Soluble SLAMF6 Receptor Induces Strong CD8⁺ T-cell Effector Function and Improves Anti-Melanoma Activity In Vivo

One of the approaches used in the search for new immunomodulatory mechanisms is to identify lymphocyte surface receptors that are likely to have a regulatory function. The clinical success of targeting the immunoglobulin superfamily receptors PD-1 and CTLA-4 fueled interest in the uncharacterized members of several structural Ig-like subfamilies (1).

One such family, the SLAM (signaling lymphocytic activation molecules) family of receptors, is typical of the homotypic-binding molecules involved in immunomodulation that are expressed on cells of hematopoietic origin.

Our previous results (2) showed differential expression of SLAMF6 on T-cell clones with identical epitope specificity but different functional capacity. These data sparked our interest in this receptor in the context of antimelanoma CD8⁺ T-cell responses and prompted us to explore whether SLAMF6 is a function-determining receptor or a secondary chaperone.

SLAMF6 is expressed on natural killer (NK), T, and B cells. It augments Th1 responses in CD4⁺ T cells and activates NK cells in a homotypic manner (3, 4). SLAMF6 is linked to immune pathology: the allele responsible for autoimmunity in lupus-prone B6.Sle1b^z/z mice is a spliced isoform of Ly108, the mouse ortholog of SLAMF6 (5).

In this study, we focused on SLAMF6 as a potential target for improving the CD8⁺ T-cell response against melanoma. We show that a 203 amino acid–long polypeptide mimicking the extracellular domain of SLAMF6 exerted a costimulatory effect on activated antimelanoma CD8⁺ T cells. The soluble ectodomain (seSLAMF6) improved CD8⁺ T-cell expansion, cytokine secretion and cytotoxicity against melanoma. In vivo, the reagent supported adoptively transferred antimelanoma CD8⁺ T cells, mediated growth delay or complete tumor clearance, and was readily tolerated. We suggest that by interfering with the homotypic binding of the SLAMF6 receptor, seSLAMF6 has the features of a costimulator and could be relevant in melanoma immunotherapy.

Nude (athymic Foxn1^−/−), C57BL/6, and BALB/c mice were purchased from Harlan Laboratories. Pmel-1 mice carry a rearranged T-cell receptor (TCR) Vbeta13, specific for the 9-mer epitope 25-32 from murine Pmel 17, the homolog of human gp100. Pmel-1 and Pmel-1-GFP were self-bred (kind gift from M. Baniyash, Hebrew University, Jerusalem). All experiments were performed with female 8- to 12-week-old mice.

Cell line 888mel (HLA-A2−/MART-1⁺/gp100⁺; ref. 6) and 624mel (HLA-A2⁺/MART-1⁺gp100⁺) were a gift from M. Parkhurst (Surgery Branch, NCI, NIH, Bethesda, MD, 2004). Human melanoma 526mel is an HLA-A2⁺ cell line. Mouse melanoma B16-F10 and B16-F10/mhgp100 (B16-F10 melanoma cells transduced with pMSGV1 retrovirus that encodes a chimeric mouse gp100 with the human gp100_25-33 sequence (gift from Ken-ichi Hanada, Surgery Branch, NCI, NIH, 2014). The cells were cultured in RPMI1640 supplemented with 10% heat-inactivated FCS, 2 mmol/L l-glutamine, and combined antibiotics (all from Invitrogen Life Technologies). Lines are kept in culture for no more than 14 to 20 days, are monthly tested for TCR-specific cognate T-cell reactivity (7), and are mycoplasma free.

DD154 is a lymphocyte line (HLA-A*0201) transduced to express a TCR cognate for gp100_154-162 nonapeptide (gift from the Surgery Branch, NCI, NIH).

Plasmid pDONR223-SLAMF6 was obtained from Addgene (deposited by William Hahn, plasmid # 23396). The full-length SLAMF6 sequence was inserted into PCDNA3.1-hygromycin plasmid (Invitrogen) using BamH1 and Xba1 restriction enzymes (NBT). 624mel human melanoma was transfected with the PCDNA3.1-hygromycin-SLAMF6 vector using lipofectamine. Hygromycin-resistant melanoma cells were subcloned, and the stably transfected cells were labeled with anti-SLAMF6 Ab and analyzed by flow cytometry.

Peripheral blood mononuclear cells (PBMC) were purified from healthy donors' buffy coats [Hadassah Blood Bank, obtained under Institutional Review Board (IRB) approval].

