SLAMF6 as a Regulator of Exhausted CD8⁺ T Cells in Cancer

T-cell exhaustion, associated with the inability to mount successful antitumor immune responses, has increasingly become a target for immunotherapy. Programmed cell death 1 protein (PD-1, CD279) is well characterized as an inhibitory receptor that is upregulated during immune evasion (1–5). PD-1 expression under physiologic conditions controls the magnitude of T-cell effector functions upon activation (6). Within the tumor microenvironment, however, overexpression of the PD-1 ligand, PD-L1, on the surface of tumor cells increases the likelihood of its binding to PD-1 on T cells, which, in turn, suppresses T-cell effector functions (6, 7). PD-1 expression is also associated with an exhausted state of T cells. Blocking the PD-1/PD-L1 axis has been demonstrated to be an effective way to remove the break on these otherwise suppressed T cells, restoring their cytotoxic capacity in the setting of advanced solid tumors, as well as in leukemias (8, 9).

Signaling lymphocyte activation molecule family 6, SLAMF6 (CD352, Ly108, NTB-A), is a homophilic cell-surface receptor, belonging to the immunoglobulin superfamily (10, 11). SLAMF6 is a type I transmembrane protein with two extracellular immunoglobin (Ig)-like domains and three cytoplasmic tyrosine-based signaling motifs, one of which is immunoreceptor tyrosine-based switch motif (ITSM; refs. 10, 12). The SLAMF6 receptor is expressed on the surface of a wide variety of hematopoietic cells, e.g., T, B, and natural killer (NK) cells (expression restricted to human), and interactions on different cell types allow for diverse immunomodulatory functions, some of which include adhesion, innate T-lymphocyte development, neutrophil function, and NK and CD8⁺ T cell–mediated cytotoxicity (13–22). Upon phosphorylation of the immunoreceptor, the two ITSMs, the SH2 domain-containing T- and NK cell adaptor SLAM-associated protein (SAP) is recruited to the SLAMF6 cytoplasmic tail (12, 21, 23). Upon engagement of SLAMF6, the ensuing signaling induces both cooperation between T follicular helper cells and germinal center B cells (18), as well as interactions between cytotoxic T-cell and B-cell targets (23). This process is dependent upon the presence of SAP. By contrast, in the absence of SAP, SLAMF6 negatively regulates both processes by recruiting the tyrosine phosphatases SHP1 and SHP2 to its cytoplasmic tail.

Because B cells do not express SAP, we previously hypothesized that triggering SLAMF6 would negatively regulate B-cell responses. Indeed, the monoclonal antibodies 13G3 and 330 directed against mouse SLAMF6 (24) react with TCL1-192 cells, a CD5⁺ chronic lymphocytic leukemia (CLL) B-cell clone that had been transferred into SCID mice (25, 26). One injection of the antibody limits the expansion of TCL1-192 due to antibody-dependent cellular cytotoxicity (ADCC) and downregulation of B-cell receptor (BCR) signaling (26). Based on these observations, we then hypothesized that antibodies directed against SLAMF6 should also affect the immunomodulatory action of cytotoxic T cells that respond to Eμ-TCL1 CLL cells. Here, we showed that anti-SLAMF6 increases the CD8⁺ T-cell responses to CLL, resulting in a significantly reduced proportion of exhausted cytotoxic T lymphocytes (CTL). We also demonstrated that anti-SLAMF6 arms CD8⁺ T cells in not only B-cell leukemias but also nonhematopoietic solid tumors such as B16 melanoma, where the tumors do not express SLAMF6. Thus, anti-SLAMF6 increases CTL responses to and affects the expansion of both the leukemic cells and solid tumors.

C57BL/6J (B6) wild-type (WT) mice were obtained from The Jackson Laboratory. Eμ-TCL1 mice (25) were kindly provided by Dr. Amy Johnson (Ohio State University, Columbus, OH). SLAMF6^−/− B6 mice were generated from Bruce4 ES cells (27). SAP^−/− mice were generated as described elsewhere (28). All animals were maintained under specific pathogen-free conditions at the Beth Israel Deaconess Medical Center (BIDMC; Boston, MA) animal facility. Experiments were performed in accordance with the guidelines and with the approval of the institutional animal care and use committee (IACUC) at BIDMC.

Frozen peripheral blood mononuclear cells (PBMC) provided by Dr. Jan Burger at The University of Texas MD Anderson Cancer Center (Houston, TX) were used for all experiments. Information on patient samples is provided in Supplementary Table S1. Written patient consent for samples used in this study was obtained in accordance with the Declaration of Helsinki on protocols that were approved by the institutional review board at the BIDMC and at The University of Texas MD Anderson Cancer Center.

