Vinculin recruitment to α-catenin halts the differentiation and maturation of enterocyte progenitors to maintain homeostasis of the Drosophila intestine
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This study examines a potential mechanosensation mechanism in fly intestinal stem cells and their terminal enteroblast progeny. The manuscript’s data clearly demonstrate a role for vinc in suppressing the proliferation of midgut stem cells and the differentiation of their terminal enteroblast progeny and suggest that this role is exerted specifically through enteroblast vinc. The authors find that similar phenotypes are induced by genetic manipulations of vinc, a-cat, and myosin, and they argue that this similarity implies that vinc activity in enteroblasts is mechanosensitive. These findings are potentially relevant to biologists interested in stem cells, tissue homeostasis, fate decisions, and mechanobiology. Initial studies of vinc null flies failed to reveal any essential functions in development or viability, so the report of an adult-specific phenotype in the intestine is notable. However, the current manuscript falls short of demonstrating a key pillar of its model – that enteroblast vinc is regulated by mechanical tension. In addition, some important experiments using either whole-animal mutants or cell-specific manipulations leave room for alternate interpretations.
(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.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Mechanisms communicating changes in tissue stiffness and size are particularly relevant in the intestine because it is subject to constant mechanical stresses caused by peristalsis of its variable content. Using the Drosophila intestinal epithelium, we investigate the role of vinculin, one of the best characterised mechanoeffectors, which functions in both cadherin and integrin adhesion complexes. We discovered that vinculin regulates cell fate decisions, by preventing precocious activation and differentiation of intestinal progenitors into absorptive cells. It achieves this in concert with α-catenin at sites of cadherin adhesion, rather than as part of integrin function. Following asymmetric division of the stem cell into a stem cell and an enteroblast (EB), the two cells initially remain connected by adherens junctions, where vinculin is required, only on the EB side, to maintain the EB in a quiescent state and inhibit further divisions of the stem cell. By manipulating cell tension, we show that vinculin recruitment to adherens junction regulates EB activation and numbers. Consequently, removing vinculin results in an enlarged gut with improved resistance to starvation. Thus, mechanical regulation at the contact between stem cells and their progeny is used to control tissue cell number.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
Bohère, Eldridge-Thomas and Kolahgar have studied the effect of mechanical signalling in tissue homeostasis in vivo, genetically manipulating the well known mechano-transductor vinculin in the adult Drosophila intestine. They find that loss of vinculin leads to accelerated, impaired differentiation of the enteroblast, the committed precursor of mature enterocytes, and stimulates the proliferation of the intestinal stem cell. This leads to an enlarged intestinal epithelium. They discriminate that this effect is mediated through its interaction with alpha-catenin and the reinforcement of the adherens junctions, rather than with talin and integrin-mediated interaction with the basal membrane. This results aligns well, as the authors note, with previous observations from Choi, Lucchetta and …
Author Response
Reviewer #1 (Public Review):
Bohère, Eldridge-Thomas and Kolahgar have studied the effect of mechanical signalling in tissue homeostasis in vivo, genetically manipulating the well known mechano-transductor vinculin in the adult Drosophila intestine. They find that loss of vinculin leads to accelerated, impaired differentiation of the enteroblast, the committed precursor of mature enterocytes, and stimulates the proliferation of the intestinal stem cell. This leads to an enlarged intestinal epithelium. They discriminate that this effect is mediated through its interaction with alpha-catenin and the reinforcement of the adherens junctions, rather than with talin and integrin-mediated interaction with the basal membrane. This results aligns well, as the authors note, with previous observations from Choi, Lucchetta and Ohlstein (2011) doi:10.1073/pnas.1109348108. Bohère et al then explore the impact that disrupting mechano-transduction has on the overall fitness of the adult fly, and find that vinculin mutant adult flies recover faster after starvation than wild types.
The main conclusions of the paper are convincing and informative. Some important results would benefit from a more detailed description of the phenotypes, and others could have alternative explanations that would warrant some additional clarification.
- Interpretation of phenotypes in vinc[102.1] mutants
The paper presents several adult phenotypes of the homozygous viable, zygotic null mutant vinculin[102.1], where the fly gut is enlarged (at least in the R4/5 region). In many cases, they correlate this phenotype with that of RNAi knockdown of vinculin in the gut induced in adult stages. This is a perfectly valid approach, but it presents the difficulty of interpretation that the zygotic mutant has lacked vinculin throughout development and in every fly tissue, including the visceral mesoderm that wraps the intestinal epithelium and that also seems enlarged in the vinc[102.1] mutant. So this phenotype, and others reported, could arise from tissue interactions. To me, the quickest way to eliminate this problem would be to express vinculin in ISCs and/or EBs the vinc[102.1] background, either throughout development or after pupariation or emergence, and observe a rescue.
We agree with the reviewer that we cannot exclude additional vinculin role(s) in other tissues during or after development that might have an impact on the intestinal epithelium. Our attempts to express a full-length Vinculin construct (Maartens et al, 2016) in the vinc102.1 flies, either in adulthood or throughout development, were not very conclusive: although we observed some degree of rescue, it was not fully penetrant. This was in contrast to the complete rescue observed with the genomic rescue of vinculin. Thus, it is possible that some form of tissue interaction contributes to the phenotype observed, for example if vinculin loss affects muscle structure. Alternatively, just like it was shown that too much active vinculin is detrimental to the fly (Maartens et al, 2016), our experiment suggests that too much vinculin may be deleterious to the intestine.