Microcultures were initiated and expanded from tumor specimens taken from resected metastases of melanoma patients, as described previously (8). Human lymphocytes were cultured in complete medium (CM) consisting of RPMI1640 supplemented with 10% heat-inactivated human AB serum, 2 mmol/l l-glutamine, 1 mmol/l sodium pyruvate, 1% nonessential amino acids, 25 mmol/L HEPES (pH 7.4), 50 μmol/L 2-ME, and combined antibiotics (all from Invitrogen Life Technologies). CM was supplemented with 6,000 IU/mL recombinant human IL2 (rhIL2, Chiron) every other day.

On day 14 after tumor-infiltrating lymphocyte (TIL) initiation, lymphocytes were washed with PBS, resuspended in PBS supplemented with 0.5% BSA, and stained with FITC-conjugated HLA-A*0201/MART-1_26–35 or HLA-A*0201/gp100_209-217 dextramer (Immudex) for 30 minutes at 4°C. Lymphocytes were then incubated with allophycocyanin-conjugated mouse anti-human CD8 (eBioscience) for an additional 30 minutes at 4°C and washed. CD8⁺ lymphocytes, positively stained by dextramer (CD8⁺dextramer⁺ cells), were sorted by a BD FACSAria and directly cloned at one or two cells per well in 96-well plates in the presence of anti-CD3 (30 ng/mL, eBioscience), rhIL-2 (6,000 IU/mL), and 4 Gy-irradiated 5 × 10⁴ allogeneic PBMCs as feeder cells. Five days later, rhIL2 (6,000 IU/mL) was added and renewed every 2 days thereafter. On day 14, the clones were assayed for IFNγ secretion in a peptide-specific manner following their coincubation with T2 cells pulsed with MART-1_26–35 or gp100_209-217 [both commercially synthesized and purified (>95%) by reverse-phase HPLC by Biomer Technology] by ELISA (R&D Systems). The MART-1_26–35– or gp100_209-217–reactive clones were further expanded in a second-round exposure to anti-CD3 (30 ng/mL) and rhIL2 (6,000 IU/mL) in the presence of 50-fold excess irradiated feeder cells. Supplementary Table S1 describes TILs used throughout the study.

Human TILs were expanded by activation with 30 ng/mL of anti-CD3 and PBMCs as feeders for 12 days (9, 10). Where indicated, cells were expanded for 3 to 5 days with plate-bound anti-CD3 (1 μg/mL) in the presence of either IL2 (3,000 IU/ml for TILs and 300 IU/mL for PBMCs) or seSLAMF6 (50 μg/mL), as detailed for each experiment. Fresh medium containing IL2 or seSLAMF6 was added on day 5 and every other day thereafter.

Pmel-1 mouse splenocytes (2 × 10⁶/mL) were activated with gp100_25-33 peptide (1μg/mL) for 6 days. Cells were incubated with either IL2 (30 IU/mL) or seSLAMF6 (50 μg/mL). Fresh medium containing IL2 or seSLAMF6 was added every other day.

Following expansion, human or mouse lymphocytes were washed, counted, and 1 × 10⁵ cells were cultured in CM supplemented with IL2 or seSLAMF6, as specified in each experiment. At the indicated time points, cells were harvested, washed, and labeled with the Annexin V apoptosis Detection Kit (eBioscience), according to the manufacturer's instructions. Cells were analyzed by flow cytometry using a BD LSRII (BD Biosciences). The data were analyzed using FCS Express software (De Novo Software).

The following proteins and antibodies were used: seSLAMF6 (recombinant human protein, extracellular domain fused with polyhistidine tag at the C-terminus, Novoprotein); anti-SLAMF6 (NT-7, Biolegend); anti-CD3 (OKT3, eBioscience), anti-CD8, anti-CD107, anti-CD137, anti-OX40, anti-PD-1, anti-2B4, and anti-GITR (all from eBioscience).

TILs were incubated with increasing concentrations from 10 to 100 μg/mL seSLAMF6 or BSA (negative control) for 45 minutes on ice, washed twice, and labeled with anti-SLAMF6 for 45 minutes on ice. Antibody binding was measured by flow cytometry.

PBMCs (2 × 10⁶) were incubated in 48-well plates for 3 hours at 37°C in the presence of anti-CD3 with either IL2 (300 IU/mL) or seSLAMF6 (25 μg/mL). Cells were collected and RNA was isolated using GenElute Mammalian Total RNA Kit (Sigma, RTN70) according to the manufacturer's protocol. RNA was then transcribed to cDNA using qScript cDNA Synthesis Kit (Quanta, 95047-100). Real-time PCR was performed using PerfeCT SYBR Green FastMIX ROX (Quanta, 95073-012). Primers used were:

• Actin: F 5′- CATTGGCAATGAGCGGTTGG R 5′-AGTGATCTCCTTCTGCATCC

• BCL2: F 5′- CATGCTGGGGCCGTACAG R 5′-GAACCGGCACCTGCACAC

Expanded TILs were washed and cocultured with target melanoma cells for 1 hour at 37°C in the presence of anti-CD107a (eBioscience). Cells were washed and analyzed by flow cytometry.

