Viral reprogramming of the Daxx histone H3.3 chaperone during early Epstein-Barr virus infection

Abstract

Host chromatin assembly can function as a barrier to viral infection. Epstein-Barr virus (EBV) establishes latent infection as chromatin-assembled episomes in which all but a few viral genes are transcriptionally silent. The factors that control chromatin assembly and guide transcription regulation during the establishment of latency are not well understood. Here, we demonstrate that the EBV tegument protein BNRF1 binds the histone H3.3 chaperone Daxx to modulate histone mobility and chromatin assembly on the EBV genome during the early stages of primary infection. We demonstrate that BNRF1 substitutes for the repressive cochaperone ATRX to form a ternary complex of BNRF1-Daxx-H3.3-H4, using coimmunoprecipitation and size-exclusion chromatography with highly purified components. FRAP (fluorescence recovery after photobleaching) assays were used to demonstrate that BNRF1 promotes global mobilization of cellular histone H3.3. Mutation of putative nucleotide binding motifs on BNRF1 attenuates the displacement of ATRX from Daxx. We also show by immunofluorescence combined with fluorescence in situ hybridization that BNRF1 is important for the dissociation of ATRX and Daxx from nuclear bodies during de novo infection of primary B lymphocytes. Virion-delivered BNRF1 suppresses Daxx-ATRX-mediated H3.3 loading on viral chromatin as measured by chromatin immunoprecipitation assays and enhances viral gene expression during early infection. We propose that EBV tegument protein BNRF1 replaces ATRX to reprogram Daxx-mediated H3.3 loading, in turn generating chromatin suitable for latent gene expression.

Importance: Epstein-Barr Virus (EBV) is a human herpesvirus that efficiently establishes latent infection in primary B lymphocytes. Cellular chromatin assembly plays an important role in regulating the establishment of EBV latency. We show that the EBV tegument protein BNRF1 functions to regulate chromatin assembly on the viral genome during early infection. BNRF1 alters the host cellular chromatin assembly to prevent antiviral repressive chromatin and establish chromatin structure permissive for viral gene expression and the establishment of latent infection.

FIG 1

BNRF1 binds Daxx on its histone-binding domain (HBD) and copurifies with histone H3.3. (A) Diagram of HA-Daxx deletion constructs with functional domains indicated. (B) Immunoprecipitation (IP) assay of FLAG-BNRF1 and HA-Daxx deletion constructs, visualized by Western blotting. The input (6%) of each IP is shown in the left. (C) IP assay of FLAG-tagged histone H3.1 or H3.3 with Daxx and BNRF1, analyzed by Western blotting. (B and C) Molecular masses (in kilodaltons) are shown on the left.

FIG 2

The Daxx interaction domain (DID) of BNRF1 forms a complex with the Daxx/H3.3/H4 complex. (A) Diagram of GST-tagged BNRF1 DID construct. (B) GST pulldown of BNRF1-interacting nuclear proteins. Purified GST-BNRF1-DID or GST alone was used to pulldown cellular proteins from Mutu I cell nuclear extract. Total nuclear extract (input) and eluted proteins from GST alone or GST-BNRF1-DID were analyzed by Western blotting and probed for Daxx, histone H3.3 and GST, with ATRX and PARP probed as negative controls. Molecular masses (in kilodaltons) are shown on the left. (C) (i to iv) Gel filtration chromatograms of BNRF1-DID and Daxx/H3.3/H4 complexes. Bacterially produced, purified BNRF1-DID, H3.3/H4 tetramer, and Daxx/H3.3/H4 complex were passed through a Superdex S75 column in different combinations. (vi to x) Fractions were collected and analyzed by SDS-PAGE and Coomassie blue staining.

FIG 3

FGARAT homology domains of BNRF1 are essential for dissociation of ATRX from Daxx. (A) Sequence comparisons of EBV BNRF1 with other homologous genes, including KSHV ORF75, MHV68 ORF75c, human FGARAT, and Salmonella Typhimurium PurL. Arrowheads denote residues on Salmonella PurL that form hydrogen bonds with ADP (54). (B) Diagram of FLAG-BNRF1 constructs with deletions in FGARAT-homology conserved regions. (C) IP-Western assay of FLAG-tagged BNRF1 FGARAT-homology mutants in transfected 293T cells and probed for association with Daxx and ATRX.

