Formalin-fixed paraffin-embedded (FFPE) samples of 23 OSCC patients who underwent surgical treatment from 2008 to 2013 in our university hospital (ASST dei Sette Laghi, Varese) were retrieved from the files of the Unit of Pathology. Included cases were reviewed by two expert pathologists according to the 8^th Edition of the American Joint Committee on Cancer (AJCC) Staging Manual and the 5^th Edition of the World Health Organization (WHO) Classification of Head and Neck Tumors (26, 27). All included cases were HPV-negative, conventional-type OSCC and provided a sufficient number of tissue slides to perform immunohistochemistry (IHC) reactions in the current study. The protocol was approved by the Institutional Ethical Committee (Comitato Etico dell'Insubria, protocol n° 19/2018).

The following clinical data were retrospectively collected for each included patient: age, sex, cervical lymph node metastasis (LN metastasis) and stage at diagnosis, and 5-year cancer-related death (CRD).

Firstly, we evaluated the degree of tumor lymphoid infiltration in hematoxylin-eosin staining slides by counting the number of lymphocytes infiltrating the tumor epithelial nests per High Power Field (HPF), and a score from 0 to 2 was defined for each case, with score 0 = no or few cells (i.e., less than 5 cells per HPF); 1 = low-moderate number of cells (between 5 and 20 cells per HPF); and 2 = high number of cells (i.e., 20 cells or more per HPF). The immune infiltration was also evaluated in the tumor stroma among epithelial nests and along tumor growth front, and the stromal area occupied by immune cells was scored from 0 to 2, with score 0 = none or low infiltration (less than 1% of stromal area); 1 = moderate infiltration (from 1 to 10%); and 2 = high infiltration (at least 10%).

The presence of T lymphocytes and monocytic-macrophage cells in both tumor epithelial nests and stroma was evaluated with appropriate staining with antibodies for specific markers: anti-CD4⁺ for TH (Ventana, rabbit, clone SP35), anti-CD8⁺ for CTL (Ventana, rabbit, clone SP57), and anti-CD68⁺ for monocyte-macrophages (DAKO, mouse, clone 1G12) (28).

Consistently with our aim, the phenotypic characterization of MHC-I and MHC-II expression was assessed on tumor cell surface and compared with the patients' clinical outcomes and the immune infiltration. In the same manner, the immune checkpoints PD-1 and PD-L1 were measured in tumor and stromal compartments. Of note, PD-L1 expression was measured both by the internationally recognized Combined Positive Score (CPS) and, separately, on immune cells and, when present, on tumor cells by the Tumor Proportion Score (TPS) (29).

IHC stainings were performed on 3 µm FFPE sections deparaffinized and rehydrated through alcohol series to water, as previously described (30). Briefly, endogenous activity was blocked with 3% aqueous hydrogen peroxide for 10 min, antigen retrieval was performed for each antigen in a domestic 750 kW microwave oven with different solutions (EDTA or citrate buffer pH 6.0). Primary antibodies ( Supplementary Table S1 ) were applied overnight at 4°C followed by a polymeric detection system (Ultravision DAB Detection System, LabVision, Värmdö, Sweden) according to the manufacturer's instructions. The immunoreaction was developed with 3.3'-diaminobenzidine tetrahydrochloride (DAB) (Sigma Aldrich, St. Louis, MO, USA) as chromogen.

For each case of OSCC, the MHC-I, MHC-II, and PD-L1 expression was evaluated on tumor cells and scored as summarized in Supplementary Table S2 . Briefly, MHC-I expression was scored as low (< 50% of immunoreactive tumor cells), moderate (immunoreactive cells ≥ 50% and < 90%), and high expression (immunoreactive cells ≥ 90%); MHC-II was scored as expressed when positive tumor cells were at least 10%; PD-L1 Tumor Proportion Score (TPS) was considered positive when immunoreactivity was present in at least 10% of tumor cells; PD-L1 Combined Positive Score (CPS) was considered positive when ≥ 1 and scored as highly expressed when it was at least 10. CD4⁺, CD8⁺, CD68⁺, and PD-1 immunoreactivities in tumor epithelial nests and stroma were categorized into low or high expression as reported in Supplementary Table S2 .

