The Rhabdoviridae family is comprised of more than 175 different currently classified viruses, which are able to infect vertebrates, invertebrates and plants. Common features shared by all rhabdoviruses are an elongated bullet-like shape and an enveloped virion that contains a single-stranded nonsegmented RNA genome. In addition, rhabdoviral genomes encode a basic set of five structural proteins: a nucleoprotein (N), a phosphoprotein (P), a matrix protein (M), a single trans-membrane glycoprotein (G), and an RNA-dependent RNA polymerase (L). Of note, plant as well as some animal rhabdoviruses can encode up to four additional nonstructural proteins [1]. In principle, these nonstructural proteins could be involved in suppression or evasion of host antiviral responses. However, to date the only known mechanisms involve structural proteins that serve dual functions in both virus replication and suppression of host responses.

Although the viruses classified in the Rhabdoviridae family have many similarities, there are also important differences in the way different members interact with the host defense mechanisms. This is presumably due to the different defense mechanisms within their hosts. Whereas in plant and insect cells micro RNA (miRNA) probably play a major role in host defense mechanism against rhabdoviral infections [2,3], in higher vertebrates the production of type I interferon (IFN) and initiation of the innate immune response is critical. Most viral pathogens of higher vertebrates are sensitive to IFN-induced antiviral proteins (see other articles in this issue), and thus, have adapted individual mechanisms to avert the innate immune response. This review will focus on type I interferon induction and evasion by two well-studied animal pathogens rabies virus (RABV) and vesicular stomatitis virus (VSV). These viruses are representative of two major Rhabdovirus genera, Lyssavirus and Vesiculovirus, respectively. These viruses are quite similar in their mechanisms of replication, but differ markedly in their host range, pathogenesis, and mechanisms of immune evasion.

More than twenty years ago type I IFN production in response to RABV infection, especially at the site of infection, was described [4]. It has also been reported that the high levels of IFN in the serum early following infection with RABV contributes to viral resistance [5]. Furthermore, mice injected with anti-mouse IFN-α/β antibody prior to infection with RABV were more sensitive to the virus than mice injected with a control antibody [4]. Likewise, when the IFN-β gene was cloned into the RABV backbone, it was seen that the recombinant virus had greatly reduced pathogenicity and viral replication in mice, however the immunogenicity of RABV was not decreased [6].

Similar observations for the importance of IFN-α/β induction have been made following VSV infection. Type I IFN has been shown to be critical for resistance of mice to infection with VSV. For example, mice deficient in molecules required for IFN-mediated signaling are more sensitive to VSV infection than wildtype mice [7,8]. Interestingly, wild-type VSV infections do not induce type I IFN production in the central nervous system of mice, however type I IFN is rapidly induced in the periphery of these animals [9], which serves to protect most tissues from viral pathogenicity. Taken together, these findings imply that type I IFN plays a major role in the viral lifecycle of RABV and VSV. Furthermore, it indicates that both viruses have developed mechanisms to interfere with the innate host response in order to establish productive infections in their hosts.

In order to better understand the host innate immune response to RABV and VSV infection, we will first briefly describe the genomic organization and the viral lifecycle. As listed above rhabdoviruses encode a “basic set” of five proteins, which are encoded on a genome that has a gene order of 3’ N-P-M-G-L 5’. In addition, all rhabdoviruses contain a noncoding 3’-leader RNA and a non-transcribed trailer sequence at the 5’ end. The rhabdoviral genes are flanked by intergenic sequences, which are in part conserved between all members of the family [10,11]. Following transcription, the viral proteins are translated from monocistronic mRNAs.

The N protein of rhabdoviruses encapsulates the RNA genome and antigenome into a tight, RNase-resistant core [12]. Only the encapsidated RNA, termed the ribonucleoprotein (RNP), is a functional template for transcription and replication. It is also thought that the level of N protein in the cells regulates the switch from transcription to replication by interactions with the leader RNA (leRNA) sequence [13]. It is speculated that if enough N protein is available within the cell, encapsidation of the leRNA sequence results in the synthesis of a full-length antigenome rather the production of individual viral mRNA, therefore switching transcription to replication [13].

The P protein serves as the non-catalytic subunit to the viral transcriptase complex. Once P is phosphorylated it was proposed to form trimers that are able to bind L and N [14]. However, more recently structural analysis revealed that P is dimmer or tetramer [15-17]. Another function of P is to serve as a chaperone for N and prevent it from aggregation [18,19]. Additionally, RABV P is able to antagonize interferon signaling, as discussed in subsequent sections of this review.

