Simultaneous membrane and RNA binding by tick-borne encephalitis virus capsid protein

  1. Lauri Ilmari Aurelius Pulkkinen
  2. Sarah Victoria Barrass
  3. Marie Lindgren
  4. Hudson Pace
  5. Anna K. Överby
  6. Maria Anastasina
  7. Marta Bally
  8. Richard Lundmark
  9. Sarah Jane Butcher

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Tick-borne encephalitis virus is an enveloped, pathogenic, RNA virus in the family Flaviviridae , genus Flavivirus . Viral particles are formed when the nucleocapsid, consisting of an RNA genome and multiple copies of the capsid protein, buds through the endoplasmic reticulum membrane and acquires the viral envelope and the associated proteins. The coordination of the nucleocapsid components to the sites of assembly and budding are poorly understood. Here, we investigate the interactions of the wild-type and truncated capsid proteins with membranes with biophysical methods and model membrane systems. We show that capsid protein initially binds membranes via electrostatic interactions with negatively-charged lipids, which is followed by membrane insertion. Additionally, we show that membrane-bound capsid protein can recruit viral genomic RNA. We confirm the biological relevance of the biophysical findings by using mass spectrometry to show that purified virions contain negatively-charged lipids. Our results suggest that nucleocapsid assembly is coordinated by negatively-charged membrane patches on the endoplasmic reticulum and that the capsid protein mediates direct contacts between the nucleocapsid and the membrane.

Article activity feed

  1. Version published to 10.1371/journal.ppat.1011125
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Manuscript number: RC-2022-01707

    Corresponding author(s): Sarah Butcher, Richard Lundmark

    1. General Statements [optional]

    We thank the reviewers for their insightful comments. The inclusion of the points raised by the referees have strengthened the manuscript. However, some of the reviewer suggestions are beyond the scope of the work (see below), but will doubtlessly be touched upon in future studies by the authors. In addition to incorporating changes relevant to answering the reviewers’ comments, we have edited the manuscript for increased clarity and precision.

    2. Description of the planned revisions

    1. Liposome flotation assay Reviewer #1 suggested that we should perform a liposome floatation assay to separate possible C protein aggregation from membrane binding: "I would strongly recommend supplementing the current liposome sedimentation assay by liposome flotation assay. In contrast to liposome co-sedimentation, the flotation assay can discriminate protein aggregates from proteins bound to liposomes. Although the SDS PAGE shown in Fig. 1A looks pretty convincing, a faint protein band in the „P" lane of the middle panel for the (-) sample is evident. Therefore, C protein aggregation cannot be ruled out and it would be indistinguishable from liposome binding examined by mere co-sedimentation assay”

    Response: We agree that this is a necessary control experiment to add, and we will perform it with liposomes containing 40 % POPS. As we detected complete C protein co-sedimentation with this lipid composition, performing the flotation experiment with the same composition will prove that the earlier result indicates lipid binding and not protein aggregation.

    3. Description of the revisions that have already been incorporated in the transferred manuscript

    1. Reviewer #1
    2. “In addition, it needs to be clarified which TBEV C protein construct, whether full-length or truncated, was used for co-sedimentation fragmentation.”

    Response: We have clarified in this section of the manuscript that the full-length C protein construct was used for the liposome co-sedimentation assays by adding “full-length” prior to instances of “C protein” e.g. in the paragraph starting line 118.

