Disruption of Host Antiviral Resistances by Gammaherpesvirus Tegument Proteins with Homology to the FGARAT Purine Biosynthesis Enzyme

Abstract

All known gammaherpesviruses encode at least one conserved tegument protein that contains sequence homology to the cellular purine biosynthesis enzyme: phosphoribosylformylglycineamide amidotransferase (FGARAT, or PFAS). While no enzymatic activity have been found on these viral FGARAT-homology proteins (vFGARAT), they are important for disarming host intrinsic antiviral machinery. Most vFGARAT proteins disrupt the intrinsic antiviral response -associated cellular subnuclear structure: ProMyelocytic Leukemia (PML) associated nuclear body (PML-NB). vFGARATs from different viruses target different components of PML-NB to prevent cellular repression of viral infection. In addition, vFGARATs of rhadinoviruses were recently found to oligomerize with the cellular FGARAT to deamidate RIG-I and repress inflammatory cytokine production. In this review we discuss the diverse mechanisms of antiviral response disruption by gammaherpesvirus vFGARATs and the significance of the enzyme homology domain.

Introduction

Herpesvirus tegument proteins constitute a major component of the viral particle, and are important for multiple steps of the viral infection life cycle. Delivered pre-made into the host cell, tegument proteins are well positioned to regulate the intracellular infection environment immediately after entry. Members of the gammaherpesvirus subfamily encode tegument proteins that contain conserved domains of sequence homology with the cellular purine biosynthesis enzyme phosphoribosylformylglycineamide amidotransferase (FGARAT, also called PFAS). Members of this gene family include the Epstein-Barr virus (EBV) major tegument protein BNRF1, Kaposi’s sarcoma associated herpesvirus (KSHV) ORF75, murine gammaherpesvirus 68 (MHV68 or MuHV4) genes ORF75a, ORF75b, and ORF75c (or viral glutamine amidotransferase, vGAT); and herpesvirus saimiri (HVS) ORF3, collectively referred to as viral FGARATs (vFGARATs). To date, efforts to elucidate the functions of vFGARATs have yet to reveal any enzymatic activity or interactions related to purine biosynthesis or metabolism. Instead, these tegument proteins disrupt various PML-NB-associated innate antiviral resistance pathways. Below, we discuss the importance of these tegument proteins in the herpesvirus life cycle, what is known about the FGARAT enzyme, the different ways that vFGARATs and other herpesviral tegument proteins disrupt innate antiviral resistance pathways, and what vFGARATs may teach us about the gammaherpesvirus infection strategy.

Tegument proteins in herpesviral biology

Herpesviruses package a collection of tegument proteins in the viral particle. Tegument proteins that are attached to the nuclear capsid are generally associated with the transport of viral capsid to the cell nucleus; while other tegument proteins can diffuse into the cytosol upon cell entry and regulate host processes during early infection, including the inhibition of several branches of intrinsic immunity pathways. Host cells possess multiple forms of intrinsic antiviral resistances that may abrogate early infection, including interferon induction, inflammasome activation, and heterochromatin assembly [1]. Herpesvirus tegument proteins have been well characterized for their ability to target and disarm these host intrinsic defenses. For example, EBV LF2 and KSHV ORF45 targets interferon regulator IRF7, to block IRF7 transactivation of type I interferon genes [2,3]. KSHV tegument protein ORF63, interacts with the host detector of viral nucleotides NLRP1, preventing its association with active components of inflammasomes, attenuating production of pyroptosis-associated cytokines IL-1β and IL-18 [4]. HSV-1 VP22 interacts with the cellular chromatin remodeler TAF-1, preventing TAF-1-associated nucleosome assembly [5]. These examples demonstrate the numerous cellular pathways targeted by viral tegument proteins during the early stages of viral infection to regulate antiviral resistances and ensure a productive viral infection.

