Host Restriction Factor Screening – Let the Virus Do the Work

Abstract

In this issue of Cell Host & Microbe, Varble et al 2013, engineer a library of RNA viruses to express small interfering RNAs and couple this with the power of virus evolution and selection to screen for host genes that when silenced resulted in greater viral infection in vivo.

Over the last five years, a variety of screening methods have been used to identify host factors that restrict viral infection. Some investigators have used genome-wide screening approaches, whereas others have targeted specific antiviral pathways, including the type I interferon (IFN) response. Experimentally, these screens have relied on short hairpin RNA-based gene silencing or ectopic gene expression largely in transformed cell culture models of virus infection. In gene silencing approaches, virus infectivity is enhanced when expression of restriction factors is diminished. In the context of treating cells with exogenous IFN, gene silencing can define the relative contribution of individual interferon-stimulated genes (ISG) to the host antiviral responses. In ectopic expression screens, particularly of genes in the IFN signaling and effector pathway, host factors that are sufficient to protect cells from virus infection have been revealed.

Each screening method has strengths and limitations. In genome-wide siRNA screening approaches, an underlying assumption is that a restriction factor will be expressed basally at levels that are sufficient to control an incoming virus. The identification of the antiviral activity of IFITM gene family members is a successful example of this strategy (Brass et al., 2009). Other genome-wide siRNA screens have been performed in the context of IFN treatment and uncovered numerous putative host restriction factors, many of which regulate cellular antiviral responses even though they are not induced by IFN (Fusco et al., 2013; Zhao et al., 2012). Two screens that specifically targeted the IFN pathway by silencing a comprehensive panel of ISGs identified novel host factors that had direct effector functions or regulated IFN response pathways (Li et al., 2013; Metz et al., 2012). In addition to gene silencing strategies, ectopic expression screens also have identified ISGs that inhibit virus infection. When hundreds of ISGs were tested for their ability to suppress virus infection (Liu et al., 2012; Schoggins et al., 2011) many known and novel genes with antiviral activity were revealed, which has helped to elucidate the complex nature of the IFN response against RNA and DNA viruses.

Identifying the full complement of cellular host restriction factors against viruses likely will require the continued and cumulative effort of both loss-of-function and gain-of-function approaches. Perhaps not surprisingly, the strongest hits from most screens to date have been genes that regulate antiviral pathways. Nonetheless, some putative and novel direct antiviral effectors have been identified (Li et al., 2013; Liu et al., 2012; Metz et al., 2012; Schoggins et al., 2011) with numerous laboratories now pursuing their respective mechanisms of action. The relevance of several ISGs identified in cell culture screens (e.g., Ifitm3, Ifit1, Ifit2, Rsad2/viperin, Irf1, and Ch25h among others) has been confirmed in vivo with demonstrable virological phenotypes observed in mice with targeted gene deletions. Given the resources required to generate transgenic mice, novel gene screening strategies that can rapidly assess antiviral activity in vivo would be particularly useful in identifying additional and physiologically relevant host restriction factors. In this issue, the tenOever laboratory has addressed this need. Varble et al. (2013) applied an innovative screening strategy to identify novel host genes that restrict viral infection in an in vivo setting.

The authors performed a large-scale siRNA screen in the context of an infection by Sindbis virus (SINV), a model arthritogenic alphavirus of the Togaviridae family. Rather than introduce genes ectopically or silence them by transfection of siRNA or plasmid expression of shRNAs, the authors constructed libraries of individual SINV, each encoding a distinct artificial microRNA (amiRNA). Unlike shRNAs, which are fully complementary hairpins that bypass the need for the cell’s microprocessor, amiRNAs are Drosha-dependent substrates that mimic endogenous hairpins and feed into the host small RNA machinery to produce gene-specific siRNAs (tenOever, 2013). Proof-of-principle studies were performed in cell culture with SINV encoding amiRNAs against GFP or the antiviral sensor RIG-I to confirm accurate processing of the amiRNA, reduction in target gene expression, and enhanced viral infection (RIG-I only). The authors then performed experiments with the SINV library in mice. Two days after subcutaneous inoculation, virus was isolated from the spleen and subjected to deep sequencing; this revealed a clear enrichment of particular amiRNAs in the viral genome. The significance of selecting SINV containing individual amiRNAs was confirmed by re-cloning the hairpin into the parental SINV, infecting recombinant viruses into naïve mice, and showing they retained selection and replicated to higher levels compared to the parent virus.

The authors performed several independent and unbiased screens using SINV libraries of 10,000 independent amiRNAs and after deep sequencing, looked for target genes that were enriched. After identifying approximately 25 of the top ‘hits’, the authors selected two siRNA target genes, Zfx and Mga, for detailed validation and mechanism of action studies. These two genes were previously uncharacterized with respect to viral infection, conferred the greatest replication advantage in vivo, and were enriched in the majority of independent screens. Using a variety of biochemical and cell-based assays, the authors provide compelling evidence that both Zfx and Mga are transcriptional factors that influence ISG induction at different stages of the signaling pathway. Consistent with a more general transcriptional effect, the impact of reducing expression of these genes was not specific to SINV, as silencing or deletion of Zfx and Mga also enhanced infection of influenza A virus, an unrelated segmented negative strand RNA virus of the Orthomyxoviridae. Interestingly, individual genes defined previously as alphavirus restriction factors in cell culture or in vivo (e.g., Bst2, Rsad2/viperin, Zc3hav1/ZAP, Isg20, and Parp12) were not identified in this screen.

By coupling RNA interference with virus replication and selection in vivo, the authors defined a novel set of host genes, that when silenced, resulted in enhanced infection in a relevant context. This strategy worked for several reasons: (a) although RNA viruses tend to select against exogenous genomic insertions that may inherently compromise fitness, the selective pressure to reduce expression of antiviral genes was greater, and thus the embedded amiRNA targeting inhibitory genes were retained; (b) SINV replicates rapidly in the absence of immune pressures (i.e., propagation in Vero or BHK21 cells lacking cell-intrinsic immunity) but is inhibited efficiently by the antiviral effects of type I IFN or host defense genes; this allows a strong and selective pressure to silence inhibitory genes; and (c) SINV, in contrast to other alphaviruses (e.g. Venezuelan and Eastern equine encephalitis viruses), is relatively attenuated in mice, possibly because its viral gene products do not efficiently antagonize and evade host innate immune restriction pathways. Thus, selection and enhanced replication of SINV variants in the context of silencing individual host genes can be observed.

The power of this RNA interference screen is its ability to identify host restriction factors in the context of virus infection in vivo by relying on virus competition and selection. For SINV, this strategy yielded a distinct set of target inhibitory genes compared to conventional gene silencing or ectopic expression screens. One can imagine the application of this screen in different contexts to address unique questions of host antiviral defense. Screens could be performed ex vivo in specific primary cell types (neurons, dendritic cells, macrophages, fibroblasts, or epithelia cells) to identify cell-type specific genes that restrict SINV infection. Alternatively, infection screens with SINV could be repeated in immunocompromised mice (i.e, Stat1^−/− or Ifnar^−/−) to focus on inhibitory genes that function independently of IFN signaling. Selected smaller libraries of SINV-miR targeting specific candidate effector ISGs could be developed to identify the hierarchy of antiviral control in vivo. Finally, this gain-of-function selection approach likely can be applied to other RNA or DNA viruses to clarify the host-pathogen interface by identifying novel host factors that inhibit replication, affect tropism, impact the establishment of persistence, and attenuate pathogenesis and virulence.