Expression of WSX1 in Tumors Sensitizes IL-27 Signaling-Independent Natural Killer Cell Surveillance

Interleukin (IL)-27 receptor WSX1 is most homologous to the IL-12 receptor β2 chain (1). WSX1 together with gp130 constitute a functional signal-transducing receptor for IL-27, whereas lack of either subunit attenuates IL-27–mediated signaling (2). WSX1 is reported to be expressed in immune cells such as monocytes, dendritic cells, T and B lymphocytes, natural killer (NK) cells, mast cells, and endothelial cells (1).

In a patent application (PCT/2007/0280905), we reveal that WSX1 is detectable in breast epithelial tumor cells. This discovery is further supported by a recent report, which revealed that WSX1 is expressed in another type of epithelial tumor, melanoma cells (3). The same as found in immune cells, WSX1 is functional in these epithelial tumors cells as indicated by the IL-27–mediated activation of STAT1 and STAT3 (3).

Clearly, the reports found in the literature suggest that WSX1 plays a role through the IL-27 signaling pathway, but the IL-27–independent role of WSX1 in promotion or inhibition of tumorigenesis has not been reported yet. Using genetically modified tumor cells, we present evidence that the expression of WSX1 in epithelial tumor cells suppresses tumor growth both in vitro and in vivo. Such inhibition of tumor growth is dependent on NK cells but independent of IL-27 signaling. Our results reveal a novel function of WSX1 in epithelial tumor cells, which is to sensitize NK-mediated antitumor immunosurveillance in an IL-27–independent manner.

Cell culture and reagents. Human cancer cell lines from different tissue origins including HELA, HT29, HCT116, and 4T1 were purchased from the American Type Culture Collection. Human breast cancer cell lines MDA468, MDA231, and MCF7 were provided by Dr. Bolin Liu (University of Colorado Denver School of Medicine). The normal colon cell line NCM460 was purchased from INCELL. UM-SCC11A, UM-SCC11B, UM-SCC17A, and UM-SCC17B are head and neck squamous cell carcinomas provided by Dr. Thomas Carey (University of Michigan; ref. 4). Mouse human papilloma virus-associated tumor cell line TC1 was provided by Dr. T.C. Wu (John Hopkins University; ref. 5). The mouse squamous cell carcinoma cell line AT84 was provided by Dr. Edward Shillitoe (State University of New York Upstate Medical School). Recombinant mouse IL-27, NKG2D/Fc, monoclonal anti-human WSX1, and anti-human MICA-Pe antibody were purchased from R&D Systems. Anti-mouse pSTAT1-701, anti-mouse IgG-PE, actin, anti-human IgG-PE, and anti-hamster IgG-PE were purchased from Santa Cruz Biotechnology. Anti-NKG2D C7 was provided by Dr. Wayne Yokoyama (Washington University School of Medicine).

Quantitative real-time PCR. Total RNA was isolated from cells using TRIzol Reagent (Invitrogen). Residual genomic DNA was removed from total RNA using the TURBO DNA-free kit (Applied Biosystems/Ambion). Two micrograms of RNA were used for cDNA synthesis using the High-Capacity RNA-to-cDNA Kit (Applied Biosystems). The relative gene expression levels were determined using quantitative real-time PCR and SYBR Green labeling method in an ABI 7300 Sequence Detector (Applied Biosystems). The reaction contained 2 μL cDNA, 12.5 μL SYBR Green PCR Master Mix (Applied Biosystems), and 200 to 250 nmol/L primer in a total volume of 25 μL. The PCR cycling conditions were as follows: 40 cycles of 15 s at 95°C and 60 s at 60°C. All samples were run in duplicates. PCR amplification of β-actin was done using 0.1 μL cDNA. The C_T value of each sample was acquired, and the relative level of gene expression was calculated by the ΔC_T method, which was normalized to the endogenous control of β-actin. Data were expressed as a n-fold relative to control. The forward and reverse primer sequences for the human β-actin and WSX1 detection are: 5′-AGAGGGAAATCGTGCGTGAC-3′ and 5′-CAATAGTGATGACCTGGCCGT-3′, WSX1: 5′-GAGCCCCCTCCGAGTTACAC-3′ (forward) and 5′-AGCTGTTCCCGAGGAATGG-3′ (reverse).

