Enhanced SLAMF7 Homotypic Interactions by Elotuzumab Improves NK Cell Killing of Multiple Myeloma

Multiple myeloma (MM) is a bone marrow plasma cell neoplasm and second most common hematologic malignancy (1). Immunomodulatory drugs (thalidomide, lenalidomide, and pomalidomide), proteasome inhibitors (bortezomib, carfilzomib, and ixazomib), and autologous stem cell transplantation have significantly improved patient survival (2–8). However, MM remains largely incurable, as most patients relapse and become refractory to existing treatments (9).

Elotuzumab (Elo) is an IgG₁ monoclonal antibody targeting signaling lymphocytic activation molecule family member 7 (SLAMF7), which is expressed on tumor cells in 95% of MM cases (10–12). SLAMF7 is also found on immune cells, including natural killer (NK) cells, monocytes, and macrophages (13–16). Elo generates effective responses toward MM in preclinical studies in vitro and in vivo (10, 11, 17) and improves progression-free survival (PFS) of relapsed/refractory (RR)MM patients when administered as an immunotherapeutic in combination with lenalidomide/dexamethasone (17, 18). Elo plus pomalidomide/dexamethasone also significantly improves PFS compared with pomalidomide/dexamethasone alone (19). Antitumor effects result from several innate immune cell activation mechanisms: (i) NK cell–mediated antibody-dependent cellular cytotoxicity (ADCC) through FcγRIIIA (CD16), (ii) FcγR-dependent macrophage-mediated antibody-dependent cellular phagocytosis (ADCP), and (iii) CD16-independent costimulation of NK cells through direct interaction with SLAMF7 (10, 11, 14, 16, 20–22).

The efficacy of ADCC-inducing antibodies, such as rituximab, in hematologic malignancies is enhanced in patients homozygous for the high-affinity polymorphic variant of CD16 [valine at position 176 (or position 158 if leader sequence is subtracted)] compared with patients with one or two alleles encoding the low-affinity variant with phenylalanine (F) at position 176 (23–25). Accordingly, in a randomized phase II clinical trial of Elo plus bortezomib and dexamethasone, 176V/V homozygous patients have higher PFS compared with 176F/F patients (26).

Like most members of SLAM family receptors, SLAMF7 serves as a self-ligand (27), but it has unique costimulatory function in NK cells (28). SLAMF7 contains an intracellular immunoreceptor tyrosine-based switch motif (ITSM), which can recruit the cytosolic EAT-2 adaptor protein (29). NK cells express EAT-2, which mediates intracellular costimulatory signaling by SLAMF7, but plasma and MM cells do not express EAT-2 and thereby lack SLAMF7 signaling (29–31). Tyrosine phosphorylated EAT-2 recruits PLC-γ1, resulting in calcium mobilization, ERK activation, and enhanced functional responses by NK cells (29, 32). SLAMF7 can also physically interact with Mac-1 to trigger activation signaling in macrophages (13). Alternative mRNA splicing generates SLAMF7-long (L) and SLAMF7-short (S) isoforms (33). SLAMF7-S lacks the ITSM, interaction with EAT-2, and activation signaling.

Previous work showed that Elo promotes cytotoxicity by NK cells independent from ADCC (22) by causing CD16-independent costimulation of NK cells through SLAMF7 (16). Here, we demonstrated that CD16-independent enhancement of cytotoxicity by Elo required SLAMF7 expression on both NK and target cells and required expression of SLAMF7-L in the NK cells. Elo had unique capacity among several SLAMF7 antibodies to enhance cytotoxicity by promoting SLAMF7–SLAMF7 interactions between NK and MM cells. In addition, an Fc mutant form of Elo lacking CD16-binding properties promoted cytotoxicity of MM target cells by primary NK cells from most healthy donors, especially when the NK cells were cultured with IL2.

