Megan Steain ,
Max O. D. G. Baker ,
Chi L. L. Pham,
Brian P. McSharry,
Margaret Sunde ,
Max O. D. G. Baker,
Chi L. L. Pham,
Brian P. McSharry,
Published: July 10, 2020
?This is an uncorrected proof.
AbstractHerpesviruses are known to encode a number of inhibitors of host cell death, including Rip Homotypic Interaction Motif (RHIM)-containing proteins. Varicella zoster virus (VZV) is a member of the alphaherpesvirus subfamily and is responsible for causing chickenpox and shingles. We have identified a novel viral RHIM in the VZV capsid triplex protein, open reading frame (ORF) 20, that acts as a host cell death inhibitor. Like the human cellular RHIMs in RIPK1 and RIPK3 that stabilise the necrosome in TNF-induced necroptosis, and the viral RHIM in M45 from murine cytomegalovirus that inhibits cell death, the ORF20 RHIM is capable of forming fibrillar functional amyloid complexes. Notably, the ORF20 RHIM forms hybrid amyloid complexes with human ZBP1, a cytoplasmic sensor of viral nucleic acid. Although VZV can inhibit TNF-induced necroptosis, the ORF20 RHIM does not appear to be responsible for this inhibition. In contrast, the ZBP1 pathway is identified as important for VZV infection. Mutation of the ORF20 RHIM renders the virus incapable of efficient spread in ZBP1-expressing HT-29 cells, an effect which can be reversed by the inhibition of caspases. Therefore we conclude that the VZV ORF20 RHIM is important for preventing ZBP1-driven apoptosis during VZV infection, and propose that it mediates this effect by sequestering ZBP1 into decoy amyloid assemblies.
Rip homotypic interaction motifs (RHIMs) are found in host proteins that can signal for programmed cell death and in viral proteins that can prevent it. Complexes stabilized by intermolecular interactions involving RHIMs have a fibrillar amyloid structure. We have identified a novel RHIM within the ORF20 protein expressed by Varicella zoster virus (VZV) that forms amyloid-based complexes with human cellular RHIMs. Whereas other herpesvirus RHIMs inhibit necroptosis, this new VZV RHIM targets the host RHIM-containing protein ZBP1 to inhibit apoptosis during infection. This is the first study to demonstrate the importance of the ZBP1 pathway in VZV infection and to identify the role of a viral RHIM in apoptosis inhibition. It broadens our understanding of host defense pathways and demonstrates how a decoy amyloid strategy is employed by pathogens to circumvent the host response.
Citation: Steain M, Baker MODG, Pham CLL, Shanmugam N, Gambin Y, Sierecki E, et al. (2020) Varicella zoster virus encodes a viral decoy RHIM to inhibit cell death. PLoS Pathog 16(7):
https://doi.org/10.1371/journal.ppat.1008473Editor: Edward Mocarski, Emory Vaccine Center, UNITED STATESReceived: March 6, 2020; Accepted: May 28, 2020; Published: July 10, 2020Copyright: © 2020 Steain et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Data Availability: All relevant data are within the manuscript and its Supporting Information files.Funding: This work was supported by funding from the Australian Research Council to M. Sunde and E.S. (DP180101275), Research Training Program support to MB, a BioMed Connect Grant from the Sydney Medical School to M. Sunde and A. A., a University of Sydney Bridging Scheme grant to M. Steain and National Health and Medical Research Council funding to A. A. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.Competing interests: The authors have declared that no competing interests exist.
IntroductionViruses have evolved a diverse range of strategies to evade host intrinsic, innate and adaptive immune responses. Members of the Herpesviridae family, Herpes simplex virus (HSV) -1 and human and murine cytomegalovirus (HCMV/MCMV), are masters at manipulating host cell death pathways such as apoptosis and necroptosis, in order to successfully spread and establish latency [1–3]. Although Varicella zoster virus (VZV) causes a significant health burden [4–6], the mechanisms employed by VZV to undermine host responses have not been fully elucidated. Primary infection with VZV leads to varicella, commonly known as chickenpox. During this infection the virus establishes latency within sensory neurons, and when VZV-specific T cell immunity wanes, the virus can reactivate to result in herpes zoster (shingles) . Complications arising from VZV reactivation include protracted pain termed post-herpetic neuralgia, encephalitis and VZV vasculopathy (reviewed in ).
VZV is a highly cell-associated virus and does not release cell-free virions into culture , necessitating cell-associated propagation of the virus in vitro. VZV is also highly species-specific, not progressing to productive infection in cells of non-human origin . Therefore no animal model exists which re-capitulates the full spectrum of VZV disease. Combined, these properties of VZV have largely hampered the characterization of VZV pathogenesis.