Purine biosynthesis homology in the gammaherpesvirus tegument

All gammaherpesviruses encode at least one essential tegument protein that contains extensive homology to the cellular enzyme formylglycineamide ribonucleotide aminotransferase (FGARAT). As this homology is not present on any alpha- or beta-herpesviruses tegument proteins, FGARAT-homology tegument proteins likely provide a function uniquely important to the gammaherpesvirus life cycle. Cellular FGARATs catalyze the fourth step in the ten-step de novo purine biosynthesis pathway, a pathway that synthesizes the purine precursor inosine 5'-monophosphate (IMP) from phosphoribosyl pyrophosphate (PRPP) (Figure 1). FGARAT hydrolyses ATP to power the transfer of an amino group (NH₃⁺) from glutamine to formylglycineamide ribonucleotide (FGAR), forming the intermediate product formylglycineamidine ribonucleotide (FGAM) [6]. The product FGAM is also called phosphoribosylformylglycinamidine, thus the enzyme FGARAT is alternatively named phosphoribosylformylglycinamidine synthetase (PFAS). FGARAT is highly conserved across prokaryotes and eukaryotes, alternatively named PurL in bacteria. The solved crystal structure of the Salmonella typhimurium FGARAT (StPurL) provides the best insight into the structure of this enzyme [7]. StPurL consists of a four-domain structure: 1) an N-terminal domain of unclear function, 2) a linker region, 3) the FGAM synthetase domain, and 4) a C-terminal glutaminase (GATase1) domain (Figure 2). The FGAM synthetase domain has a pseudo-2-fold symmetry structure of two aminoimidazole ribonucleotide synthetase (AIR_S/PurM) homology domains, with the first half (PurM1) containing the FGAM synthetase active site, and a catalytically inactive second half (PurM2) of similar structure that holds an ADP for functions that are not yet understood. The C-terminal GATase1 domain removes the amine group from glutamine and channels it to the FGAR substrate in the PurM1 active site.

Figure 1. The de novo purine biosynthesis pathway.

The synthesis of purines involves a multi-step pathway that builds the purine ring on top of the pentose ring of phosphoribosyl pyrophosphate (PRPP). In the fourth step, FGAR aminotransferase (FGARAT, or PurL in prokaryotes) catalyzes the transfer of an amino group from glutamine (Gln) on to the pre-formed ring on formylglycineamide ribonucleotide (FGAR), resulting in the formation of formylglycineamidine ribonucleotide (FGAM). The aminotransferase domain of FGARAT shares structural homology with that of the fifth step enzyme AIR Synthetase (PurM in prokaryotes), which closes the ring of FGAM to form 5-Aminoimidazole ribonucleotide (AIR). This multi-step pathway (10 steps in eukaryotes, 11 steps in other organisms) ultimately results in the complete purine ring on the purine precursor: Inosinate (IMP). In the middle steps, R denotes the 5-phospho-D-ribosyl group. Figure adapted from [61].

Figure 2. Domain alignment diagram of viral and cellular FGARATs.

Alignment of a vFGARAT (EBV BNRF1) with the known domain structure of cellular FGARAT (Salmonella typhimurium PurL, StPurL). All vFGARATs maintain the second PurM and GATase1 domains conserved. In the case of EBV BNRF1, the domain that aligns to the first PurM (FGAM synthetase active site) of FGARAT is responsible for Daxx interaction [46]. StPurL domain information adapted from [7].

A BLAST alignment of EBV BNRF1 against Homo sapiens FGARAT and Salmonella typhimurium StPurL showed 32% and 24% identity, respectively. All viral vFGARATs are of similar size with human FGARAT and StPurL at approximately 1300aa. Viral-host homologies are most conserved at the FGAM synthetase subdomain and the C-terminal glutaminase (GATase1) domain (702–954aa and 1040–1306aa in BNRF1 coordinates, Figures 2 and 3). In other domains, the homology can be aligned sporadically across the entire protein, suggesting that the overall protein structure is conserved. Calculating the phylogenic tree of cellular and viral FGARAT homologues (Figure 4) from the alignment in Figure 3, shows that all viral vFGARATs can be grouped into a cluster that is further diverged from the cellular (human and S. typhimurium) FGARATs, suggesting that the host FGARAT gene was copied by an early common ancestor of gammaherpesviruses. Since then, vFGARATs may have divergently evolved from cellular FGARAT/PurL, while retaining the overall structural fold (Figure 5).

Figure 3. Sequence alignment of vFGARATs to human FGARAT and StPurL.

