Identification of a conserved RNA secondary structure in the CHIKV 3’UTR

The CHIKV 3’UTR contains DRs that occur in lineage-specific patterns [6,18]. The four major CHIKV lineages are the epidemic viruses of the IOL, ECSA, and Asian lineages and the enzootic virus of the Western African lineage. The genomes of the IOL, ECSA and West African lineages contain shorter 3’UTRs with two DR element types (DR1, two copies; and DR2, three copies). In contrast, the Asian lineage contains a longer 3’UTR that has accumulated insertions and point mutations around DR1 and DR2 [referred to as DR (1+2)], and the duplication of an entirely new region (DR3) [18]. Interestingly, some American CHIKV isolates derived from the Asian lineage, such as those in the Caribbean, contain an additional 177 nucleotide (nt) duplication [DR (1+2)b’], displaying the longest 3’UTR described among CHIKV isolates (Fig 1A) [23].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. RNA secondary structures in the CHIKV 3’UTR.

(A) Schematic representation of the genomic organization of CHIKV (left) and its 3’UTR (right) for the ECSA, IOL, and Asian epidemic lineages and the Caribbean strain. Colored boxes represent different DRs. The DR3 region corresponding to SLY is colored in pink. (B) Sequence alignment (middle) and nucleotide conservation plot (bottom, in gray) of the 80 nt-long DR3 region that folds into SLYa and SLYb, for ten representative ECSA, IOL, and Asian lineages isolates. Pink arcs and boxes indicate the nucleotide pairings that form the stems of each Y. Conserved nucleotides are in black, and variable nucleotides are in red, blue, green and orange. The black box indicates the nucleotide position 45/49 that covariates among lineages. (C) Consensus RNA secondary structures predicted for SLYa and SLYb. Left, minimum-free energy foldings with the corresponding consensus sequences. Middle, base-pair probabilities. For unpaired regions, the color represents the probability of being unpaired. Right, the SHAPE-constrained thermodynamic model for SLYa and SLYb. Nucleotide positions are numbered every 10 nucleotides. SLY, stem-loop Y; S1, stem 1; S2, stem 2; S3, stem 3; L1, loop 1; L2, loop 2; B1, bulge 1; B2, bulge 2; B3, bulge 3; B4, bulge 4.

https://doi.org/10.1371/journal.ppat.1011352.g001

To identify conserved RNA secondary structures in the CHIKV 3’UTR, we first aligned the 3’UTR sequences of 112 natural isolates from the IOL, ECSA and Asian epidemic lineages, and used RNA folding analysis to predict their 3’UTR secondary structures. We identified at the end of the 3’UTR an 80 nt-long region within DR3a and DR3b that folds into a stable Y-shaped RNA structure (SLY henceforth). (Fig 1B and 1C). The SLY structure was present in a single copy in the 3’UTRs of ECSA and IOL lineages but duplicated in the Asian lineage. Each SLY comprises three stems: stem 1 (S1) exposes a right loop (L1) and is interrupted by one bulge (B1); stem 2 (S2) exposes a left loop (L2); and stem 3 (S3) leaves a central bulge (B2) and is interrupted by two other bulges (B3 and B4) (Fig 1C, left). SLYa and SLYb are highly stable and have similar predicted minimum free energies (−36.48 and −35.46 kcal/mol, respectively). In addition, according to modeling, base-pair probabilities for nucleotides forming both SLYs are high (Fig 1C, middle).

Next, we analyzed the SLY’s structure and nucleotide conservation, finding that almost all nucleotides involved in base-pairing are conserved (Fig 1B, bottom plot and Fig 1C, left). However, sequence variation appears at unpaired nucleotides forming loops or bulges (Fig 1B, bottom plot). For example, five nucleotides in the SLYa sequence varied among the analyzed isolates: one in a loop (nt 27 in L2), three in bulges (nt 54 in B1, nt 61 in B2, and nt 67 in B3), and one (nt 45) in S1. Half of the analyzed sequences had an adenosine at position 45, pairing uracil in position 49. In contrast, the other half had guanosine, maintaining the pairing with U49. The preservation of the overall SLY structure despite this nucleotide variation represents strong evidence for S1 formation. In SLYb, all nucleotides in the stems were completely conserved, and there were only three positions where nucleotides varied: one in a loop (nt 27 in L2) and two in bulges (nt 9 in B4 and nt 20 in B2).

The RNA secondary structure of the complete genome of a human CHIKV isolate from a 2013 outbreak on a Caribbean island (GenBank accession number MT228631) was recently mapped in solution using the selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE)-mutational profiling technique [21]. The joint analysis of the 3’UTR sequence and its SHAPE reactivity showed that virus stocks comprise viral variants with differences in their 3’UTRs. These variants differ in length and RNA structures downstream the translation stop codon but had a common sequence and secondary structure at the end of the 3’UTR. To evaluate whether their secondary structure in solution supports the SLYa and SLYb folding, we thermodynamically modeled SLYa and SLYb using the SHAPE reactivity values published by Madden et al. as constraints [21] (Fig 1C, right). For both SLYs, nucleotides in S1, S2 and S3 consistently showed low reactivity values (less than 0.3, in black), suggesting that they are in non-flexible zones of the RNA and likely involved in base pairings. Notably, single-stranded B1, B4 and L1 terminal nucleotides of SLB were also poorly reactive, indicating they are likely involved in RNA tertiary interactions. In contrast, nucleotides in L1, L2, B2 and B3 had higher SHAPE values, indicating that these regions are exposed in the tertiary RNA structure.

Altogether, our analyses support the folding of a stable SLY in the CHIKV 3’UTR that is duplicated in the Asian lineage.

The presence of a duplicated SLY contributes to 3’UTR variability after viral adaptation to mammalian cells

In a previous study, we followed the in vitro evolution of CHIKV populations, finding that new viral variants with deletions in their 3’UTRs emerge during adaptation to cells [20]. Specifically, restricting the virus to a single cell type resulted in 3’UTR variants with clean deletions of DR copies that arose with higher frequency in BHK mammalian cells than in C6/36 mosquito cells. Interestingly, many emergent variants contained only one DR3 copy in their 3’UTRs, and therefore a single SLY element. These observations raise questions about why Asian strains have retained a duplicated SLY in their 3’UTRs and whether SLY has a host-dependent function during the CHIKV life cycle.

