Methionine metabolism controls the B cell EBV epigenome and viral latency

Abstract

Epstein-Barr virus (EBV) subverts host epigenetic pathways to switch between viral latency programs, colonize the B cell compartment, and reactivate. Within memory B cells, the reservoir for lifelong infection, EBV genomic DNA and histone methylation marks restrict gene expression. But this epigenetic strategy also enables EBV-infected tumors, including Burkitt lymphomas, to evade immune detection. Little is known about host cell metabolic pathways that support EBV epigenome landscapes. We therefore used amino acid restriction, metabolomic, and CRISPR approaches to identify that an abundant methionine supply and interconnecting methionine and folate cycles maintain Burkitt EBV gene silencing. Methionine restriction, or methionine cycle perturbation, hypomethylated EBV genomes and de-repressed latent membrane protein and lytic gene expression. Methionine metabolism also shaped EBV latency gene regulation required for B cell immortalization. Dietary methionine restriction altered murine Burkitt xenograft metabolomes and de-repressed EBV immunogens in vivo. These results highlight epigenetic/immunometabolism crosstalk supporting the EBV B cell life cycle and suggest therapeutic approaches.

Keywords: dietary amino acid restriction; folate metabolism; gamma-herpesvirus; immunometabolism; lytic reactivation; methionine cycle; methionine metabolism; one-carbon metabolism; tumor virus; viral latency.

Figure 1.. Methionine restriction de-represses Burkitt latent membrane and lytic antigens

A) Schematic diagram of EBV latency programs. Six Epstein-Barr nuclear antigens (EBNA) and two latent membrane proteins (LMP) are expressed in the latency III program. The viral C promoter (Cp) and LMP promoter (LMPp) drive expression of EBNA and LMP, respectively. In latency IIa, the Q promoter (Qp) drives expression of EBNA1 and LMPp induces LMP1 and LMP2. In latency I, Qp drives EBNA1, the only EBV-encoded protein expressed. The immediate early gene BZLF1 promoter (BZLF1p) is silenced in each latency program to prevent expression of nearly 80 lytic cycle viral antigens. B) Confocal immunofluorescence analysis of LMP1 or LMP2 expression in P3HR-1 or Mutu I cells cultured in media with the indicated methionine concentrations for 3 days. Scale bar, 100 μm. Representative of n=3 independent replicates. C) Immunoblot analysis of whole cell lysates (WCL) from P3HR-1, Mutu I, or Kem I cells cultured in media with indicated concentration of methionine (Met) for 3 days, representative of n=3 experiments. Representative of n=3 independent replicates. D) Scatter plot visualization of Log2 fold-change in EBV transcript abundance (x-axis) from RNAseq analysis of Rael cells (x-axis) or P3HR-1 cells (y-axis) cultured in 10 vs. 100 μM Met for 3 days. Data are from the duplicate P3HR-1 and the triplicate Rael RNA-seq datasets. EBV immediately early (IE) lytic (red), early (blue), late (green) and latent (purple) genes are shown. E) Enrichr KEGG pathway −Log 10 (adjusted p-values) of gene sets significantly changed in 100 versus 10 μM Met, as in (D). F) FACS analysis of plasma membrane (PM) gp350 expression in P3HR-1 ZHT/RHT cells grown with 100 versus 10 μM Met for 3 days or reactivated with 4HT (1μM) and sodium butyrate (NaB, 3mM) as a positive control. Representative of n=3 independent replicates. G) qRT-PCR analysis of EBV intracellular genome copy number from cells described in (E). Total genomic DNA was extracted at 72h in the indicated media or at 48h post 4HT/NaB. P values were calculated by Student’s t-test. Mean ± SD values from n=3 replicates are shown. H) Volcano plot of metabolomic analysis of P3HR-1 cells grown in 100 versus 10 μM Met for 72 hours from four replicates. P values were calculated by Student’s t-test with the two-sample unequal variance assumption. Methionine cycle metabolites are highlighted in blue text. I) SAM/SAH ratio from cells treated as in (G). Mean ± standard deviation (SD) values from n=4 replicates are shown. P values were calculated by Student’s t-test. J) Metabolic pathway analysis plot showing the most strongly impacted metabolic pathways downregulated by MR. The x-axis represents pathway impact value computed from MetaboAnalyst 3.0 topological analysis, y-axis is the-log of the P-value obtained from pathway enrichment analysis. See also Figures S1–3.