Expanded PBMCs were harvested and washed twice with PBS. Following fixation and permeabilization (eBioscience protocol), intracellular BCL-2 was labeled with anti–BCL-2 (BioLegend) for 30 minutes at 4°C. Appropriate isotype controls were used. Cells were washed twice with permeabilization buffer, resuspended in FACS buffer, and subjected to flow cytometry.

Intracellular staining of cleaved caspase-3 was performed as described previously (11). Briefly, melanoma cells were stained with Cell Trace Far Red DDAO-SE (Life Technologies), according to the manufacturer's protocol. CellTrace Far Red DDAO-SE–labeled melanoma cells were cocultured with effector lymphocytes for 90 minutes at a 1:1 (or 2:1 for DD154 cells) effector:target ratio. The cells were then washed, fixed, permeabilized (eBioscience protocol), and labeled with rabbit anti-cleaved caspase-3-PE (BD Pharmingen). Appropriate isotype controls were included. Washed cells were subjected to flow cytometry.

TILs were activated with plate-bound anti-CD3 with and without 50 μg/mL of seSLAMF6 for 5 and 15 minutes. Cells were harvested, lysed in the presence of phosphatase and protease inhibitors (Pierce IP lysis buffer with HALT phosphatse and protease inhibitor cocktail), and lysates were analyzed using Human Phospho-Immunoreceptor Proteome Profiler (R&D Systems) according to the manufacturer's instructions. Image was captured using GelDoc (Bio-Rad). Spot intensity was quantified (Image Lab software by Bio-Rad).

Human lymphocytes or mouse splenocytes (1 × 10⁵) were cocultured overnight at a 1:1 ratio with the indicated target melanoma cells. Where indicated, seSLAMF6 50 μg/mL was added to the cultures. Conditioned medium was collected and IFNγ secretion was detected using ELISA (R&D Systems for human and BioLegend for mouse).

Half a million cells were loaded on slides using cytospin, fixed with 4% PFA in PBS, and blocked with PBS supplemented with 20% goat serum. Cells were stained with rabbit anti-SLAMF6 antibody (Proteintech) and Cy3 goat anti-rabbit (The Jackson Laboratory) was used as secondary antibody. Nuclear staining was performed with 1,5-bis{[2-(di methylamino) ethyl]amino}4,8-dihydroxyanthracene-9,10-dione (DRAQ5, Cell Signaling Technology). Images were captured with a Zeiss LSM 5 confocal microscope and analyzed with Zen software (Carl Zeiss).

The Intavis spot synthesis peptide array (Koch Institute, MIT, Cambridge, MA) permits the synthesis of a 20 × 30 array (10 × 15 cm) of up to 600 distinct peptides immobilized on a cellulose solid support. Instrument software allows generation of spot sequences and residue offsets from an intact protein sequence. The synthesized peptides are 15 amino acids long, covering the whole protein sequence, with a 1 amino acid shift. Membrane blocking was performed with 5% BSA in TBST (tris buffered saline containing 50 mmol/L Tris, 150 mmol/L NaCl, and 0.05% Tween 20) overnight, followed by overnight incubation with seSLAMF6 (1 μg/mL) in TBST, overnight incubation with 1 μg/mL mouse anti-his antibody (BioLegend) in TBST and 1 hour incubation with goat anti-mouse HRP (The Jackson Laboratory) in TBST. Development was performed with Clarity Western ECL Substrate (Bio-Rad), and results were visualized using Image Lab software.

Human TILs were expanded for 12 days as described above. At the end of activation, cells were washed and mixed at a 1:1 ratio (1 × 10⁶ cells each) with 526mel and immediately injected subcutaneously into the back of 8- to 9-week-old female nude (athymic Foxn1^−/−) mice. Tumor size was measured in two perpendicular diameters three times per week. Mice were sacrificed when tumors reached 15 mm diameter in one dimension or when the lesion necrotized. Tumor volume was calculated as L (length) × W (width)² × 0.5.

• (i) B16-F10 and B16-F10/mhgp100 mouse melanoma cells (0.5 × 10⁶) were injected subcutaneously into the left and right side of the back of C57BL/6 mice, respectively. Pmel-1 mouse splenocytes (2 × 10⁶/mL) were expanded with 1 μg/mL of gp100_25-33 peptide in the presence of IL2 30 IU/mL. Fresh medium containing IL2 was added every other day. On day 5, cells were labeled with CFSE (12). The following day, 10⁷ cells were adoptively transferred into 500 CGy-irradiated tumor-bearing mice. A total of 0.25 × 10⁶ IU/100 μL IL2 or 100 μg seSLAMF6 were administered intraperitoneally once a day for 2 days. The following day, mice were sacrificed and tumors and spleens were harvested for further analysis by flow cytometry.