In any case, because of cell-specific knockdowns in the adult gut, we are confident that EB reduction of vinculin levels or activity is sufficient to accelerate tissue turnover, at least in a specific portion of the posterior midgut. We have amended the text to acknowledge a role for tissue interactions (see page 6 (end of first paragraph), page 7 (start of last paragraph), page 12 (starvation experiments).
An experiment where this is particularly difficult is with the starvation/refeeding experiment. The authors explored whether the disruption of tissue homeostasis, as a result of vinculin loss, matters to the fly. So they tested whether flies would be sensitive to starvation/re-feeding, where cellular density changes and vinculin mechano-sensing properties may be necessary. They correctly conclude that mutant flies are more resistant to starvation, and suggest that this may be due to the fact that intestines are larger and therefore more resilient. However, in these animals vinculin is absent in all tissues. It is equally likely that the resistance to starvation was due to the effect of Vinculin in the fat body, ovary, brain, or other adult tissues singly or in combination. The fact that the intestine recovers transiently to a size slightly larger than that of the fed flies seems anecdotal, considering the noise within the timeline of fed controls. I am not sure this experiment is needed in the paper at all, but to me, the healthy conclusion from this effort is that more work is needed to determine the impact of vinculin-mediated intestinal homeostasis in stress resistance, and that this is out of the scope of this paper.
Please the new data presented in Figure 8A-B (text page 12).
- Cell autonomy of the requirement of Vinculin and alpha-Catenin
Authors interpret that Vinculin is needed in the EB to maintain mechanical contact with the ISC, restrict ISC proliferation through contact inhibition, and maintain the EB quiescent. This interpretation explains seemingly well the lack of an obvious phenotype when knocking down vinculin in ISCs only, while knockdown in ISCs and EBs, or EBs only, does lead to differentiation problems. It also sits well with the additional observation that vinculin knockdown in mature ECs does not have an obvious phenotype. However, a close examination makes the results difficult to explain with this interpretation only. If the authors were correct, one would expect that in mutant clones, eventually, vinculin-deficient EBs will be produced, which should mis-differentiate and induce additional ISC proliferation. However, the clones only show a reduction in ISC proportions; the most straight forward interpretation of this is that vinculin is cell-autonomously necessary for ISC maintenance (which is at odds with the phenotype of vinculin knockdown using the ISC and ISC/EB drivers).
We apologise that we were unclear in the text. With hindsight, the confusion may have been caused by our describing the phenotype of MARCM clones before reporting the accumulation of EBs in the vinc102.1 guts. Therefore, we swapped these two sections and improved the description of these experiments in the manuscript (see section: “The pool of enterocyte progenitors expands upon vinculin depletion” pages 6-8).
In brief, we do not think that our results are at odds with the phenotype of vinculin knockdown using the ISC and ISC/EB drivers - we realise the text was misleading and hope to have clarified our observations in the revised manuscript (pages 7 and 8). From cell conditional RNAi experiments, like the reviewer, we would predict that vinculin knockdown or loss of function in mitotic clones (MARCM experiments, Figure 4E-G) will induce accelerated differentiation of vinculin deficient enteroblasts, which in turn will increase proliferation. We observed that vinc102.1 or vinc RNAi mitotic clones contained similar number of cells compared to control clones, but reduced proportion of stem cells (Figure 4G). We interpret this as indicating that to maintain an equivalent clone size, stem cells must have divided more frequently, with some divisions producing two differentiated daughter cells. This type of symmetric division would increase the EB pool (as seen in Figure 4-figure supplement 2B), at the expense of the ISC population, in turn decreasing long term clonal growth potential. Altogether, the results obtained with MARCM clones highlight changes in tissue dynamics compatible with those observed with cell-specific vinculin knockdowns.
Also, from the authors interpretation, it would follow induce that the phenotype of vinculin knockdown in ISCs+EBs and in EBs only should be the same. However, in ISCs+EBs vinculin knockdown, differentiation accelerates, which is likely accompanied by increased proliferation (judging by the increase in GFP area, PH3 staining would be more definitive).
Indeed, the accelerated differentiation observed with esgGal4>UAS VincRNAi is accompanied by increased proliferation with the two independent RNAi lines used. We have added this result in Figure 1-figure supplement 1G (and in text, page 5).
This contrasts with the knockdown only in EBs, which leads to accumulation of EBs due to misdifferentiation, and increased proliferation, mostly of ISCs, as measured directly with PH3 staining, but not additional late EBs or mature ECs. The authors call this "incomplete maturation due to accelerated differentiation". I think that one should expect to find incomplete differentiation/maturation when the rate of the process is very slow, not the other way around. To me, these are different phenotypes, which could perhaps be explained if vinculin was also needed in the ISC to transmit tension to the EB and prevent its differentiation, and removing it only in the EB may be revealing an additional, cell-autonomous requirement in maturation.