To investigate SLY’s role during CHIKV replication and evolution, we designed recombinant viruses in which we individually deleted one (ΔSLYa or ΔSLYb) or both (ΔSLYab) SLYs from the wild type (WT) CHIKV-Caribbean 3’UTR (Fig 2A, left). First, we evaluated the replication of WT and SLY deletion mutants in mammalian cells. To this end, we synthesized full-length viral RNAs by in vitro transcription (IVT) that were transfected in equal amounts (Fig 2B, top) into BHK cells. Then, we followed viral replication as a function of time through the detection of viral antigens and titration of infectious viral particles in cell culture supernatants. We observed that deleting either or both SLYs did not negatively affect viral replication in mammalian cells. WT and mutant (ΔSLYa, ΔSLYb and ΔSLYab) viruses could infect the entire cell monolayer after three days of transfection and caused extended cytopathic effect by day 6 (Fig 2A). The quantification of viral yields in culture supernatants showed that viruses with SLY deletions had an advantage when replicating in BHK cells: ΔSLYab, ΔSLYa and ΔSLYb reached viral titers 35-, four- and four-fold higher than those reached by the WT virus on day 3. To further characterize the ΔSLYab mutant phenotype in human cells, we infected fibroblasts and Huh-7 cells with the ΔSLYab and WT viruses. Growth curves showed that both viruses replicated similarly, despite a slight advantage of the ΔSLYab over the WT in human fibroblasts (Fig 2C). These results clearly indicate that the presence of SLY in the CHIKV 3’UTR is not necessary for efficient viral replication in mammalian cells, and surprisingly, the ΔSLYab deletion might have a positive effect on viral yields.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Replication and adaptation to mammalian cells of CHIKV-Caribbean and mutant viruses with SLY deletions.

(A) Left, schematic representations of the 3’UTRs of the WT virus and mutant viruses with deletions of one or two SLY copies (ΔSLYa, ΔSLYb and ΔSLYab). Right, immunofluorescence stainings for WT and mutant viruses in mammalian BHK cells on day 3 post-transfection. Images correspond to one representative experiment out of three biological replicates. (B) Viral yields in cell culture supernatant on day 3 post-transfection for WT and mutant viruses. The symbols and bars depict the means ± standard deviations of the means from three independent experiments. Data were compared with a two-tailed, unpaired t-test. Top, in vitro transcribed (IVT) RNAs used for transfection in the experiments shown in Figs 2 and 4. (C) Comparative growth kinetics for WT and ΔSLYab viruses in human fibroblasts and Huh-7 cells. Cells were infected with MOI = 0.1 and data were compared with a two-tailed, unpaired t-test. The symbols and bars depict the means ± standard deviations of the means from two independent experiments. (D) Analysis of the 3’UTRs of the WT and mutant viruses restricted to replication in mammalian cells. Top, schematic representation of viral passages and analysis of the 3’UTRs. Middle, bar graphs showing the frequencies of viral variants in WT and mutant viral populations P1 and P5 in mammalian cells. The ratio of 3’UTR deletion variants to total clones is indicated inside the bars and corresponds to the cumulative data from three independent experiments. Data were compared pairwise with Fisher’s exact test on cumulative data. Bottom, table with the proportion of 3’UTR deletion variants from three independent experiments (n1, n2 and n3) in the WT, ΔSLYa, ΔSLYb and ΔSLYab populations. (E) Representative agarose gels for PCR amplification of the 3’UTRs of individual clones recovered from WT and mutant viral populations passaged five times in BHK cells. The sizes of DNA bands in the ladder (base pairs) are indicated on the right. (F) Schematic representation of the alignment of the 3’UTRs of viral variants comprising the WT P5 population. Red lines indicate deletions within viral variants. The corresponding sequence for each variant is shown in S1 Fig.

https://doi.org/10.1371/journal.ppat.1011352.g002

Second, we studied SLY’s role during CHIKV experimental evolution. We focused our analysis on the 3’UTR and investigated the composition of viral populations after the infection with the WT virus and mutant viruses carrying SLY deletions. Briefly, WT, ΔSLYa, ΔSLYb and ΔSLYab viral stocks recovered from transfection supernatants were serially passaged in mammalian BHK cells. Total RNA was extracted from culture supernatants for each viral population and used as a template for reverse transcription reactions with an oligo(dT) primer. The pool of viral cDNAs was used to amplify fragments corresponding to the 3’ ends of the viral genome, which were then cloned into the pCR2.1-TOPO vector. Individual plasmid clones corresponded to a single virus variant in the population and were analyzed by agarose gel electrophoresis to distinguish parental and deletion variants emerging during passaging (Fig 2D, top and 2E). Frequencies of emergent viral variants were calculated separately for WT, ΔSLYa, ΔSLYb and ΔSLYab populations (Fig 2D, bottom). As expected, after one passage in mammalian cells (P1), 3’UTR-deletion variants emerged in the WT viral population, representing 10% of the analyzed clones. Interestingly, 3’UTR-deletion variants also arose from viral mutant populations and after P1 represented 5%, 14% and 3% for ΔSLYa, ΔSLYb and ΔSLYab, respectively, not differing significantly from the WT population (Fig 2D). After five viral passages in BHK cells (P5), the WT-adapted population contained only 12% parental virus, and nearly 88% of emergent variants had 3’UTR deletions. In contrast, the P5 populations derived from SLY-deletion mutants comprised 74%, 79% and 67% parental RNAs for ΔSLYa, ΔSLYb and ΔSLYab, respectively, and a minority of variants with 3’UTR deletions (26%, 21% and 33% for ΔSLYa, ΔSLYb and ΔSLYab, respectively). Consistent with our previous results [20], WT-adapted populations comprised a repertoire of viral variants with deletions in the 3’UTR (Figs 2F and S1). Some deletions comprised the DR3 copies, yielding viral variants that carried a single SLY (clones 1, 2, 5 and 10) or the deletion of both SLYs (clones 3 and 7) in the 3’UTR. Instead, other deletions were outside DR3 and yielded viral variants that retained both SLY structures (clones 9 and 11).