Briefly, frozen CLL PBMCs were thawed, washed, and resuspended in RPMI-1640 (Gibco) + 10% fetal bovine serum (FBS) + 1× penicillin/streptomycin (Gibco) + 1× L-glutamine (Gibco). This same medium was used in all in vitro experiments. Fifty to 100,000 cells were either left unstimulated or stimulated with goat anti-human F(ab′)₂ IgM (20 μg/mL; MP Biomedical) alone, together with 1 μmol/L ibrutinib (ChemieTek) or anti-human SLAMF6 (10 μg/mL; BioLegend), or in combination with both. Cells were cultured for 24 and 48 hours, and viability was determined by Annexin V/propidium iodide (PI; BioLegend) staining by BD LSRII flow cytometry.

Degranulation was assessed in a 96-well U-bottom plate coated with anti-CD3 (5 μg/mL; OKT3, Bio X Cell) and anti-CD28 (5 μg/mL; CD28.2, BioLegend). Some wells were also coated with anti-human SLAMF6 (5 μg/mL; NT-7, BioLegend) or mouse IgG2b isotype control (5 μg/mL, BioLegend). CLL B cells were isolated from PBMCs by using CD19 microbeads (Miltenyi Biotec), and the remaining cells were plated as 2.5 × 10⁵ cells/well. Cells were then stimulated in the presence of 2 μL of anti-CD107a. After 30 minutes at 37°C and 5% CO₂, brefeldin A (1:1,000; BD Biosciences) was added, and incubation continued for 4 hours. Cells were washed twice with cold PBS and stained for extracellular markers for 30 minutes at 4°C in FACS-staining buffer [PBS, 1% BSA (Sigma-Aldrich), 2 mmol/L EDTA]. After two washes, cells were resuspended in FACS-staining buffer and analyzed by using a Cytoflex cytometer (Beckman Coulter). Data were analyzed by using Cytexpert software (Beckman Coulter). Antibodies used for these experiments are listed in Supplementary Table S1.

Fresh PBMCs were obtained from 5 healthy donors who provided written consent for the study (BIDMC). The Lymphoprep separation kit (STEMCELL Technologies) was used to isolate PBMCs from whole blood. Cells were then washed with PBS and counted. Ninety-six–well plates were coated with anti-CD3 and anti-CD28 (1 μg/mL each), and cells were plated at 2.5 × 10⁵ cells/well in RPMI medium. Every 2 days, cells were resuspended, washed, and transferred to a new 96-well plate coated with the same concentration of anti-CD3 and anti-CD28. On day 8, cells were collected and checked for exhaustion markers via flow cytometry, and degranulation assays were performed with isotype or anti-human SLAMF6 (5 μg/mL), as explained above.

Fresh or frozen splenocytes (15 × 10⁶–20 × 10⁶ per mouse) from 12- to 14-month-old leukemic Eμ-TCL1 mice were injected intraperitoneally (i.p.) into 8- to 12-week-old B6 WT or SLAMF6^−/− recipients. Mice were irradiated at 400 Rad prior to leukemic transplantation. For experiments involving anti-SLAMF6 or isotype injections, mice were bled via tail vein biweekly for 4 weeks after transfer. When leukemic burden (percentage of CD19⁺CD5⁺) reached 20% to 40% in blood, mice were randomized and injected i.p. with 200 μg/mouse anti-mouse SLAMF6 (13G3; ref. 18; anti-SLAMF6 is an agonistic antibody that can induce positive or negative signaling depending on the recruitment of its adaptor SAP or SHP1/2 on its cytoplasmic tail) or mIgG2a isotype control (clone C1.8; Bio X Cell; n = 16 for each group). Mice were injected a total of 3 times, biweekly, and were euthanized 1 week after the third injection. Cells from the peritoneal cavity (PerC) were collected by flushing 10 mL PBS + 2% FBS and subsequent drawing of the fluid from the PerC. Bone marrow cells were flushed from the femurs with 5 mL PBS + 2% FBS. Single-cell suspensions were prepared from spleen, and red blood cells (RBC) were removed by RBC lysis buffer (Sigma). Serum was obtained by centrifugation at 8,000 × g for 10 minutes at room temperature (RT).

The B16-OVA cells were kindly donated and authenticated by Dr. James Mier (Division of Hematology/Oncology at the BIDMC). The cells were cultured in RPMI medium with 10% FBS, and during the expansion phase, 10⁵ B16-OVA cells were injected subcutaneously into each B6 WT mouse. Eight days after injection, 100 μg/mouse anti-SLAMF6 or isotype control was i.p. injected, and tumor volume was monitored daily by measuring 3 diameters using a caliper. On day 15, mice were euthanized, solid tumors were extracted, and lymphocytes were isolated by digesting the tumors with collagenase IV (Gibco) for 30 minutes at 37°C (29).

