HDAC5 is a novel injury-regulated tubulin deacetylase controlling axon regeneration

Transcription

1 Manuscript EMBO HDAC5 is a novel injury-regulated tubulin deacetylase controlling axon regeneration Yongcheol Cho Corresponding author: Valeria Cavalli, Washington University in Saint Louis Review timeline: Submission date: 06 February 2012 Editorial Decision: 08 March 2012 Revision received: 11 April 2012 Editorial Decision: 02 May 2012 Revision received: 02 May 2012 Accepted: 03 May 2012 Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) 1st Editorial Decision 08 March 2012 Thank you for submitting your study to the EMBO Journal. Your manuscript has now been seen by three experts and their comments are provided below. As you can see the referee appreciate the findings reported and support publication in the EMBO journal. They raise a number of specific concerns, in particular concerning the HDAC inhibitor scriptaid. However, they shouldn't involve too much additional work to address. Given the positive comments, I would like to invite you to submit a suitably revised manuscript for our consideration. When preparing your letter of response to the referees' comments, please bear in mind that this will form part of the Review Process File, and will therefore be available online to the community. For more details on our Transparent Editorial Process, please visit our website: I thank you for the opportunity to consider your work for publication. I look forward to your revision. Yours sincerely Editor The EMBO Journal REFEREE REPORTS European Molecular Biology Organization 1

2 Referee #1 A few studies have pointed to extranuclear roles for HDAC proteins, but I do not think that any have come up with such a provocative and clear case for mechanism including upstream regulators and down stream effectors. The data for a gradient of acetyl-tubulin mediated by HDAC5 activity are clear. The experiments also support a role of PKC-dependent phosphorylation in modulating this process. The combination of in vivo and cell culture experiments along with solid biochemistry from each is a strength that should make for high interest by EMBO readers. However, there are several places where the manuscript could be strengthened and somewhere the authors take more liberty with interpretation of the data than justified by their data. The analyses of optic nerve and dorsal column lesions led the authors to conclude that tubulin deacetylation occurs in PNS but not in CNS axons. However, it appears that only 24 hour time points were assessed for both the optic nerve and spinal cord lesions. Given the delayed Wallerian degeneration in CNS and decreased regenerative capacity, I think that at least some longer duration is warranted. Moreover, this raises the issue of whether HDAC5 or p-hdac5 localizes to these CNS axons. Lack of any axonal localization of HDAC5 in CNS axons could provide a convincing case for absence of tubulin deacetylation (and mechanistic insight into why). The authors suggest that the effect of scriptaid seen is transcription-independent based on application to axon compartment of a Campenot chamber. Though these chambers have classically been used to examine axonal effects, I think the authors' case would be strengthened by directly testing for role of transcription. At best, their current data suggest transcription independence. Since proteins are clearly transported within the axon across the boundaries of these chambers, Scriptaid could be acting locally to attenuate HDAC activity yet still be a transcriptional or nuclear effect. The in vivo treatments with Scriptaid are convincing. However, the authors need to be specific when they have used a single treatment, as I presume most data are based on. Additionally, the authors should consider moving the data in Suppl Fig 5D and E to the main manuscript since the repeated treatments argue for a continued need for HDAC activity for successful reinnervation. The authors should also give some indication of the specificity of Scriptaid for HDACs and which HDACs it inhibits. The images of growth cones shown in Fig 3G are not of sufficient resolution to interpret. Similar can be argued for the supplemental videos of growth cones. The data in Fig 5 make a strong case for HDAC5 mediating some of the attenuated regenerative growth. This is extremely interesting given the work of Reviccio et al. showing that inhibition of HDAC6 allows axons to grow on non-permissive substrates. The authors note that neither HDAC5 nor HDAC6 depletion accounted for scriptaid's effects on the terminal 50 µm of the injured axon. The authors suggest that incomplete depletion of HDAC5 accounts for the difference, but chances seem equally well for other mediators. Similar can be said for the statements on pg. 14 that HDAC5 is 'sufficiently active at the site of injury in the absence of PKC signaling' - it seems like chances are equal for other deacetylases to be involved or other changes in the cytoskeleton could precipitate this (e.g., tyrosination). I do not think this detracts from the authors' data; rather this makes the system more interesting that the axon may be compartmentalizing the mechanisms regulating its cytoskeletal stability. The data in Fig 8A seem to make the case for alternate mechanisms even more since the shrna resistant HDAC5 does not completely rescue the growth deficit of the shrna and overexpression of HDAC5 by itself does not increase growth. It would be intriguing to consider the effects of a phosphomimetic form of HDAC5 on axon growth, but this seems beyond the scope of the current manuscript. Likewise, the argument that PKC-dependent activation of HDAC5 with injury accounts for this (Fig 5F) is confusing to interpret - phdac5 is not decreased by CIP treatment of the lysates (last lane of inputs). The colocalization data for phdac5 for axonal tubulin in Fig 6A needs higher magnification images to conclusively show axonal localization (preferably with XZ reconstruction). Similarly, plots of phdac5 in addition to total HDAC5 would strengthen the data in 6D. European Molecular Biology Organization 2

