Production of Interferon-Inducible Protein-10 and Its Role as an Autocrine Invasion Factor in Nasal Natural Killer/T-Cell Lymphoma Cells

Nasal natural killer (NK)/T-cell lymphoma, previously known as lethal midline granuloma (1), has distinct epidemiologic, clinical, histologic, and etiologic features. Epidemiologically, this lymphoma is common in Asian countries but is quite rare in the United States and Europe (2,6). Clinically, this lymphoma is characterized by progressive necrotic lesions mainly in the nasal cavity and a poor prognosis resulting from rapid progression of the lesion and distinct organs (2, 7, 8). Histologically, the features of this lymphoma include angiocentric and polymorphous lymphoreticular infiltrates, which are called polymorphic reticulosis (9, 10). Original cells of the lymphoma are reported to be of the NK or T cell lineages, both of which express the NK cell marker CD56 (2, 5, 6, 11).

Regarding etiologic factors, we first indicated the presence of Epstein-Barr virus (EBV) DNA, EBV oncogenic proteins, and clonotypic EBV genome in this lymphoma, suggesting that EBV may play a role in lymphomagenesis (2, 4, 12). However, little is known about its genetic features because this lymphoma is a relatively rare disease and it is often difficult to obtain a sufficient amount of tissue from necrotic lesions. Moreover, if a sufficient amount of tissue is obtained, it is difficult to analyze the profile of gene expression in biopsy samples because the tissue specimen contains too many types of cells, including tumor cells, inflammatory cells, or normal epithelial cells, which impede the examination of gene expression of individual cells. Recently, Nagata et al. (11) established two kinds of EBV-positive cell lines of the NK or T cell lineages, SNK-6 and SNT-8, from primary lesions of nasal NK/T-cell lymphoma. They are very useful to compare the gene or protein expression of these cell lines with that of other lymphoma/leukemia cell lines to find the genes or proteins that are expressed specifically in nasal NK/T-cell lymphoma. We previously showed that interleukin (IL)9 and IL-10, which were produced specifically by nasal NK/T-cell lymphoma cell lines, enhanced the lymphoma cell proliferation directly or indirectly in an autocrine manner (13, 14). Because such cytokines were not produced by EBV-negative NK cell lymphoma lines, it is suggested that EBV may contribute to lymphoma cell proliferation through the production of these cytokines.

Chemokines are a superfamily of pro-inflammatory polypeptide cytokines that selectively attract and activate different cell types. Tissue injury, allergy, cardiovascular diseases, as well as malignant tumors are prerequisites for many pathophysiologic conditions. The role of chemokines in malignant tumors is complex: Although some chemokines may enhance innate or specific host immunity against tumor implantation, others may favor tumor growth and metastasis by promoting tumor cell proliferation, migration, or neovascularization in tumor tissue (15). Especially in EBV-related hematologic malignancies such as Hodgkin's lymphoma, chemokines are likely to have beneficial effects for tumor progression rather than unfavorable ones (16, 17). Moreover, some chemokines, including CCL20 (18), RANTES (19), interferon (IFN) inducible protein 10 (IP-10; ref. 20), and IL-8 (21) were reported to be controlled by proteins derived from EBV. Therefore, in nasal NK/T-cell lymphoma, such chemokines are possibly related to tumor progression as well. However, there is no report that describes comprehensive examination of chemokines in nasal NK/T-cell lymphoma.

To determine which chemokines are specifically produced by nasal NK/T-cell lymphoma, we compared chemokine profiles in the culture supernatants from nasal NK/T-cell lymphoma cell lines to those from the other cell lines. Chemokine protein array analysis revealed that IP-10 was specifically expressed in culture supernatants of the nasal NK/T-cell lymphoma cell lines. Moreover, functional analyses revealed that IP-10 stimulated invasive phenotype of the lines in an autocrine manner. Furthermore, in vivo studies showed IP-10 expression in biopsies and sera from patients with this lymphoma. These results suggest that IP-10 is a crucial factor in the pathogenesis of nasal NK/T-cell lymphoma.

