CS1, a Potential New Therapeutic Antibody Target for the Treatment of Multiple Myeloma

Multiple myeloma is a malignant disease of plasma cells, occurring in adults with an incidence of ∼14,000 new cases per year in the United States. The median survival from diagnosis is ∼3 years (1, 2). Despite advances in therapy including stem cell transplantation and new biological agents, such as bortezomib and immunomodulatory agents, multiple myeloma is still considered an incurable disease (2–6). Thus, new therapies are needed.

Monoclonal antibody (mAb) therapy has made a major effect in the area of B-cell non–Hodgkin lymphoma therapy. In particular, anti-CD20 (rituximab) has become a standard therapeutic agent in B-cell lymphomas and has improved outcome in these diseases (7–17). However, no effective immunotherapeutic option exists yet in multiple myeloma. In an effort to develop new immunotherapeutics for the treatment of multiple myeloma, we identified CS1 (CD2 subset-1, CRACC, SLAMF7, and CD319), a member of the signaling lymphocyte activating-molecule–related receptor family (18), as a cell surface antibody target selectively expressed in plasma cells. Other members of the signaling lymphocyte activating-molecule–related receptor family include signaling lymphocyte activating–molecule (CD150), 2B4 (CD244), CD84, NTB-A (Ly-108), and Ly-9 (CD229; ref. 19). These molecules are characterized by two or four extracellular immunoglobulin (Ig)-like domains and an intracellular signaling domain with immune receptor tyrosine-based switch motifs with the consensus amino acid sequence TxYxxV/I (20, 21).

In this study, we show that normal plasma cells and multiple myeloma cells express high levels of CS1 mRNA and protein. Other normal lymphocytes subsets [natural killer (NK), NK-like T cells, CD8+ T cells, activated monocytes, and dendritic cells] also express CS1, albeit at generally lower levels than plasma cells. We generated a panel of murine and humanized mAbs to human CS1 to validate this protein as a potential target for the treatment of multiple myeloma. We show that the humanized anti-CS1 mAb HuLuc63 has potential as an immunotherapeutic by binding to multiple myeloma cells and mediating antibody-dependent cellular cytotoxicity (ADCC) with effector cells isolated from multiple myeloma patients. In addition, HuLuc63 significantly decreased established tumors in vivo in a myeloma xenograft model, an activity that seemed to be dependent on the presence of functional NK cells.

Identification of CS1 and gene expression profiling. CS1 was identified using representational difference analysis done by subtracting naïve B-cell (CD19⁺IgD⁺CD38^int/− CD27⁻) cDNA from a memory B cell and plasma cell (CD19^+/lo IgD⁻CD38 ^int/−CD27⁺) cDNA library (22, 23). Using standard molecular biology techniques, the cDNA subtraction library was ligated into a standard plasmid vector and transformed into electrocompetent Escherichia coli (DH-10B) cells. Single bacterial colonies, each representing one specific insert, were amplified using standard colony PCR. The cDNA for CS1 was identified to be preferentially expressed in the memory B cell and plasma cell cDNA library. To confirm expression of CS1 in normal plasma cells and examine expression in diseased plasma cells from patients with monoclonal gammopathies of undetermined significance and multiple myeloma, gene expression profiling was done on CD138-purified plasma cells as described by Zhan et al. (24). Gene expression profiling of nonmalignant adult tissues and cells was done using the Eos Hu03, a customized Affymetrix GeneChip, as previously described (25).

Tissue specimens and cell lines. This study was done with approval of the Institutional Review Boards of the Cleveland Clinic and University of Arkansas for Medical Sciences with informed consent. Frozen and paraffin-embedded tissues were obtained from the archives of the Division of Pathology and Laboratory Medicine of the Cleveland Clinic. Frozen tissues were maintained at −80°C. Tissue microarrays were constructed from representative tumor areas using a manual tissue microarrayer (Beecher Instruments) with two or three cores per case (1.5 or 1.0 mm cores). Multiple myeloma cell lines (L363 and OPM2) were obtained from German Collection of Microorganisms and Cell Cultures. Fresh normal donor blood, blood/bone marrow samples from multiple myeloma patients, and excess stem cell apheresis samples were used for flow cytometry and antibody dependent cellular cytotoxicity studies.

