Assembly of infectious Kaposi’s sarcoma-associated herpesvirus progeny requires formation of a pORF19 pentamer

Herpesviruses cause severe diseases particularly in immunocompromised patients. Both genome packaging and release from the capsid require a unique portal channel occupying one of the 12 capsid vertices. Here, we report the 2.6 Å crystal structure of the pentameric pORF19 of the γ-herpesvirus Kaposi’s sarcoma-associated herpesvirus (KSHV) resembling the portal cap that seals this portal channel. We also present the structure of its β-herpesviral ortholog, revealing a striking structural similarity to its α- and γ-herpesviral counterparts despite apparent differences in capsid association. We demonstrate pORF19 pentamer formation in solution and provide insights into how pentamerization is triggered in infected cells. Mutagenesis in its lateral interfaces blocked pORF19 pentamerization and severely affected KSHV capsid assembly and production of infectious progeny. Our results pave the way to better understand the role of pORF19 in capsid assembly and identify a potential novel drug target for the treatment of herpesvirus-induced diseases.


INTRODUCTION
Page 1, 8 lines from top: why 'pseudo-icosahedral capsid'? better to use 'icosahedral capsid' Due to the presence of the portal in the capsid and the fact that recent asymmetric herpesviral capsid reconstructions revealed asymmetric features within the KSHV capsid (Gong et al., 2019) we think that referring to a 'pseudoicosahedral capsid' is less misleading.
Page 2: the description of the penton components would benefit from a schematic picture for those who are not expert in the herpesvirus field, serving as road map of the different CVSC (or CAT), triplex proteins etc. on top of a schematic icosahedron. It can be a schematic figure as supplementary Fig. S1 (this would also help the interpretability of Figure 3 in the context of the virion structure).
We thank the reviewer for this idea -we have prepared such a schematic figure and appended this as new Fig. S1 to the revised manuscript. We apologize for the mislabeling of the old Fig. 3. We have now prepared a new Figure 3 that is colored according to the new Fig. S1 (see above), thereby supporting figure clarity. We have also revised the main text and the figure legend accordingly.
Page 11, 4 lines from top: the sentence 'negative-stain EM of pORF19KCTDCC demonstrated a 5-fold symmetry' is overstated. To demonstrate that the protein oligomerize forming a pentamer and thus displaying a 5-fold symmetry, a 2D classification would be more appropriated -and it is indeed recommended. Using negative-stain technique the Authors would not require a large set of images to prove their statement. Figure 4H, as it is now, is not convincing and it does not show the existence of a homogeneous population of pORF19KCTDCC that forms a stable pentamer.
We thank the reviewer for this suggestion, we have now performed an SPAanalysis on the negative stain images. 2D class averages of approximately 12.000 particles revealed top and side views of the pentameric pORF19KCTDCC and the top views clearly demonstrate the 5-fold symmetry of the molecule. These top and side views have been appended to the modified Figure 4 and an overview of the class averages is shown as new supplementary figure (Fig. S4).

DISCUSSION:
Page 19: 12 lines from top: the sentence 'The recombinant portal cap…..capsid disassembly'. It is not clear why the Authors mention this as it seems more of a plan that will possibly happen in the future. It does not add much to the discussion.
We agree with the reviewer that this sentence is not essential for the discussion and we have therefore removed it from the manuscript. We thank the reviewer for the feedback, which we have addressed above. We apologize for this technical issue -the scale bar is now visible in the top line of the fluorescent images. The images are stacked images and the individual panels were generated with ImageJ, and this information has now been appended to the figure legend.

