|Home | About | Journals | Submit | Contact Us | Français|
Influenza A virus (IAV) is an unremitting virus that results in significant morbidity and mortality worldwide. Key to the viral life cycle is the RNA-dependent RNA polymerase (RdRp), a heterotrimeric complex responsible for both transcription and replication of the segmented genome. Here, we demonstrate that the viral polymerase utilizes a small RNA enhancer to regulate enzymatic activity and maintain stoichiometric balance of the viral genome. We demonstrate that IAV synthesizes small viral RNAs (svRNAs) that interact with the viral RdRp in order to promote genome replication in a segment-specific manner. svRNAs localize to the nucleus, the site of IAV replication, are synthesized from the positive-sense genomic intermediate, and interact within a novel RNA binding channel of the polymerase PA subunit. Synthetic svRNAs promote polymerase activity in vitro, while loss of svRNA inhibits viral RNA synthesis in a segment-specific manner. Taking these observations together, we mechanistically define svRNA as a small regulatory enhancer RNA, which functions to promote genome replication and maintain segment balance through allosteric modulation of polymerase activity.
Influenza A virus (IAV) is an upper respiratory pathogen that causes both seasonal and pandemic disease throughout the world (29). Each virion contains eight RNA segments of negative polarity that enter the cell with a functional RNA-dependent RNA polymerase (RdRp) tethered to the double-stranded ends of the looped RNA (39). The RdRp is a heterotrimeric complex formed by interactions between polymerase basic protein 1 (PB1), polymerase basic protein 2 (PB2), and polymerase acidic protein (PA) (48). This complex is responsible for both transcribing viral mRNA from the incoming negative-sense template and replicating the eight genomic segments for packaging and production of viral progeny (10). The virus strictly regulates the timing of these two processes as proper progression of the viral life cycle is dependent upon a seamless transition from transcription to replication (46). The mechanisms underlying the modulation of these dual processes have long been debated and still remain only partially understood—from the intricate details enabling the molecular switch within the polymerase to the overall impact this switch has on viral RNA synthesis.
Each viral RNA (vRNA) segment contains 12 and 13 conserved nucleotides at the 3′ and 5′ ends, respectively, which themselves demonstrate partial complementarity (8); this allows the ends of each segment to fold over and interact, forming what is referred to as the panhandle or corkscrew structure (11, 19). The heterotrimeric RdRp complex utilizes this double-stranded RNA (dsRNA) promoter for docking and subsequent initiation of RNA synthesis, with the proposed interaction domain residing within PB1 (5, 13, 22). The PB1 subunit of the RdRp contains the catalytic polymerase domain that allows for RNA synthesis of both viral genome and viral mRNA (4, 33). All three components work in concert to achieve mRNA synthesis, the primary RdRp activity following viral entry (46). The cap-binding abilities of PB2 allow for host mRNA cap snatching, providing the necessary primer for transcription (14). The PA subunit contains the endonuclease domain needed for cleavage of this cap from the host transcript, taking with it the first 10 to 13 nucleotides (nt) of a host mRNA (9, 53). In order to synthesize polyadenylated viral mRNA, the polymerase maintains association with the 5′ end of the vRNA while transcribing from the 3′ end (12, 23). This transcriptional strategy results in steric hindrance for the polymerase at the end of each segment, which encodes a polyuridine (U) stretch (40). This steric hindrance at the poly(U) causes the polymerase to stutter, thereby generating a polyadenylated tail (35–37). Upon sufficient mRNA synthesis, the RdRp transitions to an activity that biases replication, using a full-length positive-sense intermediate RNA to synthesize progeny vRNAs for each segment (7, 42). Several models exist for the mechanisms underlying the controlled modulation from transcriptase to replicase activity. Most models center upon an external component: de novo RdRp synthesis that acts in trans; concentrations of either nucleoprotein (NP), nucleotides, or substrates; or simply stabilization of the complementary vRNA (cRNA) intermediate (21, 25, 26, 28, 50–52). While these factors are clearly important for the viral life cycle, the precise contribution of each, as well as the manner in which each changes the activity of the polymerase, remains unknown. Furthermore, little is understood as to how the replicase-competent polymerase maintains stoichiometric balance of each of the eight viral genomic segments.