• (ii) Melanoma cells were injected as in (i). After 5 to 7 days, the mice were irradiated to 500 CGy. The following day, 5–10 × 10⁶ gp100-specific Pmel-1-GFP lymphocytes (expanded for 6 days with IL2) were adoptively transferred intravenously (0.1 mL total volume) to the tail vein. A total of 0.25 × 10⁶ IU/100 μL IL2 or 100 μg seSLAMF6 were administered intraperitoneally twice a day for 5 days. Mice were sacrificed on day 6, and tumors and draining lymph nodes were harvested for further analysis by flow cytometry.

• (iii) B16-F10/mhgp100 mouse melanoma cells (0.5 × 10⁶) were injected subcutaneously into the back of C57BL/6 mice. After 5 to 7 days, mice were irradiated to 500 CGy. The following day, 10 × 10⁶ gp100-specific Pmel-1 lymphocytes (expanded for 6 days with IL2 or seSLAMF6) were adoptively transferred intravenously (0.1 mL total volume) to the tail vein. IL2 (0.25 × 10⁶ IU/100 μL) or seSLAMF6 (100 μL) were administered intraperitoneally twice a day for 5 days. Tumor size was measured as above. Mouse weight was measured 3 times a week.

Generation of human lymphocyte cell lines was approved by the IRB (Hadassah Medical Organization IRB, approval number 395–16.09.05). Patients and donors gave their informed consent for the production of tumor cultures, release of TIL from resected metastases, and the use of PBMCs for immune evaluation.

All animal studies were approved by the IRB (approvals MD-14-14155-5 and MD-14-13991-5).

Statistical significance was determined by unpaired t test (two-tailed with equal SD) using Prism software (GraphPad). A P value <0.05 was considered statistically significant. Analysis of more than two groups was performed using one-way ANOVA test. Analysis of tumor size was performed by comparing tumor volumes measured on the last day on which all study animals were alive (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001). For each experiment, the number of replicates performed, the number of animals per group and the statistical test used are stated in the corresponding figure legend.

Because SLAMF6 is a homotypic binding receptor, expressed on resting and activated PBMCs (Supplementary Fig. S1A–S1C), a soluble polypeptide consisting of the ectodomain sequence (seSLAMF6) was used for ligation. To confirm that seSLAMF6 associates with the extracellular domain of the naturally expressed human (Supplementary Fig. S1D) and mouse (Ly108; Supplementary Fig. S1E) receptors, we used a binding competition assay. To identify the specific binding sites of the soluble ectodomain on full-length SLAMF6, an overlapping peptide spot array was utilized. seSLAMF6 bound to sequences extending from proline at position 41 to leucine at position 127 (Supplementary Fig. S2). This binding pattern is in agreement with the homophilic interface between two engaging IgV domains, as predicted by crystallography, encompassing 12 residues between Glu-37 and Ser-90 that form the linkage (13).

Initially, we evaluated the effect of seSLAMF6 on activation-induced cell death (AICD), which is the result of vigorous activation of T cells (14). In clinical protocols of adoptive transfer, large amounts of IL2 are used and administered at high doses to overcome AICD. The associated toxicity, however, is substantial and T-cell persistence posttransfer is still unsatisfactory (15, 16). SLAMF receptors were linked to AICD by the observation that in the absence of SLAM adaptor protein (SAP), the SLAM family activating adaptor, apoptosis is impaired and massive proliferation of CD8⁺ T cells ensues (17, 18).

To test the effect of seSLAMF6 on apoptosis and on the function of rapidly proliferating T cells, we used an expansion protocol that involves TCR activation and yields a large increase in cell number (9). Bulk (Fig. 1A, left) and dextramer-selected ("cloned", Fig. 1A, right) CD8⁺ TILs were used. At the end of the activation/expansion phases, lymphocytes were maintained for five additional days with either IL2 or seSLAMF6 added to the culture medium (postexpansion). We hypothesized that in the absence of IL2, T cells would die by neglect.

Indeed, in the absence of IL2, the percentage of viable lymphocytes decreased to 31% and 37%. However, those maintained postactivation in seSLAMF6 had improved viability rates of 85% and 92% for the CD8⁺ clone and bulk TIL412. These values were higher than the viability rates of cells maintained in IL2 in these groups (63% and 65%, respectively). Lymphocytes that were maintained solely in seSLAMF6 both at the expansion and postexpansion phases, and had no IL2 added at any stage of their growth, still showed a high viability rate of 88% and 93% (Fig. 1A and B). The increase in cell number was similar in IL2 and seSLAMF6 expansion (Fig. 1C).