The M protein is able to interact both with the nucleocapsid and with the cellular plasma membrane [20]. Thus, the primary function of a rhabdoviral M protein is to bridge the viral membrane containing envelope and the nucleocapsid core during viral assembly and budding. The balance between transcription and replication is thought to be dependent on the M protein in addition to the quantity of N in the infected cell [21]. A high concentration of VSV M inhibits RNA synthesis in vitro [22]. It has been proposed that the association of RNP with VSV M results in the condensation of the RNP molecule, which renders it non-functional for transcription and replication. In addition, M has been seen to have secondary functions following VSV infection that might have been evolved later. Namely, VSV M has the ability to inhibit host defense mechanisms like type I interferon signaling, which will be discussed later in this review.

The fourth gene encodes the G protein, which is the only trans-membrane glycoprotein in this family of viruses. The G proteins trimerize in the ER [23] and the association is further stabilized by lateral interactions among trimers on the virion’s surface [24]. At lower pH values the G protein undergoes a conformational change, which exposes hydrophobic residues allowing for fusion with target membranes [25,26]. It is interesting to note that for both VSV G and RABV G, the pH dependent conformational changes required for fusion are reversible [27,28].

The catalytic subunit of the rhabdoviral polymerase complex is the large (L) protein. In addition to transcription, this large protein is capable of capping and methylating the 5’ end of the mRNA ( [29] and [30], respectively) and polyadenylating the 3’ end [31].

As with many viruses, the lifecycle of Rhabdoviridae can be divided into three distinct phases: uncoating, transcription/ replication, and viral assembly. The first phase initiates when a viral particle binds to the surface of a host cell and enters by endocytosis. Attachment to the host cell requires the G protein. However, it is less clear which cellular molecule(s) interacts with G to mediate viral entry. Of note, the nicotinic acetylcholine receptor [32], neuronal cell adhesion molecule [33] and low-affinity nerve-growth factor receptor (P75NTR) [34] has been suggested as receptors for RABV. On the other hand, VSV G protein appears to interact with negatively charged lipids on the cell surface through a combination of electrostatic and hydrophobic interactions [35-37]. Following receptor binding, rhabdoviruses are endocytosed. Once in an endosome, the viral membrane fuses with the endosomal membrane to release the genome into the cytosol. Fusion of the virus with host membranes requires a change to a lower pH within the vesicle. Consequently, membrane fusion and the release of the capsid into the cytosol can be inhibited by both ammonium chloride and chloroquine [38,39].

In the second phase, virion components are produced. Direct translation of the rhabdoviral genome cannot occur because its RNA is negative sense, and furthermore, direct translation is prevented by the encapsidation of the genome by N [40-43]. Thus, after release into the host cell cytosol the viral polymerase complex contained within the capsid is responsible for the RNA synthesis. The most widely accepted model for transcription of rhabdoviruses is the so-called stop/start model, where the polymerase stops at a signal sequence and restarts transcription at the transcription start signal sequence [44]. Reinitiation of transcription by the polymerase complex for the following gene does not always occur; therefore, attenuation of transcription occurs in a 3’ to 5’ direction [45]. Probably caused by the concentration of N protein, the viral polymerase switches to replication and transcribes a full-length anti-genomic RNA, which is also encapsidated into the N protein and serves a template for the replication of new genomic RNAs .

The last phase of the life cycle is the assembly of the viral components, budding and finally release of the virions. For this process, the RNP-complex must be engulfed in the host cell membrane. It is as of yet unknown how the capsid is transported to the cellular membrane. Budding seems to be at least partially dependent on G, as in the absence of RABV-G RABV is released 30-fold less efficiently [46] and similar finding have been made for VSV [47]. More dramatically, in the absence of RABV M viral titers were reduced as much as 500,000 fold, but this did not greatly affect viral protein expression [48]. Likewise, VSV M plays a role in seizing the endosomal-membrane fusion machinery to facilitate viral budding. This occurs via the PPPY motif near the N-terminus of VSV M, which allows for interaction with the cellular enzyme Nedd4 [49].

The hallmark response to viral infection is the induction of type I interferon (IFN). This class of cytokines is comprised of several IFN-α genes, a single IFN-β gene, and the more recently added, and less well-defined, genes such as IFN-ω, -ε, -τ, -δ, and –κ [50]. All nucleated cells have the ability to both produce type I IFN and also to respond to it via a common heterodimeric receptor [51]. Type I IFNs have been seen to induce an anti-viral state in cells, which makes them resistant to subsequent viral infection [52]. In addition, this group of cytokines helps to initiate the adaptive immune response following infection [6,53,54].

