Human cytomegalovirus deploys molecular mimicry to recruit VPS4A to sites of virus assembly

  1. Benjamin G. Butt
  2. Daniela Fischer
  3. Alison R. Rep
  4. Martin Schauflinger
  5. Clarissa Read
  6. Thomas Böck
  7. Manuel Hirner
  8. Stephen C. Graham
  9. Jens von Einem

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The AAA-type ATPase VPS4 is recruited by proteins of the endosomal sorting complex required for transport III (ESCRT-III) to catalyse membrane constriction and membrane fission. VPS4A accumulates at the cytoplasmic viral assembly complex (cVAC) of cells infected with human cytomegalovirus (HCMV), the site where nascent virus particles obtain their membrane envelope. Here we show that VPS4A is recruited to the cVAC via interaction with pUL71. Sequence analysis, deep-learning structure prediction, molecular dynamics and mutagenic analysis identify a short peptide motif in the C-terminal region of pUL71 that is necessary and sufficient for the interaction with VPS4A. This motif is predicted to bind the same groove of the N-terminal VPS4A Microtubule-Interacting and Trafficking (MIT) domain as the Type 2 MIT-Interacting Motif (MIM2) of cellular ESCRT-III components, and this viral MIM2-like motif (vMIM2) is conserved across β-herpesvirus pUL71 homologues. However, recruitment of VPS4A by pUL71 is dispensable for HCMV morphogenesis or replication and the function of the conserved vMIM2 during infection remains enigmatic. VPS4-recruitment via a vMIM2 represents a previously unknown mechanism of molecular mimicry in viruses, extending previous observations that herpesviruses encode proteins with structural and functional homology to cellular ESCRT-III components.

Article activity feed

  1. Review Commons

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

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    Summary:

    In this study the authors apply a rigorous and thorough combination of approaches including sequence analysis, deep-learning structure predictions, molecular dynamics, cell imaging and mutagenic analyses to identify a short MIM2-mimicking motif in the C-terminal region of the pUL71 protein of HCMV (and homologues in other beta-herpesviruses) that is necessary and sufficient for interaction with the ESCRT terminal ATPase VPS4A. pUL71 uses this motif to recruit, or sequester, VPS4A to the HCMV cytoplasmic viral assembly complex, though this process is dispensable for HCMV morphogenesis or replication. The identified pUL71 sequence functions as a mimic of the MIM2 motif of cellular CHMP subunits since, like MIM2, it directly binds the groove in the MIT domain found at the N-terminus of VPS4.

    Major comments:

    1). There appears to be some confusion in the coip experiment in Figure 5D. From the upper blot in 5D, the "+" above each lane suggests there should be VPS4A-FLAG protein in every sample other than the two lanes at the very left of the gel, however the anti-FLAG ip does not pull down VPS4A-FLAG from every "+" lane, but from alternating ones (and from the next to the leftmost lane, which should lack VPS4A-FLAG). Similarly, the lower "Input" blot shows VPS4A-FLAG present in alternating lanes across the blot, which does not match the "+" and "-" labeling at the top of the figure. Conversely, there is anti-HA signal in most input lanes (lower blot) though the HA-tagged pUL71 homologues should be absent from alternate lanes (top of upper blot).

    We apologise and thank the reviewer for spotting this annotation error. Figure 5D has been updated to correctly show the samples used for each lane of the IP.

    2). The Discussion is an excellent, comprehensive and scholarly assessment of the implications of this work. One appealing hypothesis is that pUL71 may be sequestering VPS4A rather than using it for envelope scission. In this regard, the authors point out that VPS4A sequestration is supported by the finding that the VPS4A MIT domain binds the isolated pUL71 vMIM2 more tightly (~ 5 fold lower Kd) than the MIM2 of CHMP6, and that pUL71 and homologues are highly abundant at later stages of viral infection, allowing them to compete effectively with endogenous CHMP6 for VPS4A. I like the sequestration model very much, but could the authors comment on the fact that this apparent sequestration is seen even in the transfection experiments in Fig. 2A and 3G, where essentially 100% of transfected WT VPS4A-FLAG is recruited to the pUL71 compartment. Even given the increased binding affinity to pUL71, this suggests that in these transfection studies pUL71 must be in excess over the sum of both endogenous and transfected VPS4. Do the authors know if this is the case, and do cells transfected with pUL71 in these experiments exhibit any cytotoxicity, or cell cycle arrest, indicative of a block in normal ESCRT function/cytokinesis?

    *With regards to the transfection experiment, the levels of pUL71 and VPS4A-FLAG expression varied across the fields of view. It is therefore hard to make a definitive statement regards the level of pUL71 expression that gives complete sequestration of VPS4A-FLAG. In transient expression experiment, cells were transfected with equal amounts of DNA for VPS4A-FLAG and pUL71 expression vectors and analysed 20 to 24 hours after transfection. Sequestration of VPS4A-FLAG by pUL71, or lack of sequestration due to mutations, was consistently observed and was thus the predominant phenotype. However, degrees in the appearance of this phenotype were noted, which were likely caused by differences in expression levels. The images in Figs. 1, 2, 3 and 5 are confocal images of selected cells that represented the predominant phenotype. In our opinion, no clear statements can be made about expression levels and their relationship with respect to sequestration of VPS4A, as transient expression gives considerable cell-to-cell variability in expression levels. *

    In MRC-5 cells transiently expressing VPS4A-FLAG under doxycycline control and infected with different strains of HCMV we see strong sequestration of VPS4A-FLAG (Fig. 6B). While VPS4A-FLAG sequestration is not always complete in the context of infection (compare WT infection in Fig. 6C and Fig. 6D), presumably because of differing VPS4A-FLAG levels, it is reasonable to assume that ectopic VPS4A-FLAG expression increases the total pool of VPS4A available. Thus, in the context of infected cells we would expect the vast majority of cellular VPS4A to be sequestered by pUL71, also considering the strong expression of pUL71 during infection. However, we note that evenly distributed signals in the cytoplasm are more difficult to visualize than concentrated signals (such as localization at the Golgi), especially in confocal images, which could contribute to the impression that almost 100% of VPS4A-FLAG is sequestered by pUL71. We have therefore added the following three sentences to the third paragraph of the discussion:

    “In the context of infection, pUL71 yields strong sequestration of ectopically expressed VPS4A-FLAG (Fig. 7A–C). As ectopic expression would be expected to increase the total pool of VPS4A present in cells, we anticipate extensive pUL71-mediated sequestration of endogenous VPS4A in HCMV-infected cells. However, we note that diffuse cytoplasmic signals are more difficult to visualise than organelle-associated signals in confocal microscopy, and it is therefore possible that some VPS4A remains free in the cytoplasm even in the presence of abundant pUL71.”

    Unfortunately, we are unaware of a high quality VPS4A antibody suitable for immunofluorescence microscopy that would allow us to probe the localisation of endogenous VPS4A directly.

