Introduction

Epstein–Barr virus (EBV), a member of the Herpesviridae family, was the first tumorigenic virus discovered in humans, which has been linked to a variety of cancers, including epithelial malignancies and lymphomas1,2. EBV-associated epithelial cancers mainly include nasopharyngeal carcinoma (NPC) and a subgroup of gastric carcinoma (GC)2,3. EBV persists in a latent state in tumor cells, which enables virus-targeted therapies. As have been reported, reactivation of EBV from the latent form to the lytic form causes tumor cell death4,5,6. In previous studies, histone deacetylase inhibitors (HDACi), DNA methyltransferase inhibitors, proteasome inhibitors, and other chemical compounds have been identified as lytic inducers5,7,8. However, the major weaknesses of lytic induction therapy are the relatively low reactivating efficiencies and the concern of infectious virion production that might aggravate virus-associated diseases or promote oncogenesis9,10,11.

Curcumin, a polyphenol derived from Curcuma longa, exhibits anticancer ability and therapeutic potential12,13. Previous studies have respectively revealed that curcumin and it analogs induce the unfolded protein response (UPR)14,15 and trigger EBV lytic cycle in GC and NPC cells16. UPR is a signal transduction pathway activated when unfolded or misfolded proteins accumulate in the endoplasmic reticulum. It is reported that the UPR induces EBV lytic cycle in Burkitt lymphoma cells17. Here we investigated the effects of C210 on the UPR and the expression of EBV lytic gene products.

Heat shock protein 90 (Hsp90) is a highly conserved molecular chaperone that promotes proper folding and stability of multiple oncogenic proteins, including those encoded by EBV18,19. Hsp90 inhibitors promote proteasomal or autophagic degradation of Hsp90 client proteins and facilitate antitumor effects20,21,22. We previously identified curcumin as a new type of Hsp90 inhibitor23 and then synthesized a list of novel curcumin derivatives24,25,26,27. One of these derivatives that have displayed more potent Hsp90-inhibiting and anticancer effects than curcumin is C210 ((1E,6E)-4-(4-hydroxy-3-methoxybenzyl)-1,7-bis (3,4,5-trimethoxy phenyl) hepta-1,6-diene-3,5-dione)27.

Signal transducer and activator of transcription 3 (STAT3) is a known client protein of Hsp9028. High levels of STAT3 have been reported to diminish the susceptibility of EBV latently-infected cells to signals that activate the lytic cycle while inhibition of STAT3 function or reduction of STAT3 levels enhance susceptibility to lytic stimuli29,30. These findings raised the possibility that C210, as an Hsp90 inhibitor, may reduce the levels of STAT3, thus enhancing the EBV-reactivating effects and anticancer ability of other conventional lytic inducers, for example, Suberoylanilide hydroxyham acid (SAHA), a histone deacetylase inhibitor.

Previous work in the Kenney laboratory has presented that Hsp90 inhibitors can also block the production of infectious EBV virions22. It has remained unexplored whether a curcumin derivative with Hsp90 inhibitory activity could significantly induce EBV lytic cycle and meanwhile inhibit the production of infectious virus particles.

We investigated the effects of C210 on EBV lytic induction and infectious virion production with or without the presence of SAHA in EBV-positive NPC and GC cells. We also studied the mechanism by which C210 exerted effects on EBV life cycle. We showed that C210 activated expression of EBV lytic gene products and inhibited production of infectious virus particles through disruption of Hsp90 function. Our findings demonstrate the importance of cellular Hsp90 in EBV lytic induction therapy and C210 or its derivatives as potentially safe, efficacious EBV lytic activators.

Materials and methods

Cell culture

HONE1-EBV31, an EBV-positive NPC cell line, and AGS-EBV32, an EBV-positive GC cell line, were obtained from Institute for Advanced Study, Central South University and used in this study. Both cell lines carry a recombinant EBV genome with an insertion of the neomycin-resistance gene and the green fluorescent protein (GFP) gene. Both cell lines were cultured at 37℃ with 5% CO2 in growth medium supplemented with 10% fetal bovine serum and 400 µg/ml G418. The HONE1-EBV cells were maintained in RPMI-1640 and AGS-EBV cells in Ham’s F12.

Reagents and antibodies

C210 was synthesized in our laboratory and prepared as a 20 mmol/L stock solution in dimethylsulfoxide (DMSO). Suberoylanilide hydroxamic acid (SAHA), phorbol 12-myristate 13-acetate (TPA) were purchased from MedChemExpress (Shanghai, China). Sodium Butyrate (NaB) was purchased from Aladdin-e (Shanghai, China). Antibodies against Zta, EaD, gp110 and gp350 were purchased from Santa Cruz Biotechnology (Dallas, Texas, USA). Antibody against Hsp90α, p-IRE1α and XPB was purchased from Abcam (Cambridge, UK). Antibody against XBP1s, Acetyl-H3 and β-actin was purchased from Cell Signaling Technology (Danvers, MA, USA). Antibody against STAT3 was purchased from Abclonal Technology (Wuhan, China).