Altogether, our experiments demonstrate that the presence of the 3’ SLY is not necessary for efficient viral replication in mammalian cells. However, less diversity is generated within viral 3’UTRs when CHIKV-Caribbean lacks one or both SLYs.

Viral RNA recombination via template-switching mechanism occurs in the 3’UTR of CHIKV-Caribbean regardless of the presence of a duplicated SLY

In RNA viruses, the composition of viral populations results from the continuous interplay between the generation of genome variability and the selective pressures imposed by the host [24]. Genome variability arises from the introduction of point mutations by an RNA-dependent-RNA-polymerase (RdRp) lacking proofreading activity and RNA recombination events generating chimeric viral genomes during replication. Our experiments showed that during CHIKV-Caribbean adaptation to mammalian cells, 3’UTR deletion variants emerged in viral populations derived from the WT virus and mutant viruses lacking one or both SLYs in their 3’UTRs. However, the proportion of new 3’UTR variants after host adaptation was much lower for SLY-mutant populations.

We have previously showed that CHIKV 3’UTR-deletion variants are generated by copy-choice recombination of the RNA viral genome through a template-switching mechanism [20]. In this process, RdRp and the nascent strand dissociate from the RNA template, and then reassociate at a different position on another RNA template, guided by sequence homology between blocks of repeated elements. Interestingly, for other RNA viruses, it has been proposed that the presence of RNA structural elements in either the template or the acceptor strand also promotes template switching [25]. Therefore, we reasoned that SLY might function to favor the rate or frequency of RNA recombination by template-switching events in the CHIKV 3’UTR. To test this hypothesis, we developed an approach that allows the detection of chimeric viral RNAs originating through the recombining of two differentially marked viral RNA templates. We introduced point mutations into CHIKV infectious clones to generate recognition sites for two unique restriction enzymes at the 5’ or 3’ end of the 3’UTR. Following this strategy, we obtained two pairs of 5’/3’end marked viral genomes: one for the WT virus and one for the ΔSLYab virus. When 5’ and 3’ marked RNAs for the same virus are mixed and co-transfected into cells, template-switching events can be unequivocally shown by restriction analysis. Viruses with a chimeric 3’UTR will contain either both or none of the marker sites compared to the 3’UTR of viruses present in the original input mixture that contained a single site (Fig 3A). Specifically, we engineered restriction sites for XbaI at position 5–10 after the translation stop codon and for AvrII at position −40–−35 before the poly(A) tail, obtaining the recombinant viruses WT-XbaI, WT-AvrII, ΔSLYab-XbaI and ΔSLYab-AvrII. First, we examined viral replication of the newly marked viruses, finding that all were infective in BHK cells (Fig 3B, top). Importantly, we tested whether the introduced mutations were stable in cell culture at least after four successive passages (Fig 3B, bottom). Culture supernatants were harvested for each recombinant virus from passage four and RNA was extracted. Next, these RNAs were used as templates in reverse transcription-polymerase chain reactions (RT-PCR) amplifying the viral genome’s last 1200 nucleotides. Then, RT-PCR products were cloned into the pCR2.1-TOPO vector. The presence of XbaI or AvrII sites was determined by enzymatic digestion of PCR products amplified from individual clones. XbaI restriction sites in individual clones can be readily assessed by a 439 or 363 nt DNA fragment after digestion of the PCR products with XbaI. Similarly, AvrII restriction sites in individual clones can be assessed by a 193 or 181 nt DNA fragment after digestion of the PCR products with AvrII or more easily, by a shift in the mobility of the remaining fragment. To test the stability of the introduced mutations, we analyzed 10 plasmid clones for WT and ΔSLYab marked viruses, observing that the restriction sites were conserved in all the clones at least after four viral passages. For illustration, digests for one representative viral clone for each WT (AvrII and XbaI) and ΔSLYab (AvrII and XbaI) passaged virus are shown (Fig 3B, bottom).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. RNA recombination in the CHIKV 3’UTR through template switching between homologous templates.

(A) Schematic representation of copy-choice recombination mechanism between WT (top) or ΔSLYab (bottom) viruses differently marked with XbaI (WT-XbaI and ΔSLYab-XbaI) or AvrII (WT-AvrII and ΔSLYab-AvrII) at the 5’ and 3 ‘ends of their 3’UTRs, respectively. After cotransfections, recombinant viruses arise at the 3’UTR (XbaI/AvrII and −/−). (B) Immunofluorescence of WT, WT-XbaI, WT-AvrII, ΔSLYab, ΔSLYab-XbaI, and ΔSLYab-AvrII viruses on day 3 post-transfection of BHK cells (top) and assessment of the presence of XbaI or AvrII marker sites after four successive passages in BHK cells (bottom). Depending on the orientation of each inserted fragment into the blunt plasmid, digestion of XbaI positive clones generated 439 or 363 nt products, and digestion of AvrII positive clones generated 193 or 181 nt products. (C) For illustration, undigested and digested products from eight representative clones are shown for each cotransfection mixture. Recombinant 3’UTR clones are indicated with an inverted gray triangle under each agarose gel. (D) Only clones spanning the full-length fragments were numbered to calculate recombination frequencies. Left, bar graphs showing the relative abundance of recombinant and nonrecombinant full-length 3’UTRs from WT and ΔSLYab viral populations. Nonrecombinant XbaI and AvrII 3’UTR variants are indicated in orange and violet plain colors, respectively, and recombinant XbaI/AvrII and −/− 3’UTR variants are indicated in squared violet/orange and gray/white patterns, respectively. Data were compared with Fisher’s exact test on cumulative data from two biological replicates. Key: ns, not significant. Right, table showing the proportion of 3’UTR variants from two independent experiments (n1 and n2).

