Structural divergence creates new functional features in alphavirus genomes

Abstract

Alphaviruses are mosquito-borne pathogens that cause human diseases ranging from debilitating arthritis to lethal encephalitis. Studies with Sindbis virus (SINV), which causes fever, rash, and arthralgia in humans, and Venezuelan equine encephalitis virus (VEEV), which causes encephalitis, have identified RNA structural elements that play key roles in replication and pathogenesis. However, a complete genomic structural profile has not been established for these viruses. We used the structural probing technique SHAPE-MaP to identify structured elements within the SINV and VEEV genomes. Our SHAPE-directed structural models recapitulate known RNA structures, while also identifying novel structural elements, including a new functional element in the nsP1 region of SINV whose disruption causes a defect in infectivity. Although RNA structural elements are important for multiple aspects of alphavirus biology, we found the majority of RNA structures were not conserved between SINV and VEEV. Our data suggest that alphavirus RNA genomes are highly divergent structurally despite similar genomic architecture and sequence conservation; still, RNA structural elements are critical to the viral life cycle. These findings reframe traditional assumptions about RNA structure and evolution: rather than structures being conserved, alphaviruses frequently evolve new structures that may shape interactions with host immune systems or co-evolve with viral proteins.

Figure 1.

The Sindbis virus genome contains a multitude of diverse RNA structures. (A) Top: schematic of the virus genome organization, with annotated elements. Middle: SHAPE data for the Sindbis virus genome, represented by the local median (55-nt window) compared with the global median. Reactivities below the x-axis indicate a region more structured than average. Gray lines denote the conserved sequence element (5′ CSE), which has low SHAPE reactivities and is highly structured. Bottom: sequence conservation at each position, based on sequence identity and gappiness, from a multiple sequence alignment of 37 alphaviruses. The protein-coding sequence contains both well-conserved (black and dark gray) and less-conserved (light gray) regions. (B) Top: median (55-nt window) Shannon entropies of base pairing across the SINV genome. Middle: Maximum squared z-score at each position in the genome, with higher values corresponding to greater structural significance. Bottom: structured regions in the SINV genome, based on the intersection of regions with low SHAPE and low z-scores. (C) SHAPE-directed structural models of SINV structured regions. Nucleotide color indicates low, medium, or high SHAPE reactivity. (D) Windowed correlation coefficients of SHAPE data between the SINV and VEEV genomes. The dashed line indicates the top 1% of correlation coefficients. SHAPE data within the 5′ CSE are among the most correlated within the genome, indicating high structural conservation within that region. (E) Distribution of windowed correlation coefficients of SHAPE data. Red: a background distribution, blue: correlation coefficients between SINV and VEEV, green: correlation coefficients of two biological replicates of a virus. Although SINV and VEEV are more correlated than expected at random, there is little overlap with the correlations of the same virus, indicating little widespread correlation. Dashed line indicates top 1% of SHAPE correlations between SINV and VEEV. (F) SHAPE data of the 5′ CSE in CHIKV, SINV, and VEEV. Within the 5′ CSE, the SHAPE profiles are very similar, representing conservation of structure, but the correlation immediately disappears outside of the 5′ CSE. (G) SHAPE-directed structural models of the CSE in CHIKV, SINV and VEEV. The 5′ CSE structure is compatible with the SHAPE data and conserved in all three viruses. (H) Distribution of alignment-derived sequence conservation scores in the entire alignment (left) and 5′ CSE only (right). Dot indicates the median, with the line extending from the 25th to 75th percentile.

Figure 2.

Structure-disrupting mutations successfully confirm function by impeding virus growth. (A) Method used to disrupt RNA structure. Trinucleotide sets are shuffled, changing the nucleotide sequence while the amino acid sequence is preserved. (B) Left: structure of the hairpin + CSE element. Start codon and stem loops 3 and 4 are indicated. Right: structure of region overlapping packaging signal. Blue circles indicate positions that are mutated to disrupt the structure. Green circles indicate previously observed GGG motifs within packaging signal structure (9). Nucleotide color represents SHAPE reactivity. Violin plot displays conservation scores within the packaging signal region. (C) Growth curves for SINV WT (black), mutated hairpin + CSE (gold), and packaging signal (blue) in Vero81 cells at a MOI of 0.01. Shading indicates standard error. Both structures are necessary for optimal virus growth.

Figure 3.

Novel virus structures tune SINV growth. (A) Structures for the new nsP1 structured region (nsP1 SR; left) and the new nsP3 structured region (nsP3 SR; right). Nucleotide color represents SHAPE reactivity. Violin plots display sequence conservation scores within each region. Blue circles indicate positions that are mutated. (B) Growth curves for SINV WT (black), the nsP1 SR (green), and the nsP3 SR (red). Mutant growth is nearly identical to WT in both Vero cells (left) and NIH/3T3 cells (right). (C) Specific infectivity of mutant viruses. The nsP1 SR mutant has a large defect in infectivity. Graph is a representative experiment of three or more replicates. (D) Genome transcription levels of WT and nsP1 SR mutants, measured by qRT-PCR. The nsP1 SR mutant has a defect for genome transcription. (E) RNA translation of WT and nsP1 SR. Expression of the nonstructural proteins is impaired in nsP1 SR compared to WT as measured by probing for nsP3.

Figure 4.

Outside of the conserved sequence element, SINV functional structures are not conserved. (A) Structure compatibility scores for the four tested sequences, with the phylogenetic tree on the left. The SINV strain used for the reference sequence and structure is bolded and in red. The structural model for the 5′ CSE is highly compatible with other alphavirus sequences, but the other functional structures are less conserved. For the new region in nsP3, the structure essentially does not exist outside of closely related strains. (B) R-scape results for structure-informed alignments. X-axis: alignment length; y-axis: percentage of base pairs in structure found by R-scape; color: number of true positives found by R-scape. Bold typeface indicates R-scape results from SINV structure-informed alignments, whereas regular typeface indicates R-scape results for alignments from (21). While no SINV region has anywhere near the amount of covariation of highly-conserved RNA structures, the region containing the 5′ CSE is notable for the highest number of covarying base pairs within SINV.