NK-92 cells were obtained from ATCC in 2005 and cultured in complete α-MEM medium as described (34), supplemented with 100 U/mL of human recombinant IL2 (teceleukin, Hoffman-La Roche Inc.). Cells were passed with fresh medium and IL2 every 3 to 4 days. The cDNAs encoding either high-affinity (176V; GenBank: BC017865.1) or low-affinity (176F; GenBank: NM_000569.6) variants of FcγRIIIA and human SLAMF7-L (GenBank: NM_021181.4) or SLAMF7–S (GenBank: NM_001282592.1) were subcloned into pBMN-NoGFP retroviral vector (35, 36). NK-92 parental cells retrovirally transduced to express either CD16 variant were previously described (34, 37) and always obtained from master stocks. NK-92 SLAMF7 knockout cells were generated using a doxycycline-inducible CRISPR/Cas9 system as previously described (38). Specific guide RNA sequences [sg1 5′-AAAGAGCTGGTCGGTTCCGT-3′, sg2 5′-GAGACACAGGAGGACCATGG-3′ (ref. 39; Integrated DNA Technologies)] were cloned into pLRG to generate lentivirus in combination with pCMV-dR8.74PAX2 and pMD2.G [from Dr. Neil Johnson, Fox Chase Cancer Center (FCCC)] in 293T cells. NK-92 cells were infected twice with doxycycline-inducible blasticidin-resistant Cas9 lentivirus, viable blasticidin-resistant clones were single-cell sorted, and clones with the highest induced Cas9 expression were selected. These were infected with lentivirus containing sgRNA and GFP, GFP⁺ cells were sorted and cultured with doxycycline for 48 hours, and SLAMF7-deficient cells were sorted. NK-92 SLAMF7 KO cells were then transduced with retrovirus encoding CD16 176V, SLAMF7-L, or SLAMF7-S and sorted for SLAMF7 expression (34).

MM cell lines were cultured in complete RPMI-1640 medium (supplemented with 10% FBS, penicillin/streptomycin, L-glutamine, sodium pyruvate, 10 mmol/L HEPES, and 50 μmol/L 2-mercaptoethanol). MM.1R and RPMI8226 cells were obtained from ATCC at the beginning of the project and reauthenticated in 2018. RPMI8226 were retrovirally transduced to express higher levels of SLAMF7-L (RPMI8226+SLAMF7), and zinc finger nucleases (product number CKOZFND19392-1KT; Sigma-Aldrich) were introduced by nucleofection to disrupt SLAMF7 expression in MM.1R cells, followed by sorting for SLAMF7 expression, as previously described (16). SKOV3 cells were obtained from the FCCC Cell Culture Facility, authenticated in 2015, cultured in complete RPMI-1640 medium, and retrovirally transduced with SLAMF7-L cDNA in pBMN-IRES-NoGFP vector to obtain SLAMF7⁺ cells after sorting.

An hSLAMF7-TCR-ζ reporter cell line was engineered by transfecting 3A9 mouse T cells with a construct encoding fusion of human (h)SLAMF7 extracellular domain and mouse CD247 (TCR-ζ) transmembrane and intracellular domains in UCOE vector (detailed in Supplementary Fig. S1). Cells were cultured in complete RPMI-1640 supplemented with 10% FBS and 800 μg/mL of hygromycin B. SLAMF7 engagement was assayed by measuring mouse IL2 secretion with ELISA according to the manufacturer (BD OptEIA, #55148).

Cell lines were authenticated using genetic short tandem repeat profiling analysis by IDEXX Bioresearch. Cell lines were thawed fresh every 2 months and found to be free of Mycoplasma at least yearly in assays by the FCCC Cell Culture Facility. Primary NK cells were enriched from healthy donor PBMCs using the EasySep NK cell–negative selection kit (STEMCELL Technologies) and confirmed to be >95% purity (CD45⁺CD56⁺CD3⁻; see Supplementary Fig. S2). Healthy donor volunteers provided written informed consent using HIPAA-compliant procedures approved by the Institutional Review Board of FCCC and experiments were performed in accordance with the Declaration of Helsinki principles.

SLAMF7 antibodies: Elo, Elo with Fc domain mutations that prevent recognition by CD16 (Elo Fc mut; ref. 16), Elo Fab fragments (produced as recombinant proteins), Elo F(ab')₂ fragments (produced by pepsin cleavage), and PDL241 (40) were provided by Bristol-Myers Squibb (BMS). The SLAMF7 antibody ChLuc90, which consists of chimerized human IgG₁ Fc fused to mouse CDR variable regions, was provided by AbbVie Inc. (11). Elo Fc mut antibody (provided by BMS) was expressed by the CHO-S cell line cotransfected with vectors pICOFSCneoK (containing a cDNA encoding Elo variable regions) and pODpurIgG1.1f (containing a cDNA encoding IgG1 heavy-chain constant region with L234A-L235E-G237A-A330S-P331S mutations, which abolish interaction with CD16, as previously described, ref. 16). Comparative binding of Elo and Elo Fc mut to myeloma cells is shown in Supplementary Fig. S3. The m24 antibody targeting activated LFA-1 and NKG2D blocking antibody, 1D11, were from BioLegend. The DNAM-1 blocking antibody was from R&D Systems (AF666; polyclonal goat IgG).