    1. “How to understand the finding that „the C protein forms a very rigid layer when adsorbed to the membrane". Can the aggregation of C-protein be ruled-out? *Following the 1M NaCl wash of C-protein-bound to SLB, the authors stated: „This shows that initial membrane recruitment of C protein is strongly dependent on its interactions with the negatively-charged lipid headgroups. However, once bound, the C protein-membrane interaction is complemented with non-electrostatic interactions such as membrane insertion or protein oligomerization": does it mean that there are several layers of C protein, the first held by electrostatic interactions, overlayed by non-electrostatically bound C protein? If yes, the illustration of single-layered C-protein adsorbed onto SLB in Fig. 2A, B is not correct. *”

    Response: We understand the confusion regarding the term “rigid” which was used as a way to describe how we interpret the relatively minor change in the dissipation upon membrane binding. What we intended to describe was that this indicates that the protein is attached in a stable way that does not add viscoelastic properties to the system. These data indicate that the protein does not form large aggregates that non-specifically attach to the membrane in different protrusive orientations. We have clarified this in the manuscript and specified that the as there is no dissipation change, there is no aggregation. We added the following to line 168 “This, in turn, indicates that the C protein does not bind as non-specific aggregates as these would have changed the viscoelastic properties of the system.”

    We do not mean that there are several layers of C protein. We consider, due to the highly charged nature of C, that the most likely explanation is that there are multiple modes of C binding but the result is only one layer, with multiple C-proteins interacting with each other within that layer. We have modified the text at line 184 to: “However, once bound, the C protein-membrane interaction is complemented with non-electrostatic interactions such as membrane insertion or protein oligomerization within the bound layer.”

    1. “The sentence: “To confirm that the C protein is biologically active, we investigated its ability to bind RNA" seems to be a little odd because it suggests the model membrane binding assays do not require biological active proteins. However, considering that the interactions leading to binding either negatively-charged lipid or negatively-charged RNA are electrostatic - this sentence must be rewritten.” Response: We thank the reviewer and have now rephrased this sentence to the following at line 249 “Since RNA binding is crucial for the NC assembly, we investigated the C protein’s ability to perform this function.”

    2. “The authors´ statement in the Abstract: „....we investigate nucleocapsid assembly..." is too speculative because the assembly was not studied in their work. It needs to be reformulated.” Response: We agree, and the statement has been removed from the abstract.

    3. “Despite this clear and valuable methodological contribution, the authors' contribution to our knowledge of the coordination of the nucleocapsid components to the sites of assembly and budding is not so obvious. Contrary to the earlier idea that the flavivirus is asymmetrically charged (that is, hydrophobic on one side (α2) and positively charged on the other side (α4), recent studies show that the entire surface of the protein is highly electropositive (Mebus-Antunnes et al., 2022). Therefore, a well-ordered neutralization of the flaviviral C proteins' highly positive surface seems critical for the proper organization and assembly of nucleocapsid. I am afraid that the authors do not shed much light on this issue.” Response: The recent structure of the TBEV C protein, published after we submitted the manuscript, shows that indeed the C protein is highly positively charged on all surfaces (updated Supplementary Figure 1 and Selinger et al., 2022). The recruitment of C protein to the membrane, that we demonstrate is dependent on negatively-charged head groups, provides a biologically relevant mechanism for charge neutralization on the C protein surface that interacts with the lipids. The remaining surface charge can be then neutralized by RNA recruitment. Mebus-Antunnes et al. made their observations with just RNA and C protein from Dengue virus in the context of artificial surfaces e.g. mica. However, our experiments utilize the TBEV C protein and specifically include a membrane, the third critical component of NC assembly. Thus, we build upon the work of Mebus-Antunnes et al. by adding a second biologically relevant charge-neutralising component and comparing with a distantly-related virus. We have changed the discussion section of the manuscript to reflect this new structure and to emphasize the advance here. Starting from line 371 we changed the text to: “Recently, it has been shown that the neutralization of the C protein surface positive charge is important for RNA binding in the distantly-related Dengue virus (DENV) (Mebus-Antunes et al, 2022). The recruitment of C protein to the membrane, that we demonstrate is dependent on negatively-charged head groups, provides a biologically relevant mechanism for charge neutralization on the C protein surface that interacts with the lipids. The remaining surface charge can be then neutralized by RNA recruitment.”