Antibodies. Female BALB/c mice (Taconic Farms) were immunized with purified CS1 protein using standard immunization schedules. mAbs were generated by standard techniques with spleen cells being fused to the P3X63Ag8 fusion partner (American Type Culture Collection). Anti-CS1 specific mAbs were identified using a variety of techniques including ELISA for CS1 protein, Western blotting, and flow cytometry analysis on CS1-expressing cell lines. Specific anti-CS1 mAbs were selected for further studies, including MuLuc63 (IgG2_a) and MuLuc90 (IgG2_b), both of which recognize the extracellular region of CS1, and 1G9 (IgG1), which recognizes the intracellular region of CS1.

HuLuc63 is the humanized version of the mouse MuLuc63 mAb. Humanization was done using methods described by Queen et al. (26) and in the Supplementary Data. A low fucose version of HuLuc63 (HuLuc63-LF) was generated by stably transfecting the rat myeloma YB2/0 cell line with the HuLuc63 plasmids (27). The carbohydrate structure of HuLuc63-LF was confirmed by mass spectrometry on purified protein. YB2/0 cells have been shown to produce low fucose antibodies with enhanced binding to Fc γ-receptors and enhanced ADCC activity (28). To make a variant of HuLuc63 with decreased ADCC activity (HuLuc63-AA), a variant expression plasmid with two mutations in IgG1 CH2 domain (L234A/L235A) was used; these mutations decrease ADCC via decreased Fc γ-receptor binding (29).

MSL109, the control IgG1 antibody used in all experiments, is a fully human anticytomegalovirus mAb (30).

Immunohistochemistry. Immunohistochemistry (IHC) was done with MuLuc63 or 1G9 using an automated immunostainer with 3,3′-diaminobenzidine detection (Ventana Medical Systems and Dako Autostainer; Dako North America). The 1G9 antibody and heat-induced epitope retrieval using CC1 retrieval solution (30 min, 95°C; Ventana) was used for paraffin tissues. 1G9 staining in tumor tissues was scored as negative (0-10%), 1+ (11-25%), 2+ (26-50%), 3+ (51-75%), and 4+ (76-100%).

Flow cytometry. Flow cytometry was done using a FACSCalibur flow cytometer (Becton Dickinson) with CellQuest acquisition and analysis software (Becton Dickinson). Directly conjugated anti-CS1 (FITC) was used in multiparameter flow cytometry with commercially available mAbs (Becton Dickinson) conjugated to phycoerythrin (PE), peridinin-chlorophyll protein (PerCP), or allophycocyanin (APC) to allow specific lymphocyte subset, plasma cell, blast, or stem cell gates. For stem cell analysis, at least 100 CD34⁺ events were collected using standardized methods (31). For assessing binding of HuLuc63 to NK (CD16+CD56+), NK-like T cells (CD16+CD56+CD3+), and T cells (CD3+CD8+ or CD3+ CD8−), the antibody cocktails used were as follows: CD3 APC/CD8 PerCPCy5.5/CD16 (16+56) PE/HuLuc63 FITC, or MSL109 FITC. For assessing of HuLuc63 FITC to CD20+HLA-DR+ B cells and CD14+HLA-DR+ monocytes, the antibody cocktails used were as follows: CD20 APC/HLA-DR PerCP/CD14 PE/HuLuc63 FITC, or MSL109 FITC. For assessing HuLuc63 binding to granulocytes (CD13+ CD45+), the antibody cocktails used were as follows: CD3 APC/CD13 PE/CD45 PerCP/HuLuc63 FITC, or MSL109 FITC. CD138+ plasma cells were identified using the following antibody cocktail: CD138 PE/CD45 PerCP/CD3 APC/HuLuc63 FITC or MSL109 FITC. CS1 expression in cell populations was assessed as percentage positive cells (relative to isotype control) or as mean fluorescence intensity (MFI, mean channel number HuLuc63-FITC antibody/mean channel isotype control).