Antibodies used for flow cytometry are listed in Supplementary Table S2. Surface staining was done by first incubation with 20% rabbit serum (Rockland Immunochemicals) + Fc block (BioLegend) for 10 minutes at RT, and the prepared antibody cocktails were incubated with cells for 30 minutes at 4°C. Samples were washed twice with FACS buffer (PBS + 2% FBS) and analyzed on a five-laser BD LSRII analyzer (BD Biosciences). Viable cells were gated as diaminidino-2-phenylindole (DAPI; Thermo Fisher) negative, and doublets were excluded. Results were analyzed by FlowJo software.

Fresh splenocytes isolated from mice were cultured with phorbol myristate acetate (PMA; 50 ng/mL) and ionomycin (Sigma-Aldrich; 1 μg/mL) for 4 hours in the presence of brefeldin A (1 μL/mL) at 37°C with 5% CO₂. After 4 hours, cells were washed twice with PBS, and cells were stained for CD3 and CD8, as mentioned in flow cytometry section. Next, cells were fixed and permeabilized using BD Cytofix/Cytoperm kit, according to the manufacturer's protocol, and intracellular staining for granzyme B, IFNγ, and IL2 was performed. For CD107a staining, the antibody was added to the culture in the beginning for cell-surface staining of CD107a upon degranulation, as described above.

Liver sections were harvested at the end of the experiment and were fixed in 10% formalin and stained with hematoxylin and eosin (H&E) at the BIDMC histology core. Assessment was done by Dr. Atul Bahn at Massachusetts General Hospital (Boston, MA).

Statistics for normally distributed mouse data sets were done by an unpaired Student t test, whereas for nonnormally distributed sets, a two-sided Mann–Whitney U test was used. For human CLL data, a nonparametric Wilcoxon signed-rank test was used. Values reported as mean ± SD, and analyses were carried out using GraphPad Prism.

To assess the role of SLAMF6 on the expansion of murine CLL cells, we transferred SLAMF6⁺ Eμ-TCL1 into irradiated SLAMF6^−/− or WT B6 mice. When recipient mice were euthanized 35 days after transfer, no significant difference in the spleen size or total number of splenocytes was detected (Supplementary Fig. S1A and S1B). As judged by the percentage of CD19⁺CD5⁺ TCL1 cells among lymphocytes in spleen, blood, PerC, and bone marrow, leukemic infiltration was unaffected by the absence of SLAMF6 interactions in the tumor microenvironment (Supplementary Fig. S1C). No difference in the percentages of CD3⁺ T cells (CD3: 6.1% ± 0.2% vs. 7.5% ± 0.6%) was observed between the two strains of recipient mice (Supplementary Fig. S1D). SLAMF6 was expressed at comparable levels among T-cell subsets from WT recipients (Supplementary Fig. S2A and S2B).

The spleens of SLAMF6^−/− recipients harbored significantly increased numbers (1.2 ± 0.4 × 10⁶ vs. 1.7 ± 0.2 × 10⁵, P < 0.0001) and proportions (4.5 ± 0.7 vs. 13.1 ± 3, P = 0.0019) of PD-1⁺CD3⁺CD44⁺CD8⁺ antigen-experienced T cells compared with WT mice (Fig. 1A). Because this suggested that the presence of SLAMF6 could negatively regulate CD8⁺ T-cell exhaustion, we assessed relevant effector functions. The CTL effector functions of CD8⁺ T cells isolated from SLAMF6^−/− recipients were impaired, as measured by CD107a [23.2% ± 3.3% vs. 18.5% ± 0.7%, not significant (ns)], and intracellular granzyme B (41.9% ± 2.5% vs. 28% ± 1.7%, P = 0.002) and IFNγ (46.5% ± 0.9% vs. 38.2% ± 1.8%, P = 0.003), whereas IL2 production remained unchanged (Fig. 1B and C). An increase in CD8⁺ T cells was reflected by a diminished CD4⁺/CD8⁺ ratio in SLAMF6^−/− mice as compared with the WT recipients (CD4⁺/CD8⁺ ratio 1.7 ± 0.12 vs. 2.2 ± 0.18, P = 0.04; Fig. 1D). Within CD4⁺ and CD8⁺ T-cell subsets, there was no significant shift in naïve (CD62L⁺CD44⁻) and effector/memory (CD62L⁻CD44⁺/CD62L⁺CD44⁺) subsets (Fig. 1E and F).

