J.T. Patton*,** * Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, United States ** Virginia-Maryland College of Veterinary Medicine, University of Maryland, MD, United States
The rotavirus (RV) capsid is a nonenveloped triple-layered icosahedron that contains a genome of 11 segments of double-stranded (ds)RNA. In the cytoplasm, the transcription activity of partially uncoated RV particles produces capped (+)RNAs. These transcripts direct protein synthesis and act as templates for dsRNA synthesis. This replication pathway predicts that RV (+)RNAs are infectious when introduced into uninfected cells. Indeed, such is the case for the Reoviridae members: mammalian reovirus, bluetongue virus, African horse sickness virus, and epizootic hemorrhagic disease virus, where fully recombinant reverse genetics (RG) systems have been developed using viral (+)RNAs derived from T7 transcription vectors. Attempts to create similar fully recombinant RV RG systems have so far been unsuccessful. However, single-gene RG systems have been described that allow genetic manipulation of select RV genome segments. Here, we review the RV replication cycle and consider progress and challenges in the development of more powerful RG systems.
The establishment of efficient reverse genetics (RG) systems opens a window to studies on the biology and pathogenesis of viruses that cannot be accomplished through other means. Efficient complete plasmid-based RG systems have been developed for four members of the Reoviridae: mammalian reovirus (MRV), bluetongue virus (BTV), African horse sickness virus (AHSV), and epizootic hemorrhagic disease virus (EHDV). These RG systems all rely on the principle that viral (+)RNAs are “infectious” because viral (+)RNAs function both as templates for protein and genome synthesis in the Reoviridae life cycle. Indeed, the MRV, BTV, AHSV, and EHDV RG systems all use the introduction of recombinant viral (+)RNAs produced from T7 transcription vectors into uninfected cells as a mechanism to launch replication of recombinant viruses. Despite extensive efforts, attempts to develop similar plasmid-only RV RG systems have yet to be successful. However, several single-gene replacement systems have been developed that enable reverse engineering of selected RV genome segments. In this chapter, we review the RV replication cycle and consider issues that may be hampering the development of fully recombinant RG systems for these viruses.
2. Rotavirus replication
2.1. Virus Structure
The RV virion is a nonenveloped icosahedral triple-layered particle (TLP) with a diameter of approximately 100 nm (Li et al., 2009). Enclosed within the virion is a genome consisting of 11 dsRNA segments ranging in size from 0.7 to 3.3 kbp (Estes et al., 1989). The (+) strand of each segment contains a 5′-cap, but lacks a 3′-poly(A) tail. In contrast, the 5′-γ-phosphate of the (−) strand is missing, reflecting a hydrolysis event occurring for an unknown reason (Imai et al., 1983). The smooth outer layer of the virion is formed by the glycoprotein VP7; emanating from it are spikes of the viral attachment protein VP4 (Settembre et al., 2011). Exposure of the virus to trypsin-like proteases cleaves VP4 into the subunits VP5* and VP8*, resulting in conformational changes that enhance virus infectivity (Estes et al., 1981; Arias et al., 1996). The intermediate layer of the virion is composed of elongate twisted columns formed by VP6 trimers (Mathieu et al., 2001), and the inner layer is composed of 12 thin decamers of the core shell protein VP2 (McClain et al., 2010). Each decamer forms one of 12 corners (vertices) of the core and serves as an attachment site for the viral RNA-dependent RNA polymerase (RdRP), VP1, and the RNA-capping enzyme, VP3 (Estrozi et al., 2013). Within the core, the genome is likely organized such that each dsRNA segment is associated with one specific VP1/VP3/VP2 decamer complex (Prasad et al., 1996; Periz et al., 2013).
2.2. Viral Transcription
RV entry is an endosome-dependent process that leads to the loss of the VP4-VP7 outer capsid layer and the release of a double-layered particle (DLP) into the cytosol (Abdelhakim et al., 2014; Arias et al., 2015) (see Chapter 2.2). The DLP acts as transcription machine, with VP1 and VP3 in the core directing synthesis of 11 species of RV (+)RNAs (Cohen, 1977). The 11 species are made at nonequimolar levels in the infected cell (Stacy-Phipps and Patton, 1987). Newly made transcripts extrude through channels at the 5-fold axes of the DLP, passing through both the VP2 and VP6 protein layers (Lawton et al., 1997). The 5′-ends of most (+)RNAs contain cap-1 structures (Imai et al., 1983); these modifications result from the guanyltransferase, N-7-methyltransferase and 2′-O-methyltransferase activities associated with the VP3 RNA capping enzyme (Ogden et al., 2014). Some RV (+)RNAs recovered from infected cells lack 5′-caps, ending instead with 5′-terminal phosphates (Uzri and Greenberg, 2013). Whether capped and uncapped viral transcripts have functional differences is not clear, although presumably the lack of caps would affect recruitment of translational initiation factors and ribosomal subunits to the 5′-end of (+)RNAs. The lack of 5′-caps increases the likelihood that RV (+)RNAs will be recognized by retinoic acid-inducible gene 1 (RIG-I), a intracellular pattern recognition receptor (PRR) that can promote IFN expression and an antiviral response (Reikine et al., 2014). The effect of RIG-I can be countered in RV-infected cells by the action of the viral protein, NSP1, a suspected viral E3 ubiquitin ligase that inhibits IFN expression by targeting key host factors for proteasomal degradation (Morelli et al., 2015).