A similar phenomenon was observed using the soluble ectodomain of human SLAMF6 in mouse lymphocytes. Splenocytes from BALB/c and C57BL/6 TCR transgenic Pmel-1 mice displayed improved viability when supplied with seSLAMF6 after anti-CD3 activation or gp100_25-32 peptide stimulation (Fig. 1D). In fact, the viability advantage of seSLAMF6 in mouse lymphocytes was greater than in human cells.

To determine whether the improved survival was related to a direct antiapoptotic effect, Bcl-2 transcript and protein were evaluated by quantitative RT-PCR and flow cytometry, respectively. The addition of seSLAMF6 restored Bcl-2 transcript in activated PBMCs to a significantly higher level than that induced by IL2 (P = 0.002; Fig. 1E). The trend was similar for the protein level, although the difference was smaller (Fig. 1F). In addition to improved viability, the antiapoptotic effect of seSLAMF6 resulted in a numerical expansion of lymphocytes at the end of the 7-day postexpansion phase. Splenocyte numbers increased about 4-fold when grown with seSLAMF6 compared with growth with IL2 (Fig. 1G). Anti-SLAMF6 had no effect on T-cell viability (Supplementary Fig. S3).

Taken together, these data demonstrate that seSLAMF6 reduces AICD and preserves the viability of rapidly proliferating T cells, exceeding the antiapoptotic effect of high IL2 levels, and, in mouse splenocytes, augmenting the T-cell proliferation rate.

Because seSLAMF6 supports T-cell viability better than IL2, we examined whether the soluble ectodomain complies with other features of an agonistic immune modulator.

To this end, human PBMCs were expanded in CM supplemented with seSLAMF6 or IL2. IFNγ secretion was measured following overnight activation with anti-CD3. The results (Fig. 2A) show increased secretion of IFNγ by PBMCs expanded in seSLAMF6 compared with those expanded in IL2 (P = 0.008). Enhanced IFNγ secretion was also observed in mice using Pmel-1 splenocytes expanded for 6 days with murine gp100_25-33 peptide, and rechallenged with B16-F10/mhgp100 mouse melanoma when supplemented with seSLAMF6 compared with IL2 (Fig. 2B).

To evaluate cytotoxicity, CD107a mobilization (Fig. 2C) and caspase-3 cleavage (Fig. 2D–F) were used as indicators of lymphocyte lysosomal degranulation and tumor cell damage, respectively. Cognate (526mel and 624mel, A2⁺) or noncognate (888mel, A2⁻) melanoma lines were incubated with a CD8⁺ clone derived from TIL209 (clone #4) or TCR-transduced DD154 lymphocytes, expanded in seSLAMF6 or IL2. Figures 2C–F show that T cells expanded in seSLAMF6 induced augmented CD107a mobilization in T cells and caspase-3 cleavage in melanoma targets compared with T cells expanded in IL2. Of note, seSLAMF6 did not induce a cytotoxic response against the noncognate 888mel target, demonstrating that specific recognition is required for the costimulatory effect to take place. The addition of seSLAMF6 to IL2 had no synergistic effect on cytotoxic activity (Supplementary Fig. S4).

Using a Winn assay, we assessed the effect of human antimelanoma CD8⁺ T cells expanded in seSLAMF6 and mixed with cognate tumor on the growth of human melanoma in athymic Foxn1^−/− (nude) mice. The A2-restricted TIL 209 was mixed with 1 × 10⁶ 526mel at a ratio of 1:1 and injected subcutaneously into the back of nude mice. TILs expanded in seSLAMF6 were as effective as TILs expanded in high IL2 concentration for preventing melanoma growth. In a control group of TILs that were not supplemented ex vivo by seSLAMF6 or IL2, 3 of 5 mice developed tumors (Fig. 2G).

Overall, these data indicate that activation and expansion of antimelanoma CD8⁺ T cells with seSLAMF6 was more effective than the standard IL2 protocol at augmenting cytokine secretion and cytotoxic capacity.

In the next series of experiments, we aimed to characterize the effect of seSLAMF6 on the expression of other immune checkpoints and differentiation markers in activated CD8⁺ T cells.

