cdc42 switches irsp53 from inhibition of actin...

The EMBO Journal Peer Review Process File - EMBO-2013-84993

© EMBO 1

Manuscript EMBO-2013-84993 CDC42 switches IRSp53 from inhibition of actin growth to elongation by clustering of VASP Andrea Disanza, Sara Bisi, Moritz Winterhoff, Francesca Milanesi, Dmitry S. Ushakov, David Kast, Paola Marighetti, Guillaume Romet-Lemonne, Hans-Michael Müller, Walter Nickel, Joern Linkner, Davy Waterschoot, Christophe Ampè, Salvatore Cortellino, Andrea Palamidessi, Roberto Dominguez, Marie-France Carlier, Jan Faix and Giorgio Scita Corresponding author: Giorgio Scita, IFOM, Fondazione Istituto FIRC di Oncologia Molecolare, Milan, Italy Review timeline: Submission date: 22 March 2013 Editorial Decision: 28 March 2013 Appeal: 28 March 2013 Additional Editorial Correspondence: 03 April 2013 Editorial Decision: 23 April 2013 Revision received: 29 July 2013 Editorial Decision: 27 August 2013 Revision received: 29 August 2013 Accepted: 30 August 2013 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.) Editor: David Del Alamo

1st Editorial Decision 28 March 2013

Thank you for the submission of your research manuscript entitled "CDC42 switches IRSp53 from actin growth inhibition to elongation by clustering VASP through IRSp53". I have read and considered your study on the background of the related literature and discussed its suitability according to the scope of The EMBO Journal within our editorial team. I regret to say that the outcome is not a positive one. We certainly appreciate your comprehensive study of the mechanisms underlying the role of IRSp53 in the generation of filopodia through its interactions with the plasma membrane and VASP. Unfortunately, previous reports have already shown that IRSp53 associates with the plasma membrane through its BAR domain and that promotes filopodia formation. Furthermore, it is known to interact with VASP for this purpose and it has been shown to interact with other members of the ENA/VASP family in a Cdc42-dependent manner. Along these lines, Cdc42 has been also involved in IRSp53-dependent filopodia formation. Finally, the fact that filopodia are required in vivo for cell


© EMBO 2

migration and wound healing has also been established before. In conclusion, taking these precedents into consideration, while we appreciate that you study will definitely enlighten our understanding of the detailed mechanistic events connecting IRSp53 to VAST recruitment and filopodia formation, we believe that your manuscript does not provide the kind of general conceptual advance that we expect in a The EMBO Journal article and we feel it might be better suited to a more cell biology oriented publication. We have therefore decided not to proceed with the review process. Please note that this decision is by no means a reflection of the quality of your work. We subject to external review only a small percentage of the manuscripts we receive. I am sorry to disappoint you on this occasion and I hope for the rapid publication of your study somewhere else. Appeal 28 March 2013

Thank you for your reply.

As you can imagine, we are really disappointed about this decision because we sincerely believe that due to its conceptual advances in the signaling pathways leading to filopodia initiation and formation that should be of broad general interest to a wide audience, this manuscript perfectly fits EMBO J requirements. We would like to impose further on your time to point out the reasons that motivated us in sending this manuscript to EMBO J, and thus appeal to your decision. As you mention the link, at least in vitro, between between Cdc42::IRSp53 and IRSp53::MENA has been suggested earlier. However, our current work reveals a number of novel and unexpected biochemical and signaling mechanisms that conceptually advances the field in many respects. We show here for the first time that: 1) The three proteins CDC42::IRSp53::VASP can form a complex in vitro and in vivo. 2) IRSp53, in addition to its membrane deforming function, acts as a KINETIC CAPPER (a totally unexpected and novel finding). We also mapped this novel activity and show that it requires in addition to the I-BAR domain also an unusual region never shown before to participates in regulation of actin dynamics. The capping activity of IRSP53 is very intriguing making this protein not only able to bind and deforme the plasma membrane but to keep filament elongation at bay until stimuli relieves this inhibitory functions. 3) CDC42 by binding to IRSp53 prevents its negative activity on filament elongation (prevents its kinetic capping activity- to our knowledge, it has never been shown that a RhoGTPase can directly relieve or control capping). 4) CDC42 also promotes the interaction between IRSp53 and VASP thereby inducing a switch in biochemical function/activity, turning a weak kinetic capper into a processive actin filament elongating complex, which is crucial for spatial and temporal regulation of filopodia formation. This latter finding are intriguing and novel also in light of the structural advances provided by the accompanying manuscript showing that IRSp53 is regulated conformationally by CDC42 binding between two states: a closed one that caps and an open one that binds filaments elongators such as VASP. 5) In vitro and in vivo IRSp53 is also crucial for directing, clustering and thus activating, in a spatially restricted fashion, VASP to specific PM sites where filopodia are initiated. This finding again represent the first demonstration as to how the activity of the multifunctional proteins VASP is controlled to function specifically in filopodia as opposed to lamellipodia or focal adhesion. We further show that IRSp53 precedes VASP at leading edge thus providing a mechanisms temporal regulation of relevant complex assemblies at filopodia initiation sites 6) Finally, we show that this circuitry has important physiological consequences in cells as well as in whole organisms, as IRSp53 null cells are impaired in directional migration and invasion, and more relevantly IRSp53 null mice display defective tissue repair. We thus though that all these novel, unexpected elements combined provide the type of molecular and biological insights that significantly and conceptually advanced the general field of cell biology.


© EMBO 3

After all, we have been able to define and dissect at the molecular, cell biological and organismal levels the events required for the initiation of filopodia.

Frankly, I thought that even in papers of the EMBO J status one rarely observes a dissection of complete signaling pathways of crucial importance supported by the characterization of numerous, novel and unexpected biochemical and signaling activities. I would be therefore grateful if you could reconsider your decision and send the manuscript out for review.

Looking forward to hearing from you. Thanks for your attention. Additional Editorial Correspondence 03 April 2013

Thank you for your e-mail asking us for reconsideration of our previous decision. We have discussed your case again and, after further consultation with an external advisor, I am glad to inform you that we have decided to send your manuscript out for review. I will get back to you again with a decision as soon as I receive the reports from the referees. Do not hesitate to contact me is you have any further questions. 2nd Editorial Decision 23 April 2013

Thank you once again for the submission of your manuscript entitled "CDC42 switches IRSp53 from actin growth inhibition to elongation by clustering VASP through IRSp53" to The EMBO Journal. We have now received the full set of reports from the referees, which I copy below. As you can see from their comments, all four referees are rather positive and recommend the publication of your manuscript, provided their concerns are properly addressed. In general, they are convinced that the evidence presented properly supports your conclusions, although a number of technical concerns, particularly from referees #1 and #3, and the need for some clarifications have arisen. Although these concerns are explicitly mentioned in the referee reports and thus I will not repeat them here, I would like to draw your attention to a few important specific points. Besides several technical issues, referees #1 and #3 are concerned with the physiological significance of some of your in vitro experiments, most notably the nucleation ability of VASP in high salt concentrations, the presence or absence of profiling in certain experiments, or the use of a chimeric version of VASP. Taking these reports into consideration, I would like to invite you to submit a revised version of the manuscript. Please be aware that your revised manuscript must address the referees' concerns, experimentally if required, and their suggestions should be taken on board. It is 'The EMBO Journal' policy to allow a single round of revision only and, therefore, acceptance or rejection of your study will depend on the completeness of your responses included in the next, final version of the manuscript. 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 initiative, please visit our website: http://www.nature.com/emboj/about/process.html We generally allow three months as standard revision time. As a matter of policy, competing manuscripts published during this period will not be taken into consideration in our assessment of the novelty presented by your study. However, we request that you contact me as soon as possible upon publication of any related work in order to discuss how to proceed. Should you foresee a


© EMBO 4

problem in meeting this three-month deadline, please let us know in advance and we may be able to grant an extension. Thank you again for the opportunity to consider your work for publication in The EMBO journal. I look forward to your revision. Please, do no hesitate to contact me in case you have any further question, need further input or you anticipate any problem during the revision process.

REFEREE REPORTS:

Referee #1 (General Remarks): This paper by Disanza and coworkers addresses the mechanism of action of IRSp53 in conjunction with VASP and CDC42 in the formation of actin structures. The main idea is that IRSp53 localizes to and helps create membrane deformations, but inhibits actin polymerization there. CDC42 relieves this inhibition and additionally enhances recruitment of VASP by IRSp53. Filament growth is enhanced via VASP action to form protrusions like filopodia. This is a solid paper with a nice complementarity between in vitro and in vivo approaches. However some aspects of the in vitro work should be made more clear/convincing. 1) Figure 1 and 2 are confused by the emphasis on nucleation by VASP. Given its dependence on low salt concentration, actin filament nucleation by VASP is probably a non-physiological phenomenon-so why confuse the issue by characterizing it here? The legends are not clear, but most of this work was done at 50 mM KCl, a concentration where VASP has some residual nucleating activity. Does IRSp53 allow VASP to nucleate at real physiological salt conditions of 100 mM KCl or more? If not then the stuff about nucleation is distracting and not in-line with what the authors say about VASP's role in filipodia formation later on, which they attribute to barbed end elongation enhancement not nucleation. Pyrene assay is always open to interpretation while TIRF assays are more conclusive, especially when trying to prove something about barbed end elongation enhancement or inhibition. Since some of the co-authors are TIRF specialists, this reviewer doesn't understand why the pyrene results of Figure 1 were performed. Why weren't these tests done by TIRF, especially for showing elongation inhibition by IRSp53 and relief of this inhibition by CDC42? 2) How do the authors explain that IRSp53 inhibition of barbed end growth "developed slowly" Figure 1E? How could you explain this mechanistically? 3) What's the physiological interest in using VASP-DdGAB with mammalian IRSp53 in the TIRF assays in Figure 2B and 2D? Even if it's much less effective as a barbed end elongator, this study would be much more meaningful if these experiments were shown with WT mammalian VASP and not the completely artificial mammalian/Dictyostelium chimera. Or does the WT mammalian VASP really not give any barbed end elongation (as appears to be the case in Figure 2D, graph)? However in that case, what purpose does recruitment of VASP by IRSp53 really serve? 3) From the experiment shown in Figure 2D, the authors conclude that IRSp53 "clusters" VASP for barbed end elongation enhancement. From this experiment, you can't really say this since just simple recruitment would do the trick. If they want to emphasize the clustering, they need to do the experiment knocking out the IRSp53 dimerization site and show that it's required. Or show that IRSp53 allows processive barbed end elongation to be seen at a lower level of VASP than would normally be needed. 4) Phalloidin experiment in Figure 1C: This doesn't really tell us much. It would be better to follow this dynamically, either with fluorescent actin or by taking points over time and then doing the phalloidin staining. As it is (even though the authors start with G-actin) this could just be recruitment, not active assembly, of small actin oligomers via the F-actin binding capacity of VASP. If it's real polymerization, the F-actin cloud should grow over time. Overall Figure 1 and 2 are excessively complicated for saying: 1) IRSp53 on its own inhibits barbed end elongation 2) this is relieved by CDC42 3) VASP and IRSp53 form a complex and 4) this does not either interfere with or increase VASP's ability to perform barbed end elongation enhancement (although why in Figure 2B, bottom panel does IRSp53 seem to inhibit elongation at 200 nM VASP but not at 50 nM VASP)? As regards the in vivo work, this all seems solid, if perhaps not entirely new/surprising, since the