In an effort to determine if regulatory small RNAs played a role in modulating polymerase activity, we performed RNA deep sequencing of virus-infected cells (31). This led to the identification of IAV-derived small viral RNAs (svRNAs), which mapped to each of the 5′ ends of the vRNA segments and were implicated in modulating the switch from transcription to replication (31). This work was further supported in an independent study that found that polymerase activity was defined by the concentration of template sources, where the levels of 5′ vRNA ends resulted in a predominant replicase mode (28). Here, we define svRNA function as it relates to RdRp structure and conclude that it serves as a segment-specific small RNA enhancer to promote polymerase-mediated replication and maintain segment balance.
Human alveolar epithelial (A549) cells and human fibroblast (HEK293) cells were grown in complete medium containing Dulbecco's modified Eagle's medium (DMEM; Gibco), 10% fetal bovine serum (FBS), and 1% penicillin-streptomycin. Virus infections of A549 cells and HEK293 cells were performed as previously described (30). A549 cells were infected at a multiplicity of infection (MOI) of 1 for the times indicated in the figures with influenza A/Puerto Rico/8/34 virus and subjected to subcellular fractionation. In brief, infected and mock-treated cells were incubated in cold hypotonic buffer (10 mM HEPES [CellGro], 50 mM NaCl, 10 mM EDTA, 5 mM MgCl2, and fresh protease inhibitors) for 30 min on ice and subsequently lysed at 1:10 in 1% NP-40 and 1:5 in 50% glycerol (final concentrations of 0.1% and 10%, respectively) on ice for 15 min. After centrifugation at 5,000 rpm for 5 min at 4°C, the supernatant (cytoplasmic fraction) was separated and resuspended 1:1 in TRIzol reagent (Invitrogen) for standard RNA extraction. The nuclear pellets were washed three times in cold 50 mM NaCl and finally resuspended in TRIzol reagent (Invitrogen) for standard RNA extraction. HEK293 cells were infected at an MOI of 10 for the times indicated in the figures with recombinant influenza A/PR/8/34 virus, and total RNA was extracted by TRIzol reagent (Invitrogen) per the manufacturer's protocol. Isolated RNA from both infected A549 cells and HEK293 cells was analyzed by Northern blotting and primer extension. Total RNA isolated from infected HEK293 cells was also analyzed by reverse transcription-PCR (RT-PCR). In brief, total RNA was reverse transcribed with either a poly(A)-specific or poly(U)-specific primer, as previously described, and subjected to PCR for neuraminidase (NA), hemagglutinin (HA), and tubulin (primer sequences available upon request) (34). Influenza A viruses were propagated in 10-day old fertilized chicken eggs, and titers were determined by plaque assay in Madin-Darby canine kidney cells.
Detection of small viral RNAs (svRNAs) was performed by Northern blot analysis as previously described (31). Briefly, RNA was resolved by 12% denaturing PAGE, transferred to Hybond-NX nylon membrane (GE Healthcare) at 350 mA for 1 h, UV cross-linked at 200,000 μJ/cm2, blocked for 1 h at 65°C in 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and 7% SDS, and subsequently probed with radiolabeled oligonucleotides (sequences available upon request). Radiolabeling was performed with T4 polynucleotide kinase (Invitrogen) and [γ-32P]ATP (Perkin-Elmer), and probes were purified by Sephadex G-24 columns (GE Healthcare). All depicted Northern blots are representative results from multiple experiments. svRNA production from viral components was assayed as previously described (31). In brief, HEK293 cells were cotransfected (1 μg of DNA each) with the indicated combinations of bidirectional plasmids depicted in Fig. 1A representing each of the eight viral segments (a kind gift from P. Palese, Mount Sinai School of Medicine, New York) and either an RNA polymerase I (pPol I)-driven segment 8 expression plasmid, a pCAGGS expression plasmid for nonstructural protein 1 (NS1) (both kind gifts from A. Garcia-Sastre, Mount Sinai School of Medicine, New York), or a pcDNA3.0 expression plasmid for a tandem affinity purification (TAP)-tagged nuclear export protein (NEP) (a kind gift from E. Fodor, University of Oxford, Oxford, United Kingdom) (38).
For NEP dependency, viral polymerase was expressed via pCAGGS-based plasmids for PB2, PB1, PA, and NP, and a negative-sense NA viral RNA template was expressed via a pPol I-driven plasmid (all kind gifts from A. Garcia-Sastre, Mount Sinai School of Medicine, New York, NY). HEK293 cells were cotransfected with 0.5 μg of each of the polymerase components, 4 μg of pPol I NA vRNA template, and either 2 μg of carrier plasmid (pcDNA3.0) or 100 ng, 0.5 μg, or 2 μg of TAP-NEP (with carrier plasmid as appropriate). For expression of NEP truncations, a pcDNA3.0-based plasmid minilibrary containing various TAP-NEPs was used (a kind gift from E. Fodor, University of Oxford, Oxford, United Kingdom) (38). HEK293 cells were cotransfected with 2 μg of each of the polymerase components, the appropriate NEP construct, and a pPol I HA vRNA template.
