STAT3 is constitutively activated in chronic active Epstein-Barr virus infection and can be a therapeutic target

Chronic active Epstein-Barr virus infection (CAEBV) is a lymphoproliferative disorder characterized by the clonal proliferation of EBV-infected T or NK cells and is related to severe systemic inflammation. This study aims to investigate STAT3 to elucidate the mechanism underlying the CAEBV development. We determined that STAT3 was constitutively activated in EBV-positive T- or NK-cell lines. We also determined that STAT3 was activated in the peripheral blood mononuclear cells (PBMCs) containing EBV-infected clonally proliferating T or NK cells in six of seven patients with CAEBV. We conducted direct sequencing of the STAT3 Src homology 2 (SH2) domain, which has previously been reported to be mutated in T- or NK-cell neoplasms. No mutation was detected in the STAT3 SH2 domain in patients with CAEBV. Next, we investigated the effects of ruxolitinib, an inhibitor of both JAK1 and JAK2, which phosphorylates and activates STAT3. Ruxolitinib suppressed the phosphorylation of STAT3 in EBV-positive T- or NK-cell lines. Ruxolitinib also decreased the viable cell number of EBV-positive T- or NK-cell lines and PBMCs from patients with CAEBV. Furthermore, ruxolitinib suppressed the production of inflammatory cytokines in the cell lines and CAEBV patient-derived cells. In conclusion, constitutively activated STAT3, which promotes survival and cytokine production, could be a therapeutic target for CAEBV.


INTRODUCTION
Chronic active Epstein-Barr virus infection (CAEBV) was first reported as sustained infectious mononucleosis accompanied by fever, lymphadenopathy, and liver dysfunction in patients without known immunodeficiency [1]. Later, it was reported that EBV-infected, clonally proliferating T or NK cells had been detected in patients with the disorder [2][3][4]. Furthermore, aggressive T-or NKcell lymphoma can develop during the clinical course of CAEBV [5]. Hence, in the WHO classification, revised in 2017, CAEBV was classified into EBV-positive T-or NKcell lymphoproliferative disorder [6].
The two characteristics of CAEBV are a T-or NKcell lymphoproliferative disorder and an inflammatory disorder [7,8]. Although EBV-infected T or NK cells in CAEBV clonally proliferate, they usually do not form solid tumors [8,9]. Instead, they can be detected in the peripheral blood and can infiltrate into organs, leading to their dysfunction. Furthermore, the transformation of these cells can lead to chemotherapy-resistant T-or NK-cell lymphoma or leukemia [5,8,10]. The other aspect of CAEBV is an inflammatory disorder. The main symptoms of CAEBV are equivalent to those of chronic inflammation, such as persistent fever, lymphadenopathy, liver dysfunction with hepatosplenomegaly, vasculitis, neuritis, and dermatitis [5]. This chronic inflammation eventually could develop into the life-threatening disorder hemophagocytic lymphohistiocytosis (HLH).
Unfortunately, the mechanism underlying the CAEBV development remains unclear. At present, no effective agent exists for CAEBV, and the only curative treatment strategy has been allogeneic hematopoietic stem cell transplantation (allo-HSCT) [9,11]. However, only a limited number of patients are suitable candidates for allo-HSCT. Furthermore, the outcomes of allo-HSCT for patients with an active disease, accompanied by fever, liver dysfunction, vasculitis, and uveitis, were significantly poorer than the outcomes for patients with an inactive disease [5,9]. Hence, in order to establish effective medical treatments, it is imperative to determine the underlying mechanism of clonal proliferation of EBV-infected cells and sustenance of systemic inflammation in patients with CAEBV.
STAT3 is a transactivation factor that mediates proliferation and anti-apoptotic intracellular signaling. Constitutively activated STAT3 has been observed in numerous primary tumor cells as well as in tumor-derived cell lines obtained from patients, including those with lymphoid malignancies [12,13]. STAT3 is constitutively activated in EBV-positive neoplasms, post-transplant lymphoproliferative diseases (LPDs) [14], EBV-positive diffuse large B-cell lymphoma (DLBCL) [15], nasopharyngeal cancer (NPC) cells [16], and extranodal NK/T-cell lymphoma (ENKL) [17]. In addition, Küçük et al. reported the presence of activating mutations in the Src homology 2 (SH2) domain of STAT3 in EBV-positive γδ T-or NK-cell lines and in ENKL patient cells [18]. Interestingly, they also reported that a JAK1/2-specific inhibitor, AZD1480, inhibited the STAT3 activation as well as the proliferation of EBV-infected T-or NK-cell lines. As CAEBV is characterized by EBV-positive T-or NK-cells, we hypothesized that STAT3 was also constitutively activated in CAEBV. In addition, STAT3 induces inflammation by promoting the production of inflammatory cytokines, such as IFN-γ and TNF-ɑ, among others and by mediating the molecular signaling from their receptors [19]. This study aims to investigate STAT3 activation and its role in CAEBV using both cell lines and cells obtained from patients with CAEBV.