3 The co-ip data shown in Fig 6F are convincing but raise questions on where HDAC5 is localized prior to its phosphorylation and association with kinesin-1. The text implies that previous studies showed HDAC5 phosphorylation results in its relocalization in the cytoplasm. It would be useful to know where the authors think that HDAC5 is coming from - is this a localized phosphorylation event (as Fig 8E suggests) and recruitment from other sites in the axon or is phdac5 getting transported from the cell body into the distal axons? Finally, the schematic shown in Fig 8E should be a separate figure. Unless I missed it, the authors have not shown any data for a role of HDAC5 activity in developmental growth of axons. This schematic should be reformulated to reflect the data shown here. Alternatively, the authors need to clearly reference the data supporting this model for developing neurons. Minor points: The figure legends and points in the results section suffer from a lack of details. For example, what is 'I slope' in Fig 2G? What do the dashed arrows indicate in images for Fig 2 and later in manuscript? I am not sure what Fig 1I adds. Could curve fitting be used for 1K? For the discussion, the authors might also consider referencing work from Wu et al (2003) IOVS showing that PKC's role in regeneration of injury-conditioned RGCs. The discussion also mentions the role of microtubule stabilizing drugs in CNS axon regeneration but their data seem to argue against this mechanism. Referee #2 Overall: The manuscript by Cho and Cavalli provides very convincing data demonstrating that microtubule deacetylation occurs after axon injury, that this is due to HDAC5, and that control of MT acetylation levels is important for axon outgrowth. The authors have taken a multifaceted approach to the issue and have really done a lot of work to put the story together. Of course HDAC5 k/o mice would be ideal for looking at the response to sciatic nerve injury, but I assume this is not possible. Failing this, the authors have done a lot already to test their ideas all the way from isolated neurons in vitro, to reinnervation of endplates. I do wish that fewer of the experiments relied on scriptaid, but the authors do use more specific tools at the end. General Comments: -Scriptaid is relied on for many of the expts, and it the only way acetylation is perturbed in in vivo regeneration assays. How much does this affect the transcriptional response to axon injury? Regeneration induces largescale transcriptional changes, and presumably HDACs are involved in these. This issue makes the in vivo regeneration experiments difficult to interpret. -I am left unsure whether HDAC5 is involved in normal axonal growth or is specific to regeneration. Certainly scriptaid seems to inhibit both (For example figure 3A shows developmental outgrowth). Is HDAC5 more specific? This issue is not explored (8C and D may be developing axons, I can't tell from legend or text), although 8E shows a diagram of it at the growth cone in development. - The English in the methods section is not very good: missing articles, subject-verb disagreement etc. -Quite a few of the conclusions in the results section seem overstated. -I found the figure on phospho-hdac5 vs total HDAC5 (Figure 6) very confusing. As this is central to the paper, it is crucial that this is cleared up- see below for more. Specific Comments: -Paragraph 1, reference to Figure 1 C- could you explain a little more the conclusion that this shows the deacetylation is mostly in axons? How can you tell that from the figure (this is probably obvious to someone who uses this system, but not necessarily to others). European Molecular Biology Organization 3

4 -Description of Figure 1 G and H: Different distances from ligation site are mentioned. For reference, how far from the ligation site is the cell body? -Scriptaid is introduced without a reference or much description. -What do the authors make of the decrease in acetylated tubulin distal to the injury site? This is even more dramatic than the proximal decrease, and is unlikely to be directly involved in regeneration. Is this also seen in CNS neurons? (this is an aside to the main story, so maybe just a little more discussion of it would be appropriate). -Figure 3D and E: This experiment is helpful, although I am not sure I would conclude that the effects of scriptaid are transcription-independent from it- it does suggest that targets are present in the axon, but does not address transcription directly. Because the previous expts on regeneration are done with the whole cell exposed to the drug, it would be really good to do the converse of the axondrug addition here. I would love to see the drug just added to the cell body. Does this have no effect? A stronger effect? This might allow some better guesses at how much of the effects seen in the previous global experiments are due to nuclear vs local targets of scriptaid. -I appreciate the rapid effects of the drug in the growth cone assay. One additional point that might be worth considering: I think adult DRGs have much smaller growth cones, and adult neurons in general have smaller growth cones. Do you think that HDAC effects could also be seen in these- as they are more the type of growth cone that might be expected during regeneration? -Is the culture assay in which DIV7 embryonic DRGs have their axons severed truly regenerative growth? Or is it simply a continuation of the active outgrowth already underway? -Figure 4A: In the image shown for HDAC5 shrna, it looks like acetylation is globally increased (green is stronger even far from injury site). Is this true? Or is this image an outlier? It is stated that it is the injury-induced deacetylation that is blocked, but information about acetylation levels in uninjured neurons are required to make this conclusion. Are the values in B and C all normalized to take out this information? Or am I just missing how to get at it?- OK< I see this is shown in Figure S10. It might be worth selecting a different picture for 4A. -Is PKC the only thing stimulated by PMA? If not, then the conclusion that phosphorylation of HDAC5 is PKC-dependent is a little strong for this data, although an inhibitor is used in a different experiment. -I am confused by Figure 6B. It shows in the upper row that phosphorylated HDAC is more abundant in the sciatic nerve after injury, but the bottom row shows that overall levels are unchanged. If this is true, what is being shown in D and E of this Figure? Is this really staining of phosphorylated HDAC5 and not total as indicated in the Figure, results text and legend? I also have a similar issue with the kinesin-1 experiment: isn't the point of the first part of the figure that phosphorylation of HDAC5 is the major response to injury? How does this relate to the analysis of total HDAC5 in the rest of the figure? -Again, the conclusion at the end of Figure 6 seems overstated: "These data also indicate that unlike HDAC6, HDAC5 functions specifically in injured axons." No experiments relating to function were shown. -missing reference: "To test whether microtubule stability was altered in injured nerves, we measured the steady-state level of microtubule polymerization, as previously described 45." -typos in Figure 7: I think labels in K and L should read unligated, not unligation Discussion of fine balance of stability and instability of MTs was helpful. I also found the discussion of previous studies on HDACs and regeneration helpful. Referee #3 The paper addresses an area of considerable practical importance, the inability of axons in the CNS to regenerate due to unidentified factors leading to a low intrinsic regenerative ability. It presents a comprehensive set of experiments based around the hypothesis that tubulin modifications after axotomy are an important determinant of the success of axon regeneration. 1. Much of the first part of the paper depends on the use of the HDAC inhibitory scriptaid. The results are validated later by knockdowns of HDACs, but nevertheless it would be useful to have a statement on the specificity of spriptaid for HDACs, and other possible side effects or long term toxic effects. Some of the effects of longer term treatment that are shown could be due to nonspecific toxicity. European Molecular Biology Organization 4