© EMBO 5

links between CDC42, IRSp53, VASP and filopodia are already well established, although this paper sheds light on the cross-talk between these different players. Referee #2 (General Remarks): In this manuscript, seventeen (!) authors present results in seven Figures, (comprising 34 panels, !!), five supplementary Figures (comprising 28 panels, !!!) and nine movies (!!!!), to convince the interested community of the important interplay of IRSp53 with CDC42 and VASP in generating filopodia on mouse cells. Thus, this manuscript is quite a challenge to the patience and time schedule of a reviewer. He not only has to judge upon the validity of the topic as well as of the quality of the data, but has to try to shape such a wealth of individual data panels into a coherent story - a task that should actually be performed by the authors prior to submitting such a manuscript. In the first part, this work comprises an extensive biochemical characterization of the interaction between purified IRSp53, CDC42, VASP and actin. This is the most elaborate part, and, notably, the title of the submitted ms only refers to this part. Results include (1) IRSp53, a lipid binding protein known to modulate membrane curvature, binds to and inhibits F-actin growth at the barbed end; (2) this inhibition is relieved by CDC42; (3) IRSp53 initiates clustering of VASP at the plasma membrane and initiates processive F-actin elongation. These in vitro results are interpreted to explain the observed filopodia formation at restricted sites and the relocalization of VASP to the filopodial tip. In the second part, using cells from a mouse strain with k.o. IRSp53, it is shown that IRSp53 is essential for filopodia formation that is correlated with directional cell migration and cell invasion. The third part demonstrates that wound healing is impaired in mice without IRSp53. The import role of IRSp53 for these processes is certainly an interesting and significant issue. However, there are major shortcomings and serious flaws in this ms which will have to be addressed in detail. First, the fusion of the biochemical part (1) together with the cell biological (2) and the pyhsiological (3) parts into one paper is not really plausible. For example, all biochemical assays are apparently performed with muscle actin, not with cytoplasmic actin, and with recombinant VASP which is not phosphorylated, as mammalian VASP is in tissues and cells. Thus, binding affinities etc. cannot directly lead to conclusions of the situation in cells and tissues. A separate paper on the biochemical data would still allow for general conclusions on the interplay between the candidate proteins, without imposing such an ill justified link. Second, in all three parts, there are inconsistencies in the data presented and experiments missing. Just some selected examples: Fig. 1, panels A and B: Why does the same VASP concentration (300 nM) give different fluorescence values at 400 sec.? Why are the murine cells from IRSp53 null and rescued mice sometimes plated on laminin (Fig. 4b-d and movies S2-5), in other experiments on fibronectin (Fig. 5d)? In the in vitro wound healing assay: was cell proliferation inhibited (for example by mitosis blockers) or not? What is the doubling time? In wound healing in mice: Why can endothelial cells apparently migrate without delay, even in the absence of IRSp53? Last but not least: the entire ms is not carefully edited, there are numerous typing and spelling errors and incomplete sentences. Example: Results, {section sign} 5, heading. Referee #3 (General Remarks): The manuscript by Disanza et al describes the cooperation of IRSp53 with VASP, a regulator of F-actin extension, in filopodia formation. The authors report that IRSp53 functions as a weak F-actin barbed end capper and this function is negatively regulated by CDC42. In turn, CDC42 positively regulates the interaction of IRSp53 with VASP and the authors provide evidence that IRSp53 is required for CDC42 induced filopodium formation. In general the experiments have been performed with great care. The authors and others had previously shown that the SH3 domain of IRSp53 interacts with VASP and Mena. It had been already shown that CDC42 induces filopodia by promoting the formation of an IRSp53:Mena complex (Krugmann et al Curr Biol 2001). The authors have published that a complex of IRSp53-VASP functions in filopodia formation by supporting actin filament bundling (Vaggi et al PLOS Comp Biol). Nevertheless, this referees believes that the function of an IRSp53-Mena/VASP complex in filopodia formation is not understood and some of the older experiments were not


© EMBO 6

properly quantified or controlled. In general I would support a publication in the EMBO Journal if the molecular mechanism of the IRSp53-Mena/VASP function in filopodia formation would be clarified. The authors can be more upfront with prior knowledge and should address and use the prior knowledge throughout the manuscript. Furthermore, the following concerns needs to be addressed before I can recommend publication. Major concerns:

1. Page 11: The authors claim that: "The formation of a VASP::IRSp53 complex has no effect on the filament elongation rate of VASP but promotes its actin nucleation activity." This claim is not supported by the data shown: The polymerization kinetics of pyrenyl actin in Fig 2A in the presence of VASP and profilin are identical with or without IRSp53 indicating that, in the presence of profilin, IRSp53 does not increase the activity of VASP. In contrast, without profilin IRSp53 increases the activity of VASP (Fig 1B). In Fig 2B the authors chose to use a mutant of VASP with a Dictyostelium VASP G-Actin binding site, which makes the comparison with human VASP very difficult. Furthermore, the assay in Fig 2B was performed without profilin whereas the experiment in Figure 2C was performed with full length VASP and profilin. Comparing Fig 2A and 2C, which have comparable assay conditions, suggests that according to Fig 2A there is no difference in VASP activity and the observed increase in filament numbers in Fig2C might arise from the setup of the experiment that IRSp53 was coated on the slide thereby maybe allowing recruitment and clustering of VASP. 2. In general the authors chose to include profilin in some experiments and not others without explanation. Profilin has an important role for VASP function and therefore it is important that when experiments are compared that this is taken into account. 3. Fig. 2D: The authors decided to use a mutant of VASP with a Dictyostelium VASP G-Actin binding site, which makes the comparison with human VASP very difficult. It would be better to repeat the experiments with human VASP (alternatively the experiments in Fig 1 and S1 should be repeated with VASP DdGAB). Measuring the elongation rates of just 10 filaments is only meaningful if the differences observed show statistically significant differences. Statistics should be performed and most likely more filaments needs to be measured. 4. Fig 3B Was IRSp53 on beads? (not stated in the Figure legend)? Is this increase in IRSp53/VASP binding statistically significant? 5. Fig 3: To understand the molecular mechanism of the IRSp53-VASP complex better, the authors should perform pyrene-actin polymerization assays with IRSp53-VASP and Cdc42-GTP or Cdc42-GDP and this should be compared with Eps8-IRSp53-VASP with Cdc42-GTP or Cdc42-GDP. 6. Fig 6A How were the cells selected for analysis if only 7 cells per movie were quantified? An analysis of relative reduction in wound area should be performed in addition. Page 16: The claim that null MEFs are impaired in extending filopodia in the Movie S9 is not obvious. 7. Fig 6B-C: Boyden chamber assays are problematic especially if not properly controlled. The number of cells that appear on the lower side of the membrane depends on the number of cells seeded in each experiment. Even if exactly the same numbers are seeded, a defect in adhesion could be interpreted as a defect in invasion. At least the numbers of cells adhering to the upper chamber should be quantified as control. Better less problematic invasion assays are available and should be used. 8. Fig. 7: The SEM error bars are huge. Is the p value of the t-test really p<0.001? The conclusions from the data are only valid if the differences are significant. 9. The conclusion sentence on page 17 is overstated since it has been only shown that macrophage numbers are significantly lower in the wounds of IRSp53 KO mice. 10. The findings of this manuscript and prior knowledge on EPS8-IRSp53 should be better integrated. In particular, the data that CDC42 reduces the binding between IRSp53 and EPS8 (mentioned in the discussion) should be shown.


© EMBO 7

11. The results of this paper and the published results of Krugmann et al Curr Biol 2001 should be introduced in the introduction. The key functional experiments should be repeated for Mena to explore whether the cooperation with IRSp53 is a common feature of the Ena/VASP protein family and the data should be compared and contrasted with the published data in the discussion. 12. The authors should address how the results from their previous publication on IRSp53-VASP-EPS8 (Vaggi et al PLOS Comp Biol) fits with the here presented findings. In particular, in that publication the authors concluded that a complex of IRSp53-VASP functions in filopodia formation by supporting actin filament bundling. Minor concerns:

1. Please add molecular weight makers on all blots. 2. Fig 4 C What defines a cluster? Is this a subjective analysis or was the cluster defined by a certain intensity threshold over background? 3. Fig S3D: Is this a site of filopodia initiation? The cluster shown is rather invaginated. The movie S7 is much better: the localization at the tips of filopodia should be shown. 4. Fig S3E missing (Fig S3E is mislabelled and should read FigS3F. 5. FigS4C Were all cells with at least one filopodia tip positive for VASP counted? 6. Fig 5C Why has IRSp53 a higher molecular weight in the WT MEFs compared to the reconstituted MEFs? Referee #4 (Remarks to the Author): Summary: The authors report that IRSp53 inhibits actin polymerization at barbed ends, and that this inhibitory activity is relieved by VASP or Cdc42. They also find that the IRSp53 SH3 domain binds to VASP, map the binding site, and find that binding is enhanced in the presence of activated Cdc42. With regard to cellular activities, they find that IRSp53 and VASP accumulate in foci at the leading edge that may represent sites of filopodia initiation. Using cells from IRSp53 null mice, they show that IRSp53 is important for VASP focus formation, filopodia formation/stability, and cell motility. IRSp53 is also important for rapid wound healing in mice. General comments:

In general this is an interesting paper that details an impressive amount of work and sheds new light on the biochemical and cell biological functions of IRSp53 and VASP. The paper is suitable for publication in EMBO J if the authors address the specific points listed below. Specific comments: 1. Page 8: There are unformatted references that need to be correctly formatted. 2. Page 11: The authors state that "bucking actin filaments grew away from the bead surface..." To this reviewer, buckling implies that both ends of the filament are fixed in place. Instead, in the video it appears that the end of the actin filament facing away from the bead is free in solution. Thus, it seems the filaments are "bent" or "curved" rather than "buckling". 3. Figures 2B, 2C, 2D, 3B: The authors should note whether the differences they observe are statistically significant, as they do in other figures. 4. Movies S2-S6: The authors state that the formation of GFP-VASP foci preceded filopodia formation. However, very few filopodia are formed in these movies. It may be valuable to observe the frequency with which clusters mature into filopodia. 5. Page 14 and movie S8: The authors mention that mCherry-IRSp53 accumulates in foci about 1.4 s before GFP-VASP. However, in the images and movies, accumulation of the two seems simultaneous. They need to better document this to clearly establish the order of recruitment. 6. Figure 5A: The authors note that IRSp53 "decorates the shaft of filopodia protrusions with a


© EMBO 8

punctate distribution". However, IRSp53 seems to be everywhere in the cell with a similar distribution, and it does not appear to specifically localize to filopodia. This should be noted. 7. Figure 7B, C: These images should be more clearly and extensively described in the text. 1st Revision - authors' response 29 July 2013

Below is a point-by-point reply to the editor’s and referees’ comments.