For cRNA expression, a positive-sense NA cRNA template was expressed via a pPol I-driven plasmid (a kind gift from F. Vreede and E. Fodor, University of Oxford, Oxford, United Kingdom) (51). HEK293 cells were cotransfected with 0.5 μg of each of the polymerase components, 4 μg of either pPol I NA vRNA template or pPol I NA cRNA template, and either 2 μg of pcDNA3.0 or 2 μg of TAP-NEP. For poly(U) cRNA expression, poly(U) track mutations in pPol I NA cRNA [and the corresponding poly(A)track mutations in pPol I NA vRNA] were generated using site-directed mutagenesis (Quick Change; Stratagene) (primer sequences available upon request). HEK293 cells were cotransfected with 1 μg of each of the polymerase components and either 4 μg of pPol I NA wild-type (WT) cRNA template or pPol I NA poly(U) cRNA template.
All transfections were performed with Lipofectamine 2000 (Invitrogen) in OptiMEM medium (GIBCO) and harvested at 24 h posttransfection unless otherwise indicated. Cell pellets were equally divided and subjected to total RNA extraction by standard TRIzol protocol (Invitrogen) or to protein extraction for Western blot analysis. Total RNA was subsequently DNase treated (DNase I; Roche) and analyzed by Northern blotting for pan-svRNA or universal-svRNA [for poly(U) experiments] and primer extension. Total RNA from HEK293 cells transfected with either pPol I NA WT or poly(U) cRNA plasmids was also analyzed by RT-PCR using oligo(dT) (Invitrogen) for NA and NP mRNAs (primer sequences available upon request).
For analysis of mRNA, cRNA, and vRNA synthesis by the polymerase, total RNA was DNase treated and subjected to RT-PCR using previously published primers (50). The resulting cDNA was resolved by 6% denaturing PAGE, transferred to Hybond-NX nylon membrane (GE Healthcare) at 350 mA for 1 h, UV cross-linked at 200,000 μJ/cm2, and viewed by autoradiogram. All depicted primer extensions are representative results from multiple experiments.
Whole-cell extracts were obtained and analyzed by Western blotting as previously described (31). Membranes were probed with anti-FLAG monoclonal antibody (Sigma) or influenza A/PR8/34 virus monoclonal NP and polyclonal NEP (BEI Resources) at 1 μg/ml in 5% dried milk in phosphate-buffered saline (PBS). Incubations were performed overnight at 4°C; samples were subsequently washed three times for 5 min each time in PBS and incubated with peroxidase-conjugated sheep anti-mouse or anti-rabbit antibody (GE) at a dilution of 1:5,000 for 1 h at room temperature. The membrane was washed again as before and visualized with an enhanced chemiluminescence (ECL) detection system as recommended by the manufacturer (Millipore).
Deep sequencing was performed on sucrose-purified virus stocks. Briefly, influenza A/PR/8/34 virus was pelleted via a 30% sucrose cushion and centrifugation at 25,000 rpm for 2 h. The viral pellet was resuspended in phosphate-buffered saline, and total RNA was isolated by standard TRIzol (Invitrogen) extraction. Small RNAs 19 to 24 nt in length were isolated from the total RNA by 12% PAGE, and cDNA libraries were generated as previously described (32, 43). Analysis of deep-sequencing data was performed as previously described (31).
Synthetic svRNAs were synthesized as previously described (31, 54), and sequences are available upon request. HEK293 cells were transfected with a total of 12 μg of Flag-tagged influenza A virus polymerase protein expression plasmids (PB2, PB1, and/or PA) (kind gifts from P. Palese and M. Shaw, Mount Sinai School of Medicine, New York). Truncations of PB1 and PA were generated by PCR-based cloning into a pFlag vector using NotI and XbaI restriction sites (New England BioLabs). PCR was performed using High Fidelity PCR Mastermix (Roche); PCR products were purified by gel extraction (Qiagen) and ligated into the cut pFlag vector with T4 DNA Ligase (New England BioLabs). Point mutations in PA were generated using site-directed mutagenesis (Quick Change; Stratagene). Protein extracts were immunoprecipitated as previously described (31) using 800 μg of whole-cell extract and 0.5 μg of unlabeled 5′-triphosphate containing synthetic svRNA. Total RNA was isolated via standard TRIzol (Invitrogen) extraction and analyzed by Northern blotting.