Amino acid alignment of known viral FGARAT-homologues to human FGARAT and Salmonella typhimurium PurL (StPurL). Alignment done with Clustal Omega program (ver. 1.2.1) hosted at The European Bioinformatics Institute (http://www.ebi.ac.uk/Tools/msa/clustalo/) [62]. Only highly homologues regions shown, regions aligning to amino acids 246–321 of BNRF1 shown in (A); while 518aa to the end shown in (B). Completely matched residues denoted by *. Putative ATPase and Glutamine binding domains as identified in [8] are marked with an underline. Residues that form hydrogen bonds with ADP at the auxiliary ADP site as listed on StPurL [7] are marked in green; while the Glutamine interacting residues of StPurL GATase1 domain are marked in red.

Figure 4. Phylogenic tree of FGARAT homology proteins.

The protein sequence alignment in Fig 3 was used to calculate the divergent distances between viral and cellular FGARAT homologues using the Phylip v3.696 software package, using the Probability Matrix from Blocks (PMB) model to calculate distances and the neighbor-joining method to calculate the distance matrix.

Figure 5. Computational Model of BNRF1 3D Structure.

BNRF1 predicted structure was modeled on the existing FGARAT structure from S. typhimurum (PDB 1T3T) using Raptor X webserver (http://raptorx.uchicago.edu). S. typhimurium FGARAT (left), EBV BNRF1 (right). The amino acids 300–600 containing the BNRF1 Daxx-Interaction Domain (DID) is highlighted in red.

Despite this overall homology, key residues of the conserved ATP and glutamine - binding motifs of cellular FGARAT [8] are missing on BNRF1, as is the case in MHV68 ORF75c [9]. However, an ADP-holding pocket in the second PurM domain of StPurL is highly conserved in vFGARATs (Figure 3). Specifically, all of the residues of StPurL that form hydrogen bonds with ADP [7] in the auxiliary ADP site (K649, E718, N722, D884, D887) can be found in BNRF1 (K627, E729, N733, D883, D886), while most of the magnesium ion ligands that stabilize ADP are also conserved on BNRF1. These ADP interacting residues are conserved in human FGARAT and all vFGARATs that we examined by sequence alignment. In the glutaminase domain, only two (G1093, Q1139 of StPurL) of the five glutamine-interacting residues are conserved in BNRF1 (G1089, Q1137); the conserved glycine and glutamine residues could also be found in all vFGARATs except HVS ORF3. Thus, vFGARATs may have lost key FGARAT catalytic residues in the FGAM synthetase and GATase1 domains, yet retaining the ADP-holding site.

Diverse Functions of viral FGARAT homologues

The imperfect conservation of catalytic residues suggests that the viral proteins have lost or modified the enzymatic activities of cellular FGARAT. Indeed, to date there are no reports of intrinsic enzymatic activity associated with vFGARAT protein. Instead, a recent study found that MHV68 ORF75c functions as a pseudo-enzyme through oligomerization with cellular FGARAT to bind and deamidate the antiviral response protein RIG-I [10]. Prior to this study, almost all gammaherpesvirus vFGARATs were reported to interact with components of PML nuclear bodies and abrogate their antiviral activities. Thus, it remains possible that vFGARATs have additional targets and functions in the viral life cycle (summarized in Figure 6).

Figure 6. Functions of vFGARATs and herpesviral PML-NB disrupters.

Gammaherpesvirus vFGARAT functions include RIG-I deamidation and the disruption of PML-NB components. Examples of alpha and beta-herpesviruses disruption of PML-NBs are shown below for comparison. Black circles denote viral proteins while red blocks denote cellular proteins. Proteins induced for degradation are drawn as slashed and grayed out.

* PML and Sp100 dispersion is only observed upon KSHV ORF75 overexpression, but not upon viral infection [44].

** HCMV IE1 specifically induces degradation of SUMO-modified isoforms of Sp100; while dispersing other isoforms of Sp100 [22,63].

PML-NBs and antiviral resistances

The majority of vFGARAT studies reveal a common function of disrupting components of the PML nuclear bodies (PML-NBs, also named ND10). PML-NBs are dynamic nuclear structures [11], where the PML protein forms a cage-like shell surrounding associated components [12,13], including Sp100, Daxx, ATRX, and HP1. PML-NBs impact diverse cellular functions linked to gene expression, telomere regulation, chromatin remodeling, apoptosis, post-translational protein modifications, and antiviral resistance [14–16].