Three human OSCC cell lines, CAL-27, SCC-25, and SCC-4, were used for the characterization of the cell surface immune phenotype by immunofluorescence and cytofluorimetry. The three cell lines represent epithelial, adherent squamous cell carcinoma cells obtained from patients affected by tongue cancer (31, 32). CAL-27 cell line was grown in ATCC-formulated Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 10% fetal bovine serum (FBS). SCC-25 and SCC-4 cell lines were grown in 1:1 mixture of DMEM and Ham's F12 medium containing 1.2 g/L sodium bicarbonate, 2.5 mM L-glutamine, 15 mM HEPES and 0.5 mM sodium pyruvate (DMEM:F12), supplemented with 400 ng/mL hydrocortisone and 10% FBS. SCC-4 cells were grown on ATCC 56-X.2™ feeder layer, MITC-STO cells, and the feeder cells (murine fibroblasts) were plated 24 hours before use at 2x10⁶/T75 in order to obtain a 30% confluent monolayer.

The mouse OSCC cell line MOC2 was used for the animal model of OSCC described in this study (33).

This cell line shows an aggressive growth phenotype (i.e., ability to form tumors with injection of as few as 10,000 cells) and is derived from a chemokine receptor CXCR3-deficient mouse on a pure C57BL/6 background (34). The MOC2 cell line was cultured in specific medium, following manufacturer's instructions. Briefly, for each Liter of medium, a 2:1 mixture of 626 mL Iscove's Modified Dulbecco's Medium (IMDM) and 313 mL Ham's nutrient mixture F10-F12, supplemented with 50 mL FBS and 10 mL Penicillin-Streptomycin was filtered in 1L filter flask. Finally, 1 mL of 5mg/mL Insulin, 800 μL of hydrocortisone solution, and 5 μL of Epidermal Growth Factor were added to the medium.

Detailed references of cell lines and reagents used in this study are reported in Supplementary Table S3 .

The cell surface expression of MHC-I, MHC-II, PD-1 and PD-L1 molecules was assessed by immunofluorescence and flow cytometry (BD FACSAria™ II Cell Sorter). Briefly, cells were washed twice with PBS, dissociated with trypsin-EDTA, and resuspended in complete medium. Cells were pelleted at 800 x g for 5 min, washed, the supernatant was discarded, and the pellet resuspended in PBS for FACS analysis. The following monoclonal antibodies were used as primary antibodies for the immune phenotype characterization of human OSCC cell lines (CAL-27, SCC-25, SCC-4): B9.12.1 anti-MHC-I (35) and D1–12 anti-MHC-II (DR) (36) used as hybridoma supernatants, followed by goat anti-mouse FITC, CD279 anti-PD-1 (clone EH12.2H7) PE, and CD274 anti-PD-L1 (clone 29E.2A3) PE, as previously described ( Supplementary Table S4 ) (37, 38).

3x10⁵ OSCC cells were plated in 6 multi-well plates and treated with 500 U/mL of IFN-γ or with its vehicle. Seventy-two hours after treatment, the cells were collected and analyzed by immunofluorescence and flow cytometry, as indicated above. As positive and negative controls, RA and HepG2 cell lines were used, respectively. RA is a glioblastoma cell line that expresses MHC-I but not MHC-II molecules on cell surface. However, after in vitro stimulation with IFN-γ, cell surface MHC-II expression is induced, as described in previous experiments (39). Conversely, HepG2 is a hepatocarcinoma cell line whose MHC-II expression on cell surface remains negative after IFN-γ treatment, due to hypermethylation of the CIITA promoter IV (CIITA pIV) region, as previously demonstrated (38).

The cell surface expression of MHC-I (H2 MHC class I), MHC-II (I-A/I-E), PD-1 and PD-L1 molecules was assessed in murine MOC2 OSCC cell line by immunofluorescence and flow cytometry at baseline and 72 hours after IFN-γ treatment (1,000 U/mL), as previously described for human OSCC cell lines. The following antibodies were used for analysis: M1/42 anti-H2 MHC class I FITC, M5/114.15.2 anti-I-A/I-E PerCP Cyanine5.5, CD279 anti-PD1 (clone J43) PE, and CD274 anti-PD-L1 (clone MIH5) PE ( Supplementary Table S4 ).

The data were analyzed by using FlowJo 9.5.2 software.