RV (+)RNAs are generally monocistronic, encoding either a single structural (VP1–VP4, VP6–VP7) or nonstructural protein (NSP1–NSP4) (Estes et al., 2007). The exception is the transcript made from RNA segment 11, which contains an open reading frame (ORF) for NSP5 and a second ORF for the smaller NSP6 (Mattion et al., 1991). Comparison of the 11 segments of RV (+)RNAs shows that the only conserved sequences among them are limited to a few residues at the 5′-(5′-GGC(A/U)n-3′) and 3′-(5′-UGUGACC-3′) termini. Notably, the 5′ and 3′ untranslated regions (UTRs) of RV (+)RNAs contain long stretches of complementary sequences, often extending into the ORF. Based on secondary structure predictions, the complementary sequences are believed to mediate long-range interactions between the ends of viral (+)RNAs, producing 5′–3′ panhandle structures (Chen and Patton, 1998; Tortorici et al., 2003). These predictions also indicate that one or more stable stem-loop structures project from the panhandles (Li et al., 2010). The panhandle and stem-loop structures and conserved 5′ and 3′ terminal sequences likely represent cis-acting signals that are critical for replication and packaging of viral RNAs (Biswas et al., 2014; Suzuki, 2014). These highly ordered structural elements of RV (+)RNAs are potentially recognized by several intracellular PRRs, including the host dsRNA sensors melanoma differentiation-associated protein 5 (MDA5), protein kinase R (PKR), and 2′–5′ oligoadenylate synthetase (OAS) (Wu and Chen, 2014). Within infected cells, RV NSP1, VP3, and other proteins likely subvert the antiviral functions of these PRRs (Morelli et al., 2015).
Transcriptionally active DLPs extrude transcripts in a 5′ to 3′ direction (Lawton et al., 1997, 2000), an orientation that may allow interaction of translation initiation factors and ribosomal subunits with the 5′-terminus before synthesis of the (+)RNA is complete. As a result, RV DLPs may promote (+)RNA-polysome formation in a manner that cannot be replicated by full-length folded RNAs containing 5′–3′ panhandles. The extent to which concurrent transcription–translation impacts the efficiency of viral protein synthesis in infected cells is not known, but could have bearing on the poor translatability noted for viral (+)RNAs transfected into cells (Richards, 2012; Richards et al., 2013). Concurrent loading of translation factors and ribosomes onto the 5′-ends of viral (+)RNAs during transcription may prevent the formation of 5′–3′ panhandle structures that are recognized by PRRs and trigger antiviral responses.
2.3. Viroplasm Formation and Function
RV (+)RNAs not only direct protein synthesis in infected cells, but also serve as templates for the synthesis of dsRNA genome segments through a process that is coordinated with the packaging and assembly of progeny cores (Trask et al., 2012a,b). Replication and assembly take place in electron-dense cytoplasmic inclusion bodies, termed viroplasms (Altenburg et al., 1980; Eichwald et al., 2012). Viroplasms are dynamic structures initially appearing in infected cells as small punctate structures that then fuse and grow in size over the course of infection (Eichwald et al., 2004a). Although viroplasms are not membrane enclosed, they are often located adjacent to vesicles formed by the endoplasmic reticulum (ER) (see Chapter 2.5).
Two RV nonstructural proteins, NSP2 and NSP5, are essential for the formation of viroplasms (Fabbretti et al., 1999). NSP2 exists in the infected cell as large doughnut-shaped octamers with deep diagonal grooves that mediate strong sequence-independent binding of single-stranded (ss)RNA (Jayaram et al., 2002; Taraporewala et al., 1999). This activity is an essential component of the helix-destabilizing activity detected for the octamer (Taraporewala and Patton, 2001). The octamer is also an enzyme, displaying phosphatase activities able to hydrolyze the γ-phosphate from NTP and RNA substrates (vis-à-vis, NTPase and RTPase activities) (Vasquez-Del Carpio et al., 2006). In contrast, NSP5 forms dimers, which may self-associate into larger decameric structures (Martin et al., 2011). NSP5 also binds RNA (Vende et al., 2002), although the activity is weak compared to that of NSP2. NSP5 undergoes two types of posttranslational modification: O-linked glycosylation and phosphorylation; the latter is believed to result from casein kinase activities (Eichwald et al., 2004a,b; Gonzalez and Burrone, 1991). NSP2 stimulates the hyperphosphorylation of NSP5 and such forms of NSP5 preferentially accumulate in viroplasms. Binding sites for NSP5 and ssRNA are located close together on the NSP2 octamer (Jiang et al., 2006), potentially allowing competition between the two to regulate the function of the octamer and the formation of viroplasms. NSP2 and NSP5 have affinity for core structural proteins, and these interactions are probably key to recruiting VP1, VP2, VP3, and VP6 to viroplasms (Arnoldi et al., 2007; Berois et al., 2003; Kattoura et al., 1994; Viskovska et al., 2014). Moreover, interactions with NSP2 and NSP5 may be important for regulating the intrinsic self-assembly tendencies of the structural proteins, thereby suppressing the formation of empty capsids. Of interest, all the nonstructural and structural proteins that accumulate in viroplasms, with the exception of VP6, have affinity for ssRNA (Patton, 1995; Patton et al., 2006).