Whole blood lymphocyte depletion experiment. Whole blood from healthy volunteers was collected in sodium heparin anticoagulant Vacutainer tubes (BD Biosciences). The samples were kept at room temperature before use and were used on the same day. Samples were incubated for 24 h with HuLuc63 (100 and 200 μg/mL), MSL109 (100 and 200 μg/mL), rituximab (40 μg/mL; Genentech/Biogen Idec), alemtuzumab (100 μg/mL; Berlex laboratories/ILEX Pharmaceuticals/Genzyme corporation), or PBS at 37°C, 5% CO2. For each experimental condition, absolute cell counts were measured using TruCOUNT tubes (BD Biosciences) using the following combinations of antibody fluorochrome conjugates: Mix-1: CD3 FITC (Clone SK7; BD Biosciences), CD16-PE (B73.1; BD Biosciences), CD56-PE (MY31; BD Biosciences), CD19-APC (SJ2SC1; BD Biosciences), and CD45-PerCP (Clone 2D1; BD Biosciences); Mix-2: CD3 FITC, CD8-PE (SK1; BD Biosciences), CD4-APC (SK3; BD Biosciences), and CD45 PerCP; Mix-3: CD3 FITC, CD27-PE (M-T271; BD Biosciences), CD45 PerCP, and CD19 APC. Whole blood staining was done according to manufacturer's specifications (BD Biosciences). Samples were analyzed using a BD FACSCalibur flow cytometer, and data analysis was done according to the manufacturer's recommended procedure for TruCount combined with Multitest reagents.

ADCC. ADCC was assessed in standard 4-h ⁵¹CrO₄ release assays. NK cell effectors were purified from healthy allogeneic donors or multiple myeloma patients using CD56 antibody–coated magnetic beads (Miltenyi Biotech) and primary myeloma cell targets were purified from multiple myeloma patients using CD138 antibody–coated beads (Miltenyi Biotech). Chromium-labeled targets were incubated with HuLuc63 or MSL109 antibody at 10 μg/mL for 30 min before adding to wells containing effectors. For blocking studies, FcR antibody or isotype control antibody (Serotec) was added to wells at 10 μg/mL. Percent-specific lysis was calculated as: [experimental release − spontaneous release]/[maximum release − spontaneous release] × 100%.

In vivo myeloma xenograft model. The antitumor activity of HuLuc63 was evaluated in 6- to 8-wk-old female IcrTac:ICR-Prkdcscid mice (Taconic Farms), severe combined immunodeficient (SCID)-beige mice (C.B-Igh-1b/GbmsTac-Prkdc^scid-Lyst^bg N7; Taconic Farms), and NOD-SCID mice with a knockout in IL-2Rg (NOD-SCID IL2Rγ<null>; Jackson Laboratory). Mice were inoculated with 1 × 10⁷ OPM2 (DSMZ-German Collection of Microorganisms and Cell Cultures) cells into the lower right flank. Tumor measurements were done twice weekly by caliper as described previously (32). When tumors reached an average size of ∼100 mm³, animals were randomized into groups of 8 to 10 mice each and were injected i.p. with 5 to 10 mg/kg of antibody thrice a week for a total of 9 doses. For depletion of NK cells, ICR-SCID mice were treated with 0.1 mL of rabbit anti–asialo-GM1 antibody (Wako Chemicals USA, Inc.) 3 d before the start of antibody treatment, then twice per week for the duration of the study. Peripheral blood was monitored for NK cell depletion by staining with anti–CD45-PerCP and DX5-FITC (BD Biosciences). Tumor growth was monitored for a period of ∼1.5 mo. One-way ANOVA with a Tukey posttest was used to compute differences between groups. Experiments were terminated once tumors reached a size of >2,000 mm³. Animal work was carried out under NIH guidelines (“Guide for the Care and Use of Laboratory Animals”) using protocols approved by the Institutional Animal Care and Use Committee at PDL BioPharma.

CS1 is highly expressed in plasma cells and multiple myeloma. To identify plasma cell–specific antibody targets, a cDNA library was constructed by subtractive hybridization of naïve B-cell cDNA from memory B-cell and plasma cell cDNA. The aim was to identify gene products up-regulated in plasma cells that localize to the cell surface for antibody accessibility but have little to no expression in vital organs to minimize undesirable side effects of a targeted antibody. Genes with the desired expression profile were triaged by extensive bioinformatic analysis to determine their structural and functional classification, as well as their potential for cell surface localization. CS1, a gene encoding a cell-surface protein previously associated with NK cells (33, 34), was one of the genes that seemed to be highly expressed in plasma cells.