Ten Japanese patients, eight men and two women, 21 to 70 years of age with a median age of 53 years, participated in the study. Nasal NK/T-cell lymphoma was diagnosed in all of them at the Department of Otolaryngology-Head and Neck Surgery, Asahikawa Medical College (Hokkaido, Japan), between 2003 and 2008. The diagnosis was carried out according to the World Health Organization classification of hematologic malignancies (22). Pertinent clinical information and follow-up data were obtained from hospital charts for all patients (Table 1). According to the Ann Arbor classification system, they were all stage I patients. The B-symptoms were observed in 3 (30) patients. The median serum lactate dehydrogenase level was 176.5 IU/L with a range of 144 to 765 IU/L (reference range 105-210 IU/L). All patients achieved complete remission after chemoradiotherapy and are healthy without any sign of relapse. The details of chemoradiotherapy were described in our previous reports (23, 24). Briefly, the MTCOP-P regimen used in two patients consists of pirarubicin hydrochloride, cyclophosphamide, vincristine sulfate, methotrexate, peplomycin sulfate, and prednisolone; the MPVIC-P regimen used in the eight remaining patients consists of ifosfamide, carboplatin, methotrexate, peplomycin, etoposide, and prednisolone. The dose of radiotherapy was 54 to 56 Gy, and a lateral opposing field was used covering the primary site. The clinical characteristics of the 10 patients are summarized in Table 1. In vivo materials used in the studies were biopsy samples and serum before and after treatment. Serum from six healthy volunteers (all men, 29-40 years of age, median 30 years) was also used. All patients signed informed consent forms for this study, which were approved by the institutional review board.

The features of the cell lines used in this study are listed in Table 2. SNK-1, SNK-6, and SNT-8 were EBV-positive cell lines established from primary lesions with nasal NK/T-cell lymphoma. The cell lines were kindly provided by Dr. Shimizu (Tokyo Medical and Dental University; ref. 11). KAI-3 originated from a patient with a severe mosquito allergy (25). EBV-negative NK cell lines NK-92 and KHYG-1 were established from patients with NK cell leukemia (26, 27). Raji was an EBV-positive B-cell line that originated from Burkitt's lymphoma (28). SNK-1, SNK-6, SNT-8, KHYG-1, and KAI-3 cells were cultured in RPMI 1640 supplemented with 10 fetal bovine serum (FBS), with 50 units/mL penicillin, 50 g/mL streptomycin (Life Technologies, Inc.), and 250 units/mL recombinant human IL-2 (Takeda Pharmaceutical Company Limited). NK-92 cells were cultured in a-MEM supplemented with 12.5 horse serum, 12.5 FBS, 50 units/mL penicillin, and 50 g/mL streptomycin and 200 units/mL recombinant human IL-2. All cell lines were incubated at 37C in an atmosphere containing 5 CO₂.

RayBio Human Chemokine Antibody Array I (RayBiotech) was used according to the instruction manual. Briefly, a membrane was incubated with 1 mL supernatant of SNK-6, SNT-8, or KHYG-1 cultures at 4C overnight. After washing, the membrane was incubated with 1 mL of primary biotin-conjugated antibody followed by 2 mL of horseradish peroxidaseconjugated streptavidin at 4C overnight. The membranes were developed by enhanced chemiluminescence and were exposed to X-ray film.

For flow cytometric analysis of surface molecules, cell lines were washed in cold PBS containing 1 albumin from bovine serum, centrifuged, and resuspended in PBS. Cells were incubated with Carboxyfluorescein-conjugated mouse anti-human CXCR3 monoclonal antibody (DAKO) for 30 min at 4C. Carboxyfluorescein-conjugated mouse IgG1 (DAKO) was used as isotype control. Samples were analyzed with FACScan (BD Biosciences).