    Reviewer #2

    1. “What results demonstrate C protein inserts into membrane? The current results support the C protein interacts with membranes with positive charge, but do not seem to demonstrate membrane insertion. If the C protein inserts into the membrane, which regions (helices) play this role?”

    Response: The Langmuir-Blodgett trough tensiometry experiments with monolayers directly measure the insertion of a protein into the monolayer. By determining the maximum insertion pressure of the C protein constructs, we also show that the membrane insertion can occur in bilayers. We show that the N-terminus is not inserting into the membrane, further work, beyond the scope of this manuscript, is needed to pinpoint the residues responsible for insertion, for instance by hydrogen-deuterium exchange or FRET measurements that would not affect folding. To clarify the use of the LB trough, we added the following at line 216: “To investigate if the C protein membrane binding includes insertion into the membrane after the initial electrostatic binding, we used Langmuir-Blodgett trough monolayer experiments. In this approach, the insertion of a protein into a lipid monolayer can be detected by following the pressure (π) of the monolayer after protein injection into the aqueous subphase, with increases in π corresponding to protein injection (Brockman, 1999; Liu et al, 2022).“

    1. “The authors should discuss several previous papers reporting the effect of partial deletions of the C gene on the replication of TBEV, West Nile virus, and other flaviviruses.” Response: We agree that this is a necessary addition, and have now added a paragraph in the discussion section starting at line 333: “N-terminally truncated flaviviral C proteins have been shown to be assembly competent and in vitro, able to bind RNA, which is consistent with our results with N-terminally truncated TBEV C protein (Khromykh & Westaway, 1996; Kofler et al, 2002; Patkar et al, 2007; Schlick et al, 2009). One role of C is in the modulation of host responses to infection and the N-terminus maybe involved in that (Yang et al, 2002; Limjindaporn et al, 2007; Colpitts et al, 2011; Bhuvanakantham & Ng, 2013; Katoh et al, 2013; Urbanowski & Hobman, 2013; Samuel et al, 2016; Slomnicki et al, 2017; Fontaine et al, 2018). The membrane insertion directly detected in our experiments is central to C protein function. Other studies have found that deletions in the hydrophobic region of the α2 helix significantly impair particle assembly (Kofler et al, 2002; Patkar et al, 2007; Schlick et al, 2009). In the light of this evidence, we consider that the α2 helix could be responsible for membrane insertion (Markoff et al, 1997; Kofler et al, 2002; Nemésio et al, 2011, 2013).”

    Reviewer #3

    1. “In Figure 4, the band (256:1) that are supposedly in the wells (red arrow) is not clear- it is only slightly darker than the other wells.”

    Response: This confusion was the result of unclear wording. We have now revised the figure legend at line 278 to : “The black arrow indicates the bands of freely-migrating RNA, and the red arrow the wells. On lanes 624:1 and 256:1, RNA has been immobilized in the wells.”

    1. “Figure S1A, the N-terminal end (which is truncated in the mutant) should be colored on the cyan molecule.” Response: We have coloured the truncated part of the cyan molecule in the figure (now S1B) according to the reviewer’s comment.

    Other

    1. As the nuclear magnetic resonance structure of the truncated TBEV C protein has recently been released (Selinger et al, 2022), we have updated the manuscript and Figure S1 to include the information from this structure. We have also generated a new homology model of the full-length TBEV C protein using this structure as a template and included that in Figure S1.

    4. Description of analyses that authors prefer not to carry out

    1. Reviewer #1
    2. “However, we do not know whether in the infected cells, the C protein is pre-bound to ER membrane or to viral RNA. Having such a unique assay in their hands, I wonder whether the authors could use the pre-bound C protein with genomic RNA (i.e. the experiment shown in Fig. 4A) ribonucleoprotein complex in the SLB binding assay. If doable, this experiment would be exciting and could bring some important information about NC assembly.”

