Intercellular Transmission of Viral Populations with Vesicles

During cell-to-cell transmission, viruses are largely thought to behave as discrete infectious units (1). While there are some known exceptions to this, such as vaccinia virus interference of superinfection by other vaccinia particles (2), investigators generally have accepted that viral particles enter and exit cells independently of one another. Upending this view, we recently discovered that enteroviruses could travel between cells, not only as independent viral particles but also as clusters of viral particles (3). We reported that members of the enterovirus family of positive-strand RNA viruses, including poliovirus, coxsackievirus, and rhinovirus, all are released nonlytically from cells in culture within large vesicles that contained up to several hundred viral particles. These vesicles then facilitated virus spread to other susceptible cells by collectively transferring multiple viral genomes into the cytoplasm (3).

Enteroviruses are categorized as nonenveloped viruses in that they do not have membranes surrounding their capsids. Consequently, it has been widely accepted that they exit cells via lysis of the plasma membrane. Over the past 25 years, however, several reports have suggested that these viruses can be released from cells through nonlytic mechanisms. For example, poliovirus can be released in a polarized fashion from enterocytes, and blocking cellular autophagy pathways, typically a degradative cellular mechanism to cope with stress, inhibits poliovirus release (4, 5). While X-ray crystallographic structures or electron micrographs of isolated enteroviral particles never revealed the presence of membranes around capsids, it is worth noting that these preparations mostly relied on detergents, freeze-thawing, or other membrane-disrupting methodologies during virus purification. In recent years, using largely nonintrusive imaging-based techniques that maintain membrane integrity, we and others have found that many of these so-called nonenveloped viruses, including hepatitis A virus, hepatitis E virus, bluetongue virus, coxsackievirus, rhinovirus, and poliovirus, all are released from cells within membrane-bound organelles (3, 6–10). These studies also revealed that the majority of the intracellular virus populations were released prior to cell lysis (3). Intriguingly, cell lysis during nonenveloped viral infections appears to be cell line dependent and may have more to do with cellular physiology and the cellular response to viral stress than an actual mechanism for virus exit (4, 11). Moreover, cell lysis within the body can trigger inflammatory immune responses (12) and may not be beneficial for the viruses. Clearly it will be critical to determine the contribution of cell lysis to viral exit in vivo.

Enteroviral RNA genomes replicate and package into capsids on the cytoplasmic leaflets of phosphatidylinositol 4 phosphate (PI4P)/cholesterol-enriched endoplasmic reticulum (ER) platforms (3, 13, 14). We found that autophagosomal membranes capture these capsids at a time coinciding with a decrease in viral RNA synthesis. The ER is likely the primary source of these membranes, as the extracellular virus-encapsulating vesicles contain resident transmembrane ER proteins as well as autophagosomal machinery (3). The mechanisms underlying capture of viral capsids by autophagosomes is unclear. One possibility is that there are yet-to-be-discovered capsid-associated signals recognized by autophagosomal machinery. Alternatively, intercapsid interactions taking place on the ER could, over time, induce the ER membrane to encapsulate the capsids. Deserno and colleagues (21) have shown that membrane curvature-inducing protein assemblies, including viral capsids docked on a membrane, can experience attractive interactions that drive their clustering concurrently with large-scale membrane curvature changes (21). We found that unlike canonical autophagosomes generated during times of metabolic stress, enterovirus-capturing autophagosomes do not fuse with lysosomes. Instead, they traffic to the cell periphery where their outer membranes fuse with the plasma membrane, and their inner membranes, carrying multiple enteroviral particles, are released to the extracellular side. The reason these autophagosomes do not fuse with lysosomes is unknown, but they lack at least some of the SNARE machinery, including syntaxin 17, which is required for this fusion process (3).

We found that the enterovirus containing extracellular vesicles were highly infectious. Using single-molecule RNA fluorescence in situ hybridization (FISH), we showed that infection with vesicles containing viral particles enabled multiple viral RNA molecules to be simultaneously transferred into the cytoplasm of a new host cell prior to the start of replication (3). This transfer, which appeared not to take place by simple fusion of vesicles with the host cell plasma membrane as virus receptor-neutralizing antibodies, still could block infection (3). Thus, the membrane surrounding the viral particles must be either completely or partially disrupted to allow receptor engagement. Whether this membrane is disrupted at the cell surface in an incompletely enclosed plasma membrane invagination or in an intracellular macropinosome remains to be determined.

