In the absence of a local inflammatory response, expression of MHC class II molecules is restricted mainly to hematopoietic cells and thymus epithelium. However, certain tumors, such as melanoma, may acquire aberrant constitutive expression of MHC class II. In a set of primary melanoma cell populations and correspondingly expanded autologous tumor-infiltrating lymphocytes (TIL), we show how MHC class II expression on melanoma cells associates with strong MHC class II–restricted CD4+ T-cell responses that are specific for tumors. Notably, we found that tumor-specific CD4+ T-cell responses were dominated by TNF production. TNF reduced CD8+ T-cell activation in IFNγ-rich environments resembling a tumor site. Conversely, direct CD4+ T-cell responses had no influence on either the proliferation or viability of melanoma cells. Taken together, our results illustrate a novel immune escape mechanism that can be activated by aberrant expression of MHC class II molecules, which by attracting tumor-specific CD4+ T cells elicit a local inflammatory response dominated by TNF that, in turn, inhibits cytotoxic CD8+ T-cell responses Cancer Res; 75(18); 3747–59. ©2015 AACR.

In the absence of a local inflammatory response, MHC class II expression is mainly restricted to hematopoietic cells and thymus epithelium (1–3). However, certain types of solid tumors, including a subset of melanomas, de novo constitutively express MHC class II molecules (1, 3). Furthermore, MHC class II can be induced in several cell types, including tumor cells, by exposure to cytokines such as IFNγ (1, 4).

MHC class II expression in melanoma has been previously associated with both shorter and longer survival (5–7). Indeed, MHC class II expression can make these tumors directly detectable by tumor-antigen-specific CD4+ T cells that we and others previously have shown to be capable of generating Th1 responses in response to autologous tumor antigens (8–11). In contrast, recent studies proposed that engagement of MHC class II by lymphocyte-activation gene 3 (LAG3), expressed respectively on tumor cells and tumor-infiltrating immune cells, may trigger prosurvival signals and tumor cell resistance to apoptosis (12). In addition, LAG3 has been characterized as an immune inhibitory receptor, and its engagement on T cells may mediate downregulation of immune responses in the tumor microenvironment during priming but not effector phase (13, 14).

In order to clarify the association of MHC class II expression with CD4+ T-cell responses and CD4+ T-cell functional patterns in melanoma, we conducted a multifunctional analysis of expanded tumor-infiltrating lymphocyte (TIL) direct recognition of a panel of 38 autologous-matched melanoma cell lines generated from individual patients with advanced disease. Our results shed light on a novel mechanism of inflammatory CD4+ T-cell attraction to the tumor microenvironment by primary MHC class II expression on melanomas, which dampen CD8+ T-cell responses via TNF-induced counteraction of IFNγ-mediated local amplification of immune responses.

Patients and samples

All the procedures were approved by the Scientific Ethics Committee for the Capital Region of Denmark. Written informed consent was obtained from patients before any procedure according to the Declaration of Helsinki. All patients were diagnosed with histologically confirmed advanced melanoma, AJCC stage IV (n = 32) or IIIB (n = 1, completely resected), IIIC (n = 5, two of those were completely resected).

Tumor cell lines from primary melanomas WM-115, FM-55-P, FM-55-M1, and FM-55-M2 were obtained from ESTDAB (http://www.ebi.ac.uk/ipd/estdab/). The characteristics of WM-115, WM-75, WM-793, and WM-266-4 have been described elsewhere (15).

Antibodies for flow cytometry

Antibody panels used for flow cytometry were:

  • Multifunctional characterization of CD4+ and CD8+ T-cell responses: CD4 QDOT705 (Life Technologies), CD8 QDOT 605 (Life Technologies), Live/Dead Fixable Dead Cell Stain Near-IR (Life Technologies), MIP-1α FITC (eBioscience), MIP-1β PerCP-eFluor 710 (eBioscience), CD107a Brilliant Violet 421, IFNγ PE-Cy7, TNF-APC, IL2 Brilliant Violet 650 (BioLegend), IL17A Brilliant Violet 510.

  • All other tumor-T cell coculture experiments: CD4 FITC, CD8 PerCP, Fixable Viability Dye eFluor 450 (eBioscience), IFNγ PE-Cy7, TNF-APC, CD107a PE.

  • MHC class I characterization: anti-HLA-ABC APC, 7-AAD

  • MHC Class II characterization: anti-HLA-DP, DR, DQ FITC, 7-AAD

Generation of TILs and melanoma cell lines

TILs used in this study were generated with a protocol extensively described in other studies (16). Briefly, TILs were initially isolated and minimally expanded in high doses of IL2 (6,000 IU/mL IL2; Proleukin from Novartis) from surgically resected melanoma tumors, which were cut in 1- to 2-mm3 fragments under sterile conditions. When a minimum of 50 × 106 TILs were obtained (typically about 14–28 days after surgical resection—the product at this stage is named “minimally cultured TILs”; ref. 17), expansion was further achieved by a standard 14-days rapid expansion protocol (REP), in which TILs are unspecifically expanded with a 200-fold excess of allogeneic irradiated peripheral blood mononuclear cells (PBMC) from at least three different healthy donors, 30 ng/mL anti-CD3 antibodies (OKT3, from Janssen-Cilag or Miltenyi Biotec). All through this article, this TIL product is named “expanded TILs” or “REP-TILs.” From some patients, “uncultured (or fresh) TILs” were obtained from tumor fragments that were digested overnight in the presence of 1 mg/mL collagenase type IV (Sigma-Aldrich) and 0.0125 mg/mL dornase alpha (Pulmozyme, Roche) and immediately cryopreserved.

Pure CD8+ or CD4+ T-cell cultures—for the indicated experiments in which sorted T-cell subpopulations were used—were generated by positive magnetical selection, respectively, with CD8 or CD4 microbeads (Miltenyi Biotec), according to the manufacturer's instructions, from unselected TILs before REP. Subsequently, sorted subpopulations were expanded with REPs separately. Only cultures with frequency of over 95% of either CD8+ or CD4+ T cells were used for the indicated experiments.

Autologous melanoma cell lines were generated separately from TILs either from tumor fragments or from a combination of cells recovered from suspension in transport medium or after mincing, as previously described (16, 18).

Analysis of T-cell responses by flow cytometry

Evaluation of T-cell responses was performed as previously described (10, 16).

Briefly, TILs were thawed and rested for 3 days (for multifunctional characterization of REP TILs), 2 days (minimally cultured TILs), or overnight (in all other cases) in RPMI-1640 (Life Technologies) supplemented with 10% AB human serum (Sigma-Aldrich), thereafter washed twice and cocultured with autologous short-term cultured melanoma cell lines. Tumor reactivity was evaluated by assessing the amount of T cells previously gated as CD4 or CD8 T cells expressing cytokines (IFNγ and TNF) or CD107a.

For screening of CD8+ and CD4+ T-cell responses to autologous tumor antigens, autologous tumors in log-phase of growth were incubated with 100 IU/mL IFNγ (Imukin, Boehringer-Ingelheim) for 72 hours and thereafter extensively washed and added to the cocultures. This was done to maximize the possibility to detect antitumor responses of low frequency to autologous tumors with MHC downregulation or with no constitutive MHC class II expression. A response was defined as the presence of a minimum of 0.5% responding cells (expressing at least one of the following T-cell functions: TNF, IFNγ, or CD107a) in the relative CD8+ or CD4+ T-cell subpopulation, with a minimum of 50 positive events acquired and a minimum of a T-cell function positive cell frequency of three times the background (i.e., unstimulated samples). The frequency of tumor-reactive cells in stimulated samples was subtracted from unstimulated samples. 0.5% was used as limit of significance.

For MHC class II blocking experiments, tumor cells were incubated for 30 minutes at 37°C with 20 μg/mL of anti-HLA DR, DP, DQ antibody (clone TÜ39, from BioLegend) or with relevant isotype control. Thereafter, tumor cells were added to cocultures with TILs without additional washing (final concentration in the T-cell stimulation cocktail was around 0.5 μg/mL).

Combined tetramer (peptide–MHC multimer PE conjugated and produced in-house, HLA-A2–restricted MART-1/Melan-A–derived peptide ELAGIGILTV) and intracellular cytokine staining (CD107a PE was in this case switched with CD107a FITC) was performed to assess the frequency of MART-1–specific T cells producing functional responses, as previously described (10).

In multifunctional characterization experiments (seven T-cell function characterization), incubation time in coculture with autologous tumor cells was extended to 12 hours in order to be able to detect both early and late cytokine production.

Where indicated, cancer cells were treated for 72 hours with 1,000 IU/mL TNF (CellGenix), 100 IU/mL IFNγ or both. Because of the high sensitivity of assessment of simultaneous positivity of multiple T-cell functions by CD8+ T cells, cells expressing simultaneously at least two functions (double positive cells) among TNF, IFNγ, and CD107a were chosen as a measure of the overall CD8 tumor reactivity in order to evaluate the effects of TNF on CD8+ T-cell recognition.

Cells were acquired with a BD FACSCanto II flow cytometer or, for multifunctional characterization of T-cell responses, with a 5 lasers BD LSR II. Flow cytometers were equipped with FACS Diva Software 6.3 (BD).