When vinculin is knocked down in EBs, cells appear bigger than controls (as judged by the RFP+ nuclei in Figure 5E). This, compared to yw and vinc102.1 guts shown in Figure 4D suggests that these cells are more advanced in their differentiation. We have removed the sentence, to not confuse the reader, and clarified the text (see page 8). The discrepancy in the differentiation index between the esgGal4 and KluGal4 experiments might result from differences in the drivers, or an additional role of vinculin in EC differentiation, which we now mention in the text (page 8).
So far, we have no evidence to support the idea that vinculin is also needed in the ISC to transmit tension to the EB and prevent its differentiation; for example, the lack of any phenotype when we knocked down vinculin specifically in ISCs (Figure 3) – notably, no increase in ISC ratio and no increase in cell density (unlike the reduction seen in Figure 1F with ISC+EB Knockdown).
Another unexpected result, considering the authors interpretation, is that the over expression of activated Vinculin (vinc[CO]) does not seem to have much of an effect. It does not change the phenotype of the wild type (where there is very little basal turnover to begin with) and it only partially rescues the phenotype of the vinc[102.1] mutants, when the rescue transgene vinc:RFP does. This again suggests that there may be tissue interactions, in development or adulthood, that may explain the vinc[102.1] phenotypes. It could also be that this incomplete rescue is due to the deleterious effect of Vinc[CO]; this is another reason for doing the vinc[102.1]; esg-Gal4; UAS-vincFL experiments suggested above). An alternative experiment to perform this rescue would be to knock down vinculin gene while overexpressing the Vinc[CO] transgene - this may be possible with the RNAi HSM02356, which targets the vinculin 3'UTR and is unlikely to affect UAS-vinc[CO].
Please refer to essential point 2c; as VincCO is not a simple overactive protein, like a constitutively active kinase, additional effects in the tissue can be expected.
The claims of the authors would be more solid if the reporting of the phenotypes was more homogeneous, so one could establish comparisons. Sometimes conditions are analysed by differentiation index, others by extension of the GFP domains, others with phospho-histone-3 (PH3), others through nuclear size or density, and combinations. I do not think the authors should evaluate all these phenotypes in all conditions, but evaluating mitotic index and abundance of EBs and "activated EBs/early ECs" to monitor proliferation and differentiation rates should be done across the board (ISC, ISC+EB, EB drivers).
To improve consistency, in all conditions we have compared cell types ratios and cellular density upon vinculin knockdown: see Figure 1E-F for ISC+EB, Figure 3B-C for ISC, and Figure 5 – figure supplement 1C-E for EB (with panel E newly added). As we did not observe any effect on ratio or density, we did not monitor cell proliferation for ISC knockdown.
Nonetheless, we added the mitotic index for the ISC+EB driver (new Figure 1- figure supplement 1G) to be consistent with the results from the EB driver (Figure 5- figure supplement 1C).
If the primary role of Vinculin is to induce contact inhibition in the ISC from the EB and prevent the EB differentiation and proliferation, one would expect that over expression of Vinc[CO] (or perhaps VincFL or sqhDD) in EBs should prevent or delay the differentiation and proliferation induced by a presumably orthogonal factor, like infection with Pseudomonas entomophila or Erwinia carotovora.
This is indeed an exciting prediction, but outside the scope of this manuscript.
- Relationship between Vinculin and alpha-Catenin
The authors establish a very clear difference in the phenotypes between focal adhesion components and Vinculin, whereas the similarity of alpha-catenin and vinculin knockdowns is very compelling. Therefore I am sure the authors are in the right path with their interpretation of this part of the paper. However, some of the alpha-Catenin experiments are not very clear. The result from the rescue experiment of alpha-Cat knockdown with alpha-Cat-deltaM1b does not seem to show what the authors claim, and differentiation does not seem affected, only the amount of extant older ECs (which may be due to other reasons as this is a non-autonomous effect).
Like the reviewer, we were surprised about the milder rescue with M1b compared to M1a and are unsure of the reasons for this. Nevertheless, quantifications of the differentiation and retention indices show significant differences for M1a and M1b compared to the FL control (Figure 6F-G), with phenotypes resembling the vinc knockdown. In Figure 6E, we have added a row of zoomed views to better highlight the similarity of phenotype between M1a and M1b and have acknowledged the mild differences in the text (bottom of page 9). For the sake of rigour, we think it is important to include results from both M1 deletions, even if there is not yet a logical reason to explain why they have different effects.
Ulrich Tepass produced a UAS-alpha-catenin construct with the full deletion of the M1 region, perhaps that would show a clearer phenotype.
This is a good suggestion, however for technical reasons this is not possible. The strategy devised by Ken Irvine and his group relies on rescuing the RNAi with an RNAi resistant construct, which is not the case for the constructs generated in the Tepass lab. Furthermore, we cannot adopt a MARCM strategy as -cat is too close to the centromere (80F).
Also, the autonomy of the phenotype is difficult to address with these experiments alone. It would be expected that the phenotype of alpha-catenin knockdown should be similar to that of vinculin knockdown in the ISCs only or EBs only.