Purified trimeric A/chicken/Nanchang/3-120/01 polymerase was generated via baculovirus overexpression and TAP purification as previously described (1, 2). The 47-nt HA cRNA minireplicon (containing both noncoding regions joined by a short linker) was generated by T7-based in vitro transcription (Ambion) from an annealed double-stranded DNA oligonucleotide containing a T7 promoter (see Table S1 in the supplemental material for the sequence). T7-generated cRNA was DNase treated per the supplied protocol and isolated via standard TRIzol (Invitrogen) extraction. Polymerase assays were performed as previously described. In brief, a 10-μl reaction (rxn) mixture containing transcription buffer (25 mM Tris-Cl, pH 7.5, 100 mM KCl, 5 mM MgCl2, 0.1 mM EDTA, 2 mM dithiothreitol [DTT], 0.25% NP-40, 12.5% glycerol), IAV polymerase complex (500 ng/rxn), 0.25 U/μl RNase Out (Invitrogen), 0.3 mM ApG (Biosynthesis), 1.6 μM T7-generated HA cRNA minireplicon, 1.6 μM synthetic svRNA, 0.16 μM [α-32P]UTP (0.5 μl/rxn) (3,000 Ci/mmole; Perkin Elmer), and 0.5 mM concentrations of the nucleoside triphosphates (NTPs; Ambion) was incubated at 30°C for 1 h and heat inactivated at 70°C for 10 min. For dose dependency, a 0.16 μM, 1.6 μM, or 16 μM concentration of either a scrambled RNA control or synthetic svRNA was used. As a control, a full-length cRNA minireplicon was generated via T7 transcription in the presence of [α-32P]UTP. Products were resolved by 6% denaturing PAGE, transferred to Hybond-NX nylon membrane (GE Healthcare) at 350 mA for 1 h, UV cross-linked at 200,000 μJ/cm2, and viewed by autoradiogram. All depicted in vitro assays are representative results from multiple experiments.
svRNAs are single-stranded RNAs, greater than 21 nucleotides (nt) in length, which originate from the 5′ end of the negative-sense viral genome and accumulate to levels greater than 100,000 copies per cell (31, 49). Synthesis of svRNA was found to require a functional RdRp, the nucleoprotein (NP), and, surprisingly, the nuclear export protein or nonstructural protein 2 (NEP/NS2, here referred to as NEP) (31). To better understand NEP dependency, we directly compared the role of NS1, NEP, and/or segment 8 vRNA (v8) in the generation of svRNA by utilizing a bidirectional plasmid expression system to evaluate the contribution of each of the IAV proteins (Fig. 1A). These bidirectional plasmids allow for expression of both the vRNA and protein for each individual segment (17). To confirm the NEP-dependent production of svRNA in trans, we expressed the eight bidirectional plasmids and subsequently excluded each segment individually (Fig. 1A). As expected, Northern blot analysis demonstrated robust production of svRNA in the presence of all eight segments; loss of hemagglutinin (HA), neuraminidase (NA), and the matrix proteins (M1 and M2) had no significant impact on svRNA levels, while loss of the RdRp or NP prevented svRNA generation. Furthermore, loss of segment 8, encoding both nonstructural protein 1 (NS1) and NEP, resulted in a loss of svRNA (Fig. 1A). To delineate segment 8 dependency, we expressed vRNA (v8, which will also produce NS1 and NEP in the context of a functional RdRp) or NS1 or NEP individually. These results confirmed that NEP was the only segment 8 component capable of rescuing the generation of svRNA, as previously described (31). These data suggest that a fully functional polymerase, which requires NP for proper elongation during RNA synthesis (18), and NEP are necessary for generation of svRNA.
To better understand the need for NEP in svRNA production, we assessed the effect of NEP expression on RdRp activity in the context of a polymerase reconstitution assay. Cells expressing RdRp and NP, as well as an NA vRNA template, were monitored for RNA synthesis in the presence of increasing amounts of NEP (Fig. 1B). Unlike the bidirectional system, this assay contains only a single genome template source, allowing for the measurement of segment-specific vRNA, cRNA, and possibly svRNA synthesis from segment 6 (encoding NA). NEP expression enhanced the synthesis of cRNA from the vRNA template in a concentration-dependent manner, in concordance with recent work indicating NEP-dependent modulation of polymerase function (24, 38). Furthermore, Northern blot analysis of these samples demonstrated a correlation between enhanced cRNA synthesis and NA-specific svRNA expression (Fig. 2B). These results implicated NEP in the synthesis of cRNA, which is subsequently required for svRNA or, alternatively, for svRNA that is essential for the generation of cRNA.