It is well recognized that in T cells, SAP is recruited to the cytoplasmic tail of SLAMF6 upon SLAMF6–SLAMF6 ligation, followed by initiation of a SLAMF6-related signaling network (23). To test whether SAP signaling was essential for expansion of exhausted CD8⁺ T cells, we transferred TCL1 cells into irradiated WT and SAP^−/− mice. We found no changes in leukemic burden (Supplementary Fig. S3A) or in the abundance of exhausted T cells in the absence of SAP (Supplementary Fig. S3B–S3D), suggesting that the PD-1⁺CD8⁺ response to TCL1 is SAP independent. Taken together, these findings suggest that expression and signaling through SLAMF6 negatively controls expansion of PD-1⁺CD44⁺CD8⁺ T cells and subsequent T-cell exhaustion in recipients of Eμ-TCL1 CLL cells.

To evaluate whether anti-SLAMF6–induced signaling would have an effect on T-cell responses and leukemic expansion, we transferred tumor-bearing splenocytes from Eμ-TCL1 mice into WT mice. When TCL1 cell infiltration in the blood approached 20% to 40%, mice were randomized into 1 of 2 groups to receive once a week i.p. anti-SLAMF6 or IgG2a isotype control (Fig. 2A). One week after the third injection, the recipient mice were sacrificed and analyzed for expansion of the transferred Eμ-TCL1 CLL cells. Spleen weight and the total number of splenocytes of anti-SLAMF6–injected recipient mice were significantly reduced compared with isotype control mice (Fig. 2B and C, left; spleen weight 0.95 ± 0.04 vs. 0.38 ± 0.1 g, P = 0.003; total splenocytes 6.1 ± 1.03 × 10⁸ vs. 0.74 ± 0.16 × 10⁸, P < 0.0001). Leukemic cell infiltration in the spleen was also significantly reduced, as judged by the percentage of CD19⁺CD5⁺ lymphocytes (79.7% ± 1.7% vs. 37.3% ± 5.1%, P < 0.0001; Fig. 2C, right). Consistent with this, anti-SLAMF6–treated recipient mice exhibited a significantly reduced number of TCL1 cells in the blood (73.4% ± 2.3% vs. 38.5% ± 3.8%, P < 0.0001) and infiltrating the bone marrow (5.5% ± 1.3% vs. 0.9% ± 0.1%, P < 0.0001; Fig. 2D). Administering anti-SLAMF6, as judged by H&E staining, also diminished tumor infiltration in the liver (Fig. 2E). Eμ-TCL1 CLL cells in the spleen and blood of anti-SLAMF6–injected mice appeared significantly more proapoptotic compared with isotype-injected mice, as measured by the percentage of Annexin V⁺ TCL1 cells (Fig. 2F). We concluded that leukemic cell expansion and viability were significantly reduced by administering anti-SLAMF6 to Eμ-TCL1 CLL cell–bearing B6 mice.

Administering anti-SLAMF6 to Eμ-TCL1 CLL–bearing WT mice caused a shift among splenic CD8⁺ T cells from naïve (CD62L⁺CD44⁻) to antigen-experienced effector memory (eff/mem; CD62L⁺CD44⁺) phenotype (naïve 30.5% ± 4.1% vs. 14.5% ± 3.1%, P = 0.02; eff/mem 45% ± 5% vs. 64.8% ± 4.3%, P = 0.02; Fig. 3A). This shift in CD8⁺ T-cell subsets was also observed in the blood (naïve 30.6 ± 3.7% vs. 8.9% ± 2.6%, P = 0.0008; eff/mem 54.9% ± 4.5% vs. 80.8% ± 3.9%, P = 0.001; Fig. 3B). This increase in the percentage of CD8⁺ T cells resulted in an increased proportion of CD3⁺ T cells in the spleen and blood (spleen: 6.6 ± 0.7 vs. 15.3 ± 2.3%, P = 0.0002; blood: 7.6 ± 0.9 vs. 17.4%, P = 0.01) of anti-SLAMF6–treated tumor recipients (Fig. 3C). This was also reflected in a significantly reduced CD4⁺/CD8⁺ ratio in the spleen of anti-SLAMF6–injected mice (spleen: 1.18 ± 0.1 vs. 0.74 ± 0.17, P = 0.04), but not in the blood (blood: 0.68 ± 0.06 vs. 0.61 ± 0.2; Fig. 3D). CD4⁺ T-cell subsets, i.e., CD62L⁺CD44⁻ or CD62L⁺CD44⁺, were not affected by the anti-SLAMF6 (Supplementary Fig. S4).