Editor

As you can see from their comments, all four referees are rather positive and recommend the publication of your manuscript, provided their concerns are properly addressed.

In general, they are convinced that the evidence presented properly supports your conclusions, although a number of technical concerns, particularly from referees #1 and #3, and the need for some clarifications have arisen. Although these concerns are explicitly mentioned in the referee reports and thus I will not repeat them here, I would like to draw your attention to a few important specific points. Besides several technical issues, referees #1 and #3 are concerned with the physiological significance of some of your in vitro experiments, most notably the nucleation ability of VASP in high salt concentrations, the presence or absence of profiling in certain experiments, or the use of a chimeric version of VASP.

Taking these reports into consideration, I would like to invite you to submit a revised version of the manuscript. Please be aware that your revised manuscript must address the referees' concerns, experimentally if required, and their suggestions should be taken on board. It is 'The EMBO Journal' policy to allow a single round of revision only and, therefore, acceptance or rejection of your study will depend on the completeness of your responses included in the next, final version of the manuscript.

Thank you very much for your letter and the possibility to submit a revised version of our manuscript.

We have tackled all referees’ comments energetically, and we are now submitting a thoroughly revised and improved manuscript. We have addressed each of the referees’ comments either experimentally or by providing the requested clarifications. We have devoted particular effort in providing additional support of the physiological relevance of some of our in vitro experiments, employing both human VASP and the closely related family member EVL in combination with IRSp53-coated beads in TIRF-based assays. Following the referees’ suggestions, we have also removed data on the effects of IRSp53 on VASP nucleation, which were obtained using 50 mM KCl, in line with previously published experiments by a number of laboratories. These experiments were initially carried out to provide evidence that the IRSp53:VASP interaction has functional and biochemical consequences. However, given the lack of consensus regarding the physiological relevance of these assays, we agree with the reviewers that removal of this set of data and the addition of new actin:spectrin seed bulk and single filament elongation TIRF assays, strengthens and simplifies the structure of manuscript without compromising the quality and the novelty of the message.

We have also performed an additional set of cell biological experiments to clarify various minor issues raised by the referees.

Referee #1 (General Remarks):

This paper by Disanza and co-workers addresses the mechanism of action of IRSp53 in conjunction with VASP and CDC42 in the formation of actin structures. The main idea is that IRSp53 localizes to and helps create membrane deformations, but inhibits actin polymerization there. CDC42 relieves


© EMBO 9

this inhibition and additionally enhances recruitment of VASP by IRSp53. Filament growth is enhanced via VASP action to form protrusions like filopodia.

This is a solid paper with a nice complementarity between in vitro and in vivo approaches.

We thank this reviewer for the positive evaluation of our manuscript

However some aspects of the in vitro work should be made more clear/convincing.

1) Figure 1 and 2 are confused by the emphasis on nucleation by VASP. Given its dependence on low salt concentration, actin filament nucleation by VASP is probably a non-physiological phenomenon-so why confuse the issue by characterizing it here? The legends are not clear, but most of this work was done at 50 mM KCl, a concentration where VASP has some residual nucleating activity. Does IRSp53 allow VASP to nucleate at real physiological salt conditions of 100 mM KCl or more? If not then the stuff about nucleation is distracting and not in-line with what the authors say about VASP's role in filipodia formation later on, which they attribute to barbed end elongation enhancement not nucleation.

As correctly pointed out by the referee and shown in supplementary Fig. 2a of the old version of the manuscript, VASP polymerization activity at 100 mM KCL is minimal, but clearly detectable at 50 mM KCL, in agreement with previous reports (Barzik et al, 2005; Hansen & Mullins, 2010; Pasic et al, 2008; Reinhard et al, 1995). Initially, we wanted to test whether IRSp53 altered any of the ascribed activities of VASP to give an indication of the biochemical and functional relevance of the interaction. We agree, however, that this set of data may confuse and dilute the overall message of the manuscript, which provides compelling mechanistic evidence that IRSp53 is essential to mediate the recruitment of VASP to solid supports and lipid membranes and to enhance VASP filament elongation activity for filopodia formation. Thus, following this referee’s suggestion and for the sake of clarity, we performed additional TIRF assays (see below), and removed all data related to IRSp53 effects on VASP nucleation.

Pyrene assay is always open to interpretation while TIRF assays are more conclusive, especially when trying to prove something about barbed end elongation enhancement or inhibition. Since some of the co-authors are TIRF specialists, this reviewer doesn't understand why the pyrene results of Figure 1 were performed. Why weren't these tests done by TIRF, especially for showing elongation inhibition by IRSp53 and relief of this inhibition by CDC42?

We agree that real-time visualization of actin filament elongation by TIRF assays is meaningful, and we have now used these assays to support the contention that recruitment of VASP to IRSP53 coated beads promotes processive filament elongation despite the presence of capping proteins. However, the bulk solution barbed end growth assays are more informative, more accurate and less laborious than TIRF assays, which are mainly illustrative. Indeed, the bulk solution assays were essential to: i) demonstrate the kinetic inhibitory effect of IRSp53 at barbed ends and the inhibition of depolymerization; ii) measure the binding affinities of proteins to barbed ends; iii) establish the differences between full-length IRSp53 and its mutated form. These bulk solution assays preceded and motivated the TIRF assays and allowed us to setup the appropriate conditions for TIRF observations.

We have now also included an entirely novel set of TIRF experiments in which the inhibition of filament elongation by IRSp53 is shown (Fig. 1g of the revised manuscript), thus providing further evidence that IRSp53 slows down filament elongation. This latter notion is reinforced by the findings that: i) the steady state amount of assembled actin remains unaffected by IRSp53 at the concentration used (Fig. S2e); ii) a mutant of IRSp53 (IRSp53-1-374) retaining the minimal surfaces for barbed end inhibition, but lacking the entire SH3 domain (and thus presumably adopts an open conformation) instantaneously inhibited barbed end growth (Fig. 1f of the revised manuscript), similarly to IRSp53-W413G that lacks a functional SH3 domain (see also our reply to point 2 below). Additionally, using the TIRF assays, we can confirm that addition of active CDC42 relieves the inhibitory effect of IRSp53 on barbed end elongation (Fig. S2f of the revised manuscript).


© EMBO 10

2) How do the authors explain that IRSp53 inhibition of barbed end growth "developed slowly" Figure 1E? How could you explain this mechanistically?

We thank the referee for highlighting this issue. Several possible explanations could account for the observation that IRSp53 inhibition of barbed end growth developed slowly. The ones we favour, and which are supported by our experimental observations, are that IRSp53 needs to undergo a “slow” conformational change for efficient binding to barbed ends, resulting in slow binding and thus “slow” inhibition of filament growth, or IRSp53 undergoes a slow conformational change following its binding to barbed ends, which results in “slow” inhibition of filament growth. In support of these models, Roberto Dominguez and colleagues have very recently shown that IRSp53 can adopt a closed conformation via an intramolecular interaction between the SH3 domain and a proline-rich stretch within an extended CRIB-PP motif (Kast et al., NSMB, under revision). This closed conformation is in constant slow equilibrium with the open conformation due to the relatively low affinity of the intramolecular interaction, and can be modulated either by CDC42 or by SH3 interactors, such as EPS8 (Kast et al., NSMB, under revision). Consequently, WT IRSp53 displays a similar affinity for actin-barbed ends as IRSp53 mutants impaired in the intramolecular interaction: IRSp53-W413G that lacks a functional SH3 domain (Fig. S1b, S1f of the revised manuscript) and IRSp53-1-374, which is devoid of the entire SH3 domain. However, the latter mutants are faster at inhibiting barbed end growth. We included these observations in Fig. 1f, Fig. S2a-b and on pg. 9-10 of the revised manuscript.

It is also possible that IRSp53 binds with higher affinity to ADP-barbed ends than to ATP-barbed ends. This possibility is supported by the fact that the capping by IRSp53 was immediate (no lag) in dilution-induced depolymerization assays (Fig. 1c of the revised manuscript), where ADP-F-actin is exposed at barbed ends.

Notably, the TIRF assays monitoring single filament growth in the absence or presence of IRSp53 also support the notion that IRSp53 is a kinetic or leaky capper that slows down actin polymerization, at the concentrations used. The TIRF results in Fig. 1g of the revised manuscript indicate that IRSp53 slows down barbed end growth by about 30% at 0.5 -1 µM, while it causes ~80% inhibition at 0.58 µM in bulk polymerization assays. It must be pointed out, however, that the TIRF data were collected in the early phase of filament elongation (between 0 and 400 s), when partial inhibition of growth was also observed in bulk polymerization actin spectrin seed assays. These data further support the slow kinetics of IRSp53 binding to filament barbed ends. Additionally, to be able to detect more extensive inhibition of barbed end growth, we would need to reach concentrations of IRSp53 much higher than those we can achieve, given the relatively low Kd (~0.3 µM) of IRSp53 for barbed ends. The low affinity of IRSp53 for barbed ends also suggests that, at variance with other cappers, the t1/2 of IRSp53 binding to filaments is likely to be shorter, possibly allowing growth of barbed ends that are only transiently occupied by IRSp53. We included these comments on pg. 9-10 of the revised manuscript

3) What's the physiological interest in using VASP-DdGAB with mammalian IRSp53 in the TIRF assays in Figure 2B and 2D? Even if it's much less effective as a barbed end elongator, this study would be much more meaningful if these experiments were shown with WT mammalian VASP and not the completely artificial mammalian/Dictyostelium chimera. Or does the WT mammalian VASP really not give any barbed end elongation (as appears to be the case in Figure 2D, graph)? However in that case, what purpose does recruitment of VASP by IRSp53 really serve?

