Evolutionary clash between myxoma virus and rabbit PKR in Australia

The success of a virus in defeating or evading cell-intrinsic immune responses contributes to its virulence, pathogenesis, and host range. A central tenet of virology holds that virus and host coevolve as each adapts to survive. Rarely, however, do we have the chance to observe this principal at play in the real world. The deliberate and repeated release of myxoma virus (MYXV) to control feral European rabbits introduced into Australia provides one such exceptional opportunity. Myxoma is a member of the poxvirus family, though unlike its more famous cousin variola, the agent of small pox, is unable to infect humans and is instead limited to infection of rabbits and hares. Following the release of a MYXV reference strain into the wild that exhibited 99.8% fatality rates in laboratory rabbits, attenuated strains were recovered from the field that had begun to outcompete the virulent parental strain. In addition, the resistance of rabbits to MYXV increased (1). The underlying molecular mechanism of both rabbit resistance and MYXV attenuation, however, remained elusive. Armed with the recently compiled genome sequence of 21 myxoma field isolates, a new study by Peng et al. in PNAS reveals that genetic alterations to the MYXV M156 protein correlate with both MYXV host specificity and changes in virulence observed in the field (2). Remarkably, more than 50% of MYXV field isolates produce an M156 variant that is no longer able to antagonize the rabbit double-stranded RNA (dsRNA)-activated protein kinase PKR, a critical component of IFN-stimulated defenses that controls protein synthesis. This not only provides the first mechanistic explanation for each of these facets of myxoma biology, but also elegantly illustrates the coevolution of host innate defense and viral virulence factors since the first release of MYXV in 1950.

In response to virus infection, a key arm of antiviral host defenses acts to cripple the infected cell’s capacity to produce the polypeptides required for virus replication and spread. This is achieved by globally inhibiting the initiation of mRNA translation and is triggered by accumulation of dsRNA, a pathogen-associated molecular pattern produced by many different viruses during their replication cycle. Upon sensing dsRNA, host PKR, which resides in uninfected cells as an inactive, unphosphorylated monomer, becomes activated as a phosphorylated dimer bound to dsRNA (Fig. 1). The ensuing site-specific phosphorylation of the eukaryotic translation initiation factor eIF2 α-subunit prevents methionine-initiator tRNA charging of 40S small ribosome subunits, inhibiting the initiation of protein synthesis and effectively preventing virus replication (3). The extraordinary efforts and diverse molecular tactics many different viruses rely upon to counteract PKR and preserve the activity of the critical initiation factor eIF2 underscore the significant role PKR plays in host defense.

Fig. 1.

Myxoma virus strategies to antagonize the host dsRNA-activated protein kinase PKR. Upon infection with MYXV, dsRNA accumulates and activates PKR. Activation entails PKR dimer assembly on dsRNA followed by autophosphorylation (phosphorylation is depicted as a green P). Once activated in infected cells, PKR phosphorylates the α-subunit of the translation initiation factor eIF2 (eIF2α), inhibiting the initiation of protein synthesis and effectively restricting virus replication. MYXV produces two proteins, M029 and M156, to antagonize PKR and prevent eIF2α phosphorylation. By binding to dsRNA, M029 prevents PKR activation. In contrast, M156 is a rabbit PKR-specific pseudosubstrate (described in the text) that binds PKR in lieu of its bona fide substrate, eIF2α. An M156 variant that lost its ability to inhibit rabbit PKR has now been identified in Australian MYXV field isolates.