To test the efficacy of the MHC-II-based vaccination strategy in OSCC, we performed an in vivo experiment in a syngeneic mouse model (C57BL/6, H-2^b genotype, Charles River Laboratories Italia srl, Calco, Italy) (25). Briefly, we generated by genetic transfer stable MHC-II-expressing MOC2 tumor cells in vitro and injected them subcutaneously (s.c.) in mice to assess tumor rejection and/or tumor growth retardation compared with parental tumor cells. Following analyses included adoptive cell transfer and in vivo cell depletion experiments to better describe the role of CD4⁺ and CD8⁺ T lymphocytes in the immune response elicited against tumors.

Each experiment was repeated at least twice using five to eight mice per group.

All animal experiments were conducted according to relevant national and international guidelines and were approved by the local Animal Welfare Ethical Committee (OPBA) and by the Italian Ministry of Health (protocol n° 249/2022-PR).

CIITA cDNA was excised with EcoRI digestion from pcf-CIITA1–1130 and cloned by ligation into EcoRI-cleaved pAIP vector (Addgene), as previously described (40).

MOC2 tumor cells were transfected with 3 μg of flag-CIITA (pAIP-fCIITA) expression vector or with pAIP empty vector (mock) by using FuGENE® HD Transfection kit (Promega, cat n° E2311), as previously described (41).

Both mock and CIITA-transfected MOC2 cells underwent puromycin selection at the concentration of 1.5 μg (Sigma Aldrich, cat n° P9620). For CIITA-transfected MOC2 cells, MHC-II-positive cells were enriched by fluorescence-activated cell sorting with a BD FACSAria™ II cell sorter and subjected to limiting-dilution cloning. The expression of the pAIP empty vector in MOC2-mock cells was assessed by RT-PCR by using the following primers to amplify Puromycin resistance cassette: forward, 5'-gcaacctccccttctacgagc-3'; reverse, 5'-gtgggcttgtactcggtcat-3'.

Syngeneic C57BL/6 female mice (H-2^b) 7–8 weeks old were injected s.c. with 1x10⁵ cells/mL of either MOC2 CIITA-transfected tumor cells (MOC2-CIITA) or MOC2 parental tumor cells (MOC2-pc) or pAIP-transfected MOC2 cells (MOC2-mock). The expression of CIITA-driven MHC-II molecules was confirmed the day of the injection by immunofluorescence and flow cytometry, as described above.

Tumor growth along with the overall health condition of the mice were checked at least twice a week. The tumors were measured weekly using a caliper and registered in mm².

The mice that did not show any tumor growth after injection with MOC2-CIITA were challenged with a s.c. injection of 1x10⁵ MOC2-pc and the tumor size was measured as above.

Each experiment was repeated at least twice using 5–8 mice per group.

Primary subcutaneous OSCC tumors were isolated and collected in DMEM 4 weeks after implantation. Mouse tumors were then diced into pieces measuring 2–4 mm. Tissue suspension was further broken down with mouse tumor dissociation kit (Miltenyi, cat n° 130–096-730) and gentleMACS™ Dissociator (Miltenyi, cat n° 130–093-235), following the manufacturer instructions.

Spleens from challenged vaccinated mice that did not show parental tumor growth after 4 weeks from injection were harvested and processed. A single cell suspension was obtained and used to purify CD4⁺ and CD8⁺ T cells by CD4⁺ and CD8a⁺ Mouse T Cell Isolation Kit (Miltenyi Biotec GmbH, Cat n° 130–095-248/130–095-236, respectively). The purity of the selected cells was confirmed by flow cytometry using 145–2C11 anti-CD3e⁺ (BD, cat n° 553066), RM4–5 anti-CD4⁺ (BD, cat n° 550954), and 53–6.7 anti-CD8a⁺ (BioLegend, cat n° 100711) antibodies (25). At the same time, normal splenocytes were obtained from naïve C57BL/6 female mice of the same age.

Naïve female mice, 7–8 weeks old, were s.c. co-injected with 1x10⁵ MOC2-pc in a 100 μL volume of IMDM without FBS, along with either 1x10⁷ total immune splenocytes, CD4⁺ T cells, CD8⁺ T cells, or total naïve splenocytes as a control, in a tumor:immune cells ratios of 1:50, 1:15, 1:10, and 1:50, respectively.

The experiment was repeated at least twice using 5–8 mice per group. The mice under experimentation were followed for 4 weeks and the data were recorded as described before.