Cell death can be initiated upon sensing of a viral pathogen as a means of prematurely aborting the virus lifecycle and thereby limiting the spread of infection. Apoptosis and necroptosis may form part of the antiviral response [2, 11–14]. The most well characterized pathway of necroptosis is that induced by tumour necrosis factor (TNF). Upon TNF binding to its receptor (TNF-R1), cellular inhibitor of apoptosis proteins (cIAP 1 and 2) mediate K63-ubiquitination of receptor interacting protein kinase 1 (RIPK1), as part of a signalling platform known as complex I, which drives NF-κB signalling (reviewed in . If cIAPs are inhibited (e.g. by SMAC mimetics), RIPK1 can associate with FADD, caspase 8 and cFLIP (termed complex II). This can lead to the cleavage of caspase 8 to drive apoptosis [16, 17]. However, if caspase 8 is also inhibited then RIPK1 can interact with receptor interacting protein kinase 3 (RIPK3) to drive necroptosis [18, 19].
RIPK1 and RIPK3 interact via highly conserved Rip Homotypic Interaction Motifs (RHIMs), leading to the assembly of the necrosome complex . The RIPK1:RIPK3 necrosome complex has a fibrillar structure and is stabilised by an amyloid cross-β core, formed by interactions between the RHIMs of the two component proteins . Generation of the necrosome results in the phosphorylation of RIPK3, which subsequently phosphorylates mixed lineage kinase domain-like protein (MLKL) . Phosphorylated MLKL drives membrane permeabilisation, resulting in rupture of the cell and necroptotic cell death [23, 24].
The cellular Z-DNA binding protein 1 (ZBP1, also known as DAI) contains at least two RHIMs (A and B) and can also interact with RIPK3 [25–27]. ZBP1 was originally identified as a cytosolic DNA sensor , however more recently, studies with Influenza A virus (IAV) and MCMV have shown that ZBP1 can sense viral RNA [29, 30] and simultaneously activate apoptosis and necroptosis in the same population of cells . This implies that the inhibition of caspase 8 is not always necessary for necroptosis to proceed via this pathway.
Some viruses, including HSV-1, HSV-2, HCMV and MCMV, have evolved mechanisms to inhibit caspase 8 and therefore extrinsic apoptosis [31–33], as well as necroptosis [2, 34, 35]. It has been suggested that necroptosis evolved as a backup mechanism to ensure the elimination of virally infected cells . This theory is supported by studies showing that mice which are deficient in RIPK3, a kinase essential for necroptosis, succumb to fatal vaccinia virus infection whereas wild type mice do not .
Many of the currently recognized viral protein inhibitors of necroptosis contain a RHIM . MCMV encodes a RHIM close to the N-terminus of a non-functional ribonucleotide reductase, M45. This virally-encoded RHIM has been shown to interact with RIPK3 to inhibit necroptosis [34, 38]. Furthermore, mutation of MCMV to remove the ability of the virus to inhibit necroptosis renders MCMV highly attenuated in vivo . We recently proposed a mechanism by which M45 subverts RHIM-based cell death signalling, by forming heteromeric decoy amyloid structures . We demonstrated that M45, like human RHIM proteins, is able to spontaneously form amyloid fibrils. We also demonstrated that M45 is capable of forming networks of hybrid, heteromeric amyloid structures with RIPK1 and RIPK3, in a manner that is more favourable than the interaction of the two human proteins with each other. It is likely that these human:viral protein complexes, by some property of their conformation, are unable to signal to downstream effectors, and thus cell death signalling is abridged.
HSV-1 and -2 also contain RHIMs in the N-terminal regions of functional ribonucleotide reductases infected cell protein (ICP)6 and ICP10 respectively [40–42]. ICP6 and ICP10 have been shown to block necroptosis in cells of human origin in response to TNF and FasL [41, 42]. Further, ICP6 has been reported to protect human cells from ZBP1-induced cell death .
VZV, HSV-1 and HSV-2 all belong to the Alphaherpesvirinae subfamily and share a high degree of homology. Thus, we sought to determine if VZV also contained a RHIM that could inhibit necroptosis. We identified a RHIM sequence within the open reading frame (ORF) 20 capsid triplex protein. Like the well-characterised RHIMs in RIPK1, RIPK3 and M45, this RHIM is able to drive the formation of amyloid structures and the ORF20 RHIM interacts with RIPK3 and ZBP1 RHIMs in vitro. Strikingly however, whilst VZV can inhibit TNF-induced necroptosis, this effect cannot be attributed to the presence of the RHIM in the VZV ORF20 protein. Instead, this newly identified viral RHIM interacts with ZBP1 and its primary function appears to be the suppression of ZBP1-driven apoptosis through the formation of decoy hybrid amyloid structures.
