Junk mail: Viral envelopes promote zoonoses

Viruses display enormous genomic diversity, employing effectively all possible routes for the expression of genetic information and lacking universally shared genes (1). This diverse molecular biology translates into equally diverse ecology with nearly all cellular life bearing at least one viral symbiont, and in most cases, many. At once, residents from every neighborhood of the virosphere may be found to inhabit a single species (2), and just one virus may infect a great variety of hosts (3). Identifying viral host range is currently a major challenge of significant fundamental interest with clear implications for public health. Current standards of practice wildly vary across the virosphere from empirically measuring the sequence space of mosquito blood meals (4), to finessed computational structure analysis of coronavirus receptor homology (5), to the minimalist comparison of nucleotide composition among phages and bacteria (6).

Surprisingly, despite considerable efforts invested in the prediction of specific viral host identity from genomic sequencing, there is presently no consensus regarding what genomic features are associated with the “first-order approximation” of this question: what makes a virus a generalist? In PNAS, Valero-Rello and Sanjuan demonstrate that simply the presence of a major virus surface structure, an envelope, is a powerful predictor of broad host range (7). The work presented suggests enveloped viruses are more likely to infect multiple hosts, including a combination of nonhuman and human hosts, and are predicted to have a greater propensity for novel zoonoses.

While the boundaries of the virosphere are fuzzy (8), the genomes of most members encode for at least one major structural protein, which assembles to build the virus capsid. Characteristic, if not essential by definition, most viruses lacking capsid genes within the viral genome itself, still mature into encapsidated virions utilizing host structural proteins. Conversely, the presence of a viral envelope is highly variable. Both enveloped and nonenveloped viruses are dispersed throughout the Eukaryotic virosphere (bacteriophages are primarily nonenveloped), and both groups infect essentially the same scope of hosts. When present, viral envelopes principally define all virion surface interactions. This patchy heterogeneity of this prominent structural component has left many fundamental epidemiological and biophysical questions about viral envelopes unanswered at the time of this writing, presenting enormous opportunity for 21st century comparative genomics.

Valero-Rello and Sanjuan demonstrate that simply the presence of a major virus surface structure, an envelope, is a powerful predictor of broad host range.

Valero-Rello and Sanjuan focused on mammalian viruses made available through the Global Virome in One Network (VIRION) (9), an open-access database, which aims to provide a comprehensive overview of known vertebrate–virus associations, seeking to establish genomic features predictive of cross-species transmissibility and zoonotic propensity. While expanding sequencing efforts will likely change the landscape in the near future, at present there are few viruses, even among human pathogens (10), for which a robust multisequence genome alignment may be constructed. For many, only short, sparse sequence fragments are available. Coupled with a robust network of host-virus associations, however, even fragments may be used to establish host range through the inference of deeply conserved basic viral characteristics.

Valero-Rello and Sanjuan considered six such virus features: the viral genome size, whether the viral genome is encoded in RNA or DNA, is single or double stranded, and is segmented or nonsegmented as well as whether the virus replicates in the cytoplasm or the nucleus and, lastly, whether the virus is enveloped or nonenveloped (see Fig. 1). These features may be inferred whenever enough sequence information is available for viral family identification, a relatively low bar.

Fig. 1.

Cartoon of the six virus features considered: viral genome size (red), viral genome encoded in RNA vs. DNA (orange), viral genome is single vs. double stranded (yellow), viral genome is segmented vs. nonsegmented (green), virus replicates in the cytoplasm vs. nucleus (blue), and virus is enveloped vs. nonenveloped (purple).

The impact of these features on cross-species transmissibility and zoonotic propensity were then assessed. Cross-species transmissibility was estimated by two measures: host breadth, the number of host species in which a virus has been found excluding humans, and multihost status, a binary variable indicating whether a virus was found in multiple host species. Similarly, zoonotic propensity was simply reported by a binary variable indicating whether a virus was found in humans and at least one additional mammalian species. Statistical analysis (negative binomial regression for host breadth and binary logistic regression for multi-host status and zoonotic propensity) demonstrated that cytoplasmic replication, eliminating the requirement for nuclear-entry, was found to be significantly associated with increased cross-species transmissibility and zoonotic propensity, in agreement with previous findings (11, 12). Genome segmentation was also found to be significantly associated with increased zoonotic propensity but not (excluding human hosts) cross-species transmissibility. Segmentation may impact virus host generalization through the maintenance of viral diversity, enabling frequent recombination through segment reassortment (11, 13).

In contrast to contemporary expert opinion (14), however, Valero-Rello and Sanjuan demonstrated the presence of a viral envelope to be the feature most strongly associated with both cross-species transmission and zoonosis. While mechanistic understanding is currently limited, a few possibilities are proposed. As discussed above, host-generalizability is likely conferred by cytoplasmic replication by obviating the need for nuclear-entry. Similarly, viral envelopes may assist in cytoplasmic entry by presenting a flexible, malleable viral surface subject to fewer structural constraints than capsid proteins. Host receptor identity varies at surprisingly short evolutionary distances (13), through mechanisms which are, perhaps, more easily achieved for enveloped viruses. Envelopes may also serve to “disguise” virions, broadly evading host immune surveillance (15) and providing opportunities for alternative modes of cell entry, including apoptotic mimicry (16).

In addition to these mechanisms impacting intrahost dynamics, envelope biophysical properties impacting interhost transmission must also be taken into consideration. Virion environmental stability plays a critical role in the determination of virus transmissibility, but is surprisingly poorly understood as is perhaps most strikingly illustrated by the lack of consensus regarding what factors permit transmission via aerosol (17). Consequently, while previous work has demonstrated that in select environments, the presence of a viral envelope is predictive of reduced environmental stability (18), mechanistic studies emphasize that virus inactivation rate bears complex dependence on multiple physical parameters (19). Valero-Rello and Sanjuan additionally emphasize that viruses which frequently cross species barriers likely also frequently lead to limited outbreaks, which without mechanistic understanding, may result in the prediction of reduced cross-species transmissibility for virus features, which, in fact, promote host generalizability.

As the diversity of the sequenced virosphere deepens, new opportunities will be presented to reassess our understanding of biological properties inferred from statistical associations among a small subset of genomes largely limited to human pathogens of epidemiological importance today. In this work, Valero-Rello and Sanjuan make full use of newly available genomes and metadata, however imperfect, to perform an unbiased assessment of what viral characteristics are predicative of host generalizability. This approach will, in turn, promote model generalizability, a key step in making the leap from the explanation of prior, to the prediction, and hopefully prevention, of future pandemics.