Transcriptionally active DLPs are proposed to serve as nucleation sites for the formation of viroplasms (Silvestri et al., 2004). In this scenario, transcripts produced by DLPs soon after infection are incorporated into nearby polysomes that direct viral protein synthesis. Due to the affinity of the viroplasm proteins for RV (+)RNAs and for each other, many will interact with and concentrate around transcriptionally active DLPs, ultimately embedding the particle within small protein-rich inclusions. With the continued synthesis of transcripts by DLPs and viral proteins by nearby polysomes, the inclusions grow into larger electron-dense structures that capture large amounts of (+)RNAs through the activity of their RNA-binding proteins. An important aspect of this scenario is the assumption that (+)RNAs used as templates for dsRNA synthesis may originate from transcriptionally active DLPs contained within the viroplasm. Consistent with this hypothesis are data from BrUTP-labeling studies indicating that transcriptionally active DLPs are present within viroplasms and that newly made transcripts accumulate in viroplasms (Silvestri et al., 2004). On a related note, studies of other members of the Reoviridae (ie, MRV, rice dwarf virus) indicate that transcriptionally active particles generated from their incoming virions are recruited into viroplasmic structures (Miller et al., 2010; Wei et al., 2006).
Transfection of RV-infected cells with gene 5- and gene 7-specific small interfering (si)RNAs knock down expression of NSP1 and NSP3, respectively, without reducing levels of genome replication and virion assembly (Silvestri et al., 2004; Montero et al., 2006). Not only do such data demonstrate that NSP1 and NSP3 have nonessential roles in virus replication, but they also reveal the presence of two pools of (+)RNAs in infected cells. One pool, which directs protein synthesis, is sensitive to siRNA-induced degradation, while the other pool, which directs dsRNA synthesis, is not subject to degradation. Moreover, these data indicate that the pool of (+)RNAs that directs protein synthesis is not the source of (+)RNAs used for dsRNA synthesis. These data also suggest that (+)RNAs contained in viroplasms—the site of genome replication—are not targeted by siRNA-induced degradation pathways, consistent with the concept that viroplasms operate as “safe houses” walled off from host antiviral effectors. Given that (+)RNAs present in the cytosol [ie, polysomal (+)RNAs] are sensitive to siRNA-induced degradation, it can be argued that the more likely source of (+)RNAs used for dsRNA synthesis are transcriptionally active DLPs contained within viroplasms. If this argument holds, then viral (+)RNAs transfected into infected cells may be poorly integrated into viroplasms (Silvestri et al., 2004).
2.4. Genome Replication and Particle Assembly
Steps in RV genome replication and particle assembly are poorly understood, but some insight has been gleaned from the study of (1) viral replication intermediates (Boudreaux et al., 2015; Gallegos and Patton, 1989), (2) the structure and function of the core proteins VP1, VP2, and VP3, and the viroplasm-building blocks NSP2 and NSP5 (Boudreaux et al., 2013; Liu et al., 1988; McDonald and Patton, 2011), and (3) the location and activities of cis-acting replication signals in template RNAs (Chen et al., 2001; Patton et al., 1996). Fig. 2.3.1 summarizes a model for replication and assembly that we believe is most consistent with available data. The model proposes that the 11 different segments of RV (+)RNAs accumulate in the viroplasm, folded in such a way that their 5′–3′ panhandle structures are present. The viral RdRP, VP1, is recruited to the 3′-end of (+)RNA panhandles due to the affinity of its template entry tunnel for the conserved 3′-terminal sequence UGUGACC (Lu et al., 2008). Due to the affinity of VP3 for the 5′-end of the RNA and possibly for VP1, VP1–VP3–(+)RNA complexes are formed (Patton and Chen, 1999). The model predicts that these complexes undergo assortment via RNA–RNA interactions mediated by conserved sequences and/or structural elements associated with 5′–3′ panhandles (Li et al., 2010). VP2 decameric plates assemble around the assorted complexes due to the protein’s affinity for VP1 and VP3, and ssRNA. Contacts made between the assembled VP2 core shell and VP1 trigger conformational changes in the priming loop of the polymerase, resulting in the initiation of (−) strand RNA synthesis (Lu et al., 2008; Gridley and Patton, 2014). This polymerase activity yields the formation of progeny cores that contain 11 dsRNA genome segments. Interaction of cores with VP6 concentrated around the periphery of viroplasms leads to DLP formation (Chen and Ramig, 1993). Newly formed DLPs can amplify the replication cycle by serving as sources of secondary transcripts or migrate to the ER due the affinity of the VP6 capsid component for the ER-transmembrane protein NSP4. The VP4–VP7 outer capsid of the TLP is acquired during or after budding of DLPs into the ER (Trask and Dormitzer, 2006). TLPs characteristically accumulate within ER vesicles during the course of infection and are then released by cell lysis. To some extent, TLPs may be released early in infection by exocytosis from the apical surface of polarized cells (Chwetzoff and Trugnan, 2006).