Quantitative RT-PCR (qRT-PCR) for detecting viral and cellular RNA

The cells were treated and harvested, and total RNA was extracted using an RaPure Total RNA Mini Kit (Magen, Guangzhou, China). The reverse transcription (RT) step was performed using Hifair AdvanceFast One-step RT-gDNA Digestion SuperMix (Yeasen, Shanghai, China) according to the manufacturer’s protocol. The primers used for qRT-PCR are listed as follows:

  • BZLF1 (Zta) F: ACATCTGCTTCAACAGGAGG.

  • BZLF1 (Zta) R: AGCAGACATTGGTGTTCCAC.

  • BMRF1 (EaD) F: CTGCCGTTGGATCTTAGTGTTAT.

  • BMRF1 (EaD) R: AGGAGATGGACTGACCGTATG.

  • BLLF1 (gp350) F: GCAACAAGTAAGCCTGGAATCT.

  • BLLF1 (gp350) R: CCTCACTACTGCCGTTATATTGG.

  • BALF4 (gp110) F: AACCTTTGACTCGACCATCG.

  • Hsp90AA1 F: TGACCATTCCATTATTGAGACCTT.

  • Hsp90AA1 R: TCCAGACTGAAGCCAGAAGA.

  • XBP1s F: TGCTGAGTCCGCAGCAGGTG.

  • XBP1s R: GCTGGCAGGCTCTGGGGAAG.

  • STAT3 F: AGAAGGACATCAGCGGTAAGA.

  • STAT3 R: GGATAGAGATAGACCAGTGGAGAC.

  • BFRF3 F: TGAACCAGAATAATCTCCCCAATG.

  • BFRF3 R: GCCGAGGCACCCCAAAAGTC.

  • BcLF1 F: GGATGCCGCCTATGAATACC.

  • BcLF1 R: CTGTACTCGTTGACCATGTTGT.

  • GAPDH F: TCTTTTGCGTCGCCAGCCGA.

  • GAPDH R: AGTTAAAAGCAGCCCTGGTGACCA.

Western blot analysis for determination of protein level

Total proteins from cells following corresponding treatment were extracted as previously described33 and the protein concentration was determined with a bicinchoninic acid (BCA) assay. Proteins were loaded onto 10% SDS polyacrylamide gels for separation and then transferred to 0.2 μm polyvinylidene fluoride (PVDF) membranes. After that, the membranes were blocked with 5% nonfat milk, then incubated with specific primary antibodies and corresponding secondary antibodies (1:5000). Protein bands were measured using an ECL detection kit and visualized using a ChemiDoc MP imaging system.

Immunoprecipitation

HONE1-EBV and AGS-EBV cells were treated with 4 μmol/ L C210 or vehicle for 12 h. The cells were collected, lysed, and subjected to centrifugation. The concentration of protein was identified utilizing a BCA assay. Afterwards, 1 mg of lysate was incubated with 4.5 μg of anti-Hsp90α antibody overnight at 4 °C. Then, 20 μL of Protein A/G Magnetic Beads (MedChemExpress, China) was added to the mixture and incubated at room temperature for 4 h. The immunoprecipitated protein complexes were washed in TBST, resuspended in SDS gel loading buffer and boiled for 5 min. The proteins were then loaded onto SDS-PAGE followed by transfer to PVDF membranes. The membranes were immunoblotted with anti-Hsp90α, anti-STAT3 and anti-XPB antibodies and the signals were detected by enhanced chemiluminescence.

Raji infection assay

To examine the production of infectious viral particles upon pharmacologic induction of EBV lytic cycle, HONE1-EBV and AGS-EBV cells were treated with C210 and/or SAHA for 5 days. After centrifuging the cells, the supernatant was subjected to filtration through 0.45 μm syringe filters. Raji cells were cultured with filtered supernatant for 3–5 days, and flow cytometry was performed to quantitate the percentage of GFP-positive Raji cells representing infectious viral particles. Raji cells cultured with fresh medium were used as blank control for gating of GFP-expressing cell population.

Quantitative PCR (qPCR) assay for the detection of EBV DNA

HONE1-EBV and AGS-EBV cells were treated with C210 and/or SAHA for 48 h. DNA was extracted from cells using a FastPure® Cell/Tissue DNA Isolation Mini Kit (Vazyme, Nanjing, China). Viral genomes were detected using Epstein–Barr Virus DNA Quantitative Fluorescence Diagnostic Kit (PCR-Fluorescence Probing) purchased from Sansure Biotech Inc (Changsha, China).