STAT3 is constitutively activated in EBV-positive T-or NK-cell lines
We investigated the STAT3 activation in EBVpositive T-or NK-cell (EBV-T/NK-cell) lines established from patients with EBV-positive T-or NK-cell lymphoid neoplasm. For the activation of STAT3, the phosphorylation of both tyrosine-705 and serine-727 is indispensable. At first, we conducted an immunoblotting assay to determine the phosphorylation of STAT3 ( Figure 1A). Figures 1B  and 1C show the relative intensity of the bands by the densitometry analysis. The serine-727 phosphorylation of STAT3 was detected in all cell lines under the maintenance condition ( Figures 1A and 1C). However, the phosphorylation of tyrosine-705 was detected in EBVpositive T-or NK-cells, not in Jurkat, MOLT4, and HPB-ALL cells, which are EBV-negative T-cell lines ( Figures  1A and 1B). In KHYG1 cells, an EBV-negative NK-cell line, a little phosphorylation of tyrosine-705 of STAT3 was detected ( Figures 1A and 1B). In addition, we investigated the localization of STAT3 in these cells, as activated STAT3 is phosphorylated and localized in the nucleus. Figure 1D shows that STAT3 was phosphorylated and detected in the cytoplasmic and nuclear fraction in EBV-T/NK-cell lines by western blotting. Figures 1E and 1F show the densitometry analysis. EBV-negative cell lines did not exhibit tyrosine-phosphorylated STAT3 in the nucleus under these conditions ( Figures 1D, 1E and 1F).