This model indicates that genome replication and core assembly are coordinated via the VP2-dependent polymerase activity of VP1, a mechanism assuring that dsRNA synthesis does not take place unless progeny cores are available to capture the RNA product. The model also predicts that (+)RNA assortment is precise and occurs prior to the synthesis of dsRNA by newly formed cores. Indeed, the 11 dsRNA genome segments are made in equimolar levels in infected cells (Patton, 1990), and based on CsCl gradient ultracentrifugation studies, incompletely packaged virus particles do not accumulate. There can be little doubt that the nonstructural proteins NSP2 and NSP5 play significant roles in genome replication and core assembly, as these proteins are major components of RV replication intermediates (Boudreaux et al., 2015; Gallegos and Patton, 1989; Campagna et al., 2005; Vascotto et al., 2004). In particular, NSP2 may prepare viral (+)RNAs for assortment and packaging into cores and facilitate (–) strand initiation (Jayaram et al., 2002).
3. Reverse genetics systems
3.1. Fully Recombinant Systems
Mechanisms of genome replication are similar for the Reoviridae, with viral (+)RNAs serving as templates for protein synthesis and dsRNA synthesis. This implies that for all members of the family, development of fully recombinant RG systems should be straightforward, simply requiring the introduction of a complete cohort of synthetic viral (+)RNAs into permissive cells. Indeed, this has been the pathway for development of RG systems for MRV, BTV, ASHV, and EHDV (Table 2.3.1). In the case of MRV, viral (+)RNAs are made in vivo by transfection of T7 transcription vectors into cells producing T7 RNA polymerase. For BTV, ASHV, and EHDV, viral (+)RNAs are made in vitro from T7 transcription vectors and then cotransfected into cells. Both approaches have yielded tractable RG systems that allow the study of virus biology and pathogenesis through mutagenesis (Boehme et al., 2013; Sandekian and Lemay, 2015), the directed formation of reassortant viruses (Celma et al., 2014; Kobayashi et al., 2010), the production of recombinant viruses that express nonviral proteins (Demidenko et al., 2013; Kobayashi et al., 2007; Shaw et al., 2013), and the exploration of next generation vaccines (Celma et al., 2013; Matsuo et al., 2011). Despite extensive efforts, a fully recombinant RV RG system has yet to be developed, even though efforts have closely patterned after those used for MRV, BTV, AHSV, and EHDV.
Fully Recombinant Reverse Genetics Systems for the Reoviridae
Source of T7 RNA polymerase
Approach for generating recombinant virus
Vaccinia virus rDIs-T7pol
Transfection of T7 transcription vectors (10-plasmid)
The plasmid-based MRV RG system has advanced through a number of improvements since its initial description in 2007. The original system required transfection of 10 plasmids—each a T7 transcription vector with a select MRV cDNA—into cells infected with a recombinant vaccinia virus encoding T7 RNA polymerase (rDIs-T7pol) (Kobayashi et al., 2007). The vectors were constructed by placing cDNAs of MRV genome segments immediately downstream of a T7 promoter and upstream from a cis-acting hepatitis delta virus (HDV) ribozyme (Rz). These T7(MRV-cDNA)Rz cassettes allowed synthesis of viral (+)RNA transcripts with authentic 5′ and 3′-termini. The efficiency of the system was subsequently improved by constructing multicistronic plasmids that contained two to four T7(MRV-cDNA)Rz cassettes, reducing the number of plasmids that needed to be transfected, from 10 to 4 (Kobayashi et al., 2010). A particularly important advance was the discovery that transfection of the MRV RG plasmids into BHK cell lines expressing T7 RNA polymerase (BHK-T7) also yielded recombinant virus, thus negating the need for continued use of rDIs-T7pol (Kobayashi et al., 2010). Interestingly, this discovery also indicated that viral transcripts lacking 5′ caps supported the formation of recombinant MRV. More recent studies show that recombinant MRV can be made in any of several cell lines by cotransfecting the MRV RG plasmid set with a plasmid encoding T7 RNA polymerase; this is an important finding for generating candidate vaccine viruses in qualified cell substrates (Komoto et al., 2014).
