Citrullination profile analysis reveals peptidylarginine deaminase 3 as an HSV-1 target to dampen the activity of candidate antiviral restriction factors

Herpes simplex virus 1 (HSV-1) is a neurotropic virus that remains latent in neuronal cell bodies but reactivates throughout an individual’s life, causing severe adverse reactions, such as herpes simplex encephalitis (HSE). Recently, it has also been implicated in the etiology of Alzheimer’s disease (AD). The absence of an effective vaccine and the emergence of numerous drug-resistant variants have called for the development of new antiviral agents that can tackle HSV-1 infection. Host-targeting antivirals (HTAs) have recently emerged as promising antiviral compounds that act on host-cell factors essential for viral replication. Here we show that a new class of HTAs targeting peptidylarginine deiminases (PADs), a family of calcium-dependent enzymes catalyzing protein citrullination, exhibits a marked inhibitory activity against HSV-1. Furthermore, we show that HSV-1 infection leads to enhanced protein citrullination through transcriptional activation of three PAD isoforms: PAD2, PAD3, and PAD4. Interestingly, PAD3-depletion by specific drugs or siRNAs dramatically inhibits HSV-1 replication. Finally, an analysis of the citrullinome reveals significant changes in the deimination levels of both cellular and viral proteins, with the interferon (IFN)-inducible proteins IFIT1 and IFIT2 being among the most heavily deiminated ones. As genetic depletion of IFIT1 and IFIT2 strongly enhances HSV-1 growth, we propose that viral-induced citrullination of IFIT1 and 2 is a highly efficient HSV-1 evasion mechanism from host antiviral resistance. Overall, our findings point to a crucial role of citrullination in subverting cellular responses to viral infection and demonstrate that PAD inhibitors efficiently suppress HSV-1 infection in vitro, which may provide the rationale for their repurposing as HSV-1 antiviral drugs.

1.It is very interesting that PAD3 protein levels are undetectable in HFF cells until after HSV-1 infection, and that HSV-1 specifically seems to need PAD3 in this cell type.However, as the authors did not test the specific PAD3 inhibitors in the other cell types included in this study (ie. the epithelial ARPE-19 cells or the SH-SY5Y neuroblastoma cells), or any other cell types for that matter, the conclusion can only be made about HFF cells.It is troublesome that there is no detectable basal levels of PAD3 in HFF cells, which begs the questions what does HSV-1 infection look like in cells with basal levels of PAD3?Clearly HFF cells express basal levels of PAD2 and 4, and HSV-1 infection appears to up regulate these isoforms as well, but with different kinetics and to different levels.Would this happen with PAD3 too, in a cell type that expresses basal levels of PAD3?Is what is observed with PAD3 in HFF cells just a cell-type specific phenomenon?The authors do show that general, pan PAD inhibition inhibits HSV-1 replication in additional cell types, but the conclusions about the PAD3 isoform specifically would be made stronger if the qPCR and westernblot analysis was done in these cell types as well.It would be greatly beneficial if the authors also included a cell type that had basal levels PAD3 expression, like HEK293T cells see citation: (Pong U, et.al. 2014(Pong U, et.al. , https://doi.org/10.1016(Pong U, et.al. /j.bbamcr.2014.02.018).02.018).The same is true for the PAD3 inhibition experiments done with , these experiments should be repeated in the ARPE-19 and SH-SYY5 cell types, and perhaps HEK293T cells.These experiments would address if the isoform specificity found with HSV-1 infection is simply a matter of the HFF cells themselves, or something that occurs in all HSV-1/PAD3 relevant cell types.
We thank Reviewer #1 for pointing out these issues related to the generalizability of our findings.As requested, we have performed qPCR and Western blot analyses using the other cell lines included in our study, namely ARPE-19 and SHSY-5Y, and, as suggested by this Reviewer, HEK293T.These new data, presented in panel D of new Fig. 2, and discussed on page 8 (lines 174-179), demonstrate that PAD3 is upregulated during HSV-1 infection (MOI 1) in all the cell lines tested.Despite minor variations between the cell lines, this suggests that the phenomenon observed in HFFs may indeed extend to other cell types.Notably, the kinetics of PAD3 consistently paralleled those of the viral protein ICP27, indicating that the timing of PAD3 production is primarily influenced by the permissiveness of the cells and, consequently, by the kinetics of viral replication rather than being driven by differences in PAD3 basal expression.
In response to the Reviewer's additional suggestion, we have assessed the impact of the PAD3 inhibitors HF4 (0.5-5 µM; see new Fig.4E) and CAY10727(0.5-1µM; see new Fig.4F) on viral plaque formation in HSV-1-infected (MOI 1) SH-SY5Y, ARPE-19, and HEK293T cells (see page 11, lines 230-233).These compounds were administered 1 h before virus adsorption and kept throughout the whole experiment.After 24 h of continuous exposure to the inhibitors, we observed a dose-dependent decrease in the number of viral particles in all treated cells.Finally, we conducted a cell viability assessment using the MTT assay in all cell lines treated under the same conditions and found that the drugs were not cytotoxic at these concentrations (refer to Suppl.Fig 3C and 3D).Collectively, these results indicate that HSV-1 infection can modulate specific PAD isoforms (e.g., PAD3) across various cell types.2A and discussed in the text lines 152-156 is normalized to GAPDH, which is an insufficient housekeeping gene to use with HSV-1 infection.It is very well documented that GAPDH is not a sufficient housekeeping gene to use for HSV-1 infection, as this transcript is virtually degraded by the vhs protein by 6hpi.see references (Hsu et.al. 2005, https://doi.org/10.1128/JVI.79.7.4090-4098.2005, and Watson, et.al. 2007https://doi.org/10.1186/1743-422X-4-130)Due to this reason, the data presented at 16hpi, 24hpi and 32hpi in Figure 2A is likely an artificially inflated mean fold change.Therefore, the experiment presented in figure 2A should be repeated using a different method, such as qRTPCR using a standard curve made from plasmids encoding each PAD isoform, or an RNA-seq experiment where each gene of interest can be addressed individually over time.This is particularly relevant because the data presented in Figure 2A makes it appear that the basal levels of the PAD isoforms are equal in HFF cells, and that the lack of PAD3 protein expression in this cell type is due to regulation after the transcriptional level.This is confusing for the reader, particularly when the authors go on to mechanistically address this phenomenon as stated on line 166, using cycloheximide and IFN gamma experiments, as stated on line 166.