Short hairpin RNA (shRNA) transfection

Lentiviral particles containing an shRNA construct targeting Hsp90AA1 and negative control lentiviral particles obtained from Shanghai Genechem Inc (Shanghai, China) were used following the manufacturer’s protocol. Stable cell lines expressing the shRNA were selected with puromycin.

Small interfering (siRNA) transfection

Cells at approximately 50% confluence were carefully transfected with control siRNA and single siRNA targeting EBV Zta and human STAT3 and XPB purchased from Hanbio (Shanghai, China) using siRNA transfection reagent (HB-RF-1000, Hanbio, Shanghai, China). Six hours after transfection, the medium with transfection reagents was removed and the cells were then cultured in new medium. C210 or SAHA was placed on cells to induce lytic reactivation 48 h following siRNA transfection.

Immunofluorescence

Cells were treated with SAHA in the presence or absence of C210 for 48 h, fixed with paraformaldehyde, permeabilized with Triton X-100, and stained with specific primary antibodies followed by fluorescent secondary antibodies. Images were captured using a Thermo Scientific Cellomics ArrayScan VTI.

Statistical analysis

All experiments were performed in triplicate. Data were expressed as mean ± SD, and the results were processed using GraphPad Prism 9.0 software. The significance was evaluated by student’s t-test and one-way ANOVA. P value < 0.05 was considered statistically significant.

Results

C210 markedly induced EBV lytic cycle in EBV-positive cancer cells

Upon treatment with C210 in HONE1-EBV cells and AGS-EBV cells, immediate early (IE) (BZLF1), early (E) (BMRF1), and late (L) (BALF1 and BLLF1) lytic transcripts levels were all elevated in the dose-dependent manner. However, C210’s upregulation of IE and E lytic gene expression was more potent than that of L genes (Fig. 1A). SAHA at 10 µM or a combination of 20 ng/mL 12-O-tetradecanoylphorbol-1,3-acetate (TPA) with 3 mM sodium butyrate (NaB) were used as positive controls and extensively upregulated expression of viral genes in all three lytic phases. Compared to 4 µM C210, SAHA induced higher expression levels of L viral genes despite its weaker upregulation of IE and E gene expression in AGS-EBV cells. A similar gene activation profile was detected in HONE1-EBV cells. C210 reactivated EBV from latency, exerting remarkable upregulating effect on IE and E viral gene expression but less effect on L viral gene expression. These results suggest that C210 may somehow inhibit L gene expression, which is discussed separately in this paper. Additionally, both cell lines were either treated with various concentrations of C210 for 24 h, or treated with 2 µM of C210 [approximately half maximal inhibitory concentration (IC50) as determined by MTT (Figure S1A)] for different durations. We observed that IE protein Zta and E protein EaD were induced by C210 in the dose- and time-dependent manner (Fig. 1B). GFP expression represents the activation level of EBV lytic cycle in HONE1-EBV and AGS-EBV cell lines16. A slight level of spontaneous lytic reactivation represented by baseline GFP expression was seen in both cell lines. C210 treatment resulted in a remarkable level of GFP induction comparable to SAHA treatment (Fig. 1C).

Fig. 1
figure 1

C210 activated EBV lytic cycle in NPC and GC cells. (A) Expression of EBV lytic RNA adjusted to GAPDH was determined by quantitative reverse transcription PCR. Cells treated with C210 for 24 h. Cells treated with SAHA or T/S for 24 h were induced as positive controls (B) Western blot analysis showed the levels of Zta and EaD proteins after treatment with C210 for the indicated time duration. Protein extracted from cells treated with T/S served as a positive control. Cellular β-actin served as loading control. (C) Cells were treated with C210 and SAHA for 24 h, and GFP expression was imaged by fluorescence microscopy. *P < 0.05, **P < 0.01, ***P < 0.001 versus the control.

Full size image

C210 activated EBV lytic cycle by inhibiting Hsp90 function and upregulating XBP1s

As C210 is a new type of Hsp90 inhibitor, in order to investigate whether C210’s EBV-reactivating capacity at least partly arises from the inhibition of Hsp90 function, we knocked down Hsp90 using Hsp90AA1 shRNA and observed whether the downregulation of Hsp90 function induces EBV lytic replication. Knockdown of Hsp90AA1 gene reduced Hsp90 RNA (Figure S2) and protein (Fig. 2B), consequently upregulated Zta RNA, and diminished C210-driven induction of Zta RNA (Fig. 2A). Paralleled results were seen on protein level (Fig. 2B). These results indicated that EBV lytic cycle could be induced by suppressing Hsp90 function through gene knockdown, and the action of C210 on lytic induction partially depended on sufficient Hsp90 availability.