To examine the plasma cell expression in detail, gene expression profiling was done on CD138-purified plasma cells using the Affymetrix DNA microarray technology. High CS1 expression was revealed in plasma cells from healthy donors, patients with monoclonal gammopathies of undetermined significance, smoldering myeloma, and >500 patients with frank multiple myeloma at diagnosis (Fig. 1A). In comparison to the plasma cell samples, NK cells expressed substantially lower levels of CS1 mRNA. Using average intensity units (AIU) as an arbitrary measure of expression level, the median AIU for NK cells was 2,662, whereas the median AIU values for the plasma cell samples were 7,886 for multiple myeloma, 9,241 for smoldering myeloma, 6,207 for monoclonal gammopathies of undetermined significance, and 7,783 for normal plasma cells. We also observed that primary myeloma cells expressed higher levels of CS1 than human myeloma cell lines (median AIU, 4,867). In contrast, 15 normal nonlymphoid tissues (lung, uterus, kidney, stomach, brain, breast, spleen, prostate, skeletal muscle, testis, thymus, liver, ovary, heart, and small intestine) tested negative and exhibited a median AIU of 152 (range, 6-741; data not shown).

CS1 expression was observed in multiple myeloma from all patient populations, including high-risk and low-risk molecular profiles and those with and without cytogenetic abnormalities. CS1 expression also seemed high in all seven molecular subtypes of multiple myeloma, which included groups characterized by activating translocations involving c-MAF/MAFB, CCND1, CCND3, and MMSET/FGFR3, as well as groups characterized by hyperdiploidy, low-bone disease, and higher gene expression associated with proliferation (24). In an analysis of 256 Total Therapy 2 patients at baseline, median CS1 expression seemed highest in the low-bone disease subgroup, followed by MMSET/FGFR3, CCND3, hyperdiploidy, higher gene expression associated with proliferation, CCND1, and c-MAF/MAFB (ordered by highest to lowest expression; Fig. 1B). In addition, CS1 expression was maintained in multiple myeloma patients after relapse from therapy. A paired analysis of a subset of multiple myeloma patients showed that CS1 gene expression was detectable in patients both at diagnosis and at relapse (Fig. 1C). Although CS1 expression was lower at relapse in some of the patients, in all cases, expression levels were still high (median AIU, 2,386; range, 895-15,738).

To examine the expression of CS1 in normal tissues in more detail, we compared gene expression levels across 332 normal adult tissue samples representing 86 different organs, as well as 148 leukocyte samples representing 12 different cell types using a customized DNA microarray (35). CS1 expression was restricted to leukocytes and was not expressed in any major body organs (data not shown). Specifically, CS1 was expressed in NK cells, as previously published, as well as a subset of T cells, activated monocytes, and activated dendritic cells. The restricted expression profile of CS1 in normal cells and tissues, as well as the high uniform expression of CS1 in multiple myeloma suggest that CS1 may be a potential therapeutic antibody target for the treatment of multiple myeloma.

CS1 protein is highly expressed in normal and neoplastic plasma cells but not in normal tissue parenchyma or in a variety of solid tumors. To study CS1 protein expression, a panel of anti-CS1 mAbs was generated. Three mAbs were selected for further study by virtue of their specificity to CS1 (Fig. 2). All three recognized CS1 protein from cell extracts by Western blotting. Although the mAb clones MuLuc63 and MuLuc90 recognize distinct epitopes in the extracellular region of CS1, clone 1G9 is specific for the intracellular domain of CS1 (data not shown). IHC investigation showed that MuLuc63 and MuLuc90 stained only fresh frozen tissues, whereas 1G9 stained formalin-fixed paraffin-embedded samples.

To examine the expression pattern of CS1 in normal tissues, a panel of normal cryopreserved samples was stained with MuLuc63. Pericellular staining, indicating cell-surface expression, was detected only on infiltrating leukocytes in the following organs: mammary gland, tonsil, lymph node, spleen, stomach, duodenum, ileum, cecum, colon, bladder, ureter, urethra, kidney, cervix, uterus, prostate, ovary, vagus nerve, trigeminal ganglia, liver, and thyroid (see Fig. 3A for staining in cecum). Tissues double-stained by IHC showed that the majority of CS1-positive infiltrating cells also express CD138, a marker for plasma cells (data not shown). In lymphoid tissue, staining was observed in reactive germinal centers and in subepithelial areas. Rare small lymphocytes were also seen that expressed CS1 both within and outside germinal centers. No staining with MuLuc63 was detected in the epithelia, smooth muscle cells or vessels of heart, liver, lung, kidney, colon, small intestine, lymph node, spleen, tonsil, peripheral nerve, brain, spinal cord, aorta, adrenal, thyroid, pituitary, pancreas, parathyroid, bladder, cervix, ovary, endometrium, mammary gland, testes, retina, prostate, skeletal muscle, and skin.