Together the cycloheximide and IFN-gamma data only demonstrate that viral protein production is required for PAD3 up regulation, and that this up regulation is independent of the IFN response. This combined with the artificially inflated qRTPCR data make it difficult to understand what is actually going on with PAD3 expression during HSV-1 infection. In order to gain mechanistic insights into PAD3 up-regulation many more experiments need to be included than what the authors have currently done. These experiments would analyze PAD3 expression at the transcriptional level, the protein translational level, the mRNA decay level, and the protein stability level. PAA experiments should also be included to determine if PAD3 up regulation is due to a viral immediate early/ early protein, or a viral late protein.
We thank the Reviewer for keenly pointing this out.As requested, we have repeated the RT-PCRs using two other housekeeping genes, namely pyruvate dehydrogenase kinase 1 (PDK1) and beta2microglobulin (B2M), which are among the most conserved during HSV-1 infection and are widely used in this type of analysis.As shown in the figures below, regardless of the reference gene employed, we observed a consistent pattern of PAD induction, with a marked upregulation of PADI3 gene starting from 16 hpi, even though the absolute values may show minor variations from one reference gene to another.In light of these findings, we would rather maintain Fig. 2A in its original form given that the main message of this experiment remains unchanged and is robustly confirmed by the Western blot analyses presented in new Figures 2B and 2D.
Overall, we concur with this Reviewer's observation that further data are needed to gain more mechanistic insights into PAD3 upregulation.To address this issue, we have performed initial experiment aimed to assess the promoter activity of the PADI3 gene by dual luciferase assay.As detailed on page 9 (lines 185-190), HFFs were transiently transfected with the luciferase reporter plasmids carrying the wild-type promoter region of PAD3.Twenty-four h after electroporation, cells were infected with HSV-1.The updated Figure 3A shows that HSV-1 infection leads to a 6-fold induction of PADI3 promoter activity, indicating that PADI3 is upregulated at the transcriptional level through promoter activation following HSV-1 infection.Furthermore, as per the Reviewer's suggestion, we conducted experiments with the viral DNA synthesis inhibitor phosphonoformic acid (PFA, foscarnet) to determine whether PAD3 upregulation was linked to a viral immediate early/early or late protein.As expected, PFA treatment of HSV-1-infected HFFs led to a significant reduction in the synthesis of the late viral protein gD, but it had only a marginal effect on PAD3 upregulation (see new Fig.3D and comments on page 9, lines 196-200).These combined findings, which imply that viral late proteins likely do not play a role in PAD3 transcriptional activation, have been integrated into a new Results section entitled "PAD3 protein levels are induced through a calciumindependent HSV-1-early mechanism".This section has been consolidated with the new results related to calcium influx (see next point and refer to pages 9-10, lines 201-210).