    *The reviewer raises an interesting point with regard to the potential blocking of cellular ESCRT functions in the presence of transfected pUL71. We did not find indications of a block in ‘normal’ ESCRT functions like cytokinesis in cells expressing pUL71 (or mutant versions thereof). We therefore investigated the function of ESCRT in pUL71-expressing cells by assessing whether expression of pUL71 can inhibit the function of VPS4 in the release of HIV Gag virus like particles (VLPs). The results of these studies have been added to the manuscript as supplemental figure 7 – they show no evidence for a functional inhibition of VPS4 by co-expression of pUL71. We have added a section to the results describing this experiment, plus the following section in the discussion: *

    “We did not observe any defect in ESCRT-mediated Gag VLP production in the presence of pUL71 (Fig. S7), suggesting that transient expression of pUL71 is not sufficient to inhibit cellular ESCRT activity. However, we note that studies analysing the role of VPS4 in ESCRT-mediated virus budding generally exploit dominant-negative forms of VPS4A or VPS4B (Corless et al., 2010; Horii et al., 2006; Pawliczek and Crump, 2009; Taylor et al., 2007). As human VPS4A and VPS4B interact with each other (Scheuring et al., 2001), overexpressing dominant-negative mutants of either protein would be expected to poison the activity of both via formation of heteromeric APTase hexamers. Studies using CRISPR/Cas9 gene editing show only modest defects in VLP budding when VPS4A or VPS4B are deleted individually, with the VPS4B deletion causing a greater VLP budding defect than VPS4A deletion (Harel et al., 2022), and the MIM2 of CHMP6 has higher affinity for the MIT domain of VPS4A than of VPS4B (Wenzel et al., 2022). While we have not investigated the interaction between pUL71 and the VPS4B MIT domain in this study, it is possible that pUL71 has higher affinity for VPS4A than VPS4B and pUL71 expression may thus lead to selective sequestration of only one VPS4 isoform.”

    *In light of the above new results and discussion, we can neither confirm nor rule out that pUL71 modulates ESCRT functions by sequestration of VPS4. We agree with the reviewer that it is an extremely interesting hypothesis but addressing it properly would require thorough experimental investigation, which we feel is a substantial study in its own right and is beyond the scope of this manuscript. Lastly, we apologise that we had inadvertently included the affinity of GST-tagged pUL71(300–325) for VPS4A in the discussion text, not the data for the pUL71(300–325) peptide. We have updated the text accordingly and confirm that all the data in Table 2 are correct. *

    Minor comments:

    In general, the text and figures are very clear and accurate, the Results section is careful to walk the reader though these studies in a clear and well written fashion and prior studies are referenced appropriately. There are some minor issues that are listed below.

    i). For clarity, please direct the reader to panel 1B when referring to the pp28 data (line 11 of Results section).

    Done

    ii). At the bottom of the page 4, the Results section states "immunoprecipitation experiments show VPS4A-FLAG to be robustly co-precipitated by wild-type pUL71 but not by the PPAA and V317D mutants". However, from Fig. 1E it appears to be the reverse. The wild type pUL71 (but not mutants) is being co-precipitated by VPS4A-FLAG, using an anti-FLAG antibody.

    Corrected – we apologise for this error and thank the reviewer for spotting it.

    iii). In Fig.1D the localization of WT pUL71 and the PPAA and V317D mutants to a juxtanuclear compartment provides a nice internal control demonstrating that the mutant proteins are at least partially functional (able to localize correctly), and the fluorescence intensities of the WT and mutant pUL71 proteins appear comparable. However, do the authors have any additional quantitative or semi-quantitative data (such as from a Western) to confirm similar expression levels for the pUL71 WT and PPAA/V317D mutant proteins?

    The relevant data is shown in Fig. 1E. Specifically, the immunoblot of the input samples shows that pUL71 mutants are expressed at similar levels to the wild-type protein. We have added a note to this effect to the Results.

    “Inspection of the immunoprecipitation input samples confirms that pUL71 mutants are expressed at similar levels to the wild-type protein.”

    *Furthermore, we added to methods following statement: “equal volumes of lysate were used for all samples”, to confirm that the signals in Co-IPs stem from equal amounts of lysates. *

    iv). In Fig. 4, An OPTIONAL experiment, which would add to the paper, would be to test the ability (or rather, lack of the ability) of the pUL71 I307R mutant to coip VPS4A from infected or transfected cells. Such a study would extend the predictive power of the elegant MD simulations and ITC studies to the "gold standard" of testing the phenotype in vivo.

    While we appreciate that this additional experiment would provide further confirmation of our computational analysis in the cellular context, we would argue that ITC is the ‘gold standard’ when it comes to the measurement of protein interaction affinities. We show in Figure 1 that ITC, coIP and immunofluorescence experiments yield the same result (compare WT and PPAA pUL71 in panels D, E and G). We have thus respectfully declined to perform this additional IP experiment as we feel that the ITC data included in the manuscript, combined with the coIP data for the P315A and P318A mutants, are sufficient to prove the predictive power of the model.

    v). The Fig. 6B TB71stop pp28 panel is not referred to in the Fig. 6 legend.

    We apologise for this oversight. We have added a description of this experiment to the Fig. 6 legend:

    “Cells infected with TBstop71 were also stained for tegument protein pp28 to confirm successful cVAC formation (bottom).”

    We have also added the relevance of this image in the Results:

    “Formation of perinuclear cVAC in cells infected with TBstop71 was confirmed via immunostaining for pp28 (Fig. 6C) (Sanchez et al., 2000b; Seo and Britt, 2007).”

    vi). In the second paragraph of the Discussion it is stated that "The pUL71 vMIM2 is necessary and sufficient to recruit VPS4A to specific membranes in co-transfected cells (Fig. 5) and to sites of virus assembly in HCMV infection (Fig. 6)". Strictly speaking, Fig. 5 (panel 5F) shows that the HCMV pUL71 region 283-361 is sufficient to localize VPS4A to a compact juxtanuclear structure in transfected cells, and Fig. 6 (panel 6C) shows that pUL71 residues 315-326 (and the two conserved prolines in this region) are necessary for VPS4A localization to a structure that appears to be the HCMV assembly compartment.

    We thank the reviewer for highlighting that we been imprecise when describing the implications of our results. We have updated the relevant sentence as follows:

    “The pUL71 vMIM2 is necessary and sufficient to recruit VPS4A to juxtanuclear structures in co-transfected cells (Fig. 5) and is necessary for VPS4A recruitment to pUL71-positive structures that have been identified as sites of virus assembly during HCMV infection (Fig. 6) (Dietz et al., 2018).”

    Reviewer #1 (Significance (Required)):

    This is the first report of a virus encoding a MIM-like domain, and of a viral motif that directly binds the VPS4A MIT domain. This will be of broad interest to those studying the cell biology of virus assembly and mechanisms of virus-host cell interaction, as well as to cell biologists and structural biologists studying the ESCRT apparatus. It is striking, and will be illuminating to virologists and ESCRT biologists, that viruses have evolved to mimic MIM2 with a motif that has a lower Kd than a conventional cellular MIM2 motif. The possibility, addressed in the Discussion, that pUL71 may be sequestering VPS4A (rather than using it) is an important issue that virologists should consider.