VZV ORF20 contains a RHIM
Given that in HSV-1 and HSV-2 the large subunit of the ribonucleotide reductase contains a functional RHIM [40–42, 44], we examined the amino acid sequence of the orthologue in VZV, which is encoded by ORF19. The C-terminal domains of the ribonucleotide reductase R1 from HSV-1 and VZV (ICP6 and ORF19 respectively) share 44% homology at the amino acid level , however we found that VZV ORF19 encodes a shorter R1 that lacks a RHIM. We then examined the remainder of the VZV genome to look for a potential RHIM. We found a RHIM-like sequence within ORF20, which encodes the capsid triplex subunit 1 (Fig 1A and 1B). This putative RHIM is located in the N-terminal region of ORF20, between L24 and Y40 (Fig 1B). ORF20 is homologous to the HSV-1 capsid triplex subunit protein VP19C . There is low conservation in the first 111 amino acids of VP19C compared to orthologue proteins in other herpesviruses, and mutational analysis has revealed that capsid assembly is not disrupted by alterations to the first 107 codons of VP19C . It is therefore possible that the N-terminal of ORF20 may serve as a cell death inhibitor without disrupting capsid function. The core tetrad of residues in this potential VZV RHIM is IQIG, conforming to the I/V-Q-I/V/L-G consensus found in cellular RHIMs , and a number of other residues are conserved. This potential RHIM was conserved within every VZV genome that we examined, including the vaccine strains Varivax and VarilRix (S1A Fig). We then extended our search and found that at least the core tetrad residues of this potential RHIM (I/V-Q-I/V-G) were conserved in other members of the Varicellovirus genus for which an ORF20 orthologue sequence was available, including Simian Varicella Virus (SVV), Pseudorabies virus (PRV), bovine herpesvirus (BHV) 1 and 5 and equine herpesvirus (EHV) 1, 4, 8 and 9 (S1B Fig). This level of conservation strongly suggests that this motif is essential for the successful dissemination of Varicelloviruses.
Fig 1. VZV ORF20 contains a RHIM that inhibits TNF-induced necroptosis in HT-29 adenocarcinoma cells.(A) Schematic diagram of RHIM-containing proteins. RHIM location indicated by yellow box. (B) Amino acid sequence alignment of the new RHIM identified in VZV ORF20 (Dumas strain) with other known human cellular RHIMs from RIPK3, RIPK1, TRIF and ZBP1 and the viral RHIMs from MCMV (M45), HSV-1 (ICP-6) and HSV-2 (ICP10), also indicating the percentage of conservation and consensus sequence, with the RHIM core tetrad boxed. (C) Cytopathic effect and immunofluorescence staining for VZV antigens (IE62, pORF29 and the gE:gI glycoprotein complex, green) in VZV-infected and mock HT-29s. Following immunostaining cells were counterstained with DAPI (blue). Scale bars indicate 20 μm (D) Viability of mock and VZV infected HT-29s (72 h post-infection) following treatments with TNF (T; 30 ng/ml), BV-6 (S; 1 μM), z-VAD-fmk (V; 25 μM) and Necrostatin-1 (Nec1; 30 μM) alone or in combination as indicated. Data was normalised to DSMO only control. Error bars show standard error of the mean, from 4 independent statistical significance was determined using a two-way ANOVA.
Varicella zoster virus can infect HT-29 adenocarcinoma cells and block TNF-induced necroptosis
Many cultured cells, such as human fibroblasts (HFs), which are commonly used to grow and study VZV infection, lack the expression of RIPK3, rendering them resistant to necroptosis . Further, VZV is highly-species specific and can only establish a full productive infection in cells of human origin . Therefore, in order to study the modulation of necroptosis pathways by VZV we had to identify a cell line that was susceptible to both VZV infection and necroptosis. The human colorectal adenocarcinoma cell line HT-29 has been used extensively to study necroptosis [19, 22, 41, 47], however to date there have been no reports of productive VZV infection in this cell line. In order to determine if VZV could infect HT-29s, cells were grown to ~70% confluence and then VZV rOKA-infected HFs were added at a ratio of 1:8. A cell-associated infection was used due to the highly cell-associated nature of VZV infection in vitro  and this approach is routinely used to infect cells in vitro [49–52]. Within 72 h cytopathic effect (CPE) was readily observed within VZV-infected HT-29s (Fig 1C) and further passaging of the virus in a cell-associated manner in HT-29s could be continued. In addition, following several passages to eliminate the infecting HF inoculum, immunofluorescence staining for VZV immediate early (IE62), early (pORF29) and late proteins (gE:gI complex) was performed. This showed that the full cascade of VZV gene expression occurred in HT-29s, and the cellular localisation of each viral antigen was typical of a productive infection [53–55] (Fig 1C) Together this shows that VZV is able to productively infect HT-29 adenocarcinoma cells.
