Introduction
The incidence of melanoma has increased over the past three decades [
1]. In addition, this tumor is resistant to standard therapies with a life expectancy of less then 1 year for metastatic melanoma patients [
2]. Tough new targeted therapies [
3] and immunotherapy trials [
4,
5], however, suggest that survival of melanoma patients may be considerably improved in the near future. Ocular melanoma, though infrequent, is a very aggressive disease with a dismal prognosis since 50% of patients die from metastatic disease [
6,
7]. Notably, the biological and molecular profiles of ocular melanoma showed relevant differences compared to cutaneous melanoma, with the first type of tumor commonly developing liver metastasis. For example, the mutated BRAF, which is detected in up to 62% of cutaneous melanoma and represents a novel target molecule [
3], is rarely found in ocular melanoma [
8]. Furthermore, the eye has been considered an “immune privileged” site maintained by a variety of immune suppressive mechanisms (e.g., TGF-β, IL-10, immune cells with negative regulatory functions) that can sustain the development of ocular tumors [
9]. However, immunogenic intraocular tumors leading to the immune-mediated rejection have been documented in animal models [
10‐
12]. It has been reported that the presence of either tumor-infiltrating lymphocytes (TILs) or tumor-infiltrating macrophages (TIMs) is associated with poor prognosis in uveal melanoma patients [
9]. Thus, it remains to be ascertained if immune stimulation may favor or control the progression of ocular tumors. In fact, it has been shown in cutaneous melanoma that tumor cells can produce immuno-modulating factors, which exhaust or block the immune response and recruit immune cells with negative regulatory activity (T regulatory cells, Tregs; myeloid-derived suppressor cells, MDSCs; suppressive macrophages, M2) to a specific site [
13,
14]. Therefore, immunotherapeutic protocols that can modify and rescue anti-tumor immune responses need to be designed for both cutaneous and ocular melanoma patients.
Different active strategies have been used to induce anti-tumor immunity in cancer patients mainly based on the targeting of a variety of tumor-associated antigens (TAAs) by active therapy (vaccination) with peptide, protein or DNA/RNA [
15‐
17]. These vaccination strategies, while increasing systemic anti-tumor immune responses, resulted in limited clinical outcome [
16‐
18], though a recently performed peptide-based phase II vaccination trial documented an increase in the median overall survival in metastatic melanoma patients [
5]. On the other hand, the adoptive T-cell therapy (ACT) can overcome the suppressive tumor milieu by using immune cells with cancer specificity grown outside of the patient and infused in large numbers following pre-conditioning by chemotherapy, alone or in combination with patient total body irradiation [
19‐
21].
ACT with TILs isolated from metastatic melanoma lesions led to objective tumor regression in 49–72% of patients with many long-term sustained responses [
20,
21]. However, TILs can be exploited for ACT studies only in melanoma patients with resectable tumors and from which T cells can be expanded ex vivo (approximately 60–70%) [
20,
22]. A recent published analysis identified the parameters, such as age, sex and the type of systemic therapy, that can negatively influence the ex vivo expansion of TILs [
22].
The persistence in vivo of the infused T cells was highly associated with clinical responses. The in vivo effectiveness of TILs was also associated with the differentiation state of the T cells, telomere length and CD27 expression, factors that are relevant for specific anti-tumor activity [
23‐
25]). These results have been confirmed in a phase II clinical study showing high frequency of objective regressions in metastatic melanoma patients [
26].
An alternative approach has been explored for patients without resectable lesions or with low number of TILs using autologous lymphocytes isolated from the peripheral blood genetically modified with genes encoding for anti-tumor T-cell receptors (TCRs) [
19,
27,
28]. These studies showed encouraging clinical responses by using T-cell receptor (TCR) specific for differentiating (Melan-A/MART-1, CEA) or cancer testis (MAGE-A3, NY-ESO-1) antigens, which could be applied to tumors having different histological origins [
27]. This strategy however permits the targeting of only a single tumor antigen by the engineered T lymphocytes with the risk of tumor variants selection. Further efforts are therefore needed to develop ACT protocols, which employ the use of T lymphocytes with the ability to target a wider range of tumors and have a greater potential of tumor site migration. This would result in treatment for a larger number of cancer patients.