Fig. 2
figure 2

Knockdown of Hsp90AA1 diminished C210-induced EBV lytic activation. Zta RNA levels were measured by qPCR (A) and protein levels by Western blot experiments (B) in HONE1-EBV and AGS-EBV cells transfected with Hsp90AA1 shRNA in the presence of absence of C210. Cellular β-actin served as loading control in Western blot assays. *P < 0.05, **P < 0.01, ***P < 0.001.

Full size image

Previous literature has reported that UPR leads to EBV lytic induction34 and curcumin activates IRE1α-XBP1 signaling pathway, a branch of the UPR35. Given that C210 is also a derivative of curcumin and may play a similar role in activating UPR, here we investigated whether this pathway involved in C210 induction of EBV lytic cycle. We found that C210 induced Zta RNA in the context of the IRE1α-XBP1 pathway activation. C210 caused phosphorylation of IRE1α and splicing of XBP136 (Fig. 3A). And the upregulation of X box binding protein 1 splicing (XBP1s), the spliced form of XBP1, occurred prior to that of Zta protein in both cell lines. Next, both cell lines were pretreated with various doses of toyocamycin, an XBP1 splicing inhibitor, for 4 h, and then followed by C210 treatment for 24 h. Toyocamycin reduced XBP1s RNA levels and diminished C210-induced Zta RNA in the dose-dependent manner (Fig. 3B). These findings revealed that C210 activated the EBV lytic cycle through activating XBP1s, which was reversed by toyocamycin.

Fig. 3
figure 3

C210 induced EBV lytic cycle via activation of the IRE1α-XBP1 pathway of UPR. (A) The levels of p-IRE1α, XBP1s and Zta proteins in cells treated with C210 at the indicated times were detected by Western blot analysis. β-actin served as loading control. (B) Quantification of expression levels of Zta and XPB1s RNA in cells pretreated with toyocamycin for 4 h, followed by 24-h C210 treatment.

Full size image

It is noteworthy that neither Hsp90 knockdown nor inhibition of XBP1s completely abolished C210-induced Zta expression, indicating that C210 may also directly or indirectly activate EBV lytic cycle through other mechanisms. As both curcumin and SAHA affect histone acetylation that can also lead to EBV reactivation5,37, we examined the effect of C210 on histone modifications. Cells were treated with C210 or SAHA for 48 h and the expression of acetylated histone was detected by Western blot analysis. The results showed that C210 inhibited histone acetylation, whereas SAHA, as a well-established HDACi, promoted it (Figure S3). In this context, both drugs effectively induced the expression of the Zta protein, suggesting that histone modification is not the mechanism by which C210 induced EBV lytic cycle.

C210 downregulated STAT3 and enhanced the EBV-reactivating capacity of SAHA

Evidence has been shown that STAT3 restrains EBV activation in B lymphocytes. Our experiment yielded a similar result in HONE1-EBV and AGS-EBV cells. Overexpression of STAT3 protein significantly repressed SAHA induction of Zta RNA (Figure S4 A, B). While knockdown of STAT3 using siRNA resulted in a nearly two-fold increase in Zta RNA and enhanced SAHA-induced Zta RNA (Figure S4 C, D).

As STAT3 is a known client protein of Hsp90, we next examined whether C210 could reduce the protein level of STAT3 and thus improve the lytic-inducing efficiency of SAHA. Expression levels of STAT3 RNA after C210 treatment was measured by qPCR and STAT3 protein by Western blot analysis. We showed that C210 downregulated the level of STAT3 protein (Fig. 4B) with little effect on the transcription (Fig. 4A). To explore the mechanism of C210-induced degradation of STAT3, we first examined whether the stability of STAT3 could be affected by C210 treatment. STAT3 protein levels were measured at various times in HONE1-EBV and AGS-EBV cells treated with cycloheximide (CHX), a protein synthesis inhibitor, with or without C210. We found that the degradation of STAT3 was accelerated by C210 when CHX restricted new protein synthesis (Fig. 4C). These results suggested C210 enhanced STAT3 degradation rather than inhibiting protein synthesis. Hsp90 client proteins are known to be degraded by the proteasome or lysosome. The inhibitors of proteasome or lysosome were used to determine through which way C210 mediated STAT3 degradation. The result showed that C210-induced degradation of STAT3 was blocked by the proteasome inhibitor MG132 (Fig. 4D). While the lysosome inhibitor NH4CL did not affect the degradation in either cell line (Fig. 4E). Using immunoprecipitation (IP) experiments, we investigated the molecular interactions between Hsp90 and STAT3 as well as the effect of C210 on the molecular chaperone functions of Hsp90 in both cell lines. The results of IP analysis indicated that C210 inhibited STAT3 from binding to Hsp90 and induced its proteasomal degradation (Fig. 4F). These data indicated that C210 could degrade STAT3 by disrupting the molecular chaperone function of Hsp90.