Establishing stable WSX1-expressing cell lines. The murine WSX1 gene was purchased from Open Biosystems and subcloned into pBMN-green fluorescent protein (GFP) plasmid (Phoenix Retrovirus Expression System). The retrovirus was produced by transfecting mWSX1/GFP constructs into Phoenix eco packaging cells. AT84 and TC1 cells were infected with retroviral containing supernatant derived from the transduced HEK293 cells. The transduction was confirmed by detecting GFP-expressing cells under the fluorescence microscope. Cell colonies with GFP expression from a single cell were picked, expanded, and further confirmed for WSX1 expression using flow cytometry. Using this approach, both WSX1/GFP- and GFP-positive TC1 and AT84 cells were obtained.

Animal procedures. All the animal procedures were approved by the Institutional Animal Care and Use Committee at Louisiana State University. Six- to 8-week-old mice were used for this study. The subcutaneous tumor models were generated by subcutaneously inoculating TC1 and AT84 tumor cells (2 × 10⁵ in a 30 μL volume per mouse) into mice. Tumor measurement and calculation were the same as described previously (6). C57BL/6 WSX1 knockout mice were provided by Dr. Fred de Sauvage (Genentech), and C57BL/6 EBV-induced gene 3 (EBI3) knockout mice were provided by Dr. Mark P. Birkenbach (Temple University School of Medicine). C57BL/6, BALB/c severe combined immunodeficient, C3H, C57BL/6 perforin, and Rag-deficient mice were purchased from commercial sources.

NK cytotoxicity assay. NK cell activity was evaluated using the CyToxiLux kit (OncoImmunin), a single-cell-based fluorogenic cytotoxicity assay (7, 8). Effector cells were prepared from spleens as described previously (8) and incubated with red fluorescence-labeled target cells at a ratio 100:1, 50:1, and 25:1 in 200 μL cell culture medium. Target cells alone were used as control for spontaneous cell death. Sixteen hours after incubation, adhesive target cells were washed with PBS. Alive red target cells (input target cells) were counted using Olympus BX41 fluorescence microscope. NK activity was calculated using the following equation: % NK cell activity = 100 × (input target cells - output target cells) / (input target cells).

Flow cytometry. Cells were stained with the indicated primary and secondary antibodies for 30 min at 4°C as indicated in each figure. The expression of the indicated genes was analyzed on FACSCalibur and CellQuest graphics software (BD Biosciences).

Cell proliferation. Cell proliferation assays were done using the luminescence ATP Lite assay detection system (Perkin-Elmer). Briefly, 500 cells were seeded in a 96-well plate; cells were lysed on days 0, 2, and 4 for measuring the ATP levels. The cell proliferation index was calculated using the following equation: cell proliferation index = [ln(d)] / [ln(d0)], where ln(d) = natural log at the day when cells were lysed and ln(d0) = natural log at the day when cells were seeded.

Soft-agar growth assay. The clonogenic assay was done as described previously (9). Briefly, genetically engineered cells (5 × 10³ for TC1-GFP and TC1-WSX1 cells and 1 × 10³ for AT84-GFP and AT84-WSX1 cells) were suspended in 0.34% agar in cell culture medium (Sigma). The mixture solution was layered on solid agar support prepared from 0.9% agar in cell culture medium. The cells were seeded in triplicates on a 6-well plate and grown for 2 weeks. Colonies were counted under a 10× dissecting microscope after staining with 0.05% crystal violet for 1 h. Images were captured using Molecular Image Gel Dox XR (Bio-Rad).

Western blot. Cells (5 × 10⁵) growing in 10% heat-inactivated fetal bovine serum-containing culture medium were treated with or without IL-27 (20 ng/mL) for the indicated times. Protein extract was obtained by directly lysing cells using 60 μL Laemmli sample buffer. Twenty microliters of total protein extract from each sample were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and incubated with primary antibody overnight at 4°C (750-fold dilution for anti-pSTAT1/701 and 1,000-fold dilution for anti-actin).

