Figure 2

Transcriptionally active integration sites associate with AC status and clone abundance in weak HBZ binders. HTLV-1 unique integration sites from Japanese asymptomatic carriers (AC, black triangles) from Kagoshima and Kumamoto were compared to those from HAM/TSP patients (blue circles, from Kagoshima). Integration sites were stratified on the basis of predicted HBZ peptide binding affinity of host HLA class I alleles (strong binders, sHBZ, filled symbols; weak binders, wHBZ, open symbols) and binned by absolute abundance. Data is expressed versus an in silico generated random integration site dataset. (A) AC individuals (Kagoshima and Kumamoto cohorts combined) had a greater proportion of integration sites in genes than HAM/TSP patients (chi-squared test). Percentage of clones with integration sites in genes was correlated with clone abundance only in individuals with HLA class I alleles which could not bind HBZ (arrow, significant chi-squared test for trend). (B) Asymptomatic carriers had a higher frequency than HAM/TSP of H3K4me2 marks, enriched in transcriptionally active areas, within 10 Kb of integration sites (Mann-Whitney U test), and (C) a higher frequency of H3K4me1 marks, associated with enhancers. Mean epigenetic mark frequency near integration sites in a bin is divided by frequency near random sites. (D) In contrast, AC had a lower frequency of H3K9me3 marks, associated with constitutively heterochromatic DNA. Spearman correlation shows a significant (arrow) association between epigenetic mark frequency and log absolute abundance. Statistical comparisons AC vs HAM by Mann-Whitney U test after correction for multiple testing: * 0.05 > p > 0.01, ** 0.01 > p > 0.001, *** p < 0.001.