In order to determine if VZV infection could confer resistance to necroptosis, VZV-infected HT-29s (72 h post-infection, 24–45% gE:gI antigen +) and mock-infected HT-29s were treated with combinations of TNF (T), the Smac mimetic BV-6 (S) and z-VAD-fmk (V) to inhibit caspase 8. The percentage of surviving cells was then determined 17–18 h post-treatment by measuring intracellular ATP levels. On average from four biological replicates, treatment of the cells with TNF alone reduced cell survival in the VZV infected cells compared to mock to a modest yet significant degree, although both mock and VZV-infected HT-29 cells were equally susceptible to apoptotic cell death induced by T+S treatment (Fig 1D). However, following treatment to induce necroptosis (T+S+V), significantly more cells from the VZV infection survived compared to mock (on average 69% vs. 39%) (Fig 1D). The addition of necrostatin-1 inhibited the cell death induced by T+S+V in both the mock and VZV infected populations, confirming the involvement of RIPK1 in this cell death pathway. Additionally, mock and VZV-infected cells treated with combinations of T+S+V were collected for immunoblot (IB) analysis. These analyses showed that phosphorylated MLKL (pMLKL) a hallmark of necroptosis induction could be readily observed in mock infected, but not in VZV infected cells, treated with T+S+V (S1C Fig). VZV infection also did not appear to alter total cellular levels of MLKL. Together, these results suggest that VZV infection confers a resistance to TNF-induced necroptosis.
VZV infection inhibits phosphorylation of MLKL
The cell-associated nature of VZV infections means that within the VZV-infected culture there is a mix of bystander and virally infected cells. In order to determine if VZV-infection prevented the phosphorylation of MLKL only within infected cells, dual immunofluorescence staining for VZV antigen and pMLKL was performed. Mock and VZV-infected HT-29s were treated with DMSO alone (control) or with T+S+V for 7–8 h, then fixed and immunostained for the VZV immediate early viral protein IE62 (green) and pMLKL (red). Within the mock-infected HT-29 cultures treated with T+S+V, pMLKL could be readily seen localised to cellular membranes (Fig 2A), with an average of 21.8% (range: 15.9–28.8%) of cells pMLKL positive over three biological replicates (Fig 2B). Within the VZV-infected HT-29 culture, pMLKL staining was observed predominantly in VZV-antigen negative cells following treatment (Fig 2A), with an average of 10.8% (range 8.6–15%) of antigen negative cells being positive for pMLKL and significantly fewer of the VZV-antigen positive cells (on average 3.4%, range 1.1–6.5%), (Fig 2B). No specific pMLKL staining was observed in DMSO control treated cells in the mock or VZV-infected cultures (Fig 2A). This data strongly suggests that VZV-infection can confer resistance to TNF-induced necroptosis prior to the phosphorylation of MLKL.
Fig 2. VZV infection inhibits phosphorylation of MLKL during TNF-induced necroptosis.(A) Immunofluorescence staining for phosphorylated MLKL (red) and VZV IE62 antigen (green) in mock and VZV infected HT-29 adenocarcinoma cells untreated (DMSO control) or treated with TNF (T; 30 ng/ml), BV-6 (S; 1 μM) and z-VAD-fmk (V; 25 μM) for 7–8 h to induce necroptosis. Following immunostaining cells were counterstained with DAPI (blue). Examples of pMLKL positive cells are shown with white arrows (B) The percentage of cells that were pMLKL positive was determined by randomly imaging 10–20 non-overlapping regions of each slide and manually counting cells from 3 independent experiments. Error bars show standard error of the mean, statistical significance was determined using a one-way ANOVA. Scale bar indicates 20 μm.
