Enterovirus C recombination groups: RNA sequence similarity and the viral polymerase underpin sexual replication mechanisms

Abstract

Enteroviruses frequently recombine with one another in nature; however, it is unclear how viral replication machinery can distinguish between related and unrelated partners during recombination. We hypothesize that viral RNA recombination involves two parental RNA templates, nascent RNA products, and their dynamic interactions with the viral polymerase-a sexual replication strategy. When nascent RNA products move from one parental RNA template to another, RNA sequence similarity may be an important factor underpinning the mechanism and efficiency of recombination. To test this hypothesis, we focused on recombination between two related group C enteroviruses, poliovirus and Coxsackievirus A21 (CVA21), using bioinformatic, biological, and biochemical approaches. Bioinformatic analyses comparing 22 prototypical group C enteroviruses delineated four recombination groups where viruses in each group exhibit high RNA sequence and amino acid similarity in their polymerase genes. ClickSeq and ViReMa methods detect recombinant forms of poliovirus with P3 genes from CVA21, analogous to recombinant circulating vaccine-derived polioviruses (cVDPV). Biochemical assays show that poliovirus and CVA21 polymerases can detect mismatched base pairs as they traverse an extended primer grip surface adjacent to the active site. Mismatched base pairs in the -2 and -3 positions destabilize polymerase elongation complexes, consistent with the predicted role of RNA sequence similarity in recombination. Two subgroup-specific genetic elements, upstream open-reading frames (uORFs) and RNase L competitive inhibitor RNAs (RNase L ciRNAs), reinforce the existence and biological relevance of enterovirus C recombination groups. Altogether, our observations suggest that enterovirus RNA replication machinery can distinguish between related and unrelated partners during recombination.

Importance: Viral RNA recombination transforms live-attenuated polioviruses into neurovirulent circulating vaccine-derived polioviruses, complicating the planned eradication of poliovirus. When humans are co-infected with poliovirus and related non-polio enteroviruses, viral replication machinery can produce recombinant viruses. However, who recombines with whom? What factors determine whether two distinct viruses can produce recombinant progeny that are fit for transmission from person to person? In this study, we clarify which viruses recombine with one another in nature and further elucidate the mechanisms by which the viral polymerase distinguishes between related and unrelated RNA templates-a sexual form of replication. Understanding these mechanisms could lead to better strategies for virus control and/or eradication.

Keywords: RDRP; RNA recombination; RNA-dependent RNA polymerase; enterovirus; picornavirus; poliovirus; positive-strand RNA virus.

Fig 1

RNA sequence similarity delineates four polymerase groups/four EV-C recombination groups. SSE (49) and SimPlot (50) were used to compare RNA sequence similarity of 22 enterovirus C complete genome sequences representing each serotype (Table 2). RNA sequence similarity was compared using a 300-base sliding window with a one nt step and plotted from the 5’ to 3’ end of RNA genomes. Reference strains from each polymerase group included CVA1 (Pol Group I, red), EV-C104 (Pol Group II, green), EV-C105 (Pol Group III, yellow), and PV1 (Pol Group IV). Lines are color-coded for viruses in recombination groups I (red); II (green); III (yellow); and IV (blue) as indicated in Table 3. A colored line for the reference strain is not evident in each panel because the reference strain would have 100% sequence similarity to itself in SimPlot. As a result, Pol Group I, II, III, and IV panels have one less strain in their respective graphs. EV-C95 in Pol Group IV was not included in the analysis here due to having only a partial genome sequence. Note how well viruses in each subgroup exhibit increased RNA sequence similarity in the P2 and P3 regions of the genome when compared with viruses in other subgroups (grey lines for viruses in other subgroups for comparison in each panel).

Fig 2

Group C enterovirus sequence similarity pairwise comparisons. (A) RNA sequence identity of P1, P2, and P3 regions. (B) RNA and amino acid sequence identity of polymerase genes. Viruses in recombination groups I, II, III, and IV are indicated by diagonal lines. Heatmap indicates amounts of sequence similarity in pairwise comparisons.

Fig 3

Enterovirus C subspecies groups compared with recombination groups. Enterovirus subspecies groups C1, C2, and C3 are based on capsid gene sequence similarity (24). Regrouping subspecies C1, C2, and C3 enteroviruses by polymerase and RNA sequence similarity delineates four subgroups. This new categorization emphasizes polymerase homology and recombination groups rather than capsid genes and serotypes. All the viruses in subspecies C2 and C3 are in polymerase group 4/recombination group 4, with one exception (EV-C96). Viruses in subspecies C1, along with EV-C96, are subdivided into three polymerase/recombination groups.

