Cardiac differentiation of human pluripotent stem cells using defined extracellular matrix proteins reveals essential role of fibronectin

  1. Jianhua Zhang
  2. Zachery R Gregorich
  3. Ran Tao
  4. Gina C Kim
  5. Pratik A Lalit
  6. Juliana L Carvalho
  7. Yogananda Markandeya
  8. Deane F Mosher
  9. Sean P Palecek
  10. Timothy J Kamp

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This is an important paper describing the work to develop defined surface coatings for cardiac cell differentiation. The described experiments are convincingly supporting the critical role of fibronectin (FN) in the formation of precardiac mesoderm. The exploration of FN as necessary for pre-cardiac mesoderm formation was explored using various other ECM conditions, endogenous FN knock-out, blocking antibodies against integrin subunits, and inhibition of ILK via small molecule cdp22. In the case of the FN knock-out, experiments included rescue conditions that established a causal link between FN and the formation of precardiac mesoderm. Particularly insightful was the tracking of FN deposition over time with or without exogenously provided FN (i.e., the LN-111 case). This work is of interest to developmental biologists, stem cell biologists, and engineers as they work to optimize defined matrices for differentiation and manufacturing protocols of all cell types including cardiomyocytes.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Research and therapeutic applications using human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) require robust differentiation strategies. Efforts to improve hPSC-CM differentiation have largely overlooked the role of extracellular matrix (ECM). The present study investigates the ability of defined ECM proteins to promote hPSC cardiac differentiation. Fibronectin (FN), laminin-111, and laminin-521 enabled hPSCs to attach and expand. However, only addition of FN promoted cardiac differentiation in response to growth factors Activin A, BMP4, and bFGF in contrast to the inhibition produced by laminin-111 or laminin-521. hPSCs in culture produced endogenous FN which accumulated in the ECM to a critical level necessary for effective cardiac differentiation. Inducible shRNA knockdown of FN prevented Brachyury + mesoderm formation and subsequent hPSC-CM generation. Antibodies blocking FN binding integrins α4β1 or αVβ1, but not α5β1, inhibited cardiac differentiation. Furthermore, inhibition of integrin-linked kinase led to a decrease in phosphorylated AKT, which was associated with increased apoptosis and inhibition of cardiac differentiation. These results provide new insights into defined matrices for culture of hPSCs that enable production of FN-enriched ECM which is essential for mesoderm formation and efficient cardiac differentiation.

Article activity feed

  1. Version published to 10.7554/elife.69028 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    This is a wonderful paper from the Kamp lab describing the work to develop defined surface coatings for cardiac cell differentiation. Kamp lab developed Matrigel overlay protocol for cardiac differentiation that is now widely used and adapted by many. Recent advances in directed cardiac differentiation resulted in a number of defined cytokine protocols that significantly advanced the field and made the cells accessible to many labs. The work on defined ECM preparation was less extensive, thus this paper contributes important new knowledge. The described experiments are convincingly supporting the role of fibronectin in early cardiac mesoderm induction, in particular the siRNA knock-down studies. I only have couple of questions that can improve this paper further.

    We are delighted that the reviewer found the manuscript of interest and importance examing defined surface coatings.

    1. In addition to direct signaling through cell surface receptors, in vivo ECM can sequester and then slowly release growth factors. It is know that cardiac differentiation cultures are sensitive to endogenously secreted signals. Besides direct signaling, is it possible that Fb is particularly suitable for sequestering and release of these factors in the context of early mesoderm induction and cardiac differentiation?

    Thank you for this important comment. Yes, it is possible and likely that fibronectin impacts growth factor signaling by sequestering and releasing these factors. We now specifically address this question in the Discussion highlighting relevant literature. Lines 470-479 state, “Furthermore, FN can provide a reservoir for growth factors as well as serve as a coactivator with growth factors (Hynes, 2009). For example, FN through its second heparin binding domain (FN III12-14) binds a variety of growth factors in the PDGF/VEGF, FGF, and TGF-β families (Martino and Hubbell, 2010; Wijelath et al., 2006). FN can also interact with binding proteins for growth factors including IGF-binding protein-3 and latent TGF-β binding protein (Dallas et al., 2005; Gui and Murphy, 2001). FN and growth factors can be costimulatory at cellular adhesions (Miyamoto et al., 1996). Although the precise role of FN-associated growth factor signaling during early development is little understood, a study in Xenopus did demonstrate that FN bound PDGF-AA provides guidance in mesenoderm migration during gastrulation (Smith et al., 2009).” This discussion highlights the multiple possible interacting signaling molecules with FN that potentially contribut to the cardiac differentiation process, but mechanistically dissecting out the role of these interactions is beyond the scope of this manuscript.