Overnight coculturing of TIL209 with 526mel showed that cell subsets coexpressing GITR, PD-1, 2B4 (SLAMF4), and 4-1BB (CD137) increased in lymphocytes expanded in seSLAMF6 (Fig. 3A). Prolonged and exclusive exposure to seSLAMF6 increased the expression of 2B4 in all CD8⁺ T cells (Fig. 3A and B). The improved functional capacity and reduced AICD (Fig. 1A) of seSLAMF6-expanded cells implies that, in this context, overexpression of 2B4 is not an indicator of exhaustion (Fig. 3C). In addition, a larger population of 4-1BB–expressing (Fig. 3D) and PD-1–expressing (Fig. 3E) CD8⁺ T cells was found in seSLAMF6-expanded T cells, whereas OX40 (Fig. 3F) showed lower expression in seSLAMF6-expanded compared with IL2-expanded cells. Figure 3G shows a distinct population of CD4⁺5RO⁺, CD62L^high/CCR7^high double positive lymphocytes that diminished in the presence of IL2 but were retained with seSLAMF6. This phenotype characterizes T cells with a central memory differentiation, which are more likely to persist (19).

In summary, the enhanced response of CD8⁺ T cells that were preexposed to seSLAMF6 was associated with a change in immune receptor expression. The altered pattern may play a role in the improved response or may merely represent the outcome of vigorous activation.

We evaluated whether the improved in vitro viability of activated CD8⁺ T cells that were supplemented with seSLAMF6 would also manifest itself in vivo. To this end, persistence and proliferation of transferred CFSE-labeled Pmel-1 T cells was measured. Labeled splenocytes were administered to melanoma-bearing irradiated mice. Posttransfer, the animals were injected with two doses of intraperitoneal IL2, seSLAMF6, or PBS, 24 hours apart. Mice were sacrificed on day 3, and the spleens were harvested.

Mice treated with IL2 or seSLAMF6 had a higher percent of CFSE-positive cells in their spleens than PBS-treated controls (P < 0.01). The percent was similar in IL2-treated mice and in seSLAMF6-treated mice (Fig. 4A and B). CFSE dilution showed that a majority of Pmel-1 T cells from mice treated with IL2 underwent 2 to 4 divisions, while T cells from seSLAMF6-treated mice were still in their first or second division cycle (Fig. 4C).

The presence of transferred T cells in the tumor and its draining lymph nodes correlates with treatment outcome (20). To detect the effect of seSLAMF6 on lymphocyte localization in draining lymph nodes, Pmel-1-GFP T cells were transferred to C57BL/6 mice bearing established B16-F10 and B16-F10/mhgp100 melanomas on each side of the back. The double tumor model was selected to identify the role of melanoma immunogenicity on T-cell accumulation in the draining lymph nodes. On day 6 after tumor implantation, mice were irradiated to 500 CGy and injected with Pmel-1-GFP T cells. Following transfer, the animals received two daily doses of IL2 or seSLAMF6, for 5 successive days. One day later, mice were sacrificed and tumors and draining lymph nodes were harvested. The tumors were largely necrotic and did not yield lymphocytes, but a distinct population of Pmel-1-GFP lymphocytes was detected in the lymph nodes draining both melanomas (Fig. 4D). A higher ratio of Pmel-1 cells was found in mice treated with seSLAMF6, in lymph nodes of both B16-F10 and the immunogenic B16-F10/mhgp100 tumors (Fig. 4E).

These experiments demonstrate that systemically administered seSLAMF6 was able to sustain adoptively transferred lymphocytes in recipient mice, with enhanced accumulation in tumor draining lymph nodes.

Our next goal was to evaluate the effect of seSLAMF6 on the antitumor activity of adoptively transferred, melanoma-cognate cytotoxic T lymphocytes (CTL) against established melanoma tumors. To assess this, peptide-activated splenocytes from Pmel-1 transgenic mice were administered to irradiated C57BL/6 mice carrying 6-day-old B16-F10/mhgp100 melanoma (Fig. 5A). During the pretransfer (expansion) phase, we compared the standard IL2 protocol with a seSLAMF6-supplemented regimen for the ability to expand and preserve viability and function of proliferating CD8⁺ T cells (Fig. 1C and F). Following transfer, we compared the effect of systemic administration of high IL2 doses and seSLAMF6 on the antimelanoma response of seSLAMF6-expanded Pmel-1 T cells. The standard IL2 expansion/IL2 injection protocol was used as a reference.

All groups of mice receiving Pmel-1 T cells showed a delay in tumor outburst (Fig. 5B and C). However, although tumors eventually grew in all mice receiving lymphocytes expanded in seSLAMF6 and supported in vivo by IL2, the systemic administration of seSLAMF6 to support seSLAMF6-expanded Pmel-1 cells resulted in complete clearance of melanoma in 2 of 5 mice; melanoma did not recur during 80 days of follow-up. The optimized reference protocol of IL2-expanded/IL2-supported adoptive transfer resulted in the longest growth delay and best clearance rate of melanoma (4/7). On day 31, the last day on which all mice were alive (except the group with no injected lymphocytes), tumor size was similar in mice treated exclusively with seSLAMF6 and those treated with the IL2 reference protocol (Fig. 5D). During the 5 days in which the mice received high-dose IL2 injections, they lost 16% of their body weight on average. In contrast, mice receiving seSLAMF6 lost less than 5% of their body weight (Fig. 5E), attesting to the reduced systemic toxicity of seSLAMF6.