    This is a rigorous, thorough and well controlled basic science study that elegantly combines a variety of approaches to provide important new insights concerning the biology of pUL71 in HCMV, other human beta-herpesviruses and a large number of mammalian and rodent cytomegaloviruses. The claims and conclusions are thoughtful and measured, and supported by the data. Data and methods are presented in such a way that they can be reproduced, and experiments are adequately replicated with appropriate statistical analyses.

    Reviewers fields of interest: Cell biology, ESCRT function, Virus assembly

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    Summary

    The manuscript from Butt et al. entitled "Human cytomegalovirus deploys molecular mimicry to recruit VPS4A to sites of virus assembly" addresses the potential role of ESCRT components in the biogenesis of HCMV virions. The topic is relevant since the ESCRT machinery has been implicated in the propagation of various herpesviruses. However, conflicting results are present in the literature regarding its role and relevance. In the present study, the authors focus on HCMV pUL71, which plays a role during the final envelopment of the viral capsids, and explore the possibility that it acts as an ESCRT-III component by recruiting the VPS4A ATPase (which induces membrane deformation and scission). To this end, they identified a motif in the C-terminal region of HCMV pUL71 that resembles the cellular type 2 MIM (MIM2) consensus sequence that is present in ESCRT-III proteins such as CHMP4B and CHMP6. They show substantial data that delineate an interaction between that pUL71 motif and the cellular ATPase using a panoply of tools (co-IF, co-IP, ITC, bimolecular fluorescence and markerless BAC mutagenesis). They also show that this interaction in present across a wide range of HCMV strains, but is absent in the case of the HSV-1 pUL51 homolog.

    Main Comments

    The manuscript is very well written, albeit line numbers would facilitate reviewing. The plethora of assays used convincingly show an interaction between pUL71 and VPS4A. They also indicate that this interaction relocalizes VPS4A to the TGN and likely the VAC. However, I do have some issues. For instance, using another viral marker, instead of only pUL71, would have been a good idea to distinguish the TGN from the VAC. This is not trivial given the reorganization of the cellular organelles by the virus. For this reason, looking at tegument and envelope viral proteins may not be optimal for this task since potentially on both compartments. However, viral capsid proteins or the viral genome may be useful here. Immuno-EM against VPS4A could also be a useful experiment to show a potential link between the ATPase and re-envelopment.

    It is well established that pUL71 is present at the cVAC of HCMV-infected cells. We apologise that we did not make this clearer in the introduction. We have now updated a sentence in the penultimate paragraph of the introduction to clarify this:

    “pUL71 and pUL103 are present at the cVAC (Ahlqvist and Mocarski, 2011; Dietz et al., 2018; Read et al., 2019; Womack and Shenk, 2010) and deficiencies in these proteins result in an accumulation of nucleocapsids at various advanced stages of envelopment (Ahlqvist and Mocarski, 2011; Schauflinger et al., 2013, 2011; Womack and Shenk, 2010), which is consistent with impaired envelopment and a block in membrane scission at the end of the envelopment process.”

    *Additionally, we have added confocal images of infected cells showing co-staining of pUL71, capsid associated tegument protein pp150, and Golgi maker GM130 in a new figure (Fig. 6A). *

    We are unaware of any commercial antibodies that recognise VPS4A and are suitable for immuno-EM, making such analysis unfeasible. However, we note that our study concurs with the data from Streck et al (2018) that VPS4 activity is not required for virus envelopment (although we do not rule out a contributory role).

    Another issue is the actual pUL71 residues interacting with VPS4A. While substantial efforts were made to map them (truncated constructs, bimolecular assay, viral mutants), the data do not always point toward the exact same residues (for example aa 314-320 by co-IF but aa 300-310 by ITC). This suggests potentially multiple binding sites or conformational issues. Hence, the statement on page 5 "that pUL71 residues 300-310 are necessary for the VPS4A interaction, in addition to the potential MIM2" may be misleading. What happens if one deleted aa 314-320 in the ITC assay? Or aa 300-310 by IF? These findings are further confounded by the lack of impact of the mutations of aa 315 and 318, predicted to be important in silico (p. 6). Moreover, in figure 7, the mutants made were a deletion 315-326 or the double point mutant P315A and P318A (not clear why in light of above results). Would a deletion of aa 300-320 not be a more appropriate and safer one to test for viral propagation?

    *We are afraid that the reviewer may have misinterpreted several of our results. In Fig. 2 we demonstrate that residues 1–320 are sufficient for co-localisation of pUL71 with VPS4A-FLAG, but residues 1–314 are not. This implies that residues 314–320 are necessary for the interaction, but it is not evidence that they are sufficient. Similarly, our ITC data shows that a purified peptide spanning residues 300–325 is *sufficient *for the interaction, but a peptide spanning residues 310–325 is not. From this we can clearly infer that residues 300–310 are *necessary for the interaction, as we state on page 5. We have expanded the sentence in question to further clarify our reasoning:

    “Further ITC analysis of pUL71 truncations purified as GST fusions (Fig. S1 and Table 1) demonstrated that pUL71 residues 300–310 are necessary for the VPS4A interaction, in addition to the potential MIM2, as GST-pUL71(300–325) was capable of binding VPS4A while GST-pUL71(310–336) was not.”

    We agree that residues 300–320 might be sufficient for the interaction, as indicated by the immunofluorescence analysis (Fig. 2A). However, out of an abundance of caution we included residues 300–325, spanning the entire MIM2-like sequence, in all of our biophysical analyses as the dynamic range of immunofluorescence experiments is limited and we wanted to avoid removal of residues that are not necessary but nonetheless contribute to the interaction. The similar affinity of GST-pUL71(283–361) and GST-pUL71(300–325) for VPS4A (2.8 and 2.3 µM, respectively) confirms that residues 300–325 contain all residues that contribute to the interaction (Figs 1 and S1, Table 1).

    With regards the mutational data presented in Fig. 4, our molecular dynamics analysis indicates (Fig. 4A–C) that single point mutations P315A and P318A do not disrupt the interaction between pUL71 and VPS4, only the double mutation (P315A+P318A; PPAA) disrupts the interaction. This is consistent with the immunoprecipitation presented in Fig. 4E: The pUL71(P315A) and pUL71(P318A) proteins are efficiently immunoprecipitated by VPS4A-FLAG, while the pUL71(PPAA) mutant is not. We have updated the penultimate sentence of the section “Model of the HCMV pUL71 in complex with VPS4A MIT” to explain this in more detail:

    “Immunoprecipitation of co-transfected pUL71 and VPS4A-FLAG confirmed this surprising result, showing that pUL71(P315A) and pUL71(P318A) are efficiently immunoprecipitated by VPS4A-FLAG whereas pUL71(PPAA) is not (Fig. 4E).”