3.1.2. Orbiviruses: BTV, AHDV, and EHDV
Fully recombinant RG systems have been developed for three orbiviruses: BTV, AHDV, and EHDV. Key to their development was the initial discovery that viral (+)RNAs synthesized by BTV core particles in vitro are infectious when transfected into permissive cells (Boyce and Roy, 2007). Boyce et al. (2008) advanced this finding by creating a set of 10 plasmids, each containing a cDNA of one of the 10 BTV genome segments. The cDNAs were positioned with an upstream T7 promoter and a unique downstream restriction enzyme site, allowing linearized plasmids to be used as templates for the synthesis of BTV (+)RNAs in vitro. Transfection of capped T7 transcripts made from the plasmids resulted in the recovery of infectious virus, creating the first fully recombinant BTV RG system. Subsequent studies revealed that the efficiency of the system was significantly improved if cells were transfected twice with BTV transcripts, spaced 18-h apart (Matsuo and Roy, 2009). These studies went on to show that the principal function of the early transfected transcripts was to promote synthesis of BTV proteins required to support RNA synthesis, the assembly of subcore structures, and the formation of viroplasms. Based on this information, the components of the early transfection were adjusted to include only those viral (+)RNAs necessary for expression of the required BTV proteins (Matsuo and Roy, 2013). Further studies showed that the early transfected transcripts could be replaced with expression vectors encoding the required BTV proteins (Matsuo and Roy, 2013). Transcription of these vectors is driven by host RNA polymerase II and generates capped (+)RNAs that lack the terminal sequences found on BTV RNAs, stressing their importance as templates for protein synthesis and not for viral genome replication. Consistently, the results showed that viral (+)RNAs introduced into cells with the second transfection served as the source of templates for genome replication. It is interesting that there have been no reports indicating that recombinant BTV can be generated using T7 expression vectors as a source of viral (+)RNAs that serve as templates for genome replication.
Lourenco and Roy (2011) developed an alternative system for producing recombinant BTV, relying on the in vitro reconstitution of infectious BTV core particles containing synthetic viral (+)RNAs. Key to the success of the system was the sequential incubation of BTV (+)RNAs with wheat-germ expressed subcore and core proteins, in a process emulating virus-assembly steps in the infected cell. Clearly, the core-reconstitution RG system is a more technically challenging approach for generating recombinant virus than are infectious RNA-based RG systems. However, in situations where RNA transfections do not yield recombinant viruses, perhaps due to the poor translation or toxicity of viral (+)RNAs (Richards et al., 2013; Trask et al., 2012a; Wentzel, 2014), core-reconstitution systems may provide a useful alternative route for creating RG systems.
The approaches taken to generate RG systems for AHSV and EHDV mirror those used for BTV. In the case of AHSV, initial experiments showed that double transfection of viral (+)RNAs into permissive cells was more efficient than a single transfection in generating recombinant virus (Matsuo et al., 2010). The AHSV system was subsequently modified, relying on the early transfection of expression vectors instead of (+)RNAs to drive the synthesis of viral proteins (Kaname et al., 2013). As with the BTV system, only six vectors were required, those encoding viral proteins associated with genome replication, subcore assembly, and the formation of viroplasms. The EHDV RG system is like the modified AHSV system: the early transfection comprised of pCI-expression vectors for six viral proteins and the later transfection comprised of 10 full-length viral (+)RNAs synthesized in vitro by T7 transcription (Yang et al., 2015).
3.1.3. RV—Attempts at Developing Fully Recombinant RG Systems
Many laboratories have attempted to establish fully recombinant RV RG systems, extrapolating from methods used to develop the MRV and orbivirus systems. The best documented efforts are described in the publication by Richards et al. (2013) and in the PhD theses of Richards (2012), Mlera (2013), and Wentzel (2014). These scientists attempted to develop RV infectious RNA systems using viral (+)RNAs produced in vitro by the polymerase activity of purified DLPs or by T7 transcription of viral cDNAs cloned into plasmids. Their work included transfecting viral transcripts of a variety of virus strains (eg, SA11, Wa, DS-1, RF) into a range of cell substrates (eg, MA104, COS-7, BSR, Caco-2, Vero, 293T), under optimized transfection conditions (Richards, 2012; Wentzel, 2014; Mlera, 2013). In their attempts to establish an infectious RNA synthesis, they also evaluated the double-transfection approach that, for orbiviruses, significantly enhanced the production of recombinant virus. In their trials, the early transfection step was varied to include the complete cohort of viral (+)RNAs or only those viral (+)RNAs encoding proteins essential for genome replication, DLP assembly, and viroplasm formation (Richards, 2012; Richards et al., 2013). In other trials, the early transfected RNAs were replaced with viral protein expression vectors, driven either by cellular RNA polymerase II (CMV promoter) or by T7 RNA polymerase expressed by recombinant fowlpox virus (FPV-T7) (Richards, 2012; Wentzel, 2014). None of the efforts resulted in the recovery of recombinant RV. A common observation has been that transfection of viral (+)RNAs results in extensive cellular cytopathic effects (CPE) (Richards, 2012; Richards et al., 2013; Wentzel, 2014; Mlera, 2013). The basis for the CPE is not clear, but may result from the recognition of the RNAs by PRRs, inducing innate immune pathways that lead to apoptosis and necrosis. The failure of transfected viral (+)RNAs to be efficiently translated may limit the expression of viral antagonists that are needed to suppress innate immune responses (Richards, 2012; Richards et al., 2013; Wentzel, 2014).
