Round 1

Reviewer 1 Report

This is a very interesting and useful study for the field. This is particularly interesting with cryo-EM studies of actin expanding in recent years.

I have no concerns about the work but the authors should make the following adjustments:

Figure 1 - legend does not accurately describe panel D. It is not clear what open and closed arrows refer to (at least on my PDF).

Author Response

Comments and Suggestions for Authors

This is a very interesting and useful study for the field. This is particularly interesting with cryo-EM studies of actin expanding in recent years.

I have no concerns about the work but the authors should make the following adjustments:

Figure 1 - legend does not accurately describe panel D. It is not clear what open and closed arrows refer to (at least on my PDF).

Response 1: We would like to thank the reviewer for the valuable suggestions. We have added three new Figures (1-3) with electron microscopy images to clarify the organization of G- and F-actin, Supplementary materials with MS data and SDS-electrophoresis to demonstrate the quality of the original protein. Moreover, we have added the X-ray diffraction data for G- and F-actin, which demonstrate the absence of double-helix organization for F-actin.

Reviewer 2 Report

The manuscript by Glyakina and coauthors proposes a new model of F-actin organization, which is inconsistent with the generally recognized double-helical structure of the filament. The classical F-actin model is proposed to be inconsistent with the data obtained by proteolysis studies, mass spectrometry, and EM imaging data. Strong statements such as the presence of principally new actin-based structures necessitate demonstration of substantial evidence, which, in my opinion, is missing. The experiments are conducted without considering actin properties and do not satisfy standards accepted in the area. Although the manuscript is overall reasonably well written, the conclusions are not fully substantiated by the experimental data. A more detailed review is provided below.

Major concerns:

• The methodology is problematic in several ways, which may lead to false conclusions.
• The incubation of actin for a day at 37^oC is harsh on actin. It is anticipated that all ATP will be hydrolyzed, which negatively affects the stability of this protein and, likely, the structure of the filaments.
• Centrifugation of actin at 10kxg for 20 minutes is sufficient only to sediment bundles, which would likely to form after a day-long incubation at 37oC. 100,000xg for at least 30 minutes (better for an hour or at higher speeds) is required for reliable pelleting of individual actin filaments.
• G-actin requires a nucleotide (ATP or at least ADP) for its stability. F-actin is less demanding, but due to the treadmilling, it will also inevitably get damaged due to the absence of the nucleotide. Therefore, solubilization of the pellet in the ammonium bicarbonate buffer without ATP with the subsequent incubation for 8 hours represent highly non-physiological conditions for this protein.
• The presence of certain peptides is considered by the authors as their protection due to the contacts in F-actin. However, the incubation in the presence of a high concentration of potent proteases should destroy the filament within minutes while the experiment was continued for 8 hours. After the protease destroys the filament, the protection cannot result from the filament contacts. The authors should at least monitor the actin degradation profile by SDS-electrophoresis. However, even the quality of the original protein is not demonstrated by SDS-electrophoresis.
• Intrinsically, mass spectrometry is not a quantitative technique, unless designed respectively. The sole peptide’s presence in the analyzed solutions cannot demonstrate protection, if not compared quantitatively with the controls. Also, the original MS data are not provided.
• Modern-day modeling requires more than putting actin subunits together to “protect” the observed peptides and, as a minimum, should entail docking with Rosetta or similar protein modeling software.
• In my opinion, the interpretation of G-actin images as rings is misleading. Just as in clouds, our mind would “recognize” familiar objects, so does it in amorphous aggregates of actin.
• It is nor clear whether F-actin was prepared for EM in the same way as for proteolysis (i.e., a day-long incubation at 37^oC). If so, actin may be damaged. Furthermore, the proposed “ring” and “spiral” actin organization modes do not seem to represent the EM images of F-actin. Accurate measurements of the filament width are not demonstrated. Therefore, I am not convinced that the thinner structures represent single-stranded filaments. To avoid the ambiguity, thorough image analysis and filament reconstruction are required. An alternative interpretation could be that the thinner filaments represent overall classical double-stranded actin filaments, while the thicker counterparts may represent two filaments aligned along their sides.
• Splitting of F-actin to individual strands, albeit only on a short distance, has been proposed and demonstrated by cryo-EM (McGough and Chiu, 1999). Negative staining is, however, is prone to aberrations. In combination with prolonged incubation at 37oC under ATP-depleting conditions, the aberrations can be dramatic and likely to not represent physiological conditions. While the classical double-stranded arrangement has received countless confirmations of its physiological authenticity, the proposed models are poorly substantiated by the provided experimental data.
• The authors are encouraged to conduct reconstruction of the observed filaments.