We also stained a variety of normal and neoplastic tissues using the formalin-fixed paraffin-embedded reactive 1G9 antibody. Normal tissue IHC showed that all tissues were negative with the exception of plasma cells and scattered lymphocytes. Tissues (five examples of each type) included brain, breast, colon, heart, kidney, liver, lung, ovary, pancreas, prostate, skin, small intestine, spleen, stomach, testis, tonsil, uterus, and urinary bladder. IHC analyses of multiple solid tumors were negative for anti-CS1 staining. The tumor samples (n = 5 for each type) included ductal breast adenocarcinoma, lobular breast adenocarcinoma, renal cell carcinoma, prostate carcinoma, endometrial adenocarcinoma, gastric adenocarcinoma, urothelial carcinoma, small cell lung carcinoma, non–small cell lung carcinoma, ovarian carcinoma, testicular germ cell tumor, head and neck squamous cell carcinoma, melanoma, and pancreatic adenocarcinoma.

Analysis of a series of lymphomas and leukemias (Table 1) showed that CS1 expression was absent in the vast majority of acute leukemias, B-cell lymphomas, and classic Hodgkin lymphomas. In contrast, CS1 seemed to be strongly expressed in extramedullary plasmacytomas (8 of 8 cases were 2+ to 4+; Fig. 3B), in the neoplastic plasma cell component of 3 of 5 cases of lymphoplasmacytic lymphoma (Fig. 3C), and in the vast majority of tumor cells in multiple myeloma bone marrow cores (19 of 20 cases, 4+; Fig. 3D). Interestingly, the multiple myeloma case exhibiting less reactivity (scored 2+) expressed the mature B-cell marker CD20 as well as CD138 but was negative for IGH/CCND1 translocation. Two of five cases of splenic marginal zone lymphoma also expressed CS1 in a minority of cells. Approximately 25% of peripheral T-cell lymphomas expressed moderate levels of CS1, perhaps reflecting an origin from T cells naturally expressing CS1.

HuLuc63, a humanized anti-CS1 mAb, binds to lymphocyte subsets and to neoplastic plasma cells by flow cytometry. The gene and protein expression data suggest that CS1 is a potential therapeutic antibody target for the treatment of multiple myeloma. To investigate this possibility and in anticipation of treating multiple myeloma patients with an anti-CS1 antibody, the MuLuc63 anti-CS1 mAb was selected for humanization. As part of a safety evaluation of the humanized mAb, HuLuc63, we examined its binding profile to various leukocyte subsets in blood and bone marrow samples using flow cytomety. In normal donors, HuLuc63 stained primarily CD3+CD16+CD56+ NK-like T cells, CD3-CD16+CD56+ NK cells, and approximately half of CD3+CD8+ T cells. In contrast, substantially lower binding was detected on CD3+CD8- T cells, and no significant binding of HuLuc63 occurred on monocytes (CD14+HLA-DR+), B cells (CD20+HLA-DR+), or granulocytes (CD13+CD45+; Table 2).

HuLuc63 binding data were observed on similar subsets of leukocytes in bone marrow samples from multiple myeloma patients, except that binding to these subsets in bone marrow was found to be more variable between subjects (Table 2). In addition, the bone marrow samples contained between 1% and 20% CD138+ plasma cells, the target cells for the anti-CS1 antibody. The great majority of plasma cells in multiple myeloma stained strongly positive with HuLuc63, the staining intensity being substantially higher than the staining intensity on normal NK cells (Table 2).

Because multiple myeloma is often treated with autologous stem cell transplantation, we examined the potential concern of HuLuc63 binding to CD34+ stem cells. Flow cytometry analysis of CD34+ stem cells from apheresis samples collected for the purposes of autologous stem cell transplantation (six lymphoma patient samples) and two multiple myeloma patient bone marrow aspirate samples showed that HuLuc63 did not significantly bind stem cells (data shown only for the multiple myeloma samples in Table 2).