https://doi.org/10.1371/journal.ppat.1011352.g003

Next, to evaluate the occurrence of recombination events producing chimeric 3’UTRs in the presence or absence of SLY, equal amounts of RNA transcripts for the WT-XbaI/WT-AvrII and ΔSLYab-XbaI/ΔSLYab-AvrII pairs were separately cotransfected into BHK cells. On day 3 post-transfection, supernatants from both mixtures were collected and used to infect fresh BHK cells. To assess recombination, total RNA was extracted after three days and processed as described above to test the presence or absence of XbaI and AvrII sites in individual plasmid clones (Fig 3C). For WT and ΔSLYab populations, we examined clones containing full-length 3’UTRs and calculated the relative abundances of viral variants. Clones containing either XbaI or AvrII were classified as nonrecombinants, while clones containing XbaI and AvrII sites or no marker sites were classified as recombinants (Fig 3D). Interestingly, about 60% (13 out of 21 total clones with full-length 3’UTR) of viruses recovered from the WT-XbaI/WT-AvrII mixture had recombined in their 3’UTR, and most had no marker sites (11 clones). A similar abundance of RNA recombinants was found with the ΔSLYab-XbaI/ΔSLYab-AvrII mixture (12 out of 20 clones bearing full-length 3’UTR), also with no marker sites (Table in Fig 3D). The fact that variants lacking artificially generated restriction sites vastly outnumbered those with both sites suggested that introducing the restriction sites diminished viral fitness. In fact, competition experiments of the AvrII vs the −/− virus showed that the single nucleotide substitution introduced to generate the AvrII recognition site at the end of the 3’UTR significantly decreased viral fitness. In turn, nucleotide substitutions that generate the XbaI recognition site had no effect on fitness compared to the −/− virus (S2 Fig). While the decreased fitness of the AvrII virus explains the composition of viral populations arising from cotransfection experiments (Fig 3D), it does not impact on the approach to analyze the recombination rate. As noted, the proportions of recombinant 3’UTR variants in the WT and mutant populations were not significantly different. These data indicate that RNA recombination occurs at high rates in the CHIKV 3’UTR, regardless of the presence of the SLY element in DR3, ruling out SLY’s role in favoring recombination. Nevertheless, in vitro evolution experiments (Fig 2D) indicated that, after P5, the WT-adapted population contained more 3’ deletion variants than the SLY-mutant populations. Therefore, we reasoned that the wider spectrum of emerging variants in the WT population was not due to different recombination rates but to strong viral selection processes in mammalian cells. According to fitness parameters (Fig 2B and 2C), it is plausible that new variants lacking DR copies have a replicative advantage over the parental WT virus, explaining their high abundance in the mammalian-adapted population. Conversely, viruses with engineered SLY deletions have per sé high fitness, explaining why in vitro evolution did not result in the selection of new variants with shorter 3’UTRs as observed with the WT virus.

Overall, our findings indicate that the SLY element impacts CHIKV 3’UTR evolution without altering template-switching rates.

SLY presence in the CHIKV 3’UTR promotes viral replication in mosquito cells

After finding that SLY is dispensable for efficient viral replication and 3’UTR recombination in mammalian cells but is nevertheless retained in the Asian CHIKV lineages, we next asked whether mosquito-specific selective pressures act on the 3’UTR to maintain SLY elements. To address this question, we evaluated the replication of the WT virus and SLY-deletion mutant viruses (ΔSLYa, ΔSLYb and ΔSLYab) in C6/36 mosquito cells. We observed that deleting one or both SLY copies drastically affected replication in mosquito cells (Fig 4A). On day 3 post-transfection, the WT virus had infected approximately 13% of cells in the monolayer, while the ΔSLYa and ΔSLYb viruses had infected <5%. No ΔSLYab virus infected cells were detected at this time point. On day 6, differences in cell infection percentages were maintained between the WT and SLY-deletion mutant viruses. The WT virus had infected the complete monolayer, while the ΔSLYa, ΔSLYb and ΔSLYab viruses had infected 73%, 68% and 15% of the monolayer, respectively. On day 9, all the mutant viruses had infected the complete monolayer. Differences in viral yields from culture supernatants consistently accompanied differences in growth between the WT virus and SLY-deletion mutant viruses (Fig 4B). For example, on day 6 post-transfection, titers were significantly lower for the ΔSLYa and ΔSLYb viruses (500- and 30-fold) than for the WT virus. However, the virus lacking both SLYs showed the most drastic effect, with viral titers about 7000-fold lower than the WT virus.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Replication of WT CHIKV and viruses with SLY deletions in mosquito cells.

(A) Left, schematic representation of the CHIKV-Caribbean WT virus and mutant viruses with deletions of one or two SLY copies. Right, immunofluorescence staining for WT and mutant viruses in mosquito C6/36 cells on days 3, 6, and 9 post-transfection. Images correspond to one representative experiment out of three biological replicates. The data were analyzed and presented as described in Fig 2. (B) Viral yields in cell culture supernatants on day 6 post-transfection for WT and mutant viruses. The symbols and bars depict the means ± standard deviations of the means from three independent experiments. Data were compared with a two-tailed, unpaired t-test. (C) Comparative growth kinetics of WT and ΔSLYab viruses in C6/36 cells. Cells were infected with MOI = 0.1, and data were compared with a two-tailed, unpaired t-test. The symbols and bars depict the means ± standard deviations of the means from two independent experiments.

https://doi.org/10.1371/journal.ppat.1011352.g004

To further confirm growth phenotypes, we used stocks of the WT and ΔSLYab viruses to infect fresh C6/36 cells using a multiplicity of infection (MOI) of 0.1 and followed viral replication. The results confirmed that viruses lacking SLY copies showed delayed growth kinetics in mosquito cells compared to the WT virus (Fig 4C). Altogether, these results indicate that the presence of SLY in the CHIKV 3’UTR is necessary for efficient viral replication in mosquito cells, where there is a positive correlation between the number of SLY copies and viral replication. Moreover, our experiments suggest that SLY has a host-dependent role during CHIKV replication: it favors viral replication in mosquito cells but is entirely dispensable in mammalian cells. Furthermore, the duplicated SLY in Asian and Caribbean lineages confers a replicative advantage in mosquito cells.