SLAMF7 expression was assayed by staining with biotin conjugated Elo antibody (EZ-Link NHS-Biotin, Thermo Fisher #20217) and APC–streptavidin (BioLegend, #405207), SLAMF7 antibody conjugated with APC or DL650 (BioLegend, clone 162.1; ref. 41), or isotype control (BioLegend, #400330). Surface expression of 176V and 176F CD16 was assessed with GRM1 antibody (Southern Biotech, #9570-31), although CD16 176V was sometimes assayed with 3G8-APC (BioLegend, #302012). SLAMF7 antibody competition experiments were performed using the CHO cell line transfected to express high levels of SLAMF7-L. Cells (100,000 cells/condition) were incubated with various concentrations of SLAMF7 antibodies (Elo, PDL241, and 162.1) on ice, washed 2× with staining buffer (0.1% Na azide and 1% FBS in HBSS) and incubated with conjugated SLAMF7 antibodies (162.1-PE, BioLegend; Elo-AF647, PDL241-AF647) on ice, washed ×2 with staining buffer and 100 ng/mL of propidium iodide (Invitrogen) in the final wash, and then subjected to analysis on a BD LSR II or Aria II flow cytometer. Data were analyzed using FlowJo software (BD).

NK cell cytotoxicity was assessed using the CytoTox 96 nonradioactive cytotoxicity assay (Promega, #G1780), which is a colorimetric assay that measures lactate dehydrogenase released when target cells are lysed, according to the manufacturer's protocol.

The xCELLigence Real-Time Cell Analyzer (RTCA; ACEA Biosciences) detects electrical impedance of adherent cells over time and can be used as a sensitive measure of kinetic cytotoxicity responses by NK cells. The upward trajectory of plots over time is characteristic of growing adherent cells, and downward plots upon addition of NK cells are characteristic of adherent target cell death (Supplementary Fig. S4). SKOV3 cells (5,000/100 μL of complete RPMI-1640) were added into each well of E16 xCELLigence plates (ACEA Biosciences) and cultured overnight at 37°C with 5% CO₂. The next day, complete medium was added to control wells, plates were inserted into an xCELLigence RTCA DP unit (ACEA Biosciences) at 37°C with 5% CO₂ and baseline impedance was measured. NK-92 cells or purified donor NK cells (100 μL of 0.5 × 10⁶/mL) and SLAMF7 antibodies (Elo, Elo Fc mut, Elo F(ab')₂, Elo Fab, PDL241, ChLuc90, 162.1; 10 μg/mL) were added to desired wells. Data were recorded at 10-minute intervals for up to 24 hours. Every condition containing NK cells was performed in duplicate wells within each experiment and a total of 3 to 6 experiments were performed for each condition. Total lysis (100%) was determined by adding lysis buffer (20 mmol/L Tris HCl pH = 8, 2 mmol/L EDTA pH = 8, 150 mmol/L NaCl, 0.2% SDS, 10% Triton X-100) to a well of target cells alone and the percent cytotoxicity for each experimental condition was normalized to this 100% value at the 10-hour time point (Supplementary Fig. S4). All kinetic plots represent the lines generated from means of all values from every experiment at individual time points. Statistical analysis was performed using Student t tests in GraphPad Prism on mean values from the duplicate wells for the percent cytotoxicity at 10 hours within each experiment performed for each condition, as detailed in figure legends, with significance defined as P < 0.05.

We tested the effect of Elo on the cytotoxicity of MM cell lines with various SLAMF7 expression levels by NK-92 cell lines expressing high (CD16-176V) or low (CD16-176F) affinity FcγRIIIA (Fig. 1A). Two matching pairs of MM target cell lines were genetically modified as previously described (16): RPMI8226 cells expressing low endogenous levels of SLAMF7 were retrovirally transduced to express high SLAMF7 levels (RPMI8226 + SLAMF7), and MM.1R cells, which constitutively express high levels of SLAMF7, were treated with zinc finger nucleases to knock out SLAMF7 expression (MM.1R SLAMF7 KO cells; Fig. 1B). Adding low concentrations of Elo (0.001–0.1μg/mL) to CD16-bearing NK-92 cell lines in the presence of MM.1R cells promoted higher cytotoxicity by CD16-176V NK-92 cells, as compared with CD16-176F NK-92 cells. Using higher concentrations of Elo (1–300 μg/mL) stimulated similar maximal response by NK cells expressing the high- or low-affinity polymorphism (Fig. 1C). A similar trend was observed when NK cells were cocultured with RPMI8226 + SLAMF7 cells, although the difference was not statistically significant (Fig. 1D). Elo did not appreciably stimulate cytotoxicity toward MM cells lacking (MM.1R KO) or expressing low levels of SLAMF7 (parental RPMI8226). These results demonstrate that low concentrations of Elo promote more potent NK cell–mediated ADCC toward SLAMF7⁺ MM cells if the NK cells express high-affinity (176V) CD16, although the response plateaus to a level of cytotoxicity equivalent for CD16-176F at concentrations above 1 μg/mL of Elo.