This is not what our understanding of cadherin-mediated adhesion would predict. Forming cadherin adhesions requires cadherins and catenins in both cells, so we would expect similar phenotypes in ISCs only and EBs only. What is exciting about our findings is that the mechanosensitive machinery is not equally important in the two adherent cells, i.e. the EB is using vinculin to measure force on the contact and regulate differentiation, whereas the ISC needs to resist that force, but does not use vinculin to sense that force and regulate its behaviour.
We have added new data showing the role of the vinculin/α-catenin interaction in ISCs or EBs by co-expressing α-Cat RNAi and α-Cat ΔM1a. We observed that absence of VBS in α-catenin has no effect in ISCs but promotes EB differentiation and increase in numbers (new Figure 6 – figure supplement 2), similar to our observations with vincRNAi (see text page 10).
Reviewer #2 (Public Review):
Vinculin functions as an important structural bridge that connects cadherin and integrin-mediated adhesions to the F-actin cytoskeleton. This manuscript carefully examined the mutant phenotype of vinc in the Drosophila intestine and found that vinc mutant in EBs causes significant increases of EB to EC differentiation, stem proliferation, and tissue growth. By analyzing the mutant phenotype of the cadherin adaptor alpha-catenin, the authors suggest that the vinc functions through the cell-cell junctions instead of cell-CEM adhesions in EBs. Finally, manipulation of myosin activity in EBs phenocopies the vinc mutant, suggesting that vinculin is regulated by the mechanical tension transduced through the cytoskeleton.
The authors claim that the vinculin mutant phenotype is opposite compared to the loss of the major integrin components, suggesting a function independent of the cell-ECM adhesions. However, the phenotype of vinc and integrin may not be completely opposite. Besides loss of ISCs, both mys and talin knockdown in ISCs clearly causes ISCs differentiation into EC cells (Fig.3A), suggesting a possible involvement of integrin in EB to EC differentiation. Therefore, it will be important to test the phenotype of integrin KD in EBs using EB-specific Gal4.
The reviewer raised an important point. To test this we had to overcome the ISC defect of mys or talin RNAi, and specifically tested their function in enteroblasts using the KluGal4 driver. This revealed a similar phenotype of accelerated differentiation, assayed with the ReDDM system (see new Figure 6 -figure supplement 4). Thus, as the reviewer suggested both integrins and cadherins function in this process, we have amended the text to indicate this (see page 10, and sentence in the discussion page 12). It appears however that, unlike vinculin, they also have a key role in ISCs.
The authors proposed a model that the cell-cell adhesion between ISC and EBs is required for vinculin mediated differentiation suppression. However, this model is not directly supported by the data as the EB-ISC adhesion and EB-EC adhesion have not been tested separately.
This is an important point and we have amended the text to address this.
We have focussed our model on EB-ISC adhesion as the adherens junctions are stronger between progenitor cells than EBs-ECs, and because of previous data from the Ohlstein lab (Choi et al, 2011) demonstrating the relationship between adherens junction stability and EB differentiation/ISC proliferation. Nonetheless we agree it is possible that EB-EC adhesion might contribute to this mechanism and have modified the last sentence of the result section (page 12) and the legend associated to the model (Figure 8) to take this into account.
In addition, previous short-term manipulation of E-cadherin in ISCs and EBs shows no change in cell proliferation (Liang J. et al. 2017), which seems to contradict the authors' model. To support the authors' conclusion, long-term manipulation of E-cadherin in ISCs and EBs must be tested.
A main feature of the vinculin phenotype is the regional accelerated differentiation observed in R4/5, potentially reflecting areas more subject to mechanical forces. Strikingly, this accelerated differentiation is rarely observed more anteriorly (such as region R4a/b studied in Liang et al, 2017). In fact, these regional differences were previously reported with E-cadherin knockdown by the Adachi-Yamada group (see Figure S1, Maeda et al, 2008). This highlights the importance of considering regional control of cell fate for the field.
To test our hypothesis further, we have knocked down E-cadherin and α-catenin in EBs only (with Klu-Gal4). As shown in new Figure 6-figure supplement 3, we observed an accumulation of EBs as early as 3 days after induction, reminiscent of vinculin loss of function phenotype. Longer E-cadherin EB knock-down with KluGal4 appears particularly detrimental for survival as all flies died after 4 days of continuous RNAi expression preventing any further observations (see new text page 10). These observations support our model that junctional stability slows down EB differentiation. Our results are also in agreement with the work described in Choi et al (2011), whereby after 6 days of E-Cadherin RNAi expression in progenitors or EBs (using a different driver from us, Su(H)Gal4), the mitotic index increases, showing a feedback regulation on ISC proliferation. Therefore, our work and the Liang et al 2017 study are not in fact contradictory: the differences in the contribution of junctions to tissue dynamics might reflect the variety of molecular mechanisms involved along the small intestine.
The result of MARCM analysis seems inconsistent with the rest of the data. In MARCM, no significant change of clone sizes is observed between WT and vinc mutant (Fig. 3E). However, vinc mutant in EBs clearly promotes ISC proliferation in other experiments such as esg>vinc-RNAi and the EB>vinc-RNAi (Fig. 1A, Fig. 4).