ELISPOT and cytotoxicity assays

IFNγ ELISPOT experiments were conducted as previously described (16). A total of 3 × 104 TILs (respectively 3 × 104 CD8+ T cells, 3 × 104 CD4+ T cells, or 1.5 × 104 CD8+ plus 1.5 × 104 CD4+ T cells, so that the total amount of TILs was identical in each well) and 3 × 103 cancer cells were added in each well. Triplicate wells were analyzed. Results are presented as number of IFNγ spots in stimulated wells minus background.

Conventional 51Cr-release assays for CTL-mediated cytotoxicity were carried out as described elsewhere (19).

Analysis of cancer cells

Analysis of MHC expression.

Semiquantitative MHC class I or II expression of cancer cells was assessed by standard staining of freshly detached cancer cells with anti-HLA-ABC or HLA-DP, DR, DQ antibodies or isotype controls, washed, and 2 μL of 7-AAD was added to each sample 5 minutes before acquisition. Cells were acquired with a BD FACSCanto II flow cytometer, and given the different autofluorescence of individual cell lines voltage parameters were adjusted for each cell line in order to obtain an APC mean fluorescence intensity (MFI) of 275 ± 15 and a FITC MFI of 65 ± 5.

Given the nonhomogeneous MHC class II staining of several cancer cell lines in our panel, melanomas were identified as MHC class II–positive when the MFI of the antibody-stained sample under study exceeded at least four times the isotype-control stained.

Analysis of cell proliferation.

Cell proliferation was evaluated using a flow cytometry–based counting method, as previously described (10). Briefly, after standard trypsinization, at day −1, melanoma cells were seeded at 5 to 8 × 104/well into 24-well plates and grown for 24 hours into standard medium. Thereafter, medium was exchanged with fresh standard medium ± indicated dilutions of supernatants from activated CD4+ T cells from corresponding patients for a final concentration of 0.5 mL/well. Baseline control wells were trypsinized at day 0 with 50 μL of trypsin solution per well and 250 μL of standard medium with 0.05 μg/mL of propidium iodide (PI; from Sigma-Aldrich) was added into each well to exclude dead cells, and the obtained suspension was counted under a standard rate for a constant amount of time (for 90 seconds at high flow rate) in a BD FACSCanto II flow cytometer equipped with BD FACS Loader carousel. After 72 hours of drug exposure, the other wells were trypsinized and the cells were counted after identical working conditions of the baseline control wells. Growth inhibition was calculated using the following formula: (T72 − T0)/(K72 − T0) × 100, where T72 is the cell count after 72 hours, T0 is the cell count of the control well at time zero, and K72 is the cell count of the control well (medium) after 72 hours. Control values were arbitrarily set to 100. Values below 0 indicated net cell loss while values between 0 and 100 indicated growth inhibition.

Supernatants from activated CD4+ T cells were obtained by incubating 5 × 106 sorted CD4+ T cells from individual patients with 5 × 105 autologous IFNγ-treated (72 hours) tumor cells in wells of a 24-well plate in a total volume of 1 mL. After 24 hours, supernatants were collected, cells were washed out by two centrifugation steps at 1,500 rpm/5 minutes, and supernatants were subsequently cryopreserved at −80°C for later use.

RT-PCR.

Total RNA was extracted from samples using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Reverse transcription reactions were performed using the Transcriptor First Strand cDNA Synthesis Kit (Roche). Real-time PCR (qPCR; for IDO-1, PDL-1, MLANA, TYR, PMEL, TAPBP, PSMB9, and GAPDH) analyses were carried out using in-house designed primers and a LightCycler Nano instrument (Roche).

Statistical analysis

The D'Agostino-Pearson normality test was performed to check for normal distribution of the values, and F tests to check whether populations had equal variances. Decisions on whether to perform parametric or nonparametric tests in relevant datasets were made on the basis of results from tests discussed above. Results were compared between different groups with two-tailed tests (either paired t tests, unpaired t tests, Mann–Whitney test, or Wilcoxon matched pairs tests). Survival analyses were conducted with log-rank test. Qualitative data were compared with the Fisher exact test.

Logarithmically transformed data were used to compare differences between groups for analysis of relative quantitative capacity to express single T-cell functions. For qPCR data, ΔCt were obtained by subtracting GAPDH Ct to target gene Ct. Wilcoxon matched-pairs signed-ranks test was used to compare differences between IFNγ-treated cells and medium-, TNF-, and TNF + IFNγ-treated cells. Data are shown as fold change relative to IFNγ-treated cells, which were set arbitrarily to 1.

In multifunctional characterization experiments (seven T-cell function characterization), flow cytometry data were initially processed with the FlowJo 9 (TreeStar Inc.) using a sequential gating strategy until identification of CD4+ and CD8+ T cells that were subcategorized in T-cell function–positive cells with parallel Boolean gating (gates drawn solely around responsive cells). Data were exported into Pestle 1.7 (courtesy of Dr. Mario Roederer, ImmunoTechnology Section, VRC/NIAID/NIH, Bethesda, MD) and properly formatted, according to Pestle instructions. Analysis and presentation of distributions was performed using SPICE version 5.2, downloaded from http://exon.niaid.nih.gov (20). In the analysis of REP TILs and minimally cultured TILs, background subtraction of unstimulated samples was performed with Pestle 1.7, according to the manufacturer's instructions. In contrast, in the analysis of uncultured TILs, background samples could not be subtracted because tumor digests unavoidably also contain uncultured tumor cells that can stimulate TILs during the 12 hours of incubation (data not shown). A threshold of at least 1% responding cells in CD4+ or CD8+ T-cell subsets was accepted, thus the impact of noise from background T-cell function expression was not more than approximately 30% to 40% (value estimated from background staining of Young TILs or REP TIL samples). Although with these thresholds, the contribution of background events to the overall response may be relatively high (especially for CD107a+ cells, where a proportion of positive cells was observed in unstimulated samples in minimally cultured TILs and REP TILs). The analysis was possible only with these conditions because the frequency of tumor-reactive cells in uncultured TILs was generally low (of five samples analyzed from patient (Pt.) 11, 15, 19, 24, and 25, the average frequency of positive cells was 2.2% ± 1.6% for CD4+ T cells, and 5.3% ± 5.1% for CD8+ T cells).

In SPICE, thresholds were set at 0.1 for analysis of REP TILs, while in all the other analyses, thresholds were set at 0.02. Comparison of distributions was performed using a Student t test and a partial permutation test as described previously (20). Other statistical analyses were performed with GraphPad Prism 5 (GraphPad Software).

Screening of CD8+ and CD4+ T-cell responses to autologous tumor antigens

Despite the fact that in vivo tumor heterogeneity may not be fully reflected by short-term cultured autologous melanoma cell lines, coculture assays of TILs/autologous tumor cell lines currently represents a gold standard as they allow the identification of T-cell responses directed against many, if not all, possible patient relevant tumor antigens. In order to obtain maximal T-cell responses, autologous tumors were pretreated with a low-dose IFNγ, which we have previously shown to increase TIL recognition and reactivity. Thereafter, tumors were exposed to autologous TILs (10).

Screening of 38 in vitro expanded TILs/autologous tumor pairs identified detectable direct CD8+ and CD4+ T-cell responses, respectively, in 34 (89%) and 18 (53%) patients (Fig. 1A; P = 0.001). The magnitude of responses, measured as the frequency of TILs expressing either TNF, IFNγ, or CD107a, was respectively 18% ± 18% (mean values, while the median was 10%) for CD8+ and 4.0% ± 9% (mean values, while the median was 0.66%) for CD4+ T cells (Fig. 1B; P < 0.0001). Thus, CD8+ T-cell responses appeared both more frequent and stronger in magnitude than CD4+ T-cell responses.

Figure 1.

CD4+ T-cell responses against melanoma. A, frequency of patients with CD4+ or CD8+ TIL responses against autologous melanoma cell lines. B, frequency of tumor-responding CD8+ or CD4+ TILs against autologous melanoma cell lines. Tumor cells were pretreated with 100 IU/mL IFNγ before coculture with TILs, as indicated in Materials and Methods. Tumor-responding cells expressing at least one of the following T-cell functions: TNF, IFNγ, or CD107a. Lines show median values. C and D, production of TNF from CD4+ T cells after MHC class II blocking on target tumor cells. C, one representative sample of six with similar results, gated on CD4+ T cells. Lines in D show median values. E, frequency of CD4+ T cells of all TILs in patients with high or no/low CD4+ T-cell responses. Lines show mean values. F, frequency of tumor-responding CD8+ T cells in patients with high or no/low CD4+ T-cell responses. Lines show mean values.

Figure 1.

CD4+ T-cell responses against melanoma. A, frequency of patients with CD4+ or CD8+ TIL responses against autologous melanoma cell lines. B, frequency of tumor-responding CD8+ or CD4+ TILs against autologous melanoma cell lines. Tumor cells were pretreated with 100 IU/mL IFNγ before coculture with TILs, as indicated in Materials and Methods. Tumor-responding cells expressing at least one of the following T-cell functions: TNF, IFNγ, or CD107a. Lines show median values. C and D, production of TNF from CD4+ T cells after MHC class II blocking on target tumor cells. C, one representative sample of six with similar results, gated on CD4+ T cells. Lines in D show median values. E, frequency of CD4+ T cells of all TILs in patients with high or no/low CD4+ T-cell responses. Lines show mean values. F, frequency of tumor-responding CD8+ T cells in patients with high or no/low CD4+ T-cell responses. Lines show mean values.

