Temperature and insulin signaling regulate body size in Hydra by the Wnt and TGF-beta pathways

How multicellular organisms assess and control their size is a fundamental question in biology, yet the molecular and genetic mechanisms that control organ or organism size remain largely unsolved. The freshwater polyp Hydra demonstrates a high capacity to adapt its body size to different temperatures. Here we identify the molecular mechanisms controlling this phenotypic plasticity and show that temperature-induced cell number changes are controlled by Wnt- and TGF-β signaling. Further we show that insulin-like peptide receptor (INSR) and forkhead box protein O (FoxO) are important genetic drivers of size determination controlling the same developmental regulators. Thus, environmental and genetic factors directly affect developmental mechanisms in which cell number is the strongest determinant of body size. These findings identify the basic mechanisms as to how size is regulated on an organismic level and how phenotypic plasticity is integrated into conserved developmental pathways in an evolutionary informative model organism.

1. It is a surprising observation that knockdown of INSR actually increases hydra's size. In most of vertebrate and invertebrate animal models, it is generally known that insulin/insulin like signalling positively regulates cellular and organismal growth, so the authors' observation is quite unexpected. As such, I feel that rigorous work should be done to more strongly support this finding. The authors should consider off-target effects when knockdown of INSR using shRNA, so additional independent shRNAs should be used to confirm the reproducibility of the data presented. Moreover, the authors should confirm the activity of molecular marker of insulin signalling when insulin signalling is perturbed.
2. I assume that in Figure 4, the authors tried to show that Wnt signal mediated tentacle formation (axis stability) is negatively correlated with organismal size associated with INSR and FOXO. To argue for the model for INSR-Wnt-TGF b cascades in size determination, the authors should show that size-modified hydra induced by knockdown of INSR or FOXO exhibits correlative tendency of tentacle formations or related traits reflecting axis stability. Present form of Figure 4 does not support that insulin signalling controls hydra's size by regulating Wnt signalling mediated axis stability. 3. Figure 5 showed that treatment of TGF-b inhibitors modifies hydra's size both in wild type and in conditions of perturbation of insulin signalling. In order to establish that insulin signalling controls hydra's size through TGF-b, the authors should show that perturbation of insulin signalling by knockdown of INSR or FOXO barely modulates hydra's size in the presence of TGF-b inhibitors. 4. Figure 5 used two different TGF-b signalling inhibitors which lead to two contradictory results. The authors noted that it may be due to the different target specificities referring to table s3, which fails to sufficiently explain the results. Why do two TGF-b inhibitors having similar target affinities have opposite effects on hydra's size regulation? 5. Figure S8 shows that treatment of GSK-3 inhibitor decreases the expression of several TGF-b signalling component genes, which cannot sufficiently support that TGF-b signalling is downstream of Wnt. To support this, the authors should shows that Wnt signalling induced changes in organismal size diminish in the presence of TGF-b inhibitors.
6. In Figure 2d the authors noted that no significant size difference was observed at 8°C , but there are no data showing this. Reviewer #2: Remarks to the Author: This study addresses an interesting biological question, how genetic and environmental factors do interact to regulate animal size. In their manuscript Mortzfeld et al use the freshwater Hydra polyp as a model and combine transcriptomics, transgenesis and pharmacological approaches to investigate this question. They claim that three distinct signaling pathways, namely the Insulin, Wnt and TGF-b pathways contribute to the regulation of cell number in Hydra. As main result, they propose that Wnt signaling acts as a size-measuring tool and controls axis stability by determining "critical size" at which budding is initiated. Once the critical size is reached, TGF-b would induce secondary axis formation through bud formation. In