USP18 is crucial for IFN-γ-mediated inhibition of B16 melanoma tumorigenesis and antitumor immunity

Adenovirus containing mouse USP18 (Ad-mUSP18) was purchased from Applied Biological Materials Inc. (Richmond, BC, Canada). We prepared lentivirus constructs containing mouse USP18 shRNA. Rabbit and goat anti-mouse USP18 antibodies were kindly provided by Dr. Ethan Dmitrovsky (Dartmouth-Hitchcock Medical center, Dartmouth College, USA) or purchased from Santa Cruz Biotechnology.

C57BL/6, NOD-SCID-IL2Rγ^-/- (NSG), Ifng^-/-, OT-1 and OT-2 C57BL/6 and pmel-1 C57BL/6 transgenic mice were purchased from Jackson Laboratory. All mice were 6- to 7 weeks of age at the time of experiment, and at least 5 mice per group were used in each experiment. Mice were housed and experimental procedures were performed in accordance with the IACUC guidelines at University of Texas MD Anderson Cancer Center and Cleveland Clinic.

Overexpression of USP18 into the tumor cell line B16 was accomplished by transduction of adenovirus Ad-mUSP18- followed by cell sorting to select GFP-positive tumor cells (B16-USP18, B16-OVA-USP18). Stable knockdown of USP18 was accomplished by lentivirus shUSP18 transduction of B16 and B16-OVA tumor cells and sorting for GFP-positive tumor cells (B16-shUSP18, B16-OVA-shUSP18).

Subcutaneous and intravenous murine melanoma models were established as described elsewhere[13]. Briefly, for the subcutaneous tumor model, 0.5- to 1.0 million B16-OVA-shCon, B16-OVA-USP18, B16-OVA-shUSP18 tumor cells were subcutaneously inoculated. For the intravenous tumor model, 0.3- to 1.0 million of tumor cells were intravenously injected, and lung tumor foci were counted to assess tumor burdens. In some experiments, the tumor model was first established, followed by adoptive transfer of antigen-specific CTLs to the tumor-bearing mice.

Cells from tumor, spleen, lymph nodes and lung were mechanically dissociated, and the red blood cells were removed by ACK lysis buffer (Lonza). Cells were first blocked with Fc antibody and then labeled with different combinations of antibodies. Data were acquired with an LSR Fortessa flow cytometer (BD Biosciences), and analysis was performed using Flowjo software. Cell sorting was performed using a BD FACSAria cell sorter (BD Biosciences). Fluorochrome conjugated CD4, CD8, CD11b, NK1.1, PD1, NKG2D, CD86, MHC I, MHC II, KLRG1, IFNγ and TNFα were bought from BD Biosciences. PE conjugated OT-1 tetramer were bought from Beckman Coulter, Inc.

Cell proliferation was assessed by a BrdU incorporation assay. Cells cultured in vitro were treated with 10 μM BrdU for 30 min or 16 hrs when the cells were in exponential growth. Cells were harvested and stained with BrdU-APC/7AAD. For apoptosis assay, cells were cultured until they reached exponential growth and then were collected and stained with Annexin-V/PI.

Cell lysates, 20 μg, were electrophoresed on 4-12% SDS-PAGE gels, and transferred onto 0.22 μm nitrocellulose membrane (Hybond, Amersham Biosciences, Inc.). The membrane was then incubated with specific antibody followed by horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences, Inc.). The expressed protein was detected with an ECL Plus Western blotting detection kit (Amersham Biosciences, Inc.).

mRNA was prepared from cells or tissues using a Qiagen RNAeasy Kit. cDNA was synthesized using an ABI Biosystems cDNA synthesis kit. Quantitative PCR was conducted with SYBR Green Master Mix (Life Technologies) and run in an ABI750 thermal cycler. The primers for USP18 are: 5′-atgcaggacagtcgacagaa-3′ and 5′- tgtcaagtctgtgtccgtga-3′.