STAT3 is constitutively activated in EBV-positive T-or NK-cells from patients with CAEBV
We validated the results mentioned above in patientderived cells. In CAEBV, EBV-positive cells are detected in the peripheral blood. In this study, 14 patients with CAEBV (aged 18-64 years; five males, nine females; CD4 type: n = 4; CD8 type: n = 4; CD56 type: n = 3; CD4 and CD56 double infection: n = 2; and CD4 and CD8 double infection: n = 1) were investigated. Table 1 presents the clinical findings, phenotype, and EBV DNA load of infected cells. The clonal proliferation of infected cells was detected in the peripheral blood mononuclear cells (PBMCs) of all patients. The EBV DNA load of the patient-derived PMBCs was 1.7×10 3 -2.6×10 5 (mean: 9.2 × 10 4 ) copies/μg DNA. We studied the STAT3 activation in the PBMCs obtained from five patients; the cells were isolated from the peripheral blood, immediately lysed, and used for western blotting (Figure 2A). Figures 2B  and 2C show the relative intensity of the bands by the densitometry analysis. The lysate of SNT8 cells was a positive control. The phosphorylation of serine-727 was detected in the samples from patients as well as those from healthy donors. On the other hand, tyrosine-705 of STAT3 was significantly phosphorylated in four of five patient cells. The non-detection of the tyrosine phosphorylation in Case 9 could present the negative or weak activation of STAT3, or could be attributed to the low rate of EBVpositive cells in the PBMCs (Table 1), because STAT3 was constitutively expressed in PBMCs. Taken together, the phosphorylation of both tyrosine-705 and serine-727 was detected in patients' samples. In addition, we and PS-STAT3 (C) bands of (A) were determined as ratio to total STAT3 by densitometry. MOLT4 was determined as a control. (D) Western blotting for STAT3 localization in EBV-T/NK-cell lines. Tyrosine-PY-STAT3 is localized in the nucleus in EBV-T/ NK-cell lines but not in EBV-negative T-or NK-cell lines. Hsp90 and YY1 are proteins that were localized to the cytoplasm and nucleus, respectively. (E and F) the relative intensities of PY-STAT3 bands (D) of cytoplasm (E) and nucleus (F). The intensites were determined as ratio to Hsp90 (E) and YY1 (F), respectively by densitometry. MOLT4 was determined as a control. www.oncotarget.com investigated the expression of STAT3-inducible genes by the microarray analysis. The fractions containing EBVinfected cells were isolated from the PBMCs of patients, Cases 1, 7, and 11, whose EBV-positive cells were CD4-, CD8-, and CD56-positive cells, respectively. For the isolation, we used antibody-conjugated magnetic beads against surface markers on infected cells. In addition, the RNA was extracted from cells and used for the microarray analysis. Figure 2D shows that the expression of STAT3inducible genes, including IFN-γ, was generally enhanced in the patient-derived T-or NK-cell fractions that included EBV-positive cells. Furthermore, we examined the activation and localization of STAT3 in EBV-positive Tor NK-cells from the patients. In Cases 3 and 11, EBVinfected cells were CD4-and CD56-positive, respectively (Table 1). We isolated CD4-or CD56-positive cells from the PBMCs of patients and healthy donors using antibody-conjugated magnetic beads. Figure 2E shows the flow cytometry for the isolated cells, confirming that a majority of the isolated cells were positive for CD4 or CD56. Figure 2F shows the localization of tyrosinephosphorylated STAT3 (PY-STAT3) in isolated CD4-or CD56-positive cells. PY-STAT3 was detected not only in the cytoplasm but also in the nucleus of patient-derived CD4-or CD56-positive cells, whereas it was not detected in CD4-or CD56-positive cells from healthy donors. EBV infection was determined by the LMP1 expression in patient-derived CD4-or CD56-positive cells. These results indicated that STAT3 was constitutively activated in PBMCs fractions containing EBV-infected T or NK cells in CAEBV. In Case 3 and 11, STAT3 was constitutively activated in EBV-infected T or NK cells. At present, however, it is not clear whether this activation is specific to EBV-infected cells, because we could not examine infected and un-infected cells separately.

Mutation analysis of the STAT3 SH2 domain in EBV-positive T-or NK-cells
The SH2 domain of STAT3 frequently exhibits gain-of-function mutations in large granular lymphocytic leukemia. Mutations in this domain have also been identified in γδ T cells from lymphoid neoplasms including ENKL [18,20]. Hence, we investigated the mutation status of the SH2 domain in EBV-positive T-or NK-cells from patients with CAEBV. Three of six EBVpositive T-or NK-cell lines: SNT8; SNT15; and SNK6, presented a mutation in the SH2 domain of STAT3 (D661Y in SNT8 and SNK6, and Y640F in SNT15; Table 2; Supplementary Figure 1). These mutations were identical to those previously reported [18]. However, no mutation was detected in the PBMCs analyzed, which included EBV-infected and clonally proliferating T or NK cells from 14 patients with CAEBV.

The effects of JAK1/2 inhibitors on the phosphorylation of STAT3 in EBV-positive T-or NK-cell lines
The results described above suggest that the constitutive activation of STAT3 might be induced by an upstream intracellular signaling molecule. JAKs are kinases that phosphorylate and activate STAT3. Hence, we investigated the effects of JAK inhibitors on the phosphorylation of STAT3 in EBV-positive T-or NK-cell lines. Interestingly, the tyrosine-705 phosphorylation of STAT3 was inhibited by ruxolitinib (a JAK1/2 inhibitor), tofacitinib (a JAK3 inhibitor), and JAK inhibitor 1 (a JAK1/2/3 inhibitor) in a dose-dependent manner even in cell lines harboring the mutation (Figure 3). The phosphorylation of serine-727 of STAT3 was not clearly suppressed after the treatment. Dimethyl sulfoxide (DMSO), a solvent for inhibitors, did not exhibit significant effects on the phosphorylation (Supplementary Figure 2). Thus, even in cells with the SH2 domain mutation, the constitutive activation of STAT3 could be inhibited in EBV-positive T-or NK-cell lines by the inhibition of JAKs.