5 2. The changes in tubulin acetylation that are described take 6 hours to occur. Yet there are various accounts in the literature describing in vivo and in vitro experiments demonstrating that peripheral nerve regeneration can occur much faster than this. The authors need to discuss demonstrate whether the changes in tubulin acetylation in axons occur before or after regeneration has started. 3. On page 8 it would be useful to know the percentage of axons that regenerate rather than the absolute number. 4. The experiment on page 14, in which microtubule stability is assayed, will have been performed on an mixture of axonal and glial tubulin. Do the authors have information on the percent of tubulin that comes from axons in these extracts? Can they give arguments showing that we are looking at axonal changes? 1st Revision - authors' response 11 April 2012 Referee #1: A few studies have pointed to extranuclear roles for HDAC proteins, but I do not think that any have come up with such a provocative and clear case for mechanism including upstream regulators and down stream effectors. The data for a gradient of acetyl-tubulin mediated by HDAC5 activity are clear. The experiments also support a role of PKCdependent phosphorylation in modulating this process. The combination of in vivo and cell culture experiments along with solid biochemistry from each is a strength that should make for high interest by EMBO readers. However, there are several places where the manuscript could be strengthened and somewhere the authors take more liberty with interpretation of the data than justified by their data. We agree with the reviewer and have modified the text throughout to avoid overinterpretation of the data. The analyses of optic nerve and dorsal column lesions led the authors to conclude that tubulin deacetylation occurs in PNS but not in CNS axons. However, it appears that only 24 hour time points were assessed for both the optic nerve and spinal cord lesions. Given the delayed Wallerian degeneration in CNS and decreased regenerative capacity, I think that at least some longer duration is warranted. Moreover, this raises the issue of whether HDAC5 or p-hdac5 localizes to these CNS axons. Lack of any axonal localization of HDAC5 in CNS axons could provide a convincing case for absence of tubulin deacetylation (and mechanistic insight into why). We absolutely agree with the reviewer and have examined the level of tubulin acetylation 48 hours following injury in the optic nerve. Similar to the 24 hour time point, we did not detect significant changes in tubulin acetylation levels 48 hours after injury (now shown in Supplemental Data 1D). We also examined whether HDAC5 localizes to CNS axons and now show in Figure 6G that, in contrast to the sciatic nerve, HDAC5 is not detected in the optic nerve. This result provides a mechanistic insight for the lack of tubulin deacetylation observed there. The authors suggest that the effect of scriptaid seen is transcription-independent based on application to axon compartment of a Campenot chamber. Though these chambers have classically been used to examine axonal effects, I think the authors' case would be strengthened by directly testing for role of transcription. At best, their current data suggest transcription independence. Since proteins are clearly transported within the axon across the boundaries of these chambers, Scriptaid could be acting locally to attenuate HDAC activity yet still be a transcriptional or nuclear effect. We appreciate the reviewer s suggestion to examine a role for HDAC in transcriptional regulation of axon growth. We absolutely agree that our data do not exclude the possibility of transcriptional roles for HDACs in addition to a local axonal role in axon regeneration and we have revised the Result and Discussion sections accordingly. European Molecular Biology Organization 5