To this aim, we characterized anti-melanoma immune responses of T cells isolated both from the peripheral blood and from the TILs of cutaneous and ocular melanoma patients. As a result, we identified a new protocol that allows for the in vitro enrichment and expansion in large scale of T lymphocytes obtained from PBMCs, which were compared to those obtained with TILs. Furthermore, this protocol benefits from the stimulation with the autologous tumor cells, which in turn leads to the selection of T cells directed against multiple TAAs. Therefore, this protocol would be suitable for ACT clinical studies in the vast majority of both cutaneous and ocular melanoma patients particularly when TILs are not available.
Materials and methods
Cell lines, PBMCs and TILs
Melanoma cell lines were established in vitro from surgically resectable tumor lesions from both cutaneous (#1061, 1067, 2710, 4478D, 49318, 0342, 25368 and 7 mel) and ocular (#2130, 4330, 4022, 1141, 37165, 48409 and 15765) melanoma patients. The informed consent was obtained from all the subjects enrolled in this study. Other cutaneous melanoma lines used for this study were JOFR-IA, DAJU (kindly provided by Dr Pierre Coulie, de Duve Institute, Université Catholique de Louvain, Brussels, Belgium), 501 mel, a gift of Dr. Paul F. Robbins, (National Cancer Institute, NIH, Bethesda, MD) and the 15392 line provided by Dr. Chiara Castelli (Istituto Nazionale Tumori, Milan, Italy). These melanoma lines were cultured in RPMI 1640 (Biowittaker, Lonza, Treviglio, Italy) supplemented with 10% FBS (Lonza), 20 mM HEPES, penicillin (200 U/ml), streptomycin (200 ug/ml) and 2 mM Glutamax (Invitrogen, Carlsbad, CA).
Melanoma cell lines used for the stimulation in vitro of autologous PBMCs were maintained in 48-h cultures with RPMI plus 10% human serum (HS).
Other cell lines used were the lymphoblastoid cell line T2 and HLA-A3-C1R, the erythroblastoid cell lines K562 (American Type Cell Culture, ATCC-LGC, LGC Standards) and the EBV-B cell line 1869 [
29].
PBMCs were obtained by Ficoll density gradient centrifugation (Ficoll-Paque PLUS, GE Healthcare Bio-Science Ab, Uppsala, Sweden) of peripheral blood drawn from both ocular and cutaneous melanoma patients, while TILs were isolated from melanoma lesions. The MHC class I and II typing of PBMCs of melanoma patients was performed by sequence-specific oligonucleotide PCR.
Immunohistochemistry
Surgical specimens from ocular melanoma patients (N = 10 primary uveal melanoma) were fixed in formalin and embedded in paraffin. Immunohistochemistry (IHC) analysis was done on 5-μm tissue sections, and the staining was performed with a sensitive non-biotin detection system (Novo Link polymer, Novacastra) and with diaminobenzidine or double stain with diaminobenzidine and alkaline phosphatase-fast red development. Heat-induced antigen retrieval was done with Tris–EDTA (pH 9.0) in water bath for 30 min. The monoclonal antibodies (mAbs) used were directed against the following antigens: anti-CD3, anti-CD4, anti-CD25 and anti-CD163 (Novocastra), anti-CD8 (DB Biotech), anti-FOXP3 (CNIO Madrid), anti-GATA3 (Becton–Dickinson; BD Biosciences, San Jose, CA, USA) and anti-T-Bet (Santa Cruz Biotechnology, Inc.). Staining was carried out by an automatic immunostainer (Autostainer 480, Thermofisher), and immunostained slides were digitalized with the Aperio (Aperio Technologies, Vista, CA, USA) slide scanner and corresponding 1 mm2 of tumor areas on serial sections was selected. Immunolabeled cells were counted and expressed as percentage on total cells obtained using the IHC Nuclear Image Analysis algorithm of the Spectrum Plus software (Aperio).