    Response: We agree that it would be very interesting to decipher if the C-protein first binds to RNA or to membranes using the QCM-D methodology. Yet, our data on pre-incubated C-protein and RNA suggests that large aggregates are formed which would hamper the interpretation of the QCM-D data. Furthermore, based on the suggested experiment, we will not be able to firmly conclude whether or not the C-protein first binds to RNA or to membranes since the time of the experiment will allow rearrangement of preformed complexes between C-protein and RNA. Additionally, the QCM-D measurement cannot differentiate if the preformed complexes bind on their own, or if excess unbound C protein binds the membrane and then recruits the complex. Therefore, addressing this question would require major adjustments to the RNA model system and methodology that we feel are beyond the scope of this study.

    Reviewer 2

    1. “The authors should use the lipids detected in the virions to confirm C protein binding experiments.”

    Response: In the mass spectrometry characterization of the TBEV virions, we detected lipids from 9 classes (Car, PE, PS, PI, PG, PC, Cer, HexCer & TG). We have tested four of them (PE, PS, PI, PC) in the liposome sedimentation assay. Additionally, we tested GalCer, which, like HexCer, are cerebrosides. Our liposome binding experiments clearly demonstrate that the C protein does not bind to a specific lipid class, but instead to lipids with negatively-charged headgroups. Therefore, we would argue that doing additional sedimentation experiments with Car, PG, Cer, and TG would not add extra insight to the manuscript.

    Additionally, while the population of lipid species in the TBEV envelope is diverse, the diversity mostly comes from differences in the lipid tails, which do not generally affect the head group-mediated binding of proteins. Therefore, performing additional lipid binding experiments with varying tail lengths would not likely lead to new observations.

    Finally, to perform the authentic experiment of testing C protein binding to liposomes formed from lipids extracted from purified virions would require orders of magnitude more virus sample than our research laboratory is capable of producing. Therefore, we argue that this experiment is beyond the scope of this study.

    1. “The study may be strengthened by performing virus mutagenesis experiments.” Response: While we agree that, ultimately, experiments on virus and cells would help to understand the role of the C protein in the biological context, we think these experiments are beyond the scope of this study. For virus mutagenesis, candidate residues should be first identified with biochemical and biophysical studies, which is already beyond the scope of this work. Additionally, the C protein has multiple functions in the host cell in addition to NC assembly, and interpreting the effect on the mutations on e.g. virus titer is difficult.

    Reviewer #3

    1. “In all figure legends, authors should write a conclusion line after the description of the experiments - what conclusion is drawn from each experiment.”

    Response: While we agree that adding such a conclusion line would make it easier for the reader to understand each figure, the format of the figure legends is highly subject to journal policy. Therefore, we think that the addition of such lines will be an editorial decision and will depend on the journal. We have, however strived to make the figure titles as informative as possible in lieu of such concluding lines.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Pulkkinen and co-authors, title: Simultaneous membrane and RNA binding by TBE virus capsid protein.

    This paper characterizes the ability of purified TBE capsid proteins to bind to different composition of lipids by biophysical methods and found that it prefers to bind to negatively charge lipids. The capsid then partially inserts into the membrane. Using mass spectrometry, they analyze the lipid composition of the purified TBE virus and showed they composed of negatively charge lipids thereby further supporting that the virus is likely first assembled where the negatively charge lipids are located in the endoplasmic reticulum. They also characterize the membrane bound capsid protein's ability to bind RNA and show they are able to bind. For all these experiments, they also included a capsid mutant with its N-terminal end deleted and show the mutant capsid protein activity doesn't not differ much from the whole capsid protein- thus showing the N-terminal end is likely not important for these processes. The experiments are well conducted and the manuscript is very clearly written.

    Comments:

    1. In Figure 4, the band (256:1) that are supposedly in the wells (red arrow) is not clear- it is only slightly darker than the other wells.

    Minor comments:

    1. In all figure legends, authors should write a conclusion line after the description of the experiments - what conclusion is drawn from each experiment.
    2. Figure S1A, the N-terminal end (which is truncated in the mutant) should be colored on the cyan molecule.