6 We believe that the experiment shown in Fig 3D using Campenot chamber strongly suggest an additional local axonal role for HDACs. Indeed, the Campenot chambers used allowed us to add vehicle to the left axon chamber and scriptaid to the right axon chamber. The cell body compartment was left untreated and axon growth monitored on both chambers. Thus, in these conditions, if HDACs played transcriptiondependent roles in outgrowth, they should be observed symmetrically in both chambers. Instead we found that growth was delayed only in the scriptaid treated chamber (Fig. 3D), suggesting local axonal role of HDACs in mediating this effect. Our results, however, do not rule out the possibility that HDACs play both transcription-dependent and independent roles in axon regeneration. However, it is important to note that the current literature on the role of HDACs in transcriptional regulation of axon regeneration indicates a repressive role of HDACs, suggesting that the effect we observed may arise from a distinct mechanism. Indeed, while we found that HDAC activity positively regulates axon growth of peripheral neurons, others have shown that HDAC transcriptional activity represses axon regeneration in several types of CNS neurons (for example Gaub et al 2010, Gaub 2011) or DRG neurons grown on inhibitory substrates (Riveccio et al 2009). While we are examining further the potential role of HDAC5 in transcriptional regulation following nerve injury, we believe that these studies are beyond the scope of the current manuscript. The in vivo treatments with Scriptaid are convincing. However, the authors need to be specific when they have used a single treatment, as I presume most data are based on. Additionally, the authors should consider moving the data in Suppl Fig 5D and E to the main manuscript since the repeated treatments argue for a continued need for HDAC activity for successful reinnervation. The authors should also give some indication of the specificity of Scriptaid for HDACs and which HDACs it inhibits. We thank the reviewer for this suggestion and have moved Supp Figure 5D and 5E now shown as Figure 4G and 4H. We have also specified in both Results and Methods sections how the scriptaid treatments were performed in vivo and provided more detailed information on the specificity of scriptaid. The images of growth cones shown in Fig 3G are not of sufficient resolution to interpret. Similar can be argued for the supplemental videos of growth cones. We agree with the reviewer and have repeated this experiment to provide higher resolution images of growth cones in Fig 3G. We believe that in combination with the video, these images support a role for HDAC activity in growth cone dynamics. The data in Fig 5 make a strong case for HDAC5 mediating some of the attenuated regenerative growth. This is extremely interesting given the work of Reviccio et al. showing that inhibition of HDAC6 allows axons to grow on non-permissive substrates. The authors note that neither HDAC5 nor HDAC6 depletion accounted for scriptaid's effects on the terminal 50 µm of the injured axon. The authors suggest that incomplete depletion of HDAC5 accounts for the difference, but chances seem equally well for other mediators. Similar can be said for the statements on pg. 14 that HDAC5 is 'sufficiently active at the site of injury in the absence of PKC signaling' - it seems like chances are equal for other deacetylases to be involved or other changes in the cytoskeleton could precipitate this (e.g., tyrosination). I do not think this detracts from the authors' data; rather this makes the system more interesting that the axon may be compartmentalizing the mechanisms regulating its cytoskeletal stability. We appreciate the reviewer s comments and have modified the text to acknowledge the possibility that other mediators including deacetylases or cytoskeletal changes could be involved in tubulin deacetylation in the axon segment immediately proximal to the injury site. The data in Fig 8A seem to make the case for alternate mechanisms even more since the shrna resistant HDAC5 does not completely rescue the growth deficit of the shrna and European Molecular Biology Organization 6

7 overexpression of HDAC5 by itself does not increase growth. It would be intriguing to consider the effects of a phosphomimetic form of HDAC5 on axon growth, but this seems beyond the scope of the current manuscript. We agree with the reviewer that studying the effect of a phosphomimetic HDAC5 on axon growth would be very interesting. However, while PKC is known to phosphorylate HDAC5 and regulate its nuclear export, additional HDAC5 phosphorylation sites have been recently identified (Greco et al., 2011). Therefore, studies involving phosphomimetic HDAC5 would first require identifying which residues are involved in the regulation of HDAC5 activity towards tubulin and, as the reviewer pointed out, would go beyond the scope of the current manuscript. Likewise, the argument that PKC-dependent activation of HDAC5 with injury accounts for this (Fig 5F) is confusing to interpret - phdac5 is not decreased by CIP treatment of the lysates (last lane of inputs). We realize that the labeling of Figure 5F and the experimental description lacked clarity and thus led to confusion. The CIP treatment was performed on the immunoprecipitated HDAC5, not on the total lysate. Thus in the input lysate (now labeled Lysates before IP, phdac5 is not decreased by CIP treatment. However, following CIP treatment of the immunoprecipitated HDAC5 (labeled as Ppt), total HDAC5 is detected with the Flag antibody, but there is no corresponding signal with a p-ser/thr antibody, reflecting that HDAC5 was indeed dephosphorylated by CIP treatment. We have improved the description for this experiment in the Materials and Methods section and in the Figure 5 legend. The colocalization data for phdac5 for axonal tubulin in Fig 6A needs higher magnification images to conclusively show axonal localization (preferably with XZ reconstruction). Similarly, plots of phdac5 in addition to total HDAC5 would strengthen the data in 6D. We agree with the reviewer and now provide higher magnification confocal images to show co-localization of HDAC5 with TUJ-1 positive axons in injured sciatic nerves. We now also provide in Figure 6D and 6E plots for p-hdac5 in addition to total HDAC5. Similar to HDAC5, p-hdac5 localizes in a gradient that correlates with the gradient of deacetylated tubulin. The co-ip data shown in Fig 6F are convincing but raise questions on where HDAC5 is localized prior to its phosphorylation and association with kinesin-1. The text implies that previous studies showed HDAC5 phosphorylation results in its relocalization in the cytoplasm. It would be useful to know where the authors think that HDAC5 is coming from - is this a localized phosphorylation event (as Fig 8E suggests) and recruitment from other sites in the axon or is phdac5 getting transported from the cell body into the distal axons? We agree with the reviewer that knowing where HDAC5 is coming from would be very interesting. We have edited the Discussion section as well as the model shown in Figure 8F to address this point and suggest that HDAC5 transported from more proximal axonal locations leads to the formation of the HDAC5 gradient. In basal conditions, low levels of HDAC5 are present in sciatic nerve axons (Figure 6B), in agreement with previous observations showing that only 50% of HDAC5 resides in the nucleus in hippocampal neurons (Chawla et al. 2003). Following axotomy, local calcium influx (Figure 7A,B) activates PKC (Figure 7C), which in turn phosphorylates axonal HDAC5 (Figure 6B). p- HDAC5 displays enhanced activity towards tubulin (Figure 5F,G) and promotes local tubulin deaceylation (Figure 5A,B,C). The retrograde propagation of calcium along the axon (Figure 7A) may then enhance HDAC5 interaction with kinesin-1 as suggested by our ionomycin experiment (Figure 6F). This increased interaction then leads to anterograde transport of HDAC5 towards the injury site, resulting in the formation of gradients of HDAC5 and tubulin deacetylation (Figure 6D,E) European Molecular Biology Organization 7