Although the efforts are poorly documented, several laboratories have attempted to develop a plasmid-based RV RG system. In our laboratory, we cloned full-length cDNAs of the 11 genome segments of SA11-4F RV into the same T7 transcription vectors used in the MRV RG system. These 11 RV plasmids were transfected into COS-7 cells previously infected with rDIs-T7pol. No recombinant RVs were recovered, even though in parallel transfection experiments using the 10-plasmid MRV RG system, we were able to recover recombinant MRV. We also reduced the number of RV plasmids transfected into cells by creating multicistronic bacmids; this allowed us to transfect cells with 4 plasmids instead of 11. This modified approach also failed to generate recombinant RV. In analyzing cells transfected with SA11-4F transcription vectors and infected with rDIs-T7pol, we detected little or no expression of RV proteins by either immunofluorescence or western blot assay. Although this raises questions as to whether the expression of RV proteins was adequate to launch infection, we noted that protein expression in transfected cells was similarly difficult to detect with the MRV RG system. In our experiments we also observed that cells transfected with RV plasmids and infected with rDIs-T7pol displayed considerable cytopathic effect (CPE) within 24 h, even though after transfection, the cells were maintained in media containing fetal bovine serum. The basis for the CPE is not fully understood, but could reflect the impact of rDIs-T7pol, transfection reagent, or viral transcripts or proteins produced by the expression vector. Further damage to the cells was noted when the cells were maintained in serum-free medium containing trypsin, conditions anticipated to promote spread of recombinant viruses.
In our attempts to develop a plasmid-based RV RG system, we found that viral protein expression was enhanced by adding extra G residues to the 5′-end of RV cDNAs inserted into T7 transcription vectors. Indeed, the level of protein expression by the modified vectors progressively increased as the number of G residues increased from one to three. Due to the importance of precise terminal sequences on RV RNAs for genome replication, it is doubtful that the modified vectors can be used in the development of functional plasmid-based RG systems. However, such modified vectors may prove useful for expressing viral proteins at levels adequate to launch the replication of viral (+)RNAs with authentic termini. Computer modeling indicated that the secondary structures of viral (+)RNAs with and without extra G residues are the same, containing 5′–3′ panhandles and associated stem loop structures. The fact that viral (+)RNAs with extra G residues were efficiently translated indicates that the highly ordered structural elements in the RNAs did not trigger innate immune responses that suppressed protein expression. From these results, we suggest that the failure of viral (+)RNAs made by unmodified T7 transcription vectors to drive efficient protein expression is not due to an innate immune response, but rather reflects an issue with the recruitment of the (+)RNAs into polysomes. In considering why authentic viral (+)RNAs are not efficiently translated, it is worth noting that RVs encoding defective NSP1 proteins induce very high levels of IFN expression in infected cells, yet these mutant viruses replicate quite well (Barro and Patton, 2005). It was also observed that transfection of BTV (+)RNAs into uninfected cells gives rise to strong innate immune responses (Wentzel, 2014), yet such methods are used to generate recombinant BTV (Boyce et al., 2008). Thus, although RV infection and RNA transfections may activate antiviral pathways, this does not provide sufficient information to explain the poor translation of RV (+)RNAs or the failure create a RV RG system.
3.2. Rotavirus Single-Gene Replacement Systems
The failure to establish a fully tractable recombinant RV RG system has led to efforts to generate single-gene RG systems that allow replacement of a single genome segment in a helper virus with an RNA of recombinant origin. Four such single-gene systems have been described, using four different helper viruses and a variety of selection methods to recover recombinant viruses from a background of parental viruses (Table 2.3.2).
Single-Gene Reverse Genetics Systems for the Reoviridae
The first single-gene replacement system described for RV was used to introduce a recombinant SA11 gene 4 (VP4) RNA into human KU RV (Komoto et al., 2006). The recombinant virus [recKU(SA11g4)] was produced by infecting COS-7 cells with rDIs-T7pol, which were then transfected with a plasmid containing a T7(SA11g4-cDNA)Rz cassette and infected with KU helper virus. To recover recombinant virus, the COS-7 cell lysates were passaged serially on MA104 cells in the presence of neutralizing antibody against KU VP4, followed by triple plaque purification. The basis for producing the recKU(SA11g4) reassortant came from earlier studies showing that KU(SA11g4) reassortant viruses grew better than the KU parent in cell culture. In a later study, the gene 4 RG system was used to replace one of the antibody neutralization epitopes in SA11 VP4 with the corresponding epitope in DS1 VP4 (Komoto et al., 2008). Interestingly, there have been no reports of gene 9 (VP7) single-gene replacement systems, despite the availability of numerous VP7 neutralizing antibody that could be used in selecting recombinant viruses.