Functional characterization of the SLY’s sequence and structural elements

According to RNA structure prediction and chemical probing in solution, the SLY contains defined structural elements, including three stems, four bulges, and two loops (Figs 1C and 5A, left). To characterize in detail the role of individual structural elements on viral replication, we designed a panel of recombinant viruses that: (i) modify the L1 sequence without altering the SLY general structure (MutL1); (ii) disrupt the SLY structure by incorporating mismatches into S1’s left strand (MutS1), (iii) disrupt the SLY structure by incorporating mismatches into S3’s left strand (MutS3) and (iv) reconstitute the SLY structure, including artificial nucleotide sequences in both S3 strands (RecS1) (Fig 5A, left). To characterize SLYa and SLYb separately and simplify the experiments’ performance and analyses, we used recombinant viruses containing only one SLY in their 3’UTR (ΔSLYb and ΔSLYa) as the parental genomes. Therefore, we individually introduced each of these mutations into SLYa in the context of the ΔSLYb infectious clone and vice versa (see panels with mutant viruses in Fig 5A, right). We synthesized viral genomic RNA molecules and transfected them in equal amounts (Fig 5B, right) into mosquito and mammalian cells. Next, we evaluated viral replication as a function of time via immunofluorescence assays and viral yields in cell culture supernatants. For SLYa mutants in the ΔSLYb backbone, we found that disrupting S1 drastically impaired viral replication in mosquito cells (Fig 5B, left). In contrast to the parental ΔSLYb virus that was readily detected on day 3 and infected the entire monolayer by day 6, the MutS1_ΔSLYb virus was only detected after day 6 and infected only 20% of the cell monolayer by day 9. Disrupting S3 also significantly delayed viral replication in mosquito cells. MutS3_ΔSLYb replication was barely detectable on day 3 post-transfection and had infected about 8% and 44% of the monolayer on days 6 and 9, respectively. Finally, modifying the L1 sequence also negatively affected viral replication since only 2.2%, 25% and 50% of infected cells were detected on days 3, 6, and 9, respectively. Viral yields consistently followed the immunofluorescence results (Fig 5B, right). On day 6 post-transfection, MutS1_ΔSLYb, MutS3_ΔSLYb and MutL1_ΔSLYb had -8000, -5000 and -3600 fold lower titers than the parental ΔSLYb virus, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Replication of recombinant viruses with mutations that maintain or disrupt the SLY structure in mosquito cells.

(A) Left, schematic representation of the mutations introduced into the SLY in order to generate recombinant viruses that maintain (MutL1 and RecS3) or disrupt (MutS1 and MutS3) its predicted Y-shape RNA structure. In each case, the sequences for WT and mutated SLYs are indicated inside gray and colored boxes, respectively. Right, a schematic representation of the 3’UTRs with SLYa mutations inserted into ΔSLYb (top) and vice versa (bottom). (B and C) Left, immunofluorescences for the mutant ΔSLYb, MutS1_ΔSLYb, MutS3_ΔSLYb, MutL1_ΔSLYb, and RecS3_ ΔSLYb viruses (B) and ΔSLYa, MutS1_ΔSLYa, MutS3_ΔSLYa, MutL1_ΔSLYa, and RecS3_ ΔSLYa viruses (C) in C6/36 and BHK cells at the indicated time points. Images correspond to one representative experiment, with data analyzed as described in Fig 2. Right, viral yields in cell culture supernatants on days 3, 6 and 9 post-transfection. The symbols and bars depict the means ± standard deviations of the means from two independent experiments. Data were compared with a two-tailed, unpaired t-test. Agarose gels with IVT RNAs used for transfection in a representative experiment are shown.

https://doi.org/10.1371/journal.ppat.1011352.g005

The results suggest that altering the global SLY structure by introducing mismatches greatly impair viral replication in mosquito cells. However, it is possible that the adverse effects result from specific sequence requirements instead of structural constraints. To show that the SLY’s host-specific function relies on the structure and not on the conservation of the nucleotide sequences that pair at the stem’s base, we designed an additional viral mutant carrying an intact SLYa but with a different nucleotide sequence. Briefly, we introduced point mutations into the S3’s right strand in the context of MutS3, which carried changes in the S3’s left strand that disrupted SLYa formation, to reconstitute base pairings and regenerate SLYa folding (RecS3_ΔSLYb, Fig 5A). As expected, the in silico RNA structures folded precisely the same way as the WT virus. Assessment of viral fitness showed that RecS3_ΔSLYb had growth kinetics and reached viral titers comparable to the parental ΔSLYb virus (Fig 5B).

Analogous mutations in SLYb in the context of the ΔSLYa backbone were also tested (Fig 5C). The spread of parental ΔSLYa virus was comparable to ΔSLYb virus. In turn, as observed for the SLYa mutants, mutations disrupting the SLY general structure severely affected viral replication in mosquito cells. MutS1_ ΔSLYa and MutS3_ΔSLYa infected 6% and 27% of cell monolayers on days 3 and 6, respectively, compared to 22% and 100% for the control ΔSLYa virus at the same time points. Mutation of the loop nucleotides (MutL1_ΔSLYa) slightly delayed viral replication, with 14% and 69% of cells infected on days 3 and 6, respectively. Mutations that reconstitute S3 pairings (RecS3_ΔSLYa) infected the entire monolayer on day 6 as the parental ΔSLYa. Viral yields for SLYb mutants recapitulated what we observed when detecting viral antigens. For example, on day 6, MutS1_ ΔSLYa and MutS3_ΔSLYa had about 250-fold lower titers than the parental ΔSLYa virus, MutL1_ΔSLYa only showed a slight decrease in viral titers and RecS3_ΔSLYa reached viral titers comparable to the parental virus (Fig 5C, right).

Finally, we tested whether discrete mutations of SLYa or SLYb in the context of full-length 3’UTR impact viral replication. Mutations disrupting the S3 of SLYa or SLYb were introduced into CHIKV genomes containing both SLYs. Point mutations in either SLY recapitulated our observations with the ΔSLY backbone (S3 Fig). Furthermore, virus yields in the supernatants paralleled the effects of the individual SLY deletions (compare titers on day 4 with those in Fig 4B). Altogether, these results indicate that a duplicated SLY structure is necessary for efficient CHIKV replication in mosquito cells.

Similar analyses were conducted in BHK mammalian cells, where all the mutant viruses could replicate to levels comparable to parental viruses, further confirming that SLY is not required for viral replication in this host (Fig 5B and 5C, middle).