This is a good question. The physiological actin concentration in cells is typically in the range of a few hundred µM (Koestler et al, 2009). The in vitro TIRF assay, however, has to be conducted in non-physiologically low actin concentrations, because at concentrations greater than 4 µM, the growing filaments would instantly cover the visible area and hamper all analyses. Thus, researches are forced to work with non-physiological G-actin concentrations of about 1 µM in the TIRF assay. We have previously shown that the G-actin-binding (GAB) site of mammalian VASP contains a rather low affinity binding site compared to the GAB of Dictyostelium VASP (Breitsprecher et al, 2011). Thus, mammalian ENA/VASP proteins are poorly charged with G-actin (~20%) at 1 µM G-actin, while ENA/VASP proteins containing the Dictyostelium GAB are almost fully charged (~80%) at the same concentration. Since the elongation rate is directly correlated with the G-actin load factor of ENA/VASP proteins (Breitsprecher et al, 2011), only the chimera achieves, at the low actin concentration used in the TIRF assay, the high filament elongation rates one would expect to see


© EMBO 11

with mammalian wild type ENA/VASP proteins in the presence of physiological actin concentrations. Thus, the hVASP-DdGAB chimera likely represents the most physiological readout possible in this assay, by combining two admittedly non-physiological parameters (DdGAB and low actin). Of note, no changes other than the replacement of the respective GABs were made in the hVASP chimera. For these reasons, we would like to retain these data in the revised version of the manuscript. Furthermore, we now show that similar to human VASP, IRSp53 is also able to recruit human EVL to the bead surface to drive processive (but slower) filament elongation in the presence of capping protein, as indicated by buckling of the elongating filaments (see movie S1). In line with this data, spontaneous growth of actin filaments in other parts of the glass cover slip is virtually not seen, because the capping protein rapidly terminates spontaneous filament growth in solution (Breitsprecher et al, 2008; Breitsprecher et al, 2011). We included these findings in Fig 2a-b of the revised manuscript and on pg.12 of the main text.

4) From the experiment shown in Figure 2D, the authors conclude that IRSp53 "clusters" VASP for barbed end elongation enhancement. From this experiment, you can't really say this since just simple recruitment would do the trick. If they want to emphasize the clustering, they need to do the experiment knocking out the IRSp53 dimerization site and show that it's required. Or show that IRSp53 allows processive barbed end elongation to be seen at a lower level of VASP than would normally be needed.

We agree with the reviewer and have reworded the conclusions derived from the result illustrated in this figure. Indeed, it is nearly impossible to dissect whether increased barbed end elongation is due to forced molecular recruitment (crowding effect), mediated by IRSp53 on limited surfaces and/or to the formation of high order cluster of VASP onto these beads. We have previously shown that inducing molecular crowding by coating carboxylated latex bead with VASP is sufficient to promote processive filament elongation, as long as a sufficiently high density of VASP on the beads is achieved (Breitsprecher et al, 2008; Breitsprecher et al, 2011). However, this assay has a rather limited dynamic range, preventing us from unequivocally determining whether a dimeric-impaired IRSp53 that retains the same binding affinity for VASP as WT IRSp53 is significantly less efficient than the latter in promoting VASP crowding onto beads (or is quantitatively more efficient with respect to uncoated latex beads onto which VASP associates). Similarly, it would be unreliable to determine unequivocally whether lower concentrations of VASP are required to observe processive filament elongation when IRSp53-coated beads are used.

Notwithstanding this, we would like to point out that our results indicate that:

i) IRSp53 induces high order VASP oligomerization in solution. This ability strictly depends on a functional IRSp53::VASP interaction.

ii) In vivo, the formation of VASP foci at the PM is impaired by genetic removal of IRSp53: a phenotype restored by expression of WT IRSP53, but not by IRSp53W413G.

iii) VASP foci are sites where the concentration of VASP locally increases and invariably precedes the formation of filopodia (Fig. 4, Fig. S4c-d and Movies S7 and 8 of the revised manuscript). Thus, the IRSp53-dependent increase in the local concentration of VASP at focal points is necessary for subsequent elongation of actin filaments into filopodia, corroborating the validity of our contention.

5) Phalloidin experiment in Figure 1C: This doesn't really tell us much. It would be better to follow this dynamically, either with fluorescent actin or by taking points over time and then doing the phalloidin staining. As it is (even though the authors start with G-actin) this could just be recruitment, not active assembly, of small actin oligomers via the F-actin binding capacity of VASP. If it's real polymerization, the F-actin cloud should grow over time.

We agree that this experiment cannot be unequivocally interpreted as promotion of actin assembly and therefore removed it as suggested. We replaced the figure with data from an experiment in which we examined individual filament elongation by TIRF microscopy.

Overall Figure 1 and 2 are excessively complicated for saying: 1) IRSp53 on its own inhibits barbed end elongation 2) this is relieved by CDC42 3) VASP and IRSp53 form a complex and 4) this does not either interfere with or increase VASP's ability to perform barbed end elongation enhancement.


© EMBO 12

We agree with the reviewer and significantly rearranged and simplified the manuscript by removing most of the experiments analysing the nucleation activity of VASP at low salt concentrations. Additionally, we added TIRF assays to quantify inhibition of barbed end elongation and processive elongation activity.

(although why in Figure 2B, bottom panel does IRSp53 seem to inhibit elongation at 200 nM VASP but not at 50 nM VASP)?

The most likely reason for the decreased rate of polymerization at 200 nM VASP (now shown in Fig. S3b of the revised manuscript) is that under this condition there is a significant increase in filament number and, thus, of nucleation events. This results in a decrease in the amount of G-actin available for elongation and, consequently, an apparent decrease in the rate of elongation. We commented on this finding on pg. 11 of the revised manuscript.

As regards the in vivo work, this all seems solid, if perhaps not entirely new/surprising, since the links between CDC42, IRSp53, VASP and filopodia are already well established, although this paper sheds light on the cross-talk between these different players.

We are pleased that the in vivo work is appreciated and agree that CDC42, IRSp53, VASP have been shown to be involved in filopodia. However, our data additionally shows that IRSp53 is essential for the recruitment VASP to PM foci for filopodia initiation and that this occurs in a CDC42-dependent manner. Furthermore, we provide evidence that IRSp53 impairs directional cell migration (not novel), invasion and wound repair in mice (novel), through its ability to initiate filopodia.


In this manuscript, seventeen (!) authors present results in seven Figures, (comprising 34 panels, !!), five supplementary Figures (comprising 28 panels, !!!) and nine movies (!!!!), to convince the interested community of the important interplay of IRSp53 with CDC42 and VASP in generating filopodia on mouse cells. Thus, this manuscript is quite a challenge to the patience and time schedule of a reviewer. He not only has to judge upon the validity of the topic as well as of the quality of the data, but has to try to shape such a wealth of individual data panels into a coherent story - a task that should actually be performed by the authors prior to submitting such a manuscript.

In the first part, this work comprises an extensive biochemical characterization of the interaction between purified IRSp53, CDC42, VASP and actin. This is the most elaborate part, and, notably, the title of the submitted ms only refers to this part. Results include (1) IRSp53, a lipid binding protein known to modulate membrane curvature, binds to and inhibits F-actin growth at the barbed end; (2) this inhibition is relieved by CDC42; (3) IRSp53 initiates clustering of VASP at the plasma membrane and initiates processive F-actin elongation. These in vitro results are interpreted to explain the observed filopodia formation at restricted sites and the relocalization of VASP to the filopodial tip.

In the second part, using cells from a mouse strain with KO IRSp53, it is shown that IRSp53 is essential for filopodia formation that is correlated with directional cell migration and cell invasion. The third part demonstrates that wound healing is impaired in mice without IRSp53.

We apologize to this referee for the complexity of the manuscript. Indeed, the work we presented is complex as it includes in vitro reconstitution, and cell and mouse biology experiments. These experiments required the combined expertise of a number of researchers who must obviously be acknowledged by authorship. To simplify the manuscript, and following referee #1’s suggestion, we have extensively revised our work and simplified the biochemical part emphasizing: 1) the role of IRSp53 in slowing down filament elongation; 2) the ability of CDC42 to relieve this inhibition; 3) the ability of VASP and IRSp53 to form a complex with CDC42; 4) the fact that IRSp53 does not


© EMBO 13

impair VASP elongation activity, but mediates its recruitment to PM and its accumulation into clusters (foci) where filopodia are initiated.

The import role of IRSp53 for these processes is certainly an interesting and significant issue. However, there are major shortcomings and serious flaws in this ms which will have to be addressed in detail. First, the fusion of the biochemical part (1) together with the cell biological (2) and the physiological (3) parts into one paper is not really plausible.

For example, all biochemical assays are apparently performed with muscle actin, not with cytoplasmic actin, and with recombinant VASP, which is not phosphorylated, as mammalian VASP is in tissues and cells. Thus, binding affinities etc. cannot directly lead to conclusions of the situation in cells and tissues. A separate paper on the biochemical data would still allow for general conclusions on the interplay between the candidate proteins, without imposing such an ill justified link

Regarding this specific concern, we remain convinced that a thorough biochemical characterization should be complemented by relevant cell biological assays and attempts to validate results in complex organisms. We therefore tried to provide interpretations and build a model that best fitted all the results and observations. We would have preferred to present the data as separate manuscripts, but this is strongly discouraged by the current strategies and guidelines of nearly all-scientific journals. To increase the clarity of the manuscript we have tried to clarify any shortcomings with experimental design and apparent over interpretations of results.

Regarding the specific issues on the use of muscle actin and on the impact of VASP phosphorylation status on its activity:

-Muscle actin has been used in hundreds of actin biochemical experiments. Additionally, the biochemical properties of muscle and cytoplasmic actin have been shown to be nearly identical in most cases (Tsukada et al, 1987), justifying the use of this reagent. Finally, in a vast number of recent studies performed to explore the role of VASP, muscle actin has invariably been used [example papers: (Applewhite et al, 2007; Barzik et al, 2005; Bear & Gertler, 2009; Breitsprecher et al, 2011; Dominguez, 2007; Ferron et al, 2007; Hansen & Mullins, 2010; Mejillano et al, 2004; Menzies et al, 2004; Pasic et al, 2008; Schirenbeck et al, 2006)]. Thus, to be able to directly compare our data with previously published results, we also employed muscle actin.