A variety of viral moieties can trigger activation of the IFN-α/β signaling cascade by numerous pathogen pattern recognition receptors (PRRs). In each case however, the signaling culminates in the binding of cytoplasmic transcription factors, namely IFN regulatory factor-3 (IRF-3) and nuclear factor kappa B (NF-κB), to the IFN-β promoter. Following signaling initiation, the C-terminus of IRF-3 is phosphorylated and the proteins dimerize. This dimerization reveals a nuclear-localization signal (NLS) on IRF-3, which allows for transport into the nucleus [55]. Alternatively, cytoplasmic NF-κB can translocate to the nucleus following signal induced ubiquitination and proteasomal degradation of inhibitor of NF-κB (IκB) (for review see [56]). Optimal binding and induction of IFN-β requires the additional binding of c-jun/ATF-2 to the IFN-β promoter. When all three transcription factors complexes assemble onto the promoter it is referred to as the enhanceasome. The enhanceasome aids in the recruitment of CREB-binding protein (CBP)/p300, which promotes the assembly of the transcriptional machinery and RNA polymerase II [57]. Each transcription factor can bind to the IFN-β promoter individually, however with limited affinity. It is generally accepted that the binding of IRF-3 is indispensable for induction, while the binding of NF-κB and c-jun/ATF-2 is not required [58]. Similarly, the transcription of most interferon-α genes requires activation of IRF-7, whose expression is induced in response to interferon-β in many cell types [59].

The upstream signaling mechanisms leading to type I interferon production vary markedly among different cell types depending on the expression of various pattern recognition receptors. The receptors that have been demonstrated to be important for the response to rhabdoviruses include Toll-like receptors (TLR), which are expressed either on the cell surface or in the endosomal compartment, and RIG-I-like receptors, which are expressed in the cytoplasm.

As mentioned above, the receptor for type I IFN is expressed on a wide range of cells. Additionally, IFN-α/β can act in both an autocrine and paracrine manner to induce an anti-viral state in the cell. When IFN-α/β binds to its heterodimeric receptor oligomerization and a subsequent conformational change occurs allowing members of the Janus kinase (JAK) family to phosporylate the IFNAR1 subunit. The phosphorylated receptor subunit can then act as a docking site for members of the signal transducers and activators of transcription (STATs), which are in turn phosphorylated by JAKs [50]. Once STAT-1 and STAT-2 are phosphorylated they dimerize and create a novel NLS, which allows for their transport into the nucleus. The STAT1/2 complex forms a heterotrimer with IRF-9 either in the nucleus or at the receptor, making what is referred to as the ISGF3 complex. The canonical activation of IFN-stimulated genes (ISGs) occurs by ISGF3 binding to the IFN-stimulated response element (ISRE) present in the promoter of most ISGs [50]. Of note, all of the components that make up the ISGF3 are IFN-inducible.

Following treatment with IFN-α/β, the transcript frequency for numerous genes increases [93]. Many of these gene products encode proteins or transcription factors involved in the aforementioned signaling cascades, thus acting as an amplification loop to increase the amount of type I IFN produced. Other genes encode for proteins with direct antiviral functions. Here we will only discuss the antiviral gene products that have been shown to play a role in a RABV or VSV infection, however that does not imply that those genes not discussed do not play a role in the innate response to these or other rhabdoviruses.

One such anti-viral protein is the Mx family of GTPases, which are highly induced in response to type I IFN. The main viral target for these proteins is nucleocapsid-like structures and after binding they appear to sequester the viral proteins and interfere with proper localization [94]. It was seen that expression of bovine Mx1 can reduce viral titers and decrease N protein levels in rabies infected VERO cells [95]. Likewise, VSV replication is inhibited by a wide variety of Mx proteins in vitro [96-98]. The human MxA protein was seen to inhibit both viral mRNA and protein levels following infection with VSV for 5 hours [98]. Furthermore, transgenic mice that express high levels of bovine Mx1 are resistant to a lethal challenge of VSV [99].

Another IFN-induced gene that affects both RABV and VSV is promyelocytic leukaemia (PML). PML is expressed in the nucleoplasm and nuclear bodies (NB) of the cell. Following IFN-α/β treatment, enhanced PML protein expression is seen and the NB size markedly increases [100,101]. The anti-viral properties of PML are still unclear, however it is known that expression is essential for type I IFN induced apoptosis [102] and that PML has growth suppressing properties [100]. Like Mx proteins, it was observed that overexpression of PMLIII induced resistance to VSV infection, by inhibiting mRNA and protein synthesis [103]. Of note, complete inhibition by PML was only seen following infections at a low multiplicity of infection [103]. Furthermore, following type I IFN treatment MEFs derived from PML-/- mice had similar resistance to VSV infection as seen in wildtype cells [101]. The anti-viral role of PML following a RABV infection has not, as of yet, been experimentally proven. Rather, an anti-viral role is hypothesized due to RABV P protein’s ability to disrupt NB organization and increased rabies virus replication, as seen by higher protein level expression and 20 times more virus, in PML-/- MEFs [104]. The details of RABV induced PML inhibition will be discussed in detail in the next section.