 

Minor comments:

• Actin concentration cannot be reliably measured at 280 nm in the presence of ATP due to a strong absorbance of the latter. OD290 with a different extinction coefficient or with the same coefficient but in the presence of 0.5 M NaOH is recommended.

Author Response

Comments and Suggestions for Authors

The manuscript by Glyakina and coauthors proposes a new model of F-actin organization, which is inconsistent with the generally recognized double-helical structure of the filament. The classical F-actin model is proposed to be inconsistent with the data obtained by proteolysis studies, mass spectrometry, and EM imaging data. Strong statements such as the presence of principally new actin-based structures necessitate demonstration of substantial evidence, which, in my opinion, is missing. The experiments are conducted without considering actin properties and do not satisfy standards accepted in the area. Although the manuscript is overall reasonably well written, the conclusions are not fully substantiated by the experimental data. A more detailed review is provided below.

Response: We thank the reviewer for your effort in critically reviewing our manuscript and for the constructive suggestions and comments. We have added three new Figures (1-3) with electron microscopy images to clarify the organization of G- and F-actin, Supplementary materials with MS data and SDS-electrophoresis to demonstrate the quality of the original protein. Moreover, we have added the X-ray diffraction data for G- and F-actin, which demonstrate the absence of double-helix organization for F-actin.

Major concerns:

• The methodology is problematic in several ways, which may lead to false conclusions.
• The incubation of actin for a day at 37^oC is harsh on actin. It is anticipated that all ATP will be hydrolyzed, which negatively affects the stability of this protein and, likely, the structure of the filaments.

Response 1: We have clarified this situation in the text (Materials and methods). EM analysis of G-actin was carried out with a freshly obtained preparation. To convert G-actin into F-actin, up to 0.1 M KCL was added to the preparation of G-actin and incubated at room temperature for about 20 hours (according to EM analysis, after incubation for 5-10 hours, not all monomeric actin is converted into filamentous form).

The incubation of actin for a day at 37^oC was used for MS analysis.

 

• Centrifugation of actin at 10kxg for 20 minutes is sufficient only to sediment bundles, which would likely to form after a day-long incubation at 37oC. 100,000xg for at least 30 minutes (better for an hour or at higher speeds) is required for reliable pelleting of individual actin filaments.

Response 2: We study the process of amyloid formation more than 10 years to determine the core of amyloid fibrils. We have found the optimum velocity for centrifugation to sediment bundles. It is not possible to sediment only individual actin filaments.

 

• G-actin requires a nucleotide (ATP or at least ADP) for its stability. F-actin is less demanding, but due to the treadmilling, it will also inevitably get damaged due to the absence of the nucleotide. Therefore, solubilization of the pellet in the ammonium bicarbonate buffer without ATP with the subsequent incubation for 8 hours represent highly non-physiological conditions for this protein.

Response 3: We repeated our experiments several times (up to 5). Each time, before proteolysis, we checked with help of EM, whether we really got F-actin. F-actin is stable at 37. Otherwise, we have no precipitation sample after proteolysis. We don't need all the actin filaments to be long.

 

• The presence of certain peptides is considered by the authors as their protection due to the contacts in F-actin. However, the incubation in the presence of a high concentration of potent proteases should destroy the filament within minutes while the experiment was continued for 8 hours. After the protease destroys the filament, the protection cannot result from the filament contacts. The authors should at least monitor the actin degradation profile by SDS-electrophoresis. However, even the quality of the original protein is not demonstrated by SDS-electrophoresis.

Response 4: As for the fact that we have to control actin degradation. Is it worth doing? After all, actin degradation products (their size, length of peptides) are not of interest and are not the essence of this study. Degradation products are peptides less than 5 residues in length. But we are interested in the F-actin core. In addition, we work with a small amount that will not be visible during electrophoresis. We have added the SDS-electrophoresis to demonstrate the quality of the original protein.

 

• Intrinsically, mass spectrometry is not a quantitative technique, unless designed respectively. The sole peptide’s presence in the analyzed solutions cannot demonstrate protection, if not compared quantitatively with the controls. Also, the original MS data are not provided.

Response 5: We have provided MS data after processing by the program.