To ascertain whether NEP dependence is a direct result of NEP's ability to enhance cRNA or svRNA synthesis, we next sought to bypass the cRNA requirement by directly supplying this template in the polymerase reconstitution assay. To this end, we compared NEP dependency of svRNA in cells expressing RdRp, NP, and either an NA vRNA or cRNA template (Fig. 1C). These results demonstrated that svRNA was NEP dependent in the context of vRNA but NEP independent with cRNA (Fig. 1C). Taken together, these data suggest that cRNA serves as a template for the synthesis of svRNA and that the basis of NEP dependence is in catalyzing cRNA production.
We next sought to explore the relationship between NEP and cRNA expression through the utilization of NEP truncation mutants in a polymerase reconstitution assay. RNA synthesis was monitored in cells expressing RdRp, NP, and an HA vRNA template in the absence or presence of various NEP truncations (Fig. 2A). Full-length NEP, in addition to truncations lacking the nuclear export sequence (NES), was capable of driving enhanced cRNA synthesis and an increase in svRNA production (Fig. 2B). In contrast, truncations that contained the nuclear export sequence yet lacked an intact C-terminal domain were unable to promote cRNA synthesis or svRNA production. These data corroborate the original findings of Robb et al. and further demonstrate that the C-terminal domain of NEP is not only responsible for enhancing cRNA synthesis by the RdRp but also necessary for promoting svRNA production in these assays (38).
In an effort to determine the cellular distribution of svRNA, we performed subcellular fractionation of IAV-infected alveolar epithelial cells (Fig. 3A). While modest contamination was evident by Northern blot analysis, svRNAs were detected predominantly in the nucleus as early as 8 h postinfection (hpi), and their localization remained consistent throughout the course of infection. In contrast, NP vRNA slowly accumulated in the cytoplasm, with levels peaking by 16 hpi. These data indicate that svRNA function is predominantly nuclear and suggest that these small RNAs are not packaged during virus egress.
To confirm the restricted nuclear localization of svRNAs suggested by cellular fractionation, we analyzed cell-free virus to ensure that svRNAs were not also found in budding virions. To this end, we performed deep sequencing on purified IAV stocks, comparing the relative amounts of svRNA to that of total vRNA reads (Fig. 3B). Comparing the total number of IAV segment 1 reads from virus-infected cells found that svRNA accounted for ~28% of total reads, whereas the same analysis from virion RNA accounted for ~1.8% (see Table S1 in the supplemental material). Taken together, these data further corroborate the hypothesis that svRNAs are maintained predominantly in the nucleus. Furthermore, with the knowledge that svRNAs are critical to the production of vRNA (31), these data imply that svRNAs are an essential component of the virus's replication machinery.
In an effort to elucidate the molecular mechanism by which svRNA impacts cRNA-dependent vRNA synthesis, we first sought to determine which RdRp subunit associated with this 22-nt RNA. To this end, we utilized synthetic svRNAs containing a 5′ triphosphate to assay RdRp binding in vitro. Cells expressing tagged polymerase subunits or tagged green fluorescent protein (GFP) were immunoprecipitated (IP) in the presence of the synthetic svRNA to measure direct association (Fig. 4A). Antibody-mediated, agarose-dependent IPs demonstrated protein-specific interactions with svRNA, as ascertained by the lack of interaction with GFP and a strong association with the PB1/PB2/PA (RdRp) trimeric complex or the heterodimeric combination of PB1 and PA (Fig. 4A).