In summary, although IL2 is the accepted protocol for preparation and maintenance of adoptively transferred CD8⁺ T cells, seSLAMF6 is a potential alternative that may be better tolerated.

To elucidate the mechanism of action of seSLAMF6, we evaluated the net effect of transactivation of the SLAMF6 receptor in the context of effector–target interaction. To this end, we engineered 624mel melanoma cells, which lack SLAMF6, to aberrantly express the full-length receptor (624mel-slamf6 cells; Fig. 6A). Melanoma cognate TILs (209, 412, 431) were cocultured overnight with aberrantly expressing 624mel-slamf6. IFNγ secretion decreased after stimulation with the 624mel- slamf6 compared with the natural target. The addition of seSLAMF6 to the coculture resulted in increased IFNγ secretion, but did not restore the original reactivity (Fig. 6B). These results imply that SLAMF6 engagement in trans reduces lymphocyte response. In this context, the soluble ectodomain lowers SLAMF6's natural inhibitory role.

To identify the initial signal induced by seSLAMF6, we used the Proteome Profiler array to follow SLAMF6 phosphorylation upon seSLAMF6 engagement. Phosphorylation was measured in anti-CD3–activated melanoma cognate TILs 209 clone #1 and clone #4 and TIL 412 at 0, 5, and 15 minutes after stimulation. The addition of seSLAMF6 was compared with basal CD3 activation.

As shown in Fig. 7, SLAMF6 is phosphorylated in nonactivated TILs. Following CD3 activation, SLAMF6 is dephosphorylated. However, although dephosphorylation induced by CD3 activation is apparent after 15 minutes, the addition of seSLAMF6 induced dephosphorylation after only 5 minutes. Thus, T-cell activation, and, to a greater extent, seSLAMF6 costimulation, induced dephosphorylation of the SLAMF6 receptor.

We conclude that SLAMF6 acts as a CD8⁺ T-cell inhibitory receptor, which is dephosphorylated upon TCR stimulation. The soluble ectodomain of SLAMF6 enhances receptor dephosphorylation acting as an agonistic immune modulator.

In this work, we showed that the soluble ectodomain of SLAMF6, which binds to sequences of the extracellular part of the receptor by ligand mimicry, in vitro and in vivo, presents features of a costimulatory immune modulator. The magnitude of the activating binding effect was substantial and unexpected, in light of the uncertain role attributed to this receptor per se, outside the context of the SLAM family. In fact, it is difficult to assess the net contribution of SLAMF6 to normal immune response because much of the data on this receptor were acquired in disease-prone animal models.

The human disease that drew attention to SLAMF receptors, X-linked lymphoproliferative disorder, indicated a dual function for SLAMF6, as a negative or a positive coreceptor, depending on competition between SAP and phosphatase SHP-1 for binding to the immunoreceptor tyrosine-based switch motif (ITSM) on its cytoplasmic tail (21). Schwartzberg and colleagues showed that although SLAMF6^−/− mice have intact antiviral CD4-mediated responses, the deletion of SLAMF6 repairs the defect that occurs in SAP deficiency (22). The observation that SLAMF6 exerts a negative effect on synapse organization in the absence of SAP was also shown in CD8⁺ T cells. Enhanced diffuse SHP-1 localization was detected in the defective immune synapse, further supporting the concept that in the absence of SAP, SLAMF6 recruits SHP-1, which consecutively dephosphorylates kinases such as Lck, which is required for TCR signaling (23).

Further support for the inhibitory role of SLAMF6 was provided by adoptive transfer of SLAMF6^−/− CD4⁺ T cells, which caused abnormalities related to systemic lupus erythematosus in a susceptible mouse strain (24). The complexity of deciphering the role of SLAMF6 was further exemplified by Veillette and his group, who showed that SLAMF6 is required for NK-cell priming in a phase prior to target encounter, and that the effect of the receptor is exerted in cis, that is, in the same cell or cell populations of the same hematopoietic ancestry (25).