    Regarding the choice of mutant viruses, we wanted to make the smallest change possible to pUL71 to avoid inadvertent removal of additional (potentially unknown) functional motifs. Both of the viruses we have used show an absence of VPS4A recruitment to the pUL71-positive cVAC in immunofluorescence (Fig. 6D) and, in the case of pUL71(PPAA), we have also shown an absence of VPS4A binding to this mutant in ITC (Fig. 1G) and coIP (Figs 1E and 4E). We feel this is sufficient evidence to confirm that these mutations either severely impair or completely abolish recruitment of VPS4A.

    Given the above, we don’t believe there is any need for additional experimentation or consideration of confounding variables when it comes to the definition of the vMIM2 motif or mutations introduced into HCMV for functional analysis.

    As the identification of the VPS4A binding motif in other herpesviruses appears to only be detected by manual inspection of the protein sequences, I wonder if other HCMV proteins or alpha/gamma viral proteins may interact with VPS4A. A good way to address this would be to do a VPS4A affinity column to see if any other viral proteins can bind. MS analyses may be required to identify the bound viral proteins. This could be a good follow-up paper...

    We thank the reviewer for suggestion and agree that it would form the basis for a good follow-up study.

    I am unfortunately unable to evaluate the outcome of the in silico analyses and cannot therefore judge their relevance or accuracy. Other reviewers can hopefully access this portion of the manuscript.

    Unless mistaken, previous work (Albecka A et al, 2017, JVI) has shown that HSV-1 pUL51 does not require its binding partner pUL7 to reach the TGN. Given that HSV-1 pUL51 does not seem to recruit VPS4A, could the pUL7/pUL51 complex be required for the recruitment of VPS4A to the TGN or VAC? Alternatively, could the lack of pUL51 binding to VPS4A reflect a different re-envelopment mechanism (absence of the CMV onion ring VAC)? These possibilities should be addressed in the manuscript.

    The reviewer is correct that, like pUL71, the HSV-1 protein pUL51 associates with TGN membranes as both proteins are N-terminally palmitoylated. Inspection of HSV-1 pUL7 does identify a potential vMIM2 sequence, spanning residues 221–232 (sequence LANnPpPVlsaL). However, these residues lie in a well-structured region at the interface with pUL51 (helices α8 and α9; see Fig. 2B of Butt et al (2020)) and would thus be unavailable to bind VPS4A. If the pUL7:pUL51 complex were required for VPS4A recruitment to sites of HSV-1 assembly, which has not been shown, then a different mechanism would be required. To test if this was the case, we performed a transfection experiment where pUL51-mCherry, or mTurquoise2-pUL7 +pUL51-mCherry, were co-transfected with GFP-VPS4A into U2-OS cells. As a positive control, we co-transfected pUL71-mCherry and GFP-VPS4A. As shown below, we observe recruitment of GFP-VPS4A to pUL71-mCherry positive membranes but do not see recruitment of GFP-VPS4A to pUL51-mCherry positive TGN membranes in the presence or absence of mTurquoise2-pUL7.

    This experiment has been performed twice with identical results. However, we have declined to include the above figure in the manuscript because our study focusses on the vMIM2 motif and the betaherpesviruses in which it is conserved. We already show that the pUL71 homologue in HSV (pUL51) does not recruit VPS4A to membranes (Fig. 5E). We believe that additional negative data on the lack of VPS4A recruitment by this HSV-1 complex complicates the story and would distract the reader. Identifying and characterising mechanisms via which other herpesvirus subfamilies may (or may not) specifically recruit VPS4A to sites of virus assembly is interesting, but it lies outside the scope of this current manuscript.

    We agree with the reviewer that the HCMV secondary envelopment pathway likely differs from that of HSV. For example, HSV envelopment is severely restricted by dominant negative VPS4 whereas HCMV is not. This indicates that, at a minimum, HCMV must have additional/redundant mechanisms that drive envelopment in the absence of a functioning ESCRT machinery. We have added a comment to this effect to the second paragraph of the discussion:

    “This lack of requirement for ESCRT activity during HCMV secondary envelopment contrasts with the situation for HSV-1, where expression of dominant-negative VPS4 (Calistri et al., 2007; Crump et al., 2007) or CHMP proteins (Calistri et al., 2007; Pawliczek and Crump, 2009) severely restricts virion production. We therefore conclude that either HCMV and HSV-1 utilise different molecular mechanisms for secondary envelopment, or HCMV can exploit additional (redundant) pathways in addition to ESCRT-mediated membrane remodelling to ensure assembly of mature virus particles.”

    Minor

    Fig 3S: I would suggest highlighting the central P residue in the aligned sequence and consensus sequence.

    We thank the reviewer for this helpful suggestion. We have highlighted both the central ‘P’ plus the other conserved hydrophobic residues of the vMIM2 in the aligned and consensus sequence in Figures 5, S3 and S4.

    Reviewer #2 (Significance (Required)):

    Not surprisingly, the biggest issue in the manuscript is that perturbing pUL71 / VPS4A binding has no detectable positive or negative impact on VAC assembly, secondary viral envelopment or viral spread (titre, plaque size). This raises the question as to the relevance of VPS4A for the virus. As mentioned above, it could be relevant to test a viral mutant lacking pUL71 aa 300-320, which may lead to different results.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    Summary

    In this study the authors investigate mechanisms by which human cytomegalovirus (HCMV) modulates ESCRT III to facilitate virus maturation. The viral protein pUL71 has been shown previously to be an important viral mediator of the so called "secondary envelopment", that is the process in which the viral capsid is budding into Golgi-derived membranes to acquire its envelope.

    pUL71 was previously shown to recruit VPS4 to the trans-Golgi upon co-transfection. In this study, the authors investigate the structural requirements for this interaction. Sequence comparison prompted the investigation of a short motif in the C-terminus of pUL71 with homology to the Type 2 MIT-Interacting Motif (MIM2) of CHMP6 that is known to bind the MIT of VPS4. Using co-localization, co-immunoprecipitation and isothermal titration calorimetry they show clearly that a peptide spanning amino acids 300-325 of pUL71 is required and sufficient for binding of VPS4A. State of the art modeling of the protein complex identifies the crucial amino acids that define the interaction on both sides. The authors validate these predictions experimentally in transfection and with purified peptides as well as in the context of HCMV infection using bimolecular fluorescence complementation. Furthermore, the authors demonstrate that not only the other human betaherpesviruses but also closely related CMVs of rat and mouse encode viral MIM2-like motifs (vMIM2)that are able to interact with VPS4A. Unexpectedly, albeit in line with previous reports, they find that mutation of this highly conserved vMIM2 domain did not alter viral progeny and focus size largely. The authors further confirm by quantification of high quality electron micrographs, that a large portion of capsids is able to complete the process of budding into and scission from cellular membranes, demonstrating that the ability of pUL71 to bind VPS4A is dispensable for secondary envelopment.

    Taken together, this study demonstrates clearly that a newly defined vMIM2 in the HCVM pUL71 protein binds cellular VPS4A. Yet, it remains unclear in which context the virus requires this novel form of molecular mimicry.