Fig. 4
figure 4

C210 degraded STAT3 and increased the efficiency of SAHA in EBV lytic induction. (A) STAT3 RNA levels were measured by qPCR in cells treated with C210 for 48 h. (B) STAT3 protein levels in cells were measured by Western blot analysis; (C) STAT3 was analyzed by Western blot in cells treated with cycloheximide in the absence or presence of C210. (D,E) The expression levels of STAT3 proteins were measured by Western blot in cells pretreated with MG132 or NH4Cl for 3 h and followed by C210 for 24 h. (F) Interaction of Hsp90α with STAT3 was evaluated by immunoprecipitation in cells treated with C210 for 12 h. EBV reactivation induced by SAHA, C210, and a combination of the 2 agents for 48 h was indicated by expression of Zta protein on Western blot (G) and quantification of Zta mRNA levels by qCR (H). S, SAHA. C, C210. S + C, SAHA + C210. ns, not significant. ***P < 0.001.

Full size image

Next, we examined whether C210 increased the efficiency of SAHA in EBV lytic induction and enhanced anticancer cytotoxicity. Both cell lines were treated with 10 µM SAHA and 2 µM C210, either separately or simultaneously for 48 h. Compared to treatment with either SAHA or C210 alone, a stronger lytic reactivation was induced by the combination of two drugs, as indicated by the increased level of Zta protein (Fig. 4G) and RNA (Fig. 4H). These results suggested that enhanced SAHA induction of EBV lytic cycle may result from the degradation of STAT3 by C210. Additionally, MTT assay showed that 1 µM C210 significantly reduced the IC50 of SAHA from 24.94 µM to 7.103 µM in HONE1-EBV cells and from 26.04 µM to 14.66 µM in AGS-EBV cells (Figure S5A). The HSA synergy scores were calculated using the online SynergyFinder software, and the results showed that the HSA synergy scores was 13.019 in HONE1-EBV cells and 17.969 in AGS-EBV cells (Figure S5B). To explore whether the cytotoxicity of C210 arose from its EBV-reactivating capability, Zta gene expression was knocked down using siRNA to abrogate EBV lytic induction. We found C210 treatment resulted in additional killing of cells transfected with control siRNA in comparison with those with Zta siRNA (Figure S1), which suggested that EBV-reactivating ability contributed to the cytotoxicity of C210 in EBV-positive epithelial cancer cells. These data taken together indicated that C210 at a low dose synergistically increased the cytotoxicity of SAHA against HONE1-EBV and AGS-EBV cells, and the enhanced EBV lytic induction might play some role in this process in addition to the inherent cytotoxic capabilities of the two drugs.

C210 inhibited production of infectious virions and expression of SM-dependent late viral genes

To examine the effect of C210 on EBV virion production, we conducted an EBV infection assay that utilized GFP-expressing virus produced by lytic induction to infect the Raji cells, a Burkitt lymphoma cell line34. As can be seen in Fig. 5A, detection of infectious virions in the supernatant from C210 and/ or SAHA-treated NPC or GC cells could be achieved by the expression levels of GFP in virion-infected Raji cells. A small amount of infectious virus was produced in the supernatant from untreated cells, which might be explained by the baseline spontaneous lytic reactivation in both cell lines (Fig. 1C). Even though C210 induced EBV lytic cycle in HONE1-EBV and AGS-EBV cells, the supernatant from both cell lines treated with C210 alone led to a lower Raji cells infection rate than that from untreated cells did (Fig. 5A,B). Contrary to C210 treatment, SAHA treatment resulted in remarkable production of infectious virions. C210 inhibited virion production from SAHA-treated cells in the dose-dependent manner (Fig. 5A,B). These data suggested that C210 inhibited EBV infectious virions produced by both spontaneous and SAHA-induced EBV reactivation.

Fig. 5
figure 5

C210 blocked infectious virion production, EBV DNA replication and SM-dependent viral gene expression. HONE-EBV and AGS-EBV cells were treated with C210 and/or SAHA for 5 days and Raji cells were exposed to the cell-free supernatants for 2 to 3 days. GFP-expressing Raji cells were imaged (A) and quantitated by flow cytometry (B). HONE-EBV and AGS-EBV cells were treated with C210 and/or SAHA for 48 h. DNA or RNA was extracted form cells. EBV DNA was detected by quantitative PCR (C) and late viral RNAs were measured by qRT-PCR (D). CTL, control; C4, 4 μM C210; S + C1, SAHA + 1 μM C210; S + C2, SAHA + 2 μM C210; S + C4, SAHA + 4 μM C210. ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001.