IP-10 protein in cell culture supernatants and serum was quantified using an ELISA kit, Quantikine Human CXCL10/IP-10 (R&D Systems, Inc.). Cell lines (2.5 10⁵/mL) were cultured in 96-well round-bottomed plates, and supernatants of cell cultures were collected after 24, 48, and 72 h. Blood samples from nasal NK/T-cell lymphoma patients were obtained at pretreatment and posttreatment periods. As the healthy control, blood samples from six healthy subjects were also obtained. Serum was separated from whole blood by centrifugation and stored at 80C. The mouse anti-human IP-10 monoclonal antibody was coated onto the bottom of supplied 96-well ELISA plates. The supernatants and serum were diluted with assay diluent and added to the each well. The plates were washed with wash buffer after 2 h of incubation at room temperature. Anti-human IP-10 polyclonal antibody conjugated to horseradish peroxidase (200 L) was added to each well and incubated at room temperature for 2 h. After washing, 200 L of a substrate solution were added and the plates were incubated for 30 min in the dark. The absorbance of each well was determined at 450 nm using a microplate reader (Nalge Nunc International). For cell culture supernatants, measurements were done in triplicate, and for serum, measurements were done in duplicate. A standard curve was generated using serial dilutions of recombinant IP-10 (R&D Systems). The results correspond to mean SD.

SNK-6 and KHYG-1 cells (2 10³ or 1 10⁴ per well) in 96-well plates were cultured in 200 L 10 FBS containing RPMI 1640 with or without 50 units/mL IL-2. Each well was treated with recombinant IP-10 (10, 100, or 1,000 ng/mL; Peprotech) or anti-human IP-10 antibody (1 or 10 g/mL; R&D Systems) for 48 h. To determine the number of viable cells, we used the Cell Titer 96 Aqueous One Solution Cell Proliferation Assay (Promega). MTS (20 L) was added to each well, and then incubated for 4 h at 37C under 5 CO₂. The absorbance at 490 nm was measured with an ELISA plate reader. Results were expressed as a percentage of untreated controls. Measurements were done in duplicate and experiments were repeated at least three times. The results correspond to mean SD.

Tumor cell invasion was assayed in 24-well BioCoat Matrigel Invasion Chambers (8 m; Becton Dickinson) according to the manufacturer's protocol. Briefly, after hydration of rehydrated Matrigel inserts with DMEM for 2 h, SNK-1, SNK-6, or KHYG-1 cells (2.5 10⁴/well) in serum-free RPMI 1640 containing recombinant human IP-10 (1 or 10 ng/mL; Peprotech), anti-humanIP-10 monoclonal antibody (1, 10 g/mL; R&D Systems), or anti-humanCXCR3 monoclonal antibody (1,10 g/mL; R&D Systems) were plated in the top chamber. RPMI 1640 with 10 FBS was placed into a bottom chamber. After incubation for 22 h, the Matrigel membranes were removed, fixed with methanol, and stained with a Giemsa-type stain (Diff-Quik, Sysmex). Migrated cells in the membranes were counted under a microscope in five arbitrary fields. As a next experiment, supernatant was obtained from culture of SNK-6 or KHYG-1 (2.5 10⁴/mL) in RPMI 1640 with 10 FBS for 48 h, and was plated in the top chamber instead of the medium with IP-10. Assays were done in triplicates independently. Results were shown as the number of cells through the Matrigel matrix.

EBV-encoded small nuclear early region type 1 (EBER1) expression in tumor cells was assessed by in situ hybridization on formalin-fixed, paraffin-embedded tissue sections using a fluorescein-conjugated peptide nucleic acid probe for EBER1 (DAKO) and PNA ISH Detection Kit (DAKO) according to the manufacturer's instructions. Expression of EBV latent membrane protein 1 (LMP1) was detected by the EnVision+ system (DAKO) using anti-LMP1 monoclonal antibody CS1-4 and according to the manufacturer's instructions. The lymphoma was considered LMP1 positive if more than 50 of the cells were stained in cytoplasm by the CS1-4 antibody (29).

