Fig. 1.
Genome-wide structure mapping of DENV and ZIKV genomes inside virions. a Schematic showing the workflow for identifying functional structural elements in eight viruses. Full-length virus genomes are probed inside their virus particles using a SHAPE-like chemical, NAI, which modifies single-stranded regions along the genome. Pair-wise interactions within the virus genomes are also interrogated using biotinylated psoralen, which crosslinks base-paired regions inside the virus. The local and pair-wise experimental data are then used to constrain computational models to derive more accurate structure predictions for DENV and ZIKV. We also identify structurally conserved virus structures and determine the functional impacts of conserved and pair-wise interactions through mutagenesis and virus fitness assays. b NAI-MaP reactivities and Shannon entropy along viral genome. Top: raw NAI-MaP reactivities along DENV1: Y-axis indicates the extent of NAI reactivity, and X-axis indicates position along the genome. Red, yellow, and black bars indicates high, medium, and low reactivities respectively. Bottom: Shannon entropy across the DENV1 genome. Low values indicate higher probability of having a defined structure. c, d NAI-MaP structure signals maps to known structures in the 5′ UTR and 3′ UTR of DENV2 virus genomes. A higher NAI-MaP reactivity is colored to be more red, indicating that a base is likely to be single-stranded. d Top: scatterplot showing the sequence identity and structure similarity derived by NAI-MaP between each pair of DENV strains in 100-nt windows. The red dots show locations whereby structure similarity is greater than sequence similar by 1 standard deviation from mean. Bottom: each row indicates the locations of the high structure/sequence similarity regions (top) in each DENV pair, along the DENV genome. The structure conserved regions are consistently located at specific locations along the genome (200-, 900-, 2000-, 3800-, 4000-, 4200-, 4600-, 6500-, 7000-, 10,400-, and 10,600-nt regions, starred), indicating that the high structure conservation is not random and is an evolutionary constraint on structure beyond constraints on sequence alone
