The extent to which LMP1 TES2 requires NF-κB activity to affect target genes has not previously been characterized on either a genome-wide scale or in a non-B-cell context. We chose a 293 cell background, because LMP1 TES2 NF-κB inhibition does not result in cell death. Surprisingly, despite robust activation of the p38, ERK, and JNK pathways, NF-κB activity was critical for LMP1 TES2 effects on all but 5 target genes. The profound IκBα superrepressor effect is at least in part due to an important direct role for NF-κB transcription factors in target gene regulation. NF-κB may serve as a master controller of transcriptional responsiveness to LMP1 TES2 signaling, as has been reported for the inflammatory response (46
). Of note, both canonical and noncanonical NF-κB may be affected by the IκBα superrepressor, given the extensive cross talk between the two pathways. Alternatively, LMP1/NF-κB may regulate expression of a target gene(s) important for transcriptional responsiveness. For example, NF-κB activity may be required for the synthesis of enzymes that directly or indirectly modify chromatin, such as histone acetyl-deacetylases, methyltransferases, ubiquitin ligases, or kinases. In this fashion, NF-κB may regulate the ability of other transcription factors to gain access to their DNA-binding sites (24
). Additional MAPK pathway dependence has been described for a limited number of LMP1 TES2-affected RNAs, including IL-8 (16
). The extent to which p38, JNK, ERK, and IRF7 pathways are also required remains an open question for most LMP1 TES2 targets. Nevertheless, these data indicate that LMP1 TES2 broadly alters cell gene regulation in an NF-κB-dependent fashion. The results presented here provide new evidence for the therapeutic importance of canonical NF-κB activation in EBV-associated malignancies and perhaps for malignancies with elevated NF-κB states more generally. Numerous canonical NF-κB pathway inhibitors are currently in clinical development and may have more pronounced effects on EBV-infected cells than previously envisaged (11
Inducible LMP1 TES2 expression significantly affected 1,916 cell RNAs >2-fold with an FDR of <0.001, upregulating 1,479 RNAs and downregulating 437 RNAs. In contrast, TNF-α treatment of 3T3 fibroblasts for 12 h upregulated only 180 RNAs and downregulated 70 RNAs, based on a 2-fold cutoff (22
). Similarly broad LMP1 effects were observed on 1,926 RNAs upon electroporation of LMP1 into purified human germinal center B cells (64
). Likewise, LMP1 upregulated by >2-fold 131 of 1,905 (7%) RNAs sampled in Burkitt lymphoma BL41 cells (10
). Twenty-seven of these 131 (21%) BL41 LMP1 target genes were also 2-fold upregulated by LMP1 TES2 in 293 cells. These included the antiapoptotic proteins BCL2, cIAP2, MCL1, and cFLIP (CFLAR), the chemokine CXCL10, the cytokine IL-32, the antigen presentation molecules TAP1 and HLA-DQB1, the transcription factors p100 and JUNB, the NF-κB negative regulators IκBα, A20, and CYLD, the ubiquitin ligase CBLB, and cell surface receptors (FAS, CD44, CD58, and CD83). Differences in microarray platform and the presence of both LMP1 TES1 and TES2 limited comparisons between our 293 cell results and available LMP1 B-cell data sets.
We observed a greater overlap between LMP1 293-upregulated genes and RelA LCLs by Chip-Seq analysis, where RelA occupancy was detectable at 58.6% of LMP1 TES2 293 cell targets. Taken together, these data suggest that there is a significant but limited overlap in LMP1 TES2-regulated genes in 293 versus B-cell contexts. Differences in chromatin accessibility between cell types may profoundly shape transcriptional responses to LMP1 TES2, as has been observed for inflammatory gene expression programs in macrophages versus fibroblasts (18
). Differences in basal expression levels between 293 and BL41 cells also contribute to the observed LMP1-induced fold changes.
The onset of LMP1 TES2 expression was tightly coupled to LMP1 effects on RNA levels. Concurrent with LMP1 TES2 expression, NF-κB and MAPK pathway activation were evident by the 6-h time point (). LMP1 TES2 affects the abundance of a small subset of RNAs by 6 h and substantially affects most target genes by the 9-hour time point. Similarly diverse kinetic responses have been observed following stimulation with proinflammatory stimuli, such as TNF-α or lipopolysaccharide (LPS) (22
). Although TNF-α and LPS trigger rapid NF-κB nuclear translocation, NF-κB recruitment to individual target gene loci is highly asynchronous (18
). NF-κB immediately binds to regulatory elements of so-called “fast gene” loci. Yet, at “slow gene” loci, binding to regulatory elements occurs hours after NF-κB nuclear translocation. Differences in chromatin configuration may underlie this phenomenon. Fast gene loci constitutively exhibit an open chromatin state, which allows NF-κB transcription factors immediate access to their binding sites. By contrast, slow gene loci must undergo chromatin remodeling prior to NF-κB binding (18
Transient changes associated with chromatin remodeling might also underlie the observation that LMP1 TES2 initially downregulates the abundance of many RNAs that are subsequently upregulated (, cluster 3) and initially upregulates RNAs that are subsequently downregulated (cluster 5). Alternatively, early LMP1-affected RNAs may encode important transcriptional regulatory functions. Genes that are expressed early may directly or indirectly affect transcription factor expression or functionality and, in aggregate, reverse the direction of affected gene transcriptional regulation. For example, one or more genes in cluster 1 may have profound transcriptional regulatory effects. Identification of the gene(s) responsible for these effects would be of considerable interest. Surprisingly, the abundance of most cell RNAs did not change substantially after 9 h. These results contrast with a prior study, in which LMP1 induction of IL-1α and IL-1β was suggested to secondarily account for most LMP1 effects on epithelial cell RNA regulation (45
). Indeed, LMP1 TES2-inducible expression activates NF-κB more robustly than culture of the cells with high concentrations of IL-1β (unpublished data).