Surprisingly, we observed significantly higher infection efficiencies, as measured by the total amount of replication (3), when cells were infected with viral particles within vesicles as opposed to equivalent titers of free viral particles. Upon closer examination with single-molecule RNA FISH, we found that when cells were infected with free viral particles, the viral RNA genomes were dispersed throughout the cell (Fig. 1A and B) and there was a steep drop in the number of viral genomes per infected cell prior to replication, as the free viral particle titer added to the cells was diluted. Even at high free virus titers added to cells, while there were many more viral RNA genomes per infected cell, they were spatially segregated from one another in the cytoplasm, consistent with viruses behaving independently during entry (3). In contrast, when infections were carried out with equivalent titers of viral particles but which were within vesicles when added to the cells, there was a significantly smaller decrease in the number of viral genomes per infected cell prior to replication, as viral titers were diluted. Moreover, by this mode of infection, entering viral RNA molecules, instead of being dispersed, were frequently found in spatially proximal clusters (Fig. 1C and D). This spatial proximity could be due to the rapid association with membranes and establishment of replication complexes by multiple viral RNA molecules entering collectively into the cytoplasm (13).

In a cell infected with a single strain of RNA virus, the replicating genomes exhibit significant sequence diversity as a consequence of the inherent error rates and lack of proofreading of viral RNA-dependent RNA polymerases. This sequence diversity within the population of viral quasispecies can generate differences within the population in terms of replication kinetics, translation efficiency, packaging, coping with innate host defenses, and responding to antiviral therapies. When many viral genomes have entered the cytoplasm, this genetic diversity could provide an environment for genetic cooperation and lead to increases in replication efficiency of multiple quasispecies. Genetic cooperation could take the form of genetic recombination among neighboring RNA templates during RNA polymerization, facilitated genome packaging, the sharing of machinery involved in replication, and modulating host innate immune defenses. Indeed, in a recent in vivo study it was reported that under selection pressure, group cooperation among viral quasispecies contributed greatly to population fitness; notably, this group cooperation was observed most often when the multiplicity of infecting viral particles was high between passages (15). Hence, our finding of increased levels of replication with vesicular viral transmission is likely due to it enabling large numbers of viral genomes to be simultaneously transferred into the host cell, which, along with the spatial proximity of the membrane-docked viral genome replication sites, provides opportunities for cooperative genetic interplay and enhancing the overall population fitness (Fig. 1C and D).

We discovered that the vesicle membrane surrounding the enterovirus particles was enriched in phosphatidylserine (PS) lipids, and masking these lipids blocked infection (3). Thus, while free enterovirus particles only appear to require the virus-specific receptors for entry, when viruses are trafficked via these vesicles, both a receptor for PS as well as the receptor for the virus appear to be needed for infection. PS lipids have been observed previously in the envelopes of vaccinia virus, Ebola virus, human immunodeficiency virus, vesicular stomatitis virus, and rabies virus and also appear to play a role in these infections as a cofactor in viral entry (16). PS lipids are recognized through many different types of receptors expressed on a multitude of cell types within the body, and the expression of these PS receptors is spatially and temporally regulated. Given this, we predict that the PS lipids on vesicles could further enhance the number of viral genomes entering a cell by targeting these vesicles to PS receptor-containing cells, particularly macrophages and dendritic cells. These cell types are master regulators of the host immune responses, where PS lipids not only act as “eat me” signals for the cells but also induce the expression of anti-inflammatory cytokines (17). We found that enteroviral particles within vesicles were not only taken up by macrophages but also, contrary to many other pathogens, able to replicate within these cells (3). Whether the PS lipids of these virus-containing vesicles also trigger an anti-inflammatory response, which would have significant implications for viral pathogenesis, is under investigation. In addition to PS, there may be other molecules present on the vesicles that regulate tropism and thereby alter the number of viral genomes entering a cell; this would be best determined through proteomic analysis of in vivo-derived vesicles.

How widespread in the virus world is this form of vesicle-mediated transmission of clusters of viruses? In addition to enteroviruses, electron micrographs of bluetongue virus-infected cells reveal extracellular vesicles containing multiple viral particles (6). Although it has yet to be demonstrated that norovirus can release nonlytically from cells, fecal preparations from norovirus-infected patients reveal multiple norovirus particles encapsulated within vesicles (18). Importantly, the transmission of multiple viral genomes may not always be in the form of viral particles: in hepatitis C infections, exosomes containing naked viral RNA are released and are capable of spreading HCV infection to new susceptible host cells (19). Similar findings have been reported for human pegivirus (20). Thus, even though the bona fide enveloped HCV and pegivirus can exit cells through conventional secretory pathways as individual particles, their genomes are also likely released in parallel as a collective of RNA quasispecies. Given this, the vesicular transfer of viral populations, either as particles or as naked RNA genomes, may be widespread and enable both nonenveloped and enveloped viruses to effectively increase the number of viral genomes entering a cell and enhance genetic interplay among quasispecies.

We thank Ellie Ehrenfeld and Geoff Holm for critical and constructive comments on the manuscript.