It is well documented that HSV-1 infection causes an influx in intracellular calcium. Seeing that PAD3 expression is induced by an influx of intracellular calcium, it is possible that PAD3 up-regulation is a result of this rather than a direct up regulation of PAD3 gene expression by the HSV-1 viral proteins themselves. HSV-1 infection progresses though the temporal cascade much quicker than HCMV infection in HFF cells, it is possible that the influx of calcium occurs to a much greater level and to a quicker extent than in HCMV infection. This could be an alternate explanation to the results surrounding PAD3 specificity in HSV-1 infection which needs to be explored further.
Experiments using artificially induced intracellular calcium levels should be included to rule out this possibility as a mechanism of action-similar to what the authors did for IFN gamma.
We are grateful to Reviewer #1 for providing us with the opportunity to address this crucial issue.Following the Reviewer's valuable suggestion, we performed experiments in which HFFs were exposed to thapsigargin (TG).TG is well-established ER stressor known for its ability to elevate cytoplasmic calcium levels by interfering with the sarcoplasmic/endoplasmic reticulum Ca2 + -ATPase (Quynh Doan et al., Curr Pharm Des. 2015, doi: 10.2174/1381612821666151002112824).In these experiments, we treated mock-infected HFFs with two different concentrations of TG and assessed PAD3 expression at both mRNA and protein levels after 16 h.As shown in new Fig.3E and F, unlike HSV-1 infection (MOI 1; 16hpi), TG exposure failed to induce a significant upregulation of PAD3 at any of the concentrations applied.Of note, TG treatment of HFFs led to a substantial increase in ATF-6 expression levels (Fig. 3E), confirming its effectiveness as an ER stressor at both concentrations used.These results have been included in the new Results section entitled "PAD3 protein levels are induced through a calcium-independent HSV-1-early mechanism" (pages 9-10, lines 201-210).
Collectively, our results suggest that the overexpression of PAD3 observed during HSV-1 infection is the result of a direct transcriptional upregulation of PAD3 gene expression by HSV-1 viral proteins rather than the result of intracellular calcium influx caused by HSV-1 infection.

Reviewer #2: None noted.
We sincerely thank the Reviewer for his/her kind appreciation of our work and thoughtful comments.

Figure 1F: The 'pretreatment' arm is not really pretreatment as it overlaps with HSV addition for 2 hours. Normally pretreatment means washing off prior to viral addition and if done in that way may provide an even clearer discrepancy in panel G with 'post -treatment and a better indication of potential mechanism.
We completely agree with the input from Reviewer #2.As suggested, we have now carried out experiments involving three different treatment schedules with Cl-A and BB-Cl (new Figs.1F and  1G).These schedules include the one recommended by the Reviewer, where the drugs are added before infection and removed just prior to viral infection (pre-treatment, Pre).In addition, we retained the previous treatment schedules: treatments maintained for the first 2 h of infection (now referred to as Pre+), and treatments extended until 24 hpi (referred to as Post).Our results show only slight differences between the Pre and Pre+ conditions, and both failed to yield a significant reduction in the number of viral particles when compared to the Post treatment schedule.These new results have been included on page 7 (lines 144-149).We thank the Reviewer for his/her valuable suggestion, which allowed us to deepen our understanding of the mechanism of action of PAD inhibitors.

