiac differentiation.

Human ESC line H1 and iPSC line DF19-9-11T were used in this study. Human PSCs were cultured on Matrigel (GFR, BD Biosciences) coated 6-well plates in mTeSR1 medium. Cells were passaged using Versene solution (Gibco) every 4–5 days as previously described (Zhang et al., 2012).

For cardiac differentiation with the matrix sandwich protocol as shown in Figure 1A, the tested matrix proteins were overlaid on the hPSC culture after 2 (d-2) and 4 (d0) days of seeding when the monolayer of cells reached 80% and 100% confluence. Human PSCs were dissociated with 1 ml/well Versene solution at 37 °C for 5 min and seeded on Matrigel, LN111 (BioLamina, 1 μg/cm²), LN521 (BioLamina, 0.5 μg/cm²) and FN (Corning 354008, 2.5 μg/cm²) coated 12-well plates at the density of 6 × 10⁵ cells/well in mTeSR1 medium supplemented with 10 μM ROCK inhibitor (Y-27632, Tocris). The medium was changed daily. The amount of matrix proteins in overlay was: 8.7 µg/cm² Matrigel; LN111, LN521, FN and COL4 was 0.3 µg/cm². For cardiac differentiation with the monolayer based protocol, the matrix overlay was not added. The cells were fed with fresh mTeSR1 medium from d-3 to d0. At d0 the medium was changed to RPMI +B27 without insulin and supplemented with growth factors. The concentration of growth factors were: Activin A (R&D) at 100 ng/ml, BMP4 (R&D) at 2.5 ng/ml, and bFGF (Waisman Biomanufacturing, UW-Madison) at 10 ng/ml. The volume of medium was 0.8 ml at d0-d1 and 1.5 ml at d1-d5. The medium was changed to RPMI +B27 with insulin, and cells were fed every other day and cultured until d15 for flow cytometry analysis.

For cardiac differentiation with the GiWi protocol, human PSCs were dissociated with 1 ml/well Versene solution at 37 °C for 5 min, and seeded on Matrigel coated 12-well plates at the density of 6–10 × 10⁵ cells/well in mTeSR1 medium supplemented with 10 μM ROCK inhibitor (Y-27632). Cells were cultured in mTeSR1 medium with medium change daily until reached 100% confluence. At d0, the medium was changed to 1 mL RPMI +B27 without insulin and supplemented with 12 µM CHIR99021 (Tocris), and changed to 2 mL RPMI +B27 without insulin after 24 hr. Seventy-two hr after addition of CHIR99021, a combined medium was prepared by collecting 1 mL of medium from the wells and mixing with same volume of fresh RPMI +B27 without insulin medium and supplemented with 5 µM IWP2 (Tocris). The medium was changed to 2 mL RPMI +B27 without insulin on d5, and to RPMI +B27 with insulin starting from d7. Cells were fed every other day and cultured until d15 for flow cytometry analysis.

Human PSCs maintained in mTeSR1 medium were used for lentivirus transfection. The cloning strategy is shown schematically in Figure 4—figure supplement 4.

For imaging with an EVOS microscope (Life Technologies), hPSCs were cultured either in 6-well plates, or seeded and differentiated in 12-well plates, coated with Matrigel or the ECM proteins. For imaging with confocal microscope (Leica SP5II), cells were seeded or plated on glass coverslips coated with Matrigel or the ECM proteins. Cells were fixed with 4% paraformaldehyde for 15 min at room temperature for labeling with different markers. For intracellular markers, cells were permeabilized in 0.2% Triton X-100 (Sigma) for 1 hr at room temperature. For labeling the ECM proteins, cells were not permeabilized. After fixation and permeablization, samples were blocked with 5% non-fat dry milk (Bio-Rad) in 0.2% Triton X-100 solution and incubated for 2 hr at room temperature on a rotator followed by two washes with PBS. Primary antibodies (Supplementary file 1a) were added in 1% BSA in PBS solution with or without 0.1% Triton X-100 depending on the markers to label, and incubated overnight at 4 °C. Samples were washed with 0.2% Tween 20 in PBS twice and 1 X PBS twice. Secondary antibodies specific to the primary IgG isotype were diluted (1:1000) in the same solution as the primary antibodies and incubated at room temperature for 1.5 hours in dark on a rotator. Samples were washed with 0.2% Tween 20 in PBS twice and 1 X PBS twice. Nuclei were labeled with Hoechst or DAPI. Confocal images were analyzed with the Leica LAS AF Lite software.