    Referees cross-commenting

    I agree with comments by Reviewers #1 and #2

    Significance

    The study here although done in an in vitro system illuminates the virus assembly process - the positively charged capsid protein binds to the negatively charge area of the endoplasmic reticulum membrane, the capsid then partially insert into the membrane, then capsid interacts with viral RNA genome to facilitate virus assembly process. This is a very detailed study of the initial steps of the virus assembly process.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Pulkkinen et al. performed biochemical and biophysical experiments to suggest (1) negatively charged lipids are required for TBEV C protein interacting with membrane, (2) the membrane-associated C protein could simultaneously bind viral RNA, (3) the first 17 amino acids are not required for (1) and (2), and (4) TBEV virions contain negatively charged lipids. The study is important and provides molecular insights in flavivirus assembly. The following points can substantiate the manuscript.

    Major points

    1. The authors should use the lipids detected in the virions to confirm C protein binding experiments.
    2. What results demonstrate C protein inserts into membrane? The current results support the C protein interacts with membranes with positive charge, but do not seem to demonstrate membrane insertion. If the C protein inserts into the membrane, which regions (helices) play this role?
    3. The study may be strengthened by performing virus mutagenesis experiments.
    4. The authors should discuss several previous papers reporting the effect of partial deletions of the C gene on the replication of TBEV, West Nile virus, and other flaviviruses.

    Significance

    This is an important study as indicated in the comments to authors.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    The authors characterized the interactions of recombinant, bacterially expressed full-length and N-terminally truncated C proteins of tick borne encephalitis virus with model membrane systems. They used a unique combination of biophysical methods, including protein liposome co-sedimentation, QCM-D measurement and Langmuir-Blodgett trough monolayer experiments. Their experiments showed that the binding of TBEV C to both liposomes and supported lipid bilayer (SLB) is strongly dependent on the presence of negatively charged lipids. They also showed that following the initial electrostatic binding to the model lipid membrane, both C protein variants absorb to the SLB and form rigid layers which are stabilized by non-electrostatic interactions. By Langmuir-Blodgett trough monolayer experiments they demonstrated that negatively charged lipids are needed for C protein membrane insertion. The SLB bound C proteins, either full-length or N-terminally truncated, were shown to bind in vitro transcribed TBEV genomic RNA. Finally, to prove their major finding that negatively charged lipid head groups are crucial for C protein interaction with the lipid membrane, the authors analyzed the lipid content of the purified virions.

    This work deals with the central role of the C protein, namely with its binding to the lipid membrane and genomic RNA. In the infected cells, this process leads to nucleocapsid assembly, a step which is poorly understood. The authors demonstrate that the membrane affinity of the C protein is conditioned by the presence of negatively charged polar heads. The text and figures are clear and accurate. The results obtained from three independent methodological approaches are solid and confirm the importance of electrostatic interactions for a contact of C protein with the membrane. As highly interesting, I considered the observation that the C protein, while bound to the model membrane (SLB), still retains its ability to bind RNA. Although their data did not show anything about the orientation of the C protein in SLB, this methodology opens the way to how, using suitable mutants of TBEV C, this can be found. I am sure that the authors are aware of the possibilities of studying a series of the TBEV C mutants with impaired membrane or RNA binding. Therefore, I assume that the authors' primary focus here is to show new methodological approaches to the simultaneous measurement of C protein interactions with model membranes and RNA, and some data obtained on the abovementioned mutants will be published afterwards.