Numerous experiments indicate that PML-NB components confer antiviral resistance. PML-NBs increase in size and number in response to viral infection [17,18], and viral-associated interferon induction increases expression of PML [19] and Sp100 [20]. Studies from HSV-1 and HCMV indicate that viral genomes localize to PML-NBs after nuclear entry [21] [22], suggesting that PML-NBs act as a site where host antiviral resistance effectors confront incoming viral genomes. PML overexpression inhibits VZV infectivity [23], PML and Sp100 knockdown rescues HSV-1 virus lacking ICP0 [24], and both PML and Daxx knockdown results in increased HCMV replication [25]. Consistent with its antiviral function, PML has been found to form spherical cage-like structures that interact with the VZV capsid surface protein ORF23, entrapping VZV nuclear capsids in PML-oligomeric enclosures in the host cell nuclease, preventing nuclear egress and infectious virion production [23,26].

Viral-associated disruption of PML-NBs was first described for alpha- and beta-herpesvirus proteins that are not homologues to vFGARATs (summarized in Figure 6). HSV-1 protein ICP0 is an immediate early gene that is also found in small amounts in the tegument [27,28]. ICP0 prevents PML and Sp100 from forming nuclear bodies [29,30], through inducing degradation of PML protein and certain isoforms of Sp100 [31]. ICP0-null mutant virus is deficient in viral gene expression while knockdown of PML and Sp100 resulted in a partial rescue of viral gene expression [24]. The VZV-encoded ICP0 ortholog, ORF61, disperses both PML and Sp100 away from nuclear bodies [23,32]. Interestingly, this disruption is less effective in human cell xenografts in mice, where PML entrapment of VZV nucleocapsids is dominant [23,26]. HCMV targets the PML-NB-associated transcription co-repressor Daxx through it’s tegument protein pp71. Upon infection of permissive cell types, pp71 localizes to PML-NBs [33], binds Daxx [34], induces Daxx degradation [35], and disperses the Daxx-interaction partner ATRX away from PML-NBs [36]. Daxx knockdown resulted in increased immediate early (IE) gene expression along with active histone markers on the viral genome (presence of H4 acetylation, absence of H3K9 di-methylation); while Daxx over-expressing cells are refractory to HCMV IE gene expression [37], emphasizing the role of Daxx in the repression of viral gene expression.

PML disruption by viral FGARAT-homology tegument proteins

MHV68 belongs to the rhadinovirus genus of gammaherpesviruses, along with the squirrel monkey virus HVS and human virus KSHV. MHV68 encodes three FGARAT homologues, ORF75a, ORF75b, and ORF75c, with ORF75c essential for viral infection in fibroblasts [38]. ORF75c can induce PML degradation (but not Daxx) in fibroblasts [9], which was attributed to its inherent ubiquitin-E3 ligase activity [39]. ORF75c protein may be multifunctional as ORF75c knockout virus also showed reduced transport of viral capsids into the nucleus [40]. Intriguingly, a recent report found another MHV68 gene ORF61, a ribonucleotide reductase (RNR) large subunit homolog, that reorganizes PML-NBs into track-like structures [41].

The squirrel monkey rhadinovirus HVS, encodes two vFGARATs: ORF3 and ORF75 [42]. Studies of HVS infection in both human and rhesus macaque fibroblasts show that. ORF3, but not ORF75, can induce proteosomal degradation of Sp100, while sparing PML and Daxx [43]. Knockdown of Sp100 partially rescues ORF3-null virus infection, while PML knockdown resulted in a stronger rescue. This suggests that while Sp100 can repress viral infection, the structural support by PML formation of nuclear bodies may aid in localizing Sp100 onto viral DNA.

The human rhadinovirus KSHV encodes one vFGARAT homologue, ORF75. A recent study showed that ORF75 can induce the degradation of ATRX, and the dispersion of Daxx, PML, and Sp100 away from nuclear bodies [44]. Knockdown of PML and Sp100 resulted in increased KSHV infection in fibroblasts. Degradation of ATRX could be induced by tegument-delivered ORF75 as de novo viral gene expression was not needed. In contrast, dispersion of PML and Sp100 was only observed with transfected ORF75, but not during viral infection. These findings suggest that KSHV ORF75 may affect PML-NB components differently at distinct stages of viral infection.