    While we thank the reviewer for summarising our study, highlighting the careful attention they paid to our work, we would like to emphasise that __we are unaware of any previous study showing that pUL71 recruits VPS4 to the trans-Golgi upon co-transfection. __It had previously been observed that VPS4 is present at the cytoplasmic virus assembly compartment (cVAC) during infection – as we state in paragraph 4 of the introduction (first sentence). However, none of these studies identified the virus protein responsible for this recruitment nor did they identify the viral motif that mediated this recruitment. Our identification of pUL71 as the HCMV protein that recruits VPS4A to the cVAC is novel.

    Major comments:

    1. Despite the very thorough analysis and conduction of the study, the presented work does not reveal a phenotype in virus infection. The authors would need to find out what the functional relevance of their discovery is.

    As described by reviewer 1, our data confirms with and extends previous studies to show that ESCRT activity seems not to be essential for HCMV secondary envelopment. This is different from other herpesviruses such as HSV-1, showing that the secondary envelopment process is not universally conserved across the herpesviruses (or that additional redundant processes are encoded by HCMV). We also identify a novel virus-encoded VPS4 recruitment motif, the vMIM2. The fact that this sequence is conserved across beta-herpesvirus pUL71 homologues strongly suggests that it has a conserved and important function in the virus lifecycle, even if we haven’t identified that function in this study. In the discussion we posit multiple hypotheses for how this motif may function during infection, including sequestration of VPS4. As reviewer 1 states, VPS4 sequestration “is an important issue that virologists should consider”. We feel that additional experiments to tease out the precise function conferred by this motif represent important future work but are beyond the scope of this current study.

    Please include statistical analysis of virus release, spread and envelopment (Figure 7 and Table 2). It would be helpful to see if the small differences observed are likely to be random or not.

    We thank the reviewer for this suggestion. We have performed relevant statistical tests for the virus growth data and virus spread (plaque size) assays.

    *A repeated measures two-way ANOVA test of the high MOI (single step) virus growth data shows that there is no significant difference between the viruses tested (P = 0.5824). A repeated measures two-way ANOVA test of the low MOI (multi-step) virus growth shows that there is a difference between viruses. A Dunnett’s multiple comparison test shows that there is no significant difference between the TB71revPPAA and wild-type virus at any time point. There are significant differences between the TB71mutPPAA virus and wild-type at 15 dpi (P “While two-way ANOVA analysis showed significant differences between wild-type and mutant virus yields at late time points in the multi-step growth curve, the TB71mutPPAA mutant had higher titres at 15 dpi whereas TB71del315-326 had lower titres at 15 and 18 dpi. Given the divergence in observed effect between the two mutants, and the fact that these differences were observed only at very late times post-infection, we do not believe they represent biologically meaningful differences in virus release.”

    A Mann Whitney test of the focus expansion assay data, performed instead of a t test because a D'Agostino & Pearson test showed the WT plaque data to not be normally distributed, showed no significant difference between the WT and TB71 mutPPAA virus (P = 0.489), which would agree with our notion that there is no in change in virus growth when interaction with VPS4A is disrupted.

    *Concerning quantitative evaluation of secondary envelopment, we have respectfully declined to include statistical analysis in the manuscript. This is because quantitative analysis of virus envelopment via electron microscopy has multiple caveats that complicate robust statistical analysis. The numbers of virus particles in the area of the cVAC in the individual cells is subject to stark variation, only a small part of the cell or the cVAC is analysed, and naked capsids are only rarely observed. Defects in HCMV secondary envelopment, as has been published for pUL71 knock-out viruses, manifest as strong shifts in the proportions of the envelopment stages; see for example Schauflinger et al. (2013). We did perform a two-way ANOVA with Dunnett’s multiple comparison test, which shows that there are significantly fewer enveloped particles (P In summary, none of our data consistently and robustly show involvement of VPS4A for HCMV assembly that could explain the conservation of this interaction among betaherpesviruses, which is consistent with previous publications indicating that HCMV secondary envelopment does not require the cellular ESCRT machinery.

    One caveat is that the presented study investigates the impact of VPS4A on HCMV only in fibroblasts. However, other studies used epithelial cells to investigate the impact of VPS4 knockout on HCMV and also did not see a reduction in virus titers. Yet, the authors could significantly improve the manuscript by testing for a cell type specific requirement of the vMIM2. The replication of the PPAA mutant virus could be analyzed in additional cell types such as macrophages or endothelial cells and using different experimental systems.

    *We thank the reviewer for this comment. We have evaluated viral growth of the PPAA mutant virus in monocyte-derived macrophages, similar to our analysis of a pp65 stop mutant (Chevillotte et al., JVirol 2009). We specifically tested macrophages as this cell type appears to restrict HCMV growth when compared to released virus yield from fibroblasts and endothelial cells. However, viral growth of the PPAA mutant is similar to that of parental and revertant virus, verifying our growth analysis in fibroblasts. Furthermore, we investigated virion morphogenesis of the PPAA mutant in macrophages by electron microscopy because pUL71 plays an important role in HCMV secondary envelopment. Consistent with our growth analysis, we could not find evidence for a role of VPS4 recruitment by pUL71 for virion morphogenesis. We have added this additional data as a new supplementary figure (Fig. S6). *

    Consider discussing if other viral and cellular proteins could compensate the loss of interaction between pUL71 and VPS4. Is a similar motif found in any other HCMV protein? Could redundancy explain the lack of a consequences for viral growth?

    The short answer is no, it is unlikely that any other HCMV protein could compensate for the loss of pUL71 binding and efficiently recruit VPS4A to the cVAC because we see a complete loss of VPS4A-FLAG recruitment to the cVAC when pUL71 is either absent (Fig. 6C) or has a defective vMIM2 (Fig. 6D). However, it is possible that additional HCMV proteins could interact with VPS4A, for example to enhance its retention at the cVAC by increasing the avidity of binding.

    We used the ScanProsite web server to identify additional proteins encoded by HCMV with the vMIM2 sequence [YLM]-{P}-{P}-x-P-x-[AVP]-[VP]-x-x-x-[LVP]. This sequence corresponds to the residues observed at each position in the vMIM2s of betaherpesvirus pUL71 homologues presented in Figures 5, S3 and S4, where proline is disallowed at the second and third position because of our identification that the first residues of the vMIM2 form an α-helix (proline residues being incompatible with α-helix formation).