Close modal

Blockade of MHC class II on tumor cells significantly reduced CD4+ T-cell responses in 6 of 6 samples analyzed (samples from 6 patients with high CD4+ T-cell responses were selected), confirming class II dependent recognition (Fig. 1C and D).

The presence of high CD4+ T-cell responses (14 patients, identified with the presence of at least 2% cells in the whole CD4+ T-cell subpopulation responding to autologous tumor) was not associated with a higher frequency of CD4+ T cells among TILs (Fig. 1E). This may indicate that the magnitude of in vivo CD4+ T-cell infiltration, which is likely to be reflected in the CD4/CD8 T-cell ratio of expanded TIL cultures, is not associated with the presence of CD4+ T-cell responses but rather the majority of CD4+ TILs represents non-tumor relevant immune cells. Notably, we have previously shown the presence of high frequencies of non-tumor–related virus-specific CD8+ T cells in the tumor microenvironment (21).

In addition, the characteristics of CD8+ T-cell responses of patients with high CD4+ T-cell responses did not significantly differ from patients with no or low CD4+ T-cell responses (Fig. 1F and Supplementary Fig. S1).

Analysis of FoxP3 expression on expanded TILs from patients with high CD4+ T-cell reactivity (using flow cytometry) showed a positive staining on a relatively high fraction of both CD4+ and CD8+ TILs immediately after thawing. However, complete loss of FoxP3 expression 7 days after resting in IL2-free media was observed (data not shown). Therefore, we interpreted this temporary FoxP3 expression as an effect of high T-cell activation caused by in vitro culture conditions, and not associated with classical regulatory T-cell functions.

Expression of MHC class II in melanoma cell lines

Previous studies have shown constitutive expression of MHC class II molecules in 43% of melanoma cell lines, but also that the vast majority of melanomas (>70%) expressed class II upon exposure to IFN-γ—a cytokine that is presumably present at high levels in the tumor microenvironment, in close association with CD8+ T-cell antitumor responses (1).

Nineteen out of 38 melanomas (50%) expressed constitutively MHC class II molecules (Table 1 and Fig. 2A). Relative fold increase of MHC class II expression after IFNγ was similar in both MHC class II constitutively positive and negative tumors (data not shown) and with high or no/low CD4+ T-cell responses (Supplementary Fig. S2A). Only three out o19 constitutively class II–negative tumors did not express class II molecules after IFNγ exposure (data not shown).

Figure 2.

MHC class I and II expression in melanoma cells and association with CD4+ T-cell responses. A, expression of MHC class II molecules (HLA DP, DR, DQ) in two representative patients in either a class II constitutively positive melanoma or a MHC constitutively negative melanoma. Dotted line, isotype control; solid gray solid line, constitutive expression; solid black line, expression after pretreatment with IFNγ 100 IU/mL for 72 hours. B, frequency of tumor-responding CD4+ T cells in patients with constitutive expression of MHC class II molecules. Lines show median values. C and D, relative constitutive or IFNγ-induced class I expression of melanoma cell lines from patients with high or no/low CD4+ T-cell responses. Lines show mean values.

Figure 2.

MHC class I and II expression in melanoma cells and association with CD4+ T-cell responses. A, expression of MHC class II molecules (HLA DP, DR, DQ) in two representative patients in either a class II constitutively positive melanoma or a MHC constitutively negative melanoma. Dotted line, isotype control; solid gray solid line, constitutive expression; solid black line, expression after pretreatment with IFNγ 100 IU/mL for 72 hours. B, frequency of tumor-responding CD4+ T cells in patients with constitutive expression of MHC class II molecules. Lines show median values. C and D, relative constitutive or IFNγ-induced class I expression of melanoma cell lines from patients with high or no/low CD4+ T-cell responses. Lines show mean values.

Close modal
Table 1.

Characteristics of TILs and melanoma cell lines

Disease stageMHC class IICD8 T cellsCD4 T cells
Patient noAJCC stageConstitutive MHC class II% CD8% Responder CD8CD8 response% CD4% Responder CD4CD4 responseHigh CD4 response
IIIC YES 58.7 21.20 YES 38.2 36.98 YES YES 
IIIB YES 97 35.79 YES 0.94 YES NO 
IIIC NO 68 51.18 YES 30.7 0.50 NO NO 
IV NO 81 68.91 YES 18 0.50 NO NO 
IV YES 77 18.21 YES 10 11.77 YES YES 
IV NO 17.3 1.52 YES 82.1 0.50 NO NO 
IV NO 3.5 2.77 YES 5.1 0.50 NO NO 
IV NO 80 24.43 YES 20 1.77 YES NO 
IV YES 52.5 0.50 NO 46.8 2.38 YES YES 
10 IV NO 78.2 8.23 YES 21.4 0.50 NO NO 
11 IV YES 15 61.11 YES 82 21.48 YES YES 
12 IIIC NO 12.8 3.51 YES 85.1 0.50 NO NO 
13 IV YES 77.8 30.97 YES 21.5 0.50 YES NO 
14 IIIC YES 10.4 0.50 NO 88.3 4.45 YES YES 
15 IV YES 64 4.12 YES 33 2.46 YES YES 
16 IV NO 69.6 9.22 YES 29 0.64 YES NO 
17 IIIC YES 97 40.46 YES 4.27 YES YES 
18 IV YES 47.5 21.38 YES 51.8 36.05 YES YES 
19 IV YES 73 29.66 YES 24 3.61 YES YES 
20 IV NO 92.2 45.89 YES 0.50 NO NO 
21 IV NO 49.9 30.42 YES 28.9 0.50 NO NO 
22 IV YES 37.3 1.57 YES 10.1 0.50 NO NO 
23 IV NO 41.3 0.51 YES 56.7 0.50 NO NO 
24 IV NO 60 6.10 YES 31 5.47 YES YES 
25 IV NO 19.8 4.90 YES 66.5 2.27 YES YES 
26 IV YES 57.7 8.50 YES 37.7 0.50 NO NO 
27 IV YES 84 11.45 YES 14 0.50 NO NO 
28 IV YES 25.1 0.50 NO 72.6 2.27 YES YES 
29 IV YES 71 1.74 YES 27 0.67 YES NO 
30a IV YES 76 26.22 YES 23 0.50 NO NO 
31 IV NO 77.9 25.41 YES 20.2 0.50 NO NO 
32 IV NO 45 0.50 NO 40 0.50 NO NO 
33 IV NO 83 25.00 YES 10 0.50 NO NO 
34 IV YES 34.1 5.25 YES 61 2.91 YES YES 
35a IV YES 96.7 30.87 YES 2.5 8.44 YES YES 
36 IV NO 34.3 1.38 YES 64.9 1.11 YES NO 
37 IV NO 80 11.62 YES 18 1.76 YES NO 
38 IV NO 5.1 2.00 YES 91 0.50 NO NO 
Disease stageMHC class IICD8 T cellsCD4 T cells
Patient noAJCC stageConstitutive MHC class II% CD8% Responder CD8CD8 response% CD4% Responder CD4CD4 responseHigh CD4 response
IIIC YES 58.7 21.20 YES 38.2 36.98 YES YES 
IIIB YES 97 35.79 YES 0.94 YES NO 
IIIC NO 68 51.18 YES 30.7 0.50 NO NO 
IV NO 81 68.91 YES 18 0.50 NO NO 
IV YES 77 18.21 YES 10 11.77 YES YES 
IV NO 17.3 1.52 YES 82.1 0.50 NO NO 
IV NO 3.5 2.77 YES 5.1 0.50 NO NO 
IV NO 80 24.43 YES 20 1.77 YES NO 
IV YES 52.5 0.50 NO 46.8 2.38 YES YES 
10 IV NO 78.2 8.23 YES 21.4 0.50 NO NO 
11 IV YES 15 61.11 YES 82 21.48 YES YES 
12 IIIC NO 12.8 3.51 YES 85.1 0.50 NO NO 
13 IV YES 77.8 30.97 YES 21.5 0.50 YES NO 
14 IIIC YES 10.4 0.50 NO 88.3 4.45 YES YES 
15 IV YES 64 4.12 YES 33 2.46 YES YES 
16 IV NO 69.6 9.22 YES 29 0.64 YES NO 
17 IIIC YES 97 40.46 YES 4.27 YES YES 
18 IV YES 47.5 21.38 YES 51.8 36.05 YES YES 
19 IV YES 73 29.66 YES 24 3.61 YES YES 
20 IV NO 92.2 45.89 YES 0.50 NO NO 
21 IV NO 49.9 30.42 YES 28.9 0.50 NO NO 
22 IV YES 37.3 1.57 YES 10.1 0.50 NO NO 
23 IV NO 41.3 0.51 YES 56.7 0.50 NO NO 
24 IV NO 60 6.10 YES 31 5.47 YES YES 
25 IV NO 19.8 4.90 YES 66.5 2.27 YES YES 
26 IV YES 57.7 8.50 YES 37.7 0.50 NO NO 
27 IV YES 84 11.45 YES 14 0.50 NO NO 
28 IV YES 25.1 0.50 NO 72.6 2.27 YES YES 
29 IV YES 71 1.74 YES 27 0.67 YES NO 
30a IV YES 76 26.22 YES 23 0.50 NO NO 
31 IV NO 77.9 25.41 YES 20.2 0.50 NO NO 
32 IV NO 45 0.50 NO 40 0.50 NO NO 
33 IV NO 83 25.00 YES 10 0.50 NO NO 
34 IV YES 34.1 5.25 YES 61 2.91 YES YES 
35a IV YES 96.7 30.87 YES 2.5 8.44 YES YES 
36 IV NO 34.3 1.38 YES 64.9 1.11 YES NO 
37 IV NO 80 11.62 YES 18 1.76 YES NO 
38 IV NO 5.1 2.00 YES 91 0.50 NO NO 

aPt 30 and Pt 35 represent two spatially and temporally distinct metastases obtained from the same patient.