Please refer to point 2a, essential revisions. We do not think that our results are at odds with the phenotype of vinculin knockdown using the ISC and ISC/EB drivers - we realise the text was misleading and hope to have clarified our observations in the revised manuscript (pages 7-8).
In Fig. 4H, the authors suggest that vinculin mutant prevent terminal EC formation. However, this may be simply caused by longer retention of Klu expression in the newborn ECs. To test if EB differentiation is indeed affected, the EC marker pdm1 staining will provide more convincing evidence. Another experiment to strengthen the conclusion will be the tracking of clone sizes generated from a single EB cell using the UAS-Flp system (such as G-trace).
These are good suggestions to strengthen our findings. Unfortunately, we have not managed to obtain a working Pdm1 antibody (or other commercially available EC marker), which is why we assayed nuclear size and the tracking of KluReDDM cells. Therefore, we have not been able to test if Klu is retained in newborn ECs.
As we agree this section of the text was misleading, we have rephrased and highlighted that the phenotype seen with KluGal4ReDDM resembles the accumulation of activated EBs and newborn ECs observed in vinc102.1 guts. (page 8).
In Fig. 6D, the survival rate of WT and vinc mutant flies were compared. However, as there is no additional assay about the feeding behavior or metabolic rate, the systematic mutant of vinc does not provide a direct link between animal survival and intestinal EBs. Therefore, an experiment with vinc level specifically manipulated in fly intestine using esg>vinc-RNAi or BE>vinc-RNAi will be more relevant.
This experiment has now been added in Figure 8B and the text modified to acknowledge the limitations of the survival experiments with whole mutant flies (see point 3, essential revisions above).
Reviewer #3 (Public Review):
Prior work had identified essential roles for Integrin signaling in regulating intestinal stem cell (ISC) proliferation, and the authors studies were motivated by trying to understand whether Vinculin (Vinc) might participate in this. However, Vinc is involved in mechanotransduction at both focal adhesions (FA) and adherens junctions (AJ), and their results revealed that Vinc phenotypes do not match those of FA proteins like Integrin. Conversely, they do match a-catenin (a-cat) RNAi phenotypes, and together with the localization of Vinc and the phenotypes associated with a-cat mutants that can't bind Vinc, this led to the conclusion that Vinc is acting at AJ rather than FA in this tissue. The results here are convincing, with clear presentation, nice images, and appropriate quantitation. It's also worth emphasizing that initial characterization of Vinc mutant flies failed to reveal any essential roles for this protein in Drosophila, so finding a mutant phenotype of any sort is significant.
While the manuscript is strong as a descriptive report on the requirement for Vinc in the Drosophila intestine, it doesn't provide us with much understanding of the mechanism by which Vinc exerts its effects, nor how its requirement is linked to intestinal physiology.
There is always more to learn, and the importance of our work so far is that it demonstrates a very specific role for vinculin as a mechanoeffector in regulating cell fate decisions in specific regions of the midgut, and provide the foundation for future work addressing the detailed mechanism of this function and physiological role.
Prior work has shown that mechanical stretching of intestines stimulates ISC proliferation (presumably through Integrin signaling), which is opposite to what Vinc does here.
We would like to stress that very little mechanistic knowledge is available regarding how mechanical stretching stimulates ISC proliferation, in Drosophila or mammalian systems. To our knowledge, the only work linking gut mechanical stretching to cell fate decisions in Drosophila identified Msn/Hippo pathway (Li et al., 2018) and the ion channel Piezo requirement (He et al., 2018). We agree with the reviewer that integrin signaling would most likely contribute, especially given the composition of gels for organoid cultures (Gjorevski et al, 2016), yet the actual molecular mechanisms remain to be elucidated.
There is a suggestion that Vinc is involved in maintaining homeostasis, but how its regulated remains a bit murky. The authors report that reductions in myosin activity result in phenotypes reminscent of Vinc phenotypes, which they interpret as supporting a model where Vinc's role is to help maintain tension at AJ. Of course it could also be reversed - maybe they are similar because tension is needed to maintain Vinc recruitment to AJ? They lack of epistasis tests and lack of analysis of whether Vinc localization to AJ in EBs is affected by tension or the M2 deletion of a-cat leaves us uncertain as to the actual basis for the relationship between Vinc and myosin phenotypes.
Thank you for all these suggestions. New experiments have been done to test the relationship between cellular tension and vinculin at junctions (see essential point 1).