The main enhancing effects of seSLAMF6 in our systems, which focused on CD8⁺ T cells, were improved viability and enhanced cytotoxicity against melanoma targets. We chose to compare SLAMF6 with IL2 because this cytokine, itself a product of strongly costimulated T cells (26, 27), circumvents the need for costimulation and drives CD8⁺ T-cell function to a theoretical maximum. IL2 is also a cytokine approved for use in the clinic, for the preparatory phase and posttransfer support of T cell–based therapies. In the early phases of development, strategies of ACT were mainly based on TILs in melanoma (28). Since then, the field of ACT has expanded to many other cancers, with the introduction of genetically engineered TCRs, leveraging the recognition of unique tumor antigens or surface markers by antibodies (29–31). In the future, it is not unlikely that “off-the-shelf” TCRs and hybrids may become part of cancer care, especially because regressions of epithelial cancers have occurred and mutated oncogenes have been successfully targeted (32). However, the problem remains that even with inserts of optimized costimulatory signals, the resistance of engineered cells to apoptosis still poses a challenge, as rebound overexpression of inhibitory receptors results in decreased function (33), and IL2 is still required (31). Our data suggest that seSLAMF6 can be given systemically and could replace IL2, simultaneously providing costimulation and opposing suppressive signals. The use of seSLAMF6 highlights the difference between a costimulator and a cytokine, as with costimulation only antigen-specific CTLs are subjected to function enhancement, whereas nonactivated T cells remain unaffected.

The mechanism of action of seSLAMF6 is addressed indirectly in our data. The reduced antimelanoma activity of CTLs against SLAMF6-expressing melanoma implies that ligation of this receptor in trans exerts an inhibitory effect. It is possible that the same effect occurs in cis, between lymphocytes. We show that SLAMF6 in nonactivated T cells is in the phosphorylated form. Rapid dephosphorylation, which occurs upon activation, is accelerated by seSLAMF6. We suggest that SLAMF6 acts as a rheostat: close contact between lymphocytes provides binding opportunities for the receptor with extracellular domains of other cells, leading to SLAMF6 phosphorylation. In this state, lymphocytes are less susceptible to activating signals. seSLAMF6 interferes with this autoregulatory status, increasing the magnitude of T-cell functional capacity and making the lymphocytes less susceptible to apoptosis.

Although inhibitory receptors can be targeted by mAbs, our literature review showed little study of SLAMF6-targeting antibodies that augment immune response. Only Valdez and colleagues, who used a mixture of mAbs against human SLAMF6 in CD4⁺ T cells, showed improved IFNγ secretion in naïve T cells, but not in memory cells (3). From the peptide binding map, we conclude that seSLAMF6, which is a monomer at relevant concentrations, adheres to the native receptor in a spatial conformation that may not be reproducible by an antibody. In this way, seSLAMF6 suppresses an autoregulatory effect, unleashing an armed phenotype in addition to external TCR triggering.

In summary, the soluble ectodomain of SLAMF6 activates CD8⁺ T cells in a manner that complies with the characteristics of costimulation: when signal 1 is generated, lymphocytes exposed to seSLAMF6 display improved capacity to secrete IFNγ and lyse tumor targets. Increased expression of BCL-2, induced by seSLAMF6, combats activation-induced cell death, and as a result sustains lymphocyte proliferation and numerical expansion. When given to melanoma-bearing mice, antigen-specific CTLs supported in vivo by seSLAMF6 inhibit melanoma growth and circumvent the need for cytokine support. This highlights seSLAMF6 as an attractive component in strategies of adoptive T-cell transfer.

A. Machlenkin is the head of Immuno Oncology at Compugen Ltd. M. Lotem has received speakers bureau honoraria from BMS and is a consultant/advisory board member for MSD. No potential conflicts of interest were disclosed by the other authors.

Conception and design: G. Eisenberg, R. Uzana, A. Machlenkin, M. Lotem

Development of methodology: G. Eisenberg, R. Uzana, A. Rutenberg

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Eisenberg, R. Engelstein, A. Geiger-Maor, E. Hajaj, A. Rutenberg

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Eisenberg, R. Engelstein, A. Geiger-Maor, E. Hajaj, S. Frankenburg, G. Frei, T. Peretz, M. Lotem

Writing, review, and/or revision of the manuscript: G. Eisenberg, R. Engelstein, A. Geiger-Maor, S. Merims, S. Frankenburg, M. Lotem

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Eisenberg, S. Frankenburg

Study supervision: G. Eisenberg, M. Lotem

This work was supported by research grants from Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (AMRF); the Canadian Institutes of Health Research (CIHR), the International Development Research Centre (IDRC), the Israel Science Foundation (ISF), the Azrieli Foundation; Melanoma Research Alliance; Deutsche Forschungsgemeinschaft (DFG); and The Israel Cancer Research Fund, Rosetrees Trust, and the Perlstein Family Fund.

The authors wish to acknowledge the devoted technical work of Inna Ben David, Anna Kuznetz and Yael Gelfand. We thank Andre Veillette, Ofer Mandelboim, Eli Pikarsky, Philipp Beckhove, and Tillmann Michels for helpful discussions. Special gratitude to Stewart Greenberg for his endless support of this work.

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.