-VASP is indeed regulated by phosphorylation. However, this posttranslational modification does not seem to affect some of its key biochemical activities as shown in our manuscript and in a previous publication (Breitsprecher et al, 2008). For instance, phosphorylation of recombinant VASP in vitro had no effect on actin filament elongation in the TIRF assay (Breitsprecher et al, 2008). We do agree, however, that this post-translational modification introduces an additional level of complexity in the regulation of ENA/VASP that would need to be taken into account and specifically investigated. Doing so, however, would require performing a much larger number of experiments that, we feel, are beyond the scope of the present manuscript and if carried out would render the current work even more complicated. Moreover, in a number of recent publications on VASP (Applewhite et al, 2007; Barzik et al, 2005; Bear & Gertler, 2009; Breitsprecher et al, 2011; Dominguez, 2007; Ferron et al, 2007; Hansen & Mullins, 2010; Mejillano et al, 2004; Menzies et al, 2004; Pasic et al, 2008; Schirenbeck et al, 2006), bacterially produced, unphosphorylated VASP has been systematically used. It was therefore, necessary to use similar conditions for meaningful comparisons with the published literature.

Second, in all three parts, there are inconsistencies in the data presented and experiments missing.

We deeply respect the reviewer’s opinion, yet we are puzzled as to what we could do to address this issue. On the one hand, we are advised to separate the biochemical story from the cell biological and in vivo part; on the other we are asked to include more experimental work in a variety of different directions. Once again, we apologize if we failed to concisely convey the core of our finding and hope that the revision of the data and their reorganization is a significant step forward


© EMBO 14

in the direction suggested by this reviewer.

Just some selected examples:

Fig. 1, panels A and B: Why does the same VASP concentration (300 nM) give different fluorescence values at 400 sec.?

The absolute fluorescent of the pyrenyl assays changes from experiment to experiment primarily due to different efficiencies in pyrenyl labelling of actin and/or slightly different concentrations of labelled actin used in the assays. Indeed, experiments shown in Fig. 1a and b (old version of the manuscript) were performed with different batches of pyrenyl actin. Thus, the absolute fluorescent value can change, but the effects on the actin polymerization dynamics are comparable and meaningful. This is why in each of these experiments various positive and negative controls (e.g. actin alone, VASP alone, IRSp53 alone) are systematically introduced.

Having said this and as advised by this referee and referee #1, we have removed the experiments on the effects of IRSp53 on nucleation activity of VASP (see reply to referee #1) and restructured the part on the biochemical analysis of IRSp53 effects on VASP.

In the in vitro wound healing assay: was cell proliferation inhibited (for example by mitosis blockers) or not? What is the doubling time?

We measured the doubling time of IRSp53 KO and IRSp53 KO MEFs reconstituted with IRSp53. We found no difference in the doubling time between the various genotypes. We added this information in Fig. S6a of the revised manuscript.

Why are the murine cells from IRSp53 null and rescued mice sometimes plated on laminin (Fig. 4b-d and movies S2-5), in other experiments on fibronectin (Fig. 5d)?

Laminin is known to promote filopodia formation (Dent et al, 2007; Pertz, 2011), while fibronectin is ideally suited for cell spreading, and allowed us to compare directly the data on IRSp53 KO MEFs with published data generated using VASP loss-of-function mutants (Applewhite et al, 2007). Indeed, filopodia can be differentially induced in response to diverse ECM substrates or soluble stimuli. Importantly, the different stimuli elicit filopodia formation by triggering a diverse set of signalling pathways. Thus, we felt it was important to explore multiple modes of inducing these migratory structures in order to evaluate the relevance of a specific protein (IRSp53) in their formation.

In wound healing in mice: Why can endothelial cells apparently migrate without delay, even in the absence of IRSp53?

There are multiple mechanisms through which filopodia can be generated (Dent et al, 2007; Faix et al, 2009; Ridley, 2011; Yang & Svitkina, 2011). For example, VASP family members have been shown to be required for filopodia in neurons through genetic analyses. However, these protrusions can be rescued by expression of Myosin X or by plating neuronal cells on laminin (Dent et al, 2007), indicating alternative mechanisms of filopodia formation exist. Additionally, it has emerged that there are numerous modes of migration, enabling cells to adapt plastically to environmental conditions and to overcome migration blockades (Friedl et al, 2012). Endothelial cells, for example, can move collectively (Trepat & Fredberg, 2011; Vitorino et al, 2011). It is possible that through this mode of locomotion endothelial cells adapt to interfering migratory (impairment of filopodia) conditions, thus enabling them to overcome loss of factors important for filopodia formation, such as loss of IRSp53.

Last but not least: the entire ms is not carefully edited, there are numerous typing and spelling errors and incomplete sentences. Example: Results, {section sign} 5, heading.

We apologize and we have now carefully edited the manuscript, paying particular attention to the formatting of the references.


© EMBO 15


The manuscript by Disanza et al describes the cooperation of IRSp53 with VASP, a regulator of F-actin extension, in filopodia formation. The authors report that IRSp53 functions as a weak F-actin barbed end capper and this function is negatively regulated by CDC42. In turn, CDC42 positively regulates the interaction of IRSp53 with VASP and the authors provide evidence that IRSp53 is required for CDC42 induced filopodium formation.

In general the experiments have been performed with great care. The authors and others had previously shown that the SH3 domain of IRSp53 interacts with VASP and Mena. It had been already shown that CDC42 induces filopodia by promoting the formation of an IRSp53:Mena complex (Krugmann et al Curr Biol 2001). The authors have published that a complex of IRSp53-VASP functions in filopodia formation by supporting actin filament bundling (Vaggi et al PLOS Comp Biol). Nevertheless, this referees believes that the function of an IRSp53-Mena/VASP complex in filopodia formation is not understood and some of the older experiments were not properly quantified or controlled. In general I would support a publication in the EMBO Journal if the molecular mechanism of the IRSp53-Mena/VASP function in filopodia formation would be clarified. The authors can be more upfront with prior knowledge and should address and use the prior knowledge throughout the manuscript. Furthermore, the following concerns needs to be addressed before I can recommend publication.

We appreciate that this referee is in principle supportive of publication. We further sincerely appreciated her/his careful critical reading and fair evaluation of our results in the context of previous published work.

We have cited the work by Krugmann et al. in the introduction rather than at the beginning of the Results section and discussed, as suggested, some of the previous findings with respect to the present set of results throughout the manuscript as requested.

Major concerns:

1. Page 11: The authors claim that: "The formation of a VASP::IRSp53 complex has no effect on the filament elongation rate of VASP but promotes its actin nucleation activity." This claim is not supported by the data shown: The polymerization kinetics of pyrenyl actin in Fig 2A in the presence of VASP and profilin are identical with or without IRSp53 indicating that, in the presence of profilin, IRSp53 does not increase the activity of VASP. In contrast, without profilin IRSp53 increases the activity of VASP (Fig 1B).

The experiments in Fig. 1b and 2a are very different. In Fig. 1b, VASP actin polymerization was measured in the absence of spectrin:actin seeds and thus both the nucleation and elongation activity can account for VASP-induced actin assembly. Under these conditions, IRSp53 enhances VASP activity. As addressed in the replay to referee #1, point 1, these experiments were performed at 50 mM KCL, under conditions in which the nucleation activity of VASP is indeed detectable. In the experiment of Fig. 2a, actin polymerization is instead initiated by spectrin:actin seeds, enabling the detection of elongation activity alone (the preformed spectrin:actin seeds overcome the thermodynamic barrier required for the formation of actin dimers and trimers that trigger subsequent actin filament elongation). Under these conditions, i.e. under conditions in which nucleation is not a decisive factor, IRSp53 has no effect on the polymerization activity of VASP. On the contrary, VASP prevents IRSp53 from capping and relieves IRSp53-mediated slowing down of actin polymerization. Thus, indeed IRSp53 promotes actin nucleation of VASP at 50 mM salt. However, following the suggestion of referee #1, we removed this set of experiments and re-crafted the manuscript to emphasize: i) the role of IRSp53 in slowing down filaments elongation; ii) the finding that CDC42 and VASP relieves this effect; iii) and the ability of IRSp53 to promote VASP recruitment to solid and lipid phases and thus to enhance VASP processive activity to overcome the inhibition by the capping protein.

In Fig 2B the authors chose to use a mutant of VASP with a Dictyostelium VASP G-Actin binding site, which makes the comparison with human VASP very difficult.


© EMBO 16

Please see our detailed response to a very similar question regarding the use of the VASP chimera in point #3 of referee #1. Moreover, we added experiments using human VASP, as well as human EVL, which corroborate our previous observations with hVASP-DdGAB.

Furthermore, the assay in Fig 2B was performed without profilin whereas the experiment in Figure 2C was performed with full length VASP and profilin. Comparing Fig 2A and 2C, which have comparable assay conditions, suggests that according to Fig. 2A there is no difference in VASP activity and the observed increase in filament numbers in Fig2C might arise from the setup of the experiment that IRSp53 was coated on the slide thereby maybe allowing recruitment and clustering of VASP.

The referee is correct in stressing the notion that irrespective of the presence or absence of profilin, IRSp53 has marginal effects on filament elongation rate by VASP, but enables the recruitment of this protein on solid surfaces, facilitating its clustering and processive elongation activity of VASP despite the presence of capping proteins as described in the text (see also below the response to point 2).

2. In general the authors chose to include profilin in some experiments and not others without explanation. Profilin has an important role for VASP function and therefore it is important that when experiments are compared that this is taken into account.

We agree that profilin is likely to be relevant in regulating some VASP activities. Indeed, one of our first concerns was to establish whether IRSp53 and profilin binds VASP in a competitive manner. Experiments in Fig. S1g, however, indicate that IRSp53 and profilin do not compete for VASP binding. Finally, we previously showed that at least under the experimental conditions tested, VASP-mediated actin filament elongation was unaffected by profilin (Breitsprecher et al, 2008). Similarly, we did not detect any effects of profilin under the conditions used for the TIRF assays in Fig. S3b (compare ‘no profilin’ with ‘plus profilin’).

3. Fig. 2D: The authors decided to use a mutant of VASP with a Dictyostelium VASP G-Actin binding site, which makes the comparison with human VASP very difficult. It would be better to repeat the experiments with human VASP (alternatively the experiments in Fig 1 and S1 should be repeated with VASP DdGAB). Measuring the elongation rates of just 10 filaments is only meaningful if the differences observed show statistically significant differences. Statistics should be performed and most likely more filaments needs to be measured.

We performed a new set of experiments using hVASP and hEVL, and provided statistics that indicate that differences are significant according to the t-test (see Fig. 2a-b and Movie S1 of the revised manuscript). At least 16 filaments per condition were measured in each experiment and error bars represent s.e.m. from three independent experiments (see Fig 2a-b).