Because exhausted CD8⁺ T cells display a loss of effector function in CLL patients (1), we further analyzed the in vivo effect of anti-SLAMF6 on CD8⁺ T-cell exhaustion. The percentage of activated CD3⁺CD44⁺CD8⁺ T cells in the spleen that expressed the well-known marker PD-1 in the anti-SLAMF6–injected group during the response to TCL1 cells was significantly lower compared with that in isotype control mice (3.1% ± 0.5% vs. 11.7 ± 1.6, P = 0.0002; Fig. 4A, left), and the absolute number of PD1⁺CD3⁺CD44⁺CD8⁺ T cells was also reduced in the spleen (0.1 ± 0.03 vs. 0.6 ± 0.1 × 10⁶, P = 0.0003; Fig. 4A, right). This effect of anti-SLAMF6 was also observed in the blood (5% ± 1% vs. 11.1% ± 1.2%, P = 0.001; Fig. 4B). In addition to PD-1, the proportion of CD3⁺CD44⁺CD8⁺ T cells expressing additional markers of exhaustion, CD160, LAG3, and KLRG1, was also significantly reduced in the spleen after the injection of anti-SLAMF6 (Fig. 4C).

To test whether the phenotypical changes in the exhausted CD8⁺ T-cell compartment of anti-SLAMF6–injected mice correlated with an increase in effector functions, splenocytes isolated from anti-SLAMF6–injected mice were in vitro stimulated with PMA/ionomycin in the presence of brefeldin A for 4 hours. Intracellular staining of CD8⁺ T cells showed significantly increased lysosomal CD107a (6.7% ± 0.3% vs. 8.8% ± 0.50%, P = 0.003) and granzyme B (14.3% ± 0.6% vs. 21.8% ± 1.5%, P = 0.001; Fig. 4D), and IFNγ-expressing (35.5% ± 2.9% vs. 46.8% ± 1.7%, P = 0.003) and IL2-expressing (8.7% ± 0.9% vs. 12% ± 0.5%, P = 0.009) CD8⁺ T cells were significantly increased (Fig. 4E). From these data, we concluded that SLAMF6 is a negative checkpoint inhibitor, which restricts CD8⁺ T-cell exhaustion in response to murine CLL cells. Consequently, signaling induced by anti-SLAMF6 reduced the number of exhausted CD8⁺ T cells, which unleashes CTL responses against leukemic cells.

Previously, we have shown that anti-SLAMF6 fails to reduce the leukemic burden in the PerC of SCID mice into which the aggressive CLL clone TCL1-192 had been transferred (26). Upon administering anti-SLAMF6, a significantly reduced number of leukemic cells (3.1 ± 0.68 vs. 1.1 ± 0.26 × 10⁷, P = 0.01), as well as a significantly reduced percentage of leukemic engraftment (91.4% ± 1.1% vs. 79.8% ± 2.7%, P = 0.001), was found in the PerC of WT recipients of Eμ-TCL1 cells (Fig. 5A). Similar to leukemic cells in spleen and blood, TCL1 cells residing in PerC of anti-SLAMF6–injected mice were significantly more proapoptotic compared with isotype-injected recipients (Fig. 5B).

Knowing that ADCC does not work in the PerC (30), we speculated that this reduction in leukemic engraftment might be due to activation of T cells. We found that the percentage of T cells had increased (1.5% ± 0.3% vs. 2.8% ± 0.4%, P = 0.09) and the CD4⁺/CD8⁺ ratio had decreased in anti-SLAMF6–injected mice compared with the control group, although it did not reach statistical significance (Fig. 5C). Although we did not find a change in CD4⁺ T-cell subsets (Supplementary Fig. S4), we found a shift from naïve to eff/mem phenotype in CD8⁺ T cells in the PerC of anti-SLAMF6–injected mice (naïve 38.7% ± 4.1% vs. 11.8% ± 2.04%, P = 0.0004; eff/mem 52.5 ± 4 vs. 75.8 ± 1.9, P = 0.0009; Fig. 5D). These activated CD8⁺ T cells seen in the PerC with anti-SLAMF6 treatment led to increased frequency of PD-1 expression (percentage: 9.5 ± 1.1 vs. 3.6 ± 1.9, P = 0.03; absolute number: 1.7 ± 0.4 vs. 0.5 ± 0.3 × 10⁴, P = 0.09; Fig. 5E). Because PD-1 expression can be correlated with either activation or exhaustion, it is plausible that the increase in PD-1⁺CD8⁺ T cells in the PerC may be a sign of activation at the time of analysis. In conclusion, as a single agent, anti-SLAMF6 reduced leukemic burden and engraftment not only in spleen, blood, and bone marrow but also in the PerC through activation of cytotoxic CD8⁺ T cells.