METHODOLOGY:
Page 24: 'Structure analysis': The Author should state why the simulated map from the crystal structure was set to 5 Å resolution for the fitting into the portal density. The claimed resolution in the EMD-20430 is 7.6 Å for the asymmetric 3D reconstruction and it is likely that the resolution of the portal cap is slightly lower so why has the Author filtered the simulated map at 5 Å and not 7.6 or 8 Å?
As we have re-prepared figure 3 (see above), we also repeated the fitting procedure and for this purpose we fitted the monomeric and pentameric forms using a "SegFit" rotational search in UCSF Chimera into the asymmetric (C1) KSHV portal vertex reconstruction (EMD 20431). As the claimed resolution in EMD 20431 is 5.2 Å, but the resolution in the portal cap region of this map is likely to be considerably lower, we filtered the simulated map at 7.6 Å as suggested.
The authors didn't submit the PDB validation reports for their X-ray structures. In this case Table  S2 is pretty convincing but it would a good practice for authors describing X-ray structures to include in their submission the corresponding PDB validation report: https://www.wwpdb.org/validation/validation-reports All three validation reports are now added as accompanying items.
Reviewer #2: Summary Herpesviral capsids are complex icosahedral assemblies composed of many copies of several viral proteins. One of these, the outer capsid protein conserved among all known herpesvirusestermed UL25 in HSV-1 -reinforces the capsid to allow it to withstand the pressure of the packaged genome and has also been implicated in genome packaging, cleavage, and retention. Prior to this study, the structural information has only been available for the HSV-1 UL25 homolog. Here, Naniima et al present the structures of three additional homologs of UL25: KSHV pORF19, MHV68 pORF19, and HCMV UL77. Interestingly, the KSHV pORF19 crystallized as a pentamer. By fitting the pentameric crystal assembly into the cryo-EM densities for the portal cap on the KSHV capsid, the authors conclude that the pORF19 pentamer observed in the crystals represents the portal cap. Mutations at the pentameric interface reduce viral titer and abolish capsid assembly, prompting the conclusion that formation of the pORF19 pentamer is essential for capsid assembly. The structural data are solid, but there are concerns regarding the overinterpretation of the biochemical and the virological data, the inadequate description of experimental procedures, and the missing controls.
revealed the presence of the portal cap and suggested that it contains UL25. However, the presence of UL25 within the portal cap has not yet been shown conclusively. The authors reference both Newcomb 2001 and Trus 2004 in the introduction (refs 9 and 10) as studies that agree with these proposed density assignments in the EM studies, yet both references focus on the HSV-1 portal composed of UL6 and do not examine the presence or involvement of UL25 at the portal cap. To be able to conclude that pORF19, a UL25 homolog, is a component of the portal cap (and that the pentameric structure reported here could be important for its formation), the authors need to show that the portal, indeed, contains pORF19, e.g., by immunogold labeling, or, alternatively, show that the deletion of pORF19 eliminates the portal cap from the capsids. Short of this, the conclusion that the pentameric pORF19 structure presented here is present in the portal cap should be toned down throughout the manuscript (including the title and the abstract) and presented as a hypothesis rather than a conclusion.
While we agree with the reviewer that the presence of pORF19 in the KSHV portal cap has not been conclusively shown and therefore understand the reviewer's concern, we consider it difficult to address this issue by immunogold labeling studies to unambiguously show the presence of pORF19 in the portal cap. Such an endeavour will be problematic for two reasons: 1) the absence of a good labelling antibody and 2) the simultaneous presence of pORF19 in the penton vertices that will make it difficult to distinguish the different vertices. We have therefore toned down our conclusions and offer the notion that our pentamer represents the portal cap as a likely, but not yet proven hypothesis. In addition, we have clarified the referencing in the revised manuscript.
2. The observation that mutations at the pentameric interfaces in pORF19 abolish capsid assembly in infected cells contradict the author's own data that show capsid assembly even in the absence of pORF19 in insect cells expressing capsid proteins. A bigger concern is the fact that this finding contradicts a body of literature on capsid assembly in alphaherpesviruses, such as HSV-1 and PRV, that shows that A-and B-capsids can form in infected cells even in the absence of UL25 (McNab 1998, Stow 2001, Mettenleiter 2006, Baines 2014. For example, C-terminal truncations in HSV-1 UL25 do not preclude formation of A-or B-capsids (Baines 2014). Furthermore, KSHV capsids can form in vitro the absence of the portal (Grzesik 2017), a finding confirmed by the authors.
We respectfully disagree with this reviewer in this point. It is true that in a baculovirus overexpression system, KSHV capsids can form in the absence of pORF19, but such an experiment only poorly reflects the capsid assembly process in vivo. The massive protein overexpression observed during baculovirus infection facilitates assembly of regular capsids even in the absence of the portal as stated by this reviewer, which has not yet been described in infected cells. It appears reasonable that the available protein levels in the host cell can affect or determine KSHV capsid formation in insect cells and we have discussed this hypothesis in the revised manuscript (lines 406-412). While we cannot exhaustively explain the observed difference to the alphaherpesvirus literature, we do not claim that such mutations in a putative pentameric pUL25 interface will inhibit HSV-1 capsid assembly and we agree with the reviewer that there likely are unanticipated differences between alphaand gammaherpesviruses in this respect (see reviewer's comment below). Our pORF19 knockout mutant is truncated at a similar position upstream of the globular domain as a knockout mutant described in HSV-1 for pUL25 (McNab et al., 1998), however, the observed phenotype is strikingly different with no capsid formation in KSHV vs. A-and B-capsid formation in HSV-1. These data clearly indicate functional differences between HSV-1 pUL25 and KSHV pORF19 and we have now included that in the Results section of the revised manuscript.
One potential explanation is that the mutant viruses contain additional, unintended mutations that disrupt capsid assembly. This interpretation is consistent with poor in-trans complementation with WT pORF19.
The sequences of the entire bacterial artificial chromosomes used in this study have been verified by Next Generation Sequencing as stated in the Results section and we can therefore exclude unintended mutations that disrupt capsid assembly. As we state in the Results section we interpret the incomplete transcomplementation as a result of the presence of mutant pORF19, which likely acts as dominant-negative competitor in pORF19 pentamerization (at the portal) or formation of the CVSC helical bundle (at the penton) due to residual interactions with wt pORF19.
Alternatively, UL25 homologs could have differing roles in capsid formation during infection such that whereas HSV-1 UL25 is not essential for capsid formation, its KSHV homolog pORF19 is. This finding would be very interesting but requires that the authors perform additional controls, namely, a positive control showing that capsids do not form in the absence of pORF19 as well as a negative control showing that mutations located away from the pentameric interface has no effect on capsid formation.
We thank the reviewer for this constructive comment and as stated above we agree with the reviewer in this point. We already show the pORF19 deletion experiment suggested by the reviewer: the phenotype resulting from the pORF19 knockout is lack of capsid formation (Figs. 6-8), which is in stark contrast to the phenotype observed by McNab and colleagues for the HSV-1 KUL25NS mutant. It will be interesting to analyse in more detail whether this phenotype reflects an actual functional difference or whether it is due to secondary effects (e.g., differences in protein expression levels). With respect to the requested negative control, we consider usage of such a mutant virus with 'mutations located away from the pentameric interface' difficult to interpret. In view of the poorly understood pORF19 functions, it will be problematic to confirm that a particular mutation does not accidentally also impair a yet unknown important pORF19 function.
Along the same lines, it is difficult to reconcile the observed differences in the mutant phenotypes observed in Fig. 6 vs. Figs. 7 and 8. For example, the DQ and VL mutants appear to have a similar defect in progeny production (Fig. 6). However, only DQ appeared to form capsids (Fig. 7), predominantly A-and B-capsids, whereas the VL mutant produced no capsids (as observed in Figs, 7 and 8). Proper controls are needed to resolve these discrepancies.