    *We identified eight additional HCMV proteins with potential vMIM2 sequences: pUL31, pUL57, pUL72, pp28 (a.k.a. pUL99), pUL141, pUS22, pUS29 and pUS30. Of these, pUL141 could be immediately discounted because the vMIM2 sequence is located in an extracellular portion of the protein and thus would be incapable of binding the cytosolic VPS4A MIT domain. pUL31, pUL57 and pUS22 could similarly be discounted because inspection of AlphaFold2 models of these proteins (https://www.bosse-lab.org/herpesfolds/) reveal the potential vMIM2 sequences to lie within regions of the protein that are predicted to be well-ordered and are buried and/or form secondary structures incompatible with binding the VPS4A MIT domain. The vMIM2 motifs of the remaining four proteins were in regions of the proteins that lacked tertiary structure and were predicted with low confidence, indicating that these regions are likely to have little intrinsic structure in the absence of a binding partner. Additionally, we observed that the potential vMIM2 sequences of pUL72 and pUS29 were predicted to have a helix-then-extended conformation, like pUL71. *

    To probe whether the pUL72, pp28, pUS29 and pUS30 sequences that matched the vMIM2 consensus were likely to bind VPS4A, we used AlphaFold2 to predict structures of the relevant 26 amino acid regions from these proteins in complex with the VPS4A MIT domain. Analysis of the pLDDT scores show that the interaction between VPS4A and pUL72 is plausible, although this interaction is predicted with less confidence than the VPS4A:pUL71(300–325) interaction. The other models are predicted with very low confidence, suggesting that these regions are unlikely to interact. This agrees with our data for pp28, where we demonstrated using transient expression experiments that pUL71 but not pp28 could sequester VPS4A (Fig. 1B).

    Further inspection of the VPS4A:pUL72(potential vMIM2) prediction showed that several residues in the potential interacting region are predicted to contribute to the pUL72 folded domain, forming the final strand of a β-sheet. AlphaFold-Multimer prediction of a complex between the VPS4A MIT domain and full-length pUL72 failed to yield models where the potential vMIM2 interacted with VPS4A, suggesting that steric clashes between VPS4A and the globular domain of pUL72 would prevent pUL72 from binding VPS4A in cells.

    While it is theoretically possible that the potential vMIM2 motifs identified above could interact with VPS4A, the interaction is clearly not sufficient to effectively recruit VPS4A to the cVAC in the absence of pUL71 or in the presence of a pUL71 mutant with a defective vMIM2 (Fig. 6C,D). There are also several HCMV proteins that have ‘late domains’ and could in theory compensate for the absence of pUL71 via recruitment of ‘upstream’ ESCRT machinery components (Streck 2020), but these are similarly incapable of efficiently recruiting VPS4A in the absence of pUL71. It is possible that a small amount of residual VPS4 recruitment via late domain containing proteins could functionally compensate for the absence of the vMIM2 but, given the published evidence that VPS4 activity is dispensable for virus envelopment, it is more likely in our opinion that an alternative non-ESCRT mechanism drives HCMV envelopment.

    We have added a paragraph at the end of the results section “VPS4A binding is conserved amongst cytomegaloviruses and human β-herpesviruses” and a new supplemental figure (Fig. S5) describing the other potential vMIM2 sequences in HCMV. We have also added a section to the discussion where we describe our interpretation of these results, outlining our reasons for concluding that other HCMV sequences that match the vMIM2 consensus are very unlikely to play a role in envelopment (although we admit that we cannot entirely discount this hypothesis):

    “While other HCMV proteins have sequences that match the vMIM2 consensus, none are able to recruit VPS4A to the cVAC when pUL71 is absent (Fig. 6C) or has a defective vMIM2 (Fig. 6D). It is therefore unlikely that these sequences are functionally redundant to the pUL71 vMIM2, although we cannot formally discount this hypothesis.”

    We thank the reviewer for asking this interesting question.

    Would the small differences (if significant) in virus titer be sufficient to provide enough of an evolutionary advantage to explain the sequence conservation? It would be interesting to try an in vitro selection assay and test if wildtype would outcompete the PPAA mutant after some passages.

    This is an interesting suggestion, but it is not really supported by our data, as all our data indicate that sequestration of VPS4 by pUL71 has no growth advantage (see also answer to point 3). This is further supported by our results from electron microscopy. Even the minimal differences in growth are not significant at most time points. In addition, to our knowledge, there is no established assay for HCMV for the proposed analysis and therefore no reliable data regarding the significance.

    Albeit far beyond the original scope of the study:

    In the very thoughtful discussion, the option is discussed that other MIT domain containing proteins could be the actual targets of the pUL71 MIM2-domain. It would be interesting to use the generated expression constructs to identify other cellular targets by co-IP and mass spectrometry.

    We agree this would be an interesting avenue of future work and it is one we intend to pursue in the future. However, identification of novel binding partners for the vMIM2 and biochemical plus functional characterisation of these interactions is a large study and is thus outside the scope of this current manuscript.

    Minor comments: None, the presented experiments are well conducted and presented. The work is adequately discussed.

    Reviewer #3 (Significance (Required)):

    Significance section

    General assessment:

    The performed experiments are well described and the high quality is revealed by the abundant primary data shown. Multiple independent methods were used to investigate the central findings. The claims made are therefore well supported. Especially the data supporting the direct interaction between the MIM2-like domain and the MIT of VPS4 are excellent and unequivocally demonstrate a direct interaction. On the other hand, the lack of effect of this interaction in the context of viral infection questions the significance of the finding. Possibly, by testing additional cell lines to assess virus spread, the authors could increase relevance of the findings.

    Advance:

    The impact of VPS4 and the ESCRT machinery on HCMV secondary envelopment has been a matter of debate since a study by Tandon et al. in 2009 seemed to contradict the first publication on the topic by Fraile-Ramos et al in 2007. The current study by Butt et al. now supports a more recent report by Streck et al., which suggested that VPS4 is not required for secondary envelopment. The fact that the two studies use different experimental systems with similar outcome, suggests that virus maturation is indeed independent of VPS4. However, Streck et al. observe an effect of dominant negative ESCRT mutants on virus spread, suggesting that the interaction of the HCMV tegument with ESCRT is required only under special conditions, which still remain to be defined. Albeit the present study cannot fill all the gaps of our understanding of this topic, the high quality of the data is a good basis for further investigations.

    In addition, the description of a viral MIM2-like motifs might spur the investigation of similar motifs in other viruses, potentially bringing more cases of molecular mimicry to light.

    Audience:

    This study is of interest to basic researchers investigating aspects of modulation of cellular membranes by viruses or interested the cellular components and interactors of the ESCRT complexes.

  2. 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

    Summary

    In this study the authors investigate mechanisms by which human cytomegalovirus (HCMV) modulates ESCRT III to facilitate virus maturation. The viral protein pUL71 has been shown previously to be an important viral mediator of the so called "secondary envelopment", that is the process in which the viral capsid is budding into Golgi-derived membranes to acquire its envelope.

    pUL71 was previously shown to recruit VPS4 to the trans-Golgi upon co-transfection. In this study, the authors investigate the structural requirements for this interaction. Sequence comparison prompted the investigation of a short motif in the C-terminus of pUL71 with homology to the Type 2 MIT-Interacting Motif (MIM2) of CHMP6 that is known to bind the MIT of VPS4. Using co-localization, co-immunoprecipitation and isothermal titration calorimetry they show clearly that a peptide spanning amino acids 300-325 of pUL71 is required and sufficient for binding of VPS4A. State of the art modeling of the protein complex identifies the crucial amino acids that define the interaction on both sides. The authors validate these predictions experimentally in transfection and with purified peptides as well as in the context of HCMV infection using bimolecular fluorescence complementation. Furthermore, the authors demonstrate that not only the other human betaherpesviruses but also closely related CMVs of rat and mouse encode viral MIM2-like motifs (vMIM2)that are able to interact with VPS4A. Unexpectedly, albeit in line with previous reports, they find that mutation of this highly conserved vMIM2 domain did not alter viral progeny and focus size largely. The authors further confirm by quantification of high quality electron micrographs, that a large portion of capsids is able to complete the process of budding into and scission from cellular membranes, demonstrating that the ability of pUL71 to bind VPS4A is dispensable for secondary envelopment.
    Taken together, this study demonstrates clearly that a newly defined vMIM2 in the HCVM pUL71 protein binds cellular VPS4A. Yet, it remains unclear in which context the virus requires this novel form of molecular mimicry.