-
Evaluation Summary:
This study examines a potential mechanosensation mechanism in fly intestinal stem cells and their terminal enteroblast progeny. The manuscript’s data clearly demonstrate a role for vinc in suppressing the proliferation of midgut stem cells and the differentiation of their terminal enteroblast progeny and suggest that this role is exerted specifically through enteroblast vinc. The authors find that similar phenotypes are induced by genetic manipulations of vinc, a-cat, and myosin, and they argue that this similarity implies that vinc activity in enteroblasts is mechanosensitive. These findings are potentially relevant to biologists interested in stem cells, tissue homeostasis, fate decisions, and mechanobiology. Initial studies of vinc null flies failed to reveal any essential functions in development or viability, so …
Evaluation Summary:
This study examines a potential mechanosensation mechanism in fly intestinal stem cells and their terminal enteroblast progeny. The manuscript’s data clearly demonstrate a role for vinc in suppressing the proliferation of midgut stem cells and the differentiation of their terminal enteroblast progeny and suggest that this role is exerted specifically through enteroblast vinc. The authors find that similar phenotypes are induced by genetic manipulations of vinc, a-cat, and myosin, and they argue that this similarity implies that vinc activity in enteroblasts is mechanosensitive. These findings are potentially relevant to biologists interested in stem cells, tissue homeostasis, fate decisions, and mechanobiology. Initial studies of vinc null flies failed to reveal any essential functions in development or viability, so the report of an adult-specific phenotype in the intestine is notable. However, the current manuscript falls short of demonstrating a key pillar of its model – that enteroblast vinc is regulated by mechanical tension. In addition, some important experiments using either whole-animal mutants or cell-specific manipulations leave room for alternate interpretations.
(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.)
-
Reviewer #1 (Public Review):
Bohère, Eldridge-Thomas and Kolahgar have studied the effect of mechanical signalling in tissue homeostasis in vivo, genetically manipulating the well known mechano-transductor vinculin in the adult Drosophila intestine. They find that loss of vinculin leads to accelerated, impaired differentiation of the enteroblast, the committed precursor of mature enterocytes, and stimulates the proliferation of the intestinal stem cell. This leads to an enlarged intestinal epithelium. They discriminate that this effect is mediated through its interaction with alpha-catenin and the reinforcement of the adherens junctions, rather than with talin and integrin-mediated interaction with the basal membrane. This results aligns well, as the authors note, with previous observations from Choi, Lucchetta and Ohlstein (2011) …
Reviewer #1 (Public Review):
Bohère, Eldridge-Thomas and Kolahgar have studied the effect of mechanical signalling in tissue homeostasis in vivo, genetically manipulating the well known mechano-transductor vinculin in the adult Drosophila intestine. They find that loss of vinculin leads to accelerated, impaired differentiation of the enteroblast, the committed precursor of mature enterocytes, and stimulates the proliferation of the intestinal stem cell. This leads to an enlarged intestinal epithelium. They discriminate that this effect is mediated through its interaction with alpha-catenin and the reinforcement of the adherens junctions, rather than with talin and integrin-mediated interaction with the basal membrane. This results aligns well, as the authors note, with previous observations from Choi, Lucchetta and Ohlstein (2011) doi:10.1073/pnas.1109348108. Bohère et al then explore the impact that disrupting mechano-transduction has on the overall fitness of the adult fly, and find that vinculin mutant adult flies recover faster after starvation than wild types.
The main conclusions of the paper are convincing and informative. Some important results would benefit from a more detailed description of the phenotypes, and others could have alternative explanations that would warrant some additional clarification.
- Interpretation of phenotypes in vinc[102.1] mutants
The paper presents several adult phenotypes of the homozygous viable, zygotic null mutant vinculin[102.1], where the fly gut is enlarged (at least in the R4/5 region). In many cases, they correlate this phenotype with that of RNAi knockdown of vinculin in the gut induced in adult stages. This is a perfectly valid approach, but it presents the difficulty of interpretation that the zygotic mutant has lacked vinculin throughout development and in every fly tissue, including the visceral mesoderm that wraps the intestinal epithelium and that also seems enlarged in the vinc[102.1] mutant. So this phenotype, and others reported, could arise from tissue interactions. To me, the quickest way to eliminate this problem would be to express vinculin in ISCs and/or EBs the vinc[102.1] background, either throughout development or after pupariation or emergence, and observe a rescue.
An experiment where this is particularly difficult is with the starvation/refeeding experiment. The authors explored whether the disruption of tissue homeostasis, as a result of vinculin loss, matters to the fly. So they tested whether flies would be sensitive to starvation/re-feeding, where cellular density changes and vinculin mechano-sensing properties may be necessary. They correctly conclude that mutant flies are more resistant to starvation, and suggest that this may be due to the fact that intestines are larger and therefore more resilient. However, in these animals vinculin is absent in all tissues. It is equally likely that the resistance to starvation was due to the effect of Vinculin in the fat body, ovary, brain, or other adult tissues singly or in combination. The fact that the intestine recovers transiently to a size slightly larger than that of the fed flies seems anecdotal, considering the noise within the timeline of fed controls. I am not sure this experiment is needed in the paper at all, but to me, the healthy conclusion from this effort is that more work is needed to determine the impact of vinculin-mediated intestinal homeostasis in stress resistance, and that this is out of the scope of this paper.
- Cell autonomy of the requirement of Vinculin and alpha-Catenin
Authors interpret that Vinculin is needed in the EB to maintain mechanical contact with the ISC, restrict ISC proliferation through contact inhibition, and maintain the EB quiescent. This interpretation explains seemingly well the lack of an obvious phenotype when knocking down vinculin in ISCs only, while knockdown in ISCs and EBs, or EBs only, does lead to differentiation problems. It also sits well with the additional observation that vinculin knockdown in mature ECs does not have an obvious phenotype.