addition the authors also explore the regulation of cell size that seems to be rather temperature-dependent but independent of Wnt and TGF-b signaling. Unfortunately, the paper lacks coherence and is not convincing in the present state. Transcriptomics do not suffice to evidence changes in growth factor signaling. Also the knock-down assays of the insulin receptor and FoxO, obtained through shRNA transgenic constructs, are not validated and provide results at the phenotypic and transcriptomic levels that are rather patchy and descriptive. The transcriptomic analyses of these animals only provide a list of putative target genes of insulin signaling, while the pharmacological experiments designed to modulate the activities of the Wnt and TGF-b pathways do not suffice to decipher their role. In addition several technical issues need to be solved to convincingly demonstrate a correlation between temperature and animal size. Therefore, the significance of this article is currently too limited for the broad readership of Nature C ommunications.
Here are listed the main criticisms that should be addressed by the authors: 1. The authors demonstrate that animals are significantly larger at lower temperatures. However, animal size varies with the culture conditions such as for example the number of animals in a dish in a given volume. We noted some strong variations in the average size of the wild-type (Fig 1b) and control animals (Fig 2b, 3b, Supplementary Figure S3b, d). The authors definitely need to verify throughout their experiments the stability of animal size in the culture conditions they use. 2. The opposing effects of FoxO-KD and INSR-KD in size regulation suggests that Insulin signaling negatively regulates Hydra FoxO. To confirm this results the authors should validate their KD assays and show that FoxO activity is indeed enhanced when INSR is knocked-down. 3. The scale of the plots does not seem consistent between the different experiments. The authors should explain why in all experiments GFP negative control animals that are supposed to be equivalent to the wild-type condition are always smaller. It is quite difficult to evaluate the role of INSR-KD in the regulation of cell number, if at 22°C the control animals are 50% smaller than the wild-type ones. 4. The authors propose that Wnt signaling responds to Insulin signaling independently of the temperature effect (Fig 7). The argument is the up-regulation of Wnt11 in FoxO-KD animals at 12°C and 18°C . However, Wnt11 expression does not seem to be modulated in INSR-KD animals kept at 18°C ( Table 1). The expression of several other Wnt ligands should be tested in FoxO-KD and INSR-KD animals. In addition, the authors need to investigate the activity of the Wnt pathway in these different contexts to convincingly establish a hierarchical link between Insulin and Wnt signaling. 5. The authors propose that the activity of the TGF-b pathway is under the control of Wnt, as proposed by Watanabe et al in 2014 (should be cited for this point). To support this claim, the authors need to provide functional evidences. The results shown in supplemental figure 8 rather suggest that Wnt signaling negatively regulates TGF-b signaling. 6. Figure 4b requires better explanation to adequately evaluate the data. Again, it is difficult to understand why the GFP negative control animals of the different KD lines differ so much in size and tentacle number after ALP treatment. The presented results indicate that these controls are not equivalent to each other and also not equivalent to the AEP F2A4 wild-type animals. 7. The authors show opposite effects of the two TGF-b receptor inhibitors: a reduction in the number of epithelial cells with K02288 and an increase with LDN-193189. However, the specificity of these drugs in Hydra is not tested. Also their respective cellular effects is not tested, as for example a possible increase in apoptosis upon K02288 exposure. In addition, there is no indication on the temperature(s) where these drug treatments were performed. This information is needed given the temperature impact on animal size.
Minor points: • lines 335 and 338: the authors write that the increase in time before budding explains the increase in size of the KD animals, refering there to Figure 6. This is misleading as Figure 6 only contains information about dT and the graphs representing the animal size are in Figure