Eight-week old mice were injected intraperitoneally (i.p.) -9, -7, -5, -2, -1 days prior to tumor injection (day 0) and +2, +5, +7 days after, with 100 μL/injection of anti-CD4⁺ (aCD4 GK 1.5) or anti-CD8⁺ (aCD8 2.43.5) antibodies, or with control rat polyclonal Ig (24). For CD4⁺ depletion we performed an additional injection at day -12. To verify the efficacy of the procedure before tumor injection, CD4⁺ and CD8⁺ T cells depletion was tested by immunofluorescence and flow cytometry on splenocytes derived from treated mice, incubated with anti-mouse CD4⁺ and anti-mouse CD8⁺ antibodies, and compared with splenocytes isolated from untreated mice.

Treated and untreated mice were injected s.c. with either 1x10⁵ MOC2-CIITA or MOC2-pc at day 0. The size of the tumor was measured weekly, as described before.

The hypothesis of independence between categorical variables was tested by Chi-square test or Mann-Whitney test. Variables were considered dependent when p ≤ 0.05.

The other analyses were performed using GraphPad Prism 10 (GraphPad Software, http://www.graphpad.com), and the multiple unpaired Student's t-test was conducted to determine significance.

Comparisons between groups were considered statistically significant when the corresponding p-value was ≤ 0.05.

The 23 patients affected by OSCC were 13 males and 10 females, with a mean age of 72.52 ± 12 years, without differences between the sexes (p=0.217). All the included samples were HPV-negative, conventional-type OSCC ( Table 1A ).

Based on AJCC and WHO updated diagnostic criteria, the oral cancer samples analyzed were at different stage of tumor progression. Seven patients showed LN metastasis at diagnosis, while six died within 5 years for cancer-related disease (CRD). Among deceased patients, three did not show LN metastasis at diagnosis and developed recurrent disease after primary treatment.

Based on the presence of immune cellular infiltrate in tumor epithelial islands and stroma, tumors were classified as 0, 1+, and 2+ (see Materials and Methods). The degree of cellular infiltration in tumor epithelial nests did not correlate with either LN metastasis at diagnosis (p=0.530) or 5-year CRD (p=0.343) ( Table 1B ). A similar trend was observed in relation to the degree of inflammation in the stroma (LN metastasis p=0.533; 5-year CRD p=0.666).

The expression of both MHC-I and MHC-II molecules was assessed on OSCC tumor tissues by IHC, as stated in Materials and Methods.

MHC-I antigens were detectable on tumor cells in all 22 analyzed samples; one sample was not evaluable due to exhaustion of tumor area. However, the expression of MHC-I staining was variable, with 7, 7, and 8 patients showing low, moderate, and high degree of expression, respectively ( Table 2 ). The level of MHC-I expression did not statistically correlate with clinical outcomes, even though low MHC-I expression levels were associated with a higher incidence of LN metastasis at diagnosis and more advanced stage of disease. Indeed, in patients with stage IV oral cancer MHC-I molecules were significantly hypoexpressed on tumor cells (p=0.025, Table 2 ).

MHC-II-positive tumor cells were found in 43.5% of the recruited patients (10 out of 23), particularly in females (p=0.024, Table 2 ). As observed for MHC-I, MHC-II expression level on tumor cells did not correlate with either LN metastasis at diagnosis (p=0.074) or 5-year CRD (p=0.708, Table 2 ).

We then assessed whether a correlation existed between MHC molecules expression on tumor cells and the presence of specific immune cell infiltration in tumor epithelial nests and stroma ( Table 3 ). We did not find statistically significant correlations between MHC-I molecule expression and tumor epithelial nest infiltration, considering both total immune infiltrate (p=0.263) and specific cell subpopulations (CD4⁺ p=0.325; CD8⁺ p=0.970; CD68⁺ p=0.137). Similar findings were observed in the stroma (total immune cells p=0.575; CD4⁺ p=0.911; CD8⁺ p=0.537; CD68⁺ p=0.911) ( Table 3 ).

Interestingly, although MHC-II expression did not statistically correlate with the total immune cell infiltrate in tumor islands (p=0.097), it strongly correlated with a high presence of both CD4⁺ TH (p=0.047) and CD8⁺ CTL (p=0.012), but not with CD68⁺ cells (p=0.099) ( Table 3 , Figure 1 ). As for the stroma, MHC-II expression on tumor cells was associated with the presence of high total immune cells (p=0.026) and high CD4⁺ TH (p=0.026), but not with CD8⁺ CTL (p=0.072) and CD68⁺ cells (p=0.062) ( Table 3 ).