EBV encodes a single vFGARAT termed BNRF1, which is essential for the early infection of B-lymphocytes by EBV. BNRF1 null virus (ΔBNRF1) is incapable of viral gene expression and induction of B cell proliferation, although capable of lytic replication in transfected 293 cells [45]. Unlike rhadinovirus vFGARATs, BNRF1 did not induce detectable degradation of any PML-components. Rather, BNRF1 interacts with Daxx at the PML-NBs, and disperse ATRX away from Daxx and the PML nuclear bodies [46]. The Daxx-interaction domain of BNRF1 (360–600aa) [46] spatially maps to the first PurM domain of StPurL, overlapping with the putative FGAR synthetase active site. This suggests that the FGAR substrate holding domain may have divergently evolved to hold Daxx instead.

The complex of Daxx and ATRX was recently reported to form a histone variant H3.3 chaperone [47,48]. Using FRAP to measure histone mobility, BNRF1 was found to increase the free histone H3.3 pool in the nucleus, and thus have histone chaperone-like activity [49]. These data suggest a model where BNRF1 displaces ATRX to form an alternative histone chaperone complex of BNRF1/Daxx/H3.3/H4, decreasing the deposition of repressive chromatin on viral genomes [49]. This ensures the presence of permissive chromatin on the first activated viral promoter Wp, permitting viral gene expression during the earliest phase of nuclear infection. Thus, BNRF1 functions to disarm a form of antiviral repression-associated with viral chromatin assembly.

vFGARAT-regulation of the cellular RIG-I antiviral response pathway

While most gammaherpesvirus vFGARATs converge on functions involving the disruption of PML-NBs, it is not clear how this relates to FGARAT. A recent study provides the first report of a vFGARAT recruiting the enzymatic activity of cellular FGARAT to regulate the cellular antiviral gene RIG-I [10].

RIG-I belongs to the RIG-I-like receptor (RLR) family of pattern recognition receptors (PRRs), which recognizes viral RNA in the cytosolic compartment and trigger the activation of innate immune responses [50,51]. Upon binding viral RNA, RIG-I activates the downstream adaptor MAVS/Cardif/IPS-1/VISA[52,53]. Through downstream signaling cascades, RLR-MAVS signaling activates transcription factors such as interferon-regulator factor 3 and 7 (IRF3/7) and NF-κB [54], to induce transcription of interferons and inflammatory cytokines. While RIG-I is an RNA sensor, it may also respond to DNA virus infection. The EBV encoded EBER family of non-coding RNAs was reported to interact with and activate RIG-I signaling, activating NF-κB and inducing interferon production [55]. RIG-I is also important for the induction of interferons upon KSHV and HSV-2 infection [56,57], while the recognized viral ligand is yet unknown in both cases.

In the case of MHV68, He et al. reports that the vFGARAT ORF75c (referred to as viral glutamine amidotransferase, or vGAT by the authors) directly activates RIG-I by recruiting cellular FGARAT to deamidate RIG-I [10]. ORF75c acts as a “pseudo-enzyme” that does not have enzymatic activity, yet utilizes its potential structural homology to oligomerize with the host FGARAT enzyme, thus recruiting cellular amidotransferase activity to deamidate and activate the host antiviral component RIG-I. FGARAT-mediated deamidation removes amine groups from three specific amino acids on RIG-I, resulting in Q10E, N245D, and N445D editing of RIG-I. ORF75c effectively alters the substrate specificity of cellular FGARAT (PFAS) from glutamine to the ORF75c bound RIG-I. ORF75c-induced RIG-I deamidation coincides with RIG-I activation signs such as K63-linked poly-ubiquitination on RIG-I, RIG-I oligomerization, and RIG-I interaction with the downstream adaptor MAVS. This signaling cascade results in activation of NF-κB. Knockout of RIG-I or drug-mediated inhibition of FGARAT enzymatic activity results in loss of NF-κB activation, suggesting that RIG-I and cellular FGARAT are both important for ORF75c induction of NF-κB activation.