Flow cytometry analysis
The HLA class I and II expression by melanoma cell lines was measured by flow cytometry using anti-HLA-A2, HLA-ABC and HLA-DR mAbs conjugated with fluorescein isothiocyanate (FITC), phycoerytrin (PE) and peridinin chlorophyll protein (PerCP)-Cy-5.5 (BD Biosciences), respectively. For melanoma line immuno-phenotyping, the following mAbs were used: anti-MICA, anti-MICB and anti-ULBP1-4 (provided by Amgen, Thousand Oaks, CA, USA), anti-MAGE 57B (provided by Dr. Giulio Spagnoli, University of Basel, Switzerland), anti-NY-ESO-1 E978 (Zymed Laboratories, San Francisco, CA, USA), anti-Survivin (SVV) 8E2 (Thermo Fisher Scientific), anti-Melan-A/MART-1 M2-7C10, anti-gp100 HMB45 and anti-IL13Ra2 B-D13 (Santa Cruz Biotechnology) and the polyclonal Ab anti-COA-1 (Protein Expert; Marseille, France). The goat anti-mouse IgG PE conjugated or the goat anti-rabbit IgG FITC (Dako Italia SpA, Milan, Italy) was used as secondary Abs.
Phenotypic characterization of T-cell cultures was done by the multiparametric flow cytometry analysis using the following mAbs: TCRγ/δ, CD4, CD62L, CD57 and IFN-γ FITC conjugated; TCRα/β, CCR7, CD127, CD107a and perforin PE conjugated; NKG2D, CD27, CD56 and CD137 allophycocyanin (APC) conjugated; CD8 and CD16 APC-H7 conjugated; CD25 APC-Cy7 conjugated; CD134 PE-Cy5 conjugated; CD45RO PE-Cy7 conjugated; CD45RA PE Texas Red (ECD) conjugated; CD28 PerCpCy5.5 and CD3 Pacific Blue (PB) conjugated (BD Biosciences).
Lymphocytes were incubated with mAbs to surface markers at 4°C for 30 min, washed with PBS 5% FBS buffer and fixed with 0.5% paraformaldehyde. Intracellular staining was performed using BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD Bioscience Pharmingen). Data were acquired on the LSRII flow cytometer (BD) and analyzed with FCS Express Software (Denovo Software, Los Angeles, CA, USA) or Kaluza Software (Beckman Coulter, Brea, CA, USA). Results are expressed as MRFI, representing the ratio between the mean fluorescence intensity of cells stained with the selected mAb and that of cells stained with isotype-matched control mouse immunoglobulins or, in the case of multiparametric phenotype analysis, as the percentage of positive cells subtracted above background.
Mixed lymphocytes tumor cell culture (MLTC)
PBMCs isolated from melanoma patients were thawed and incubated overnight at 37°C. PBMCs (106 cells/well) were cultured in the presence of autologous irradiated (150 Gy) melanoma cells at a lymphocyte to tumor ratio of 5:1 in 24-well plates with X-VIVO15 and 5% HS. Different cytokine combinations were added to these MLTCs in order to select the most suitable in vitro culture conditions in terms of T lymphocyte expansion and efficiency in tumor cell recognition: (1) 120 IU/ml rhIL-2 (Proleukin, Novartis Farma, Origgio, Italy); (2) 120 IU/ml rhIL-2 and 10 ng/ml rhIL-15 (Peprotech, Rocky Hill, NJ, USA); (3) 10 ng/ml rhIL-15; (4) 5 ng/ml rh-IL-7 (Peprotech); (5) 120 IU/ml rhIL-2 and 5 ng/ml rhIL-7; (6) 120 IU/ml rhIL-2 and 10 ng/ml rhIL-21 (Peprotech); (7) 10 ng/ml rhIL-21 alone. Fresh medium containing the above indicated cytokines was replaced every 3 days. These MLTCs were stimulated weekly with irradiated autologous melanoma cells, and their reactivity was tested starting from the 3rd week of culture.