In order to clarify whether the aberrant expression of MHC class II was an exclusive event of metastatic melanomas, two cell lines originated from primary melanomas (WM-115 and FM-55-P) were characterized for MHC expression. Both cell lines showed a similar pattern of MHC class I expression compared with cells from metastatic origin (constitutive expression, increased expression after IFNγ). WM-115 was constitutively positive for MHC class II while FM-55-P was negative, and both stained positive (WM-115 with increased expression) after exposure to IFNγ (data not shown). Two additional cell lines, obtained from separate metastatic lesions of the same patient FM-55 (FM-55-P representing the primary melanoma), namely FM-55-M1 and FM-55-M2, were analyzed. Both showed no constitutive expression of MHC class II (but positive after IFNγ) exactly as their primary counterpart.

To extend the analysis and confirm our results, the ESTDAB database (http://www.ebi.ac.uk/ipd/estdab/) was examined. Indeed, our data on constitutive expression of WM-115 and FM-55-P matched those reported at ESTDAB. In addition, two out of two other cell lines derived from primary melanoma (WM-75 and WM-793) were reported to constitutively express at least one MHC class II isoform, exactly as the WM-266-4 that is a cell line originating from a metastatic lesion diagnosed 18 months after establishment of the primary cell line (WM-115) of one of the patients that we analyzed. These results confirm previous in situ data from Bröcker and colleagues (22), and indicate that activation of constitutive MHC class II expression may be a very early event in melanoma.

Association of CD4+ T-cell responses and constitutive expression of MHC class II

In order to characterize whether constitutive tumor expression of MHC class II is associated with increased frequency of CD4+ tumor-specific T cells in TILs, we examined CD4+ T-cell responses in matched samples (expanded TILs).

A higher frequency of tumor specific CD4+ T-cell responses were detected in TILs from tumors expressing MHC class II constitutively (mean 7.5% ± 11% for class II constitutively positive vs. 1% ± 1% for class II constitutively negative; P = 0.002; high CD4+ T-cell responses in 12 of 19 class II constitutively positive tumors vs. 2 of 19 class II constitutively negative tumors; P = 0.002; Fig. 2B; Table 1). In all cases, tumor recognition assays were performed after IFNγ pretreatment, allowing eventual detection of T-cell responses if tumor-specific CD4+ T cells were present, because the vast majority of tumors expressed high levels of MHC class II molecules regardless of constitutive positivity (see above).

Surprisingly, further quantitative analyses revealed that melanomas associated with a high CD4+ T-cell response upregulated surface MHC class I expression to a significantly higher extent upon IFNγ exposure compared with melanomas with low/no CD4+ T-cell responses (relative MFI 42 ± 19 vs. 29 ± 17, P = 0.026; relative fold increase 3.0 ± 0.8 vs. 2.4 ± 0.9, P = 0.047; Fig. 2C; Supplementary Fig. S2B). However, both melanomas had similar constitutive expression of MHC class I (relative MFI 15 ± 7 vs. 13 ± 8; P = 0.41; Fig. 2D).

These data indicate that class II expression is an early event that is frequently followed by infiltration of tumor antigen–specific CD4+ T cells. However, this is not an absolute requirement as the absence of strong CD4+ T-cell responses in some melanomas with constitutive class II expression (7 out of 19′ 37%) indicate that this is not sufficient, and the presence of high-frequency CD4+ T-cell responses in two melanomas with no constitutive class II expression (Pt. 24 and Pt. 25; 2 out of 19; 10.5%) indicate that this is not mandatory.

Of particular interest, two samples of our cohort (Pt. 30 and Pt. 35) were obtained sequentially from the same patient (Table 1). The first metastasis (named Pt. 30) was resected for inclusion in a TIL trial (clinicaltrials.gov identifier: NCT00937625). Shortly after, the patient was infused with the expanded TILs and this treatment resulted in >80% regression of all the preexistent tumor lesions, but one new metastatic lesion appeared at first evaluation (data not shown). This lesion continued to grow, and was resected approximately 6 months after infusion of TILs (named Pt. 35). As shown in Table 1, both cell lines were expressing MHC class II constitutively, but only the treatment-refractory and progressive metastatic lesion was infiltrated with tumor-specific CD4+ T cells, at high frequencies (Table 1). This is an interesting anecdotal observation indicating that the infiltration of tumor-specific CD4+ T cells can be subjected to a spatiotemporal dynamic regulation, and can be associated with disease progression without obvious changes in MHC class II expression on tumor cells.

Multifunctional characterization of CD4+ and CD8+ T-cell responses to melanomas

In order to determine potential functional patterns of CD4+ and CD8+ T-cell responses, we selected expanded TILs from 8 patients with strong responses in both subsets (>2% of CD4+ or CD8+ gated TILs simultaneously expressing TNF and IFNγ, and a subpopulation frequency of at least 10% CD4+ and/or CD8+ T cells; from Pts. 1, 5, 11, 15, 18, 19, 24, and 25) and performed a multifunctional analysis of their effector functions based on seven different known antitumor activities, in order to determine intrapatient variability of responses related to T-cell subsets.

To determine the relative quantitative capacity to express single T-cell functions, we compared the MFI of tumor-reactive cells. Tumor-reactive cells were identified by strict gating strategies of CD4+ and CD8+ T cells, gated solely on responsive (T cell-function positive) T cells.

These analyses revealed a higher intrinsic ability, on a single cell basis, of CD4+ T cells to produce TNF (Supplementary Fig. S3) and IL2, but comparable IFNγ, MIP-1α, and MIP-1β production (Fig. 3A). On the other hand, IL17A production was detected in CD4+ T-cell populations from only 1 patient and never in CD8+ T cells, while CD107a mobilization was much more associated with CD8+ T-cell activity (Fig. 3A). To summarize, a single TNF-producing CD4+ T-cell produces, on average, much more TNF than a TNF-producing CD8+ T cell. Similar results were obtained with IL2 production but, as expected, opposite results with CD107a mobilization.

Figure 3.

Multifunctional characterization of CD4+ and CD8+ T-cell responses to melanoma. A, semiquantitative comparison of the expression of seven individual T-cell functions from tumor-reactive CD4+ and CD8+ T cells. For any individual T-cell function, the relative expression value showed in the bar chart was obtained with the following formula: 100/[(MFI of all CD4+ T cells expressing the individual function in the stimulated sample/MFI of unstimulated CD4+ T cells)/(MFI of all CD8+ T cells expressing the individual function in the stimulated sample/MFI of unstimulated CD8+ T cells)]. Values over 1, the individual T-cell function is predominantly expressed by CD8+ T cells (e.g., in the case of a cytokine, cytokine-producing CD8+ T cells produce on a single cell basis more cytokine than cytokine-producing CD4+ T cells). Values below 1, the individual T-cell function is predominantly expressed by CD4+ T cells. B, CD4 or CD8 T-cell subpopulations were gated on cells expressing at least one of the seven T-cell functions analyzed, and the frequency of cells expressing any of the seven cell functions was assessed. Figure shows that the vast majority of tumor-responding CD4+ T cells (i.e., expressing at least one of the seven T-cell functions) also produce TNF. Columns show the average values ± error bars. C and D, SPICE T-cell function pattern analysis of cells expressing at least one of the seven T-cell functions shows very different functional patterns of CD4+ and CD8+ T cells. #, P < 0.05.

Figure 3.

Multifunctional characterization of CD4+ and CD8+ T-cell responses to melanoma. A, semiquantitative comparison of the expression of seven individual T-cell functions from tumor-reactive CD4+ and CD8+ T cells. For any individual T-cell function, the relative expression value showed in the bar chart was obtained with the following formula: 100/[(MFI of all CD4+ T cells expressing the individual function in the stimulated sample/MFI of unstimulated CD4+ T cells)/(MFI of all CD8+ T cells expressing the individual function in the stimulated sample/MFI of unstimulated CD8+ T cells)]. Values over 1, the individual T-cell function is predominantly expressed by CD8+ T cells (e.g., in the case of a cytokine, cytokine-producing CD8+ T cells produce on a single cell basis more cytokine than cytokine-producing CD4+ T cells). Values below 1, the individual T-cell function is predominantly expressed by CD4+ T cells. B, CD4 or CD8 T-cell subpopulations were gated on cells expressing at least one of the seven T-cell functions analyzed, and the frequency of cells expressing any of the seven cell functions was assessed. Figure shows that the vast majority of tumor-responding CD4+ T cells (i.e., expressing at least one of the seven T-cell functions) also produce TNF. Columns show the average values ± error bars. C and D, SPICE T-cell function pattern analysis of cells expressing at least one of the seven T-cell functions shows very different functional patterns of CD4+ and CD8+ T cells. #, P < 0.05.