To follow-up studies showing CD16-independent enhancement of NK cell–mediated natural cytotoxicity by Elo (10, 11, 14, 16, 20–22), we utilized NK-92 cells lacking endogenous CD16 but expressing SLAMF7. In the presence or absence of Elo, knocking out SLAMF7 on MM.1R decreased cytotoxicity by NK-92 cells and overexpressing SLAMF7 on RPMI8226 boosted natural cytotoxicity by NK-92 cells (Fig. 2A), suggesting that SLAMF7–SLAMF7 interactions potentiate cytotoxicity. Interestingly, adding Elo to NK-92 further enhanced cytotoxicity of RPMI8226 + SLAMF7 in a dose-responsive manner (Fig. 2A, bottom). Although it is unclear why this effect was not observed for MM.1R targets, the enhancement toward RPMI8226 + SLAMF7 targets by Elo suggests that this effect may occur toward only a subset of SLAMF7⁺ MM target cells and further indicates that Elo promotes SLAMF7–SLAMF7 interactions between NK and myeloma cells.

Sensitive kinetic measurement of adherent target cell cytotoxicity can be performed using xCELLigence RTCA assays (see Materials and Methods). Unfortunately, available MM cell lines are minimally adherent and could not provide a sensitive enough signal in the xCELLigence system, even when wells were coated with adhesion ligands, such as fibronectin. Therefore, we utilized the adherent human ovarian cancer cell line SKOV3, which undergoes minimal natural cytotoxicity by NK-92. We further generated SKOV3 cells that stably expressed SLAMF7 [SKOV3(+SLAMF7); Fig. 2B]. Adding Elo at 0.1 to 1 μg/mL with CD16-expressing NK-92 cells significantly enhanced cytotoxicity of SKOV3(+SLAMF7) target cells to a similar level of cytotoxicity upon addition of trastuzumab, an ADCC-inducing antibody that binds the SKOV3 surface marker, Her2 (Supplementary Fig. S4A and S4B). Adding Elo to a coculture of parental NK-92 cells lacking CD16 and parental SKOV3 cells lacking SLAMF7 expression did not affect target cell death (Fig. 2C, top; 2D, left). Similarly, trastuzumab did not promote cytotoxicity by parental NK-92 cells, due to their lack of CD16 (Fig. 2C, top). In contrast, Elo significantly enhanced killing of SKOV3(+SLAMF7) cells by parental NK-92 cells, but trastuzumab had no effect (Fig. 2C, bottom; 2D, right). To confirm the Fc-independent enhancement of cytotoxicity, the addition of a mutant form of Elo (Elo Fc mut), which has five amino acids altered in the Fc region to abrogate binding to CD16 (see ref. 16), also increased cytotoxicity of SKOV3(+SLAMF7) cells by CD16-deficient NK-92 cells (Supplementary Fig. S5, bottom), and this effect required SLAMF7 expression on the SKOV3 cells (Supplementary Fig. S5, top). Therefore, in the absence of CD16 interactions, Elo still enhances NK cell cytotoxicity apparently by promoting SLAMF7–SLAMF7 interactions between NK and target cells (Fig. 2E), analogous to our results with RPMI8226 + SLAMF7 target cells (Fig. 2A, bottom).

To further investigate the observed effect of Elo to promote NK cell cytotoxicity toward SKOV3(+SLAMF7) cells, we generated SLAMF7-knockout NK-92 cells (SLAMF7 KO, CD16-) using a doxycycline-inducible CRISPR/Cas9 system (see Materials and Methods). The loss of SLAMF7 expression was confirmed by flow cytometry (Fig. 3A). Elo had no effect on the cytotoxicity of SLAMF7 KO NK-92 cells toward SKOV3(+SLAMF7) target cells (Fig. 3B and C, top). In contrast, adding Elo to NK cells that expressed SLAMF7 [NK-92(SLAMF7⁺,CD16⁻)] significantly enhanced the killing of SKOV3(+SLAMF7) cells (Fig. 3B and C, top), consistent with our previous results toward RPMI8226 + SLAMF7 cells (Fig. 2A, bottom). Elo Fc mut had a similar effect on enhancing NK cell cytotoxicity (Fig. 3B and C, bottom). To further rule out a role for the Fc domain of Elo in the response, we generated bivalent F(ab')₂ fragments, which had similar capacity to stimulate cytotoxicity as intact Elo (Fig. 3D). Alternatively, monovalent Fab fragments of Elo were incapable of stimulation, demonstrating a requirement for bivalent association with SLAMF7 to stimulate cytotoxicity (Fig. 3D). These results showed that enhanced NK-mediated killing of SLAMF7⁺ tumor cells by Elo required SLAMF7 expression on NK cells and the bivalent structure of Elo, but was independent of the Fc domain (summarized in Fig. 3E).