Statistical analysis. For in vivo experiments, univariate repeated-measures ANOVA was used to analyze the difference among treatments using SAS version 9.1.3. When appropriate, Tukey's HSD test was done for interaction affects. For in vitro results, Student's t test analysis was conducted.

A much higher level of functional WSX1 is expressed in most of the tested epithelial tumor cells than in normal epithelial cells. It is well known that the IL-27 receptor WSX1 is expressed mainly in immune cells and the only other type of cells expressing this gene are endothelial cells. However, recently we and others have found WSX1 expression in breast and melanoma epithelial tumor cells lines (3). To determine whether the WSX1 expression in epithelial tumor cells plays an important function, we have compared the magnitude of WSX1 expression between normal and tumor epithelial cells. The quantitative analysis result showed that WSX1 was present not only in breast cells but also in colon, cervical, and squamous cell carcinoma tumor cells, suggesting that WSX1 was expressed in most human epithelial tumor cells (Fig. 1A). However, the expression level of this gene varied greatly among the different cell lines when compared with the normal epithelial cell line NCM460 (Fig. 1A). A few cell lines such as HT29 and UM-SCC17A showed 6.9- to 8.4-fold lower expression of WSX1 when compared with a normal epithelial cell line, NCM460, whereas most of the cell lines such as HELA, HCT116, and UM-SCC11A showed much higher levels of expression (ranging from 13 to 78 times higher than NCM460). The high level of WSX1 expression in most of the epithelial tumor cells but not in the normal epithelial cells suggests that WSX1 may play a role in regulating tumor progression. The level of WSX1 protein expression was positively associated with the level of mRNA (Fig. 1B).

Because WSX1 is the receptor of IL-27, one obvious question is whether the high level of gene expression is associated with a high level of function. We used phosphorylation of STAT1 by IL-27 as a functional WSX1 end point. After 10 min of incubation with IL-27 (lane 2 in each panel; Fig. 1C), IL-27 induced phosphorylation of STAT1; such an increase correlates well with the presence of WSX1 expression but does not correlate with the absolute level of WSX1 expression (Fig. 1B). The human cell lines HT29 and UM-SCC17A lacking WSX1 expression showed very low to no detectable STAT1 phosphorylation, whereas cell lines such as HeLa, HCT116, and UM-SCC11A with WSX1 expression showed an increase in phosphorylation of STAT1. However, a similar level of phosphorylation was detected in both low-level (NCM460 and UM-SCC17B) and high-level (HCT116 and HELA) WSX1-expressing cells (Fig. 1C versus Fig. 1A).

WSX1 reduces tumorigenicity and proliferation of epithelial tumor cells. The results from others exclusively illustrate that WSX1 plays a role in inducing an immune response through IL-27 signaling in immune cells. The result from Fig. 1B confirms that the IL-27/WSX1 signaling occurs in epithelial tumor cells. However, the high level of WSX1 expression does not correlate to the STAT1 phosphorylation in most epithelial tumor cells (Fig. 1C). Moreover, the endogenous IL-27 is undetectable in either serum or splenocytes and therefore may not initiate any signaling in either tumor or host cells during tumorigenesis and development. These facts suggest that a high level of WSX1 expression alone may affect tumorigenesis, which is the central hypothesis to be tested below.

To determine whether increased WSX1 expression alone may regulate IL-27 signaling-independent tumor development, a clonogenic assay was done to determine the tumorigenicity and proliferative ability of tumor cells engineered with WSX1 or GFP control genes. This method has been known to be effective in determining these end points (10). Flow cytometry analysis confirmed the expression of WSX1 in the stable transfected cell clones (Fig. 2A). The clonogenic assay results illustrated that expression of WSX1, but not GFP, dramatically reduced the ability of cells to grow in soft agar in both TC1 and AT84 cells (Fig. 2B and C).

To further confirm the inhibitory effect of WSX1 expression on tumor cells, cell proliferation was determined. Similar to the clonogenic assay, WSX1 significantly reduced the proliferation of TC1 and AT84 cells, but the inhibition of AT84 proliferation by WSX1 expression was at a much lower magnitude when compared with TC1 (Fig. 2D).