Close modal

Subsequently, selective analysis of relative proportions of cells generating at least one of the seven T-cell functions revealed a significant skewing of CD4+ T cells toward production of TNF (Fig. 3B). Indeed, over 85% of CD4+ T cells generating at least one function were positive for TNF production. Thus, production of TNF seems a condition sine qua non that CD4+ T cells have to satisfy, in order to generate direct responses to melanomas. This is very different from what was observed for CD8+ T cells in the same TIL products, as only about 50% of responding cells were positive for TNF (Fig. 3B). Other T-cell functions, except for IFNγ production from CD8+ T cells (only a trend was observed), mirrored results of quantitative expression assessments. Indeed, similar frequency of MIP-1α- and MIP-1β-producing cells, slightly higher proportion of CD4+ T cells expressing IL2, but significantly more CD8+ T cells mobilizing CD107a were detected (Fig. 3B). Globally, these results indicate that regardless which function is expressed by CD4+ T cells, TNF production seems to be universal.

Simplified Presentation of Incredibly Complex Evaluations (SPICE) is a novel bioinformatics tool that allows in-depth dissection of diverse “patterns” of T-cell effector functions, with potential identification of a multitude of distinct cell populations (not evidenced with classical sequential gating strategies) by combinatorial Boolean gating and complex algorithms of data analysis (20, 23). To this end, we have subjected our samples to supervised analysis of T-cell response quality with SPICE.

Analysis of complex functional patterns revealed strong intrapatient differences between expanded CD4+ and CD8+ T cells reacting to autologous melanomas (P = 0.0006). Again confirming a significant CD4+ T-cell skewing toward TNF production—in particular, over 50% of CD4+ T cells generating detectable responses were producing only TNF—while CD8+ T-cell responses appeared much more complex, multifunctional and mostly based on IFNγ production or CD107a mobilization (Fig. 3C and D). Supplementary Fig. S4 shows a graphical representation of SPICE data analysis displaying all possible combinations of T-cell functions (n = 128) in a bar chart (Supplementary Fig. S4A), NPlot for CD4+ (Supplementary Fig. S4B), or CD8+ (Supplementary Fig. S4C) T cells or CoolPlot (Supplementary Fig. S4D).

Because of the technical complexity and the availability of large TIL samples required for the experiments described, expanded TILs were used as previously indicated. However, it is not known whether the observed functional patterns are maintained during several logs expansion. Thus, we conducted similar analyses with minimally expanded TILs as well as uncultured TILs, and compared the results with expanded TILs obtained from the same patients. We conducted seven T-cell function experiments, but as with expanded TILs, only a minority of T cells expressed IL2, IL17A, MIP-1α, or MIP-1β (data not shown). Therefore, only three functions (TNF, IFNγ, and CD107a) were analyzed.

Keeping in mind potential technical pitfalls in the analysis of uncultured TILs (see Materials and Methods/Statistical Analysis), this analysis showed that similar patterns were observed with all TILs analyzed. Close similarities with previous multifunctional result from expanded TILs only where obtained. Indeed, when comparing CD4+ with CD8+ T-cell subsets within each type of TIL culture, a higher proportion of TNF-producing cells was observed in CD4+ T cells (Supplementary Fig. S5A). In addition, CD4+ T cells and CD8+ T cells patterns were similar in different types of TIL cultures (Supplementary Fig. S5A and S5B), Supplementary Fig. S6 shows the T-cell functional patterns for each individual patient, again showing close intrapatient similarities between types of TIL culture. Thus, we concluded that the main functional differences between CD4+ and CD8+ T-cell subsets observed in expanded TILs were stable during in vitro expansion, thus mirrored T-cell responses in the tumor microenvironment.

On a different note, it seemed that a higher proportion of CD107a+ cells along with a lower proportion of IFNγ+ cells were present in uncultured TILs (Supplementary Fig. S5A and S5B). However, given the small number of patients analyzed as well as technical issues in the analysis of uncultured TILs (see Materials and Methods/Statistical Analysis), this incidental observation will need further investigation, but it does not change the overall conclusion that CD4+ T cells produce more TNF also in uncultured TILs.

Effects of TNF on CD8+ T-cell recognition

Because our data indicated that TNF production was the main effector function of melanoma-specific CD4+ T cells, we asked whether exposure of autologous melanoma to TNF could influence recognition by expanded CD8 TILs. To this end, melanoma cells were treated with TNF, IFNγ, or both TNF and IFNγ simultaneously. This was chosen on the basis of previous SPICE analyses, in order to reproduce a tumor microenvironment with strong CD4+ T-cell responses, CD8+ T-cell responses or both.

Landsberg and colleagues (24) has previously demonstrated that TNF induces melanoma reversible dedifferentiation with reduced T-cell recognition of melanoma differentiation antigens (MDA). Analysis of T-cell responses revealed, as expected, that TNF significantly reduced recognition of MDAs (Fig. 4A and Supplementary Fig. S7—figures show a representative patient, experiments were conducted with samples from three different patients and similar results were obtained; data not shown). However, TNF did not significantly influence global CD8+ T-cell reactivity (Fig. 4A and B). As expected, IFNγ significantly increased both global CD8+ and CD4+ T-cell reactivity (Fig. 4B–D and Supplementary Fig. S8) as previously demonstrated by our group (10). Of note, it has similarly been shown that IFNγ neither increases or reduces autologous or allogeneic tumor recognition by MDA-specific cells, such as CD8+ T cells recognizing MART-1EAA or gp-100YLE peptides (10)—although in this case, it is not fully elucidated whether the effects are antigen class–specific or rather peptide-specific because of the IFNγ-induced shift from proteasome to immunoproteasome (25, 26).

Figure 4.

Effects of cytokines on tumor recognition of CD4+ and CD8+ TILs. A, TNF reduces tumor recognition of MART-1–specific CD8+ TILs without significantly modifying tumor recognition of other-specific CD8+ TILs. FACS plots of cytokine expression from a representative patient of three tested with similar results are shown in the figure. B, effects of IFNγ, TNF, or simultaneous exposure of TNF and IFNγ on tumor recognition of bulk CD8+ TILs. Lines show mean values of the frequency of CD8+ TILs expressing at least two functions simultaneously among TNF, IFNγ, and CD107a. C and D, effects of IFNγ, TNF, or simultaneous exposure of TNF and IFNγ on tumor recognition of bulk CD4+ TILs. Lines show mean values *, P < 0.05; **, P < 0.01.

Figure 4.

Effects of cytokines on tumor recognition of CD4+ and CD8+ TILs. A, TNF reduces tumor recognition of MART-1–specific CD8+ TILs without significantly modifying tumor recognition of other-specific CD8+ TILs. FACS plots of cytokine expression from a representative patient of three tested with similar results are shown in the figure. B, effects of IFNγ, TNF, or simultaneous exposure of TNF and IFNγ on tumor recognition of bulk CD8+ TILs. Lines show mean values of the frequency of CD8+ TILs expressing at least two functions simultaneously among TNF, IFNγ, and CD107a. C and D, effects of IFNγ, TNF, or simultaneous exposure of TNF and IFNγ on tumor recognition of bulk CD4+ TILs. Lines show mean values *, P < 0.05; **, P < 0.01.

Close modal

Surprisingly, exposure to TNF significantly reduced the well-known IFNγ-mediated increase of CD8+ T-cell responses to melanoma (Fig. 4B and Supplementary Fig. S8). On the other hand, TNF increased (although not significantly) CD4+ T-cell responses compared with untreated tumors (Fig. 4C and D).

These data indicate that on one hand TNF tends to marginally amplify itself with a positive feedback mechanism (tumor exposure to TNF increase TNF production from CD4+ T cells), but on the other hand, it dampens CD8+ T-cell reactivity in an IFNγ-rich tumor microenvironment, as it occurs when strong CD8+ T-cell responses are present simultaneously.

Effects of TNF on gene expression of melanoma cells

To clarify whether the observed influences on tumor recognition by CD8+ T cells were associated with changes in gene expression, a panel of genes involved in immune recognition was analyzed by qPCR in eight cell lines (from Pts. 1, 5, 11, 13, 15, 17, 18, and 19). Besides the well-known effects of IFNγ to increase the expression of immunosuppressive genes, the addition of TNF increased expression of genes such as indoleamine 2,3 dioxygenase-1 (IDO-1) or programmed death ligand-1 (PD-L1) by about 5- to 10-fold (Supplementary Fig. S9A). In addition, a trend toward reduced expression of MDAs was evidenced (Supplementary Fig. S9B). On the other hand, a trend toward increase in expression of the MHC class I processing and presentation pathway was observed with the addition of TNF by 2- to 5-folds, but notably IFNγ itself increased gene expression by about 10 folds relative to control (Supplementary Fig. S9C).

These data may suggest that TNF reduces melanoma immune-sensitivity in an IFNγ-rich tumor microenvironment through augmentation of tumor cell immunosuppressive capacity.

Effects of tumor-specific CD4+ T cells on melanoma cell proliferation and viability or on short-term CD8 IFNγ responses

In order to characterize whether CD4+ T-cell responses could influence melanoma cell proliferation or viability, we conducted additional proliferation and cytotoxicity assays.