3.2.2. Gene 8 (NSP2)
The most extensively used RV single-gene replacement system allows substitution of the mutant gene 8 (NSP2) RNA in the SA11 temperature-sensitive (ts) mutant tsE with recombinant RNA (Trask et al., 2010). A critical element of this system is that tsE NSP2 functions at low temperature (30°C), but not at elevated temperature (39°C). To generate SA11 viruses using this system, COS-7 cells were infected with rDIs-T7pol and then transfected with a plasmid containing a T7(g8-cDNA)Rz cassette that directs the synthesis of gene 8 transcripts that encode non-ts NSP2. Afterward, the cells were infected with the tsE helper virus and maintained at 30°C. To select for the recovery of viruses containing recombinant gene 8 RNA, COS-7 cell lysates were passaged serially ∼3-times at elevated temperature (39°C) in MA104 cells expressing an siRNA (MA104/g8D) that targets the tsE gene 8 RNA of the helper virus. Recombinant viruses were then isolated by triple plaque purification. The gene 8 RG system has been used to produce more than 20 recombinant viruses, including those that contain chimeric gene 8 RNAs (Trask et al., 2010), gene 8 RNAs with viral sequence duplications, and gene 8 RNAs with insertions of nonrotaviral sequences (Navarro et al., 2013) (Table 2.3.3). Notably, genetically stable recombinant RVs were produced that contained gene 8 RNAs with sequences for the FLAG tag, Hepatitis C Virus E2 epitope (HCV2), and Cricket Paralysis Virus internal ribosome entry site (CrPV IRES) inserted into the 3′ UTR (Navarro et al., 2013). The ability to engineer foreign sequences into the RV genome opens for the door for the potential development of these viruses as expression vectors, potentially leading to the generation of next generation RV vaccines that can induce protection against other enteric pathogens. Given that mutant strains of RVs have been described with ts lesions that map to nearly all eleven genome segments, it seems likely that ts-based RG systems could be developed for other RV genes (Criglar et al., 2011).
Contains a Pac I site at the NSP2 ORF/3’-UTR junction
3’-UTR contains a 25-nt sequence duplication
3’-UTR contains a 50-nt sequence duplication
3’-UTR contains a 100-nt sequence duplication
3’-UTR contains a 200-nt sequence duplication
3’-UTR contains a nonexpressing Flag sequence
3’-UTR contains a nonexpressing HCV2-epitope sequence
3’-UTR contains a nonexpressing Flag sequence and 25-nt g8 sequence duplication
3’-UTR contains a nonexpressing Flag sequence and 50-nt g8 sequence duplication
3’-UTR contains a nonexpressing HCV2-epitope sequence and 25-nt g8 sequence duplication
3’-UTR contains a nonexpressing HCV2-epitope sequence and 50-nt g8 sequence duplication
3’-UTR contains a CrPV IRES
3’-UTR contains a FLAG expression sequence downstream of a CrPV IRES
3’-UTR contains a recognition element for PP7 protein
3’-UTR contains two tandem recognition elements for PP7 protein
3’-UTR contains a recognition element for MS2 protein
3’-UTR contains recognition elements for MS2 and PP7 proteins
3’-UTR contains HIV TAR element
3.2.3. Gene 7 (NSP3)
RVs with rearranged genome segments, resulting from head-to-tail sequence duplications, can be generated in cell culture by serial passage of virus at high multiplication of infection (MOI) (Hundley et al., 1985; Patton et al., 2001; Arnold et al., 2012). Viruses with rearranged genome segments have also been detected in immunocompromised children (Hundley et al., 1987; Gault et al., 2001). Analysis of viral progeny produced in cells coinfected with wild type viruses and viruses with rearranged segments have indicated that there can be a selective advantage favoring packaging of the mutant RNAs (Troupin et al., 2011). Troupin et al. (2010) used this concept as the basis for developing a single-gene replacement system that allowed the introduction of recombinant gene 7 (NSP3) RNA into bovine RF RV. These investigators engineered two plasmids containing T7(g7-cDNA)Rz cassettes, one directing transcription of the rearranged gene 7 RNA of the human M1 strain and, the second, directing transcription of the rearranged gene 7 RNA segment of the human M3 strain (Gault et al., 2001). To generate recombinant viruses, the plasmids were transfected into COS-7 cells, which were later infected with rDIs-T7pol and RF viruses. Recombinant viruses contained in COS-7 infected cell lysates were amplified by serial passage 18-times at high MOI in MA104 cells. RF viruses containing M1 and M3 rearranged gene 7 (NSP3) RNAs were purified from 18th passage lysates by triple plaque purification. In addition to viruses containing rearranged gene 7 segments, viruses have been isolated that have rearranged gene 5 (NSP1), 6 (VP6), 8 (NSP2), 10 (NSP4), and 11 (NSP5/NSP6) segments (Patton et al., 2001; Arnold et al., 2012; Desselberger, 1996). Based on experience with the gene 7 reverse genetics system, it may be possible to use the packaging preference of rearranged RNAs to generate RG systems for these other segments as well. A limitation of this packaging preference is that it cannot be used to produce viruses with wild type genome segments; instead, it can only be used to generate viruses with rearranged segments.