WSX1 suppresses tumor growth in vivo in both TC1 and AT84 tumor models. Although the clonogenic assay is a good predictor of tumorigenicity in vitro, it does not read any host cell-induced cytotoxic endpoints that occur in a true tissue environment (11). To avoid this problem, the effect of WSX1 expression on tumor growth was tested in syngeneic mice by subcutaneously inoculating with GFP- or WSX1-expressing tumor cells. In agreement with the in vitro assay result, WSX1 expression almost completely abolished TC1 tumor growth (Fig. 3A). Likewise, WSX1 expression also inhibited AT84 tumor growth in a different mouse strain (Fig. 3B). The remarkable difference in the tumor growth rate shown in TC1 and confirmed in AT84 strongly indicates that WSX1 has a tumor-suppressive role in epithelial tumor cells.

WSX1-mediated suppression of tumor growth is dependent on NK cells. The direct antitumor mechanism by WSX1 was not found in the literature, but one possible explanation could be due to the reduction of tumor cell proliferation as observed in vitro (Fig. 2D). However, the dramatic difference in tumor growth in vivo (Fig. 3A) when compared with the small difference of proliferation in vitro (Fig. 2D) between TC1-WSX1 and TC1-GFP indicates that cell proliferation differences alone may not be the major cause for the diminished tumor growth in the presence of WSX1. An alternative assumption is that WSX1 expression in tumor cells may enhance the immune surveillance by the host immune cells. To distinguish whether the effect of cell proliferation or the immune system might be the major mechanism that accounts for the WSX1-dependent tumor growth inhibition in vivo, we tested tumor growth in wild-type, perforin (NK), and Rag (T and B) knockout mice.

Similar to wild-type mice, tumor growth reduction between TC1-GFP and TC1-WSX1 was found also in T and B knockout mice (Fig. 4A,, left versus middle). However, the absence of perforin almost completely impaired the ability of WSX1 to inhibit tumor growth, as there is no statistically significant difference in tumor growth between GFP and WSX1 tumors (Fig. 4A,, right). This result suggests that WSX1 may sensitize NK cell surveillance for inhibiting WSX1-positive tumor growth. To further support this statement, tumors engineered with control GFP gene grew at a similar rate regardless of the presence or absence of NK or T cells (Fig. 4B,, left). In contrast, WSX1-positive TC1 tumors disappeared in 3 of 4 wild-type mice, grew very slowly in T- and B-deficient mice (reaching 50 mm³ by day 40 after tumor inoculation), and developed aggressively in NK-deficient mice (averaging 450 mm³ by day 40, almost 9 times higher than in T and B knockout mice; Fig. 4B,, right). To confirm that this observation was not dependent on a single clone, another independent clone (TC1-WSX1-CL2) was tested in vivo. Similar to the other engineered WSX1-positive TC1 clone, TC1-WSX1-CL2 was eradicated in 4 of 4 wild-type mice, whereas it reached 200 mm³ by day 38 in NK-deficient mice (Fig. 4C). These findings suggest that our observation is not clone-dependent.

Similar to the TC1 model, the WSX1-positive AT84 tumor cells grew slower than the control AT84-GFP in both wild-type and T- and B-cell–deficient mice (Fig. 4D,, left versus middle). To test the WSX1-mediated NK cell dependence for the AT84 tumor model, STAT1 knockout mice were used because STAT1 is an essential transcription factor for NK cell function. In the absence of STAT1, these mice show impaired NK activity in vitro and fail to reject NK-sensitive tumors in vivo (12). As expected, no difference in the growth of control and WSX1-positive AT84 tumors was detected in these NK-defective STAT1-deficient mice (Fig. 4D , right).

WSX1 suppression of tumor growth is independent of IL-27. The presented results strongly suggest that WSX1 may play an antitumor role independent of IL-27 because the endogenous level of IL-27 is undetectable in mice. To exclusively confirm the role of IL-27, because the cooperation among WSX1 expression in tumor cell and NK cell is needed for antitumor activity, and IL-27 signals in both tumor and immune cells, WSX1-mediated tumor growth inhibition was compared in EBI3 knockout mice. Because EBI3 is a subunit of IL-27, the lack of EBI3 would result in inhibition of IL-27 signaling in both the host and the tumor. As expected, lack of endogenous IL-27 did not affect WSX1-mediated tumor growth suppression and a similar difference in tumor growth between TC1-GFP and TC1-WSX1 was found in wild-type mice as was found in IL-27 EBI3 knockout mice (Fig. 4A,, left versus Fig. 5A).