    Major comments:

    1. One of the fundamental challenges of the work with flaviviral capsid proteins is that they tend to form amorphous aggregates to neutralize their highly positive surface charge. As the authors state themselves, „ We cannot rule out that the C protein preparation is heterogeneous..." I would strongly recommend supplementing the current liposome sedimentation assay by liposome flotation assay. In contrast to liposome co-sedimentation, the flotation assay can discriminate protein aggregates from proteins bound to liposomes. Although the SDS PAGE shown in Fig. 1A looks pretty convincing, a faint protein band in the „P" lane of the middle panel for the (-) sample is evident. Therefore, C protein aggregation cannot be ruled out and it would be indistinguishable from liposome binding examined by mere co-sedimentation assay. In addition, it needs to be clarified which TBEV C protein construct, whether full-length or truncated, was used for co-sedimentation fragmentation.
    2. In section: Initial C protein recruitment to the membrane is of an electrostatic nature How to understand the finding that „the C protein forms a very rigid layer when adsorbed to the membrane". Can the aggregation of C-protein be ruled-out?

    Following the 1M NaCl wash of C-protein-bound to SLB, the authors stated: „This shows that initial membrane recruitment of C protein is strongly dependent on its interactions with the negatively-charged lipid headgroups. However, once bound, the C protein-membrane interaction is complemented with non-electrostatic interactions such as membrane insertion or protein oligomerization": does it mean that there are several layers of C protein, the first held by electrostatic interactions, overlayed by non-electrostatically bound C protein? If yes, the illustration of single-layered C-protein adsorbed onto SLB in Fig. 2A, B is not correct.

    1. C protein inserts into membranes It is beyond the frame of this work; however, it would be nice to show whether mutations of amino acid residues within the hydrophobic segment of TBEV C, which are in other flaviviral C proteins considered responsible for hydrophobic interaction, can abolish the membrane interaction.
    2. Membrane-bound C protein can recruit TBEV genomic RNA. The sentence „ To confirm that the C protein is biologically active, we investigated its ability to bind RNA" seems to be a little odd because it suggests the model membrane binding assays do not require biological active proteins. However, considering that the interactions leading to binding either negatively-charged lipid or negatively-charged RNA are electrostatic - this sentence must be rewritten.
    3. The authors state, "These data show that membrane-bound C protein is capable of recruiting TBEV genomic RNA at the membrane, suggesting that this also happens in the context of NC assembly". However, we do not know whether in the infected cells, the C protein is pre-bound to ER membrane or to viral RNA. Having such a unique assay in their hands, I wonder whether the authors could use the pre-bound C protein with genomic RNA (i.e. the experiment shown in Fig. 4A) ribonucleoprotein complex in the SLB binding assay. If doable, this experiment would be exciting and could bring some important information about NC assembly.

    Minor comments:

    The authors´ statement in the Abstract: „....we investigate nucleocapsid assembly..." is too speculative because the assembly was not studied in their work. It needs to be reformulated.

    Referees cross-commenting

    I agree with the Reviews by reviewers #2 and #3

    Significance

    This manuscript's major novelty and originality are in using a unique combination of biophysical methods, including quartz crystal microbalance with dissipation monitoring and Langmuir-Blodgett trough. Using quartz crystal microbalance with dissipation, the authors confirmed the necessity of negatively charged lipid components of the model lipid membrane for C-protein binding. Furthermore, this method also allows them to measure the formation of a rigid layer of C protein stabilized by non-electrostatic interactions. By Langmuir-Blodgett trough monolayer experiments, they demonstrated the insertion of TBEV C protein into the model membrane. However, I do not have sufficient expertise to evaluate the correctness of the experiments done by these two methodologies.

    Despite this clear and valuable methodological contribution, the authors' contribution to our knowledge of the coordination of the nucleocapsid components to the sites of assembly and budding is not so obvious. Contrary to the earlier idea that the flavivirus is asymmetrically charged (that is, hydrophobic on one side (α2) and positively charged on the other side (α4), recent studies show that the entire surface of the protein is highly electropositive (Mebus-Antunnes et al., 2022). Therefore, a well-ordered neutralization of the flaviviral C proteins' highly positive surface seems critical for the proper organization and assembly of nucleocapsid. I am afraid that the authors do not shed much light on this issue.

  6. Version published to 10.1101/2022.10.06.511088v1 on bioRxiv