8 Finally, the schematic shown in Fig 8E should be a separate figure. Unless I missed it, the authors have not shown any data for a role of HDAC5 activity in developmental growth of axons. This schematic should be reformulated to reflect the data shown here. Alternatively, the authors need to clearly reference the data supporting this model for developing neurons. We agree with the reviewer and have revised our schematic model now shown in Fig 8F to strictly reflect the data presented. Minor points: The figure legends and points in the results section suffer from a lack of details. For example, what is 'I slope' in Fig 2G? What do the dashed arrows indicate in images for Fig 2 and later in manuscript? We agree and have revised figure legends and result section to include more detailed explanations. I am not sure what Fig 1I adds. We agree and have removed Fig 1I Could curve fitting be used for 1K? We agree that this could be an interesting additional analysis of the data in Fig. 1K, However, given that the manuscript is not solely focused on the temporal regulation of tubulin deacetylation following injury, we believe such analysis would go beyond the scope of the current manuscript. Furthermore, our in vitro data suggest that tubulin deacetylation may actually occur more rapidly than the data in Fig 1K indicates, yet our ability to detect this in vivo is limited by the spatial resolution of in vivo sciatic nerve preparation (the smallest fragment that can be analyzed biochemically being 3mm). For the discussion, the authors might also consider referencing work from Wu et al (2003) IOVS showing that PKC's role in regeneration of injury-conditioned RGCs. We thank the reviewer for this suggestion and have included the Wu et al 2003 reference in the discussion section. The discussion also mentions the role of microtubule stabilizing drugs in CNS axon regeneration but their data seem to argue against this mechanism. We thank the reviewer for raising this point. We have clarified the Discussion section to better explain the proposed effects of the microtubule stabilizing drug taxol on CNS axon regeneration, which is not solely mediated by its activity on microtubule on axons, but is also proposed to act on the surrounding glia. Further, whether the low amounts of taxol used in theses studies actually alter microtubule stability has not been determined. Referee #2 Overall: The manuscript by Cho and Cavalli provides very convincing data demonstrating that microtubule deacetylation occurs after axon injury, that this is due to HDAC5, and that control of MT acetylation levels is important for axon outgrowth. The authors have taken a multifaceted approach to the issue and have really done a lot of work to put the story together. Of course HDAC5 k/o mice would be ideal for looking at the response to sciatic nerve injury, but I assume this is not possible. Failing this, the authors have done a lot already to test their ideas all the way from isolated neurons in vitro, to reinnervation of endplates. I do wish that fewer of the experiments relied on scriptaid, but the authors do use more specific tools at the end. European Molecular Biology Organization 8