Cells were detached from cell culture plates by incubation with 0.25% trypsin-EDTA (Invitrogen) plus 2% chick serum (Sigma) for 5 min at 37 °C. Cells were vortexed to disrupt the aggregates followed by neutralization by adding an equal volume of EB20 medium (Zhang et al., 2009). Cells were fixed in 1% paraformaldehyde in a 37 °C water bath for 10 min in the dark, permeabilized with ice-cold 90% methanol for 30 min on ice. Cells were washed once in FACS buffer (PBS without Ca/Mg²⁺, 0.5% BSA, 0.1% NaN₃) plus 0.1% Triton, centrifuged, and the supernatant was discarded leaving about 50 μl. See Supplementary file 1a for the primary antibodies information. For labeling cTnT, the primary antibody was diluted in 50 μl FACS buffer plus 0.1% Triton and added to each sample for a total sample volume of 100 μl. Samples were incubated with the primary antibodies overnight at 4 °C. For co-labeling of Brachyury and Sox17, the cells were incubated with the conjugated antibodies for half an hour at room temperature in dark. Cells were washed once with 3 ml FACS buffer plus 0.1% Triton and resuspended in 300–500 μl FACS buffer plus Triton for analysis. Please refer to Supplementary Material for details of the primary antibodies. For the secondary antibody labeling for cTnT, cells were washed once with 3 ml FACS buffer plus 0.1% Triton after the primary antibody labeling, centrifuged, and supernatant discarded leaving ~50 μl. Secondary antibody specific for the primary IgG isotype was diluted in FACS buffer plus Triton in a final sample volume of 100 μl at 1:1000 dilution. Samples were incubated for 30 min in the dark at room temperature, washed in FACS buffer plus Triton and resuspended in 300–500 μl FACS buffer plus Triton for analysis. Data were collected on a FACSCaliber flow cytometer (Beckton Dickinson) and analyzed using FlowJo.

The FN immuno-fluorescence images of d-3, d-2, d-1, d0 matrix sandwich culture, and d0 monolayer culture were evenly split into 4 squares with online image splitting tool (https://www.imgonline.com.ua/eng/cut-photo-into-pieces.php), and imported to ImageJ Fiji software. The control images without primary antibody labeling were also imported and analyzed in the same way by ImageJ Fiji software. The gray value was used to represent the intensity of fluorescence, ranging from 0 (black, low intensity) to 256 (white, high intensity). For mean and standard deviation of gray values, the entire area of each image was selected, and the values were retrieved by selecting Menu bar >Analyze > Measure. The mean gray value was generated from ImageJ by summing each pixel’s gray value divided by the number of pixels in the selected area. The final mean fluorescence value was obtained by subtracting the mean gray value of the control image from FN immuno-labeled images. The mean of fluorescence intensity of each quartiles from each biological replicate in each group were represented in the box plots using Origin v9.

Cell samples were collected using 0.25% trypsin-EDTA (Invitrogen) to remove the cells from cell culture plates. Total RNA was purified using QIAGEN RNeasy Mini kit. Possible genomic DNA contamination was removed by RNase-Free DNase Set (QIAGEN) with the RNeasy columns, or by DNase I (Invitrogen) treatment for 15 min at room temperature. A total of 500 ng of total RNA was used for Oligo(dT)₂₀ – primed reverse transcription using SuperScript III First-Strand Synthesis System (Invitrogen). Quantitative RT-PCR was performed using Taqman PCR Master Mix and Gene Expression Assays (Applied Biosystems, Supplementary file 1b) in triplicate for each sample and each gene. A total of 0.5 μl of cDNA from RT reaction was added as template for each Q-PCR reaction. The expression of genes of interest was normalized to that of GAPDH. RT-PCR was carried out using Platinum Taq DNA Polymerase (Invitrogen) or Gotaq Master Mix (Promega) and then subjected to 2% agarose gel electrophoresis. PCR conditions included denaturation at 94 °C for 30 s, annealing at 60 °C for 30 s, and extension at 72 °C for 1 min, for 35 cycles, with 72 °C extension for 7 min at the end. ACTB (β-actin) was used as an endogenous control.