The VZV ORF20 RHIM does not inhibit TNF-induced necroptosis
In order to determine if the RHIM-like sequence identified within VZV ORF20 was responsible for the observed inhibition of TNF-induced necroptosis, we sought to express ORF20 in isolation in HT-29s. Initially we attempted a lentivirus-based stable-transduction approach and tested three different lentivirus constructs with ORF20 being driven by either a CMV or EF1α promoter. We also tested a transient transduction system using replication deficient adenoviruses. However with both approaches we found we could not achieve a high percentage of HT-29s expressing VZV ORF20, despite each vector driving robust ORF20 expression in 293T cells and/or HFs. Thus, we went on to mutate the ORF20 RHIM within VZV via “recombineering” using the VZV bacterial artificial chromosome (BAC). A tetra alanine substitution was made to the core amino acids of the ORF20 RHIM (IQIG to AAAA), as this mutation is commonly used to disable RHIM functions (Fig 1A). Previous studies of the viral RHIMs in MCMV and HSV-1 have shown that a tetra alanine substitution in the RHIM core largely abolishes the ability of these viruses to inhibit necroptosis [34, 41]. Following the production of both the parental virus and the ORF20 RHIM mutated virus (VZV-RHIMmut) in ARPE-19 epithelial cells, the virus was transferred to HT-29s and passaged several times (5+) to ensure there were no carry over ARPEs remaining. Cell survival was then assessed in mock and VZV infected HT-29s following treatments with combinations of T, S and V as outlined above. To ensure that results could be compared between the parent (BAC-derived pOKA) and the mutant virus, the percentage of HT-29s expressing VZV gE:gI glycoprotein complex was determined by flow cytometry at the beginning of each assay for both viruses (S2 Fig). This analysis demonstrated that the percentage of VZV antigen positive cells in the parental and VZV-RHIMmut infected HT-29 cultures was very similar. As seen previously for VZV rOKA (Fig 1), the BAC derived VZV pOKA strain was still capable of inhibiting necroptosis compared to mock (50.7% cell survival versus 29.9%). Surprisingly however, mutated virus with the tetra alanine substitution of the RHIM (VZV-RHIMmut) protected cells almost to the same extent (45% cell survival) (Fig 3). To further investigate the role of the RHIM, we constructed a virus that had the entire 20 amino acids of the ORF20 RHIM removed (VZV-RHIMKO). This virus was also able to inhibit necroptosis when compared to mock HT-29, however to a slightly lesser extent than what was seen for VZV pOKA (S2B Fig). This result suggests that the VZV ORF20 RHIM does not play a major role in the inhibition of TNF-induced necroptosis in HT-29 cells, and that other viral mechanisms are involved.
Fig 3. The VZV ORF20 RHIM is not involved in TNF-induced necroptosis.(A) Viability of mock, VZV and VZV RHIM mutated (VZV-RHIMmut) virus infected HT-29s (72 h post-infection) following treatments with TNF (T; 30 ng/ml), BV-6 (S; 1 μM), z-VAD-fmk (V; 25 μM) and Necrostatin-1 (Nec1; 30 μM) alone or in combination as indicated. Data was normalised to DSMO only control. Error bars show standard error of the mean, from 4 independent replicates and statistical significance was determined using a two-way ANOVA.
The VZV ORF20 RHIM inhibits ZBP1-driven apoptosis during infection
Pathways leading to cell death can originate from more than one upstream initiator [35, 37]. Considering that the VZV RHIM appeared to have a limited effect in modulating TNF-directed necroptosis, we sought to determine whether it could influence cell death signalling pathways originating from other triggers in the context of infection. Recent research examining intrinsic immune responses to viral infection has demonstrated the nucleic acid sensing-protein ZBP1 as a key initiator of a variety of cell death pathways in response to MCMV, HSV-1, and IAV [27, 29, 43, 56]. To test if ZBP1 plays a role in initiating programmed cell death during VZV infection, we performed infectious centre assays using HT-29s that were engineered to express ZBP1 via lentivirus transduction, or empty vector transduced control HT-29s. Monolayers of ZBP1-expressing or control HT-29s were infected at a ratio of 1 infected HT-29 (wild type) to 10 uninfected (ZBP1/control) cells with the VZV BAC-derived parent strain or the RHIM mutated virus (VZV-RHIMmut) and plaque formation was assessed at 72 h post-infection. Cells were immunostained for VZV IE63 and infectious centres visualised by fluorescence microscopy.
The parental VZV virus (pOKA) formed plaques in both the ZBP1-expressing and control empty vector cells (Fig 4A), with no significant difference in plaque size seen between the two cell types (Fig 4B). VZV-RHIMmut readily formed plaques in the control empty vector HT-29s, which were on average the same size as plaques formed by pOKA (Fig 4B). However in contrast, VZV-RHIMmut virus spread was severely restricted in the ZBP1-expressing cells (Fig 4A), with plaque size significantly reduced (Fig 4B). The VZV-RHIMKO virus formed larger plaques on average in the empty vector control HT-29s, however similarly to the RHIM mutant virus, plaque size was also significantly reduced in the ZBP1-expressing cells (Fig 4). Given that during Influenza A infection ZBP1 has been shown to drive both apoptosis and necroptosis , and as our prior results suggested that VZV could not inhibit caspase 8, we tested if either or both pathways were being triggered during infection with the two mutant viruses. Monolayers of ZBP1-expressing or control empty vector HT-29s were infected at a ratio of 1:10 and then the pan-caspase inhibitor z-VAD-fmk (25 μM), or the MLKL inhibitor necrosulfonamide (NSA, 1 μM), was added to the cultures. After 72 h of infection, it appeared that the addition of necrosulfonamide had little impact on the spread of any of the viruses (Fig 4). Treatment of the ZBP1-expressing HT-29s with T+S+V confirmed that these cells retained the ability to undergo necroptosis (S3 Fig). In contrast, the addition of z-VAD led to an increase in average plaque size for all three viruses in ZBP1 expressing cells (Fig 4B), and increased the average plaque size for the two RHIM mutant viruses to similar levels as seen in the untreated empty vector control cells. This suggests that the ORF20 RHIM functions to inhibit ZBP1-induced apoptosis during VZV infection. This is the first time a viral RHIM protein has been associated with the inhibition of a non-necroptotic cell death pathway.