Because mutant virus replication was only evident several days post-transfection, we asked whether the recovered viruses had introduced spontaneous mutations in their 3’UTRs to reconstitute SLY structures. To evaluate this possibility, we performed RT-PCRs and sequencing analyses of viral 3’UTRs on day 9 post-transfection. Interestingly, the engineered sequences of the mutant viruses were retained, and no new changes appeared in the entire 3’UTR region. This result suggests that the recovered viruses can replicate in cell culture with an attenuated viral phenotype. However, we cannot rule out that spontaneous mutations have arisen in other regions of the genome to compensate for defects in viral replication.

In conclusion, the structure of both SLYa and SLYb in the 3’UTR of CHIKV-Caribbean must remain intact for efficient viral replication in mosquito cells. Our results with mutant viruses carrying SLY structures designed with artificial sequences clearly show that the role of this RNA element in mosquito cells depends on its structure and not on the sequence of the base-pairings. The loops’ sequences were also critical for viral replication in mosquito cells, indicating that specific nucleotides should be exposed at the top of SLYs. Interestingly, the effects were more pronounced for SLYa mutants, suggesting that the SLY’s 5’ copy is more sensitive to nucleotide changes.

This is the first report of a host-specialized RNA element in the CHIKV 3’UTR, and supports the view that selective mosquito pressures act on DR3 sequences to maintain SLY’s folding.

The 3’UTR of distant alphaviruses supports efficient replication of CHIKV in mosquito cells

It has recently been observed that the CHIKV´s replicase can trans-amplify a template RNA carrying the Sindbis virus (SINV) promoter sequences in both mammalian and mosquito cells [26]. In contrast, CHIKV template replication by the SINV replicase is inefficient [26]. A systematic analysis using chimeric templates showed that major determinants for efficient replication by the SINV replicase mapped to the 5’UTR of the RNA template [26].

SINV is the prototype of the alphavirus genus. Like CHIKV, it alternates between mammalian and mosquito hosts. The 5’ and 3’ ends of its viral genome have been extensively studied [27]. The SINV 3’UTR contains three RSEs copies about 40nt long (Fig 6A, left) that have been predicted to fold into three identical SLYs [28]. Noteworthy, these RNA motifs have a host specific role stimulating translation of SINV genome in mosquito cells [28]. Despite divergence in nucleotide sequences, other members of the SINV group, such as Babanki (BABV), Whataroa (WHAV) and Aura (AURAV) viruses, also contain three copies of the SINV RSE in their 3’UTRs [19]. Moreover, the Western equine encephalitis virus (WEEV), a recombinant virus derived from a SINV-like virus and the Eastern equine encephalitis virus, contains only two RSEs copies (Fig 6A, right) [27]. Considering the common architecture of the 3’UTR of viruses in the SINV group, we wondered whether the SLYs described for SINV were a conserved feature among other group members.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. RNA secondary structures in the 3’UTR of Sindbis group viruses.

(A) Left, schematic representation of the genomic organization of SINV and its 3’UTR with the RSEs indicated as light blue boxes and regions folding into SLYs in white boxes. Right, phylogenetic relationship between the Sindbis group members (SINV, Babanki ⦍BABV⦎, Whataroa ⦍WHAV⦎, Western equine encephalitis ⦍WEEV⦎, and Aura ⦍AURAV⦎ viruses) and their 3’UTRs. (B) Sequence alignment (middle) and nucleotide conservation plot (bottom, in gray) of the RSEs that fold into SLYa, SLYb, and SLYc for SINV (NC_001547.1), BABV (MF409178.1), WHAV (NC_016961.1), AURAV (NC_003900.1), and WEEV (NC_003908.1). Pink arcs and boxes indicate the nucleotides that pair to form the stems of each Y. Conserved nucleotides are in black, and variable nucleotides are in blue, red, green, and orange. Black boxes indicate the nucleotides that covary among viruses. Key: S, stem; L, loop. (C) Predicted RNA secondary structures for the RSEs in the 3’UTRs of SINV, BABV, WHAV, AURAV and WEEV. Base-pair probabilities are color-coded, and the minimum-free energy for each SLY is shown below. (D) Schematic representation of the SLY regions conserved (in light blue) or covarying (in black) between different SINV group viruses and SLY copies.

https://doi.org/10.1371/journal.ppat.1011352.g006

To answer this question, we aligned the 3’UTRs of SINV, BABV, AURAV, WHATV, and WEEV and identified sequences corresponding to RSEs. Next, we individually predicted the RSEs secondary structure, finding that all the members in the SINV group contain 60nt-long regions that fold into highly stable SLYs. The SLY element comprises an upper right stem (S1) that exposes the L1 loop, an upper left stem (S2) that exposes the L2 loop, a central bulge (B1), and a basal stem (S3). Consistent with the number of RSE copies, SLY appears to be triplicated (SLYa, SLYb and SLYc) in SINV, BABV, WHAV, and AURAV and duplicated (SLYa and SLYb) in WEEV. All SLY copies had similar predicted minimum free energies (around -20 kcal/mol) and high base-pair probabilities for individual nucleotides (above 0.8), except those involved in forming S3 in WHAV SLYb (Fig 6C).

The finding of multiple SLY copies in SINV group members prompted us to analyze the nucleotide conservation of SLYa, SLYb, and SLYc together with SLY folding. Interestingly, while nucleotides in L1, S2, and L2 were conserved among species and SLY copies, most nucleotides involved in forming S1 and S3 were highly variable. Strikingly, despite poor nucleotide conservation, base-pair complementarity is fully maintained among copies resulting in structurally identical SLY elements (Fig 6B and 6D). This form of structural conservation, where different nucleotide sequences form similar structures, is known as covariation and provides strong evidence for the formation of functional RNA elements. Here, covariation analysis of the 3’UTRs of SINV and related alphaviruses supports the folding of 2–3 copies of highly stable and conserved SLYs, resembling our analysis with the CHIKV 3’UTR.

Based on accumulated observations, we asked whether the SINV 3’UTR would function in place of the CHIKV 3’UTR in the context of full-length viruses to support replication in mosquito cells. To address this question, we constructed a viral chimera containing the SINV 3’UTR in the context of the CHIKV-Caribbean genome (Fig 7A, left). Equal amounts of RNA transcripts of parental and chimeric viruses were individually transfected into BHK or C6/36 cells. Next, virus replication was monitored by detecting viral antigens using immunofluorescence assays and quantifying viral yield as a function of time. In both C6/36 and BHK cells, monolayers were infected to a similar extent by parental and chimeric viruses at each time point analyzed (Fig 7A). When we titrated viruses in culture supernatants, we observed that the chimeric virus reached titers about 30-fold higher than the parental virus on day 6 in C6/36 cells (Fig 7B) and similar to the parental virus at the time points tested in BHK cells (Fig 7C). These results show that the SINV 3’UTR was fully functional for viral replication in the CHIKV genomic background.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Replication of CHIKV chimeric viruses with the SINV 3’UTR in mosquito and mammalian cells.