Fig 4

Poliovirus recombination in the field and in the lab. (A) cVDPV2 from 2022. Diagram of Rockland NY 2022 cVDPV2 RNA genome (OP265178). SnapGene alignment of Rockland NY 2022 (OP265178) & OPV2 (AY184220.1) reveals a crossover site adjacent to the P2-P3 junction in the viral open-reading frame. This new strain of cVDPV2 arose in 2022, has been implicated in paralytic disease, and was detected by wastewater testing in the US/Canada/London and Israel (36). Location of the crossover site is in blue. (B) Recombination between poliovirus and CVA21. HeLa cells were co-transfected with PV∆GDD RNA + CVA21 sgRNA (four independent replicates). Virus recovered from co-transfected cells was amplified by one passage in HeLa cells, detected by plaque assay and sequenced to detect crossover sites (Blue arrow lines). Table S1 indicates locations of the most abundant PV-CVA21 crossover sites, with >250 cDNA reads.

Fig 5

Diagram of asexual and sexual RNA replication strategies. (A) Asexual RNA replication with one parental template. Replicative form (RF) and replicative intermediate (RI) RNAs. (B) Sexual RNA replication is the same as asexual RNA replication, except there are two parental templates (e.g., Polio & NPEV) with template-switching during negative-strand RNA synthesis. Two parental templates are copied into one chimeric (-) strand, which is then copied into multiple (+) strand progeny.

Fig 6

Enterovirus polymerases detect nascent RNA products and sequence similarity adjacent to the active site. (A) Polymerase elongation complex highlighting extended primer grip residues [K375/R376 (red), M392 (orange), L420 (yellow)] interacting with nascent RNA products adjacent to the active site [YGDD motif (fuchsia)]. Diagram shows nine bases of nascent RNA (N⁻¹ to N⁻⁹) in dsRNA products adjacent to the active site. (B) RNA sequence similarity of CVA21 and PV1 using a nine-base sliding window and a 1-base step across the viral ORFs. RNA sequence similarity (y-axis on the left) ranges from 0% to 100% for each 9-base window across the ORF. Mismatched base pairs (y-axis on the right) range from 0/9 to 9/9 for each 9-base window across the ORF.

Fig 7

Mismatched template-primer design and in vitro polymerase assays. (A) Template-primer pairs tethered by a polyethylene glycol (PEG) linker to minimize bias for short duplex formation. 17-base template (in blue); 6-residue PEG linker (in red); 4-base primer to mimic nascent RNA (in green). m0 RNA has four base-pairs with zero mismatched, whereas m1, m2, and m3 RNAs incorporate a single mismatched base at the first (N⁻¹), second (N⁻²), and third (N⁻³) positions, respectively. Ctrl RNA is the same as m0, but with a three nt deletion to separate it on gels. These template-primer pairs were designed to characterize mismatched base pairs across the extended primer grip of the polymerase. (B) Polymerase initiation assay with 5 μM polymerase, 0.75 μM m# RNA, 0.25 μM Ctrl RNA, and 100 μM GTP incubated from 0 to 120 min, showing +1 products from both precursor RNAs (R). Bar chart shows +1 and +2 initiation times for all m# RNAs on CVA21 wildtype and L420A polymerases, expressed relative to the initiation time for the internal Ctrl RNA in each individual reaction. P-values for comparisons with m0 in each data subgroup are indicated with * <0.05, ** <0.01, and ns being not significant. (C) Elongation complex stability experiment in which pre-initiated +1 or +2 complexes were diluted into high-salt buffer to prevent RNA rebinding and then periodically assayed to see what fraction of the complexes remain intact and can rapidly chase the +1 RNA to a + 2 product. Plots with single exponential curve fits show selective EC destabilization when mismatches are translocated into the N⁻² and N⁻³ sites but not the N⁻⁴ site, and the L420A polymerase mutation exacerbates these effects by additionally destabilizing all complexes. Overall, the stability of a N⁻² or N⁻³ mismatched RNA complex with L420A poliovirus 3D^pol is destabilized ~150 fold compared to a no-mismatch RNA with the wildtype polymerase. (D) Summary of elongation complex stability data for all four m# RNAs with both +1 and +2 initiation reactions on both wildtype and L420A mutants of PV and CVA21 polymerases. Exact values and fold effects are listed in Table 4 and P-values relative to m0 data are indicated as in panel B.

Fig 8

RNase L ciRNA is conserved in recombination group IV. (A) RNase L ciRNA, an RNA structure in the 3C ORF (71, 72). (B) RNA sequence similarity of the 3C gene using PV1 reference for SimPlot (100-base sliding window with 10-base step) and the representative viruses from polymerase groups I (red), II (green), III (yellow), and IV (blue). Location of RNase L ciRNA highlighting 5’ and 3’ portions (solid blue bars) and intervening sequence (dashed line). RNase L ciRNA sequence alignments (Fig. S2) show loop E motif polymorphisms incompatible with functional activity in Pol Group I, II, and III viruses.