4. Fig 3B Was IRSp53 on beads? (not stated in the Figure legend)?

Yes, IRSp53 was immobilized on NTA beads. We added this information to the revised figure legend.

Is this increase in IRSp53/VASP binding statistically significant?

The increase is indeed significant. We added t-test data in the revised manuscript.

5. Fig 3: To understand the molecular mechanism of the IRSp53-VASP complex better, the authors should perform pyrene-actin polymerization assays with IRSp53-VASP and Cdc42-GTP or Cdc42-GDP and this should be compared with Eps8-IRSp53-VASP with Cdc42-GTP or Cdc42-GDP.

This is a relevant point of great interest as it has become clear that interactions between different actin regulatory complexes having similar or different activities is key to the regulation, in space and time, of actin polymerization.


© EMBO 17

We tested the addition of CDC42-GTP and CDC42-GDP in actin:spectrin seed assays and showed that GTP-loaded-CDC42, but not CDC42-GDP is capable of relieving IRSp53 kinetic capping activity (Fig. 1e of the revised manuscript). Since the effect of IRSp53 on VASP activity could be unmasked primarily after clustering of the latter onto IRSp53-coated beads in TIRF assays, we attempted to add CDC42-GTP under these conditions. Unfortunately, the limited dynamic range of these assays as described in the reply to point 4 of referee #1, prevented us from reliably measuring quantitative differences. Attempts to reconstitute the CDC42-IRSp53-VASP complex on solid surfaces are on-going, but the setting up of the conditions necessary to obtain reliable results requires substantially more time and effort, and would certainly delay publication.

We omitted the use of EPS8 at this stage because the interpretation of results under these conditions would be complex and far from unequivocal. EPS8 has capping activity, and in the presence of IRSp53 is also able to bundle filaments (Disanza et al, 2004; Disanza et al, 2006; Vaggi et al, 2011). This complicates the interpretation of results obtained using the combination of these proteins, which possess multiple activities on actin. Additionally, we showed that EPS8 and VASP do compete for IRSp53 binding. To address the role of EPS8 in this context we would most likely need to set up single molecule analysis of differentially fluorescently-labelled EPS8, VASP and actin molecules, and perform multicolour TIRF. We are aiming to set up the conditions necessary to perform these experiments in the near future. However, with our current TIRF setup, we are unable to address these questions.

6. Fig 6A How were the cells selected for analysis if only 7 cells per movie were quantified?

Cells of identical shape and morphology at the wound edge were picked and the experimenter was blind with respect to the genotypes. A minimum of 20, and not 7, cells for each condition were analysed as specified in the figure legends. This implies that in most of the experiments we analysed more than 20 cells. Indeed, we generally selected 30–50 cells per experimental condition.

An analysis of relative reduction in wound area should be performed in addition.

We included this new measurement in Fig. 6a (bottom right graph) of the revised manuscript.

Page 16: The claim that null MEFs are impaired in extending filopodia in the Movie S9 is not obvious.

We agree and rephrased this statement. We also would like to stress that cells reconstituted with WT IRSp53 extend polarized protrusions into the wound areas, while IRSp53 KO cells do not display this polarized phenotype.

“Importantly, MEFs null cells are impaired in extending polarized protrusions in the wound space, and moved randomly rather than directionally (Movie S9)”. Pg.17 of the revised manuscript.

7. Fig 6B-C: Boyden chamber assays are problematic especially if not properly controlled. The number of cells that appear on the lower side of the membrane depends on the number of cells seeded in each experiment. Even if exactly the same numbers are seeded, a defect in adhesion could be interpreted as a defect in invasion. At least the numbers of cells adhering to the upper chamber should be quantified as control. Better less problematic invasion assays are available and should be used.

We agree that the interpretation of Boyden chamber invasion assay can be problematic since this assay is the result of potentially different processes, including proliferation, cell adhesion to the matrix and cell migration. However, IRSp53 null cells do not show any difference in proliferation nor in adhesion as shown in Fig. S6a and 5d of the revised manuscript. Additionally, we performed an additional invasion assay, in which single cells invading into a native type I collagen matrix towards soluble PDGF were monitored in real time. This experimental setup allows the determination of the invasive chemotactic index (Forward migration index-FMI- of Invasion). The removal of IRSp53 reduces the invasive FMI by >30%, and this result is consistent with the results obtained with Boyden chambers. We have included this result in Fig. S6B of the revised manuscript.


© EMBO 18

8. Fig. 7: The SEM error bars are huge. Is the p value of the t-test really p<0.001?

The conclusions from the data are only valid if the differences are significant.

We apologize for the mistake. Indeed, the t–test indicates that the difference is significant with a p<0.05, rather than a p<0.001.

9. The conclusion sentence on page 17 is overstated since it has been only shown that macrophage numbers are significantly lower in the wounds of IRSp53 KO mice.

We agree and rephrased this sentence as follows on pages 18-19 of the revised manuscript.

“Microphotographs and morphometric quantification and H&E analysis of wounds revealed that IRSp53 KO animals displayed a marked delay in wound closure relative to control animals (Fig. 7a-b). Indeed, a thick clot was still present at 7 days after wounding, the wound limits were larger while the re-epithelisation process was still incomplete (Fig. 7b). Notably, we detected no differences in the cellular composition and overall morphology between IRSP53 KO and WT mice, indicating that IRSp53 is dispensable for skin development, but required for wound re-epithelisation (Fig. 7c). Importantly, IRSp53 is expressed in epidermal keratinocytes, macrophages, isolated dermal fibroblasts and endothelial cells (Fig. 7d-e). Removal of IRSp53 also delays wound closure in vitro also of keratinocytes (not shown). Furthermore, quantitative analysis of cells at wound sites revealed that while endothelial cells were recruited in similar number and generated vases of similar size regardless of the genotypes, macrophages were significantly reduced in IRSp53 KO mice (Fig. S5), in keeping with previous finding showing that IRSp53 removal also impairs macrophage chemotactic cell motility (Abou-Kheir et al, 2008). Collectively, these results suggest that IRSp53 impairs the complete repopulation into the wound sites of epithelial keratinocytes, macrophages and dermal fibroblasts, the major cells involved in wound healing in vivo”.

10. The findings of this manuscript and prior knowledge on EPS8-IRSp53 should be better integrated. In particular, the data that CDC42 reduces the binding between IRSp53 and EPS8 (mentioned in the discussion) should be shown.

We introduced the reference suggested by the referee in the introduction. We would like to point out that we did mention this manuscript at the beginning of the Results section in the previous version of the manuscript.

Regarding exploring the effect on Mena, we made various attempts to produce recombinant full length MENA, but due to its much larger size compared with VASP and EVL or its repeat elements, MENA could not be expressed. We do, however, fully agree with the referee that exploring whether IRSp53 and its related genes are able to interact with VASP, MENA and EVL family members is of great interest. Thus, we expressed recombinant human EVL and now show that IRSp53 can bind to EVL (Fig. S1a of the revised manuscript), and promote its recruitment and clustering to drive processive filament elongation (Fig. 2a-b of the revised manuscript). These results and the data obtained by Krugmann et al. suggest that IRSp53 can interact with all of the members of the ENA/VASP family and mediate their recruitment to solid surfaces.

11. The results of this paper and the published results of Krugmann et al Curr Biol 2001 should be introduced in the introduction. The key functional experiments should be repeated for Mena to explore whether the cooperation with IRSp53 is a common feature of the Ena/VASP protein family and the data should be compared and contrasted with the published data in the discussion.

We introduce the reference suggested in the introduction. We would like to point out that we did mention this manuscript also at the beginning of the results section in the previous version of the manuscript.

Regarding exploring the effect on Mena, we made various attempt to produce recombinant full length MENA, but apparently due to its much larger size as compared to VASP and EVL or its repeat elements, MENA could not be expressed. We do, however, fully agree with the reviewer that exploring whether IRSp53 and its related genes are able to interact with VASP, MENA and EVL


© EMBO 19

family members is of great interest. Thus, we expressed recombinant human EVL and now shown that IRSp53 can bind to EVL (Fig. S1a of the revised manuscript), promote its recruitment and clustering to drive processive filament elongation (Fig. 2a-b of the revised manuscript). These and the data obtained by Krugmann et al. results therefore suggest that IRSp53 can interact with and mediate the recruitment to solid surfaced of all the members of the ENA/VASP family.

12. The authors should address how the results from their previous publication on IRSp53-VASP-EPS8 (Vaggi et al PLOS Comp Biol) fits with the here presented findings. In particular, in that publication the authors concluded that a complex of IRSp53-VASP functions in filopodia formation by supporting actin filament bundling.

We addressed this point in the revised discussion (pg. 22) as reported below.

“ … Notably, we have previously shown that IRSp53 may contribute to filopodia elongation by promoting the bundling activity of a CDC42-IRSp53-VASP complex through oligomerization (Vaggi et al, 2011). Thus, the formation of this complex may not only be required to promote processive filaments elongation, an activity that is likely essential to initiate the growth of actin filaments. These filaments must also be bundled to support the extension of filopodia. This latter activity is though to be primarily carried out by fascin (Vignjevic et al, 2006). Nevertheless, bundling particularly close to the barbed ends, to which the IRSp53::VASP complex is restricted may subsequently favour the tighter crosslinking by fascin to provide sufficient stiffness for filopodia to resist counter PM tension and extracellular forces…”

Minor concerns:

1. Please add molecular weight makers on all blots.

We added the MW markers as requested in the revised figures of the manuscript.

2. Fig 4 C What defines a cluster? Is this a subjective analysis or was the cluster defined by a certain intensity threshold over background?

A cluster was defined as the area with at least a 3-fold increase in fluorescence intensity with respect to the signal of VASP along the adjacent leading edge. We added this information in the Material and Methods section of the revised manuscript.

3. Fig S3D: Is this a site of filopodia initiation? The cluster shown is rather invaginated. The movie S7 is much better: the localization at the tips of filopodia should be shown.

In order to demonstrate that the foci to which IRSp53 and VASP localize are the sites of subsequent filopodia initiation, we added an additional set of stills from the very same movie, showing that the IRSp53 and VASP foci indeed precede the formation of a bona fide growing filopodium. We also indicated by arrows the timing of the appearance of IRSp53 and VASP at the tip of the nascent filopodium (Fig. S4D and Movie S8 of the revised manuscript).

3. Fig S3E missing (Fig S3E is mislabelled and should read FigS3F).

We amended this mistake.