(A) Left, schematic representation of CHIKV-Caribbean virus and its 3’UTR with two SLY copies (WT) and a chimeric CHIKV- Caribbean virus with the SINV 3’UTR (chimeric virus). Right, immunofluorescence stainings of WT and chimeric viruses in mosquito C6/36 cells on days 3, 6, and 9 post-transfection and in mammalian BHK cells on days 1 and 3 post-transfection. Images correspond to one representative experiment out of two biological replicates. Data were analyzed as described in Fig 2. Viral yields in cell culture supernatants for the WT and the chimeric virus in C6/36 cells (B) and BHK cells (C). The symbols and bars depict the means ± standard deviations of the means from two independent experiments. An agarose gel with IVT RNAs used for transfection. Data were compared with a two-tailed, unpaired t-test. Bottom, a representative agarose gel of the PCR amplification of the 3’UTRs of individual chimeric virus clones after P5 in BHK cells. A DNA ladder (M) was used as the reference. The sizes of DNA bands in the ladder (in base pairs) are indicated on the right.

https://doi.org/10.1371/journal.ppat.1011352.g007

To assess whether the SINV 3’UTR changed during host adaptation, we performed five successive viral passages in BHK cells and then analyzed the 3’UTRs of the chimeric viral population. Unlike our observations with the CHIKV 3’UTR, which was prone to losing DR copies [20], we found that >85% of the chimeric virus’s progeny contained an intact 323nt-long 3’UTR. These results suggest that the SINV 3’UTR is stable in the CHIKV genome context during adaptation to mammalian cells in cell culture (Fig 7C, bottom panel). Altogether, these data indicate that the SINV 3’UTR can functionally replace the 3’UTR of CHIKV-Caribbean, assuring optimal viral replication in mammalian and mosquito cells.

In summary, in this study we characterized a Y-shaped RNA element in the CHIKV 3’UTR that (i) modulates viral evolution in mammalian cells without altering recombination rate, (ii) assures optimal viral replication in mosquito cells, and (iii) can be replaced by 3’UTRs from distant alphaviruses to replicate efficiently in both hosts. Importantly, SLY’s function in mosquito cells depends highly on its structure and specific top-loop nucleotides.

Sequence alignment and RNA secondary structure prediction

Complete genome sequences of CHIKV natural isolates were obtained from the ViPR database and cured manually to select those with complete 3’UTRs. 112 3’UTR sequences from the East Central and South Africa (ECSA), Indian-Ocean (IOL), and Asian lineages (S1 Table) were aligned using an online version of the Clustal Omega program. The RNA secondary structure predictions were done using the RNAalifold program. The SHAPE constrained thermodynamic model was done with RNAstructure program (version 6.4, Mathews lab) considering the consensus SLYa and SLYb sequences and the SHAPE reactivity values obtained by Madden et al. [21]. Sequences of SINV, Babanki (BABV), Whataroa (WHAV), Aura (AURAV), and Western Equine Encephalitis (WEEV) 3’UTRs were downloaded from GeneBank public database and aligned with Clustal Omega to identify the RSEs. The RSE secondary structures were predicted using the RNAfold program.

Recombinant CHIKVs

We constructed a 3’UTR cloning cassette spaning from the stop codon of ORF encoding for viral structural proteins to the poly(A) tail to facilitate mutations within the CHIKV-Caribbean 3’UTR infectious clone. The cloning cassette is flanked by SacI and NotI unique restriction sites and allows the exchange of the WT sequence by mutant 3’UTRs. Mutant 3’UTRs of ΔSLYa, ΔSLYb, ΔSLYab, WT-AvrII, ΔSLYab-AvrII, MutS1_ΔSLYb, MutS3_ΔSLYb, MutL1_ΔSLYb, MutS1_ΔSLYa, MutS3_ΔSLYa, MutL1_ΔSLYa, RecS3_ΔSLYb, RecS3_ΔSLYa, MutS3-SLYa, and MutS3-SLYb were generated by overlapping PCRs. The templates and the forward and reverse oligonucleotides used to introduce the specific mutations are given in S2 Table. The overlapping PCRs were performed with the common outside oligonucleotides 5’-TCAGCAGGCACTAAGAGCTCGACAATTAAGTA-3’ and 5’-CGAAACAAGCGCTCATGAGC-3’. The 3’UTRs with XbaI restriction site (WT-XbaI and ΔSLYab-XbaI) were obtained by PCR amplification of the 3’UTR using the sense primer 5’-TCAGCAGGCACTAAGAGCTCTAGAGACAATTAAGTA-3’ and the reverse primer 5’-CGAAACAAGCGCTCATGAGC-3’. The chimeric virus 3’UTR was constructed through PCR amplification using the SINV infectious clone as template, the sense primer 5’-CGAAGATGAGCTCTACGCCCCAATGA-3’, and the reverse primer 5’-TTAGCGGCCGCTTTTTTTTTTTTTTTTTTTTTTTTTTGAAATGT-3’. The PCR products and the CHIKV-Caribbean infectious clone were digested with SacI and NotI restriction enzymes, ligated together and used for transformation of competent E.coli cells. Plasmidic infectious clones were mini-prep purified and Sanger sequenced to verify the presence of the desired mutations.

Recombinant constructs DNAs were linearized by digestion with NotI and used as templates for transcription with SP6 polymerase in the presence of m⁷G(5’)ppp(5’)G cap structure analog, using the mMessage Machine transcription kit (Thermo Fisher). The IVT RNAs were gel-quantified and their integrity was verified by electrophoresis on agarose gels. In each case, equal amounts of RNA transcripts (3 μg/well) were used for transfection in cell culture. Note that IVT reactions result in two RNA products probably due to a cryptic leaky SP6 terminator in the CHIKV-Caribbean infectious clone.