To map the interaction between svRNA and the PB1/PA heterodimer, we generated truncations to determine the minimal domains required for association (Fig. 4B). These results demonstrated that the minimal components required to maintain svRNA association include the PB1 interaction domain of PA (residues 155 to 716) in addition to the catalytic domain of PB1 (residues 1 to 500) (Fig. 4B) (6, 41, 45). In contrast, neither the PB2 interaction domain of PB1 nor the endonuclease domain of PA was required for svRNA association (Fig. 4B) (9, 53). Interestingly, truncations of both PA and PB1 that retain their heterodimeric structures were sufficient to mediate interaction with svRNA (15). Given the requirement for the PB1/PA heterodimeric structure for svRNA association, we examined the reported partial crystal structure of these proteins to gain insights into putative interactions (16). X-ray crystallography of the PA/PB1 complex (residues 257 to 716 and 1 to 25, respectively) revealed a cleft on the PA surface, which was described to include the flexible α3-α4 loop of PA, a highly basic 25-Å cavity containing conserved basic residues K328, K539, and R566 and ending in a smaller 14-Å channel within the C terminus of PA (16). In order to test if the putative PA RNA binding cleft mediated the interaction between the RdRp and svRNA, we mutated the conserved basic residues and performed protein-RNA IPs (Fig. 4C). A single point mutation at R566 dramatically reduced binding of the PB1 and PA heterodimer to svRNA, while single mutations at K328 and K539 had minimal individualistic effects. In contrast, double mutations of K328 and K539 substantially reduced binding of the heterodimer to svRNA, as did all other double and triple mutations assayed. Taken together, these binding assays demonstrate that the PA RNA binding cleft is critical to the svRNA-RdRp interaction and suggest that R566 is a key residue in the loading of svRNA into PA, with residues K328 and K539 playing minor roles in mediating this interaction. Interestingly, mutation of the PA RNA binding cleft rendered the polymerase incapable of both transcription and replication as assayed by polymerase reconstitution (data not shown), suggesting that this cleft is also critical to overall polymerase activity.
As IAV produces svRNAs from cRNA (Fig. 1C), we next sought to investigate if svRNAs could act upon this template to modulate RdRp activity. To this end we performed in vitro RdRp assays, utilizing purified trimeric polymerase, truncated HA cRNA, and synthetic svRNAs (Fig. 5A). For this assay, we also utilized an NP-independent minireplicon to prevent RdRp from generating svRNA de novo (Fig. 1A). ApG-primed in vitro RdRp reactions resulted in little production of full-length vRNA from the supplied cRNA template or from reaction mixtures containing increasing amounts of a scrambled control RNA. In contrast, addition of equimolar amounts of a 22-nt synthetic HA svRNA-cRNA template resulted in enhanced RdRp activity, as well as synthesis of full-length product. Interestingly, this in vitro assay did result in aberrant 30- to 35-nt products, reminiscent of svRNA production in the absence of NP (Fig. 1A), suggesting that the baseline activity of the RdRp can result in small RNA production.
To ascertain the manner in which svRNAs function to promote full-length vRNA synthesis by the RdRp, we utilized both short synthetic svRNAs as well as synthetic svRNAs having a methylated 3′ hydroxyl group (Fig. 5B, denoted with an asterisk), such that extension from the svRNA oligomer cannot occur. Administration of the first 13 nt of svRNA was sufficient to mediate full-length vRNA synthesis and elongation, demonstrating that the enhancer function of svRNA lies within the conserved sequence present in each svRNA population. Interestingly, the 3′ hydroxyl svRNA mutants blocked production of the aberrant 30- to 40-nt products without impacting full-length vRNA synthesis. This suggests that, in the context of this in vitro assay, the smaller products represent nonspecific, primer-dependent elongation of the scrambled and svRNA mimetics. In contrast, as production of full-length vRNA was not impacted by the lack of a free 3′ hydroxyl group, we can conclude that svRNA acts as a guide for RdRp-dependent genomic amplification. Taken together, these data suggest that svRNAs modulate polymerase activity via allosteric, rather than primer-mediated, enhancement of vRNA synthesis and that this activity is largely coordinated by the first 13 conserved nucleotides of svRNA.
Inhibition of full-length svRNA was previously demonstrated to block vRNA synthesis in a segment-specific manner, a phenotype also observed upon mutation of the poly(U) track necessary for polyadenylation of IAV transcripts (31, 34). Upon replacement of the poly(U) track with a poly(A) track, IAV demonstrated deficiencies in both segment-specific mRNA export and vRNA synthesis (34). To determine if the loss of vRNA synthesis was due to the lack of svRNA production from the corresponding mutation in the cRNA promoter, we generated an NA cRNA plasmid bearing the same poly(U) track (Fig. 6A, Poly U). svRNA production was monitored in cells expressing RdRp and NP, in addition to either wild-type cRNA (WT) or poly(U) cRNA with a universal svRNA probe that hybridized to only the shared 5′ 13 nt nucleotides of svRNA. Strikingly, svRNA synthesis was abrogated by loss of the canonical poly(A) track at both 24 and 48 h posttransfection compared to cells expressing WT cRNA, suggesting that the poly(A) track is essential to svRNA production by the RdRp (Fig. 6B).