    Major comments:

    1. Despite the very thorough analysis and conduction of the study, the presented work does not reveal a phenotype in virus infection. The authors would need to find out what the functional relevance of their discovery is.
    2. Please include statistical analysis of virus release, spread and envelopment (Figure 7 and Table 2). It would be helpful to see if the small differences observed are likely to be random or not.
    3. One caveat is that the presented study investigates the impact of VPS4A on HCMV only in fibroblasts. However, other studies used epithelial cells to investigate the impact of VPS4 knockout on HCMV and also did not see a reduction in virus titers. Yet, the authors could significantly improve the manuscript by testing for a cell type specific requirement of the vMIM2. The replication of the PPAA mutant virus could be analyzed in additional cell types such as macrophages or endothelial cells and using different experimental systems.
    4. Consider discussing if other viral and cellular proteins could compensate the loss of interaction between pUL71 and VPS4. Is a similar motif found in any other HCMV protein? Could redundancy explain the lack of a consequences for viral growth?
    5. Would the small differences (if significant) in virus titer be sufficient to provide enough of an evolutionary advantage to explain the sequence conservation? It would be interesting to try an in vitro selection assay and test if wildtype would outcompete the PPAA mutant after some passages.

    Albeit far beyond the original scope of the study:

    1. In the very thoughtful discussion, the option is discussed that other MIT domain containing proteins could be the actual targets of the pUL71 MIM2-domain. It would be interesting to use the generated expression constructs to identify other cellular targets by co-IP and mass spectrometry.

    Minor comments: None, the presented experiments are well conducted and presented. The work is adequately discussed.

    Significance

    General assessment:

    The performed experiments are well described and the high quality is revealed by the abundant primary data shown. Multiple independent methods were used to investigate the central findings. The claims made are therefore well supported. Especially the data supporting the direct interaction between the MIM2-like domain and the MIT of VPS4 are excellent and unequivocally demonstrate a direct interaction. On the other hand, the lack of effect of this interaction in the context of viral infection questions the significance of the finding. Possibly, by testing additional cell lines to assess virus spread, the authors could increase relevance of the findings.

    Advance:

    The impact of VPS4 and the ESCRT machinery on HCMV secondary envelopment has been a matter of debate since a study by Tandon et al. in 2009 seemed to contradict the first publication on the topic by Fraile-Ramos et al in 2007. The current study by Butt et al. now supports a more recent report by Streck et al., which suggested that VPS4 is not required for secondary envelopment. The fact that the two studies use different experimental systems with similar outcome, suggests that virus maturation is indeed independent of VPS4. However, Streck et al. observe an effect of dominant negative ESCRT mutants on virus spread, suggesting that the interaction of the HCMV tegument with ESCRT is required only under special conditions, which still remain to be defined. Albeit the present study cannot fill all the gaps of our understanding of this topic, the high quality of the data is a good basis for further investigations.
    In addition, the description of a viral MIM2-like motifs might spur the investigation of similar motifs in other viruses, potentially bringing more cases of molecular mimicry to light.

    Audience:

    This study is of interest to basic researchers investigating aspects of modulation of cellular membranes by viruses or interested the cellular components and interactors of the ESCRT complexes.

  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 #2

    Evidence, reproducibility and clarity

    Summary

    The manuscript from Butt et al. entitled "Human cytomegalovirus deploys molecular mimicry to recruit VPS4A to sites of virus assembly" addresses the potential role of ESCRT components in the biogenesis of HCMV virions. The topic is relevant since the ESCRT machinery has been implicated in the propagation of various herpesviruses. However, conflicting results are present in the literature regarding its role and relevance. In the present study, the authors focus on HCMV pUL71, which plays a role during the final envelopment of the viral capsids, and explore the possibility that it acts as an ESCRT-III component by recruiting the VPS4A ATPase (which induces membrane deformation and scission). To this end, they identified a motif in the C-terminal region of HCMV pUL71 that resembles the cellular type 2 MIM (MIM2) consensus sequence that is present in ESCRT-III proteins such as CHMP4B and CHMP6. They show substantial data that delineate an interaction between that pUL71 motif and the cellular ATPase using a panoply of tools (co-IF, co-IP, ITC, bimolecular fluorescence and markerless BAC mutagenesis). They also show that this interaction in present across a wide range of HCMV strains, but is absent in the case of the HSV-1 pUL51 homolog.

    Main Comments

    The manuscript is very well written, albeit line numbers would facilitate reviewing. The plethora of assays used convincingly show an interaction between pUL71 and VPS4A. They also indicate that this interaction relocalizes VPS4A to the TGN and likely the VAC. However, I do have some issues. For instance, using another viral marker, instead of only pUL71, would have been a good idea to distinguish the TGN from the VAC. This is not trivial given the reorganization of the cellular organelles by the virus. For this reason, looking at tegument and envelope viral proteins may not be optimal for this task since potentially on both compartments. However, viral capsid proteins or the viral genome may be useful here. Immuno-EM against VPS4A could also be a useful experiment to show a potential link between the ATPase and re-envelopment.

    Another issue is the actual pUL71 residues interacting with VPS4A. While substantial efforts were made to map them (truncated constructs, bimolecular assay, viral mutants), the data do not always point toward the exact same residues (for example aa 314-320 by co-IF but aa 300-310 by ITC). This suggests potentially multiple binding sites or conformational issues. Hence, the statement on page 5 "that pUL71 residues 300-310 are necessary for the VPS4A interaction, in addition to the potential MIM2" may be misleading. What happens if one deleted aa 314-320 in the ITC assay? Or aa 300-310 by IF? These findings are further confounded by the lack of impact of the mutations of aa 315 and 318, predicted to be important in silico (p. 6). Moreover, in figure 7, the mutants made were a deletion 315-326 or the double point mutant P315A and P318A (not clear why in light of above results). Would a deletion of aa 300-320 not be a more appropriate and safer one to test for viral propagation?

    As the identification of the VPS4A binding motif in other herpesviruses appears to only be detected by manual inspection of the protein sequences, I wonder if other HCMV proteins or alpha/gamma viral proteins may interact with VPS4A. A good way to address this would be to do a VPS4A affinity column to see if any other viral proteins can bind. MS analyses may be required to identify the bound viral proteins. This could be a good follow-up paper...

    I am unfortunately unable to evaluate the outcome of the in silico analyses and cannot therefore judge their relevance or accuracy. Other reviewers can hopefully access this portion of the manuscript.