The biological effects of ruxolitinib on EBVpositive T-or NK-cell lines and CAEBV patientderived cells
From the three inhibitors, we selected ruxolitinib, the effective concentration of which was identical to that of its clinical use. We investigated the effects of ruxolitinib on the proliferation and survival of EBVpositive T-or NK-cells. The XTT assay demonstrated that ruxolitinib suppressed the viable cell number of EBV-positive T-or NK-cell lines in a dose-dependent manner ( Figure 4A). In addition, ruxolitinib suppressed a viable cell number of CAEBV patient-derived PBMCs in a dose-dependent manner ( Figure 4B). Ruxolitinib also suppressed the number of healthy donor-derived PBMCs. However, under the concentration of 0.25 μM, which can be actually achieved in patients, the number of patientderived PBMCs was decreased whereas that of healthy donor-derived PBMCs was not. As shown in Figure 4C, ruxolitinib induced cellular apoptosis in SNT16, although it was not remarkable in SNT8, SNT15, and SNK6 cells. Finally, we investigated the effects on cytokine production. Figures 5A and 5B show that ruxolitinib suppressed the mRNA production of inflammatory cytokines, IFN-γ and TNF-ɑ in EBV-positive T-or NKcell lines. Furthermore, ruxolitinib suppressed the mRNA production of inflammatory cytokines IFN-γ ( Figure 5C) and TNF-ɑ ( Figure 5D) in the CAEBV patient-derived PBMCs. The mRNA of interleukin-6 (IL-6) was not detected in these cells (data not shown). The viability of each cells was approximately 80%, and 60-80%, in the cell lines and patients' cells, respectively (Supplementary Figure 3).

DISCUSSION
Our results indicated that STAT3 was constitutively activated in EBV-positive T-or NK-cells from patients with CAEBV. This study did not detect activating mutations of STAT3 in these cells. Furthermore, the tyrosine phosphorylation of STAT3, as well as the survival, was suppressed by ruxolitinib even in cell lines harboring the activating mutation of STAT3. Although mutations detected in the SH2 domain upregulated the dimerization of STAT3 [18], our results indicated that the activation of STAT3 depended on the tyrosine phosphorylation.
The underlying mechanism for the phosphorylation of STAT3 in CAEBV could be attributed to several factors. First, the activation of JAK kinases which directly phosphorylates and activates STAT3, should be examined. Previously, gain-of-function mutations have been reported in JAK1 or JAK3 in some neoplasms, leukemia, breast cancer, lung cancer, hepatocellular carcinoma [21], and NK-lymphoma cells [22][23][24]. Interestingly, an association between the JAK1 expression and the STAT3 activation has been reported in T-cell lymphomas [25]. In addition, Src family kinases can activate STAT3. Reportedly, LMP1 activates Src kinases in EBV-infected B-cells [26], and this activity should be investigated in EBV-positive T-or NK-cells in CAEBV.
Several studies have suggested that STAT3 might be activated downstream of LMP1 through the activation of NF-κB. In EBV-infected B cells or epithelial cells, NF-κB is activated by C-terminal-activating region 1 (CTAR1) of LMP1 [27,28]. In addition, we recently determined that NF-κB was constitutively activated through LMP1, promoting survival in EBV-positive  T-or NK-cells obtained from patients with CAEBV [29]. Chen et al. reported that NF-κB activates STAT3 through IL-6 production in EBV-infected epithelial cells [30]. They mentioned that IL-6 associates with IL6R/ gp130 expressed on cells and activates STAT3 through the JAK activation in cells by autocrine mechanisms. In addition, IL-6 is elevated in the serum of patients with CAEBV [19]. Although the mRNA of IL-6 was not detected in EBV-positive T-or NK-cell lines in this study, IL-6 can be produced by other inflammatory cells such as macrophages in CAEBV. Reportedly, activated NF-κB induces the expression of a receptor tyrosine kinase epidermal growth factor receptor (EGFR), leading to the STAT3 activation by the phosphorylation on Y-705 in NPC cells. According to the study by Nosbaum et al., EGFR can be inducibly expressed on T or NK cells [31]. Thus, EGFR could also mediate the STAT3 activation in EBV-positive T-or NK-cells. The expression and the activation of EGFR in CAEBV, and the effects of ruxolitinib on EGFR warrant further investigation. Overall, NF-κB might induce STAT3 activation through IL-6 or EGFR in EBV-infected T or NK cells. Furthermore, other molecular pathways can contribute to the activation of STAT3 in EBV-infected T or NK cells. Erk is reportedly activated downstream of LMP1 and induces the serine phosphorylation of STAT3 [32].
Our recent study determined that Erk was constitutively activated in EBV-infected T or NK cells (our unpublished data). Additional studies in this regard are ongoing in our laboratory.
In this study, we determined that ruxolitinib suppresses the phosphorylation of STAT3, decreasing the survival and cytokine production in EBV-positive cells of patients with CAEBV. However, ruxolitinib-induced apoptosis was not clear in SNT8, SNT15 and SNK6 cells ( Figure 4C). Additional mechanisms, such as necrosis or other types of programmed cell death, might be induced by the treatment. It is indispensable to confirm the effects using PBMCs of patients with CAEBV.
The two features of CAEBV are a neoplastic disease and an inflammatory disease. Since the concentration of inflammatory cytokines in the serum of patients with CAEBV is closely related to the disease status [19], these results suggest that ruxolitinib is effective for CAEBV treatment by suppressing both neoplastic cells and inflammation. Interestingly, two studies recently reported that JAK1/2 inhibition was effective in treating a mouse model of HLH [33,34]. HLH, a severe form of inflammation, occurs in 24% of patients with CAEBV and can be lethal [9]. Inhibition of the JAK/STAT pathway might inhibit the development of HLH and could be an attractive candidate for CAEBV treatment. Nonetheless, further extensive studies using clinical samples are needed to confirm the results and apply these findings clinically.