ELISPOT assay
To determine the specific recognition of tumor cells by T lymphocytes, the ELISPOT assay was performed as previously described [
30]. Anti-human IFN-γ mAb (1-D1 K), the secondary biotinylated anti-IFN-γ mAb (7-B6-1) and the secondary alkalin phosphatase-streptavidin were purchased from Mabtech (Naka Stand, Sweden). The specificity of T-cell recognition was determined by the inhibition of the IFN-γ release after pre-incubation of target cells with the W6/32 (anti-HLA class I) or L243 (anti-HLA class II DR molecules) mAbs (ATCC). In addition, T lymphocytes were pre-treated with the anti-NKG2D mAb (clone M585, kindly provided by Amgen) before the incubation with target cells. T lymphocytes incubated with phytohemagglutinin (PHA) and concanavalin-A (ConA) (Sigma-Aldrich) were used as a positive control for IFN-γ secretion. The cell lines T2 (HLA-2
+), 1869 EBV-B (HLA-A3
+ and A24
+) and C1R (HLA-A3 +) were used as antigen-presenting cells and pulsed with 10 μg/ml of Melan-A/MART-1-, gp100-, MAGE-A2- and A3-, NY-ESO-1- or SVV-HLA-A2-restricted peptides (JPT Peptide technologies, Berlin, Germany) or MAGE-A1-, gp100- and COA-1-HLA-A3-restricted peptides [
29,
30]. In addition, HLA-A3-C1R cells transiently transfected by electroporation with expression vectors (pcDNA3) coding for COA-1 and SVV antigens were used as target cells as well.
Statistical analysis of the differences between means in the cytokine release assays was performed using two-tailed t test (P < 0.05).
In vitro rapid expansion of T lymphocytes
The rapid expansion of TILs or of MLTC-derived T cells was performed using the rapid expansion protocol (REP) as previously described by Dudley et al
. [
31]. Briefly, T lymphocytes were cultured in T25 flasks in X-VIVO15, 5% HS and in the presence of 200-fold excess of irradiated (50 Gy) feeder cells that were isolated and pooled from three healthy donors. On day 4, 30 ng/ml of anti-CD3 (OKT-3, Ortho Clinical Diagnostics, Rochester, USA) mAb and 6,000 U/ml of rh-IL-2 were added. The culture media containing rh-IL-2 was replaced every 3 days. At day 14 following the in vitro expansion, the specific reactivity by T lymphocytes against autologous and/or HLA-matched allogeneic melanoma lines and melanoma-associated epitopes was determined by IFN-γ release assay (ELISPOT) as described above.
Assessment of the cytotoxic activity
The cytotoxic activity of T lymphocytes was determined by the CD107a mobilization assay [
32] and intracellular detection of perforin. T lymphocytes were co-cultured with autologous or allogeneic HLA-matched tumor cells at a 4:1 ratio in polystyrene tubes. Control tubes contained lymphocytes alone or co-cultured with HLA-mismatched melanoma cells. Positive controls were comprised of cells stimulated with PHA/Con-A or OKT3. CD107a-PE (BD Pharmingen) mAb was added to the T-cell cultures. After 1 h of incubation at 37°C, 1 μl/tube monesin (Golgi-Stop, BD Bioscience) was added as per kit protocol to the cultures then incubated at 37°C for an additional 3 or 5 h. At the end of the incubation, time cells were stained for the surface markers CD3 and CD8, permeabilized and stained with anti-perforin PE and/or anti-IFN-γ FITC mAbs (BD Pharmingen). Samples were then analyzed by flow cytometry as described above.
Discussion
ACT represents a promising therapeutic approach for metastatic melanoma. In fact, the infusion in metastatic patients of autologous TILs has led to significant responses in a high number of clinical cases [
20,
21]. A multicenter confirmatory trial has been proposed to show that this complex technique of ex vivo expansion of TILs can be applied by multiple institutions [
35]. In addition, a simplified protocol for the isolation and the growth in vitro of TILs has been set up to improve both the efficiency of isolation of these T lymphocytes and their anti-tumor potential for ACT studies [
26,
36]. Nevertheless, some features such as age, sex, tumor location and any prior systemic therapy of patients can affect the success in TIL isolation [
22]. Therefore, we sought to identify an alternatively protocol, when TILs are not available, which involves the isolation of polyclonal anti-melanoma T lymphocytes from the PBMCs of both cutaneous and ocular melanoma patients. We have successfully and reproducibly (7/7 patients) isolated anti-tumor circulating T cells by the stimulation in vitro of PBMCs with irradiated autologous tumor cells (MLTCs; Fig.
