Morphogen-driven human iPSCs differentiation in 3D in vitro models of gastrulation is precluded by physical confinement

  1. Haneen S. Alsehli
  2. Errin Roy
  3. Thomas Williams
  4. Alicja Kuziola
  5. Yunzhe Guo
  6. Jeremy Green
  7. Eileen Gentleman
  8. Davide Danovi

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

In early human development, gastrulation is tightly associated with lineage specification. The interplay between mechanical forces and biochemical signals during these processes is poorly understood. Here, we dissect the effects of biochemical cues and physical confinement on a 3D in vitro model of gastrulation that uses spheroids formed from human induced pluripotent stem cells (hiPSCs). First, we compare self-renewing versus differentiating media conditions in free-floating cultures, and observe the emergence of organised tri-germ layers. In these unconfined cultures, BMP4 exposure induces polarised expression of SOX17 in conjunction with spheroid elongation. We then physically confine spheroids using PEG-peptide hydrogels and observe dramatically reduced SOX17 expression, albeit rescued if gels that soften over time are used instead. Our study combines high-content imaging, synthetic hydrogels and hiPSCs-derived models of early development to define the drivers causing changes in shape and emergence of germ layers.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    The authors do not wish to provide a response at this time.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    The authors used hIPSCs to generate spheroids capable of elongation when cultured in KSR and exposed to BMP4. After examining all three developmental lineages of these treatments, the spheroids were embedded in different hydrogels to restrict movement and observed how this affected elongation and lineage differentiation. This manuscript was difficult to review because of the the style of the writing, the sometimes confused narrative of results, the problematic statistics and questionable interpretation of the literature. Authors should actively address the issues presented below with a significant overhaul of the text and presentation.

    Overall comments

    Major concerns - The body of the manuscript should be re-written to offer a coherent narrative that delivers a clear and condensed message . In fact, it was quite surprising given the well written and concise format of the introduction. However, there is some confusion about the interpretation of the first reference (https://doi.org/10.1038/s41586-020-2383-9) in which it was already presented that BMP4 treatment did not induce elongation in a hIPSC-derived gastruloid model. This is not to say that there is anything wrong with repeated experiments and different results, but the results from this reference were interpreted as if BMP4 treatment had induced elongation. The use of the methodology present in this manuscript as opposed to the validated model of using Wnt agonism to induce elongation is not convincingly justified. The results themselves have potential if reworded and condensed but in their current format they are not convincing. Statistics have substantial problems. According to the methodology, only one-way ANOVAs were used, when many of the circumstances would call for t-tests (if parametric) or at least mention the normality tests used to justify the ANOVAs. There is no clear mention of the number of independent differentiations and only one cell line was used. As a result of this, and the state of the statistics and interpretation of the data, the conclusions presented in this paper are inconclusive or misrepresented. The article should be re-written in a coherent scientific style, including not writing in the first person/active voice, and addressing the grammatical errors mentioned below.

    Minor concerns - Throughout the methods there are issues with inconsistent notation, e.g. acronyms aren't always in brackets (HBSS, PBS, KSR), units are sometimes spaced from the value (10ng/mL or 10 µg/mL), units are incorrectly written (u instead of µ). There are also typographical errors such as basic grammar (capital letters and full stops), double spacing, typos, strange sentence construction, as well as interchangeable use of commas and semi colons when listing antibodies. These issues were not confined to the methods, but were particularly noticeable there, and so the whole manuscript requires thorough proofreading. Positive comments - The introduction was well written and concise, the images were mostly clear and easy to interpret, illustrating a result even if not directly the one described.

    Specific Comments

    Introduction - Comments

    The introduction was well written and easy to follow. However, certain aspects should be refined in order to highlight the argument as to why the mechanical forces and environment are of importance to lineage determination in hIPSCs. For example, the paragraph describing 2D micropattern cultures could be reduced or integrated into the introduction of 3D gastruloid models. In its place, further exploration of examples of mechanical force's impact on stem cell/embryonic differentiation, both in vitro and in vivo, would be beneficial (perhaps including the geometric shape aspect mentioned in the 2D micropattern paragraph here instead). This would emphasise the necessity for investigation of environmental forces on 3D in vitro differentiation, tying it back to the first paragraph's broader developmental questions. Currently, it lacks the reiteration of the overarching purpose of this investigation.

    Materials and Methods - Comments

    hIPSC culture - The paper suggests that only one hIPSC line was used for this research; considering the variability innate to each stem cell line and their differentiation capabilities, at least two cell lines should be used. Please reference that stem cell validation was conducted by HipSci. It is concerning that antibiotics were used for tissue culture on such a short protocol and should not be used in future. Similarly, were these hIPSCs tested regularly for mycoplasma? Reference briefly the method in which hIPSCs were passaged and what range of passage number was used.