Full size image

It is known that the production of EBV virions takes place after the replication of EBV genomes. Based on the fact that C210 efficiently initiated the EBV lytic cycle but did not lead to production of infectious virus particles, we hypothesized that C210 might induce an incomplete EBV DNA replication process in EBV-associated tumor cells. Therefore, we investigated the effect of C210 on EBV DNA replication by using an Epstein–Barr virus DNA diagnostic kit. DNA was extracted from HONE-EBV and AGS-EBV cells treated with C210 and/or SAHA and detected by qPCR. We found that SAHA treatment resulted in strong induction of EBV DNA replication in HONE1-EBV cells (15.50-fold) and AGS-EBV cells (15.39-fold) compared to the control group (Fig. 5C). Although C210 induced EBV lytic induction in high proportion of cells (Fig. 1C), only a mild increase in viral DNA copy number was shown in HONE1-EBV cells (2.69-fold) and AGS-EBV cells (2.10-fold). Moreover, C210 caused suppression of SAHA-induced DNA replication and the suppressing-effect was dose-dependent (Fig. 5C). These results suggested that C210 might somehow suppress EBV DNA replication after it initiated the EBV lytic cycle.

EBV capsid, tegument and glycoproteins encoded by late viral genes are essential for virion production and infectivity. As mentioned before, C210 led to a much milder increase in late viral RNA than SAHA did (Fig. 1A). Here we further studied whether C210 inhibited SAHA-induced late lytic gene expression. SM is a regulatory viral protein expressed in the early phase and preferentially enhances the expression of several late genes, including BcLF1 and BLLF1, which are indispensable for viral infectivity38,39. EBV late lytic genes were categorized into SM-dependent and SM-independent ones. SAHA remarkably upregulated expression of both types of genes. While the effect of C210 on late lytic genes was gene-specific, inhibiting SAHA-induced SM-dependent RNA [BcLF1 and BLLF1 (gp350)], with no obvious effect on SM-independent RNA [BALF4 (gp110) and BFRF3] (Fig. 5D). Immunofluorescence analysis showed that SAHA upregulated the expression of both SM-dependent protein gp110 and SM-independent protein gp350. In the presence of C210, gp350 protein (SM-dependent) expression was inhibited, while gp110 (SM-independent) expression was unaffected (Figure S6), which was consistent with the effect of C210 on corresponding RNAs (Fig. 5D).

C210 degraded XPB which is essential for EBV virion production and SM-dependent gene expression

A previous study showed that the xeroderma pigmentosum group B-complementing protein (XPB) protein is essential for SM-dependent gene expression and EBV virions production40. A similar result was observed in our study. SAHA-induced production of EBV infectious virus particles significantly decreased in HONE1-EBV and AGS-EBV cells depleted of XPB by siRNA inhibition (Fig. 6A). Meanwhile, the expression levels of SM-dependent genes (BcLF1 and BLLF1) decreased, while that of SM-independent genes (BALF4 and BFRF3) were not significantly affected (Fig. 6B).

Fig. 6
figure 6

Depletion of XPB abrogated production of EBV infectious virions and expression of late genes. (A) Virion production was examined by infecting Raji cells with induced cell supernatants from cells transfected with XPB siRNA or negative control siRNA. SAHA was added 48 h following XPB siRNA transfection and the cells were incubated for another 5 days before Raji infection assay. (B) Quantification PCR showed the expression levels of EBV late viral RNA isolated from cells in the presence or absence of XPB siRNA. SAHA was placed 48 h after XPB siRNA transfection and the cells were incubated for another 48 h before RNA isolation. ns, not significant, ***P < 0.001.

Full size image

The effect of XPB knockdown paralleled the effect of C210 (Fig. 5), which suggested that depletion of XPB might be the mechanism by which C210 inhibited EBV infectious virion production and SM-dependent late gene expression. The effect of C210 on XPB was thus evaluated. C210 did not obviously affect XPB RNA (Fig. 7A), but reduced the level of XPB protein (Fig. 7B) in both HONE1-EBV and AGS-EBV cell lines. C210 treatment caused a faster degradation of XPB upon new protein synthesis inhibition by CHX (Fig. 7C). The degradation of XPB was completely blocked by treatment with the proteasome inhibitor MG132 (Fig. 7D) but not affected by the lysosome inhibitor NH4CL (Fig. 7E). IP analysis. demonstrated that XPB could bind with Hsp90 to form a complex and C210 reduced the abundance of XPB in the Hsp90-XPB complex (Fig. 7F). These data indicated that XPB might be a client protein of Hsp90. C210 bound competitively to Hsp90, disassociated XPB from Hsp90, and caused the free XPB to be degraded by the proteasome. The hypothesis was verified that C210 reduced the level XPB protein, selectively inhibiting SM-dependent late gene expression and thus blocking infectious virion production.

Fig. 7
figure 7

C210 degraded XPB. (A) XPB RNA levels were measured by qPCR. XPB protein levels were measured by Western blot analysis in cells treated with increasing concentrations of C210 alone for 48 h (B) or cycloheximide in the absence or presence of C210 at different time points (C). (D,E) Western blot results showing the expression levels of XPB proteins in HONE1-EBV and AGS-EBV cells treated with MG132 or NH4Cl for 3 h prior to 24-h C210 treatment. (F) Interaction of Hsp90α with XPB was evaluated by immunoprecipitation in HONE1-EBV cells treated with C210 for 12 h. ns, not significant.