Diagnosis of CAEBV
CAEBV was diagnosed on the basis of following criteria: (1) the presence of characteristic clinical findings such as persistent systemic inflammation, fever, liver dysfunction, lymphadenitis, vasculitis, uveitis, severe mosquito bite allergy (sMBA), and hydroa vacciniforme-like eruption (HV); (2) a high EBV load detected in the PBMCs by quantitative PCR (> 10 2.5 copies/μg of EBV DNA); and (3) EBV infection in T or NK cells [5,36]. The clonality of the infected cells was examined by Southern blotting.

Detection of the clonality
The clonal proliferation of EBV-infected cells was detected by Southern blotting for EBV-terminal repeat. [39].

Isolation of the EBV-infected cell fraction from patients with CAEBV
For assays, we isolated the fraction containing EBVinfected cells from the PBMCs using antibody-conjugated magnetic beads against a surface marker on infected cells (IMag Anti-Human CD4, 8, or 56 Particles-DM; BD Biosciences). Table 1 describes the marker used to isolate infected cells from each patient.

Apoptosis detection assay
Apoptosis was detected using a TACS ® Annexin V-FITC apoptosis detection kit (Trevigen, Gaithersburg, MD), according to the manufacturer's instructions.

Antibodies
For western blotting, we used the following antibodies: phospho-STAT3 (

Cell lysis for western blotting
The assay was performed as described previously [40]. For total cell lysis, cells were lysed in SIP buffer (50 mM Tris-HCl [pH 7.5], 5 mM EDTA, 100 mM NaCl, 50 mM NaF, 1 mM Na 3 VO 4 , 40 mM β-glycerophosphate, and 1% Triton X-100) after washing with phosphate buffered saline (PBS). The resulting lysate was centrifuged at 15,000 rpm for 10 min, and the supernatant was collected and subjected to western blotting.
Furthermore, proteins were separated into cytoplasmic and nuclear fractions, as described later. After washing with PBS, cells were lysed in hypotonic buffer (50 mM Tris-HCl [pH 7.5], 5 mM EDTA, 10 mM NaCl, 1 mM NaF, 1 mM Na 3 VO 4 , and 0.05% NP-40). Lysates were centrifuged at 700×g for 2 min, and supernatants were collected as cytoplasmic protein fractions. Pellets were washed three times with hypotonic buffer and lysed in SIP buffer (50 mM Tris-HCl [pH 7.5], 5 mM EDTA, 100 mM NaCl, 50 mM NaF, 1 mM Na 3 VO 4 , 40 mM β-glycerophosphate, and 1% Triton X-100). Lysates were centrifuged at 15,000 rpm for 10 min, and supernatants were used as nuclear protein fractions. Proteins were quantified using the DC protein assay kit (Bio-Rad, Hercules, CA) and equivalent protein quantities were subjected to western blotting.