To safeguard eIF2 function and effectively contain PKR, viruses frequently enlist multiple molecular strategies. MYXV is no exception, producing two proteins, M029 and M156 that are orthologs of Vaccinia virus (VACV) E3 and K3, and which inhibit the activation and activity of PKR (Fig. 1). Although VACV was used as the smallpox vaccine and is the best-studied proto-typical poxvirus, its precise origins and natural hosts remain uncertain, limiting our understanding of how E3 and K3 interact with PKR in a genuine biological host. MXYV proteins, in contrast, are readily studied in naturally infected rabbits, offering physiological insights into how host PKR interacts with viral antagonists. Both MYXV M029 and VACV E3 inhibit PKR activation by binding dsRNA, thereby precluding its detection by PKR. MYXV lacking M029 is severely attenuated, does not cause disease in inoculated rabbits, and cannot replicate in cultured cells unless PKR is depleted (4). Relying on a distinct mechanism, VACV K3 and MYXV M156 both display homology to eIF2α. Indeed, K3 acts as a PKR pseudosubstrate that binds PKR but cannot be phosphorylated, effectively diverting PKR from its bona fide substrate eIF2α, preserving eIF2 function and allowing infected cell protein synthesis to proceed (5–7). However, although MYXV M156 was predicted to function similarly to K3, previous studies surprisingly revealed that M156 was unable to inhibit human PKR in a yeast-based assay and its importance in vivo remained unclear (5). An important clue came from findings that VACV K3 inhibited PKR in a species-specific manner and may be a host-range determinant (8, 9). Now Peng et al. (2) establish that M156 is important for MYXV productive growth in rabbit cells and its deletion further impairs replication of the M029 mutant. The authors further show, using both a heterologous yeast assay and transfected human cells, that M156 can only suppress rabbit PKR activity but not that of other mammals, mirroring the host range of MYXV (2). Thus, the specificity of M156 for rabbit PKR likely contributes to MYXV host range and its inability to infect humans, particularly given that its other characterized PKR-antagonist, the dsRNA binding protein M029, can antagonize human PKR (4).

Although the original MYXV release in Australia in 1950 resulted initially in exceptionally high lethality, causing a precipitous drop in rabbit numbers by nearly 95% in some regions, its virulence waned within 5 y. Comparison of stocks of the original released virus with wild isolates showed clear attenuation; newer strains were associated with a more prolonged illness, potentially benefitting MYXV transmission, which involves mosquitos and other biting arthropods (10). However, the underlying molecular basis of MYXV attenuation has hitherto remained unclear. Although the recent comparison of wild Australian MYXV isolates from the 1990s with the experimental strain released more than 50 y ago identified many changes in the MYXV genome (11), identifying which of these are associated with the observed attenuation of virulence posed a significant challenge. Peng et al. now show that one such mutation in M156 (L98P) abolishes its ability to inhibit PKR and cannot rescue replication of a M156-deficient MYXV in rabbit cells (2). Remarkably, this mutation lies within the region predicted to interact with PKR and is present in more than half of wild viral samples tested (11), supporting the notion that it is a determinant of attenuation.

How M156 deficiency and the L98P variant affect MYXV virulence in rabbits is an important next step. Duplications of the M156 gene were also identified in several wild myxoma isolates (11), perhaps foreshadowing the appearance of better-adapted variants. In support of this possibility, genomic duplications involving the K3L gene were recovered in laboratory evolution studies using VACV, accelerating the emergence of adaptive variants better able to counteract PKR (12). Besides M156 and M029, ancillary MYXV factors may also control eIF2α phosphorylation. In addition to E3 and K3, VACV mRNA decapping enzymes (D9, D10) accelerate mRNA decay, thus limiting dsRNA accumulation, PKR activation, and contributing to virulence in vivo (13, 14). Although MYXV encodes D9 and D10 orthologs (15), their contribution to regulating PKR and eIF2α phosphorylation remains unexplored. Furthermore, the molecular basis underlying the resistance of rabbits to MYXV is likewise unknown. Testing wild rabbits over multiple generations with a MYXV laboratory strain in the late 1950s indicated that host resistance contributed to diminishing lethality (1). As the subject of such fierce selective pressure from viral antagonists, vertebrate PKR has evolved more rapidly than the bulk of the host genome, particularly within the region near the eIF2α binding site where the host cell must attempt to differentiate between its natural substrate and a viral impostor (8, 9). Perhaps the arms race with MYXV has left its mark on PKR in Australian rabbits or indeed genes for other host proteins, such as mRNA decay factors that can regulate the detection of dsRNA in poxviruses (14).

The findings of Peng et al. (2) highlight the importance of PKR inhibition as a major determinant of viral virulence and host specificity in cell culture, and point toward its likely role in the wild as well. Moreover, their functional characterization of M156 as an inhibitor only of rabbit PKR should sound a cautionary note to those studying immune-modulatory viral functions to select their experimental systems wisely or risk missing fundamental physiological activities.