    1. The experiments to delineate the role of integrin beta1 and ILK signaling in mediating effects of fibronectin coating are conceptually sound and well executed. However, beta1 integrins and ILK signaling are implicated in cell survival pathways. It is well know that cardiac differentiation cultures are cell density sensitive. Is it possible that blocking of integrin beta1/ILK results in lower cell viability that translates into lower cell density, ultimately resulting in the outcome of the cardiac differentiation?

    We agree with the reviewer that the monolayer hPSC-CM differentiation protocols are cell density sensitive. Cell density is typically optimized for the confluence of hPSCs at the initiation of differentiation (day 0 in our manuscript) which was the starting point for testing integrin subunit and ILK block. How cell density impacts differentiation after day 0 as the cells begin to undergo the epithelial-to-mesenchymal transition is not clear to us from the existing literature as this is difficult to experimentally manipulate independent of the cell density at day 0. There is a dramatic change in cell viability and density during this early time window (see Figure 5B). Nevertheless, we agree that a decrease in cell viability or increased apoptosis in the presence of the blocking interventions may be a critical mechanistic feature. For that reason, we performed additional experiments examining cell survival using annexin V labeling to identify apoptotic cells and PI for cell viability in the presence and absence of cpd22 (see new Figure 8). The results show that cpd22 addition on day 0 induces a concentration-dependent increase in apoptosis after 18 hr on top of the high background level of apoptosis in both the matrix sandwich and GiWi protocols. Although the increase in apoptosis by cdp22 correlates with the inhibition of cardiac differentation in this time window by cpd22, it does not clarify whether the impact on differentation is due to a reduction in density of the surviving cells or apoptosis of the essential population of cells forming mesoderm. Regardless, the new data confirm the reviewer’s suspicion that changes cell density after day 0 can help explain the impact of blocking integrin β1 and ILK signaling. Lines 358-362 introduce these new data, “Because inhibition of ILK by cpd22 significantly inhibited cardiac differentiation in the Matrix Sandwich protocol, we next investigated apoptosis after 18 hours 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…”

    Reviewer #3 (Public Review):

    This paper is clearly written and represents a significant contribution to stem cell biology as it links fibronectin (FN) accumulation to mesoderm differentiation of PSCs. The exploration of FN as necessary for pre-cardiac mesoderm formation was explored using various other ECM conditions, endogenous FN knock-out, blocking antibodies against integrin subunits, and inhibition of ILK via small molecule cdp22. In the case of the FN knock-out, experiments included rescue conditions that established a causal link between FN and the formation of precardiac mesoderm. Particularly insightful was the tracking of FN deposition over time with or without exogenously provided FN (i.e., the LN-111 case).

    We appreciate the reviewers positive assessment of the study and presentation.

    There were several major weaknesses in the study. First, many of the studies were conducted with a single hiPSC line. Some studies were conducted with an hESC line, but not the most critical experiments including demonstration of a lack of FN at early time points when cultured on LN-111. These two lines with all critical experiments are required at a minimum; inclusion of multiple lines of each (ESC and iPSC) is suggested.

    We agree with the reviewer that confirming key findings in multiple cell lines is critical to demonstrate robust findings. We tested hiPSC line DF19-9-11T and hESC line H1 for almost every experiment, but given space constraints much of the confirmatory data were presented as supplemental figures. We apologize that the text did not always clearly state when multiple cell lines were used, but we have revised where necessary to indicate this. In addition, we have updated some figure legends to clearly state which cell lines were used. In addition, we have provided additional confirmatory data for our integrin antibody blocking experiments in both lines. Here is a summary of the experiments presenting data for experiments done in parallel in 19-9-11 and H1 and our revisions to refer to both sets of data more clearly:

    a) Test of defined ECM proteins (LN111, LN521 and FN) in Figure 1B, C for 19-9-11 and Figure 1 - figure supplement 1, 3 for H1. Line 134-135 revised to state, “…we tested overlay of defined ECM proteins in the matrix sandwich protocol using DF19-9-11T iPSCs and H1 ESCs.”

    b) Endogenous FN production on LN111 for DF19-9-11T iPSCs (Figure 2C) and H1 ESCs (Figure 2- figure supplement 1 ). Line 168-171, “…no detectable FN ECM at day -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).” The time course is more abbreviated in Figure 2 – figure supplement 1 as our experience showed minimal detectable FN under any conditions on day -3 and -2. Therefore we focused on day -1 and day 0 for the supplemental experiments in H1 which confirmed the increase in FN from day -1 to day 0.