Overall Impression
-Interesting, solid work that makes an important contribution to the literature

Major Issues
-Well written -Novel work -Methodologically sound in general but can the authors clarify how they controlled for multiple statistical comparisons that may lead to a Type I error (i.e. false positive findings) ? -The discussion should try to tie in the human work in this area to make the findings more relevant across species. While the authors briefly mention genetic influence over human body size there is a body of work on FOXO3, body size and longevity in humans that support the manuscript's current findings and could strengthen the conclusions. This deserves to be mentioned in another paragraph or two in the discussion. Some relevant papers appear below.

Minor Issues
-Line 319: the word "neither" should be "either." Reviewer #1 (Remarks to the Author): 1. It is a surprising observation that knockdown of INSR actually increases hydra's size. In most of vertebrate and invertebrate animal models, it is generally known that insulin/insulin like signalling positively regulates cellular and organismal growth, so the authors' observation is quite unexpected. As such, I feel that rigorous work should be done to more strongly support this finding. The authors should consider off-target effects when knockdown of INSR using shRNA, so additional independent shRNAs should be used to confirm the reproducibility of the data presented. Moreover, the authors should confirm the activity of molecular marker of insulin signalling when insulin signalling is perturbed.
Thank you for the comment, we agree that an increase of body size by the KD of the INSR was a surprising finding. However, KD of the INSR caused only mild changes in body sizes. To account for off-target effects, we performed a BLAST search for the insR-hairpin (HP) construct using an e-value threshold of 10 and no other major criteria to obtain all possible off-targets present in the new Hydra transcriptome and checked their expression levels from our RNA-seq experiment ( Supplementary Fig. S5c). We could not detect any downregulation in any of the potential target genes except for the INSR itself. Hence, we have no reason to believe that off-target effect for this construct exist. Furthermore, we generated another line bearing the same HP-construct which showed a comparable large size phenotype and added these data in Supplementary Figure 4, the authors tried to show that Wnt signal mediated tentacle formation (axis stability) is negatively correlated with organismal size associated with INSR and FOXO. To argue for the model for INSR-Wnt-TGF b cascades in size determination, the authors should show that size-modified hydra induced by knockdown of INSR or FOXO exhibits correlative tendency of tentacle formations or related traits reflecting axis stability. Present form of Figure 4 does not support that insulin signalling controls hydra's size by regulating Wnt signalling mediated axis stability. Thank you for this comment. In order to give a better explanation of the dataset and to convey the intention of the plot in a clearer way, we split Figure 4 into several subfigures and added Supplementary Fig. S10 to the manuscript. Indeed, we saw a correlation of Wnt signal mediated tentacle formation (axis stability) and the maximum size in different wild type genotypes (blue to magenta triangles, and diamond), animals reared at different temperatures (triangles), knockdown of the INSR (squares), and the knockdown of FoxO (circles, diamonds). The plot thus clearly shows, that the Wnt signaling by the means of generation of ectopic tentacles is altered in the INSR-KD or FoxO-KD Hydra lines. However, their number of ectopic tentacles is still proportionate to the maximum size and fits to the regression line. Therefore, we can show that Wnt signaling is altered with the interference of the insulin and the FoxO signaling pathway and that Wnt is the key regulator for controlling the maximum size in the polyp. Taken together, we introduce first evidence of the hierarchy of Insulin/FoxO/Temperature -Wnt is controlling size. We edited text in lines 294-309 of the manuscript and added a Supplementary Fig. S10 including an explanatory caption.

I assume that in
3. Figure 5 showed that treatment of TGF-b inhibitors modifies hydra's size both in wild type and in conditions of perturbation of insulin signalling. In order to establish that insulin signalling controls hydra's size through TGF-b, the authors should show that perturbation of insulin signalling by knockdown of INSR or FOXO barely modulates hydra's size in the presence of TGF-b inhibitors. Thank you for this recommendation! In order to put the results better into perspective, we added controls in the plots of Fig. 5, edited the captions and edited text in lines 341-342 and lines 349-354 of the manuscript. Figure 5 used two different TGF-b signalling inhibitors which lead to two contradictory results. The authors noted that it may be due to the different target specificities referring to table s3, which fails to sufficiently explain the results. Why do two TGF-b inhibitors having similar target affinities have opposite effects on hydra's size regulation? Thank you for this important question. At first we were also puzzled by this observation and could not explain these surprising results. In order to get a better understanding of the mode of action of the two inhibitors, we performed a detailed analysis of the ATP-binding domain of human and Hydra TGF-β receptors. We were able to confidently predicted similar target specificties of both inhibitors for the human TGF-β receptors, but also found differential affinities for the Hydra TGF-β receptor ATP-binding domains of the two inhibitors. This result on the one hand suggests activity of the inhibitors in Hydra and on the other hand gives an explanation of different effects of the inhibitors, as the inhibitors block different TGF-β receptors and thus different parts of the pathway. Uncovering the exact mechanism and how budding is induced by the interplay of TGF-β pathway components is beyond the scope of this paper but will be worthwhile to look into in the near future as it might reveal conserved interactions between the different parts of the TGF-β pathway. We added these results in lines 329-333 and provided an additional supplementary figure S12. Figure S8 shows that treatment of GSK-3 inhibitor decreases the expression of several TGF-b signalling component genes, which cannot sufficiently support that TGF-b signalling is downstream of Wnt. To support this, the authors should shows that Wnt signalling induced changes in organismal size diminish in the presence of TGF-b inhibitors. Thank you for this critical comment and the suggestion for further experiments. We extended the evidence, that place the Wnt signaling cascade upstream of the TGF-β pathway by the following experiments. First, we treated β-catenin overexpression animals with K02288. The overexpressed β-catenin in these animals was truncated and lacks the amino acid which is usually phosphorylated by the APC for proteolytic degradation in the absence of a Wnt signal. The blocked degradation of the β-catenin led to the constitutive activation of the Wnt signaling pathway and the formation of multiple axis 2 . Treatment with K02288 rescued this phenotype and restored a single axis in these animals. Second: we treated animals with low concentrations of ALP over a period of 10-14 days and observed a reduction in the number of epithelial cells, confirming our notion of Wnt being the key regulator in size determination in Hydra. We next co-treated animals with ALP and LDN-193189 to overwrite the ALP effect with the TGF-β inhibitor. We were able to rescue the small size phenotype in these animals and even observed an increase in epithelial cell number. We included these results in two additional paragraphs in lines 376-395 and added two new figures (Fig. 7, original Fig. 7 is now Fig. 8 and Supplementary Fig.  S14) to the manuscript. Figure 2d the authors noted that no significant size difference was observed at 8°C , but there are no data showing this. Reviewer #2 (Remarks to the Author):