Cells and viral transfections

Mammalian BHK cells (Mesocricetus auratus hamster kidney, ATCC, CCL-10) were grown at 37°C in MEM alpha medium in a 5% CO₂ atmosphere supplemented with 10% of fetal bovine serum (FBS) and penicillin-streptomycin. Huh-7 cells (Human hepatocyte cell line, ATCC, CVCL_0336) were cultured in D-MEM high-glucose medium supplemented with 10% of FBS and penicillin-streptomycin. A primary line of human skin fibroblasts was established at the Instituto de Medicina Translacional e Ingeniería Biomédica [CONICET, Hospital Italiano de Buenos Aires] and grown at 37°C in D-MEM medium.

Mosquito C6/36 cells (Aedes albopictus, ATCC, CRL-1660) were grown at 28°C in Leibovitz L-15 medium supplemented with 10% of FBS, 10% of tryptose phosphate, penicillin-streptomycin and amphotericin B. For RNA transfections, cell lines were grown at 60–70% confluence and transfected in 24-well plates using Lipofectamine 2000 (Invitrogen), following manufacturer’s instructions.

Immunofluorescence assays

BHK and C6/36 cells were seeded in a 24-well plate with a 1cm² coverslip inside and transfected WT or mutant RNAs. Immunofluorescence assays (IF) were performed at different times post-transfection. At each time point, cells were fixed in methanol and stained with a 1:1000 dilution of mouse anti-CHIKV monoclonal antibody NBP2-53111 (Novus) to detect viral antigens. Alexa Fluor 488 goat anti-mouse (Molecular Probes) was used as a secondary antibody at 1:1000 dilution. Estimation of the number of positive cells and calculation of relative percentages of infection were performed with the ImageJ program, using five images for each experimental condition.

Growth curves

Subconfluent HuH-7, BHK and C6/36 cells and human fibroblasts were seeded in a 24-well plate and infected with equal amounts of WT or mutant viruses. An MOI of 0.1 was used. One hour post-infection, cells were washed five times with PBS and 500 μl of growth media were added. At each time point, cell supernatants were collected and frozen at -70°C. For quantification of infective viral particles, culture supernatants were serially diluted and plaque assays were performed on Vero cells, as described below.

Plaque assays

Viruses from culture supernatants were quantified by plaque assays to determine viral yields. 10⁵ Vero cells (Cercopithecus aethiops kidney, ATCC, CCL-81) were seeded per well in 24-well plates and allowed to attach overnight. After serial dilution, 0.1 ml of cell culture supernatant was added to the wells and incubated for 1 hour at 37°C. Then, 1 ml of overlay (1x D-MEM medium supplemented with 2% of FBS, penicillin-streptomycin, and 0.4% of methylcellulose) was added to each well. On day 3 post-infection, cells were fixed with 10% formaldehyde and stained with crystal violet.

Experimental host adaptation and sequencing

WT, ΔSLYa, ΔSLYb and ΔSLYab IVT RNAs were transfected into BHK cells in three independent experiments. After two days, viruses were harvested and five successive infections were performed in the same cell line using an MOI of 0.5. Viral RNAs were Trizol-extracted from P1 and P5 culture supernatants and used for RT-PCR reactions with primer 115. Next, PCR reactions were carried out using primers 115 and 116 (5’-CTAATCGTGGTGCTATGC-3). Products were ligated into pCR2.1-TOPO vector and used to transform E.coli XL-1 Blue strain. The total number of clones analyzed for the WT, ΔSLYa, ΔSLYb and ΔSLYab viruses were 52, 20, 21, and 34 for P1; and 69, 31, 33, and 49 for P5, respectively. The lengths of individual viral 3’UTRs were estimated by resolving the 115–116 PCR products in 1.5% agarose gels. For WT P5 populations, 10 individual plasmid clones were sequenced by Sanger method using universal M13R and M13F primers.

Assessment of recombination after cotransfection of XbaI and AvrII marked RNAs

Two mixtures of in vitro transcribed RNAs were prepared: one with WT-XbaI and WT-AvrII RNAs, and the other with ΔSLYab-XbaI and ΔSLYab-AvrII RNAs. RNA mixtures were separately transfected into BHK cells in two independent experiments. Viral RNAs were Trizol-extracted from culture supernatants on day 3 and used for RT reactions with primer 115. Next, PCR reactions were carried out using primers 115 and 123 (5’-ACTAACGCCGTCACTATTCG-3’). Then, PCR products were cloned into pCR2.1-TOPO vector. A total of 50 and 33 individual clones were analyzed for WT-XbaI/WT-AvrII and ΔSLYab-XbaI/ΔSLYab-AvrII mixtures, respectively. To test XbaI and AvrII restriction sites in individual clones, full-length 3’UTRs were amplified by PCR reactions using universal primers M13F (5’-GTAAAACGACGGCCAGT-3’) and M13R (5’-GTCCTTTGTCGATACTG-3’). Finally, PCR products were digested with XbaI or AvrII restriction enzymes and analyzed by electrophoresis on 1.5% agarose gels. Because the PCR fragments can be inserted with two possible directions, and M13F and M13R primers anneal at different distances from the insertion site in the pCR2.1-TOPO vector, the sizes of the bands generated by digestion depend on directionality of the cloned insert. In addition, the multiple ligation site of the pCR2.1-TOPO vector has an XbaI restriction site, which impacts the size of the XbaI digestion products.

Growth competition experiments of marked versus unmarked −/− viruses

Viral RNAs of unmarked WT (−/−), WT-XbaI, and WT-AvrII were obtained by IVT and quantified. RNAs were mixed pairwise with ratios of 1 WT (−/−) to 4 WT-XbaI, or 1 WT (−/−) to 4 WT-AvrII. The mixtures were transfected into BHK cells. Viruses were recovered from culture supernantants and used to reinfect fresh cells. After three passages, total RNA was Trizol-extracted from the supernatants and used for RT-PCR reactions. Fragments corresponding to the last ~1200 nucleotides of the viral genome were obtained, cloned into pCR2.1-TOPO vector, digested with XbaI or AvrII restriction enzymes, and analyzed by agarose gel electrophoresis. The relative abundance for each virus in the population was calculated.