Paraffin-embedded tumor tissues were processed with pH9.0 antigen retrieval buffer. The slides were blocked with 3% H₂O₂ and the serum of the species of secondary antibody. The primary antibody was incubated at 4°C overnight. Slides were washed and secondary antibody was incubated for 20 min at room temperature. Slides were washed three times with 1 × PBST and then developed with DAB and counterstained with H&E and mounted for microcopy analysis.

B16-OVA-GFP or B16-OVA-USP18 tumor cells were cocultured with or without IFN-γ sensitization (5 ng/ml) and subjected to low dose (10 Gy) irradiation. OT-1- and OT-2-naive T cells were cocultured with irradiated tumor cells for 48- to 72 hours. T-cell proliferation was analyzed by a CFSE dilution assay or ³H-incorporation assay. IL-2 and IFN-γ secretion from cocultured T cells was analyzed by ELISA.

IFN signaling plays important roles in cell activity, including cell growth, differentiation, proliferation, and immune regulation. IFN signaling also is critical for immunosurveillance for tumorigenesis and metastasis[5], including surveillance for induction of MHC class-I molecule expression on tumor cells[14]. We first verified that IFN-α/-β or IFN-γ stimulation increased MHC class-I expression on tumor cells (Additional file1: Figure S1A). Inoculated tumor cells in IFN-γ signaling-deficient (Ifng^-/-) mice had reduced MHC class-I expression (Additional file1: Figure S1B).

IFN signaling is regulated by various molecules and the canonical factors are the IFN-sensitive response elements (ISREs)[8]. USP18 is an IFN stimulated gene that has been reported to regulate type-I IFN signaling in the anti-viral immune response[12]. To our knowledge, very few studies have investigated USP18 regulation of type-II IFN signaling in tumor immunosurveillance. We assessed USP18 mRNA and protein levels in B16 tumor cells after IFN-α or IFN-γ stimulation and found that IFN-γ was a more potent inducer than IFN-α for USP18 expression (Figure 1A and B), as well as Stat1 phosphorylation, which is IFN-γ signaling downstream gene (Figure 1B). Western blot analysis showed that IFN-γ induced USP18 protein expression in other tumor cell lines such as 4 T1 and EMT6 tumor cells (Figure 1C). To investigate whether endogenous IFN signaling induces USP18 expression during tumor development in vivo, we inoculated C57BL/6 mice with B16-GFP tumor cells. Tumor cell USP18 expression was increased in vivo as compared with in vitro conditions (Figure 1D). To examine exogenous IFN signaling, such as CTL secretion of IFN-γ, we performed adoptive transfer of activated OT-1 cells, which resulted in stronger USP18 expression in tumor cells in vivo (Figure 1D). Similar results were found in in vitro coculture (Figure 1E). These results demonstrated that USP18 was expressed in tumor cells during tumor development regardless of whether IFN signaling was endogenous or exogenous.

To determine whether tumor cell USP18 expression contributes to tumorigenesis in vivo, we stably silenced or ectopically expressed USP18 in B16-OVA tumor cells (Additional file1: Figure S2A and B). As IFN signaling was required for USP18 expression in tumor cells, knockdown of USP18 was confirmed after IFN-γ stimulation (Additional file1: Figure S2A). We found that USP18 knockdown or overexpression did not affect the proliferation or apoptosis status of the cells, either with or without IFN-γ stimulation (Additional file1: Figure S2C and D). B16-OVA-shCon (control), B16-OVA-shUSP18 (USP18 knockdown), or B16-OVA-USP18 (USP18 overexpression) cells were then intravenously inoculated into immunocompetent C57BL/6 mice. Knockdown of USP18 expression significantly increased tumor burden in C57BL/6 mice (Figure 2A and B, and Additional file1: Figure S3) and shortened mouse survival time (Figure 2C). Conversely, overexpression of USP18 in B16-OVA tumor cells significantly reduced tumor burden in C57BL/6 mice (Figure 2D and E, and Additional file1: Figure S3) and prolonged mouse survival time (Figure 2F). As tumor cells were intravenously injected into mice, the change in tumor burden was not due to the difference of invasion or migration of tumor cells (Additional file1: Figure S4A-C).