3 and also Figure 1S of supplementary online data). The MLTCs cultures isolated in vitro corresponded to non-terminally differentiated T
EM [
33,
34] expressing high levels of co-stimulatory molecules, a subpopulation of T cells that can exhibit efficient melanoma-specific reactivity and in vivo persistence [
24,
25].
In few cases (2/6), we could also molecularly identify the TAAs (Melan-A/MART-1 and MAGE-A3) recognized by the tumor-specific MLTC T cells. Nevertheless, the cytokine release in the presence of the autologous tumor recognition was higher compared to the TAA-directed reactivity, suggesting that our protocol can achieve an ex vivo enrichment of T lymphocytes reacting to multiple TAAs, thus avoiding the selection of tumor immune variants. Along this line, it has been documented that TILs, reactive with a broad array of TAAs, isolated from melanoma patients can be successfully exploited for ACT [
20,
37,
38], possibly targeting the heterogeneity of tumor cells.
Very little information is available on the immune-mediated control of ocular melanoma [
10,
11], and in addition, this aggressive disease has limited therapeutic options [
7,
8]. Therefore, immunotherapy-based treatment, as ACT, can represent a promising approach for the control of this disease. Interestingly, the application of our MLTC-based protocol has led to the isolation ex vivo of tumor reactive T cells from one ocular metastatic melanoma patient (Fig.
3 and Figure 1S of supplementary data). In addition, the phenotype of these T cells was of T
EM type with the expression of high level of co-stimulatory molecules, similarly to MLTCs from cutaneous melanoma patients (data not shown and Fig.
5). Of note, these T cells, in addition to IFN-γ secretion (Figs.
3 and 1S), exerted cytotoxic activity (Figs.
4) against the autologous tumor cells. Similar data were obtained for cutaneous melanoma-derived MLTCs (data not shown). Ocular melanoma is a rare disease, and therefore, the availability of tumor cell lines and PBMCs in an autologous setting is limited. However, our encouraging results demonstrate a potential future wider exploitation of our protocol that can be of benefit also for these melanoma patients. Since we could not molecularly identify the TAAs recognized by the anti-ocular melanoma MLTC cultures, we will further characterize these immune responses exploiting T cells as probes for the molecular cloning of the target molecule.
In order to determine whether our MLTC-derived immune responses could be exploited for ACT protocols, we have compared their anti-tumor activity and phenotype with those of TILs. We have isolated TILs from both cutaneous (N = 5) and ocular (N = 4) melanoma patients and found that they specifically exerted autologous tumor recognition (representative data are shown in Fig.
7). The phenotype analysis showed that non-terminally differentiated T
EM can be isolated from both TILs and MLTCs. However, higher levels of co-stimulatory molecules were found in MLTC lymphocytes, suggesting that efficient anti-tumor activity and persistence in vivo can be associated with MLTCs rather than with TILs.
By IHC, we also analyzed the presence in vivo of the immune infiltrate in 10 ocular melanoma surgical samples, where we could detect TILs in 9/10 tissues, mostly CD4
+T cells (6/10 tissues). Only 3 cases showed high numbers of CD8
+ lymphocytes. We found an heterogeneous expression of TH1 and TH2 type T cells, while the detection of CD25
+ and FOXP3
+ Treg cells was directly associated with high level of CD4. Moreover, in all of the analyzed samples (10/10), high levels of CD163
+ macrophages were observed. The presence of these cells in tumor tissues has been described to be correlated with poor prognosis and with suppressive immune functions [
39,
40]. Therefore, our observations indicate that ocular melanoma potentially represent an immune suppressive environment, in accord with previous published data [
9,
10], and similarly to cutaneous melanoma [
13,
14]. However, the ex vivo short-term culture of TILs can rescue their anti-tumor activity (Fig.
7). Thus far, with our protocol based on the ex vivo enrichment of circulating T cells (MLTCs), we have demonstrated that we can overcome the anergic state of T cells induced by the tumor milieu.