Staining for IP-10 or CXCR3 was done as follows: formalin-fixed, paraffin-embedded sections were obtained from pretreatment biopsy samples of 10 patients with nasal NK/T-cell lymphoma. The sections were deparaffinized and treated with microwave irradiation for 8 min at 750 W for antigen retrieval. After prevention of nonspecific staining by Protein Block Serum-Free (DAKO), the sections were incubated overnight at 4C with 2.5 g/mL goat anti-human antibody IP-10 (R&D Systems) or with 1:100 rabbit anti-human antibody CXCR3 (Chemicon). After washing with PBS, the sections were incubated with Histofine Simple Stain MAX PO (G) (Nichirei Bioscience, Inc.) on IP-10 staining or EnVision+ horseradish-labeled dextran polymer (DAKO) on CXCR3 staining for 30 min at room temperature. IP-10 or CXCR3 was visualized by the DAKO liquid with the 3,3-diaminobenzidine tetrahydrochloride substrate chromogen system (DAKO). For double staining of IP-10 or CXCR3 and CD56, the sections stained with IP-10 or CXCR3 were incubated overnight at 4C with 1:50 mouse anti-human CD56 monoclonal antibody (Novocastra) followed by incubation with anti-IgG Mouse Goat Poly Alkaline phosphatase (Chemicon) as a secondary antibody for 30 min at room temperature. CD56 was visualized by freshly prepared Fast Red substrate solution (DAKO). Finally, the sections were counterstained with Lillie-Mayer's hematoxylin. A case in which more than 30 of the CD56-positive cells were also IP-10 or CXCR3 positive was defined as IP-10 or CXCR3 positive (30).

Two group comparisons were tested using nonparametric test procedures such as the Mann-Whitney U test and Wilcoxon signed rank test. Statistical tests were based on a level of significance of P < 0.05.

For screening of difference in profile of chemokine production among SNK-6, SNT-8, and KHYG-1 cells, we used a chemokine protein array (Fig. 1A). The measurements of the amount in culture supernatants of the cell lines revealed that SNK-6 and SNT-8 produced more IP-10 than KHYG-1. According to quantification of chemokine expression by using Image J software, IP-10 signal of SNK-6 and SNT-8 was 2.808-fold and 4.267-fold higher than the positive control signal, respectively. However, the IP-10 signal in KHYG-1 was one seventh of the positive control signal. For confirmation of the results, we examined IP-10 production using ELISA (Fig. 1B). ELISA analysis confirmed that SNK-1, SNK-6, and SNT-8 produced IP-10 in a time-dependent manner, but KHYG-1 and NK-92 did not.

Next, to examine whether nasal NK/T-cell lymphoma cell lines express CXCR3, the receptor of IP-10, we performed flow cytometry (Fig. 1C). The analysis revealed that CXCR3 were expressed on nasal NK/T-cell lymphoma cell lines SNK-1, SNK-6, and SNT-8, as well as in the EBV-positive NK-cell line KAI-3 and the EBV-negative NK cell lines KHYG-1 and NK-92. However, they were not detected in Raji cells as previously reported elsewhere (31).

According to the results above, it became clear that nasal NK/T-cell lymphoma cell lines produce IP-10 and express CXCR3. The next step was to investigate whether IP-10 plays a role as a growth factor in an autocrine manner; thus, we performed MTS assays on SNK-6 cells under culture conditions with exogenous IP-10 or antiIP-10-neutralizing antibody. KHYG-1, which has CXCR3 but does not produce IP-10, was also used for MTS assays. Administration of exogenous IP-10 affected cell growth in neither SNK-6 nor KHYG-1 (Fig. 2A). Similarly, treatment of IP-10neutralizing antibody changed cell growth in neither SNK-6 nor KHYG-1 (Fig. 2B). Even alternating the concentration of IP-10 or IP-10neutralizing antibody did not make a difference in the cell growth.

Because the results above indicate that IP-10 does not act as a cell growth factor for nasal NK/T-cell lymphoma cells, we subsequently performed an invasion assay using Matrigel-coated filters, which are widely used to examine invasive migration (32). Administration of exogenous IP-10 increased the number of the cells that migrated through the Matrigel membrane in SNK-1, SNK-6, and KHYG-1. The number increased in a dose-dependent manner (Fig. 2C). On the other hand, treatment with antiIP-10-neutralizing antibody or antiCXCR3-blocking antibody inhibited the number of migrated cells in SNK-1 and SNK-6 (Fig. 2D). The inhibition was enhanced in a dose-dependent manner. Because KHYG-1 did not produce IP-10, the antibodies did not affect the number of migrated cells in KHYG-1.

For the next trial, to examine whether endogenous IP-10 produced by SNK-6 has a functional role in cell invasion, we performed an invasion assay using supernatant culture fluid of SNK-6 or KHYG-1 instead of serum-free medium in cell suspension (Fig. 3). The number of migrated cells in SNK-1, SNK-6, and KHYG-1 increased in the presence of the culture supernatant of SNK-6 but did not change in the presence of culture supernatant of KHYG-1. Furthermore, the effect of the culture supernatant of SNK-6 was inhibited by antiIP-10-neutralizing antibody and antiCXCR3-blocking antibody.