In order to confirm this finding, we subcutaneously inoculated B16-OVA-shCon, B16-OVA-shUSP18, or B16-OVA-USP18 tumor cells into C57BL/6 mice. Tumor growth in mice receiving B16-OVA-USP18 cells was significantly inhibited, whereas tumor growth was enhanced in those receiving B16-OVA-shUSP18 tumor cells (Figure 2G-I).

To explore whether regulation of tumorigenesis by USP18 expression in tumor cells was due to host immune status, we intravenously inoculated B16-OVA-GFP or B16-OVA-USP18 tumor cells into NSG (NOD-SCID/IL2Rγ^-/-) mice, which are deficient in T, B and NK cells. In contrast to immunocompetent mice, ectopic expression of USP18 did not inhibit tumorigenesis in NSG mice (Figure 3A and B), and no significant difference of survival in tumor-bearing mice was noted (Figure 3C). However, the tumor burden was higher in NSG mice compared with immunocompetent mice for both the B16-OVA-USP18 and B16-OVA-GFP inoculation groups, suggesting that lack of the immune system or immunosurveillance enabled the tumor cells to be more aggressive[15].

CD4⁺ and CD8⁺ T cells and NK cells are the principal IFN-γ-producing cells, and their anti-tumor activity depends on the robustness of IFN-γ production. To determine how tumor cell USP18 affects these cells and thus the immune response, we injected B16-OVA-GFP or B16-OVA-USP18 tumor cells into C57BL/6 mice that were depleted of CD4⁺ or CD8⁺ T cells or NK cells and monitored tumorigenesis. CD8⁺ T cell depletion significantly compromised the effect of USP18 expression on tumorigenesis (Figure 3D-E). However, CD4⁺ T cell or NK cell depletion was less important in USP18-mediated tumorigenesis (Figure 3D-E). To further investigate the function of USP18 in IFN-γ-mediated immune surveillance against tumorigenesis, B16-OVA-GFP and B16-OVA-USP18 cells were intravenously injected into IFN-γ-knockout or wide-type C57BL/6 mice. Tumor nodule formation was similar when B16-OVA-GFP and B16-OVA-USP18 cells were inoculated (intravenously) into IFN-γ-knockout (KO) mice (Figure 3F-G). These results showed that regulation of tumorigenesis by USP18 depended on IFN-γ signaling, most likely through the IFN-γ-mediated immune response.

During tumor development, a variety of mechanisms inhibits the host immune response, including secretion of inhibitory cytokines and chemokines[16] and cell-cell contact suppression[17]. To explore USP18 function in immune suppression, we analyzed the immune status in the tumor microenvironment after USP18 overexpression in tumor cells (B16-OVA-USP18), which were then intravenously injected in C57BL/6 mice. CD45⁺ cell infiltration of the lung increased (Figure 4A). The percentage and the absolute number of NK cells also increased, but CD11b⁺ myeloid cell infiltration decreased (Figure 4A). Monitoring of SIINFEKL MHC class-I-restricted tetramer-positive CD8⁺ cells in the lung, spleen and draining lymph nodes revealed an increased antigen-specific CTL response in B16-OVA-USP18 tumor-bearing mice (Figure 4B).

Host immunosurveillance against tumorigenesis is not only determined by the potency of the antigen-specific immune response, but also by the persistence of immune effector cells[18]. Tumor cells have developed specific mechanisms to induce tolerance, inhibition, and exhaustion of immune effector cells, by altering the expression of PD-1 and KLRG1 on T cells[19, 20]. Analysis of PD-1 expression in CD8⁺ T cells from B16-OVA-USP18 tumor-bearing mice showed a reduced PD-1 expression in the blood and draining lymph nodes compared to B16-OVA-GFP tumor-bearing mice (Figure 4C).