Autologous tumors were incubated for 72 hours with supernatants from sorted CD4+ T-cell cultures from 6 patients with a high level of CD4+ T-cell reactivity (Pts. 1, 5, 11, 15, 18, and 19) as described above (activated with autologous tumors), and cell proliferation was assessed with a standard flow cytometry–based counting assay as previously described (10). Meaningful antiproliferative properties of CD4+ T-cell supernatants were evident in only 2 out of 6 melanoma cell lines (Fig. 5A).

Figure 5.

Effects of tumor-responding CD4+ T cells on tumor cells or CD8 IFNγ responses. A, effects of supernatants of activated CD4+ T cells from patients with high CD4+ T-cell responses on proliferation of autologous melanoma cells. Dots show average values with error bars. B, autologous tumor cytotoxicity of CD8+ or CD4+ TILs from patients with high CD4+ T-cell responses. CD4+ T-cell cultures are shown in gray while CD8+ T cells are shown in black. C and D, IFNγ ELISPOT analysis of selected CD8+, CD4+, or combined CD8+ and CD4+ TILs. C and D, results from individual patients (average of triplicate observations and error bars; C), while D shows ELISPOT wells from a representative patient.

Figure 5.

Effects of tumor-responding CD4+ T cells on tumor cells or CD8 IFNγ responses. A, effects of supernatants of activated CD4+ T cells from patients with high CD4+ T-cell responses on proliferation of autologous melanoma cells. Dots show average values with error bars. B, autologous tumor cytotoxicity of CD8+ or CD4+ TILs from patients with high CD4+ T-cell responses. CD4+ T-cell cultures are shown in gray while CD8+ T cells are shown in black. C and D, IFNγ ELISPOT analysis of selected CD8+, CD4+, or combined CD8+ and CD4+ TILs. C and D, results from individual patients (average of triplicate observations and error bars; C), while D shows ELISPOT wells from a representative patient.

Close modal

CD4+ T cells with direct cytotoxic activity to melanomas have been previously shown in mice. In order to assess whether tumor cytotoxic CD4+ T cells could be detected in humans, a standard 4-hour cytotoxicity assay was performed with sorted CD4+ or CD8+ T cells from 5 patients (Pts. 1, 5, 11, 15, and 19) with high CD4+ T-cell responses. CD8+ T cells displayed the ability to kill autologous tumors (>10% lysis at a effector:target ratio of 90:1) in 4 out of 5 cases, while CD4+ T cells did not in any of the cases (Fig. 5B).

These results indicate that CD4+ T-cell responses do not seem to have strong direct antitumor activities in the effector phase of the immune response.

To assess whether production of IFNγ from CD4+ and CD8+ T cells would be influenced by the presence of the other cell subset and/or whether the presence of CD4+ T cells may enhance CD8 responses, ELISPOT assays with the addition of either CD4+, CD8+ or CD4+, and CD8+ T cells from 6 individual patients (Pts. 1, 5, 11, 15, 18, and 19) with high CD4+ T-cell reactivity were conducted. The number of IFNγ specific spots obtained with the addition of both cell subsets was consistently around the average of the number of spots obtained with either CD4+ or CD8+ T cells alone and no multiplicative effects were observed (Fig. 5C and D). Therefore, we concluded that CD4+ and CD8+ T cells do not seem to interact or influence one another in short-term assays in regard to tumor recognition and relative IFNγ production.

Outcome of patients with constitutive and high CD4+ T-cell responses

Finally, we attempted to assess whether the presence of high CD4+ T-cell responses, constitutive class II positivity, or high CD4/CD8 T-cell ratios in expanded TILs (this latter with a larger group of patients) were associated with known prognostic factors in melanoma or patient survival. Plasma lactate dehydrogenase (LDH) was chosen as it currently represents the most accurate prognostic circulating marker in metastatic melanoma.

Strong differences in plasma LDH levels were not observed (Supplementary Fig. S10A and S10E), except for a higher LDH value in patients with MHC class II constitutively negative tumor (Supplementary Fig. S10C). No differences in overall survival were observed between patient groups (Supplementary Fig. S10B, S10D, and S10F). Keeping in mind the relatively small number of patients, these data suggest that the factors examined are not strong independent prognostic indicators in melanoma.

Anecdotal cases of tumor regression mediated exclusively by CD4+ T-cell responses have been reported in humans with metastatic melanoma (8, 9, 27). However, little is known regarding how natural effector CD4+ T-cell responses are generated, and how CD4+ T cells influence CD8+ T-cell responses in the tumor microenvironment.

Constitutive expression of MHC class II is usually restricted to professional antigen-presenting cells (APC; ref. 28). However, previous studies have shown that constitutive expression of MHC class II in melanoma via abnormal transcription of class II transactivator (CIITA) might not be a random event, but rather linked to neoplastic transformation (29). CIITA activation was not involved per se in tumor progression (30), suggesting that its abnormal induction is initiated and maintained by more complex networks dependent on MHC class II expression.

In this study, we have shown that, despite not being necessary or sufficient, de novo constitutive expression of MHC class II in melanoma cells is a powerful stimulus to promote accumulation of tumor-specific CD4+ T cells in the tumor microenvironment. Our data indicate that, presumably, this event already occurs at an early stage—that is, localized melanoma. CD4+ T cells directly recognize tumor antigens presented in association with MHC class II molecules on the surface of melanomas, and exert effector functions dominated by TNF production. Indeed, endogenous proteins can be presented on class II via autophagy and/or nonclassical antigen processing (31, 32).

We have shown that the effector functions generated by tumor-specific CD4+ T cells do not significantly influence melanoma proliferation or viability, with parallel effective cytotoxicity of CD8+ T cells. This implies that, under standard conditions of cell expansion, direct antitumor activities of tumor-specific CD4+ T cells are not very much effective in limiting the growth of melanoma in vitro, at least in the majority of cases. This may reflect an in vivo condition where the majority of tumor-specific CD4+ T cells are not able to mediate powerful direct antitumor effects.

Previous studies have shown the deleterious effects of chronic exposure of melanoma tumors to TNF. Landsberg and colleagues (24) have shown that a TNF-rich inflammatory environment can reduce the immunogenicity and promote immune escape of melanoma by inducing reversible dedifferentiation. In this study, we have confirmed that chronic exposure of melanoma tumors to local TNF downregulate recognition of MDA-specific CD8+ T cells. However, when using bulk TIL cultures, it was interesting to see that preexposure to TNF did not significantly influence the global CD8+ T-cell reactivity. This may indicate that the contribution of MDA-specific T cells to the overall CD8+ T-cell reactivity is relatively small.

In contrast, exposure to TNF significantly reduced the well-known IFNγ-mediated increase of CD8+ T-cell responses to melanoma. This suggests that in an IFNγ-rich environment, such as a tumor microenvironment containing IFNγ-producing CD8+ T cells, TNF from tumor-specific CD4+ T cells (or other sources) may reduce CD8+ T-cell reactivity and, in parallel, increase CD4+ T-cell responses through a positive feedback, thereby amplifying local TNF signaling.

In an attempt to characterize the molecular events leading to the observed phenomena, we analyzed the expression of a set of genes capable of influencing immune sensitivity by melanoma. Although the addition of TNF further increased the immunosuppressive capacity of tumor cells as evidenced by induction of IDO-1 and PD-L1 and reduced the expression of MDA genes, other genes associated with MHC class I antigen processing and presentation pathway seemed to be induced in parallel. Despite difficulties to quantify the relative contribution of different immune-activating/immune-suppressive pathways, an increased immunosuppressive capacity of tumor cells may explain the reduced immune sensitivity observed in this setting.

Interestingly, these data support a model of tumor progression where some melanomas, in order to escape effective CD8+ T-cell responses, activate CIITA and consequently express MHC class II constitutively. This recruits tumor-specific CD4+ T cells in the tumor microenvironment, which, in turn, dampen CD8+ T-cell responses via production of TNF.

Notably, MHC class II has several features in common with other tumor-associated immunosuppressive molecules such as, for instance, IDO and PD-L1. In fact, as shown in this and several other studies, MHC class II is aberrantly activated in some melanomas, and exactly as IDO and PD-L1 (33, 34), it is upregulated by IFNγ-mediated immune responses. Thus, in situ detection of MHC class II in melanoma may represent a constitutive expression in melanoma cells or induced by the presence of IFNγ-secreting cells (e.g., tumor antigen–specific CD8+ T cells), or both.

Several other mechanisms to evade CD8+ T-cell responses have been characterized so far (35). CD8+ T cells recognize tumor antigens expressed on the surface of tumor cells in association with MHC class I molecules. Therefore, defects in the class I antigen processing and presentation pathway, as well as downregulation of MHC class I molecules may per se dampen CD8+ T-cell responses. Indeed, several studies have demonstrated that reduced expression of class I molecules is associated with a worse prognosis in several types of solid tumors, including melanoma (5).