Finally, we examined whether IP-10 was detected in the biopsy tissues and sera from patients with nasal NK/T-cell lymphoma. The pathologic findings were shown in Fig. 4A to F and the results are summarized in Table 1. In immunohistologic single or double staining with anti-CD56 and antiIP-10 antibodies, we found that a number of CD56-positive lymphoma cells coexpressed IP-10 in the cytoplasm in 7 of 10 patients tested (Fig. 4C and D). Similarly, CXCR3 was expressed on the lymphoma cell surface in 7 (70) of 10 patients tested (Fig. 4E and F). EBER1 was detected in all 10 patients, and LMP1 was expressed in 6 (60) of 10 patients. All 6 patients with LMP1 expression showed IP-10 expression, whereas 3 (75) of 4 patients without LMP1 expression did not express IP-10 (P < 0.05; Table 1).

ELISA results showed that IP-10 existed in serum samples from all 10 patients before treatments (Table 1). The serum IP-10 levels of the patients before treatments were significantly higher than those from healthy volunteers (P = 0.01; 81-1,635 pg/mL with median 521 pg/mL and 41-175 pg/mL with median 83 pg/mL, respectively; Fig. 4G). The serum IP-10 levels significantly decreased during the complete remission phase after treatments (P = 0.01; 84-278 pg/mL with median 154 pg/mL; Fig. 4C). Patients with LMP1 expression on the lymphoma cells showed significantly higher serum IP-10 level than the patients without LMP1 expression did (P < 0.05; 150-1,635 pg/mL with a median of 698 pg/mL and 81-579 pg/mL with a median of 258 pg/mL, respectively; Table 1).

IP-10 was first described as an -chemokine induced by IFN- in U937 monocyte-like cell line (33). IP-10 is produced by various types of cells such as human fibroblasts, endothelial cells, keratinocytes, mesangial cells, astrocytes, and neutrophils, and the production is enhanced by IFN-, IFN-, or IFN- (34). The main biological function of IP-10 is chemoattraction of human monocytes, activated T cells, and NK cells, on which CXCR3, a major receptor for IP-10, is expressed (35, 36). In hematopoietic tumors, IP-10 expression is detected in Reed-Sternberg cells of Hodgkin's lymphoma (37) and multiple myeloma cells (38). With regard to the function of IP-10 in tumor cells, there are few reports showing beneficial effects for tumor progression. Giuliani et al. (38) reported that IP-10 induced antiapoptotic effects in myeloma cells and cell lines. Additionally, Zipin-Roitman et al. (39) reported that IP-10 promoted invasive phenotype in human colorectal carcinoma cells.

In the present study, we analyzed the chemokine production of nasal NK/T-cell lymphoma cell lines SNK-6 and SNT-8 using a chemokine protein array and found that IP-10 was abundantly produced in both SNK-6 and SNT-8, compared with KHYG-1, which was established from EBV-negative NK cell leukemia. We confirmed, by using ELISA, that SNK-1, SNK-6, and SNT-8 produced the IP-10 protein in a time-dependent manner, but EBV-negative NK cell leukemia and lymphoma cell lines KHYG-1 and NK-92 did not produce the IP-10 protein. Furthermore, we clearly showed, using flow cytometry, that CXCR3, a major receptor of IP-10, was expressed on SNK-1, SNK-6, and SNT-8 cells. Because these results show that IP-10 and CXCR3 are coexpressed in nasal NK/T-cell lymphoma cell lines, we next examined, using immunohistologic double staining, whether IP-10 and CXCR3 are expressed on lymphoma cells in vivo. We succeeded to show that CD56-positive lymphoma cells coexpressed IP-10 as well as CXCR3 in a majority of biopsy samples from the patients tested.