To investigate the direct contribution of svRNAs to viral replication, we rescued a recombinant virus carrying the mutant poly(U) NA segment [referred to as poly(U) IAV]. Cells infected with either WT IAV or recombinant poly(U) IAV were monitored for both vRNA synthesis and svRNA production. Northern blot analysis revealed diminished svRNA production for poly(U) IAV compared to WT infection as loss of NA-specific svRNA would result in an overall decrease in the total svRNA population (Fig. 6C). In agreement with the hypothesis that svRNA function is restricted to vRNA production, the poly(U) IAV demonstrated no defects at the level of NA mRNA (Fig. 6B and andD).D). In contrast to mRNA levels, primer extension assays demonstrated that poly(U) IAV was incapable of robust synthesis of NA-specific vRNA in comparison to WT (Fig. 6E). Furthermore, loss of NA svRNA, while impacting NA vRNA, did not impact the levels of vRNA for HA, NP, and NS in comparison to WT virus during infection (Fig. 6E). These data suggest that loss of NA svRNA impairs the ability of the IAV RdRp to regulate vRNA synthesis during infection and that the role of svRNA in promoting vRNA synthesis is segment specific.
Modulation of IAV polymerase activity is a contested issue, with many seemingly disparate mechanisms having been proposed as to how the virus transitions from primary transcription to genome replication. Here, we demonstrate that IAV-derived svRNAs function to promote replicase activity and template elongation in a segment-specific manner by interacting with the RNA binding cleft within the PA subunit, allowing the polymerase to maintain the stoichiometric balance of all eight genomic segments. This function, as an allosteric small RNA enhancer, sets svRNAs apart from other classes of small regulatory RNAs. While its activity is somewhat reminiscent of that of microRNA (miRNA) miR-122 for hepatitis C virus (HCV), whereby miR-122 directly interacts with the 5′ untranslated region (UTR) of HCV to promote translation and viral replication, svRNAs are produced from the viral genome and directly interact with the viral polymerase (20). Apart from miRNAs, svRNAs could be, and indeed have been, compared to the leader RNAs of paramyxoviruses and rhabdoviruses (49). However, this also seems inappropriate as leader RNAs have not been found to directly associate with RdRps and generally exist for both the 5′ and 3′ ends of the viral RNA genome (3). As svRNAs have been found to directly interact with the IAV RdRp and do not require incorporation into nascent vRNAs for activity, they are unlikely to function as leader RNAs and most likely behave as segment-specific enhancers.
Given that the cRNA intermediate could serve as a template for svRNA synthesis, it is likely that svRNA production relies solely on viral proteins. While the roles of PB2, PB1, and PA could be extrapolated from their known viral functions, the need for NP and NEP was less clear. The dependence on NP may be a result of needing a functional polymerase with the ability to elongate products (18); likewise, as svRNA is approximately the same length as an RNA molecule wrapped over an NP monomer, it is also possible that NP serves as a size marker for generation of svRNA of the appropriate length (26). In addition, NEP has been ascribed a number of distinct functions, the primary activities being nuclear export of progeny vRNPs, the enhancement of cRNA synthesis, and, most recently, determining host tropism (27, 38). As dose dependency and truncation analysis demonstrated a correlation between svRNA production and nuclear export-independent cRNA synthesis, we conclude that the NEP requirement is the result of inefficient cRNA concentrations to template svRNA production.
Following cRNA-dependent svRNA synthesis, this small regulatory RNA loads into the RdRp to promote replicase activity. Truncation analysis and available structural information for the PA/PB1 heterodimer revealed a putative RNA binding cleft, which could accommodate svRNA. Subsequent mutation of these key basic residues abrogated binding to synthetic svRNA, demonstrating the crucial role this RNA binding cleft plays in mediating polymerase-svRNA interactions. As neither the 5′ triphosphate nor nucleotides beyond position 13 were required for enhancing polymerase activity, it is likely that the sequence and structure of the 5′ end of svRNA are sufficient to mediate its function. While sequence-specific interactions between proteins and dsRNAs are rare, given the uniform face of the minor groove and the depth of the major groove, the binding of the corkscrew structure by the RdRp has been proposed to be sequence specific, with interactions occurring at the ACAA at positions 8 to 11 (44, 47). This suggests that the interactions between the PA RNA binding cleft and svRNA involve both sequence and structural elements, which may alter the overall conformation of the polymerase. Furthermore, as the PA RNA binding cleft mutants were incapable of both transcription and replication, it is possible that substrate-specific occupancy may determine polymerase activity. Interactions between the host 5′ cap and the cleft may promote viral transcription, whereas interactions with svRNA promote replication. This is in concordance with work demonstrating that changes in the concentrations of both the 5′ host cap and 5′ vRNA end have the ability to bias polymerase activity to transcription and replication, respectively (28).