    Spheroids derivation from hIPSCs - There is no mention of hIPSC confluency upon differentiation. Based on the brief 4 minutes of Accutase exposure, colonies would either be small to achieve single cell dissociation (which is not in the best interest of hIPSC colonies) or that single cell dissociation was not achieved. Seeding density is also close to double seeding density for currently circulating gastruloid protocols (see https://doi.org/10.1242/dev.150391) with no explanation as to why this cell count was chosen. Make mention of whether or not they were fed during the initial 48 hours. There is also no reference to how many independent differentiations were executed and therefore replicability is of significant concern considering the use of an unvalidated protocol.

    Gastrulation-like induction of 3D hIPSCs - Difficult to follow timeline of treatments without reference to Figure 1 and therefore requires a rewrite for clarity. In addition, media 3 and 4 are formatted differently from 1 and 2, making it more confusing. Do not rely on Figure 1 to explain the treatment course, it should be clear in the methodology.

    Live Imaging - No need to state each media, particularly if it is out of numerical order as described previously. Can be written as "Live hIPSC spheroids were imaged for 96 hours....". Also include magnification and time frame of each hour, e.g. between 48 and 96 hours. Immunofluorescence staining - Other than the grammatical errors referenced above, as there are numerous in this paragraph in particular, the catalogue number for each antibody should also be supplied. Considering the number of antibodies, perhaps this would be best supplied as a supplemental table. No mention either here or previously as to how non-encapsulated spheroids were mounted and if cleared.

    Results - Comments

    As mentioned previously, the results are written in a style which is not sufficiently rigorous for a scholarly article and should therefore be edited to reflect the standard expected quality. BMP4 signalling induces axial elongation in 3D gastrulation-like models -

    Figure 1 - Nice and clean diagram but unnecessarily large, could be shrunk and placed at the top of Figure 2. Also is lacking key methodological details e.g. seeding density, reference to media changes in time line, etc.

    An example of a statement that lacks rigor and specificity is the following: "We used defined medium conditions that provided consistent shape variation in spheroid morphology. E8 medium promotes self-renewing conditions, while KSR BMP4 triggers differentiation [31,32]". Self-renewing conditions of what? Differentiation of what? Into what cells?

    These statements are too vague and remain without connection to the research This kind of issues are unfortunately recurrent throughout the manuscript. They are easily rectifiable with concise text rather than expecting the reader to search through references. "Consistent with our previous results..." - not referenced. Mention of size is brief in text and instead description of results is found in Figure 2 legend.

    Figure 2 - Although Figure 2.A and 2.B are clear images, Figure 2.B is unnecessary and doesn't add anything to the message. Figure 2.C: The statistics for this table are hard to believe, except between E8 and KSR BMP4, due to the spread of the data. Change references to "size" to "sphericity", as size would arguably be better investigated using area as a metric, again be precise in choice of language. There isn't even a description of what Figure 2.D is portraying, nor error bars or notation of statistical testing.

    Figure 3 - Figure 3.A Writing on scale bars is too small to be useful and unnecessary if mentioned in the legend. The images highlight the issue with the size/sphericity issue, although the images themselves are reasonably clear and will highlighted (if slightly overexposed and not cleared). The title of this figure is also misleading - E8 BMP4 is not emulating gastruloid development, and the gastruloid-like entity of KSR BMP4 has yet to be validated. A more appropriate title would be "Exposure to BMP4 in spheroid and elongated spheroid culture increases SOX17 expression". There is no mention of how cells were counted in methods and considering the variable fluorescence observed in Figure 3.A OCT4 for E8, this could easily be misconstrued. Inappropriate dictation of p value, should only be reporting the alpha. Much like the similar issue with sphericity, the error bars in Figure 3.E in particular make it difficult to believe the statistical significance achieved. Overall it seems inappropriate to focus solely on endodermal lineage and leave ectoderm and mesoderm to supplementals, when classically axial elongation in gastruloids is punctuated not just by SOX17 but also by BRACH expression throughout the extended region. This suggests that the authors may not have sufficiently evaluated the literature referenced. The justification provided by the authors for choosing SOX17 reinforces this fact by declaring that SOX17 is expressed during early development, as if SOX2 and BRACH isn't! Most concerningly, claims of SOX2's absence in E8 media spheroids as if to be a positive should actually be worrisome, considering SOX2 is a Yamanaka factor. Arguably, without further validation, this result undermines the foundation of this work.