Although NF-kB activation is associated with inflammatory and anti-viral responses, the ORF75c-induced deamidation of RIG-I negates inflammatory cytokine production. Knockdown of RIG-I, or FGARAT leads to an increase of inflammatory cytokines such as CCL5 and IL-6 in MHV68-infected cells. He et al. also tested a mutant construct, RIG-I-TD, where all three deamidation targets are pre-mutated into the deamidation product amino acids. RIG-I-TD is more highly K63-linked poly-ubiquitinatinated and oligomerized then wild type RIG-I, supporting deamidation as an alternative route of RIG-I activation. All data considered, RIG-I deamidation, RIG-I activation, and NF-κB activation are all required for MHV68 ORF75c to abrogate inflammatory cytokine production. While NF-κB may also activate antiviral pathways, MHV68 lytic transactivator Rta can induce degradation of the RelA component of NF-κB to dampen the inflammatory response [58]. RIG-I deamidation and NF-κB activation is not unique to MHV68. He et al. reported that exogenous expression of KSHV ORF75 or HSV-1 infection can also induce deamidation of RIG-I and activation of NF-κB, while EBV BNRF1 does not [10]. This confirms a previous screen for KSHV-encoded NF-κB-activators, which found ORF75 as the top activator of NF-κB [59]. These studies highlight some of the complex confrontations between virus and host innate antiviral response.

Ending Remarks

Our knowledge of vFGARAT tegument proteins has increased profoundly in the last few years, yielding many new surprises. Almost all viral FGARAT homologues can disrupt PML-NBs, while the vFGARATs of KSHV and MHV68 can induce FGARAT-dependent deamidation of RIG-I. While structural homology with FGARAT is likely needed for MHV68 ORF75c to oligomerize with and recruit FGARAT, it is less clear why vFGARATs are utilized by almost all gammaherpesviruses for PML-NB disruption. vFGARAT proteins may suggest a biochemical link between PML-NB-associated anti-viral resistance and nucleotide metabolism. In support of this speculation, the MHV68 ribonucleotide reductase ORF61 involved in nucleotide metabolism can also target PML for degradation [41]. However, as alpha- and beta-herpesviruses do not utilize FGARAT-homology proteins to disrupt PML-NBs, PML-associated antiviral resistance may be linked to purine metabolism in a gammaherpesvirus-specific host cell type, such as B-lymphocytes.

There are significant functional and structural differences between vFGARATs of rhadinoviruses (KSHV, MHV68, and HVS) and lymphocryptovirus (EBV). First, deamidation of RIG-I is likely a function unique to rhadinoviruses as this was only found with MHV68 ORF75c and KSHV ORF75; but not EBV BNRF1 [10]. However, it is yet unknown if EBV BNRF1 can interact with cellular FGARAT and potentially deamidate an alternative target. Secondly, the vFGARATs of rhadinoviruses all induced degradation of at least one PML component; however there is no evidence of lymphocryptovirus vFGARAT (EBV BNRF1) inducing any protein degradation. Thirdly, while tegument-delivered BNRF1 supports early infection [45], KSHV ORF75 is de novo produced through out early infection, latency, and expressed at higher levels upon lytic reactivation, suggesting that functions of ORF75 are not limited to early infection [44]. He et al. used ORF75c-deleted MHV68 where the deletion virus was produced in ORF75c-stabily expressing cells [10]. The ΔORF75c virus has the ORF75c gene deleted from the viral genome, yet most likely packaged complementing ORF75c protein in the tegument. These deletion-complemented virus was shown to be defective in their ability to induce deamidation of RIG-I [10], suggesting that tegument-delivered ORF75c was not sufficient for this function, de novo produced ORF75c is required. This is in contrast to EBV BNRF1, where complementation of BNRF1 protein in the tegument can rescue viral gene expression [46]. These differences may point to important variations between the lymphocryptovirus and rhadinovirus families in life cycle, viral gene regulation, and their repression of cellular antiviral processes.

Much remains to be discovered with vFGARATs, and why they have acquired and retained the structural homology of a complex host purine biosynthetic enzyme. The recent discovery of MHV68 ORF75c recruiting cellular FGARAT enzymatic function provides a new paradigm for viral modulation of host processes. Viral mimics of host genes are always fascinating and typically instructive [60]. Further studies of the vFGARATs have potential to reveal new mechanistic links between nucleotide metabolism and gene expression, how these are linked to viral infection, and how viruses can reprogram these cellular processes.

Highlights.

1. Viral encoded enzyme-like proteins function to disarm host antiviral resistance

2. Viral FGARATs interact with and alter host PML-NB components

3. Viral FGARATs link purine biosynthesis with anti-viral resistance