Immunofluorescence staining
Cells were fixed on slides by immersion in 4% paraformaldehyde for 10 min, followed by three washes in PBS and were incubated with mouse monoclonal www.oncotarget.com anti-LMP1 and rabbit polyclonal anti-phospho-STAT3 (Tyr705) antibodies at room temperature. Then, the slides were treated at room temperature with a Cy5-conjugated AffiniPure donkey anti-mouse antibody (No.715-175-151, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) to label the anti-LMP1 antibody and a PE-conjugated goat anti-rabbit antibody (No. 4030-09; Southern Biotech, Birmingham, AL) to label the anti-phospho-STAT3 antibody. The nuclei were counterstained with ProLong Gold and DAPI (Invitrogen, Carlsbad, CA).

Microarray analysis
The total RNA was extracted from the EBV-infected cell fraction of the PBMCs obtained from patients with CAEBV using TRIzol Reagent (Thermo Fisher Scientific K.K, Waltham, MA) and purified using the SV Total RNA Isolation System (Promega, Madison, WI), according to the manufacturer's instructions. We quantified RNA samples by an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE) and confirmed the quality with the Experion System (Bio-Rad Laboratories). The cRNA was amplified, labeled using the Low Input Quick Amp Labeling Kit, and hybridized to a SurePrint G3 Human Gene Expression Microarray 8×60K v2 (Agilent Technology, Santa Clara, CA), according to the manufacturer's instructions. All hybridized microarray slides were scanned by an Agilent scanner. Furthermore, relative hybridization intensities and background hybridization values were calculated using Agilent Feature Extraction Software (9.5.1.1; Agilent Technology).
We calculated raw signal intensities and flags for each probe from hybridization intensities (gProcessedSignal) and spot information (gIsSaturated), according to the manufacturer's instructions (Agilent). In addition, raw signal intensities were normalized by a quantile algorithm using the "preprocessCore" library package in the Bioconductor software [41,42].

Flow cytometry
The PBMCs were isolated and analyzed by a FACS Calibur flow cytometer (Becton Dickinson and Company, Franklin Lakes, NJ) with FITC-conjugated mouse antihuman CD4 and PE-conjugated mouse anti-human CD56 antibodies (No.561842, No.555516; Becton Dickinson and Company).

qRT-PCR
The RNA was extracted with ISOGEN II (Nippon Gene, Tokyo, Japan) from the cell lines and PBMCs, and cDNA was generated using Transcriptor Universal cDNA Master (Roche, Basel, Switzerland). The DNase treatment was performed using shrimp DNase recombinant (Affymetrix, Santa Clara, CA). In addition, we performed qRT-PCR analyses on a Light Cycler 480 (Roche) using TaqMan ® Gene Expression Assays (Applied Biosystems). The primers for interferon-γ and TNF-ɑ were Hs00989291_m1 and Hs01113624_g1, respectively.

DNA sequencing
In this study, genomic DNA was extracted using the DNeasy Blood & Tissue Kit (QIAGEN, Hilden, Germany). PCR analyses of genomic DNA were performed in a 50 μL reaction volume containing 5× prime STAR GXL Buffer, a dNTP mixture, a forward primer (10 μM), a reverse primer (10 μM), and Prime STAR GXL DNA polymerase (TAKARA BIO, Shiga, Japan); the primers used in this study have been previously described [20]. The PCR product was run on an agarose gel, excised with a scalpel, and purified using the MinElute Gel Extraction Kit (QIAGEN, Hilden, Germany). Furthermore, sequencing was performed using Ion PGM (Life Technologies, CA).

Ethics statement
The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the ethical committee of the Tokyo Medical and Dental University (TMDU, Tokyo, Japan). We obtained written informed consent from all patients before enrollment in this study.