Although SLAMF6 expression is restricted to hematopoietic cells, we next asked whether anti-SLAMF6 could also empower the CTL response against nonhematopoietic tumors. To this end, we subcutaneously transplanted SLAMF6-negative B16 melanoma cells into B6 WT mice. Eight days after transplantation, mice were injected i.p. with 100 μg/mouse anti-SLAMF6 or isotype control. Tumor volume was monitored daily, and mice were euthanized on day 15 for analysis of tumor-infiltrating lymphocytes (TIL; Fig. 6A). As measured by solid tumor weight, mice injected with anti-SLAMF6 had significantly smaller tumors compared with isotype controls (0.8 ± 0.2 vs. 0.25 ± 0.07, P = 0.01; Fig. 6B). Upon analysis of CD45⁺ lymphocytes in the tumor, we found significantly less percentages of CD8⁺PD-1⁺ TILs in anti-SLAMF6–injected mice compared with isotype control, suggesting activation of CD8⁺ T cells in the tumor (11.3 ± 1.6 vs. 2.9 ± 1.03, P = 0.004; Fig. 6C). These findings suggest that anti-SLAMF6 is applicable not only to B-cell tumors but also to other cancer types.

Next, we wanted to test whether our findings using anti-mouse SLAMF6 on exhausted CD8⁺ T cells could also be observed using a human monoclonal antibody. To do this, we chronically stimulated PBMCs with anti-CD3/anti-CD28 three times and on day 8, generated exhausted CD8⁺ T cells in vitro from healthy donors. Compared with freshly isolated PBMCs, we found significantly increased CD8⁺PD1⁺ T cells with chronic stimulation (Fig. 7A). We then checked the effect of mouse anti-human SLAMF6 (mαhSLAMF6) on the degranulation capacity of fresh and exhausted CD8⁺ T cells. Cells were stimulated in vitro with anti-CD3 and anti-CD28 along with isotype or anti-hSLAMF6 for 4 hours, and degranulation was measured by CD107a staining. Comparing fresh and exhausted CD8⁺ T cells receiving isotype antibody showed significantly reduced degranulation capacity in exhausted CD8⁺ T cells compared with fresh cells (Fig. 7B, left). Although mαhSLAMF6 had no effect on the degranulation of fresh CD8⁺ T cells, it significantly increased degranulation of exhausted CD8⁺ T cells (Fig. 7B, middle and right). This suggested that signaling through SLAMF6 may be altered in exhausted T cells and could be reversed using a monoclonal antibody.

To begin to translate our results for therapeutic purposes, we set up in vitro experiments with a humanized anti-human SLAMF6 (αhSLAMF6). First, we tested the capacity of the antibody on CD8⁺ T cells of CLL patients. When T cells were stimulated in vitro with anti-CD3/anti-CD28 in the presence of isotype or αhSLAMF6 for 4 hours, CD8⁺ T cells had significantly increased degranulation capacity with αhSLAMF6 treatment, as judged by CD107a staining (Fig. 7C). Degranulation from CD8⁺ T cells of CLL patients was comparable to that of exhausted T cells generated in vitro from healthy donors (Fig. 7A and B), suggesting that exhausted CD8⁺ T cells could be activated through a SLAMF6 signaling–related mechanism. This also demonstrated that the effect of anti-human SLAMF6 appears to be comparable irrespective of the Fc being mouse or human.

Because we know that SLAMF6 has an effect on BCR signaling, we determined the viability of human CLL cells with αhSLAMF6 in the presence of and in combination with ibrutinib. We used PBMCs from IGHV-unmutated CLL patients with varying levels of IgM surface expression (Supplementary Table S1). Ibrutinib alone served as a positive control for the system, as its interference with BCR-induced survival in vitro is well established (31). CLL cells were either stimulated with anti-human IgM or left unstimulated, and ibrutinib, anti-hSLAMF6, or both were added to the cells. CLL viability was determined as a percentage of Annexin V/PI-negative cells 24 or 48 hours after stimulation (Fig. 7D). We found that upon BCR stimulation, the addition of anti-hSLAMF6 alone reduced viability to comparable levels as those of ibrutinib, and this was enhanced if the two were combined, suggesting that the improved effect of combining anti-SLAMF6 and ibrutinib we observed in vivo also applies to human CLL cells in an in vitro setting.

Overcoming immune evasion, a phenomenon by which pathogens and cancer cells escape the host immune system, is being actively pursued to identify immune checkpoints for therapeutic targeting. Chief among the successfully exploited mechanisms is the well-characterized PD-1/PD-L1 axis. Blockade of this pathway has proven to be effective in solid tumors, as well as hematologic malignancies (8, 9).