Full size image

Discussion

EBV lytic induction therapy functions through viral reactivation to facilitate tumor cell lysis and selective elimination of EBV-infected cancer cells, which has shown preliminary results for advanced refractory nasopharyngeal carcinoma in clinical trials41. The relatively low reactivating-efficiencies and the safety issues of promoting viral infection have hindered the translation of this virus-targeted therapeutic approach to clinical settings. In order to optimize the outcomes of this therapy, alternative EBV-reactivating agents or regimens featuring high inducing-efficiency without infectious virion production are needed.

Over the past years, a variety of mechanisms have been uncovered in inducing EBV lytic cycle, including modulation of signaling pathways of the host, cellular stresses, epigenetic regulation, and modulation of host and viral micro RNAs4,5,8,42. Among these mechanisms, XBP1s, an active transcription factor43, binds to and transactivates the ZII region of EBV BZLF1 promotor, Zp, thus initiating the lytic switch44,45. Curcumin and curcuminoids have been reported to induce EBV reactivation, but the specific mechanisms underlying the lytic induction remains obscure16. Our findings confirm and extend the previous report that XBP1s upregulation leads to EBV reactivation, which is of great importance for C210 effects. Furthermore, while histone deacetylase inhibitors (HDACi) typically promote histone acetylation to reactivate silenced EBV immediate early genes, our findings suggest that C210 functions differently from established HDACis like SAHA. Previous studies on curcumin’s role in histone modifications have been inconsistent, with some reporting HDACi activity46 and others indicating inhibition of histone acetyltransferases (HAT)37,47. Our results demonstrate that C210 acts specifically as a HAT inhibitor, which further supports that histone modification is not the primary mechanism driving EBV lytic reactivation by C210.

We also found that knockdown of Hsp90 induced the expression of EBV lytic RNA and protein and diminished C210-mediated EBV lytic activation, which indicated that C210 induced the latent-lytic switch in part by inhibiting Hsp90 function. The mechanism by which Hsp90 activates EBV lytic cycle has not been explored before and our understanding of this phenomenon is also limited. A possible explanation is that C210 reduces the expression level of an Hsp90 client protein, STAT3 which has been identified as a negative regulator of EBV lytic induction29,30. However, we observed that depletion of STAT3 alone merely induced a two-fold increase of lytic RNA (Figure S4). Hence, there might be other undefined molecular targets of Hsp90 in EBV reactivation, which needs further investigations. This question aside, the enhancement of SAHA-induced EBV lytic gene expression by either C210 or knockdown of STAT3 is clear. Hence, degradation of STAT3 may possibly explain this synergistic effect of C210.

Due to the concern of viral dissemination aroused by EBV lytic activation, conventional lytic-inducing compounds have been used in combination with antiherpetic drugs such as acyclovir and ganciclovir that might cause drug resistance and significant myelosuppressive or nephrotoxic side effects48. C210 is of special value as an EBV lytic activator because it initiates EBV lytic cycle but no infectious virus particles are produced. Not only that, the fact that C210 inhibited both spontaneous and SAHA-induced infectious virion production further enhances the attractiveness of this activator because it exhibits additional potential as a novel antiviral drug in EBV lytic induction therapy. Curcumin has been demonstrated to have potent antiviral capacity against a wide spectrum of viruses, including but not exclusive to human immunodeficiency virus (HIV), influenza viruses, herpes simplex virus (HSV), hepatitis B virus (HBV) and human papillomaviruses (HPV)49,50,51,52,53. The present study here adds promising results to the current body of data on this topic and may promote a potential extension to the antiviral spectrum of curcumin and its derivatives.