Figure 2.. Methionine restriction alters the host and EBV epigenetic landscape.

A) 5mC dot blot analysis of DNA extracted from P3HR-1 cultured in 100 vs 10 μM Met for 72h. As a loading control, membranes were stained with ethidium bromide. B) Immunoblot analysis of WCL from P3HR-1 cells cultured in cultured in 100 vs 10 μM Met for 72h. C) Volcano plot of host epigenetic factor RNAseq from P3HR-1 (left) or Rael (right) cells cultured in 100 vs 10 μM Met for 72h, using the curated EpiFactors database (Medvedeva et al., 2015). P-value and Log2 fold change data were generated from n=2 replicates (P3HR-1) or n=3 replicates Rael by DESeq2, using default settings with Wald test and normal shrinkage, respectively. D) ChIP-qPCR analysis of IgG control versus anti-UHRF1 abundances in chromatin extracted from P3HR-1 cells cultured in 100 vs 10 μM Met for 72h. Shown are C promoter (Cp) qPCR mean ± SD values from n=3 replicates. P values were calculated by 2-way ANOVA. E-F) Immunoblot analysis of WCL from P3HR-1 or Rael cultured in 100 vs 10 μM Met for 72h. G) Immunoblot analysis of puromycin incorporation into newly synthesized polypeptides (top) or of the indicated proteins from WCL prepared from the indicated B cell lines cultured in 100 vs 10 μM Met for 72h. H) 5mC MeDIP analysis of chromatin from P3HR-1 cultured in 100 vs 10 μM Met for 72h. Shown are mean ± SD values from n=3 replicates of qPCR analysis of Cp, the LMP1 promoter (LMPp), the BZLF1 promoter (BZLF1p) or the BMRF1 promoter (BMRF1p). Cells were maintained in acyclovir (100μg/ml) to prevent lytic DNA synthesis. P values were calculated by 2-way ANOVA. I) ChIP for H3K9me3 or H3K27me3 of chromatin from P3HR-1 cells cultured in 100 vs 10 μM Met for 72h, followed by qPCR with primers specific for the C, LMP1, BZLF1 or BMRF1 promoters. Cells were maintained in acyclovir (100μg/ml) to prevent unchromatinized lytic DNA synthesis. Shown are mean ± SD values from n=3 replicates. P values were calculated by 2-way ANOVA. All blots are representative of n=3 replicates. See also Figure S3.

Figure 3.. Methionine cycle flux is necessary for maintenance of EBV latency I

A) Methionine cycle schematic. MTR, methionine synthase; MAT2A, methionine adenosyltransferase 2A; MTs, methyltransferases; AHCY, adenosylhomocysteinase; dcSAM, S-adenosylmethioninamine; MTA, 5’-methylthioadenosine; MTAP, S-methyl-5’-thioadenosine phosphorylase; SAM, S-adenosylmethionine; SAH, S-adenosylhomocysteine; Vit B12, vitamin B12. Schematic figure was modified from (Sanderson et al., 2019). B) Immunoblot analysis of WCL from P3HR-1 or Rael cells expressing the indicated control or MAT2A sgRNAs for six days. Blots are representative of n=3 experiments. C) FACS analysis of PM ICAM1 expression in P3HR-1 cells expressing control or MAT2A sgRNAs for six days. D) Mean ± SD PM ICAM1 MFI values from n=3 replicates of P3HR-1 cells expressing the indicated sgRNAs, as in (b). P values were calculated by Student’s t-test. E) Immunoblot analysis of WCL from P3HR-1 or Rael cells expressing control or AHCY sgRNAs for six days. Blots are representative of n=3 experiments. F) RNAseq analysis of Rael cells expressing control, AHCY, or MAT2A sgRNAs. The heatmap depicts Z-score standard deviation variation from the mean value for each EBV gene across 3 replicates. G) Mean ± standard error of the mean (SEM) intracellular methionine, methionine sulfoxide, homocysteine, SAM, SAH levels and SAM/SAH rations in Rael cells expressing control, MAT2A or AHCY sgRNAs. Intracellular metabolites were extracted at day 6 post sgRNA expression, prior to the obvious cell death. Mean ± SEM are shown for n=4 replicates. P values were calculated by 2-way ANOVA. H) qPCR analysis of EBV intracellular genome copy number from Rael cells expressing control, AHCY or MAT2A sgRNAs. Total genomic DNA was extracted at Day 8 post lentivirus transduction. Mean ± SD values from n=3 replicates are shown. P values were calculated by 2-way ANOVA. I) 5mC MeDIP analysis of DNA from Rael cells expressing control, AHCY, or MAT2A sgRNAs for six days, followed by qPCR using primers specific for the LMP1 or BZLF1promoters. Mean ± SEM are shown for n=3 replicates. P values were calculated by 2-way ANOVA.