 

• Modern-day modeling requires more than putting actin subunits together to “protect” the observed peptides and, as a minimum, should entail docking with Rosetta or similar protein modeling software.

Response 6: We did the docking using the Yasara program. In the future, we can try this program too.

 

• In my opinion, the interpretation of G-actin images as rings is misleading. Just as in clouds, our mind would “recognize” familiar objects, so does it in amorphous aggregates of actin.

Response 7: We have added three new figures with EM images to clarify this situation and some text in the Result section.

 

• It is nor clear whether F-actin was prepared for EM in the same way as for proteolysis (i.e., a day-long incubation at 37^oC). If so, actin may be damaged. Furthermore, the proposed “ring” and “spiral” actin organization modes do not seem to represent the EM images of F-actin. Accurate measurements of the filament width are not demonstrated. Therefore, I am not convinced that the thinner structures represent single-stranded filaments. To avoid the ambiguity, thorough image analysis and filament reconstruction are required. An alternative interpretation could be that the thinner filaments represent overall classical double-stranded actin filaments, while the thicker counterparts may represent two filaments aligned along their sides.

Response 8: We have clarified this situation in the text (Materials and methods). EM analysis of G-actin was carried out with a freshly obtained preparation. To convert G-actin into F-actin, up to 0.1 M KCL was added to the preparation of G-actin and incubated at room temperature for about 20 hours (according to EM analysis, after incubation for 5-10 hours, not all monomeric actin is converted into filamentous form).

The incubation of actin for a day at 37^oC was used for MS analysis.

We have added three new Figures (Figures 1-3) with EM images (see our answer above).

In the places where F-actin are bent, actin monomers are adsorbed on the film by the lateral surfaces and structures in the form of a "ladder" are observed (Figure 3B, closed arrows). The diameter in such areas of the fibril is about 3-4 nm. In both projections, the monomers interact by ring to ring slightly overlapping with each other. It is logical to assume that if F-actin consisted of two filaments, then we would observe fibrils only with a diameter of about 6-7 nm. Thus, the EM analysis data indicate that F-actin is most likely to have a single-stranded organization. Moreover, we have added X-ray diffraction data for G- and F-actin, which demonstrate the absence of double-helix organization for F-actin.

The filaments are heterogeneous and it is not possible to do reconstruction.

 

• Splitting of F-actin to individual strands, albeit only on a short distance, has been proposed and demonstrated by cryo-EM (McGough and Chiu, 1999). Negative staining is, however, is prone to aberrations. In combination with prolonged incubation at 37oC under ATP-depleting conditions, the aberrations can be dramatic and likely to not represent physiological conditions. While the classical double-stranded arrangement has received countless confirmations of its physiological authenticity, the proposed models are poorly substantiated by the provided experimental data.

Response 9: We did not use prolonged actin incubation at 37 degrees prior to mesh application.

 

• The authors are encouraged to conduct reconstruction of the observed filaments.

Response 10: The filaments are heterogeneous and it is not possible to do 2-D reconstruction as we did for Abeta peptide and its fragments (doi: 10.3233/ADR-180063; doi: 10.1021/acs.langmuir.7b03393).

 Minor comments:

• Actin concentration cannot be reliably measured at 280 nm in the presence of ATP due to a strong absorbance of the latter. OD290 with a different extinction coefficient or with the same coefficient but in the presence of 0.5 M NaOH is recommended.

Response 11: The actin concentration was measured on a two-beam spectrophotometer, where the control was a solution with the same ATP content, but without protein. The protein concentration is measured quite accurately.

Reviewer 3 Report

In this paper, Glyakina and co-workers propose a new model of stacking actin monomers in filamentous actin. The authors first analyze some existing PDB structures that were used to propose a double-helix model. The structures, shown in Fig. 1, are essentially similar, as revealed by their small RMSD, which in this case measures the divergence between the structures. The authors then image actin filaments purified from rabbit skeletal muscle. Their analysis of the images, particularly those seen in a lateral fashion, allow them to conclude that the filaments are apparently made of linear actin monomers. To substantiate this extraordinary claim in a more convincing fashion, the authors perform limited proteolysis on G-actin and F-actin. The overall idea is that proteolysis-protected regions are part of the interactive region between monomers, whereas exposed parts are not. Calculation of the accessibility area allow the authors to conclude that their data do not fit the current model of actin filament, which involves a double helix. Hence, they propose two new models, a ring-like and helix-like model.