9 We agree with the reviewer that studies involving HDAC5 KO mice would be ideal, but we currently are not in the position to perform such experiments. General Comments: -Scriptaid is relied on for many of the expts, and it the only way acetylation is perturbed in in vivo regeneration assays. How much does this affect the transcriptional response to axon injury? Regeneration induces largescale transcriptional changes, and presumably HDACs are involved in these. This issue makes the in vivo regeneration experiments difficult to interpret. We absolutely agree with the reviewer that the effect of scriptaid on in vivo regeneration experiments cannot be solely interpreted in terms of tubulin deacetylation in the axon and we have revised the Results section to carefully address this point. However, it is important to note that the current literature on the role of HDACs in transcriptional regulation of axon regeneration indicates a repressive role of HDACs, suggesting that the effect we observed may arise from a distinct mechanism. Indeed, HDAC transcriptional activity represses axon regeneration in several types of CNS neurons (Riveccio et al 2009, Gaub et al 2010), and in DRG neurons grown on inhibitory substrates (Riveccio et al 2009), and HDAC inhibition in retinal ganglion cells enhances their survival but not axonal regeneration after optic nerve crush (Gaub et al Together with our results, these studies suggest that HDACs may play dual roles in axon regeneration (i) HDACs may be repressing a regeneration or survival program and HDAC inhibition can unlock gene expression to promote growth and/or survival following stress or injury (ii) HDAC activity is required locally in axons to promote local growth. -I am left unsure whether HDAC5 is involved in normal axonal growth or is specific to regeneration. Certainly scriptaid seems to inhibit both (For example figure 3A shows developmental outgrowth). Is HDAC5 more specific? This issue is not explored (8C and D may be developing axons, I can't tell from legend or text), although 8E shows a diagram of it at the growth cone in development. We absolutely agree with the reviewer that the data presented indicate that scriptaid blocks both axon growth and regeneration, but that HDAC5 regulates more specifically axon regeneration. Indeed, shrna mediated HDAC5 knock down, shown in Fig 8A and supplemental data S6, does not significantly affects DRG outgrowth (observed at DIV7, 4 days following knock down), but does affect regeneration following in vitro axotomy. Further, HDAC5 KO mice are viable and fertile and show no abnormalities at early ages (Chang et al. 2004), suggesting that HDAC5 does not pay a critical role in neuronal development. The experiment shown in Figure 8B-D was done with DRG neurons at DIV3, 2 days following knock down or overexpression. We have revised the Figure Legend accordingly. We have also revised the schematic diagram shown in Fig 8F to reflect the specific role of HDAC5 on axon regeneration suggested by the data. - The English in the methods section is not very good: missing articles, subject-verb disagreement etc. We thank the reviewer for this point and have revised the method section accordingly. -Quite a few of the conclusions in the results section seem overstated. We agree with the reviewer and have revised the results section to avoid overstatements. -I found the figure on phospho-hdac5 vs total HDAC5 (Figure 6) very confusing. As this is central to the paper, it is crucial that this is cleared up- see below for more. We absolutely agree and have revised the text as well as provided additional experiments to strengthen Figure 6D and 6E. Specific Comments: European Molecular Biology Organization 9

10 -Paragraph 1, reference to Figure 1 C- could you explain a little more the conclusion that this shows the deacetylation is mostly in axons? How can you tell that from the figure (this is probably obvious to someone who uses this system, but not necessarily to others). We appreciate the reviewer suggestion and have edited the text and figure legend to better explain the assay used and how conclusions on axonal effects can be drawn. -Description of Figure 1 G and H: Different distances from ligation site are mentioned. For reference, how far from the ligation site is the cell body? We have revised Figure 1G to indicate how far from the cell body the ligation site is located. -Scriptaid is introduced without a reference or much description. We have edited the text and provided reference for the specificity of Scriptaid. -What do the authors make of the decrease in acetylated tubulin distal to the injury site? This is even more dramatic than the proximal decrease, and is unlikely to be directly involved in regeneration. Is this also seen in CNS neurons? (this is an aside to the main story, so maybe just a little more discussion of it would be appropriate). We agree that the decrease in acetylated tubulin distal to the injury is indeed dramatic. As stated by the reviewer, events occurring distal to the injury are more directly linked to axon degeneration rather than regeneration. We have focused the current manuscript on the events proximal to the injury as we focus on axon regeneration. Although it would be very interesting to know if deacetylation occurs distal to an injury in the CNS, we believe that this goes beyond the scope of the current paper. -Figure 3D and E: This experiment is helpful, although I am not sure I would conclude that the effects of scriptaid are transcription-independent from it- it does suggest that targets are present in the axon, but does not address transcription directly. Because the previous expts on regeneration are done with the whole cell exposed to the drug, it would be really good to do the converse of the axon-drug addition here. I would love to see the drug just added to the cell body. Does this have no effect? A stronger effect? This might allow some better guesses at how much of the effects seen in the previous global experiments are due to nuclear vs local targets of scriptaid. We agree with the reviewer, and have edited the text related to Fig 3D and 3E to acknowledge the possibility of additional transcription-dependent roles of HDAC5. While we agree that adding scriptaid to the cell body compartment only would be interesting, this might not represent a large conceptual advance for the current manuscript. The emphasis of the current paper is on local roles of HDAC in the axon and a transcriptional role for HDACs has already been well described in the literature. Indeed, others have shown that HDAC transcriptional activity represses axon regeneration in different types of CNS neurons (for example Gaub et al 2010, Gaub et al 2011, Pelzel 2010) and in DRG neurons grown on inhibitory substrates (Riveccio et al 2009). Further, we believe that while our experiments do not exclude transcriptional roles for HDAC in axon regeneration, the experiment shown in Fig 3D using Campenot chamber strongly suggest an additional, local axonal role for HDACs. Indeed, the Campenot chambers used allowed us to add vehicle to the left axon chamber and scriptaid to the right axon chamber. The cell body compartment was left untreated and axon growth monitored on both chambers. Thus, in these conditions, if HDACs played transcription-dependent roles in outgrowth, they should be observed symmetrically in both chambers. Instead we found that growth was delayed only in the scriptaid treated chamber. European Molecular Biology Organization 10