To further study whether Elo promotes SLAMF7–SLAMF7 interactions, a mouse reporter T-cell line expressing a chimeric receptor consisting of human SLAMF7 extracellular domain and mouse CD247 (TCR-ζ) intracellular domain was generated. The hSLAMF7-TCR-ζ reporter cells produced robust levels of mouse IL2 (mIL2) when exposed to plate-bound Elo or three other plate-bound SLAMF7 antibodies (ChLuc90, 162.1, and PDL241; Fig. 4A). We found that Elo, PDL241, and 162.1 bind to distinct epitopes on SLAMF7, because they do not compete (Supplementary Fig. S6). To test for the capacity of these antibodies to mediate SLAMF7–SLAMF7 interactions, the reporter cell line was added to culture wells containing plate-bound recombinant human SLAMF7 in the presence of varying concentrations of each antibody. Elo concentrations of 0.1 to 3 μg/mL were highly effective in inducing mIL2 secretion by the hSLAMF7-TCR-ζ reporter cells, whereas the three other SLAMF7 antibodies (ChLuc90, 162.1, and PDL241) induced very low levels of mIL2 at concentrations of 1 to 3 μg/mL (Fig. 4B). In contrast, incubating hSLAMF7-TCR-ζ reporter cells with Elo did not affect mIL2 secretion in plates coated with a negative control recombinant protein (mesothelin) or if soluble recombinant hSLAMF7 was added (Fig. 4C). These data demonstrate that Elo has unique capacity among these antibodies to orient SLAMF7-TCR-ζ in an appropriate oligomerized form to induce cell signaling through the chimeric reporter construct, apparently by promoting SLAMF7–SLAMF7 interactions with the plate-bound SLAMF7 molecules.

The four SLAMF7 antibodies (Elo, ChLuc90, 162.1, and PDL241) were also compared in xCELLigence assays for their capacity to costimulate NK-92 cells through interaction with SLAMF7 and enhance cytotoxicity of SLAMF7⁺ SKOV3 target cells. Again, only Elo promoted CD16-independent NK cell–mediated cytotoxicity (Fig. 5A; Supplementary Fig. S7A). In fact, the other SLAMF7 antibodies slightly decreased cell death of SKOV3(+SLAMF7) by NK-92 cells (although not to a statistically significant degree). In contrast, the Elo-induced enhancement of cytotoxicity was not evident with SLAMF7 KO NK-92 cells (Fig. 5B; Supplementary Fig. S7B). Taken together, our results indicated that Elo has unique capacity among SLAMF7 antibodies to promote SLAMF7–SLAMF7 interactions between NK cells and target cells, which resulted in enhanced cytotoxicity responses.

To better understand the mechanism by which Elo enhances cytotoxicity in a CD16-independent manner, we tested for upregulation of a variety of activation markers and receptors on NK-92 cells when cultured for 24 hours with SLAMF7⁺ SKOV3 cells ± Elo. The expression of CD69, NKG2D, CD25, ICAM-1, and the activated form of LFA-1 were significantly increased under these conditions (Supplementary Fig. S8A), demonstrating NK cell activation. Most of these increases required or were more pronounced in the presence of SLAMF7⁺ SKOV3 target cells. The increased expression of ICAM-1 and activated LFA-1 suggests improved target cell adhesion upon Elo treatment, although we could not detect enhanced conjugation by Elo of NK-92 cells cultured with either MM1.R or RPMI8226 + SLAMF7 target cells (Supplementary Fig. S8B). Nonetheless, upregulation of adhesion molecules by Elo may be playing a modest role in the effect. The increase in NKG2D levels by Elo was also intriguing, particularly when we found that SKOV3 target cells expressed ligands for both NKG2D (ULBPs) and DNAM-1 (Nectin 2 and PVR; Supplementary Fig. S8C). Therefore, we tested whether blocking either NKG2D or DNAM-1 with antibodies could disrupt Elo-induced cytotoxicity. Whereas NKG2D blockade partially reduced the cytotoxicity, DNAM-1 blocking antibodies were ineffective (Fig. 5C and D; Supplementary Fig. S7C and S7D). Taken together, the CD16-independent enhancement of cytotoxicity by Elo appears to be mediated through multiple mechanisms, including the activation of LFA-1, upregulation of ICAM-1 and NKG2D, and potentially upregulation of CD69 and CD137, which can act as activating receptors on NK cells (42, 43).