4. Fig S4C Were all cells with at least one filopodia tip positive for VASP counted?

VASP in WT MEFs is present in the vast majority of filopodia tips (>80% of filopodia clearly display VASP at their tip in >90% of cells). The remaining filopodia are presumably positive for VASP, but we may have failed in detecting it as some filopodia tips are out of the focal plane. Removal of IRSp53 reduces, but not completely abolishes, the formation of VASP foci. Indeed, filopodia are significantly reduced in IRSp53-/- cells. In addition, ~80% of filopodia in IRSp53-/- cells do not show any VASP at their tips, while the remaining 20% have at least one VASP-positive filopodium. Reconstitution of MEFs with WT IRSp53 restored WT phenotypes. In essence, in the vast


© EMBO 20

majority of cases whenever filopodia form, they display VASP at their tips, unless IRSp53 is removed. We included this information in the Supplementary Experimental Procedures section (pg. 6 of the supplementary information of the revised manuscript).

6. Fig 5C Why has IRSp53 a higher molecular weight in the WT MEFs compared to the reconstituted MEFs?

There are four distinct splicing isoforms of IRSp53, which differ in their very C-terminal regions. These isoforms are nearly ubiquitously expressed (Scita et al, 2008). Importantly, all isoforms retain the I-BAR, CRIB-PP and SH3 domains, but either possess a WH2 domain or a PDZ binding motif at their C-termini. The isoform we introduced into IRSp53 KO MEFs is BAIAP2-S that has a PDZ motif at its C-terminus. Thus, the reconstituted cells express only one isoform of IRSp53. We included this information in the legend to Fig. 5c of the revised manuscript.


Summary:

The authors report that IRSp53 inhibits actin polymerization at barbed ends, and that this inhibitory activity is relieved by VASP or Cdc42. They also find that the IRSp53 SH3 domain binds to VASP, map the binding site, and find that binding is enhanced in the presence of activated Cdc42. With regard to cellular activities, they find that IRSp53 and VASP accumulate in foci at the leading edge that may represent sites of filopodia initiation. Using cells from IRSp53 null mice, they show that IRSp53 is important for VASP focus formation, filopodia formation/stability, and cell motility. IRSp53 is also important for rapid wound healing in mice.

General comments:

In general this is an interesting paper that details an impressive amount of work and sheds new light on the biochemical and cell biological functions of IRSp53 and VASP. The paper is suitable for publication in EMBO J if the authors address the specific points listed below.

Specific comments:

1. Page 8: There are unformatted references that need to be correctly formatted.

We apologize for the unformatted references. This part of text has been removed as suggested by referee #1.

2. Page 11: The authors state that "bucking actin filaments grew away from the bead surface..." To this reviewer, buckling implies that both ends of the filament are fixed in place. Instead, in the video it appears that the end of the actin filament facing away from the bead is free in solution. Thus, it seems the filaments are "bent" or "curved" rather than "buckling".

Attachment by two-fixed point is not necessarily required for buckling. Buckle formation is an instant evidence for processive filament elongation, and is triggered by any resistance to elongation of an actin filament growing with its barbed end attached to a fixed point (Kovar & Pollard, 2004; Breitsprecher et al, 2008). In our bead assays, the growing barbed ends of the filament remain continuously attached to VASP clustered on the bead surface, whereas the pointed ends in principle move away freely from the surface. However, if they encounter an obstacle (transiently or permanently), or become actively or passively attached to the coverslip surface, they begin to buckle. This is now indicated in Movie S1 by white arrows. Moreover, the observed filament growth on the bead surface must be processive, because only processive filament elongation is resistant to the presence of capping proteins (Breitsprecher et al, 2008; Breitsprecher et al, 2011).


© EMBO 21

3. Figures 2B, 2C, 2D, 3B: The authors should note whether the differences they observe are statistically significant, as they do in other figures.

We included statistics, where appropriate, in the revised manuscript .

4. Movies S2-S6: The authors state that the formation of GFP-VASP foci preceded filopodia formation. However, very few filopodia are formed in these movies. It may be valuable to observe the frequency with which clusters mature into filopodia.

Movies S2 and S6 include IRSp53 KO MEFS (Movies S3 and S5 central panel) that do form a reduced number of filopodia. Conversely, WT or reconstituted KO MEFs display a significant higher number of filopodia and microspikes that are considered filopodia structures embedded into lamellipodia.

As suggested by the referee, we measured the frequency with which VASP foci mature into filopodia in WT MEFs. Nearly 80% ± 0.19 (mean ± s.e.m. n= 55) of foci will mature to form filopodia o microspikes. We included this information in the revised caption of Movie S3. A similar frequency of foci/filopodia formation was detected in IRSp53 KO MEFs reconstituted with WT IRSp53. As additional information for this referee, we have included as an addendum to this rebuttal a movie –frequency of filopodia- and series of stills, in which the formation of filopodia emanating from VASP foci is more easily visualized (see Addenda FigRev#4 and Movie Rev#4) (Note from the editor: movie not included in this document)

5. Page 14 and movie S8: The authors mention that mCherry-IRSp53 accumulates in foci about 1.4 s before GFP-VASP. However, in the images and movies, accumulation of the two seems simultaneous. They need to better document this to clearly establish the order of recruitment.

It must be pointed out the 1.4 s delay corresponds to only 3 frame differences. To better visualize this point, we included “time arrows” that mark the accumulation of IRSp53 and subsequently of VASP into foci, the intensity of which was defined by the imagining analysis algorithm (rainbow coloured still frame and Movie S8-where red pixels mark the local and timely accumulation of the protein of interest into foci defined as having an at least a 3-fold increase signal intensity with respect to adjacent leading edge signals).

6. Figure 5A: The authors note that IRSp53 "decorates the shaft of filopodia protrusions with a punctate distribution". However, IRSp53 seems to be everywhere in the cell with a similar distribution, and it does not appear to specifically localize to filopodia. This should be noted.

We agree that IRSp53 displays a punctate distribution throughout the cells with a localization also along the filopodia shaft. We included this comment into the text on pg. 16 of the revised manuscript.

7. Figure 7B, C: These images should be more clearly and extensively described in the text.

We included an additional description in the text of the revised manuscript on pg 18-19).

“Microphotographs and morphometric quantification and H&E analysis of wounds revealed that IRSp53 KO animals displayed a marked delay in wound closure relative to control animals (Fig. 7a-b). Indeed, a thick clot was still present at 7 days after wounding, the wound limits were larger while the re-epithelisation process was still incomplete (Fig. 7b). Notably, we detected no differences in the cellular composition and overall morphology between IRSP53 KO and WT mice, indicating that IRSp53 is dispensable for skin development, but required for wound re-epithelisation (Fig. 7c). Importantly, IRSp53 is expressed in epidermal keratinocytes, macrophages, isolated dermal fibroblasts and endothelial cells (Fig. 7d-e). Removal of IRSp53 also delays wound closure in vitro also of keratinocytes (not shown). Furthermore, quantitative analysis of cells at wound sites revealed that while endothelial cells were recruited in similar number and generated vases of similar size regardless of the genotypes, macrophages were significantly reduced in IRSp53 KO mice (Fig. S5), in keeping with previous finding showing that IRSp53 removal also impairs macrophage chemotactic cell motility (Abou-Kheir et al, 2008). Collectively, these results suggest


© EMBO 22

that IRSp53 impairs the complete repopulation into the wound sites of epithelial keratinocytes, macrophages and dermal fibroblasts, the major cells involved in wound healing in vivo”

References Abou-Kheir W, Isaac B, Yamaguchi H, Cox D (2008) Membrane targeting of WAVE2 is not sufficient for WAVE2-dependent actin polymerization: a role for IRSp53 in mediating the interaction between Rac and WAVE2. J Cell Sci 121: 379-390 Applewhite DA, Barzik M, Kojima SI, Svitkina TM, Gertler FB, Borisy GG (2007) Ena/VASP Proteins Have an Anti-Capping Independent Function in Filopodia Formation. Mol Biol Cell Barzik M, Kotova TI, Higgs HN, Hazelwood L, Hanein D, Gertler FB, Schafer DA (2005) Ena/VASP proteins enhance actin polymerization in the presence of barbed end capping proteins. J Biol Chem 280: 28653-28662 Bear JE, Gertler FB (2009) Ena/VASP: towards resolving a pointed controversy at the barbed end. J Cell Sci 122: 1947-1953 Breitsprecher D, Kiesewetter AK, Linkner J, Urbanke C, Resch GP, Small JV, Faix J (2008) Clustering of VASP actively drives processive, WH2 domain-mediated actin filament elongation. EMBO J Breitsprecher D, Kiesewetter AK, Linkner J, Vinzenz M, Stradal TE, Small JV, Curth U, Dickinson RB, Faix J (2011) Molecular mechanism of Ena/VASP-mediated actin-filament elongation. EMBO J 30: 456-467 Dent EW, Kwiatkowski AV, Mebane LM, Philippar U, Barzik M, Rubinson DA, Gupton S, Van Veen JE, Furman C, Zhang J, Alberts AS, Mori S, Gertler FB (2007) Filopodia are required for cortical neurite initiation. Nat Cell Biol 9: 1347-1359 Disanza A, Carlier MF, Stradal TE, Didry D, Frittoli E, Confalonieri S, Croce A, Wehland J, Di Fiore PP, Scita G (2004) Eps8 controls actin-based motility by capping the barbed ends of actin filaments. Nat Cell Biol 6: 1180-1188 Disanza A, Mantoani S, Hertzog M, Gerboth S, Frittoli E, Steffen A, Berhoerster K, Kreienkamp HJ, Milanesi F, Di Fiore PP, Ciliberto A, Stradal TE, Scita G (2006) Regulation of cell shape by Cdc42 is mediated by the synergic actin-bundling activity of the Eps8-IRSp53 complex. Nat Cell Biol 8: 1337-1347 Dominguez R (2007) The {beta}-Thymosin/WH2 Fold: Multifunctionality and Structure. Ann N Y Acad Sci Faix J, Breitsprecher D, Stradal TE, Rottner K (2009) Filopodia: Complex models for simple rods. Int J Biochem Cell Biol 41: 1656-1664 Ferron F, Rebowski G, Lee SH, Dominguez R (2007) Structural basis for the recruitment of profilin-actin complexes during filament elongation by Ena/VASP. EMBO J 26: 4597-4606 Friedl P, Sahai E, Weiss S, Yamada KM (2012) New dimensions in cell migration. Nature reviews Molecular cell biology 13: 743-747 Hansen SD, Mullins RD (2010) VASP is a processive actin polymerase that requires monomeric actin for barbed end association. J Cell Biol 191: 571-584 Koestler SA, Rottner K, Lai F, Block J, Vinzenz M, Small JV (2009) F- and G-actin concentrations in lamellipodia of moving cells. PLoS One 4: e4810