    PEG-peptide hydrogel encapsulation disturbs SOX17 patterning - The interchangeable use of "control" and E8 makes for difficult reading. Although no further substantial issues were taken with Figure 4 that had not already been addressed (scale bars, overexposure), the final comment of this section is simply inaccurate: "Overall, this suggests that embedding hiPSCs in a confining environment blocks morphogenesis, and despite the addition of BMP4, this was not sufficient to enable SOX17 expression.". It does not define that the hIPSCs in this case are spheroids and determines this assumption based off data in Figure 6.B and Figure 6.C which could have been easily integrated into Figure 4. Rather than comparing non-deg PEG to deg-PEG, in order to make this statement, it is necessary to compare the non-embedded spheroids to the embedded spheroids. There is also reference to a Figure 3.C. that does not exist.

    Modulating PEG-peptide degradability promotes SOX17 expression - This section had similar issues to the one above, predominantly flitting between Figure 5 and 6 when it could have been condensed. The microscope images are also particularly poor, with high background in the 488 channel, overexposure for Hoechst and incomplete light penetrance in the centre for both E8 and E8 BMP4. Despite the lack of methodological explanation of how counting was conducted, these aspects are still likely to have had significant impact on cell count regardless of how the count was conducted. The last comment of this section is again questionable, as there is no definitive comparison as to what the deg-PEG is improving upon with regards to SOX17 expression.

    Biochemical cues regulate morphogenesis via cell proliferation and cellular tension - There is no explanation as to why the manuscript pivots away from hydrogels to look at proliferation and F-actin. It does not appear to benefit the overarching goal of this manuscript (i.e. investigating the impact of confinement on 3D hIPSC-derived elongation models), with proliferation investigations feeling like an afterthought and F-actin investigations being placed about 3 sections too late. It was surprising to see EdU to check cell proliferation in this manner; similar can and should be achieved by using KI67 staining. This was a significantly missed opportunity to do pulse-chase experiments to understand potential cell cycle changes, and suggests the experimental design was not thought through. However, it is more reasonable to examine the F-actin network as shown by the manuscript. Although the pipeline in Figure 7.D is an interesting take on quantifying F-actin staining, no statistical analysis was undertaken to confirm a link between conditions and F-actin, nor F-actin and SOX17 expression, despite what this manuscript suggests. This should be coupled with information prior to the introduction to hydrogels, and then re-evaluated in those confined to hydrogels.

    Overall these results need to be substantially edited for clarity, rigor and completeness. The narrative of the results is lost half way through, the statistics are questionable or non-existent and the logic behind certain experiments are debatable.

    Discussion - Comments

    To consider the models generated in this manuscript as gastruloids is misleading and inaccurate and should be replaced with something that reflects what was actually generated, e.g., BMP4-induced elongated spheroids. A quick search of key gastruloid papers, such as https://doi.org/10.1242/dev.113001, indicates how far detached the spheroids examined in this paper are, only sharing similar polarised SOX17 expression. In fact, the above referenced paper already examined aspects of classic gastruloid differentiation when exposed to BMP4, albeit using mESCs, and indicated that elongation only occurred in a proportion of the population. Similarly in the first reference which used hIPSCs specifically, https://doi.org/10.1038/s41586-020-2383-9, it was reported that BMP4 treatment did not induce elongation. At no point in this manuscript were these results discussed, which is concerning. Instead, this manuscript claims to have "robustly" concluded many of the results which have been questioned in detail above, with little reference to the variability across multiple differentiations or adequate statistical analysis. Noticeably they have also not mentioned their results on proliferation, which again makes the question of its relevance to the paper's aims.

    Significance

    The authors used hIPSCs to generate spheroids capable of elongation when cultured in KSR and exposed to BMP4. After examining all three developmental lineages of these treatments, the spheroids were embedded in different hydrogels to restrict movement and observed how this affected elongation and lineage differentiation. This manuscript was difficult to review because of the the style of the writing, the sometimes confused narrative of results, the problematic statistics and questionable interpretation of the literature. Authors should actively address the issues presented here with a significant overhaul of the text and presentation.

    Our expertise is in stem cell biology, 2D and 3D model systems, microscopy, and single cell analysis.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Within the manuscript "Morphogen-driven human iPSCs differentiation in 3D in vitro models of gastrulation is precluded by physical confinement" by Alsehli et al, the authors aim to dissect the effects of biochemical cues and physical confinement using gastruloid models. Using this model, the authors find some indications that physical confinement prevents morphological changes and reduces lineage commitment of cells (as assayed by Sox 17).

    While the manuscript addresses an interesting and important question, I do not think that the setup and data as present in the manuscript provide conclusive insight in any of the questions studied. First of all, the authors use a very different system (BMP4 induction) as compared to mainstream conditions for gastruloid formation. Based on the images and movies within the manuscript, symmetry breaking and elongation seem to be very sub-optimal under the conditions used by the authors, which makes any further functional assaying highly challenging. Importantly, it remains unclear why the authors did not (at least) include mainstream conditions (Chiron pulse) within their assays for comparative purposes, I feel this is a missed chance. On top of that, KSR is known to be poorly compatible with gastrulation and further development, at least in mouse (https://pubmed.ncbi.nlm.nih.gov/35988542/), so it remains unclear why the authors include a KSR condition.