CLL is well known to generate impaired immune responses in the host, with the malignant clone residing in well-vascularized tissues and circulating in peripheral blood but also in close proximity to effector cells that are capable, if activated appropriately, of carrying out a cytotoxic response. Defective T-cell responses have been observed in CLL patients, including imbalance of T-cell subsets, inability to form immune synapses between CLL B cells and T cells, increased expression of inhibitory receptors (e.g., PD-1, CD160, and LAG3), and loss in proliferation and cytotoxic capacity (2, 3, 7, 32, 33). These, in turn, result in ineffective antitumor responses. Thus, treatment protocols are based on agents with the ability to generate an immune response, e.g., anti-CD20, checkpoint inhibitors, or cellular therapies (34).

Currently, the most effective method of studying the tumor microenvironment is through the use of murine models (35). In this study, we used the Eμ-TCL1 adoptive transfer model to assess the relevance of SLAMF6 in the murine CLL microenvironment and the usefulness of targeting this receptor with a monoclonal antibody to SLAMF6 to improve its therapeutic action on leukemic cell expansion and enhance CD8⁺ T-cell functions. When Eμ-TCL1 leukemic cells were made to reside in a tumor microenvironment that lacks SLAMF6, i.e., the SLAMF6^−/− mouse, we observed an expansion of a PD-1⁺ subset of CD3⁺CD44⁺CD8⁺ T cells with reduced cytotoxic functions. This supports the concept that SLAMF6–SLAMF6 interactions, and possibly intracellular signaling initiated from this interaction, are important in the development of CD8⁺ T-cell functions.

This concept was further supported by the outcomes of our experiments with anti-SLAMF6, which in vivo reduced the leukemic burden in the Eμ-TCL1 adoptive transfer model. Administering anti-SLAMF6 reduced the number and proportion of PD1⁺CD3⁺CD44⁺CD8⁺ T cells concomitantly with an increase in cytotoxicity, as determined by expression of CD107a and granzyme B. Possible reasons for reduction in exhausted PD1⁺CD8⁺ T cells include induction of ADCC and/or downregulation of PD-1 from the cell surface after SLAMF6 ligation. SLAMF6 serves as a costimulatory receptor in T cells and recruits SAP to its cytoplasmic tail. We found no effect of SAP deficiency in this system when murine CLL cells were transferred into SAP^−/− mice. This suggests that our findings are independent of SAP, but whether there is a direct link between SLAMF6 and PD-1 signaling needs further investigation.

The change in the CD4⁺/CD8⁺ T-cell ratio appears to be associated with CLL progression (1, 5, 36, 37). This association is complemented by the findings that both CD4⁺ and CD8⁺ T cells acquire a PD-1⁺ phenotype that is associated with their inability to perform effector functions (5, 7, 36–38). Although the shift in the CD4⁺/CD8⁺ ratio appeared to be inverted in our system, anti-SLAMF6 was able to restore the phenotype and functions of exhausted CD8⁺ T cells, which would support the notion that exhausted T cells need to be reactivated within the suppressive tumor microenvironment for antitumor immunity. Thus, empowering CD8⁺ T-cell effector functions adds to the ability of anti-SLAMF6 to control disease by removing the CLL cells by ADCC and downregulating BCR signaling.

This dual activity of anti-SLAMF6 is in contrast with the mechanism by which anti-CD20 and ibrutinib control CLL in patients (34) Primarily, anti-SLAMF6 induces activation of ADCC mediated by engagement of antibody Fc portion with Fc receptors on macrophages and NK cells (26). Secondly, signaling of SLAMF6 in CLL B cells reduces proximal BCR signaling and survival in both murine and human models. Lastly, antibody binding to SLAMF6 on the surface of T cells causes reduction in the number of PD-1⁺CD3⁺CD44⁺CD8⁺ T cells and results in increased effector functions of remaining CD8⁺ T cells. This finding of the anti-SLAMF6 applies to other models as well. Previously, our lab demonstrated that injections of anti-SLAMF6 inhibit antibody responses in an immunization model (24). Inhibition of antibody production coincides with our finding that anti-SLAMF6 reduced BCR signaling, thus survival of B cells. We also found that i.p. injection of anti-SLAMF6 to subcutaneously growing B16 melanoma cells significantly reduced the tumor growth. Because melanoma cells do not express SLAMF6, this was through the reduction in the percentage of PD1⁺CD8⁺ T cells that infiltrated the tumor, suggesting activation of T-cell responses. This suggests that there is a selection by the antibody toward the PD-1⁺ subpopulation in activated CD8⁺ T cells, and this was irrespective of the disease model used.