Fig 4. ZBP1 restricts VZV-RHIMmut and VZV-RHIMKO spread in culture.Empty vector or ZBP1 expressing HT-29s were inoculated with parental HT-29 cells either mock or infected with parent VZV (pOKA), VZV-RHIMmut or VZV-RHIMKO. z-VAD–fmk (25 μM) or necrosulfonamide (NSA; 1 μM) was added at the time of inoculation where indicated. (A) After 72 h cells were fixed and immunostained for VZV IE63 (red) to assess virus spread. Images are representative from at least 2 independent replicates. Scale bar indicates 500 μm. (B) The area of 18–20 plaques per virus (as indicated) was calculated using Zen 3.1 Blue edition (Zeiss), and statistical significance calculated using a one-way ANOVA. Line indicates mean and error bars represent standard error of the mean.
RIPK3 is required for ZBP1-induced cell death to proceed in response to VZV infection
Previous reports describing ZBP1 induction of apoptosis in response to viral infection have indicated that, depending on cellular conditions, cell death can proceed in both a RIPK3-dependent and RIPK3-independent manner . In order to elucidate the specific molecular pathways required for ZBP1-induced apoptosis in the context of VZV infection, we sought to determine whether RIPK3 was necessary for cell death. To this end, we performed infectious centre assays with ARPE-19 cells, a cell line known to be permissive to VZV infection that possesses a full complement of caspases but does not express RIPK3 . For this assay, only the VZV parent strain and the VZV-RHIMmut were compared, since we had established that the IQIG core tetrad is crucial for the prevention of ZBP1-induced cell death (Fig 4). ZBP1-expressing ARPE-19 cells were engineered by lentiviral transduction and empty vector-transduced ARPE-19 cells were used as control. Monolayers of ZBP1-expressing or control ARPE-19 cells were infected at a ratio of 1 infected ARPE-19s (wild type) to 10 uninfected (ZBP1/control) cells, with the VZV BAC-derived parent strain or the RHIM mutated virus (VZV-RHIMmut), and plaque formation was assessed at 72 h post-infection. Cells were then immunostained for VZV IE63 and infectious centres were visualised by fluorescence microscopy (Fig 5). The parental VZV virus formed plaques in both the ZBP1 and control empty vector ARPE-19 cells to the same size and number. However, unlike in HT-29 cells expressing ZBP1, the VZV-RHIMmut was also able to form plaques equally as well in control and ZBP1-expressing ARPE-19 cells (Fig 5). The addition of zVADfmk did not change the phenotype for either VZV parent or VZV-RHIMmut (Fig 5). These data indicate that the presence of ZBP1 and caspase 8 is insufficient for activation of cell death and RIPK3 is most likely necessary for cell death in response to VZV infection.
Fig 5. ZBP1 does not restrict the spread of VZV or VZV-RHIMmut in RIPK3-deficient, apoptosis-capable cells.Empty vector or ZBP1 expressing ARPE-19s were inoculated with parental ARPE-19s cells infected with parent VZV (pOKA) or VZV-RHIMmut. z-VAD-fmk (25 μM) was added at the time of inoculation where indicated. After 72 h cells were fixed and immunostained for VZV IE63 (red) to assess virus spread. Images are representative from at least 2 independent replicates. Scale bars represent 400 μm.
VZV ORF20 interacts with RIPK3 and ZBP1 in cells, forming insoluble complexes with ZBP1
Herpesvirus inhibition of host RHIM protein signalling requires the direct interaction of host and viral proteins. For the viral inhibitors of necroptosis M45 and ICP6, both proteins interact directly with ZBP1 and RIPK3 through their respective RHIMs, an interaction that is ablated by a mutation of the core tetrad of the viral protein to AAAA [27, 38, 41]. We recently demonstrated that for M45, the interaction with RIPK3 or ZBP1 results in the generation of an insoluble amyloid structure containing both viral and host proteins, which acts to sequester the host protein and prevent downstream functions . We sought to determine whether ORF20 might interact with RIPK3 and ZBP1 in the same manner. 293T cells were co-transfected with plasmid constructs expressing either V5-tagged ORF20 or ORF20-RHIMmut and GFP or ZBP1-GFP or RIPK3-GFP, and interactions between the cellular and viral RHIM-containing proteins in both the soluble and insoluble fractions were investigated by immunoprecipitation (IP). Both ORF20 and ORF20-RHIMmut were found to interact with ZBP1 in the soluble (Fig 6A) and insoluble fractions (Fig 6B), indicating the formation of soluble complexes as well as insoluble supramolecular assemblies. The ability of the ORF20-RHIMmut to interact with ZBP1 was surprising but suggests that some residues within the RHIM, but outside of the core tetrad, can maintain an interaction between the proteins. Interaction between ORF20 and RIPK3, and ORF20-RHIMmut and RIPK3, was detected only in the soluble fraction derived from cell lysates (Fig 6C), despite multiple analyses of the insoluble fraction. We sought to examine interactions between ORF20 and RIPK1 by this method, and were unable to detect co-immunoprecipitation despite multiple attempts. We attempted to detect the interactions between ORF20 and both ZBP1 and RIPK3 during productive VZV infection of cells, but were unable to demonstrate these consistently, due to limitations of reagents available to perform these analyses.