    Referees cross-commenting

    Like me, both other reviewers indicated (major) shortcomings of the current paper, and indicated potential directions for improvement. Both these other reviews are very balanced as well. Altogether, there are significant changes required for the manuscript, and it depends on the authors which direction they would like to take this manuscript. Hence, I do not know what else to conclude than that in the current format there too many shortcomings to be of added value to current literature (similar to reviewer 3, and in line w/ reviewer 1 but who is somewhat more mild in their evaluation). If the authors decide to move on with this story, this paper requires a re-evaluation of a much improved version.

    Significance

    Secondly, I am not convinced by the data quality and representation in the manuscript. As outlined above, symmetry breaking and elongation seem to be very sub-optimal. Also, Fig 4b and Fig 5b hardly show any visible SOX17, in particular also not in the deg-PEG E8 BMP4 condition. Yet, the authors claim high expression of SOX17 in this condition in Fig 6C, as well as large and significant differences between conditions. To my feeling, Fig 6C does not represent the Sox17 observations of the IF. Similarly, I am not convinced by the EdU staining in Fig 7A. Importantly, the authors base their conclusions on only a single cell line, very few spheroid of this cell line, and an unknown number of biological experimental replicated (which to me seems single experiments). Altogether, the setup and data lacks reliability and robustness.

    Currently, the efficiency of gastruloid formation is a important discussion in the field. Therefore, it is important to report on this in the current manuscript.

    Altogether, while the manuscript makes a few interesting observations, it is very preliminary and not-well worked out or validated. Without being more robust and conclusive, I do not think the manuscript is of much added value to current literature.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary: This manuscript developed a BMP-based culturing protocol for gastruloid-like tissue that eexpresses markers from ectoderm, mesoderm, and endoderm. Through modulating the mechanical properties of extracellular environments, the expression of endodermal marker, SOX17, was found to correlate with the mechanical properties of the extracellular environment. However, the data presented only constitute a crude characterization of the cell lineages within the tissue and, therefore, the biological relevance of the gastruloid-like tissue, together with the associated discoveries, is not immediately evident.

    Major comments

    1. Extra explanation should be provided to justify why mechanical property of extracellular environments is a relevant/important factor for either early embryo development or gastruloid technology.
    2. (OPTIONAL) To demonstrate the impact of mechanical property of extracellular environments, the authors compared two different gels, a stiffer gel with non-significant stress relaxation, and a softer gel with non-trivial stress relaxation. However, such comparison can be somewhat non-intuitive since a variety of mechanical properties are modulated simultaneously between two groups, including both tissue stiffness and its time-dependency. Since it has been well established that both elasticity and viscoelasticity can affect cell behavior, respectively, I would suggest the authors consider adding an extra comparison between a stiff elastic gel and a soft elastic gel.

    Minor comments

    1. The author provided individual staining of SOX2, BRA and SOX17, it would be more helpful to co-stain these markers to demonstrate their relative spatial distribution.
    2. The quantification method for Fig. 7D&E seems missing from the method section, and more explanation for the method would be helpful. What is the area ratio between the inner and outer ROI? What does the y-axis number mean in Fig.7E?
    3. In the last 2nd paragraph of "Biochemical cues regulate morphogenesis via cell proliferation and cellular tension", it was mentioned that "spheroids cultured in E8 medium exhibited organised and tightly packed F-actin with a homogenous network orientation", yet in the next paragraph, it was claimed that "F-actin orientation in round spheroids in E8 medium is mostly distributed in the inner core". Aren't these statements contradictory? Also in this paragraph, it was mentioned "Whereas in KSR conditions, F-actin was distributed in the periphery without significant differences compared to KSR medium conditions", I assume the authors intended to say "KSR BMP4" here?
    4. The authors mentioned that "cells in the KSR BMP4 spontaneously elongate mirroring A-P elongation", yet based on the data presented, the elongation orientation appears more consistent with the D-V embryo axis?

    Significance

    This manuscript demonstrates a rudimentary investigation towards how mechanical properties of the extracellular environment may affect the marker expression within a gastruloid-like tissue. Such information may provide certain useful knowledge for the field of gastruloid. However, compared with the established gastruloid models in the field, the advantage of the tissue model developed in this study isn't very clear. Further, the physiological relevance of extracellular mechanics (and particularly, matrix viscoelasticity, which is studied in this manuscript) is not immediately evident.

  5. Version published to 10.1101/2023.03.29.534685v1 on bioRxiv