    Unless mistaken, previous work (Albecka A et al, 2017, JVI) has shown that HSV-1 pUL51 does not require its binding partner pUL7 to reach the TGN. Given that HSV-1 pUL51 does not seem to recruit VPS4A, could the pUL7/pUL51 complex be required for the recruitment of VPS4A to the TGN or VAC? Alternatively, could the lack of pUL51 binding to VPS4A reflect a different re-envelopment mechanism (absence of the CMV onion ring VAC)? These possibilities should be addressed in the manuscript.

    Minor

    Fig 3S: I would suggest highlighting the central P residue in the aligned sequence and consensus sequence.

    Significance

    Not surprisingly, the biggest issue in the manuscript is that perturbing pUL71 / VPS4A binding has no detectable positive or negative impact on VAC assembly, secondary viral envelopment or viral spread (titre, plaque size). This raises the question as to the relevance of VPS4A for the virus. As mentioned above, it could be relevant to test a viral mutant lacking pUL71 aa 300-320, which may lead to different results.

  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 #1

    Evidence, reproducibility and clarity

    Summary:

    In this study the authors apply a rigorous and thorough combination of approaches including sequence analysis, deep-learning structure predictions, molecular dynamics, cell imaging and mutagenic analyses to identify a short MIM2-mimicking motif in the C-terminal region of the pUL71 protein of HCMV (and homologues in other beta-herpesviruses) that is necessary and sufficient for interaction with the ESCRT terminal ATPase VPS4A. pUL71 uses this motif to recruit, or sequester, VPS4A to the HCMV cytoplasmic viral assembly complex, though this process is dispensable for HCMV morphogenesis or replication. The identified pUL71 sequence functions as a mimic of the MIM2 motif of cellular CHMP subunits since, like MIM2, it directly binds the groove in the MIT domain found at the N-terminus of VPS4.

    Major comments:

    1. There appears to be some confusion in the coip experiment in Figure 5D. From the upper blot in 5D, the "+" above each lane suggests there should be VPS4A-FLAG protein in every sample other than the two lanes at the very left of the gel, however the anti-FLAG ip does not pull down VPS4A-FLAG from every "+" lane, but from alternating ones (and from the next to the leftmost lane, which should lack VPS4A-FLAG). Similarly, the lower "Input" blot shows VPS4A-FLAG present in alternating lanes across the blot, which does not match the "+" and "-" labeling at the top of the figure. Conversely, there is anti-HA signal in most input lanes (lower blot) though the HA-tagged pUL71 homologues should be absent from alternate lanes (top of upper blot).
    2. The Discussion is an excellent, comprehensive and scholarly assessment of the implications of this work. One appealing hypothesis is that pUL71 may be sequestering VPS4A rather than using it for envelope scission. In this regard, the authors point out that VPS4A sequestration is supported by the finding that the VPS4A MIT domain binds the isolated pUL71 vMIM2 more tightly (~ 5 fold lower Kd) than the MIM2 of CHMP6, and that pUL71 and homologues) are highly abundant at later stages of viral infection, allowing them to compete effectively with endogenous CHMP6 for VPS4A. I like the sequestration model very much, but could the authors comment on the fact that this apparent sequestration is seen even in the transfection experiments in Fig. 2A and 3G, where essentially 100% of transfected WT VPS4A-FLAG is recruited to the pUL71 compartment. Even given the increased binding affinity to pUL71, this suggests that in these transfection studies pUL71 must be in excess over the sum of both endogenous and transfected VPS4. Do the authors know if this is the case, and do cells transfected with pUL71 in these experiments exhibit any cytotoxicity, or cell cycle arrest, indicative of a block in normal ESCRT function/cytokinesis?

    Minor comments:

    In general, the text and figures are very clear and accurate, the Results section is careful to walk the reader though these studies in a clear and well written fashion and prior studies are referenced appropriately. There are some minor issues that are listed below.

    i). For clarity, please direct the reader to panel 1B when referring to the pp28 data (line 11 of Results section).

    ii). At the bottom of the page 4, the Results section states "immunoprecipitation experiments show VPS4A-FLAG to be robustly co-precipitated by wild-type pUL71 but not by the PPAA and V317D mutants". However, from Fig. 1E it appears to be the reverse. The wild type pUL71 (but not mutants) is being co-precipitated by VPS4A-FLAG, using an anti-FLAG antibody.

    iii). In Fig.1D the localization of WT pUL71 and the PPAA and V317D mutants to a juxtanuclear compartment provides a nice internal control demonstrating that the mutant proteins are at least partially functional (able to localize correctly), and the fluorescence intensities of the WT and mutant pUL71 proteins appear comparable. However, do the authors have any additional quantitative or semi-quantitative data (such as from a Western) to confirm similar expression levels for the pUL71 WT and PPAA/V317D mutant proteins?

    iv). In Fig. 4, An OPTIONAL experiment, which would add to the paper, would be to test the ability (or rather, lack of the ability) of the pUL71 I307R mutant to coip VPS4A from infected or transfected cells. Such a study would extend the predictive power of the elegant MD simulations and ITC studies to the "gold standard" of testing the phenotype in vivo.

    v). The Fig. 6B TB71stop pp28 panel is not referred to in the Fig. 6 legend.

    vi). In the second paragraph of the Discussion it is stated that "The pUL71 vMIM2 is necessary and sufficient to recruit VPS4A to specific membranes in co-transfected cells (Fig. 5) and to sites of virus assembly in HCMV infection (Fig. 6)". Strictly speaking, Fig. 5 (panel 5F) shows that the HCMV pUL71 region 283-361 is sufficient to localize VPS4A to a compact juxtanuclear structure in transfected cells, and Fig. 6 (panel 6C) shows that pUL71 residues 315-326 (and the two conserved prolines in this region) are necessary for VPS4A localization to a structure that appears to be the HCMV assembly compartment.

    Significance

    This is the first report of a virus encoding a MIM-like domain, and of a viral motif that directly binds the VPS4A MIT domain. This will be of broad interest to those studying the cell biology of virus assembly and mechanisms of virus-host cell interaction, as well as to cell biologists and structural biologists studying the ESCRT apparatus. It is striking, and will be illuminating to virologists and ESCRT biologists, that viruses have evolved to mimic MIM2 with a motif that has a lower Kd than a conventional cellular MIM2 motif. The possibility, addressed in the Discussion, that pUL71 may be sequestering VPS4A (rather than using it) is an important issue that virologists should consider.

    This is a rigorous, thorough and well controlled basic science study that elegantly combines a variety of approaches to provide important new insights concerning the biology of pUL71 in HCMV, other human beta-herpesviruses and a large number of mammalian and rodent cytomegaloviruses. The claims and conclusions are thoughtful and measured, and supported by the data. Data and methods are presented in such a way that they can be reproduced, and experiments are adequately replicated with appropriate statistical analyses.

    Reviewers fields of interest: Cell biology, ESCRT function, Virus assembly

  5. Version published to 10.1101/2024.01.04.572781v1 on bioRxiv