Overall, the paper is interesting and very provocative, as it goes against the well-established models of actin polymerization. The controversial nature of the data itself is worth publishing, but the authors need to provide additional information and testing to substantiate their claims in their entirety.

KEY ISSUES

1. Figure 1 needs some form of quantification/statistics. One possibility is to show the distribution of the filament thickness in regions of interest selected by the ability of the authors to isolate single chains (as shown in Fig. 1). Wide-laying 7-8nm filaments should account for approximately 50% of the filaments, and lateral-laying (3-4nm) filaments should be the other 50%, by the laws of probability.
2. The EM images in Fig.1 are beautiful in their heterogeneity, but raise some concerns as to the generality of the authors’ considerations. Is their conclusion of the position of the filament solely based on the thickness of the filament? Given that the prep is not made of pure recombinant actin, but rabbit muscle actin, there could be some actin cross-linkers associated that change the appearance of the filament. Hence, the authors need to show that there are no cross-linkers in the part of the images/preps they are using.
3. Concern #2 also translates to the proteolytic protection assays. If actin cross-linkers form interactive surfaces with actin monomers, it stands to reason that these actin surfaces are protected from proteolysis. I think it’s essential that the authors assess the amount of cross-linker contamination in their samples and/or analyze a filament prep using recombinant actin.
4. Regarding the proteolytic assay, the mixture of proteases includes proteinase K. How it is possible that one peptide is protected in the mixture (343-348) and not in the proteinase K? If the same protease is present, it should cut. Is it due to the rate of protease: substrate?
5. Given the potential importance of these findings, an extended discussion is needed.
6. Also, the grammar and overall use of English throughout the papers needs to be improved.

Author Response

Comments and Suggestions for Authors

In this paper, Glyakina and co-workers propose a new model of stacking actin monomers in filamentous actin. The authors first analyze some existing PDB structures that were used to propose a double-helix model. The structures, shown in Fig. 1, are essentially similar, as revealed by their small RMSD, which in this case measures the divergence between the structures. The authors then image actin filaments purified from rabbit skeletal muscle. Their analysis of the images, particularly those seen in a lateral fashion, allow them to conclude that the filaments are apparently made of linear actin monomers. To substantiate this extraordinary claim in a more convincing fashion, the authors perform limited proteolysis on G-actin and F-actin. The overall idea is that proteolysis-protected regions are part of the interactive region between monomers, whereas exposed parts are not. Calculation of the accessibility area allow the authors to conclude that their data do not fit the current model of actin filament, which involves a double helix. Hence, they propose two new models, a ring-like and helix-like model.

Overall, the paper is interesting and very provocative, as it goes against the well-established models of actin polymerization. The controversial nature of the data itself is worth publishing, but the authors need to provide additional information and testing to substantiate their claims in their entirety.

Response: We would like to thank the reviewer for the valuable suggestions. We have added three new Figures (1-3) with electron microscopy images to clarify the organization of G- and F-actin, Supplementary materials with MS data and SDS-electrophoresis to demonstrate the quality of the original protein. Moreover, we have added X-ray diffraction data for G- and F-actin, which demonstrate the absence of double-helix organization for F-actin.

KEY ISSUES

1. Figure 1 needs some form of quantification/statistics. One possibility is to show the distribution of the filament thickness in regions of interest selected by the ability of the authors to isolate single chains (as shown in Fig. 1). Wide-laying 7-8nm filaments should account for approximately 50% of the filaments, and lateral-laying (3-4nm) filaments should be the other 50%, by the laws of probability.

Response 1: This is not true. Wide-laying 7-8 nm filaments are preferred. We have added three new Figures with EM images to clarify this situation and new text.

 

2. The EM images in Fig.1 are beautiful in their heterogeneity, but raise some concerns as to the generality of the authors’ considerations. Is their conclusion of the position of the filament solely based on the thickness of the filament? Given that the prep is not made of pure recombinant actin, but rabbit muscle actin, there could be some actin cross-linkers associated that change the appearance of the filament. Hence, the authors need to show that there are no cross-linkers in the part of the images/preps they are using.

Response 2: We have provided SDS-electrophoresis to demonstrate the quality of the original protein. We did not observe any actin cross-linkers, only actin with purity more than 95%.