FIG 4

WT BNRF1 mobilizes histone H3.3, whereas the d26 or dATPase mutants do not. (A) Images of nuclei expressing GFP-H3.3 only, or GFP-H3.3 and WT or d26 BNRF1, at three different time points. Time (T) = 0 s is immediately prior to photobleaching, T = 1 s is immediately after photobleaching. The relative fluorescence of the photobleached region at T = 1 s is representative of the free pool of GFP-H3.3. (B and C) FRAP analysis of histone H3.3 in cells expressing detectable levels of RFP-BNRF1 compared to cells with no detectable RFP signal. Line graphs present the relative normalized fluorescence intensity of the photobleached nuclear region plotted against time; averages ± the standard errors of the mean (SEM) are indicated. Signal readouts from cells that coexpressed detectable levels of GFP-H3.3 and WT (B) or d26, dATPase RFP-BNRF1, or free RFP (C) are shown in red; signal readouts from cells that expressed detectable levels of only GFP-H3.3 are shown in green. (D) Bar graphs present the free pool of GFP-H3.3 in cells expressing detectable levels of WT, d26, or dATPase RFP-BNRF1 or free RFP, relative to the levels in cells expressing detectable levels of only GFP-H3.3 on the same coverslip; averages ± the SEM are shown. *, P < 0.05; **, P < 0.01. (E) Average slow GFP-H3.3 exchange rate in cells expressing detectable levels of WT, d26, or dATPase RFP-BNRF1, or free RFP, relative to cells expressing detectable levels of only GFP-H3.3 on the same coverslip; averages ± the SEM are shown. *, P < 0.05; **, P < 0.01.

FIG 5

Viral gene expression in the prelatency phase is repressed by ATRX in the absence of BNRF1. (A) Western blot validation of shRNA-mediated Daxx or ATRX depletion in EBV-negative Akata cells, with nontargeting shNeg as a negative control. (B) qPCR assay of viral gene expression in Daxx/ATRX-depleted EBV-infected cells. EBV-negative Akata cells with Daxx or ATRX depleted were infected with ΔBNRF1-bacmid produced virus and collected 48 hpi for qRT-PCR quantification of mRNA levels. **, Statistical significance (P < 0.01) as determined by a one-tailed Student t test. Error bars indicate the standard deviations (SD). (C) Confirmation of siRNA-mediated BNRF1 depletion in virus producer 293 cells by Western blotting. Wild-type or BNRF1-knocked-down virus was collected from 293 producer cells stably transfected with EBV genomes and treated with control (siCtrl) or BNRF1-specific siRNA (siBNRF1). This batch of virus was used for the ChIP assays in Fig. 8A. (D) qPCR assay of latent viral gene expression in siCtrl or siBNRF1 EBV-infected primary B lymphocytes collected at 72 hpi.

FIG 6

Tegument-derived BNRF1 is stable and interacts with Daxx during early infection, whereas no de novo expression of BNRF1 could be detected. Mutu I strain EBV-infected primary B lymphocytes in the prelatency phase were collected in a time course and subject to Western blot analysis of BNRF1 protein (A) and qPCR assay of viral gene transcripts (B). Error bars indicate the SD. (C) EBV-infected B cells were collected 2 days postinfection and subject to immunoprecipitation with BNRF1 antibody-conjugated beads.

FIG 7

BNRF1-depleted virus infection showed less Daxx/ATRX dispersion and more viral DNA located next to Daxx/ATRX foci. siCtrl, siBNRF1 EBV-infected or mock-infected primary B lymphocytes in the pre-latent phase (72hpi) were subject to IF-FISH to visualize the infecting viral DNA in green, in conjunction with ATRX (A) or Daxx (B) in red. Cell nuclei were visualized with the blue DNA stain, DAPI. In siBNRF1 infections, arrowheads indicates where viral DNA is localizing adjacent to ATRX (Aiii) or Daxx (Biii) foci, insets showing zoomed-in views of the adjacent event. (C) Bar graph showing percentage of infected cells (green positive) with dispersed (high nuclear background red signals with weak foci) ATRX or Daxx signals. (D) Percentage of infected cells with ATRX or Daxx adjacent to viral DNA (red foci next to green foci). Approximately 60 to 100 infected cells counted per slide. Statistical significance was determined by using a one-tailed Student t test; error bars indicate the SD.

FIG 8

BNRF1 decreases histone H3.3, Daxx, and ATRX binding and promotes H3K4me3 accumulation in viral genomic regions. BNRF1-depleted EBV-infected primary B lymphocytes in the prelatency phase were tested for histone H3.3, Daxx, and ATRX binding on viral DNA. Human primary B lymphocytes infected with siCtrl- or siBNRF1-treated cell-produced virus were collected at 72 hpi and subjected to ChIP to test for H3.3, H3K4me3, and H3K9me3 (A) and Daxx or ATRX (B) binding on viral genomic regions by qPCR. The regions tested included the following: a CTCF binding site close to the terminal repeats (CTCF 166); the W promoter (Wp), which drives EBNA2 expression in the prelatency phase; the Z promoter (Zp), which drives the immediate-early gene Zta; and a region in the EBNA2 gene body. Error bars indicate the SD. (C) Bar graph of ChIP signals of H3K4me3 or pan-H3 on siCtrl or siBNRF1 virus-infected primary B cells, normalized to the average pan-H3 values.