© EMBO 23

Kovar DR, Pollard TD (2004) Insertional assembly of actin filament barbed ends in association with formins produces piconewton forces. Proc Natl Acad Sci U S A 101: 14725-14730 Mejillano MR, Kojima S, Applewhite DA, Gertler FB, Svitkina TM, Borisy GG (2004) Lamellipodial versus filopodial mode of the actin nanomachinery: pivotal role of the filament barbed end. Cell 118: 363-373 Menzies AS, Aszodi A, Williams SE, Pfeifer A, Wehman AM, Goh KL, Mason CA, Fassler R, Gertler FB (2004) Mena and vasodilator-stimulated phosphoprotein are required for multiple actin-dependent processes that shape the vertebrate nervous system. J Neurosci 24: 8029-8038 Pasic L, Kotova T, Schafer DA (2008) Ena/VASP proteins capture actin filament barbed ends. J Biol Chem 283: 9814-9819 Pertz O (2011) Filopodia: Nanodevices that sense nanotopographic ECM cues to orient neurite outgrowth. Commun Integr Biol 4: 436-439 Reinhard M, Jouvenal K, Tripier D, Walter U (1995) Identification, purification, and characterization of a zyxin-related protein that binds the focal adhesion and microfilament protein VASP (vasodilator-stimulated phosphoprotein). Proc Natl Acad Sci U S A 92: 7956-7960 Ridley AJ (2011) Life at the leading edge. Cell 145: 1012-1022 Schirenbeck A, Arasada R, Bretschneider T, Stradal TE, Schleicher M, Faix J (2006) The bundling activity of vasodilator-stimulated phosphoprotein is required for filopodium formation. Proc Natl Acad Sci U S A 103: 7694-7699 Scita G, Confalonieri S, Lappalainen P, Suetsugu S (2008) IRSp53: crossing the road of membrane and actin dynamics in the formation of membrane protrusions. Trends Cell Biol 18: 52-60 Trepat X, Fredberg JJ (2011) Plithotaxis and emergent dynamics in collective cellular migration. Trends in cell biology 21: 638-646 Tsukada T, Tippens D, Gordon D, Ross R, Gown AM (1987) HHF35, a muscle-actin-specific monoclonal antibody. I. Immunocytochemical and biochemical characterization. The American journal of pathology 126: 51-60 Vaggi F, Disanza A, Milanesi F, Di Fiore PP, Menna E, Matteoli M, Gov NS, Scita G, Ciliberto A (2011) The Eps8/IRSp53/VASP network differentially controls actin capping and bundling in filopodia formation. PLoS Comput Biol 7: e1002088 Vignjevic D, Kojima S, Aratyn Y, Danciu O, Svitkina T, Borisy GG (2006) Role of fascin in filopodial protrusion. The Journal of cell biology 174: 863-875 Vitorino P, Hammer M, Kim J, Meyer T (2011) A steering model of endothelial sheet migration recapitulates monolayer integrity and directed collective migration. Molecular and cellular biology 31: 342-350 Yang C, Svitkina T (2011) Filopodia initiation: focus on the Arp2/3 complex and formins. Cell Adh Migr 5: 402-408

Legend to Addenda Figure Rev#4

Still images from fluorescence spinning-disc confocal analysis of migrating cells. MEFs were electroporated with GFP-VASP and plated on laminin (40µg/ml). 24 hours after electroporation cells were subjected to time-lapse analysis (time interval 5 seconds). Numbered white arrowheads indicate the starting point of VASP discrete accumulation and the corresponding emerging filopodia. The frequency with which VASP foci mature into filopodia in WT MEFs was measured.


© EMBO 24

Nearly 80 % ± 0.19 (mean ± s.e.m. n= 55) of foci will mature to form filopodia or microspikes. Similar frequency of foci/filopodia was detected in IRSp53 KO MEFs reconstituted with WT IRSp53 (Figure 4b and Movie S3 of the revised manuscript).

3rd Editorial Decision 27 August 2013

Thank you for the submission of your revised manuscript to The EMBO Journal and please accept my apologies for the delay in our response, due to the holiday season and the difficulty to contact referees this time of the year. It has been sent back to two of the original reviewers, who now consider that their major concerns have been properly addressed and your manuscript is almost ready for publication. Notwithstanding their positive view, referees still point out to some issues, as you will see below,

T0 T5s T10s T15s T20s

T75s T80s T85s T90s T95s

T100s T105s T110s T115s T120s



T125s T130s T135s T140s T145s

T150s T155s T160s T165s T170s

T175s T180s T185s T190s T195s

T200s T205s T210s T215s T220s

T225s T230s T235s T240s T245s


© EMBO 25

that will need your attention before your manuscript can be finally accepted. Although referee #3 suggests some minor changes and clarifications that will also need to be addressed, I would like to particularly draw your attention to the suggestion of referee #1 of moving as much as possible of the discussion of the supplementary results into the supplementary figure legends in order to simplify the paper for the potential reader. I would also like to mention that we now encourage the publication of source data, particularly for electrophoretic gels and blots, with the aim of making primary data more accessible and transparent to the reader. Although optional at the moment, would you be willing to provide a PDF file per figure that contains the original, uncropped and unprocessed scans of all or key gels presented? The PDF files should be labeled with the appropriate figure/panel number and should have molecular weight markers; further annotation could be useful but is not essential. The files will be published online with the article as supplementary "Source Data" files. If you have any questions regarding this initiative do not hesitate to contact me. Thank you very much again for your patience. I am looking forward to seeing the revised, final version of your manuscript. REFEREE REPORTS:

Referee #1 (General Remarks): The authors have adequately responded to my comments. However the authors might want to take Referee #2's comment about the paper complexity a bit more into account. While I don't agree with separation of biochemistry from cells from mice, I do agree that the paper is heavy and difficult to digest. In part this is due to excessive discussion of Supplementary Material in the main text. For example the first page of the results is entirely about Supp Fig 1. This makes for a rather weak start. Also this information seems to go with the DLS data, which is a punchier figure panel (Figure 2C) that is easier to explain. I would suggest that this DLS panel be integrated into Figure 1. After the DLS discussion, the authors could then reference Figure S1, but transfer the very long explanation of that data to the supplementary figure legend. This is just one example that jumped out at me. The authors might be able to do similar reorganizations elsewhere. In my mind, supplementary data is discussed at most with a single sentence-otherwise it's just a maneuver to squeeze in more figures. Referee #3 (General Remarks): The revised version of the manuscript Disanza et al. has been substantially improved and most of my concerns have been addressed. The authors have turned it into a very nice convincing paper and I recommend publication in the EMBO Journal after my remaining concerns have been addressed. Page 11; Fig S3b. Is there really a statistically significant increase in filament numbers? If not the term "...nucleated by VASP..." should not be used. An alternative, much more plausible explanation for the reduction in elongation rate observed in S3B upon addition of increasing concentrations of IRSp53 is that IRSp53 competes with VASP for barbed end elongation by capping the barbed ends. It is not clear to this referee how the TIRF data in S3b and S3c using the VASP DdGAB adds anything to the really clear and convincing data in Fig S3a. Therefore, this referee believes that the data rather confuses the clear line of thought and should be removed. Fig. 2B What is the statistical test used? Only His-IRSp53+VASPDdGAB shown indicated to be significantly different but it is not clear to which group it is compared. The materials and methods section states that t-test has been used. One-way ANOVA should be used to compare all groups (this applies to all panels in which several groups are compared). Fig 4A What is the statistical test used? The materials and methods section states that t-test has been used. One-way ANOVA should be used to compare all groups.


© EMBO 26

2nd Revision - authors' response 29 August 2013

Reviewer#1

1. She/he suggested to reduced the text referring to supplementary figure to a minimum and to rearrange some of the figure panel to improve readability of our manuscript

- We have move DLS experiments to Figure 1A and reduced substantially the text referring to Figure S1 (page 1 of the revised manuscript)

- Following reviewer #3 suggestions, we have eliminated the panel of S3B and C and the text referring to this panel (page 10/11 of the revised manuscript) and the related legend

- We have removed the lengthy explanation about the collagen type I invasion assays (related to Figure S6b that was requested by one of the original reviewers). We left the reference where similar assays were conducted (page 17 of the revised manuscript).

- We rearrange the order of some of the Supplementary figure panels and their legends (2-4) according to the outlined above changes.

Reviewer #3

Page 11; Fig S3b. Is there really a statistically significant increase in filament numbers? If not the term "...nucleated by VASP..." should not be used. An alternative, much more plausible explanation for the reduction in elongation rate observed in S3B upon addition of increasing concentrations of IRSp53 is that IRSp53 competes with VASP for barbed end elongation by capping the barbed ends.

It is not clear to this referee how the TIRF data in S3b and S3c using the VASP DdGAB adds anything to the really clear and convincing data in Fig S3a. Therefore, this referee believes that the data rather confuses the clear line of thought and should be removed.

R. There is statistical significant (p<0.01, t-test) difference in the number of filaments between the sample in the presence of 200 nM IRSp53 and the one on the absence of IRSp53. Similarly, the elongation rate of the sample in the absence of IRSP53 or in the presence of 200 nM IRSp53 are significantly different (p<0.01, t-test).

However, following this reviewer’s suggestion, we agree that the data in Figure S3A are sufficient to provide unequivocal support to the notion that IRSp53 does not affect VASP elongation rates and for sake of clarity we removed panel S3B and C and the text referring to it.

Fig. 2B What is the statistical test used? Only His-IRSp53+VASPDdGAB shown indicated to be significantly different but it is not clear to which group it is compared. The materials and methods section states that t-test has been used. One-way ANOVA should be used to compare all groups (this applies to all panels in which several groups are compared).

Fig 4A What is the statistical test used? The materials and methods section states that t-test has been used. One-way ANOVA should be used to compare all groups.

R. In both cases, t-test was used to calculate the statistical differences since we compared in a pairwise manner each experimental condition to control. In Fig 2B, we compared the elongation rate of His-IRSp53+VASPDdGAB with that of human VASP, which are identical to the elongation rate obtained with actin alone (~ 10 subunits/sec).

In Figure 4A, each value of fluorescence intensity of the labelled proteins binding to liposome (marked with asterisks) was compared to the fluorescence value of the control (buffer). We explicitly stated this in Figure 2B and in the legend of Figure 2B and 4A of the revised manuscript. We added similar statement to Fig. 2A.

cdc42 switches irsp53 from inhibition of actin...

Documents