An interesting and rather unexpected finding in this research was that C210 may target XPB through disruption of Hsp90 function to inhibit the expression of SM-dependent late viral genes and thus the production of virions. XPB is a subunit of Transcription Factor II H (TFIIH) complex and SM is a regulatory viral protein indispensable for EBV DNA replication and the expression of a specific subgroup of late lytic transcripts39,56. Among these late transcripts, BcLF1 encoding the major viral capsid antigen (VCA) and BLLF1 encoding gp350 are required for virion formation and infectivity. The importance of XPB in expression of SM-responsive late viral genes and production of virions has been detailed in other studies40,56. Briefly, SM-dependent EBV lytic promoters require XPB activity to proceed. SM interacts with XPB and recruits it to SM target promoters, specifically facilitating gene transcription that are important in virion assembly and infectivity. Our study highlighted the impact of XPB on this biological process. Therefore, we propose that C210 depletes XPB, thereby specifically suppressing SM-mediated gene transcription and reducing infectious virus production. The relationship between XPB and Hsp90 has not been fully investigated. Hsp90, as a molecular chaperone, was reported to associate with approximately 7% of cellular transcription factors57. Gary Flom et. al performed a genetic screen on Saccharomyces cerevisiae, finding that Ssl2, the yeast homologue of XPB, and Hsp90 coexisted in the same protein complexes in vivo58. Hsp90 inhibitors have exhibited great potential for antiviral therapy and they were previously shown to decrease the levels of EBV proteins, including EBNA1, LMP1 and PK20,22,59. Here we showed that C210, an Hsp90 inhibitor, also degraded intracellular STAT3 and XPB, which have been proved to be therapeutic targets against EBV. Moreover, Hsp90 inhibitors have also been shown to prevent the virus-encoded DNA polymerase from entering the nucleus and thereafter abrogate the DNA replication of EBV and HSV-160,61.

Despite the fact that a growing number of EBV lytic inducers have been identified in screens, only the combination of lytic-inducing drugs, gemcitabine and valproic acid, along with ganciclovir have been tested in clinical setting and shown a promising result41. Lack of transition to the clinical setting necessitates finding new effective EBV-reactivating candidates. Desired effect on EBV lytic induction was achieved by C210 because the agent causes positive signals in the immediate early phase that is initiated by the BZLF1 gene62, but exerts negative effect on subsequent events, that is, EBV DNA replication, expression of a subgroup of late viral genes, and production of infectious virions. Interestingly, the Hsp90 inhibitory ability of C210 may involve in both the positive and negative modulation. Upon EBV reactivation by C210 treatment, decreased or low levels of full EBV late lytic gene expression are indicative of abortive lytic replication, whereas even abortive reactivation may trigger cell death likely via expression of immediate early and early viral gene products5. Apart from being used alone, C210 also amplified the viral-reactivating and antitumor activity of SAHA, and eliminated the negative outcome (infectious virion production) of SAHA-induced lytic reactivation. These results demonstrate the potential of combining C210 and SAHA as a future development strategy for lytic induction therapy.

Curcumin and its derivative C210 are well-known Hsp90 inhibitors with similar biochemical properties. Although curcumin has been reported to activate EBV lytic replication, its effects on late lytic gene expression and the production of infectious viral particles remain undocumented. Based on this, we infer that curcumin may exert effects similar to C210 in these aspects, yet C210 has demonstrated superior potency in both Hsp90 inhibition and anti-tumor activity compared to curcumin27. Thus, while curcumin shows promise, C210 remains the more effective therapeutic option for targeting EBV-related malignancies.

Although C210 is a noteworthy compound due to its inability to produce infectious viral particles, it may still raise oncogenic concerns. Literature has shown that cellular oncogenesis can mediated by an abortive lytic phase, where early lytic gene proteins play a vital role, while late lytic proteins are almost non-essential63. EBV lytic proteins, especially IE an E ones, promote the initiation and maintenance of tumors by enhanced survival, immunomodulation and inflammation. For example, BHRF1 inhibits apoptosis by interfering with pro-apoptotic signals and promoting survival of cancer cells64. BARF1, regarded as a key EBV oncogene in NPC development, modulates immune responses65 and immortalizes human epithelial cells66. Zta enhances inflammatory cytokines like IL-8 and IL-13, contributing to an inflammatory tumor microenvironment that promotes tumorigenesis through autocrine and paracrine signaling67,68. Indeed, the carcinogenic effects of anticancer drugs are a well-documented topic. While anticancer drugs are designed to eliminate cancer cells, their mechanisms can sometimes cause unintended consequences, such as DNA damage in normal cells, raising the risk of secondary malignancies, especially with prolonged use or high doses69. Given this risk, the potential oncogenic effects of EBV lytic inducers including C210 warrant careful investigation and thorough risk–benefit analyses need to be conducted during therapeutic applications of these drugs. Future studies should aim to elucidate the specific roles of lytic proteins in cancer development and explore strategies to mitigate their adverse effects, while harnessing their therapeutic potential.

Our in vitro results indicate that C210 might be used to activate EBV lytic cycle and inhibit virion production in humans, whereas the safety and efficacy of this agent used in lytic induction therapy will next merit assessment in animal models of EBV-infected tumors.

Conclusion

In summary, C210 functions via inhibition of Hsp90 and upregulation of XBP1s to induce EBV lytic cycle. Moreover, disruption of Hsp90 function also mediates C210-driven degradation of STAT3 and XPB, through which mechanism viral reactivation is enhanced and infectious virions production is abolished. These findings provide a supplement for the study of EBV lytic activators and lay a foundation for further in vivo studies as well as potential clinical use of C210 in lytic-phase directed therapies for EBV-associated malignancies.