In contrast to many physiologic NF-κB stimuli that transiently activate NF-κB and are then silenced by multiple negative feedback regulatory loops, LMP1 TES2 constitutively activated NF-κB. TNF-α stimulation of murine embryonic fibroblasts induced more heterogenous RNA changes and distinctive RNA changes (22
). TNF-α-upregulated RNAs peaked at 30 min or 2 h, or they continued to increase across 12 h of TNF-α stimulation. TNF-α transiently upregulated many RNAs, despite continued TNF-α stimulation (22
). Transcription and mRNA stability contributed to these RNA level effects (22
), as likely underlies some of the RNA changes observed here. In multiple instances, TNF-α transiently upregulated the same target genes that LMP1 TES2 persistently upregulates. TNF-α signaling pathways may have evolved to be dependent on multiple factors in addition to NF-κB to limit the duration of inflammatory or innate immune responses. In contrast, LMP1 TES2 may have evolved to provide EBV-transformed cells with relatively persistent NF-κB and MAPK target gene stimulation to better enable cell survival and growth.
Negative regulators of canonical NF-κB, including IκBα, CYLD, A20, TAX1BP1, ABIN1, optineurin, TANK, PPM1B, and ZC3H12C, were among the earliest and most highly TES2-upregulated genes. Nevertheless, LMP1 ligand-independent high-level constitutive forward signaling overcomes much of this underlying negative feedback regulation. Consistent with this possibility, TES2 also induces multiple LMP1/NF-κB pathway activation pathway components, including the kinases TPL2, TAK1, IKKα, and IKKβ. These NF-κB pathway components () may facilitate robust forward signaling. Small interfering RNA knockdown of each kinase impairs TES2-mediated NF-κB activation in HEK-293 cells (unpublished data). Alternatively, TES2 may alter the activity of negative regulators through undefined posttranscriptional mechanisms.
LMP1 is expressed at a high level in latency III infection of LCLs, where it is essential for lymphoblastoid cell growth, both through TES1 noncanonical and TES2 canonical NF-κB activation. LMP1 is also expressed at very high levels in HD and at a variable level in NPC. A robust inflammatory response is commonly observed in EBV latency III-associated lymphoproliferative diseases, HD and NPC. Indeed, NPC was initially described as a lympho-epithelial malignancy. Proinflammatory cytokines promote tumor development, progression, and metastasis (2
). LMP1 TES2 upregulates both chains of the IL-6 receptor complex. While TES2 did not significantly upregulate IL-6, LMP1 TES1 has been reported to upregulate IL-6 in epithelial cells (15
). IL-6 provides a link between inflammation and cancer (30
). IL-6 levels are elevated in both serum and tumor tissues of NPC patients and decrease with successful tumor therapy (59
). Thus, LMP1 TES1 and TES2 may function in concert to provide significant growth factor support to cells of patients with latency III-associated lymphoproliferative diseases, HD and NPC. Anti-IL-6 monoclonal antibodies are currently in trial for treatment of EBV-positive posttransplant lymphoproliferative disorders (47
IL-8, CCL20, and CXCL10 are among the RNAs most highly upregulated by TES2. CCL20 is a powerful chemoattractant for immature dendritic cells and effector/memory B and T lymphocytes, particularly at skin and mucosal surfaces (52
). An IL-8 promoter polymorphism has been associated with NPC susceptibility and aggressiveness (5
). CXCL10 is highly upregulated in primary NPC clinical samples and is LMP1 dependent (36
). LMP1 TES2 canonical NF-κB activation appears to underlie CXCL10 expression. Furthermore, LMP1 TES2 highly upregulates IL-32, a recently discovered cytokine that further activates MAPK and NF-κB pathways. IL-32 is also implicated in the development of epithelial cancers (17
). IL-32 is likewise upregulated in BL41 cells upon LMP1 expression (10
). Encode Project LCL Chip-Seq data have identified high-level RelA occupancy at the IL-32 locus, together with RNA polymerase 2 and histone modifications, suggestive of active transcription (31
). Small-molecule inhibitors of LMP1-mediated canonical NF-κB activation may have antineoplastic activity by blocking both intracellular and intercellular effects that sustain NPC or HD cell growth or survival.
In conclusion, the studies presented here provide the first whole-genome kinetic analysis of LMP1 TES2 transcriptional effects and demonstrate a surprisingly robust LMP1 TES2 NF-κB dependence. The extent to which LMP1 TES1 also requires NF-κB to regulate target genes remains incompletely defined. Likewise, the amount of overlap between LMP1 TES1 and TES2 transcriptional targets remains unknown. While several LMP1 target genes, such as TRAF1 and the epidermal growth factor receptor, are preferentially induced by LMP1 TES1, other targets, such as Fas and intercellular adhesion molecule 1, are similarly upregulated by both LMP1 TES1 and TES2 (13
). Given the importance of both TES1 and TES2 signaling in EBV-associated malignancies, genome-wide analysis of TES1 transcription effects is of significant interest.