    c) FN1 shRNA knockdown clones were generated in both H1 ESCs and DF19-9-11T iPSCs (Figure 4 – figure supplement 2, 4).

    d) FN knockdown studies were conducted in multiple clones from both 19-9-11 and H1 transgenic lines. FN knockdown and exogenous FN rescue experiments were carried out in multiple clones of H1 as shown in Figure 4 and Figure 4 – figure supplement 3. With the evidence of inhibition of cardiac differentiation by knockdown of FN, gene expression and flow cytometry for Brachyury+ cells were examined in the same H1 knockdown clones (Figure 5). Furthermore, flow cytometry evaluation of Brachyury+ cells was tested in the DF19-9-11T FN knockdown clones as well (Figure 5 – figure supplement 1, 2).

    e) The integrin antibody blocking for β1, α5, αV (added in revision), and α4 (added in revision) were tested in both 19-9-11 (Figure 6) and H1 (Figure 6 – figure supplement 1). Lines 304-318 “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)... and block of integrin α4 showed significant inhibition of hPSC cardiac differentiation (Figure 6D, Figure 6 – figure supplement 1C).”

    f) The ILK inhibitor, cpd22, was tested using 19-9-11 line in Matrix Sandwich protocol (Figure 7A, B) and GiWi protocol (Figure 7C, D), and using H1 line in MS protocol (Figure 7 – figure supplement 1).

    g) Newly added in the revision, the effect of ILK inhibition by cpd22 on apoptosis was examined in both 19-9-11 (Figure 8A-C) and H1 (Figure 8 – figure supplement 1) in the Matrix Sandwich protocol, as well as in the GiWi protocol using 19-9-11 (Figure 8D-F).

    Second, it is not surprising that blocking of beta1 integrin inhibits cardiomyocyte differentiation. Beta 1 integrin subunit is necessary for engagement of most ECM proteins and therefore downstream outcomes can in no way be linked directly to FN. An opportunity is missed to identify the heterotrimer necessary for the differentiation outcome observed. In addition, it is likely that many cells underwent anoikis in the presence of the antibody making relative quantification meaningless. Further, a rescue condition is not included.

    We greatly appreciate this comment and have done additional experiments to address it. In the revised manuscript, we test a series of blocking antibodies to integrin α subunits in addition to α5 that are known to bind FN as a heterodimer with integrin β1. In the revised manuscript, we now state (lines 294-300), “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 iPSCs 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 β1with various developmental defects impacting the heart (Hynes, 2002), so we focused our studies on these integrins…” In short, we found that blocking integrin α4 or αV inhibits cardiac differentiation (revised Figure 6D, Figure 6 – figure supplement 1). Thus we conclude on lines 326 and 327, “…integrin α4β1 and αVβ1 heterodimers are likely key mediators of the FN effect on early differentation stages.”

    Third, the studies with cpd22 are weak. There is no small-molecule control, no direct knockdown to a void off-target effects of the small molecule, and no indication of whether the effect is linked to GSK3b or PI3K or both. With the identification of the critical integrin heterodimer(s) above, it would be more compelling to block these and look at downstream phosphorylation of ILK and other potential downstream signaling players.

    We appreciate the reviewers’ important mechanistic questions. Indeed many questions remain regarding downstream signaling. Because there are no published and available chemical analogues to use as a small molecular control, we were unable to perform this experiment. We also were unable to generate an effective ILK inducible knockdown PSC line. However, we did focus our attention on signaling pathways known to be downstream of ILK including GSK3β and AKT looking at both total protein and phosphorylation of these kinases at residues correlated with activity pGSK3β (Ser9) and pAKT (Ser473). Our results show that treatment of Day 0 differentiating cells with cpd22 lead to a statistically significant reduction in pAKT at 2 hr relative to control, but there were no significant changes in pGSK3β induced by cpd22 treatment (new Figure 9). The reduction in pAKT correlates with an increase in apoptosis observed in response to cpd22 treatment (Figure 8). Although we did make progress identifying α4β1 and αVβ1 as the likely critical integrin heterodimers, we were unable to perform blocking antibody experiment time courses with the series of blocking integrin antibodies due limitations in available reagents and timely revision during the pandemic.

    Finally, there was a missed opportunity for a thorough investigation of the ECM present in each condition via mass spectrometry or another proteomics approach. The only ECM that was specifically probed overtime was FN and with some limited analysis of LN.

