The open chromatin landscape of Kaposi's sarcoma-associated herpesvirus

Abstract

Kaposi's sarcoma-associated herpesvirus (KSHV) is an oncogenic gammaherpesvirus which establishes latent infection in endothelial and B cells, as well as in primary effusion lymphoma (PEL). During latency, the viral genome exists as a circular DNA minichromosome (episome) and is packaged into chromatin analogous to human chromosomes. Only a small subset of promoters, those which drive latent RNAs, are active in latent episomes. In general, nucleosome depletion ("open chromatin") is a hallmark of eukaryotic regulatory elements such as promoters and transcriptional enhancers or insulators. We applied formaldehyde-assisted isolation of regulatory elements (FAIRE) followed by next-generation sequencing to identify regulatory elements in the KSHV genome and integrated these data with previously identified locations of histone modifications, RNA polymerase II occupancy, and CTCF binding sites. We found that (i) regions of open chromatin were not restricted to the transcriptionally defined latent loci; (ii) open chromatin was adjacent to regions harboring activating histone modifications, even at transcriptionally inactive loci; and (iii) CTCF binding sites fell within regions of open chromatin with few exceptions, including the constitutive LANA promoter and the vIL6 promoter. FAIRE-identified nucleosome depletion was similar among B and endothelial cell lineages, suggesting a common viral genome architecture in all forms of latency.

Fig 1

FAIRE-seq analysis of PEL (BCBL1). (A) Schematic of the KSHV genome (depicted linearly). Boxes indicate open reading frames (ORFs). ORFs on the upper strand are transcribed in rightward directions (with corresponding, predicted, or known regulatory regions on the left), and the lower strand corresponds to ORFs transcribed in the leftward direction (with corresponding, predicted, or known regulator regions on the right). (B) Read coverage data for FAIRE across the KSHV genome (BCBL1). Genome position is indicated on top. The maximal peak height was 289 reads covering a single nucleotide. (C) Read coverage of viral DNA from BCBL1 cells processed and sequenced as in panel B but not subjected to formaldehyde cross-linking. (D) Overlay of read coverage from formaldehyde cross-linked (black) and mock (gray)-treated BCBL1 cells. For comparison, raw count data were cube root transformed, median centered, and divided by their interquartile range/1.349. The normalized counts were then averaged using a 40-nucleotide sliding window. This procedure allows for a direct comparison of peak location, even though only approximately 1/10 of signal intensity was generated in the absence of cross-linking. (E) Predicted GC content across the KSHV reference genome NC_009333.

Fig 2

Regions of latent open chromatin across the KSHV genome (BC1). (A) FAIRE-seq reads from BC1 cells mapped to the KSHV reference genome NC_009333 are shown as in Fig. 1. Regions of increased coverage density correspond to regions of KSHV open chromatin. (B) FAIRE peaks in BC1 are identified as blocks and correspond to nucleotide-level resolution latent open chromatin as determined by MACS2. Regions of open chromatin occur in KSHV lytic replication origins (Lyt) and promoter regions and at intronic/intragenic sites during latency. (C) FAIRE-seq coverage of the viral terminal repeat (TR) region of KSHV. CTCF and LANA ChIP-seq enrichments (from GEO data sets GSM941710 and GSM941712, respectively; indicated by asterisks) are included for comparison. LANA binding sites (LBS; indicated by #) in the TR, as determined by Garber et al. (56), are also indicated. Numbers on the right indicate the scale of maximum coverage at the TR region. Nucleotide coordinates are based on NC_009333.

Fig 3

Regions of open chromatin are conserved in latent KSHV-infected endothelial and B cells. A heat map of normalized coverage counts across five cell lines (BC1, BCBL1, BJAB carrying latent KSHV, KSHV-HUVEC carrying latent KSHV, and L1-TIVE cells) is shown. Darker hues indicate nucleosome depletion, i.e., higher FAIRE coverage (averaged over a 40-bp sliding window). On top, the average coverage across all cell lines is shown.

Fig 4

Regions of open chromatin occur near activated histone modifications. Significant regions of overlapping tiled probe enrichment for histone modifications on the latent KSHV genome were annotated from previous work (20). Regions of H3K9/K14-ac and H3K4-me3 activating marks are shown in green and pink, respectively. FAIRE peaks are shown in light blue. H3K27-me3 and H3K9-me3 (not observed at these loci) modifications are shown in gray and red, respectively. The KSHV OriLyt-L region (A), the lytic control region (RTA promoter) (B), and the KSHV latency locus (C) are shown. Nucleotide boundaries in NC_009333 for significant tiled window enrichment from histone ChIP-chip and significant FAIRE-seq enrichment are indicated.

Fig 5

Regions of KSHV open chromatin coincide with CTCF binding sites. The published CTCF ChIP-seq enrichment (GSM941710) (23) is shown in relation to FAIRE-seq coverage. (A) The KSHV genomic region spanning the vIL6p region is shown (nt 11000 to 19000) along with two mapped transcription start sites depicted by black arrows. (B) The approximately mapped transcription start sites for the K12p and the antisense to latent transcripts (ALT) RNA (7) are indicated with gray arrows, and the OriLyt-R is shown (nt 117000 to 123000). (C) The bidirectional transcription start sites for the lytic LANApi and K14p promoters are indicated by gray arrows, and the constitutive LANApc is denoted with a black arrow (nt 123000 to 128000). Note changes in scale between panels. All nucleotide coordinates are based on NC_009333.

Fig 6

RNA PolII designates KSHV latent promoters. (A to D) Comparison of PolII (A), FAIRE (B), CTCF (C), and LANA (D) enrichment across the KSHV genome. The horizontal axis represents the genome location; the vertical axis represents the relative enrichment score over a 100-bp sliding window. Dots indicate significant peaks. (E) Normalization process. Shown on the horizontal axis is the unit and on the vertical axis is the cumulative density. Data are shown as deep sequence-derived coverage counts (raw), normalized counts {[n^1/3 − median (n^1/3)]/(interquartile range n^1/3/1.349)}, and log₁₀ of normalized counts, which are approximately normally distributed.