11 While we are examining further the potential role of HDAC5 in transcriptional regulation following nerve injury, we believe that these studies are beyond the scope of the current manuscript. -I appreciate the rapid effects of the drug in the growth cone assay. One additional point that might be worth considering: I think adult DRGs have much smaller growth cones, and adult neurons in general have smaller growth cones. Do you think that HDAC effects could also be seen in these- as they are more the type of growth cone that might be expected during regeneration? We thank the reviewer for raising this point. We have not examined the effect of HDAC inhibition on adult DRG growth cones in vitro. We have started to examine growth cone morphology in vivo in control or scriptaid treated nerves and our preliminary data suggest that the growth cone morphology is affected by scriptaid. Unless the reviewer strongly feels otherwise, these data will be a part of a separate manuscript. -Is the culture assay in which DIV7 embryonic DRGs have their axons severed truly regenerative growth? Or is it simply a continuation of the active outgrowth already underway? This assay is truly a regenerative growth: the severed axon has to be transformed into a new growth cone. This involves the same steps that would occur for a severed axon in an adult animal, i.e. repair of the damage axon involving resealing of the plasma membrane and formation of a new growth cone. -Figure 4A: In the image shown for HDAC5 shrna, it looks like acetylation is globally increased (green is stronger even far from injury site). Is this true? Or is this image an outlier? It is stated that it is the injury-induced deacetylation that is blocked, but information about acetylation levels in uninjured neurons are required to make this conclusion. Are the values in B and C all normalized to take out this information? Or am I just missing how to get at it?- OK< I see this is shown in Figure S10. It might be worth selecting a different picture for 4A. We thank the reviewer for noting this; the image shown was an outlier. We have replaced the image to truly represent the quantification shown in Fig 5B. We have also emphasized in the accompying text to Figure 5B that global level of tubulin acetylation is unchanged by knock down of HDAC5 (shown in Figure S10). -Is PKC the only thing stimulated by PMA? If not, then the conclusion that phosphorylation of HDAC5 is PKC-dependent is a little strong for this data, although an inhibitor is used in a different experiment. We thank the reviewer and have revised the text to avoid overstating the conclusion on PKC in Figure 5, since PMA treatment likely affects other pathways in addition to PKC. -I am confused by Figure 6B. It shows in the upper row that phosphorylated HDAC is more abundant in the sciatic nerve after injury, but the bottom row shows that overall levels are unchanged. If this is true, what is being shown in D and E of this Figure? Is this really staining of phosphorylated HDAC5 and not total as indicated in the Figure, results text and legend? I also have a similar issue with the kinesin-1 experiment: isn't the point of the first part of the figure that phosphorylation of HDAC5 is the major response to injury? How does this relate to the analysis of total HDAC5 in the rest of the figure? We appreciate the reviewers comment and have provided additional data and explanation for Figure 6. The experiment shown in Fig 6B is done in vivo. Following a 2 hour sciatic nerve ligation, the level of phosphorylated HDAC5 in a 3mm segment proximal to the ligation is increased, while the total level of HDAC5 is unchanged. In Fig 6D, we present data showing that following a 3.5 hour axotomy in vitro, a gradient of HDAC5 and p- HDAC5 is formed at the injury site. The difference between in vivo and in vitro data European Molecular Biology Organization 11

12 likely arise from the sample size analyzed. While in vivo, a 3mm segment is analyzed, in vitro gradient formation is measured over only a 300µm segment directly proximal to the injury. We believe that the interaction of HDAC5 with kinesin-1 provides a mechanistic insight as to how a gradient of HDAC5 is formed after injury. Following injury, localized phosphorylation of HDAC5 in the axon can occur. Anterograde transport of HDAC5 from more proximal locations may assist in building the gradient of HDAC5 in injured axons. -Again, the conclusion at the end of Figure 6 seems overstated: "These data also indicate that unlike HDAC6, HDAC5 functions specifically in injured axons." No experiments relating to function were shown. We thank the reviewer and have revised the text accordingly. -missing reference: "To test whether microtubule stability was altered in injured nerves, we measured the steady-state level of microtubule polymerization, as previously described 45." -typos in Figure 7: I think labels in K and L should read unligated, not unligation We thank the reviewer for noting these typos and have edited the text accordingly. Discussion of fine balance of stability and instability of MTs was helpful. I also found the discussion of previous studies on HDACs and regeneration helpful. We thank the reviewer for this comment Referee #3 The paper addresses an area of considerable practical importance, the inability of axons in the CNS to regenerate due to unidentified factors leading to a low intrinsic regenerative ability. It presents a comprehensive set of experiments based around the hypothesis that tubulin modifications after axotomy are an important determinant of the success of axon regeneration. 1. Much of the first part of the paper depends on the use of the HDAC inhibitory scriptaid. The results are validated later by knockdowns of HDACs, but nevertheless it would be useful to have a statement on the specificity of spriptaid for HDACs, and other possible side effects or long term toxic effects. Some of the effects of longer term treatment that are shown could be due to non-specific toxicity. We thank the reviewer for these suggestions and have modified the text to include description of the specificity and toxicity of scriptaid and added the related reference. 2. The changes in tubulin acetylation that are described take 6 hours to occur. Yet there are various accounts in the literature describing in vivo and in vitro experiments demonstrating that peripheral nerve regeneration can occur much faster than this. The authors need to discuss demonstrate whether the changes in tubulin acetylation in axons occur before or after regeneration has started. We appreciate the reviewer s comment and have included a discussion on the timing of the observed events with relation to the onset of regeneration. The injury induced tubulin deacetylation and decreased microtubule stability that we observed in proximity of the injury site may mark the new end of microtubules and contribute to de novo growth cone formation. Indeed, previous studies suggested that microtubule destabilization strongly promotes plasma membrane resealing after axotomy (Xie & Barrett, 1991). Most axons within the spinal root reseal their damaged axon tip European Molecular Biology Organization 12