Although the antiviral effects induced by Mx/ PML is similar for RABV and VSV, there are a number of antiviral proteins that have only been reported to play a role following VSV infection, namely PKR and ISG15. PKR and ISG15 have completely unique mechanisms to alter the host response to a viral infection. PKR is constitutively expressed in all tissues, however it is activated and upregulated by either dsRNA or type I IFN treatment. PKR phosphorylates eukaryotic translational initiation factor (eIF)-2α, which results in the sequestering of eIF2-B. Thus, translation initiation is halted [105]. Mice deficient in PKR are much more sensitive to intranasal VSV infections than wildtype mice [106-108].

Alternatively, ISG15 is an ubiquitin homologue that activates its substrates, rather than targeting them for degradation [109]. ISG15 substrates include, among others, signaling components in the JAK/STAT and RIG-I pathways, MxA, and PKR [110]. The requirement for ISG15 in the innate response to VSV is still unknown, due to conflicting results. Lu et al. noted that ISG15-/- MEFs are more permissive to VSV replication, however there was no notable differences in the knockout cells’ sensitivity to the antiviral effect of IFN-α/β [109]. On the other hand, using a knockout mouse that was only deficient in ISG15 conjugation, not free ISG15, there was no difference in the susceptibility of cells to VSV in an IFN-protection assay [111]. Of note, ISG15 is regulated by USP18, which catalyzes its hydrolysis to remove it from conjugated proteins [112,113]. Interestingly, USP18-/- mice show an increased survival following intracerebral infection with VSV as compared to wildtype mice [114].

In order for an effective infection to occur, RABV and VSV must antagonize the interferon signaling cascade and block induction of antiviral molecules. However, RABV and VSV have completely different mechanisms by which they antagonize the immune response. RABV specifically targets signaling molecules in order to block IFN induction and subsequent responses. On the other hand, VSV takes a more global approach and inhibits host-cell transcription and translation. In both cases the IFN antagonism is due to viral proteins that play key roles in virus replication and have separate functions involved in inhibition of host antiviral responses, the RABV P protein and the VSV M protein.

Here we have discussed the different ways by which type I IFN is induced by RABV and VSV, the host antiviral proteins that inhibit the viral lifecycle, and the mechanisms by which RABV and VSV antagonize IFN-α/β pathways. While both rhabdoviruses can induce IFN-α/β by the RIG-I pathway, it seems they use different TLRs. VSV can signal through TLR-7 and TLR-4, while TLR-3 is dispensable for inducing IFN. On the other hand, signaling via TLRs is less well studied following a RABV infection but there is data to indicate that TLR-3 may play a role. Both RABV and VSV are sensitive to Mx GTPases and PML, while only VSV seems to be affected by PKR and ISG15. Lastly, these two rhabdoviruses have strikingly different mechanisms to antagonize IFN signaling. RABV specifically targets IRF-3 to minimize IFN-α/β production and also inhibits the antiviral response in infected cells by binding STAT-1 and -2 and disrupting PML nuclear bodies. On the other hand, VSV inhibits cellular transcription pathways and blocks mRNA export to prevent an antiviral response.

Understanding the host response to VSV and RABV is important for developing vaccines against rhabdoviral induced disease but also for developing potential rhabdoviral vaccine vectors. A vaccine strain of RABV expressing IFN-β greatly reduces viral pathogenesis and replication but not the immunogenicity of the vector [6]. Thus, providing a potentially self-limiting and highly immunogenic vaccine vector. Additionally, both RABV and VSV based vaccine vectors have been shown to be efficient vaccines against pathogenic SHIV_89.6P challenge [140,141]. Furthermore, VSV has recently been considered for potential use as an oncolytic therapy agent. Many transformed cells have impaired IFN pathways [142], thus VSV can replicate to a much greater extent in those cells than normal cells that have intact IFN signaling. In fact, it has been reported that transformed cells are more susceptible to VSV infection then normal cells following IFN-α/β treatment [143,144]. Thus, ongoing efforts to understand the type I IFN response to rhabdoviruses have obvious implications for improving current vaccines and therapies.