To further characterize the expression of immune markers of relevance for immunotherapy approaches, we analyzed the presence of immune checkpoint PD-1 and its major ligand PD-L1.

PD-1 was expressed on the surface of immune cells intermingled among tumor epithelial cells and infiltrating the stroma. On the other hand, PD-L1 was present on both immune and tumor cells. In particular, 11/23 (i.e., 47.8%) tumors showed PD-L1 expression on tumor cells (TPS mean: 49.09 ± 24.06%) whereas CPS score, which includes PD-L1 expression both in tumor and immune cells, was positive in almost all cases (i.e., 19/20), with high expression detected in 13/19 patients (CPS mean: 40.3 ± 34.84).

We next investigated a possible correlation between the expression of the immune checkpoints on immune cells (PD-1) and tumor cells (PD-L1) and the clinical outcomes ( Table 4A ). The expression of these molecules did not correlate with clinical outcomes, except for PD-1-positive immune cells in tumor nests that correlated with the presence of LN metastasis at diagnosis (p=0.015) ( Table 4A ).

We next verified whether the expression of MHC-I and/or MHC-II molecules on tumor cells was associated to the presence of immune checkpoints ( Table 4B ). While there was a higher number of PD-1-positive immune cells in the stroma of MHC-II-positive tumors (p=0.029), this correlation did not exist with MHC-I (p=0.854).

Similarly, the CPS was higher in MHC-II-positive tumors (p=0.033) and was independent from MHC-I antigen expression (p=0.918). At variance with PD-1, PD-L1 expression on stromal cells did not correlate either with MHC-I (p=0.970) or MHC-II molecules (p=0.580). Interestingly, however, most of the PD-L1-positive tumors expressed a high level of MHC-II molecules (p < 0.001) ( Table 4B , Figure 2 ).

Finally, we correlated the expression of both PD-1 and PD-L1 and the different immune cell subpopulations infiltrating the tumor epithelial nests or present in the stroma ( Table 4C ). High detection of PD-1 in immune cells infiltrating tumor epithelial islands correlated with the presence of high expression of CD8⁺ CTL (p=0.003), but not with CD4⁺ TH and CD68+ cells (p=0.604 and p=0.907, respectively). Similar findings were observed for the stroma (CD8⁺ CTL p=0.05; CD4⁺ TH p=0.472; CD68⁺ cells p=0.591) ( Table 4C ).

A statistically significant correlation between PD-L1 expression and high total immune cells infiltrating the stroma (p=0.029) was observed, although it was not related to specific T cell subpopulation ( Table 4C ).

In order to investigate similarities or differences in the cell surface phenotype between OSCC in tumor tissues and in isolated OSCC cell lines, we studied three human OSCC cell lines, namely CAL-27, SCC-25, and SCC-4 ( Figure 3 ).

As far as the MHC cell surface expression, we found that MHC-I molecules were expressed in all three cell lines, although at different levels, mimicking what we observed on tumor tissues. In particular, the MHC-I expression detected on CAL-27 cell line was significantly lower when compared to that of SCC-25 and SCC-4. On the contrary, MHC-II molecules were slightly expressed only on the surface of CAL-27 cells ( Figure 3 ). Again, this result resembled what we observed in human tissues, with a subgroup of patients showing MHC-II-positive tumors.

The expression of MHC genes could be rescued/increased by treatment with the inflammatory cytokine IFN-γ. As far as MHC-I expression, it increased to reach similar levels in all three cell lines after treatment with the cytokine. This was particularly relevant for CAL-27 ( Figure 3 , bold line). As far as MHC-II, it should be underlined that the IFN-γ-dependent rescuing of MHC-II gene expression is mediated by a direct action on the expression of CIITA, the MHC class II transactivator, which in turn activates the expression of MHC-II genes (20). MHC-II expression could be rescued in all three cell lines although to different extent ( Figure 3 , bold line), and this correlated to de novo CIITA expression ( Supplementary Figure S1 ).

We next investigated the expression of PD-1 and PD-L1 in the three OSCC cell lines before and after treatment with IFN-γ. As expected, the PD-1 molecule, which is usually expressed by immune cells, was not detected on the cell surface of the OSCC cell lines, neither before nor after treatment with IFN-γ ( Figure 3 ).