Monoclonal antibodies in supernatant without sodium azide were used for the blocking experiments. Blocking antibodies, P5D2 (anti- human integrin β1), P1D6 (anti-human integrin α5), P3G8 (anti-human integrin αV) and P4G9 (anti-human integrin α4) (all from Developmental Studies Hybridoma Bank), and ILK inhibitor cpd22 (EMD Millipore) were diluted in the media to make the final concentrations and added to the culture at the indicated time points.

Cell pellets were re-suspended in modified RIPA buffer (50 μl per 10⁶ cells) containing 50 mM Tris-HCl pH 8, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA pH 8, 1 mM PMSF, and 1 X Protease/Phosphatase Inhibitor Cocktail (Cell Signaling Technology, cat# 5872) by pipetting solution up and down several times and incubated 30 min at 4 °C mixing by end-over-end rotation. Lysates were sonicated using a Fisherbrand Model 120 Sonic Dismembrator (4×5 s bursts at 30% amplitude with 10 s between bursts) and centrifuged (16,100×g, 20 min, 4 °C) to clarify. The supernatants were recovered and protein content was determined using Bio-Rad Protein Assay Dye Reagent Concentrate (cat# 5000006) following the manufacturer’s instructions.

To assess changes in AKT and GSK3β phosphorylation, as well as changes in the expression of ILK, AKT, and GSK3β, 25 μg of cell lysate from each condition was resolved by SDS-PAGE. Proteins were transferred to Amersham Hybond P Western blotting membrane (0.2 μm pore size, PVDF, Sigma-Aldrich, cat# GE10600021) at 10 V overnight in a cold room (4 °C). After transfer, total protein loading was assessed by staining with Revert 700 Total Protein Stain (LI-COR, cat# 926–11010) in adherence with the manufacturer’s instructions. To block, membranes were incubated in TBST with 10% (w/v) nonfat dry milk for 1 hr at room temperature followed by incubation in primary antibody solution containing either 5% (w/v) BSA (AKT-Ser473 and GSK3β-Ser9) or nonfat dry milk (all other antibodies) and diluted primary antibodies in TBST overnight at 4 °C (see Supplementary file 1c for primary antibody information). The membranes were removed from primary antibody solution and washed 5×5 min with TBST followed by incubation in TBST with 3% (w/v) nonfat dry milk and 1:1,000 diluted secondary antibodies for 1 hr at room temperature. Subsequently, membranes were washed 5×5 min with TBST, overlayed with Amersham ECL Prime Western Blotting Detection Reagent (Sigma-Aldrich, cat# GERPN2236), and imaged using an Odyssey XF Imaging System (LI-COR).

Membranes were probed first with phospho-specific antibodies, stripped using Restore Western Blot Stripping Buffer (Thermo Scientific, cat# 21059) in accordance with the manufacturer’s instructions, and reprobed with pan-protein antibodies. Band densities were quantified using Image Studio Lite Quantification Software (Ver 5.2, LI-COR). The phospho-specific and pan-protein signals were normalized to total protein. Data are from three biological replicates with one or more technical replicates, and are presented as mean ± SEM.

Data are presented as mean ± SEM. Technical and biological replicates are as indicated for each dataset. For datasets with normal distributions, statistical significance was determined by Student’s t-test for two groups or one-way ANOVA for multiple groups with post-hoc test using Bonferroni method. Statistical analysis was performed using Origin, v9, p<0.05 was considered statistically significant.