Interestingly, in this study those melanomas that did not attract strong CD4+ T-cell responses displayed a defective upregulation of MHC class I in response to IFNγ. This may indeed be interpreted as an additional immune escape mechanism, thereby suggesting the existence of two different subsets of tumors: (i) melanomas with high CD4+ T-cell responses (attracted in most cases by de novo constitutive expression of class II) and normal class I upregulation after exposure to IFNγ; (ii) melanomas with defective class I upregulation in response to IFNγ, which do not attract strong CD4+ T-cell responses. According to this model, both melanomas efficiently downregulate CD8+ T-cell responses either indirectly through attraction of inflammatory CD4+ T cells (melanoma subset i), or directly through defective upregulation of MHC Class I (melanoma subset ii).

As an indirect confirmation of this theory, patients with strong CD4+ T-cell responses (or with constitutive MHC class II–positive tumors) did not seem to have a worse prognosis than patients without CD4+ T-cell responses—with the latter displaying on average a defective upregulation of MHC class I molecules.

The vast majority of patients in our study (85%) were diagnosed with AJCC stage IV melanoma (distant metastases). However, we detected strong CD4+ T-cell responses also in 3 out of 6 patients with earlier disease stage (IIIB/IIIC). As indicated by a recent study (36), in those cases where prompt and complete resection of locoregional melanoma is performed, the impact of local immune responses may be irrelevant in regard to the clinical outcome. In contrast tumor-intrinsic features may dictate whether the patient will develop distant metastases shortly. Thus, it is more likely that our observations can be applied to a disease stage where complete resection is not feasible.

It should be highlighted that, despite the natural in situ responses to melanomas from CD4+ T cells and the associated chronic exposure to TNF may reduce antitumor CD8+ T-cell responses in the majority of cases (i.e., most tumor-specific CD4+ T cells), this does not exclude a potential beneficial role of selected CD4+ T-cell subpopulations or the possibility to take advantage of these responses therapeutically, as recently suggested by Tran and colleagues (37) in colangiocarcinoma (i.e., polyfunctional mutated-antigen specific CD4+ T cells) or by previous single-case reports (9, 27). On this note, Linnemann and colleagues (11) have recently characterized mutated (neo-) antigen-specific CD4+ T cells in melanoma. In their study, the authors describe in 4 out of 5 patients a (relatively) small subset of CD4+ T cells that produce IFNγ, but not other cytokines, upon exposure to neo-antigens. This is particularly intriguing, as our data show instead that the vast majority of tumor-specific CD4+ T cells in melanoma produce TNF and not IFNγ, but also that a very small subset of melanoma-specific CD4+ T cells produce only IFNγ (Fig. 3C and D). It is anyway difficult to draw firm conclusions, because most T-cell recognition analyses in Linnemann and colleagues (11) were performed with autologous B cells (presenting neo-antigens) as targets and not tumor cells. It is indeed known from previous studies that several factors can affect T-cell functionality, including antigen density (38, 39) that may of course be different in peptide-loaded APCs and tumor cells. Future studies will determine whether functional differences exist between CD4+ T-cell subsets specific to different classes of tumor antigens.

In addition, the role of tumor antigen–specific CD4+ T cells is certainly not limited to direct recognition of tumor cells. For instance, tumor antigen–specific CD4+ T cells may promote the cross-priming of tumor antigen–specific CD8+ T cells through CD40-ligand mediated activation of antigen presenting cells (40). Recently, it was shown in three independent murine models that, unexpectedly, the majority of the immunogenic mutanome is recognized by CD4+ T cells (41). If recognition of tumor antigens by CD4+ T cells is so common in cancer, it could be speculated that during tumor progression, cancer cells are shaped in order to take advantage of, or at least not suffer from, such features of the immune system. Our results suggest that this may be realized through generation of local conditions (e.g., aberrant expression of MHC class II) inducing such CD4+ T cells to accumulate in the tumor microenvironment in order to support, instead of inhibiting, tumor growth.

In conclusion, several mechanisms of immune escape have been identified and strategies to counteract tumor immune inhibition show impressive results in the clinical setting (42–45). In this study, we have identified a potential mechanism by which de novo aberrant expression of MHC class II on melanomas may contribute to tumor escape via recruitment of inflammatory tumor antigen–specific CD4+ T cells. However, the fundamental role of MHC class II molecules in the initiation and maintenance of immune responses does not immediately warrant strategies directly counteracting this pathway, and future studies must address this issue.

No potential conflicts of interest were disclosed.

Conception and design: M. Donia, F. Nicoletti, P. Thor Straten, I.M. Svane

Development of methodology: M. Donia, P. Fagone, S. Munir

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Donia, R. Andersen, I.M. Svane

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Donia, R. Andersen, P. Fagone

Writing, review, and/or revision of the manuscript: M. Donia, R. Andersen, P. Fagone, F. Nicoletti, M.H. Andersen, I.M. Svane

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Donia, J.W. Kjeldsen, I.M. Svane

Study supervision: M. Donia, I.M. Svane

The authors thank Drs. Filippo Fraggetta (Department of Pathology, Cannizzaro Hospital, Catania, Italy) and Morten Hansen (Center for Cancer Immune Therapy, Herlev Hospital, Denmark) for useful discussions. Dr. Niels Junker (Department of Oncology, Herlev Hospital, Denmark) is acknowledged for having established the complex TIL methods at Herlev Hospital.

The studies were supported by grants from the Italian Foundation for Cancer Research (FIRC), Aase and Ejnar Danielsens Foundation, the Danish Cancer Society, the Lundbeck Foundation, the Beckett-Fonden, the Herlev Hospital Research Council, and the Capital Region of Denmark Research Foundation.

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.