However, a close examination makes the results difficult to explain with this interpretation only. If the authors were correct, one would expect that in mutant clones, eventually, vinculin-deficient EBs will be produced, which should mis-differentiate and induce additional ISC proliferation. However, the clones only show a reduction in ISC proportions; the most straight forward interpretation of this is that vinculin is cell-autonomously necessary for ISC maintenance (which is at odds with the phenotype of vinculin knockdown using the ISC and ISC/EB drivers).
Also, from the authors interpretation, it would follow induce that the phenotype of vinculin knockdown in ISCs+EBs and in EBs only should be the same. However, in ISCs+EBs vinculin knockdown, differentiation accelerates, which is likely accompanied by increased proliferation (judging by the increase in GFP area, PH3 staining would be more definitive). This contrasts with the knockdown only in EBs, which leads to accumulation of EBs due to misdifferentiation, and increased proliferation, mostly of ISCs, as measured directly with PH3 staining, but not additional late EBs or mature ECs. The authors call this "incomplete maturation due to accelerated differentiation". I think that one should expect to find incomplete differentiation/maturation when the rate of the process is very slow, not the other way around. To me, these are different phenotypes, which could perhaps be explained if vinculin was also needed in the ISC to transmit tension to the EB and prevent its differentiation, and removing it only in the EB may be revealing an additional, cell-autonomous requirement in maturation.
Another unexpected result, considering the authors interpretation, is that the over expression of activated Vinculin (vinc[CO]) does not seem to have much of an effect. It does not change the phenotype of the wild type (where there is very little basal turnover to begin with) and it only partially rescues the phenotype of the vinc[102.1] mutants, when the rescue transgene vinc:RFP does. This again suggests that there may be tissue interactions, in development or adulthood, that may explain the vinc[102.1] phenotypes. It could also be that this incomplete rescue is due to the deleterious effect of Vinc[CO]; this is another reason for doing the vinc[102.1]; esg-Gal4; UAS-vincFL experiments suggested above). An alternative experiment to perform this rescue would be to knock down vinculin gene while overexpressing the Vinc[CO] transgene - this may be possible with the RNAi HSM02356, which targets the vinculin 3'UTR and is unlikely to affect UAS-vinc[CO].
The claims of the authors would be more solid if the reporting of the phenotypes was more homogeneous, so one could establish comparisons. Sometimes conditions are analysed by differentiation index, others by extension of the GFP domains, others with phospho-histone-3 (PH3), others through nuclear size or density, and combinations. I do not think the authors should evaluate all these phenotypes in all conditions, but evaluating mitotic index and abundance of EBs and "activated EBs/early ECs" to monitor proliferation and differentiation rates should be done across the board (ISC, ISC+EB, EB drivers).
If the primary role of Vinculin is to induce contact inhibition in the ISC from the EB and prevent the EB differentiation and proliferation, one would expect that over expression of Vinc[CO] (or perhaps VincFL or sqhDD) in EBs should prevent or delay the differentiation and proliferation induced by a presumably orthogonal factor, like infection with Pseudomonas entomophila or Erwinia carotovora.
- Relationship between Vinculin and alpha-Catenin
The authors establish a very clear difference in the phenotypes between focal adhesion components and Vinculin, whereas the similarity of alpha-catenin and vinculin knockdowns is very compelling. Therefore I am sure the authors are in the right path with their interpretation of this part of the paper. However, some of the alpha-Catenin experiments are not very clear. The result from the rescue experiment of alpha-Cat knockdown with alpha-Cat-deltaM1b does not seem to show what the authors claim, and differentiation does not seem affected, only the amount of extant older ECs (which may be due to other reasons as this is a non-autonomous effect). Ulrich Tepass produced a UAS-alpha-catenin construct with the full deletion of the M1 region, perhaps that would show a clearer phenotype. Also, the autonomy of the phenotype is difficult to address with these experiments alone. It would be expected that the phenotype of alpha-catenin knockdown should be similar to that of vinculin knockdown in the ISCs only or EBs only.
-
Reviewer #2 (Public Review):
Vinculin functions as an important structural bridge that connects cadherin and integrin-mediated adhesions to the F-actin cytoskeleton. This manuscript carefully examined the mutant phenotype of vinc in the Drosophila intestine and found that vinc mutant in EBs causes significant increases of EB to EC differentiation, stem proliferation, and tissue growth. By analyzing the mutant phenotype of the cadherin adaptor alpha-catenin, the authors suggest that the vinc functions through the cell-cell junctions instead of cell-CEM adhesions in EBs. Finally, manipulation of myosin activity in EBs phenocopies the vinc mutant, suggesting that vinculin is regulated by the mechanical tension transduced through the cytoskeleton.