We speculated that USP18 expression in tumor cells might lead to secretion of soluble factors that also contribute to the immune activation. Cytokine array analysis of culture supernatant from USP18 overexpression and knockdown B16-OVA tumor cells did not show significant differences (Additional file1: Figure S5A), but analysis of a chemokine panel showed that in B16-OVA-USP18 cells, increased secretion of chemokines related to NK cell migration was observed (Additional file1: Figure S5B). In regard to NK cells, not only did the ratio and absolute number of infiltrating NK cells increased in B16-OVA-USP18 tumor-bearing mice (Figure 4A), but the activation status of the infiltrating NK cells also increased, as evidenced by detection of the activation receptor NKG2D (Figure 4D).

ISG15 is an IFN-induced gene that is regulated by USP18[21]. In free form, ISG15 can activate NK cells and antigen presentation cells[22, 23], and a recent study showed that human ISG15 deficiency leads to mycobacterial disease[24]. To detect ISG15 in the tumor microenvironment, tumor lysate was prepared from B16-OVA-GFP, B16-OVA-USP18 or B16-OVA-shUSP18 tumors, and ELISA or western blot analysis showed that the ISG15 level was higher in B16-OVA-USP18 tumor than in B16-OVA-GFP tumor, but lower in B16-OVA-shUSP18 tumor (Figure 4E, Additional file1: Figure S6). We further investigated dendritic cell activity in the tumor microenvironment. CD11c⁺ dendritic cells were tolerized in the tumor stroma during tumor development[25]. Isolation of CD11c⁺ dendritic cells from B16-OVA-USP18 tumor bearing mice showed that these cells were activated instead of tolerized, because the expression of CD86 and MHC class-II was higher (Figure 4F), and more TNF-α but less IL-10 were secreted after re-stimulation in vitro as compared with B16-OVA-GFP tumor bearing mice (Figure 4G).

As USP18 expression in tumor cells affects CD8⁺ T-cell function in vivo, B16-OVA-GFP or B16-OVA-USP18 cells were irradiated and then cocultured with OT-1 T cells in vitro to analyze T-cell activation. B16-OVA-USP18 tumor-cell coculture significantly increased OT-1 T-cell proliferation (Figure 5A), but also increased OT-1 T-cell secretion of IL-2 and IFN-γ (Figure 5B). Moreover, secretion of IL-2 and IFN-γ was even greater when OT-1 T cells were cocultured with IFN-γ-sensitized B16-OVA-USP18 tumor cells as compared with B16-OVA-GFP cells (Figure 5C). Activated T cells express inhibitory molecules, such as PD-1, to prevent over-activation[26]. We found that ectopic expression of USP18 in B16-OVA cells significantly inhibited PD-1 expression on cocultured OT-1 T cells (Figure 5D). This phenomenon was also seen in T cells from B16-OVA-USP18 cell inoculated mice (Figure 4C). Moreover, ectopic USP18 expression in B16-OVA tumor cells significantly increased the CTL activity of activated OT-1 cells in vitro (Figure 5E). CD4⁺ T cells also played a role in tumor immunosurveillance. Irradiated B16-OVA-USP18 cells stimulated OT-2 T cells to proliferate (Figure 5F) and secrete more IL-2 and IFN-γ (Figure 5G) after the coculture. Furthermore, overexpression or knockdown USP18 in B16-OVA cells did not affect OVA antigen expression level in tumor cells (Additional file1: Figure S7), excluding the possibility that USP18 could regulate antigen expression.