Loading of svRNA into the PA RNA binding cleft likely modifies the polymerase to become replication competent, promoting full-length vRNA synthesis from the cRNA intermediate. While only trace amounts of vRNA synthesis can be detected in the absence of svRNA, administration of various synthetic svRNAs demonstrates enhancement of full-length vRNA synthesis. As these in vitro reactions lack recombinant NP, it is not surprising that de novo svRNA synthesis is inhibited, which relies on NP for proper production. Therefore, synthetic svRNAs must be supplied in order to detect vRNA synthesis from the cRNA template. This is in stark contrast to the RdRp reconstitution assays, in which basal levels of svRNA production may be sufficient to mediate vRNA synthesis in the absence of NEP. In an NP-free system, svRNA is capable of promoting full-length vRNA synthesis, suggesting that it may function to enhance the processivity of the RdRp. While NP has been characterized to promote polymerase elongation (18), it does so in a non-segment-specific manner, whereas svRNA can promote processivity specifically for its cognate cRNP, ensuring proper vRNA synthesis for a given RNP. Although the short synthetic svRNA is capable of promoting full-length vRNA synthesis in vitro, it may not be beneficial to the virus to utilize such a promiscuous regulatory system in vivo.
As in vitro analysis demonstrated that svRNAs were capable of promoting full-length vRNA synthesis by the RdRp, we next wanted to investigate the contribution of svRNA to viral infection. To ascertain this, we engineered an IAV strain incapable of generating svRNA for a single genomic segment. Loss of the ability to produce svRNA from the viral promoter impaired vRNA synthesis in a segment-specific manner, such that the lack of NA svRNA resulted in a loss of NA vRNA without impacting the other viral segments. These data suggest that in addition to regulating vRNA synthesis by the polymerase, svRNA also functions to maintain the stoichiometric balance of each of the eight segments. In this way the IAV polymerase can produce equivalent amounts of each genomic segment to ensure that all eight vRNAs can be packaged into progeny virions.
Here, we propose a model in which PA is loaded with svRNA to convert the RdRp from a transcriptase into a replicase. Upon infection, the incoming polymerase, unoccupied by svRNA but associated with the vRNA promoter, acts predominantly as a transcriptase, utilizing host mRNAs as primers for synthesis of capped viral transcripts. As viral transcripts of NEP accumulate, cRNA is synthesized in what has been proposed to be a stochastic manner (27, 38, 50), with newly synthesized RdRp serving to stabilize the genomic intermediate (Fig. 7). Nascent cRNA provides a template for svRNA synthesis which, once synthesized, associates with the basic cleft of PA (Fig. 7). svRNA-loaded PA then allows the RdRp to interact with the cognate cRNA template and synthesize full-length vRNA (Fig. 7). In this model, svRNA loading would generate segment-specific replication-competent RdRp populations as the 3′ complementarity of each svRNA would be distinct for a cRNA template. This would ensure the maintenance of the stoichiometric balance between each genomic segment as one RdRp complex would be devoted to the replication of one genomic segment. Taking these observations together, we have demonstrated that svRNA interacts with the RNA binding cleft within PA to promote full-length synthesis of vRNA in an allosteric manner, switching the polymerase from a transcriptase to a replicase. This finding is the first example of a small RNA capable of controlling RNA-dependent RNA polymerase activity and suggests that svRNA is a master regulator of IAV replication.
We thank the Palese, García-Sastre, and Shaw labs (Mount Sinai School of Medicine, New York, NY) for reagents and comments during the course of this work.
This work was supported by the National Institutes of Health (grant number A1093571-01 to B.R.T.) and by the New York Influenza Center of Excellence (NIAID HHSN266200700008C), which provides support for B.K. and S.A. J.T.P was supported, in part, by the Ruth L. Kirschstein NRSA fellowship.
Published ahead of print 3 October 2012
Supplemental material for this article may be found at http://jvi.asm.org/.