To further extend this exclusive confirmation, tumor growth in WSX1 knockout mice was also tested. Because WSX1 is a specific receptor for IL-27, the lack of WSX1 in these mice should impede IL-27 signaling in the immune cells. Similar to EBI3 knockout mice, WSX1-mediated tumor growth suppression was retained in WSX1 knockout mice (Fig. 5B). Moreover, WSX1-positive tumors grow at a similar rate in wild-type, EBI3, and WSX1 knockout mice (Fig. 5C). These results support our hypothesis that WSX1 retains its ability to impede tumor growth in the absence of IL-27 signaling in either tumor or host cells.

WSX1 sensitizes NK cell surveillance by inducing NKG2D ligand expression in tumor cells. Our data above clearly show that the ability of WSX1 to suppress tumor growth is dependent on NK cells and independent of IL-27 signaling in either the tumor or the host. The question is whether the presence of WSX1 in tumor cells directly sensitizes NK cell-mediated cytotoxicity. To accomplish this goal, NK cell cytotoxicity was compared between TC1-GFP and TC1-WSX1 cells. These assays revealed that TC1-WSX1 cells are more efficiently lysed when compared with the TC1-GFP cells (Fig. 6A,, left). To rule out the role of CD8 T cells, we used perforin and Rag knockout splenocytes. Similar to wild-type splenocytes (Fig. 6A,, left), splenocytes from Rag knockout mice lysed TC1-WSX1-positive cells more efficiently than TC1-GFP (Fig. 6A,, middle). Contrarily, the lack of perforin eliminated the enhanced cytolytic activity against TC1-WSX1 (Fig. 6A,, right). Similarly, cytotoxicity against AT84-WSX1 was significantly higher than AT84-GFP (Fig. 6B). These results indicate that WSX1 expression in tumor cells provokes a direct NK cell surveillance.

One possible hypothesis is that WSX1 increases NK cell cytotoxicity by promoting the interaction between tumor cells and NK cells. Given that NKG2D is one of the primary receptors that promotes tumor cell surveillance (13), we determined whether the expression of WSX1 in tumors up-regulates expression of NKG2D ligands. Flow cytometry analysis using a NKG2D/Fc binding assay (a reagent that detects cell-surface expression of all known NKG2D ligands) confirmed the hypothesis that WSX1, but not GFP, greatly enhanced the expression of NKG2D ligands (Fig. 6C,, left and middle). Such ability of WSX1 to induce the expression of NKG2D ligands is more pronounced in the TC1 model than in the AT84 model (Fig. 6C , right).

To definitely confirm the hypothesis that up-regulation of NKG2D ligands by WSX1 expression is the central mechanism to enhance NK-mediated cytolytic activity against WSX1-positive tumor cells, NKG2D receptors on NK cells were blocked using anti-mNKG2D C7 in the cytolytic assay (14). As expected, addition of a NKG2D neutralization antibody, anti-mNKG2D C7, substantially reduced NK cell lysis against both TC1-WSX1 and AT84-WSX1 target cells to levels comparable of GFP counterparts (Fig. 6D,, left and middle). Furthermore, WSX1 expression is strongly correlated (R² = 0.7293) to MICA expression in human cancer cell lines (Fig. 6D , right).

Although it is well known that WSX1 is expressed in T and NK cells and is a critical receptor for triggering immune responses via IL-27 signaling in these cells (1, 2, 15), our recent results revealed that WSX1 is also expressed in epithelial tumor cells such as breast tumor cell lines, whereas others have found its expression in human melanoma (3) and leukemia cells (16). Neither us nor others have quantified the level of WSX1 expression and compared the level of expression between epithelial tumor cells and normal epithelial cells. In this study, using a quantitative real-time PCR assay, we surprisingly found that WSX1 is not only expressed in most of the tested epithelial cells (8 of 10) but also expressed 13- to 78-fold higher than normal epithelial cells (Fig. 1A).