Figure 4.. Methionine restriction inhibits EBV driven primary human B cell transformation.

A) Relative mRNA (bar chart) and protein (line chart) MAT2A or AHCY levels detected by RNAseq and proteomic analysis of primary human B cells at the indicated day post EBV infection (DPI) (Wang et al., 2019a; Wang et al., 2019b). B) Ion intensity abundance measurements of the indicated methionine cycle metabolites from primary human peripheral blood B cells at rest or at 1, 2, or 10 Days post-EBV infection. Mean ± SEM are shown for n=3 replicates. P values were calculated by Student’s t-test. C) Ion intensity abundance measurements of SAH (left) and of the SAM/SAH ratio (right) as in B. Mean ± SEM are shown for n=3 replicates. P values were calculated by Student’s t-test. D) Immunoblot analysis of WCL from primary human B cells at the indicated DPI and grown in 100 vs 10 μM Met, representative of n=3 replicates. DDX1 was used as a load control, since its protein levels do not change significantly following primary B cell infection by EBV (Wang et al., 2019b; Wang et al., 2019c). Blots are representative of n=3 experiments. E) FACS analysis of the percentage of PM CD23+ cells from EBV mock infected or EBV infected primary human B cells at 48 hours post infection (hpi) and grown in 100 vs 10 μM Met. Mean ± SEM are shown for n=3 replicates. P values were calculated by Student’s t-test. F) EBV transformation assays of primary human B cells grown in 100 vs 10 μM Met. Shown are fitted non-linear regression curves with means ± SEM from n=3 replicates. P values were calculated by Student’s t-test. G) FACS CSFE dye-dilution analysis of primary human B cells untreated, or stimulated as indicated by EBV multiplicity of infection (MOI)= 10, αIgM (1μg/ml), CpG (0.5mM), MEGA-CD40L (50ng/ml), IL-4 (20ng/ml) and cultured in RPMI with 100 vs 10 μM Met for 5 days. Representative of n=3 replicates. H) Mean ± SD percentage of proliferating primary human B cells from n=3 experiments as in G, obtained from three independent donors. P values were calculated by 2-way ANOVA. See also Figure S4.

Figure 5.. Folate cycle flux is necessary for maintenance for EBV antigens.