At least two differentially spliced isoforms of SLAMF7 are found in humans, which differ based on long (-L) or short (-S) cytoplasmic domains that contain or lack the ITSM, respectively. SLAMF7-S does not transduce activation signals (33, 44). We reconstituted NK-92 (SLAMF7KO, CD16⁻) cells with these SLAMF7-L or SLAMF7-S isoforms and tested the ability of Elo to stimulate cytotoxicity toward SKOV3 + SLAMF7 cells using the xCELLigence assay. Although SLAMF7-S was expressed at slightly lower levels than SLAMF7-L (Fig. 6A), Elo significantly increased cytotoxicity by NK-92(+SLAMF7-L, CD16⁻) cells, whereas Elo was entirely incapable of potentiating cytotoxicity by NK-92(+SLAMF7-S, CD16⁻) or NK-92(SLAMF7KO,CD16⁻) cells (Fig. 6B and C). The results suggest that the potentiated cytotoxicity required costimulatory signaling within the NK cells, because the cytoplasmic ITSM signaling motif that recruits EAT-2 was only present in the SLAMF7-L isoform.

To validate our results from NK-92 cells, we used purified healthy donor NK cells to test the CD16-independent effect of Elo Fc mut on promoting natural killing of parental SKOV3 and SKOV3(+SLAMF7) target cells in xCELLigence assays. Neither Elo nor Elo Fc mut was able to promote healthy donor NK cell–mediated killing of parental SLAMF7-deficient SKOV3 cells (Supplementary Fig. S9). In contrast, Elo promoted complete lysis of SKOV3(+SLAMF7) cells within 8 to 10 hours by purified donor NK cells from healthy donor #1 (HD#1; Fig. 7A). On the other hand, Elo Fc mut had no effect on cytotoxicity of SKOV3(+SLAMF7) cells by NK cells from HD#1 (Fig. 7A). After screening seven healthy donors, we observed that Elo Fc mut promoted an increase in cytotoxicity by NK cells from only three of the donors, assessed as percent cytotoxicity at 10 hours in xCELLigence assays (Fig. 7B, left), with HD#8 being a representative example (Fig. 7C, left). Therefore, Elo Fc mut was able to potentiate cytotoxicity of fresh primary human NK cells in a subset of donors (3 of 7).

The finding that Elo is therapeutically ineffective in treating RRMM as a single agent (18), yet shows significant efficacy in combination with lenalidomide and dexamethasone (17) suggests that coordinated immune stimulation is important to achieve optimal antitumor effect by Elo. Lenalidomide is an immunomodulatory drug that stimulates NK and T-cell activation, in part by increasing IL2 production by T cells (21, 45). Furthermore, our NK-92 cells are routinely cultured in IL2. Therefore, we hypothesized that culturing NK cells in the presence of IL2 might enhance the capacity of Elo Fc mut to stimulate cytotoxicity. When purified NK cells from HD#3 and HD#8 were assayed after culture for 48 hours in the presence of 100 U/mL recombinant human IL2, Elo Fc mut more potently increased cytotoxicity toward SKOV3(+SLAMF7) cells, as compared with freshly isolated NK cells from these donors (Fig. 7C and D, left vs. right). As summarized in Fig. 7B, whereas Elo Fc mut potentiated cytotoxicity by freshly prepared primary NK cells in only 3 of 7 healthy donors, cytotoxicity was boosted in 5 of 7 healthy donors after culture of the NK cells in IL2. Hence, the CD16-independent costimulation of cytotoxicity by Elo engagement with SLAMF7 can be demonstrated in primary human NK cells, most prominently when NK cells are activated by IL2.

We identified an additional mechanism by which Elo promotes NK cell cytotoxicity toward MM cells independent of CD16-mediated ADCC. SLAMF7 functions as a costimulatory receptor on NK cells that mediates intracellular signaling via recruitment of cytosolic EAT-2 adaptor protein (12, 29). Our previous work demonstrates that Elo can costimulate calcium mobilization responses triggered by the aggregation of NKp46 ± NKG2D activating receptors (16). Here, we demonstrated that Elo can also promote SLAMF7–SLAMF7 interactions between NK and MM cells to enhance NK cell–mediated cytotoxicity through costimulation in addition to facilitating ADCC.

At lower concentrations, Elo (1 ng/mL–1 μg/mL) promoted higher cytotoxicity by NK cells expressing the high-affinity CD16 176V polymorphism versus NK cells expressing the low-affinity 176F polymorphism. In contrast, at higher concentrations of Elo (up to 300 μg/mL; similar to the observed plasma concentrations in Elo-treated patients, which are routinely above 70 μg/mL; ref. 46), no effect of CD16 polymorphism on NK cell ADCC was observed. Despite the lack of differential response at high Elo concentrations, CD16 176V/V homozygous patients respond better to Elo plus bortezomide/dexamethasone compared with 176V/F or 176 F/F patients (26). These results are consistent with better ADCC by NK cells and improved clinical efficacy in response to rituximab in follicular lymphoma patients possessing the CD16 176V/V genotype (23, 47). It is currently unclear if the high-affinity CD16 polymorphism is also beneficial in patients treated with Elo in combination with lenalidomide/dexamethasone.