1.
Mendez
R
,
Aptsiauri
N
,
Del Campo
A
,
Maleno
I
,
Cabrera
T
,
Ruiz-Cabello
F
, et al
HLA and melanoma: multiple alterations in HLA class I and II expression in human melanoma cell lines from ESTDAB cell bank
.
Cancer Immunol Immunother
2009
;
58
:
1507
15
.
2.
Reith
W
,
LeibundGut-Landmann
S
,
Waldburger
J-M
. 
Regulation of MHC class II gene expression by the class II transactivator
.
Nat Rev Immunol
2005
;
5
:
793
806
.
3.
Altomonte
M
,
Fonsatti
E
,
Visintin
A
,
Maio
M
. 
Targeted therapy of solid malignancies via HLA class II antigens: a new biotherapeutic approach?
Oncogene
2003
;
22
:
6564
9
.
4.
LeibundGut-Landmann
S
,
Waldburger
J-M
,
Krawczyk
M
,
Otten
LA
,
Suter
T
,
Fontana
A
, et al
Mini-review: specificity and expression of CIITA, the master regulator of MHC class II genes
.
Eur J Immunol
2004
;
34
:
1513
25
.
5.
Anichini
A
,
Mortarini
R
,
Nonaka
D
,
Molla
A
,
Vegetti
C
,
Montaldi
E
, et al
Association of antigen-processing machinery and HLA antigen phenotype of melanoma cells with survival in American Joint Committee on Cancer stage III and IV melanoma patients
.
Cancer Res
2006
;
66
:
6405
11
.
6.
Van Duinen
SG
,
Ruiter
DJ
,
Broecker
EB
,
van der Velde
EA
,
Sorg
C
,
Welvaart
K
, et al
Level of HLA antigens in locoregional metastases and clinical course of the disease in patients with melanoma
.
Cancer Res
1988
;
48
:
1019
25
.
7.
Zaloudik
J
,
Moore
M
,
Ghosh
AK
,
Mechl
Z
,
Rejthar
A
. 
DNA content and MHC class II antigen expression in malignant melanoma: clinical course
.
J Clin Pathol
1988
;
41
:
1078
84
.
8.
Friedman
KM
,
Prieto
PA
,
Devillier
LE
,
Gross
CA
,
Yang
JC
,
Wunderlich
JR
, et al
Tumor-specific CD4+ melanoma tumor-infiltrating lymphocytes
.
J Immunother
2012
;
35
:
400
8
.
9.
Robbins
PF
,
El-Gamil
M
,
Li
YF
,
Zeng
G
,
Dudley
M
,
Rosenberg
SA
. 
Multiple HLA class II-restricted melanocyte differentiation antigens are recognized by tumor-infiltrating lymphocytes from a patient with melanoma
.
J Immunol
2002
;
169
:
6036
47
.
10.
Donia
M
,
Hansen
M
,
Sendrup
SL
,
Iversen
TZ
,
Ellebæk
E
,
Andersen
MH
, et al
Methods to improve adoptive T-cell therapy for melanoma: IFN-γ enhances anticancer responses of cell products for infusion
.
J Invest Dermatol
2013
;
133
:
545
52
.
11.
Linnemann
C
,
van Buuren
MM
,
Bies
L
,
Verdegaal
EME
,
Schotte
R
,
Calis
JJ
, et al
High-throughput epitope discovery reveals frequent recognition of neo-antigens by CD4+ T cells in human melanoma
.
Nat Med
2014
;
21
:
1
7
.
12.
Hemon
P
,
Jean-Louis
F
,
Ramgolam
K
,
Brignone
C
,
Viguier
M
,
Bachelez
H
, et al
MHC class II engagement by its ligand LAG-3 (CD223) contributes to melanoma resistance to apoptosis
.
J Immunol
2011
;
186
:
5173
83
.
13.
Matsuzaki
J
,
Gnjatic
S
,
Mhawech-Fauceglia
P
,
Beck
A
,
Miller
A
,
Tsuji
T
, et al
Tumor-infiltrating NY-ESO-1-specific CD8+ T cells are negatively regulated by LAG-3 and PD-1 in human ovarian cancer
.
Proc Natl Acad Sci U S A
2010
;
107
:
7875
80
.
14.
Baitsch
L
,
Baumgaertner
P
,
Devêvre
E
,
Raghav
SK
,
Legat
A
,
Barba
L
, et al
Exhaustion of tumor-specific CD8+ T cells in metastases from melanoma patients
.
J Clin Invest
2011
;
121
:
23
5
.
15.
Rodeck
U
,
Herlyn
M
,
Menssen
HD
,
Furlanetto
RW
,
Koprowsk
H
. 
Metastatic but not primary melanoma cell lines grow in vitro independently of exogenous growth factors
.
Int J cancer
1987
;
40
:
687
90
.
16.
Donia
M
,
Junker
N
,
Ellebaek
E
,
Andersen
MH
,
Straten
PT
,
Svane
IM
. 
Characterization and comparison of “Standard” and “Young” tumor infiltrating lymphocytes for adoptive cell therapy at a Danish Translational Research Institution
.
Scand J Immunol
2012
;
75
:
157
67
.
17.
Tran
KQ
,
Zhou
J
,
Durflinger
KH
,
Langhan
MM
,
Shelton
TE
,
Wunderlich
JR
, et al
Minimally cultured tumor-infiltrating lymphocytes display optimal characteristics for adoptive cell therapy
.
J Immunother
2008
;
31
:
742
51
.
18.
Selvan
SR
,
Carbonell
DJ
,
Fowler
AW
,
Beatty
AR
,
Ravindranath
MH
,
Dillman
RO
. 
Establishment of stable cell lines for personalized melanoma cell vaccine
.
Melanoma Res
2010
;
20
:
280
92
.
19.
Andersen
MH
,
Bonfill
JE
,
Neisig
A
,
Arsequell
G
,
Sondergaard
I
,
Valencia
G
, et al
Phosphorylated peptides can be transported by TAP molecules, presented by class I MHC molecules, and recognized by phosphopeptide-specific CTL
.
J Immunol
1999
;
163
:
3812
8
.
20.
Roederer
M
,
Nozzi
JL
,
Nason
MC
. 
SPICE: exploration and analysis of post-cytometric complex multivariate datasets
.
Cytometry A
2011
;
79
:
167
74
.
21.
Andersen
RS
,
Thrue
CA
,
Junker
N
,
Lyngaa
R
,
Donia
M
,
Ellebæk
E
, et al
Dissection of T-cell antigen specificity in human melanoma
.
Cancer Res
2012
;
72
:
1642
50
.
22.
Bröcker
E
,
Suter
L
,
Sorg
C
. 
HLA-DR antigen expression in primary melanomas of the skin
.
J Invest Dermatol
1984
;
82
:
244
7
.
23.
Lugli
E
,
Roederer
M
,
Cossarizza
A
. 
Data analysis in flow cytometry: the future just started
.
Cytometry A
2010
;
77
:
705
13
.
24.
Landsberg
J
,
Kohlmeyer
J
,
Renn
M
,
Bald
T
,
Rogava
M
,
Cron
M
, et al
Melanomas resist T-cell therapy through inflammation-induced reversible dedifferentiation
.
Nature
2012
;
490
:
412
6
.
25.
Morel
S
,
Lévy
F
,
Burlet-Schiltz
O
. 
Processing of some antigens by the standard proteasome but not by the immunoproteasome results in poor presentation by dendritic cells
.
Immunity
2000
;
12
:
107
17
.
26.
Cangemi
G
,
Morandi
B
,
Agostino
AD
,
Peri
C
,
Conte
R
,
Damonte
G
, et al
IFN-a mediates the up-regulation of HLA class I on melanoma cells without switching proteasome to immunoproteasome
.
Int Immunol
2003
;
15
:
1415
21
.
27.
Hunder
NN
,
Wallen
H
,
Cao
J
,
Hendricks
DW
,
Reilly
JZ
,
Rodmyre
R
, et al
Treatment of metastatic melanoma with autologous CD4+ T cells against NY-ESO-1
.
N Engl J Med
2008
;
358
:
2698
703
.
28.
Glimcher
LH
,
Kara
CJ
. 
Sequences and factors: a guide to MHC class-II transcription
.
Annu Rev Immunol
1992
;
10
:
13
49
.
29.
Deffrennes
V
,
Vedrenne
J
,
Piskurich
J
,
Barbieri
G
,
Jenny
P
,
Charron
D
, et al
Constitutive expression of MHC class II genes in melanoma cell lines results from the transcription of class II transactivator abnormally initiated from Its B cell-specific promoter
.
J Immunol
2001
;
167
:
98
106
.
30.
Baton
F
,
Deruyffelaere
C
,
Chapin
M
,
Prod'homme
T
,
Charron
D
,
Al-Daccak
R
, et al
Class II transactivator (CIITA) isoform expression and activity in melanoma
.
Melanoma Res
2004
;
14
:
453
61
.
31.
Muranski
P
,
Restifo
NP
. 
Adoptive immunotherapy of cancer using CD4+ T cells
.
Curr Opin Immunol
2009
;
33
:
759
68
.
32.
Matsuzaki
J
,
Tsuji
T
,
Luescher
I
,
Old
LJ
,
Shrikant
P
,
Gnjatic
S
, et al
Nonclassical antigen-processing pathways are required for MHC class II-restricted direct tumor recognition by NY-ESO-1-specific CD4(+) T cells
.
Cancer Immunol Res
2014
;
2
:
341
50
.
33.
Spranger
S
,
Spaapen
RM
,
Zha
Y
,
Williams
J
,
Meng
Y
,
Ha
TT
, et al
Up-regulation of PD-L1, IDO, and T(regs) in the melanoma tumor microenvironment is driven by CD8(+) T cells
.
Sci Transl Med
2013
;
5
:
200ra116
.
34.
Taube
JM
,
Anders
RA
,
Young
GD
,
Xu
H
,
Sharma
R
,
McMiller
TL
, et al
Colocalization of inflammatory response with B7-h1 expression in human melanocytic lesions supports an adaptive resistance mechanism of immune escape
.
Sci Transl Med
2012
;
4
:
127ra37
.
35.
Cavallo
F
,
De Giovanni
C
,
Nanni
P
,
Forni
G
,
Lollini
P-L
. 
2011: the immune hallmarks of cancer
.
Cancer Immunol Immunother
2011
;
60
:
319
26
.
36.
Quintana
E
,
Piskounova
E
,
Shackleton
M
,
Weinberg
D
,
Eskiocak
U
,
Fullen
DR
, et al
Human melanoma metastasis in NSG Mice correlates with clinical outcome in patients.
Sci Transl Med
2012
;
4
:
159ra149
.
37.
Tran
E
,
Turcotte
S
,
Gros
A
,
Robbins
PF
,
Lu
Y-C
,
Dudley
ME
, et al
Cancer immunotherapy based on mutation-specific CD4+ T cells in a patient with epithelial cancer
.
Science
2014
;
344
:
641
5
.
38.
Almeida
J
,
Sauce
D
,
Price
D
,
Papagno
L
,
Shin
S
,
Moris
A
, et al
Antigen sensitivity is a major determinant of CD8+ T-cell polyfunctionality and HIV suppressive activity
.
Blood
2009
;
6
:
P246
.
39.
Chiu
YL
,
Shan
L
,
Huang
H
,
Haupt
C
,
Bessell
C
,
Canaday
DH
, et al
Sprouty-2 regulates HIV-specific T cell polyfunctionality
.
J Clin Invest
2014
;
124
:
198
208
.
40.
Schoenberger
SP
,
Toes
REM
,
van der Voort
EIH
,
Offringa
R
,
Melief
CJM
. 
T-cell help for cytotoxic T lymphocytes is mediated by CD40–CD40Linteractions
.
Nature
1998
;
393
:
480
3
.
41.
Kreiter
S
,
Vormehr
M
,
van de Roemer
N
,
Diken
M
,
Löwer
M
,
Diekmann
J
, et al
Mutant MHC class II epitopes drive therapeutic immune responses to cancer
.
Nature
2015
;
520
:
692
6
.
42.
Topalian
S
,
Hodi
F
. 
Safety, activity, and immune correlates of anti–PD-1 antibody in cancer
.
N Engl J Med
2012
;
366
:
2443
54
.
43.
Wolchok
JD
,
Kluger
H
,
Callahan
MK
,
Postow
MA
,
Rizvi
NA
,
Lesokhin
AM
, et al
Nivolumab plus ipilimumab in advanced melanoma
.
N Engl J Med
2013
;
369
:
122
33
.
44.
Topalian
SL
,
Sznol
M
,
McDermott
DF
,
Kluger
HM
,
Carvajal
RD
,
Sharfman
WH
, et al
Survival, durable tumor remission, and long-term safety in patients with advanced melanoma receiving nivolumab
.
J Clin Oncol
2014
;
32
:
1020
30
.
45.
Hamid
O
,
Robert
C
,
Daud
A
,
Hodi
FS
,
Hwu
W-J
,
Kefford
R
, et al
Safety and tumor responses with lambrolizumab (anti–PD-1) in melanoma
.
N Engl J Med
2013
;
369
:
134
44
.