A) Schematic of interconnected folate and methionine metabolism one carbon cycles. THF, tetrahydrofolate. SHMT, serine hydroxymethyltransferase 1 and 2. MTHFD, methylene tetrahydrdofolate dehydrogenase 1 and 2. B) Immunoblot analysis of WCL from the indicated B cells, treated with DMSO, SHIN1 (10μM), or SHIN1(10μM) and sodium formate (1mM) for 3 days. Representative of n=3 replicates. C) FACS analysis of PM ICAM1 expression in Mutu I cells treated with DMSO, SHIN1 (10μM) or SHIN1 (10μM) and sodium formate (1mM) and cultured with RPMI with undialyzed FBS (bottom) versus dialyzed FBS (top) for 3 days. Representative of n=3 experiments. D) RNAseq heatmap analysis of P3HR-1 cells treated with DMSO, SHIN1 (10μM) and/or sodium formate (1mM) for 3 days, as indicated. Heatmap rows depict Z-score standard deviation variation from the mean value for each EBV gene, using data from n=2 replicates. E) Enrichr pathway analysis of gene sets significantly upregulated in SHIN1 versus DMSO-treated Rael, as in (E). Enrichr adjusted p-values are from n=2 samples. F) Scatter plot showing log2 fold changes in mRNA abundance in SHIN1 versus DMSO-treated P3HR-1 (y-axis) versus log2 fold changes in mRNA abundances in P3HR-1 grown in 100 vs 10 mM methionine for three days, as in Fig. 1, from n=2 replicates. G) Mean ± SEM intracellular serine, AICAR, IMP, methionine, methionine sulfoxide, SAM, SAH, homocysteine ion intensity abundance measurements in Rael cells treated with DMSO, SHIN1(10μM), or SHIN1 (10μM) and sodium formate(1mM) for 3 days. Mean ± SEM are shown for n=4 replicates. P values were calculated by Student’s t-test. H) Immunoblots of WCL from Mutu I cells cultured in serine and glycine free RPMI and dialyzed FBS for four days, with add back of serine and/or glycine, as indicated. Representative of n=3 replicates. I) 5mC MeDIP analysis of Mutu I cultured in replete, serine and/or glycine depleted media for four days, as indicated, followed by qPCR using primers specific for the LMP1p. Mean ± SEM are shown for n=3 replicates. P values were calculated by Student’s t-test. See also Figure S5. J) ChIP for H3K9me3 or H3K27me3 of Rael cells cultured in RPMI or serine depleted RPMI medium for 72h, followed by qPCR with primers specific for the LMP1 or Z promoters. Cells were maintained in acyclovir (100μg/ml) to prevent unchromatinized lytic DNA synthesis. Shown are mean ± SD values from n=3 replicates. P values were calculated by 2-way ANOVA.

Figure 6.. Dietary MR or pharmacological folate cycle blockade de-represses EBV immunogens in vivo.

A) Schematic of mouse xenograft control versus MR diet intervention. B-C) Volcano plot analysis of −Log10 (p-value) versus Log2 foldchange of metabolite abundances in n=3 tumors (B) or plasma samples (C) collected from mice fed MR vs control diets for 14 days. D) qRT-PCR analysis for the indicated EBV genes from tumors explanted from mice fed control or MR diets for 7 or 14 days. Mean ± SEM values in each group from n=6 tumors, obtained from bilateral flank tumors of n=3 mice, are shown. β-actin was used as the internal control. P values were calculated by Student’s t-test. E) Immunoblot analysis of WCL from tumors explanted from control or MR diet fed mice. Blots are representative of n=3 analyses. F) Immunohistochemical analysis of BMRF1 expression in tumors of mice fed control or MR diet for two weeks. G) Mean ± SEM numbers of BMRF1+ cells per 10X field as in (F). Circles represent values from each of fourteen 10X fields analyzed from two control versus MR tumors. P values were calculated by Student’s t-test. H) 5mC MeDIP analysis of DNA purified from tumors of mice fed control or MR diets for two weeks. Mean ± SEM values from n=3 tumors are shown. P values were calculated by Student’s t-test. I) Schematic of Burkitt xenograft SHIN2 experiments. Mice were treated with vehicle or SHIN2 200mg/kg intraperitoneally. Samples were collected at days 3 or 7 (d3 or d7) post-treatment. J) Immunoblot analysis of WCL from tumors taken from vehicle or SHIN2-treated mice at the indicated times. Blots are representative of n=3 replicates. See also Figures S5, S6 and S7.

Figure 7.. Schematic model of interconnected folate and methionine metabolism cycle roles in control of the Burkitt B cell EBV epigenome, latent and lytic gene expression.