13 within 2 hours in vivo (Howard et al, 1999) and resealing of cultured rat septal neurons occurs within minutes (Xie & Barrett, 1991). Although we did not directly determined whether tubulin deacetylation occurs prior to plasma membrane resealing, our in vitro results suggest that tubulin modification occur early in the regeneration process, at the time when the growth cone need to be re-formed. 3. On page 8 it would be useful to know the percentage of axons that regenerate rather than the absolute number. We absolutely agree and now provide an additional panel (in Figure 8C) with the quantification of the percentage of regenerating axons. 4. The experiment on page 14, in which microtubule stability is assayed, will have been performed on an mixture of axonal and glial tubulin. Do the authors have information on the percent of tubulin that comes from axons in these extracts? Can they give arguments showing that we are looking at axonal changes? We agree with the reviewer that knowing how much tubulin derives from axons in the nerve is important for the interpretation of this experiment. We thus stained longitudinal sections of sciatic nerve with TUJ1 to label axons, S100β to label schwann cells and α- tubulin to label total tubulin. The relative intensity of α-tubulin co-localizing TUJ1- positive or S100β-positive area was calculated. This experiment showed that α-tubulin co-localizing with S100-positive area was less than 20% of the TUJ1-positive area. Thus, at least 80% of the tubulin in sciatic nerve derives from axons. This data is presented in Supplemental Figure S13. 2nd Editorial Decision 02 May 2012 Thank you for submitting your revised manuscript to the EMBO Journal. Your study has now been seen referees #1 and 2, and their comments are provided below. As you can see both referees appreciate the introduced changes and support publication in the EMBO Journal. Referee #1 has a few suggestions before acceptance here. Once we receive the revised version, we will proceed with the acceptance of the paper for publication here. Thank you for submitting your interesting study to the EMBO Journal! Yours sincerely Editor The EMBO Journal REFEREE REPORTS Referee #1 The authors have strengthened this work and overall addressed all of my major concerns. This new function for HDAC5 and signaling mechanism underlying this should be of broad interest to the Journal's readers. The contrast with HDAC6 and notion of a gradient of activity are particularly compelling. Also, the lack of any effect of Ca2+ on tyrosination and the discordinate acetylation and tyrosination is very intriguing (and seems to be different than seen previously). The few points that I offer below are for clarification. Results, page 6, sentence referring to acetyl-histone H3 - the authors should point out here that the acetylated histone used as a positive control for scriptaid effects on nerve is likely coming nonneuronal nuclei in the nerve. The general reader may not be familiar with peripheral nerve structure. Results section, page 8-9, referring to new images of the growth cones in scriptaid treated cultures - I think the authors could be more definitive in their description. These new images show the European Molecular Biology Organization 13

14 morphology much better and it appears that microtubules are invading the lamellapodia to a much higher extent with scriptaid treatment. Discussion section, page 17, referring to membrane sealing - work from the Fishman lab also showed that Ca2+ is needed for membrane sealing after severing neurites in differentiated PC12 cells (Detrait E et al. J Neurosci Res 62:566). Minor points: pg 5, next to last paragraph, last sentence - should read 'to peripheral axons and fails to occur' pg 8, 1st paragraph, sentence beginning 'The cumulative intensity of' - 'intensity of fluorescence intensity' could just read 'fluorescent intensity'. pg 17, 2nd paragraph, sentence beginning 'Although we did not' - 'determined' should be 'determine'. pg 20, 1st paragraph, sentence beginning 'In addition to these' - formatting for reference is off. Referee #2 I appreciate the reviewers' thoughtful responses and added clarification and data. I am enthusiastic about the manuscript! 2nd Revision - authors' response 02 May 2012 Please find enclosed our revised manuscript entitled "HDAC5 is a novel injury-regulated tubulin deacetylase controlling axon regeneration (EMBOJ R). We thank the reviewers for their detailed suggestions regarding a few clarification points. We have edited the text as suggested by the reviewer on pages 6, 8-9 and 17 to include these clarifications and to correct several typographic errors on pages 5,8,17 and 20. We hope that this version is now acceptable for publication. European Molecular Biology Organization 14