Supplementary Fig 1F:
the protein citrullination pattern according to molecular weight differs from Figure 1A.Why?
We thank this Reviewer for his/her thorough examination of our results.Regrettably, our RhPG experiments show some partial inconsistencies, primarily due to slight variations in blot running time, gel homemade preparation, and the lots of probes used.Nevertheless, we firmly believe that the essential aspects remain consistent regardless of these conditions.These include the high intensity bands observed between 37 and 50 kDa in all the samples, as well as the observed increase in signal intensity between 75 and 150 kDa during infection.These consistent data, combined with mass spectrometry data (new Fig. 5A), attests to the robustness of our results.

Figure 2c:
In Discussion the authors speculate that HSV immediate early proteins are responsible for stimulating the PADs and citrullination, a key question.To differentiate IE/E proteins from late structural proteins the experiment should be repeated with foscarnet (and if confirmed, eventually with multiple IE/E protein mutants).
In line with the valuable suggestions put forth by Reviewers #1 and 3, we have duly conducted the experiments with foscarnet.Please see our response to Reviewer #1's point 3.

Part III -Minor Issues: Editorial and Data Presentation Modifications
Reviewer #1 1. Image J should be used to quantify the data presented in the westerblots, particularly surrounding the PAD2 and PAD4 isoforms that do have basal levels in mock cells.This way all western blots, even the ones not shown in the paper could be used the quantification.
As suggested by Reviewer #1, we have now presented the densitometric analysis to quantify the data presented in Fig. 2B relative to PAD2 and PAD4 isoforms (new Supplementary Fig. 2B) 2. The westerblots for PAD3 in Figure 2 look underexposed in comparison to the westerblots for PAD2 and PAD4.This is concerning as the authors go to great lengths to point out that there is no PAD3 protein expression in mock infection.The difference in background color on the blots between PAD2, PAD3, and PAD4, particularly in Figure 2E, is questionable.
We apologize for any misunderstanding that may have arisen regarding the quality of our Western blots.In all our acquisitions, we used the standard settings for automatically optimizing exposure time, both under high sensitivity and high resolution conditions.Regarding PAD3, when we manually changed the exposure time, no bands appeared in the mock even though the existing bands became saturated, with their signals extending to cover the mock lane when the exposure times were further increased (see the figures below).We trust that this clarification, coupled with the new Western blot results presented in new Fig.2D for the other cell lines, provides sufficient evidence to assure the Reviewer that, under our experimental conditions, PAD3 protein was undetectable by Western blots in mock-infected HFFs.
3. The authors point out that many viral proteins were found to be citrullinated in their screen at 16 and 24 hpi, in addition to the IFIT1 and IFIT2 proteins among others in Figure 4 A and B. The authors use an IP with an anti-peptidylcitrulline antibody to pull down IFIT1 and IFIT2 to show they are citrullinated during infection.As UL54 (ICP27) was shown to be citrullinated at 16hpi in Figure 4B, the authors should be able to pull ICP27 down with the antibody used for the IP in Figure 4C.They should do this to confirm their results in their volcano plot, and it should be easy as they already have all regents required.
We thank the Reviewer for this important comment, which significantly enhanced our citrullination profile results.In response to the suggestion, we conducted an ICP27 pull-down experiment using the same protein extract and the same antibody used for the IP of the cellular proteins IFIT1 and IFIT2.As shown in the new Fig.5D, the IP confirmed the findings from the volcano plot, showing that viral ICP27 is somewhat undergoing citrullination at least at 16 hpi.This result has now been included in the last paragraph of the Results section (page 12, lines 270-276).
4. The authors should add to the discussion after line 315 to speculate why there are citrullinated viral proteins during HSV-1 infection.
We are grateful to the Reviewer for raising this interesting topic for discussion.Speculation about citrullinated viral protein were added in the new manuscript (pages 14-15, lines 326-332).
5. The way the figures and the supplementary figures are presented is very confusing, and it makes it difficult for the reviewer/reader to find the figures that are being discussed.Many of the figures and the supplementary figures and legends have been separated.For example, Figure 1 A, B and C are presented as a separate figure from Figure 1 D, E, F and G.They each have their own legend, and therefore should be renumbered to be their own figures.The authors have done this with supplementary figures as well.The figures and supplementary figures should be re-numbered so each