    Thank you for this comment, and we agree that we have not done a complete characterization of the dynamic changes in ECM at various stages of differentiation. There is certainly a great deal of biology to be yet.

  3. eLife

    Evaluation Summary:

    This is an important paper describing the work to develop defined surface coatings for cardiac cell differentiation. The described experiments are convincingly supporting the critical role of fibronectin (FN) in the formation of precardiac mesoderm. The exploration of FN as necessary for pre-cardiac mesoderm formation was explored using various other ECM conditions, endogenous FN knock-out, blocking antibodies against integrin subunits, and inhibition of ILK via small molecule cdp22. In the case of the FN knock-out, experiments included rescue conditions that established a causal link between FN and the formation of precardiac mesoderm. Particularly insightful was the tracking of FN deposition over time with or without exogenously provided FN (i.e., the LN-111 case). This work is of interest to developmental biologists, stem cell biologists, and engineers as they work to optimize defined matrices for differentiation and manufacturing protocols of all cell types including cardiomyocytes.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    This is a wonderful paper from the Kamp lab describing the work to develop defined surface coatings for cardiac cell differentiation. Kamp lab developed Matrigel overlay protocol for cardiac differentiation that is now widely used and adapted by many. Recent advances in directed cardiac differentiation resulted in a number of defined cytokine protocols that significantly advanced the field and made the cells accessible to many labs. The work on defined ECM preparation was less extensive, thus this paper contributes important new knowledge. The described experiments are convincingly supporting the role of fibronectin in early cardiac mesoderm induction, in particular the siRNA knock-down studies. I only have couple of questions that can improve this paper further.

    1. In addition to direct signaling through cell surface receptors, in vivo ECM can sequester and then slowly release growth factors. It is know that cardiac differentiation cultures are sensitive to endogenously secreted signals. Besides direct signaling, is it possible that Fb is particularly suitable for sequestering and release of these factors in the context of early mesoderm induction and cardiac differentiation?
    2. The experiments to delineate the role of integrin beta1 and ILK signaling in mediating effects of fibronectin coating are conceptually sound and well executed. However, beta1 integrins and ILK signaling are implicated in cell survival pathways. It is well know that cardiac differentiation cultures are cell density sensitive. Is it possible that blocking of integrin beta1/ILK results in lower cell viability that translates into lower cell density, ultimately resulting in the outcome of the cardiac differentiation?

  5. eLife

    Reviewer #2 (Public Review):

    This paper is clearly written and represents a significant contribution to stem cell biology as it links fibronectin (FN) accumulation to mesoderm differentiation of PSCs. The exploration of FN as necessary for pre-cardiac mesoderm formation was explored using various other ECM conditions, endogenous FN knock-out, blocking antibodies against integrin subunits, and inhibition of ILK via small molecule cdp22. In the case of the FN knock-out, experiments included rescue conditions that established a causal link between FN and the formation of precardiac mesoderm. Particularly insightful was the tracking of FN deposition over time with or without exogenously provided FN (i.e., the LN-111 case).

    There were several major weaknesses in the study. First, many of the studies were conducted with a single hiPSC line. Some studies were conducted with an hESC line, but not the most critical experiments including demonstration of a lack of FN at early time points when cultured on LN-111. These two lines with all critical experiments are required at a minimum; inclusion of multiple lines of each (ESC and iPSC) is suggested.

    Second, it is not surprising that blocking of beta1 integrin inhibits cardiomyocyte differentiation. Beta 1 integrin subunit is necessary for engagement of most ECM proteins and therefore downstream outcomes can in no way be linked directly to FN. An opportunity is missed to identify the heterotrimer necessary for the differentiation outcome observed. In addition, it is likely that many cells underwent anoikis in the presence of the antibody making relative quantification meaningless. Further, a rescue condition is not included.

    Third, the studies with cpd22 are weak. There is no small-molecule control, no direct knockdown to avoid off-target effects of the small molecule, and no indication of whether the effect is linked to GSK3b or PI3K or both. With the identification of the critical integrin heterodimer(s) above, it would be more compelling to block these and look at downstream phosphorylation of ILK and other potential downstream signaling players.

    Finally, there was a missed opportunity for a thorough investigation of the ECM present in each condition via mass spectrometry or another proteomics approach. The only ECM that was specifically probed overtime was FN and with some limited analysis of LN.

  6. Version published to 10.1101/2021.04.09.439173v1 on bioRxiv