To confirm the relevance of our findings in the mouse models, we used both CLL CD8⁺ T cells as well as generated exhausted CD8⁺ T cells in vitro and demonstrated that anti-human SLAMF6 increased degranulation of these exhausted cells. In these short cultures, it was not possible to dissect the effect of antibody on numbers of exhausted CD8⁺ T cells. It was previously shown that SLAMF6 partakes in restimulation-induced cell death (RICD) involving interactions through SAP and LCK (39). It is also possible to think that anti-SLAMF6 and anti-hSLAMF6 induced RICD in exhausted CD8⁺ T cells in our models.

Recently, Ayers and colleagues (40) identified IFNγ-related gene-expression profiles that could predict response to PD-1 checkpoint blockade in a variety of tumor types. Starting with 19 and validating in 62 melanoma patients, a “preliminary expanded immune” 28-gene set correlating with IFNγ signatures was identified. One of these genes was SLAMF6. IFNγ signaling is known to associate with a T cell–inflamed microenvironment that responds to anti–PD-1 therapy. This may be a platform to test combination of anti-SLAMF6 together with anti–PD-1 for a better T-cell response against various tumors.

We found that the leukemic burden and infiltration of Eμ-TCL1 cells in the PerC was lowered by anti-SLAMF6. Leukemic cells, as well normal B cells, show different characteristics when they are in the PerC or in the spleen, and one idea is that the PerC is a hypoxic niche that favors adhesion and growth of tumors (41–44). We previously demonstrated no ADCC effect of anti-SLAMF6 in PerC in a T cell–independent system, similar to that observed with anti-CD20 injections (26, 30). Although the ADCC function of an antibody is eliminated in this niche due to the microenvironment, signaling through SLAMF6 in T cells was able to reduce tumor infiltration. Anti-SLAMF6 was selectively targeting PD-1⁺CD3⁺CD44⁺CD8⁺ T cells in the spleen and blood and possibly induced their killing by ADCC. However, because the antibody was unable to induce ADCC in the PerC, it enriched for PD1⁺CD8⁺ T cells. This suggests that there is a correlation between PD-1 and SLAMF6 signaling, and understanding the mechanisms behind this may be relevant therapeutically. PD-1 contains a cytoplasmic tail with an ITSM that binds SHP1 and SHP2, similar to that in SLAMF6 (45–47). There may be a competition for binding of SHP1/2 that, in turn, dictates the responses from T cells. Understanding these niche-dependent changes to tumor killing is important, as there may be niches in the human body that are not accessible for certain type of treatments.

Peritoneal metastases of ovarian, pancreatic, gastric, and colon cancers are common, and in most cases, are of poor prognosis and survival rate if left untreated (48–51). Targeting and treatment of peritoneal tumors remain a challenge due to delivery, pharmacology, and efficacy (51). Regional chimeric antigen receptor (CAR) T-cell infusion for peritoneal carcinomatosis has shown to be effective, suggesting that the regional activation of T cells is important in clearing of tumors in the peritoneum (52, 53). It would be of interest to test in different models of peritoneal tumors whether targeting SLAMF6 could play a role in clearing these tumors.

Overall, besides the ability of anti-SLAMF6 to induce ADCC, improvement in effector CD8⁺ T-cell responses makes SLAMF6 an intriguing candidate for therapy, as it appears to work not only on tumors in secondary lymphoid organs and the peripheral blood but also in hypoxic niches such as the PerC or in solid tumors. We propose that anti-SLAMF6 alone or in combination should be further explored in B-cell leukemias and lymphomas in a clinical setting.

No potential conflicts of interest were disclosed.

Conception and design: B. Yigit, N. Wang, E. ten Hacken, A. Suarez-Fueyo, J.A. Burger, R.W. Herzog, P. Engel, C. Terhorst

Development of methodology: B. Yigit, N. Wang, A.Suarez-Fueyo, N. Chiorazzi, R.W. Herzog, P. Engel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Yigit, N. Wang, E. ten Hacken, A.K. Bhan, A. Suarez-Fueyo, E. Katsuyama, N. Chiorazzi, C. Terhorst

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Yigit, N. Wang, E. ten Hacken, A. Suarez-Fueyo, E. Katsuyama, G.C. Tsokos, C.J. Wu, J.A. Burger, C. Terhorst

Writing, review, and/or revision of the manuscript: B. Yigit, N. Wang, E. ten Hacken, A. Suarez-Fueyo, G.C. Tsokos, C.J. Wu, J.A. Burger, C. Terhorst

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Yigit, S.-S. Chen

Study supervision: B. Yigit, C. Terhorst

This work was supported by grants from the NIH PO1-AI065687 to C. Terhorst, N. Wang, and P. Engel. A. Suarez-Fueyo was supported by AI074549 (to G.C. Tsokos). E. ten Hacken is a Special Fellow of the Leukemia & Lymphoma Society.

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.