The authors claim that the vinculin mutant phenotype is opposite compared to the loss of the …
Reviewer #2 (Public Review):
Vinculin functions as an important structural bridge that connects cadherin and integrin-mediated adhesions to the F-actin cytoskeleton. This manuscript carefully examined the mutant phenotype of vinc in the Drosophila intestine and found that vinc mutant in EBs causes significant increases of EB to EC differentiation, stem proliferation, and tissue growth. By analyzing the mutant phenotype of the cadherin adaptor alpha-catenin, the authors suggest that the vinc functions through the cell-cell junctions instead of cell-CEM adhesions in EBs. Finally, manipulation of myosin activity in EBs phenocopies the vinc mutant, suggesting that vinculin is regulated by the mechanical tension transduced through the cytoskeleton.
The authors claim that the vinculin mutant phenotype is opposite compared to the loss of the major integrin components, suggesting a function independent of the cell-ECM adhesions. However, the phenotype of vinc and integrin may not be completely opposite. Besides loss of ISCs, both mys and talin knockdown in ISCs clearly causes ISCs differentiation into EC cells (Fig.3A), suggesting a possible involvement of integrin in EB to EC differentiation. Therefore, it will be important to test the phenotype of integrin KD in EBs using EB-specific Gal4.
The authors proposed a model that the cell-cell adhesion between ISC and EBs is required for vinculin mediated differentiation suppression. However, this model is not directly supported by the data as the EB-ISC adhesion and EB-EC adhesion have not been tested separately. In addition, previous short-term manipulation of E-cadherin in ISCs and EBs shows no change in cell proliferation (Liang J. et al. 2017), which seems to contradict the authors' model. To support the authors' conclusion, long-term manipulation of E-cadherin in ISCs and EBs must be tested.
The result of MARCM analysis seems inconsistent with the rest of the data. In MARCM, no significant change of clone sizes is observed between WT and vinc mutant (Fig. 3E). However, vinc mutant in EBs clearly promotes ISC proliferation in other experiments such as esg>vinc-RNAi and the EB>vinc-RNAi (Fig. 1A, Fig. 4).
In Fig. 4H, the authors suggest that vinculin mutant prevent terminal EC formation. However, this may be simply caused by longer retention of Klu expression in the newborn ECs. To test if EB differentiation is indeed affected, the EC marker pdm1 staining will provide more convincing evidence. Another experiment to strengthen the conclusion will be the tracking of clone sizes generated from a single EB cell using the UAS-Flp system (such as G-trace).
In Fig. 6D, the survival rate of WT and vinc mutant flies were compared. However, as there is no additional assay about the feeding behavior or metabolic rate, the systematic mutant of vinc does not provide a direct link between animal survival and intestinal EBs. Therefore, an experiment with vinc level specifically manipulated in fly intestine using esg>vinc-RNAi or BE>vinc-RNAi will be more relevant.
-
Reviewer #3 (Public Review):
Prior work had identified essential roles for Integrin signaling in regulating intestinal stem cell (ISC) proliferation, and the authors studies were motivated by trying to understand whether Vinculin (Vinc) might participate in this. However, Vinc is involved in mechanotransduction at both focal adhesions (FA) and adherens junctions (AJ), and their results revealed that Vinc phenotypes do not match those of FA proteins like Integrin. Conversely, they do match a-catenin (a-cat) RNAi phenotypes, and together with the localization of Vinc and the phenotypes associated with a-cat mutants that can't bind Vinc, this led to the conclusion that Vinc is acting at AJ rather than FA in this tissue. The results here are convincing, with clear presentation, nice images, and appropriate quantitation. It's also worth …
Reviewer #3 (Public Review):
Prior work had identified essential roles for Integrin signaling in regulating intestinal stem cell (ISC) proliferation, and the authors studies were motivated by trying to understand whether Vinculin (Vinc) might participate in this. However, Vinc is involved in mechanotransduction at both focal adhesions (FA) and adherens junctions (AJ), and their results revealed that Vinc phenotypes do not match those of FA proteins like Integrin. Conversely, they do match a-catenin (a-cat) RNAi phenotypes, and together with the localization of Vinc and the phenotypes associated with a-cat mutants that can't bind Vinc, this led to the conclusion that Vinc is acting at AJ rather than FA in this tissue. The results here are convincing, with clear presentation, nice images, and appropriate quantitation. It's also worth emphasizing that initial characterization of Vinc mutant flies failed to reveal any essential roles for this protein in Drosophila, so finding a mutant phenotype of any sort is significant.
While the manuscript is strong as a descriptive report on the requirement for Vinc in the Drosophila intestine, it doesn't provide us with much understanding of the mechanism by which Vinc exerts its effects, nor how its requirement is linked to intestinal physiology.
Prior work has shown that mechanical stretching of intestines stimulates ISC proliferation (presumably through Integrin signaling), which is opposite to what Vinc does here. There is a suggestion that Vinc is involved in maintaining homeostasis, but how its regulated remains a bit murky. The authors report that reductions in myosin activity result in phenotypes reminscent of Vinc phenotypes, which they interpret as supporting a model where Vinc's role is to help maintain tension at AJ. Of course it could also be reversed - maybe they are similar because tension is needed to maintain Vinc recruitment to AJ? They lack of epistasis tests and lack of analysis of whether Vinc localization to AJ in EBs is affected by tension or the M2 deletion of a-cat leaves us uncertain as to the actual basis for the relationship between Vinc and myosin phenotypes.
-