As we found that endogenous IFN-γ signaling affected tumorigenesis through IFN-γ-induced USP18 expression in tumor cells and regulated immune-cell function in the tumor microenvironment, we wanted to explore whether USP18 expression in tumor cells affected exogenous IFN-γ-secreting CTL activity. Antigen-specific IFN-γ-secreting CTLs play an important role in antitumor immunity[27]. To analyze CTL activity in B16-OVA-shCon and B16-OVA-shUSP18 tumors, we intravenously injected 1 × 10⁶ B16-OVA-shCon cells or B16-OVA-shUSP18 cells into CD45.1⁺ C57BL/6 mice for 8 days, followed by adoptive transfer of 5 × 10⁶ activated OT-1 (CD45.2⁺) cells into tumor-bearing mice. Monitoring for persistence of CD45.2⁺ OT-1 cells in the lung, draining lymph nodes and spleen of tumor-bearing mice showed that OT-1 CTLs were rapidly exhausted in B16-OVA-shUSP18 tumor-bearing mice (Figure 6A). Tetramer (Figure 6B) and intracellular staining for IFN-γ and TNF-α (Figure 6C) demonstrated a loss of CTL activity in B16-OVA-shUSP18 tumor-bearing mice. Analysis of the T-cell exhaustion markers PD-1 and KLRG1 showed that the expression of these molecules was greater on OT-1 cells in B16-OVA-shUSP18 tumor-bearing mice than B16-OVA-shCon tumor-bearing mice (Figure 6D). These data indicated that USP18 expression in tumor cells regulated the IFN-γ-secreting activity and persistence of the CTLs in the tumor microenvironment.

Our study showed that USP18 expression in tumor cells regulated tumorigenesis and IFN-γ secretion by exogenous CTLs as part of the anti-tumor immune response. Next, we investigated the possibility that use of USP18 could enhance CTL activity against subcutaneous B16 melanoma. After B16-OVA-GFP, B16-OVA-USP18, or B16-OVA-shUSP18 tumor cells were subcutaneously inoculated into C57BL/6 mice, tumor growth in mice receiving B16-OVA-USP18 cells was significantly inhibited, whereas tumor growth was promoted in those receiving B16-OVA-shUSP18 tumor cells (Figure 2G). Next, we determined whether MHC class-I and PD-L1 expression differed in subcutaneous B16-OVA-USP18 tumor from that of B6-OVA-GFP tumor. Results showed that ectopic expression of USP18 increased MHC class-I and reduced PD-L1 expression (Figure 7A). As MHC class-I and PD-L1 expression on tumor cells significantly affected host immunosurveillance, we also detected greater OT-1 tetramer-positive CD8⁺ CTL infiltration in B16-OVA-USP18 tumor than those in B16-OVA-GFP tumor (Figure 7B). Moreover, more effector T cells were activated in B16-OVA-USP18 tumor and spleen compared with B16-OVA-GFP tumor and spleen (Additional file1: Figure S6A and B). However, there was no difference in infiltrating regulatory T cells between B16-OVA-USP18 and B16-OVA-GFP subcutaneous tumor (Additional file1: Figure S6C).

Next we explored whether delivery of USP18 into the tumor microenvironment enhances CTL activity against subcutaneous B16-OVA tumor. A subcutaneous B16-OVA tumor model was established in C57BL/6 mice. At day 5, activated OT-1 CTLs were adoptively transferred to the mice, together with intra-tumor injection of adenovirus vector expressing USP18 (Ad-USP18). Adoptive transfer of OT-1 CTLs induced partial tumor shrinkage (Figure 7C). Ad-USP18 injection increased OT-1 CTL activity and tumor shrinkage (Figure 7C). Detailed analysis of the immune status in tumor-bearing mice showed that Ad-USP18 significantly increased the duration of CTL persistence in the spleen and tumor sites of tumor-bearing mice (Figure 7D).

As tumor antigens are weakly immunogenic, dendritic cell vaccination and IL-2 treatment are usually used to increase CTL activity clinically[28]. Using our B16 tumor model, we co-administered pmel-1 CTLs, Ad-USP18, IL-2 and dendritic cells into tumor. Injection of Ad-USP18 significantly increased pmel-1 CTL activity and caused tumor shrinkage (Figure 7E).