We further demonstrated that Elo can significantly enhance the natural cytotoxicity of SLAMF7-expressing target cells by CD16-deficient NK-92 cells, and this effect required SLAMF7 expression on both the NK and target cells. This result is consistent with the work of Collins and colleagues, showing that Elo induces CD16-independent target cell killing by NK-92 that requires SLAMF7 expression on the target cells (22). Our work extended their results by showing a requirement for expression of signaling-competent SLAMF7 in the NK cells, indicating a costimulatory signaling mechanism. The requirement for SLAMF7 expression on both effector and target cells suggests that Elo is capable of either (i) bridging SLAMF7 between the two cells or (ii) enhancing the SLAMF7–SLAMF7 interactions between the two cell types. This enhancement only occurred toward SLAMF7⁺ RPMI8226 and SLAMF7⁺ SKOV3, but not SLAMF7⁺ MM.1R myeloma target cells. We hypothesize that Elo engagement with SLAMF7 effectively costimulates additional NK cell–activating receptors engaging with ligands expressed on RPMI8226, but absent on MM.1R. This hypothesis is supported by our previous work showing that Elo engagement with SLAMF7 cannot induce calcium signaling in NK cells on its own, but Elo can potentiate calcium signaling triggered by engagement of the ITAM-linked NK cell receptor, NKp46, either with or without coengagement of NKG2D (16). Consistent with our previous work, we showed here that NKG2D plays at least a partial role in the enhanced cytotoxicity induced by Elo. Of note, the CD16-independent effect occurs at Elo concentrations in the range of 0.1 to 300 μg/mL, whereas ADCC responses are observed in the range of 1 ng/mL to 1 μg/mL. These concentrations are both well below the high serum concentrations achieved for Elo at therapeutically effective dosing regimens (46).

Surprisingly, of four SLAMF7 antibodies tested, only Elo enhanced SLAMF7–SLAMF7-dependent activation responses. This indicates that Elo binds to a unique epitope on SLAMF7, whereby it is either better oriented to effectively bridge the receptor between effector and target cell or the binding induces a conformational change that enhances homotypic interactions between SLAMF7 molecules on opposing cells. Furthermore, we showed that the costimulatory impact of Elo requires bivalent binding to SLAMF7, because F(ab')₂ is effective, but Fab fragments are not. All four tested antibodies were found to bind distinct epitopes on SLAMF7. Elo reportedly binds to the membrane proximal C2 domain of SLAMF7, within an epitope encompassing amino acids 170-227 (U.S. patent 7842293B2). In contrast, PDL241 and ChLuc90 are reported to interact with the V domain (U.S. patent 7842293B2 and ref. 40), which mediates homotypic interactions between SLAM family members (48). These two antibodies may thereby block SLAMF7–SLAMF7 interactions, consistent with their capacities to slightly reduce cytotoxicity in our experiments. We could not detect enhanced conjugation between NK-92 cells and MM target cells by Elo or Elo Fc mut using standard flow cytometry–based assays (49). This result is consistent with a previous report that SLAMF7 signaling in YT-S NK cell line could potentiate killing of SLAMF7⁺ K562 target cells, but not enhance conjugate formation (50). Although the exact mechanism by which Elo promotes SLAMF7 interactions will require further experimentation, Elo has unique capacity to mediate this biological effect. It was shown that Elo can also facilitate macrophages to mediate ADCP toward MM cells in an Fc-dependent manner (14). As SLAMF7 is also highly expressed on subsets of macrophages (notably M1 macrophages; ref. 15), it is possible that Elo could also contribute to macrophage-mediated killing of SLAMF7⁺ tumors through a CD16-independent mechanism by promoting SLAMF7–SLAMF7 interactions.

We further demonstrated that CD16-independent costimulation by Elo requires the ITSM-containing cytoplasmic domain of SLAMF7 in NK cells, because Elo enhanced cytotoxicity by NK-92 cells reconstituted with SLAMF7-L, but not with the truncated receptor lacking ITSM (SLAMF7-S). Although we cannot rule out that another motif in the cytoplasmic domain is mediating the costimulation, our data suggest that the ITSM is responsible. Previous work showed that NK cell–mediated natural cytotoxicity toward SLAMF7⁺ target cells is potentiated via tyrosine phosphorylation of the ITSM motif on SLAMF7, which recruits EAT-2 (50) to recruit phospholipase Cγ-1, resulting in intracellular calcium signaling (50). This is consistent with the calcium signaling costimulation by Elo that we previously observed (16). Our results also suggest that extensive SLAMF7-S isoform expression on NK cells from MM patients may affect their clinical outcomes.