In
1. The authors demonstrate that animals are significantly larger at lower temperatures. However, animal size varies with the culture conditions such as for example the number of animals in a dish in a given volume. We noted some strong variations in the average size of the wild-type (Fig 1b) and control animals (Fig 2b, 3b, Supplementary Figure S3b, d). The authors definitely need to verify throughout their experiments the stability of animal size in the culture conditions they use. 3. The scale of the plots does not seem consistent between the different experiments. The authors should explain why in all experiments GFP negative control animals that are supposed to be equivalent to the wild-type condition are always smaller. It is quite difficult to evaluate the role of INSR-KD in the regulation of cell num ber, if at 22°C the control animals are 50% smaller than the wild-type ones. Thank you for this comment. We again would like to point out, that different lines, especially those gained through transgenesis, are the product of sexual reproduction. It is true that lines with different genetic backgrounds exhibit different sizes. However, it was by chance that the two intensively studied transgenic lines were smaller than the wild type counterpart. In order to show evidence for this claim, we included data for an additional INSR-KD (C6-2) and an additional FoxO-KD line (E11) into our data set. Both transgenic control lines show similar maximum sizes as the wild type F2A4 line (Supplementary Fig. S7a, Supplementary Fig. S10a), thus the smaller maximum sizes of INSR-KD C1-1 ctrl and FoxO-KD D11a ctrl animals are not caused by the transgenesis of the animals. 4. The authors propose that Wnt signaling responds to Insulin signaling independently of the temperature effect (Fig 7). The argument is the up-regulation of Wnt11 in FoxO-KD animals at 12°C and 18°C. However, Wnt11 expression does not seem to be modulated in INSR-KD animals kept at 18°C ( Table 1). The expression of several other Wnt ligands should be tested in FoxO-KD and INSR-KD animals. In addition, the authors need to investigate the activity of the Wnt pathway in these different contexts to convincingly establish a hierarchical link between Insulin and Wnt signaling. Thank you for this critisism and the suggestion for further experiments to support our data set.

Thank you for this important comment. While it is correct that the culture conditions can affect the size regulation of the animals, we want to highlight that culture conditions were comparable between all treatments, KD animals, and control animals as well as between all lines. Differences between the control animals can be explained due to genetic variation between the lines. Transgenic animals are generated by DNA construct injection into the 1-4 cell stage of the animals, and thus requires sexual reproduction of the polyps. Transgenic hatchlings are usually mosaic, and fully transgenic animals are generated by continuous bud selection resulting in animals containing only transgenic cells 1 . The same technique is applied to generate control animals of the same line (same genetic background), by simply selecting animals which do not contain transgenic cells 2 . This explains the differences in the control lines between different experiments and is an effect of different genetic backgrounds
In new experiments, we tested the expression of several Wnt pathway components via qRT-PCR in the INSR-KD and FoxO-KD animals to confirm the notion of insulin and FoxO dependent Wnt signaling (Supplementary Fig. S9). Indeed, we revealed Wnt11 to be most probably INSR dependent (p = 0.09) using this more sensitive method, while we also identified Wnt8 to be insulin signaling dependent. We further confirmed the FoxO dependent Fzd1/7 and Wnt11 expression of the transcriptome analysis and even revealed the previously undetected Wnt3a to be FoxO dependent. This result, combined with the transcriptome analysis and the changed tentacle formation after ALP treatment place insulin and FoxO signaling upstream of the Wnt pathway. To include these findings into the manuscript we added lines 259-266 to the text. 6. Figure 4b requires better explanation to adequately evaluate the data. Again, it is difficult to understand why the GFP negative control animals of the different KD lines differ so much in size and tentacle number after ALP treatment. The presented results indicate that these controls are not equivalent to each other and also not equivalent to the AEP F2A4 wild-type animals. Thank you for this comment. Reviewer #1 had a similar question and we added a figure in the supplements (Supplementary Fig. S10) which shows the data set in more depths. We hope that the corresponding explanatory caption and the editions in the main text of the manuscript (lines 294-309) help understanding the results which supports our conclusions. Again, we would like to stress that the size differences of the different transgenic control lines and the wildtype AEP F2A4 is due to differences in the genetic background of these animals and that different genotypes (lines) show specific maximum sizes. The control lines of INSR-KD C1-1 and FoxO-KD D11a are smaller compared to AEP F2A4 by chance and we added two additional lines (INSR-KD C6-2 and FoxO-KD E11), that show same phenotypes but where the control lines are of similar size as the wild type animalsm, thus equivalent ( Supplementary Fig. S7a,  Supplementary Fig. S10a). However, we do not believe that animals with different genetic backgrounds have to be equal in size. This is the reason we chose the corresponding control line to each KD line to compare to and not the AEP F2A4 line.