Fig 6. ORF20 interacts with human RHIM proteins associated with cell death signalling.293T cells were transfected with ORF20 constructs and ZBP1-GFP or GFP alone. Cells were harvested and immunoprecipition (IP) performed on the soluble and insoluble fractions using GFP as bait. Western blot (IB) was performed on the immunoprecipitated and input cell lysates. (A) Immunopreciptiation of ZBP1-GFP in the soluble fraction of cell lysates. (B) Immunopreciptiation of ZBP1-GFP in the insoluble fraction of cell lysates. Each blot is representative of at least 2 independent biological replicates. (C) Immunopreciptiation of RIPK3-GFP in the soluble fraction of cell lysates. Each blot is representative of at least 2 independent biological replicates. Arrows indicate protein size markers.
The ORF20 RHIM supports the formation of amyloid fibrils
The RHIMs within host and viral proteins have been shown to be responsible for functional amyloid fibril assembly by these proteins [21, 39, 58]. Having discovered that ORF20 and ZBP1 interact with each other to form large insoluble heteromeric complexes, we sought to determine whether these assemblies have the distinctive β-sheet rich substructure characteristic of an amyloid fibril . The 114 N-terminal residues of ORF20, containing the RHIM, were expressed recombinantly in a fusion protein with His-tagged ubiquitin (Ub-ORF201-114). This construct spontaneously assembled into fibrils that exhibited the long, straight and unbranching morphology characteristic of amyloid fibrils (Fig 7A). A fusion construct containing the mCherry fluorophore (mCherry-ORF201-114), instead of ubiquitin, likewise exhibited typical amyloid morphology (Fig 7B). When incubated with the small-molecule amyloid sensors Thioflavin T (ThT) and Congo red, the presence of Ub-ORF201-114 fibrils gave rise to increased fluorescence emission at 485 nm with ThT and an increase in absorbance at 540 nm with Congo red, confirming the presence of a cross-β amyloid structure (Fig 7C and 7D). Protein assemblies with a fusion protein containing the AAAA mutation within the ORF20 RHIM (Ub-ORF201-114mut) displayed increased ThT emission at 485 nm and Congo red absorbance at 540 nm (Fig 7C and 7D), indicating that some cross-β substructure is present, however the rate of increase in ThT fluorescence was much slower than that for the WT RHIM construct. This construct also self-assembled into large amorphous aggregates that lacked a clear fibrillar structure (Fig 7E). Likewise, a mCherry-tagged ORF20 AAAA mutant protein (mCherry-ORF201-114mut) formed amorphous structures lacking a clear fibrillar morphology (Fig 7F). In summary, the ORF20 RHIM is able to form amyloid assemblies with characteristic morphology and structure and the AAAA mutation to the core tetrad impairs the formation of a regular, extended fibrillar structure.
Fig 7. The ORF20 RHIM forms homomeric amyloid structures.(A) Transmission electron microscopy image of structures formed by Ub-ORF201-114. Scale bar represents 200 nm. (B) Transmission electron microscopy image of structures formed by mCherry-ORF201-114. Scale bar represents 200 nm. (C) ThT fluorescence over time of His6-Ub-ORF20 constructs after dilution from 8 M urea into assembly buffer. Buffer sample contains equimolar ThT but no protein. Curves are derived from three independent replicates. Error bars indicate standard deviation. (D) Absorbance spectra of solutions containing Congo red and Ub-ORF201-114 and Ub-ORF201-114mut after dialysis against assembly buffer. Buffer refers to a Congo red sample in assembly buffer with no protein. (E) Transmission electron microscopy image of structures formed by Ub-ORF201-114mut. Scale bar represents 200 nm. (F) Transmission electron microscopy image of structures formed by mCherry-ORF201-114mut. Scale bar represents 200 nm.