In contrast, as we observed in histopathological samples, PD-L1 was differently expressed by the three OSCC cell lines, with CAL-27 and SCC-4 clearly expressing the marker and SCC-25 not in normal cell culture condition. After treatment with IFN-γ, PD-L1 expression was rescued in SCC-25 and upregulated in the other two OSCC cell lines ( Figure 3 , bold line).

Our previous investigations have established the strong potential of an immune vaccination against tumors by using cellular tumor models of various histotype in mice (24, 25, 41). The rational was to render tumor cells MHC-II-positive by genetic transfer of CIITA and then assess the in vivo behavior of CIITA-tumor cells. Given the importance of oral squamous cell carcinoma both as incidence and poor therapeutic armamentarium in human, and the rather scanty availability of experimental systems to assess innovative therapies, we wanted to test whether murine OSCC tumor cells genetically modified to express constitutively MHC-II molecules could be recognized and rejected and/or delayed in their growth when injected into syngeneic immunocompetent recipients.

The murine OSCC cell line MOC2 of H-2^b genetic background was first characterized for the cell surface expression of MHC-I and MHC-II, as well as PD-1 and PD-L1. MHC-I antigens were expressed on MOC2 cell surface, while MHC-II molecules were not detectable ( Figure 4 , MOC2-pc). However, as observed in OSCC human cell lines, IFN-γ treatment upregulated MHC-I expression and induced MHC-II molecules on the cell surface of MOC2 cell line ( Figure 4 , MOC2-pc, bold line). The expression of MHC-II molecules was driven by the induction of CIITA gene transcription upon IFN-γ treatment ( Supplementary Figure S2 ).

As far as the immune checkpoints, PD-1 was not detected either before or after treatment with the cytokine. Differently from what we could observed in OSCC human cell lines, MOC2 cell line did not express PD-L1 molecule ( Figure 4 , MOC2-pc). However, expression of PD-L1 was clearly induced after treatment with IFN-γ, consistent with what we observed in human cell lines ( Figure 4 , MOC2-pc, bold line).

MOC2 cell line was selected to be transfected with CIITA and used in in vivo experiments. A transfectant stably expressing CIITA-driven MHC-II cell surface molecules (MOC2-CIITA) was selected by cell sorting. Figure 4 (MOC2-CIITA) shows the MHC-II cell surface expression of MOC2-CIITA compared to the parental untransfected control (MOC2-pc). Of note, MOC2-CIITA transfected cell line did not express PD-L1, differently from results obtained in parental cell line after treatment with IFN-γ ( Figure 4 , MOC2-pc, bold line). It should be emphasized that stable expression of CIITA in MOC2 tumor cells did not affect their growth rate in vitro ( Supplementary Figure S3 ).

MOC2-CIITA cells were s.c. injected into naïve syngeneic C57BL/6 mice and tumor growth was monitored over time, as previously described. At 4 weeks after tumor injection, MOC2-CIITA tumors were rejected in 40% of mice with respect to their MOC2-pc counterpart ( Figure 5A ). Importantly, in the remaining 60% of MOC2-CIITA injected mice the kinetics of tumor growth was strongly delayed and after four weeks the average tumor size was seven-fold lower (p<0.0001) compared to parental MOC2 tumors ( Figure 5B ). Notably, similar findings were obtained when comparing the tumor growth of MOC2-CIITA cells with MOC2 cells stably transfected with pAIP empty vector (MOC2-mock). The expression of pAIP empty vector was assessed by RT-PCR for the presence of Puromycin cassette in MOC2-mock cells compared to MOC2 parental cells ( Supplementary Figure S4A , compare lane 2 and 3). Indeed, 50% of MOC2-CIITA injected mice rejected the tumor ( Supplementary Figure S4B ) and the remaining 50% showed a significant delay in tumor growth compared to MOC2-mock tumor-bearing mice ( Supplementary Figure S4C ). At four weeks the MOC2-CIITA tumor size was seven-fold lower compared to MOC2-mock tumors (p<0.0001).

To investigate whether the rejection of CIITA-tumors in mice was attributable to the generation of an adaptive immune response that could protect the animals from a challenge with untransfected parental (pc) tumors, MOC2-CIITA tumor-free mice were challenged with MOC2-pc. Results clearly showed that animals immunized and protected from CIITA-tumors fully rejected, in 100% of the cases, the untransfected MHC-II-negative parental tumors ( Figure 5C ).