Previously, Ohshima et al. (40) performed cDNA array analysis on biopsy tissues of nasal NK/T-cell lymphoma patients and showed a markedly increased expression of IP-10 mRNA. Teruya-Feldstein et al. (41) also showed IP-10 mRNA expression in biopsy tissues from seven nasal NK/T-cell lymphoma patients using reverse transcriptase-PCR. They found IP-10 expression in endothelial cells lining the capillary vessels, macrophages, and fibroblasts, but they failed to find IP-10 expression on the lymphoma cells in the tissue sections. We used immunohistologic double staining with anti-CD56 and antiIP-10 antibodies for tissue sections and succeeded to prove that the lymphoma cells express IP-10. On the other hand, it is an accepted fact that nasal NK/T-cell lymphoma cells express CXCR3 on the cell surface. Yagi et al. (42) showed CXCR3 expression on the cutaneous NK/T-cell lymphoma cells in all five patients tested with immunohistologic staining. Similarly, Ishida et al. (43) showed CXCR3 expression on the nasal NK/T-cell lymphoma cells in 4 (15) of 27 patients tested with immunohistologic staining. In this study, we also found CXCR3 expression on the lymphoma cells in biopsy tissues from the majority of the patients tested.

It is reported that IP-10 acts as a cell growth factor inhibiting apoptosis in myeloma cell lines (38) and as an invasion factor in human colorectal carcinoma cells (39). Coexpression of IP-10 and CXCR3 in nasal NK/T-cell lymphoma cell lines found here suggests that IP-10 may have functional roles in autocrine cell growth and/or invasion. As the results of the functional analyses, IP-10 does not act as a cell growth factor for the cells but acts as an autocrine cell invasion factor. Results of the in vivo analyses showed that IP-10 and CXCR3 were expressed on the CD56-positive lymphoma cells in biopsy tissues and that serum IP-10 levels of the patients was much higher than those from healthy volunteers. Moreover, the level dramatically decreased during the complete remission phase after treatments. These findings suggest that the role of IP-10 as a autocrine cell invasion factor possibly works not only in vitro but also in vivo.

We found here that IP-10 was expressed and produced only in EBV-positive NK-cells or T-cell lines SNK-1, SNK-6, SNT-8, and KAI-3, but never in EBV-negative NK cell lines KHYG-1 and NK-92. These findings suggest that EBV may contribute to IP-10 production in nasal NK/T-cell lymphoma. Although we cannot completely explain how EBV may contribute to IP-10 production, it can be inferred that the EBV oncogenic LMP1 may be one of the candidates involved in inducing IP-10. Recently, Vockerodt et al. (20), using LMP1-transfected cell lines derived from Burkitt's and Hodgkin's lymphomas, showed that LMP1 is sufficient to induce IP-10 expression in lymphoma cells involving transcriptional (NF-B) and posttranscriptional (p38/SAPK2) mechanisms. They further showed that LMP1-mediated IP-10 activation is independent from autocrine tumor necrosis factor-, IFN-, or IL-18, which have been described as inducers of IP-10 (33, 34). In our experiments, all of the nasal NK/T-cell lymphoma cell lines (SNK-6, SNK-1, and SNT-8) used here express LMP1 (11). After analyzing in vivo materials, we found that LMP1 expression of lymphoma cells closely correlated with high IP-10 expression in the biopsy tissues and high IP-10 levels in sera. This is evidence that LMP1 may act as an IP-10 inducer. Alternatively, it is possible that some cytokines, e.g., tumor necrosis factor-, IFN-, IL-18, or others, which has been proved to induce IP-10 (33, 34), may stimulate IP-10 production in an autocrine manner, as previously found in EBV-infected lymphoblastoid cells (44). Indeed, it has been reported that various EBV-related lymphoma cells produce such cytokines (45, 46). Furthermore, we previously confirmed that the nasal NK/T-cell lymphoma cell lines produce several cytokines such as IFN-, IL-9, and IL-10 (13, 14).

In conclusion, we present here experimental evidence that IP-10 is specifically produced by nasal NK/T-cell lymphoma cell lines, has a potential role as an autocrine invasion factor for these cell lines, and is detected in tissues and sera from patients with nasal NK/T-cell lymphoma. These results suggest that the IP-10 signaling pathway may be a new pharmacologic target in the treatment of the patients with nasal NK/T-cell lymphoma.

No potential conflicts of interest were disclosed.