 

3. Concern #2 also translates to the proteolytic protection assays. If actin cross-linkers form interactive surfaces with actin monomers, it stands to reason that these actin surfaces are protected from proteolysis. I think it’s essential that the authors assess the amount of cross-linker contamination in their samples and/or analyze a filament prep using recombinant actin.

Response 3: We get a fairly pure protein without impurities. MS analysis of the actin band after electrophoresis does not register additional proteins and impurities.

 

4. Regarding the proteolytic assay, the mixture of proteases includes proteinase K. How it is possible that one peptide is protected in the mixture (343-348) and not in the proteinase K? If the same protease is present, it should cut. Is it due to the rate of protease: substrate?

Response 4: In the case of mixture of proteases the concentration of Proteinase K is 3 times less. We have checked the relative content of this fragment. It turns out that this fragment has a small contribution (on the verge of reliability).

 

5. Given the potential importance of these findings, an extended discussion is needed.

Response 5: Thank you. We have added.

 

6. Also, the grammar and overall use of English throughout the papers needs to be improved.

Response 6: We have corrected.

Reviewer 4 Report

This is a carefully done study and the findings are of considerable interest. As this paper seems already type edited, and no serious criticisms regarding methodology, results and interpretation of results. I think this manuscript is now acceptable for publication.

 

 

 

Round 2

Reviewer 2 Report

In the revised version of the manuscript, the authors expanded their original findings by presenting more EM images. The EM images are of a high and rather impressive for negative staining resolution. The observed novel structures of actin filaments deserve further investigation. Yest, several aspects of the study require clarification, and several of my original concerns were not carefully addressed.

- 0.2mg/ml of actin (~5uM) should polymerize within 30-60 minutes under the indicated conditions. The fact that incubation for 5-10 hours was not sufficient to convert all the actin into filaments is a concern for several reasons. That may point to a low quality actin preparation, but more likely on the unnecessarily long and potentially damaging incubation. It is not clear how the conclusion on incomplete polymerization of actin was made. Also, there is no need to have only F-actin to observe and analyze filaments, and, in fact, a small fraction of actin (~0.1uM) will always remain in the G-form even with fresh actin in the presence of physiological salt and Mg2+-ATP concentrations.

- I recommend the authors should analyze the filament structure after 30-60 minutes of polymerization and compare it with those polymerized for 20 hours and presented on Figs. 2-3. Such a comparison should help to clarify the source of the observed structural difference and will be very beneficial to the readers. I have little doubt that the authors will see the classical forms of the filaments after shorter incubation.

- Heterogeneity of the samle makes reconstruction more difficult, but does not preclude it as a single molecule approach can be and has been applied in such situations (see Curr Opin Struct Biol  2007 17(5):556-61. Single-particle reconstruction from EM images of helical filaments).

- I still don’t see how the proposed models may accurately represent the overall straight filaments demonstrated on Figs. 2-3.

- The addition of the SDS-electrophoresis profile of the original protein is very helpful. As to whether it is worth controlling the course of degradation, I certainly think so since the experiment is meant to reveal contacts in filamentous actin. If the experiment continues for hours after the filaments are degraded, the argument is not sufficiently justified.

- Explanation of the disulfide bonds requires minor adjustments to the filament structure and points on conformational flexibility rather than a completely different structure. In fact, to my knowledge, the disulfide crosslinking data largely agree with and support the classical double-stranded structures of F-actin.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors have improved the manuscript significantly and the new EM images are beautiful and insightful. Please have the manuscript re-checked for English grammar, but now the article is scientifically acceptable for publication.

I'm recommending this paper for highlight in the journal due to its groundbreaking nature.

Author Response

Comments and Suggestions for Authors

The authors have improved the manuscript significantly and the new EM images are beautiful and insightful. Please have the manuscript re-checked for English grammar, but now the article is scientifically acceptable for publication.

I'm recommending this paper for highlight in the journal due to its groundbreaking nature.

Response: We would like to thank the reviewer for the valuable suggestions. We have re-checked the manuscript for English grammar.

Round 3

Reviewer 2 Report

If the authors' statements about the polymerization behavior of actin in their hands are correct, then the purified actin is of low quality.  The polymerization of rabbit actin is exceptionally well documented. Its critical concentration for polymerization (Cc) at the barbed end is ~0.1uM (0.004mg/ml). At concentration 3mg/ml, it should polymerize fully in a matter of few minutes. Actin should be applied to em grids in concentrations no higher than 1-2 uM to avoid overcrowdedness.