3.2.4. Gene 10 (NSP4)
A single-gene replacement system has been described that allows modification of gene 10 (NSP4) of bovine CHLY RV (Yang et al., 2012) (see www.chinaagrisci.com/CN/abstract/abstract17260.shtml) In this system, MA104 cells were infected with CHLY helper virus and then transfected with two plasmids, one expressing T7 RNA polymerase and the other expressing gene 10 transcripts via a T7 promoter. Selection for CHLY virus containing a recombinant gene 10 RNA was accomplished by serial passage of transfected-infected cell lysates on MA104 cells expressing an siRNA that targeted the CHLY gene 10 RNA. Recombinant viruses were recovered by triple plaque purification, and the gene 10 sequences of the viruses were confirmed by sequencing. The gene 10 RG system is technically important because it indicates that the formation of recombinant viruses can be driven by T7 RNA polymerase expressed by a plasmid instead of by a vaccinia virus (rDIs-T7pol). Also, the gene 10 system suggests that viral transcripts need not be capped by vaccinia virus capping enzymes to serve as templates for the production of recombinant viruses. However, because the gene 10 transcripts may have been capped in trans by the VP3 capping enzymes of the CHLY helper virus, the results do not exclude the possibility that capped gene 10 transcripts are essential for recovery of recombinant virus.
3.3. Summary and Future Directions
The absence of a fully tractable recombinant RG system for RVs is a significant impediment to the pursuit of studies required to gain a clearer understanding of RV biology. The reasons for the lack of progress in creating such a system are not known, and indeed may be multifactorial, but a principal issue seems to be the inability of synthetic RV (+)RNAs to undergo/elicit efficient translation when introduced into uninfected cells. This holds true for in vitro synthesized (+)RNAs transfected into cells and for (+)RNAs made within cells by T7 transcription vectors. In contrast, when BTV (+)RNAs are transfected into cells, they are translated into readily detectable levels of viral proteins. In fact, viroplasm-like structures can be detected in many cells transfected with BTV (+)RNAs, quite unlike what is seen when cells are transfected with RV (+)RNAs. The 5′-terminal sequences of RV, BTV, and MRV (+)RNAs are different (GGC, GUU, and GCU, respectively); it is possible that these differences influence the ability of translation initiation factors in uninfected cells to engage the (+)RNAs and facilitate their introduction into polysomes. Only when the ends of RV (+)RNAs are modified, by the addition of a few extra 5′-terminal G-residues, does expression of RV proteins in transfected cells become easily detected. Unfortunately, the extra G residues create template (+)RNAs that are not effectively replicated in RG systems.
In the RV replication cycle, the outer capsid protein layer of virions is lost, producing DLPs that express low levels of viral (+)RNA at early times of infection. Translation of the (+)RNAs produces proteins that promote the formation of viroplasms, alter the host translational apparatus, and inhibit the development of an antiviral state. The synthesis of viral proteins during early times of infection when intracellular levels of viral (+)RNAs are low probably allows the virus to gain control of intracellular processes before viral RNAs accumulate to levels that are readily sensed by host PRRs and induce IFN and ISG expression. Given the large amount of RV (+)RNA introduced into cells as part of most RG experiments, it is probably unavoidable that the RNAs will be sensed by PRRs, triggering innate immune pathways that if not rapidly counteracted by viral antagonists will induce an antiviral state. The failure of transfected viral (+)RNAs to undergo efficient translation creates a situation where there is not only a lack of viral proteins needed to support genome replication and virus assembly, but also a lack of viral proteins necessary to control and redirect intracellular processes to establish an environment favorable for virus replication.
It is perplexing that transfection of synthetic (+)RNAs does not launch productive infection, while transfection of DLPs does. This contrast suggests that the DLP itself, or the process by which the particle produces transcripts, represents a variable that contributes importantly to initiation of infection. Since DLPs extrude nascent (+)RNAs in a 5′ to 3′ direction during transcription, the translational machinery may have an opportunity to recruit the 5′-end of the growing transcript into polysomes before the RNA is fully elongated. Such cotranscription-translation could prevent viral (+)RNAs from folding into structures that are poorly translated and/or highly immunogenic. It is also possible that DLPs localize to sites in the cytosol that favor translation of its (+)RNA products, or indeed, physically engage host factors that effect the structure of (+)RNAs or influence the loading of translation initiation factors onto the 5′-end of (+)RNAs. Clearly, studies are needed that focus on DLP transcription in the early stages of infection, and how this process connects to effective translation of viral (+)RNAs. Not only would these experiments provide important information of the RV biology, but could be key to the development of RV plasmid only-based RG systems. Given the importance of RG systems for probing mechanisms of virus virulence and pathogenesis, and for rational vaccine design, the continued pursuit of a fully tractable recombinant RV RG system is warranted.
The unrecognized efforts of Anne Kalbach, Zenobia Taraporewala, Shane Trask, Aitor Navarro, and Lauren Williamson to develop RV RG systems are greatly appreciated. This work was supported in part by the Intramural Research Program of the National Institute of Allergy and Infectious Diseases, National Institutes of Health.
Note added in proof
Recently, a reverse genetics system in which viral (+)RNAs are made in vivo by transfection of T7 transcription vectors into cells producing T7 RNA polymerase has also been established for bluetongue virus.