The core tetrad of the ORF20 RHIM controls the size and morphology of complexes formed with RIPK3 and with ZBP1
Following the observed interactions between ORF20 and human RIPK3 and ZBP1 in cells, we further investigated the nature of complexes formed between these proteins in vitro. Single molecule fluorescence confocal spectroscopy was used to confirm interactions at a single molecule level. This method uses lasers focused on a very small confocal volume (250x250x800 nm) to detect fluorescent signals from two different proteins, labelled with two different fluorophores  (Fig 8A). Coincidence of fluorescent signals in mixtures of proteins at low concentration indicates the formation of hybrid complexes containing both protein partners. Fluorescently-labelled fusion proteins were prepared, with YPet or mCherry fused to host or viral proteins respectively. We observed large signals in the fluorescence intensity traces for ORF201-114, ORF201-114mut, RIPK3387-518 and ZBP1170-355 proteins when analysed alone, indicating homo-oligomer formation by these proteins (Fig 8B, 8C, 8D and 8G). When the ORF20 proteins were mixed with the human proteins under assembly competent conditions both ORF201-114 and ORF201-114mut were observed to form hetero-oligomers with RIPK3387-518 and ZBP1170-355, indicated by coincidence of fluorescent signals in both channels (Fig 8E, 8F, 8H and 8I). We also performed confocal spectroscopy experiments to analyse interactions between ORF20 and RIPK1 (S4A Fig). In accordance with our inability to detect an interaction between ORF20 and RIPK1 by co-immunoprecipitation, no interaction between ORF20-1-114-mCherry and the minimum region of RIPK1 that has been shown to be required to interact with RIPK3 (YPet-RIPK1497-583)  was detected. Likewise, no interaction between YPet-RIPK1497-583 and ORF201-114mut-mCherry was detected (S4A Fig). These data indicate that ORF20 and RIPK1 are unlikely to interact directly.
Fig 8. The VZV ORF20 RHIM forms heteromeric complexes with RHIMs from RIPK3 and ZBP1.(A) Schematic representation of fluorescence detection from dilute solutions containing two different fluorophores in a nanolitre confocal volume. Samples are excited simultaneously with two overlapping lasers and emission from the YPet and mCherry fluorophores is recorded separately. Coincident bursts in the two channels indicate formation of heteromeric complexes containing two different proteins (B-I). Representative fluorescent time traces collected from ORF201-114-mCherry, ORF201-114mut-mCherry, YPet-RIPK3387-518 or YPet-ZBP1170-355 fusion proteins, alone or mixed in pairs under conditions that allow co-assembly. The proteins present in each mixture are indicated for each part of the figure. Inserts show detail of 1 s dual fluorescence recording, indicated by star on the full time trace.
https://doi.org/10.1371/journal.ppat.1008473.g008Considering that we had previously established that the interactions between ZBP1 and ORF20 were the most relevant in a biological setting, we analysed the sizes of the heterocomplexes formed between these constructs. Photon counting histograms reflect the distribution of particle sizes detected in the confocal volume. Analysing the signal for mCherry from the single molecule traces, we determined the size distribution of ORF201-114 wild type and ORF201-114mut oligomers alone and in combination with ZBP1170-355. For ORF20 RHIM wild type, there was no difference in the size of particles formed by ORF201-114 with or without ZBP1170-355 (S5A Fig). This indicates that the ORF20 RHIM is capable of forming large molecular assemblies by itself and in concert with ZBP1. ORF201-114mut-mCherry formed smaller structures than its wild type counterpart (S5B Fig). Further, together ORF201-114mut-mCherry and ZBP1170-355 form structures smaller than the mutant ORF20 alone, and substantially smaller than the combination of wild type ORF20 and ZBP1, reflected in the histograms and in fluorescence correlation analysis of the single molecule traces (S5C Fig).
We also assessed interactions between ORF20 and mutant forms of YPet-ZBP1, where either the first RHIM of ZBP1 was mutated to AAAA (YPet-ZBP1170-355mutA) or the second RHIM was mutated to AAAA (YPet-ZBP1170-355mutB) (S4B Fig). Mutation of the first RHIM of ZBP1 reduces the ability of the protein to self-assemble into homomeric amyloid structures and little co-assembly of ORF20 with YPet-ZBP1170-355mutA was observed. Mutation of the second RHIM does not affect self-assembly and YPet-ZBP1170-355mutB interacted strongly with ORF20 (S4B Fig). These findings are in agreement with the reported effects of RHIM mutations on ZBP1 activity in cells, where the first RHIM is primarily associated with interaction and assembly of RHIM-based structures [27, 41, 43].
Mixtures of ORF20 and ZBP1 form heteromeric amyloid assemblies with fibrillar morphology
We recently proposed that M45 successfully inhibits host RHIM:RHIM interactions by trapping host proteins in alternative stable hybrid amyloid structures . In order to determine if a similar mechanism was occurring during ORF20-ZBP1 interactions, we examined mixtures of Ub-ORF201-114 and Ub-ORF201-114mut with YPet-ZBP1170-355 via electron microscopy (Fig 9A). YPet-ZBP1170-355 alone forms short, clustered fibrils