We obtained a remarkable high frequency of both long-term (perpetual) and short-term (1–4 months) cell lines in 72 and 57% cutaneous and ocular melanoma, respectively. A low rate of TILs isolation (38%) was observed from our cutaneous melanoma patients as compared to the previous published results (60–70%) [
20,
22]. This difference may depend on the limited population size we have analyzed and/or on the small dimensions (0.5–1.5 cm diameter) of the tumor fragments we have received. In ocular melanoma, there are no documented studies involving the isolation of TILs on a large patient base. Nevertheless, we were able to isolate and successfully culture TILs in 28.5% of our patients. Thus, our data indicate that the application of the MLTC method, which is based on tumor cell availability, can be of interest since it is applicable to a large number of melanoma patients from whom TILs cannot be obtained and, in addition, to ocular melanoma subjects for whom few studies documenting TIL isolation and characterization are available [
9,
12].
Notably, the REP stimulation [
31] applied to the in vitro isolated MLTCs indeed led to the successful expansion of these T cells (Fig.
8). The final number of T cells obtained from MLTCs was comparable to that of TILs (Fig.
8); moreover, we also demonstrated that after the REP, these T lymphocytes retained their initial specific anti-tumor activity (Figure 3S of supplementary data available online). Thus, we identified a new protocol that allows the isolation of large number (2–10 billions) of T lymphocytes to be infused in melanoma patients.
Durable clinical responses have been recently documented in stage IV melanoma patients infused with polyclonal anti-tumor PBMCs, pre-stimulated in vitro with the autologous tumor cells, and given in combination with low-dose IFN-α [
41]. This indicates that by the exploitation of the immunogenicity of tumor cells, it is possible to efficiently isolate and direct systemic anti-tumor immune responses against the residual tumor cells.
A limitation of our approach is the availability of autologous tumor cells, although as we have shown above (Figs.
3 and 1S of supplementary results) that short-term cultured melanoma cells can be used. Alternatively, to overcome the lack of autologous tumor cell lines, HLA-matched/-semimatched allogeneic melanoma lines could be used as stimulators of PBMCs. Indeed, tumor reactive TILs have been selected ex vivo by the usage of allogeneic tumor cells as stimulators [
42].
Altogether, we have carried out a detailed characterization of immune responses of T cells from the peripheral blood vs tumor tissues of both cutaneous and ocular melanoma patients that allowed us identifying a novel MLTC-based protocol to isolate and successfully expand ex vivo polyclonal anti-tumor circulating T cells. In addition, our protocol can overcome the possible anergic state of TILs due to the immunesuppressive tumor environment. We envision that our method may be suitable for ACT protocols, when TILs are not available, for both cutaneous and ocular melanoma patients.
Acknowledgments
We are indebted with Mrs Gloria Sovena (Unit of Immuno-biotherapy of Melanoma and Solid Tumors, San Raffaele Scientific Institute, Milan, Italy) for the technical assistance in the establishment in vitro of melanoma lines and the isolation and expansion of T lymphocytes and with Mrs Ylenia Papa (Unit of Pathology, San Raffaele Hospital, Milan, Italy) for assistance in tissue preparation and staining for IHC analysis. We thank Dr. Katherina Fleischhauer (Unit of Molecular and Functional Immunogenetics, San Raffaele Scientific Institute, Milan, Italy) for the HLA typing analysis of cancer patients, Dr. G.C. Spagnoli (Institute of Surgical Research and Hospital Management, University Hospital, Basel, Switzerland) for providing the anti-MAGE 57B and 6C1 mAbs, and Dr. Pierre Coulie (de Duve Institute, Université Catholique de Louvain, Brussels, Belgium) for providing the JOFR-1A and DAJU melanoma cell lines and PBMCs. We thank Dr. Alessio Palini Unit of Cytometry, San Raffaele Scientific Institute, Milan, Italy for the editorial revision of the manuscript. This work was supported by the Italian Association for Cancer Research (Milan) and by the Alliance against Cancer Project 3 (Rome), Grant to G. Parmiani.