The authors propose that
7. The authors show opposite effects of the two TGF-b receptor inhibitors: a reduction in the number of epithelial cells with K02288 and an increase with LDN-193189. However, the specificity of these drugs in Hydra is not tested. Also their respective cellular effects is not tested, as for example a possible increase in apoptosis upon K02288 exposure. In addition, there is no indication on the temperature(s) where these drug treatments were performed. This information is needed given the temperature impact on animal size. Thank you for this important comment. We performed detailed sequence based in-silico analysis to provide good evidence for target specificity and an explanation for the contradictory phenotypes induced by the inhibitors (Supplementary Fig. S12) On the one hand, we show that the ATP binding domain is well enough conserved from Hydra to human to serve as specific target for both inhibitors and on the other hand we reveal that differences in the ATP binding pocket of human and Hydra TGF-β receptors explain different target specificities of the K02288 and LDN-193189, which gives an explanation for the opposite effects of the inhibitors. We included the results in Supplementary Fig. S12 Fig 6b) before bud initiation compared to untreated polyps. Third, the mere mechanics of proliferation and apoptosis, which are exponential in nature, would not provide a system which is stable enough to maintain a specific maximum size (a manuscript explaining these effects in detail is in preparation).
Minor points: • lines 335 and 338: the authors write that the increase in time before budding explains the increase in size of the KD animals, refering there to Figure 6. This is misleading as Figure 6 only contains information about dT and the graphs representing the animal size are in Figure 5. Thank you, we added a reference to Fig. 5 in the mentioned passage of the text.
• Line 117: Please clarify how the 20 genes were selected. The 20 genes were representatives of the Hydra transcriptome which where previously cloned and sequenced in the lab and therefore verified by experimental evidence.
• In lines 256 and 260 the authors refer to the Supplementary Figure S4 that should represent the fractions of orphan genes in the analysis. This figure does not seem to exist. Thank you, we are sorry for the inconvenience and added the right figure in the supplements.
• Accession numbers are missing. All raw sequencing data should be made available. Thank you, it is correct that we forgot to include the SRA accession number. We included the information in the methods section (lines 541-543). The data will be released upon publication but the SRA provides reviewer links to the metadata. Please see below: ftp://ftptrace.ncbi.nlm.nih.gov/sra/review/SRP133389_20190125_113520_37d5c0b6b354bc3c790d2696b4 2756c9 • The references for Supplementary Figure S2 and S3 are inverted in the text. Thank you, we corrected the references in the text.
Reviewer #3 (Comments to the author): Methodologically sound in general but can the authors clarify how they controlled for multiple statistical comparisons that may lead to a Type I error (i.e. false positive findings) ? Thank you for this comment and we are happy to elaborate our statistical methods. We stated in the methods section that we used either the two-tailed Student's-t-test or the Mann-Whitney-Utest, where it was applicable, in order to test for statistical significance between the treatments. We agree that controlling for multiple comparisons is an important and often neglected part of biostatistics. Especially when dealing with next generation sequencing data, p-value correction is very important in order to distinguish significant differences from false positive findings and artifacts. Therefore, wherever multiple testing occurred in our data, we adjusted the p-values using the Benjamini-Hochberg correction 1 .
The discussion should try to tie in the human work in this area to make the findings more relevant across species. While the authors briefly mention genetic influence over human body size there is a body of work on FOXO3, body size and longevity in humans that support the manuscript's current findings and could strengthen the conclusions. This deserves to be mentioned in another paragraph or two in the discussion. Some relevant papers appear below. Thank you for this suggestion and we agree that FoxO related work on humans is a bit underrepresented in the discussion. Therefore, we added a small text passage referencing the latest literature in the FoxO field (lines 430-433). However, we feel the scope of this manuscript is not to give detailed insights into FoxO research which was discussed in numerous excellent reviews in the last few years.
Minor Issues -Line 319: the word "neither" should be "either." Thank you, we corrected the typo.