We used an Fc mutant form of Elo (Elo Fc mut) to demonstrate CD16-independent enhancement of natural cytotoxicity by NK cells from healthy donors toward SLAMF7⁺ target cells. Elo Fc mut enhanced cytotoxicity of fresh NK cells from three out of seven healthy donors. Interestingly, Elo Fc mut boosted NK cell cytotoxicity of five of these healthy donors if the cells were cultured for 48 hours in IL2, demonstrating that the CD16-independent effect of Elo on natural cytotoxicity is dependent upon the state of NK cell activation. This finding is of significant interest, because Elo is ineffective as a single agent in MM patients, but highly effective when coadministered with lenalidomide and dexamethasone (17). The immunomodulatory agent lenalidomide binds the E3 ubiquitin ligase cereblon and targets the degradation of the IKZF1 and IKZF3 transcription factors, resulting in increased production of IL2 by T cells (51). Therefore, it is interesting to speculate that lenalidomide coadministration with Elo may, at least in part, improve Elo activity toward MM cells by elevating IL2 levels to activate NK cells and thereby increase their susceptibility to Elo-mediated promotion of SLAMF7–SLAMF7 interactions. This is consistent with previous reports demonstrating lenalidomide can enhance Elo-mediated killing of MM cell lines in coculture with PBMCs and this activity can be partially blocked by antibodies neutralizing the effects of IL2 (anti-CD25/IL2R) on NK cells (21). Consistent with this, we also showed that Elo can induce IL2Rα (CD25) on NK cells, which may also contribute to enhanced function by lenalidomide in patients.

Our data further elucidate the mechanism of action of Elo beyond the previously established mechanism of ADCC. We showed that Elo coactivates NK cells and enhances cytotoxicity toward SLAMF7⁺ tumor cells by promoting SLAMF7–SLAMF7 interactions. Furthermore, these findings are specific to Elo compared with other SLAMF7 antibodies tested demonstrating the unique mechanistic attributes of NK coactivation and ADCC dually mediated by Elo.

T. Pazina reports receiving a commercial research grant from Bristol-Myers Squibb. A.Y. Kearney is Scientist II at Bristol-Myers Squibb. N.A. Bezman has ownership interest (including patents) in Bristol-Myers Squibb. M.D. Robbins has ownership interest (including patents) in Bristol-Myers Squibb. A.D. Cohen reports receiving a commercial research grant from Bristol-Myers Squibb, other commercial research support from Novartis, and is a consultant/advisory board member for Bristol-Myers Squibb, Celgene, Janssen, Takeda, Kite Pharma, Oncopeptides, GlaxoSmithKline, Seattle Genetics, and Array Biopharma. K.S. Campbell reports receiving a commercial research grant from Bristol-Myers Squibb and has ownership interest (including patents) in NantKwest. No potential conflicts of interest were disclosed by the other authors.

Conception and design: T. Pazina, A.M. James, A. Jhatakia, N.A. Bezman, M.D. Robbins, A.D. Cohen, K.S. Campbell

Development of methodology: T. Pazina, A.M. James, A. Gale, A. Jhatakia, N.A. Bezman, K.S. Campbell

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Pazina, A.M. James, K.B. Colby, A.Y. Kearney, K.S. Campbell

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Pazina, A.M. James, K.B. Colby, A. Jhatakia, A.Y. Kearney, R.F. Graziano, M.D. Robbins, A.D. Cohen, K.S. Campbell

Writing, review, and/or revision of the manuscript: T. Pazina, Y. Yang, A. Gale, R.F. Graziano, N.A. Bezman, M.D. Robbins, A.D. Cohen, K.S. Campbell

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.M. James, Y. Yang, A.Y. Kearney, K.S. Campbell

Study supervision: A.D. Cohen, K.S. Campbell

The authors thank Dr. Audie Rice for fostering initiation of this project, Drs. John Krais and Neil Johnson for help generating the CRISPR/Cas9 SLAMF7 KO NK-92 cell line, Lena Shally and Sanjna Shelukar for technical support, John Dowling for developing cytotoxicity assays using xCELLigence, and the following FCCC core facilities: Cell Culture, Flow Cytometry, Cell Sorting, and High-Throughput Screening. The authors also thank Bristol-Myers Squibb and AbbVie for valuable reagents, Bristol-Myers Squibb for financial support, and the Biological Resources Branch of NCI-Frederick for recombinant human IL2. This study was supported by Collaborative Science Center of Excellence grants CA204-013, CA-204-156, and CA204-214 to A.D. Cohen and K.S. Campbell from Bristol-Myers Squibb, FCCC Bucks County and Main Line Boards of Associates, NIH Training Grant T32 CA9035 (A.M. James), and NCI Comprehensive Cancer Center Support Grant CA06927 (FCCC).

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.