Remarkably, in mice in which MOC2-CIITA tumors were delayed in their growth in vivo, the expression of MHC-II molecules was significantly reduced compared to that of MOC2-CIITA cells the day of injection, as demonstrated by FACS analysis of tumor cells isolated from MOC2-CIITA tumors explanted at 4 weeks after injection ( Figure 6 ).

From these results, we conclude that MOC2 tumor cells of aggressive behavior in vivo not only can be rejected or strongly delayed in their growth after de novo expression of CIITA-mediated MHC-II molecules, but also induce an immunological memory capable to confer resistance to challenge with MHC-II-negative parental tumors.

To assess whether protection from tumor growth after vaccination with MOC2-CIITA was associated to a change in the phenotype of T cells with respect to tumor-bearing mice, we have analyzed the spleen cells from animals bearing the MOC2 parental tumor as compared to animals protected by vaccination with MOC2-CIITA. Supplementary Figure S5 shows that MOC2 tumor-bearing mice display a preferential TH2 phenotype, as demonstrated by an increased secretion of IL-4, particularly when stimulated in vitro with MOC2 parental or MOC2-CIITA cells, while MOC2-CIITA vaccinated mice express a preferential TH1 phenotype when stimulated in vitro with either parental MOC2 or, more importantly, with MOC2-CIITA cells. These results suggest a preferential triggering of TH1 responses following vaccination with CIITA-modified tumor cells.

To define more precisely the contribution of the immune cell subpopulations involved in the triggering and maintenance of the protective antitumor response after vaccination with MOC2-CIITA cells, experiments of adoptive cell transfer (ACT) of immune splenocytes were performed ( Figure 7 ). Total spleen cells, purified CD4⁺ and CD8⁺ T cells derived from mice that had rejected MOC2-CIITA and further rejected MOC2-pc after challenge, were selectively co-injected with MOC2-pc into naïve C57BL/6 mice. The ratio of tumor cells versus total splenocytes, CD4⁺ and CD8⁺ T cells was 1:50, 1:15, and 1:10 respectively, to mimic the relative proportion of the various lymphocyte subpopulations of the spleen. Tumor growth in the distinct groups of mice was followed for 4 weeks.

Importantly, immune CD4⁺ T cells were able to strongly reduce MOC2-pc tumor growth in 89% of the animals and confer tumor protection in the remaining 11% of mice ( Figure 7 ). Virtually superimposable results were obtained by ACT with immune CD8⁺ T cells. Interestingly, the protective effect of immune CD4⁺ and CD8⁺ T cells was comparable to the effect obtained with total immune spleen cells, further reiterating the importance of both T cell subpopulations in the anti-tumor response ( Figure 7 ).

To provide further support to the role of T cells in protecting the mice from MOC2 tumor growth in vivo, we induced depletion of either CD4⁺ or CD8⁺ T cells by treating naïve mice with specific monoclonal antibodies, as described in Materials and Methods. As control, naïve mice were treated with irrelevant, isotype-matched monoclonal antibodies. Depletion of the relevant T cell subpopulation was measured by immunofluorescence and FACS analysis in spleen cells of sample mice ( Supplementary Figure S6 ). It should be noted that while the depletion of CD8⁺ T lymphocytes was complete, treatment with the specific antibody resulted in an approximate 50% reduction of CD4⁺ T lymphocytes ( Supplementary Figure S6 ). Treated mice were then injected with the same number of MOC2-CIITA tumor cells capable to induce protection and/or strong retardation in tumor growth, as described above (see Figure 5 ). Interestingly, while depletion with control antibody did not affect the protective immune response of the mice injected with MOC2-CIITA, depletion of CD4⁺ T cells and CD8⁺ T cells drastically reduced the immunogenic and protective effect against MOC2-CIITA tumors. Indeed, at four weeks post-tumor injection, mice depleted of CD4⁺ or CD8⁺ cells developed tumors, unlike MOC2-CIITA-injected mice, where 25% of the animals remained tumor-free ( Figure 8A ). The mean tumor size was 3.8-fold and 10-fold larger respectively as compared to that of control group (CD4⁺-depleted p<0.0001, CD8+-depleted p<0.000001) ( Figure 8B ). The less marked loss of protection observed after depletion of CD4⁺ T cells as compared to depletion of CD8⁺ T cells is most likely due to the residual presence of CD4⁺ T cells after the depletion treatment with the anti-CD4⁺ monoclonal antibody used for this experiment ( Supplementary Figure S6 ).