Whereas the function of WSX1 in immune cells has been studied extensively, its function in epithelial tumor cells has hardly been studied. In immune cells, it is generally accepted that IL-27 possesses both proinflammatory and anti-inflammatory properties (17–25). For example, IL-27 induces key components to Th1 commitment such as synergistic induction of IFN-γ with IL-12, proliferation of naive CD4⁺ T cells, induction of T-bet, and IL-12 receptor β expression and possesses T- and NK-cell-mediated antitumor activities (15, 20–25); however, parasitic studies show that the absence of WSX1 triggers aberrant cytokine production (26, 27). In contrast to previous studies in immune cells, our findings assign a new role to WSX1 in epithelial cancer cell biology that expression of WSX1 inhibits epithelial tumor growth in vitro and in vivo (Figs. 1, 3, and 4). Such an observation was confirmed in two independent tumor models, TC1 and AT84. However, this observation is different in tumors derived from immune cells in which WSX1 elicited antiapoptotic and mitogenic signals (16) and transformed two leukemia cell lines, 32D and BaF3.

Different from the recent report in which WSX1 expression in epithelial melanoma cells requires IL-27 to inhibit its tumor cell proliferation and tumor growth (3), our results provide a strong case of IL-27–independent antitumor activity by WSX1 (Figs. 2 and 5). First, IL-27 independence was shown by our trial showing that WSX1 expression inhibited tumor growth in mice that lacked the IL-27 subunit EBI3, the same as found in wild-type mice (Fig. 5A). Second, WSX1 expression in epithelial tumor cells inhibits tumor growth in WSX1 knockout mice (Fig. 5B).

Currently, no linkage between WSX1 expression in tumor cells and NK cell-mediated surveillance is reported in the literature. Using two different tumor models, TC1 and AT84, we show that WSX1 suppresses tumor growth via NK cells. In perforin and STAT1 knockout mice, WSX1-dependent suppression of tumor growth is impaired (Fig. 4), whereas in vitro WSX1 tumor cells are more sensitive to NK cell cytotoxicity (Fig. 6). Such phenomena are dependent on the NKG2D pathway, as the presence of WSX1 directly enhances the expression of NKG2D ligands (Fig. 6) thereby enhancing NK cell-mediated recognition and release of cytotoxic molecules toward tumor cells.

Considering that WSX1 expression in our tumor models inhibits tumor growth, whereas expression of WSX1 in certain leukemia cell lines confers transformation (16), further work is needed to investigate the molecular mechanism downstream of WSX1 that leads to opposing consequences between epithelial and blood-derived tumors.

This link between WSX1 and NKG2D ligands, which are the intrinsic sensors of oncogenic transformation that induce innate immunosurveillance (28), suggests that WSX1 expression in epithelial tumor cells might play a role to prevent tumorigenesis but was eventually overridden by the defaulted pathway or immune escape mechanism. Such a scenario is also seen with the MICA-NKG2D surveillance system in which a high level of MICA expression in tumor cells was subjected to NKG2D surveillance (29, 30), but this surveillance system was overridden by the aggressive tumors due to the development of multiple immune escape mechanisms (31, 32). The connection of WSX1 to innate immunosurveillance could explain the observed discrepancy among the two tumor models used in this study: WSX1 is more effective in inhibiting tumor growth and increasing NKG2D ligand expression in TC1 model, a human papilloma virus-transformed normal cell line, than in aggressive AT84 models, a spontaneously arising oral squamous cell tumor of C3H mice, albeit WSX1 expression is higher in AT84 than in TC1.

In summary, this study exposed a novel function of WSX1 in epithelial tumor cells, linked WSX1 to NK-mediated antitumor immunosurveillance, and, most importantly, revealed that WSX1 induces an antitumor function by up-regulating the expression of NKG2D ligands and this process is independent of the IL-27 signaling pathway.

No potential conflicts of interest were disclosed.

Grant support: NIH grant RO1CA120895 and NIH/National Institute of Biomedical Imaging and Bioengineering grant R21EB007208.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Fred de Sauvage for providing TCCR/WSX1-deficient mice.