To determine the effects of HuLuc63 on cell counts of lymphocytes and their major subsets, whole blood samples from 8 healthy individuals were incubated with HuLuc63 (100 or 200 μg/mL) for 24 hours at 37°C. Several control antibodies were used in this study, including MSL109 as a negative control, rituximab (Rituxan; Genentech/Biogen Idec), an anti-CD20 antibody that depletes B cells in vitro and in vivo (36, 37), and alemtuzumab (Campath1H; Berlex/ILEX Oncology Inc/Genzyme), an anti-CD52 antibody that depletes various lymphocytes in vivo (38) as positive controls. Samples were tested for depletion of T, NK, B, and memory B cells as measured by a decrease in absolute cell counts by flow cytometry. The results indicated no apparent effect of HuLuc63 on total lymphocytes, T-, and B-cell counts (data not shown). On average, NK cell counts were decreased by 20% at both doses of HuLuc63, although the observed decline was variable between donors. In contrast, rituximab treatment resulted in an average decrease of 70% of all B cells, and alemtuzumab treatment resulted in depletion of all lymphocyte subsets (range of 65-100%). Overall, these data show that HuLuc63 treatment of normal healthy blood did not result in substantial depletion of major lymphocyte subsets.

HuLuc63 mediates ADCC and in vivo antimyeloma activity that is dependent on NK cells. Having shown expression of CS1 on the surface of neoplastic plasma cells at relatively high levels compared with normal NK and T cells, we sought to determine whether HuLuc63, a humanized IgG1 antibody, could mediate ADCC toward primary myeloma cells isolated from multiple myeloma patients. The results showed that HuLuc63 induced specific myeloma-cell lysis in multiple assays using purified NK cells from healthy allogeneic donors (five assays) or autologous NK cells from multiple myeloma donors (six assays) as effectors (Fig. 4A). Similar results were obtained by Tai et al. (32) using peripheral blood mononuclear cells (PBMC) as effectors. Blocking the Fc receptor (CD16) on NK cells with an anti-CD16 antibody significantly inhibited this activity (Fig. 4A). In separate experiments using human PBMC effectors, depletion of NK cells significantly impaired HuLuc63-mediated ADCC toward the OPM2 myeloma cell line (Fig. 4B). In contrast, depleting T cells, B cells, or monocytes from a PBMC effector population had no effect on HuLuc63-mediated ADCC (data not shown), suggesting that NK cells are the dominant effectors for this activity.

In parallel with our studies, Tai et al. (32) showed that HuLuc63 exhibits significant dose-dependent antitumor activity using myeloma xenograft models in vivo in mice. Because our in vitro data suggested that HuLuc63-mediated ADCC activity was dependent on NK cell and Fc receptor function, we tested the contribution of the Fc region of HuLuc63 to in vivo activity. Two modified forms of HuLuc63 were produced; one form had low-fucose content (HuLuc63-LF) and bound with high affinity to human CD16a (affinity constant (K_D), 0.6 μmol/L compared with 2.1 μmol/L for wild-type HuLuc63), whereas the second form contained a double L234A/L235A mutation (HuLuc63-AA) and had a lower affinity toward human CD16a (KD, 7.6 μmol/L). The modified forms of HuLuc63 were tested with wild-type HuLuc63 in the OPM2 xenograft model for in vivo antitumor activity. The results show that higher antitumor activity correlated with increased affinity of antibody to the Fc receptor. Treatment with wild-type HuLuc63 resulted on average in a 70% decrease in tumor volumes compared with control isotype antibody treatment, whereas HuLuc63-LF caused an average of 85% decrease in tumor volume, although these groups were not significantly different from each other (Fig. 4C). In contrast, HuLuc63-AA did not exhibit any antitumor activity compared with wild-type HuLuc63 (Fig. 4C). These results combined with the in vitro ADCC results suggest that HuLuc63 exerts antimyeloma activity through NK cell–mediated ADCC via binding of CD16 expressed on the surface of NK cells.

To further explore this hypothesis, HuLuc63-mediated antitumor activity was explored in mice that had been either depleted of NK cells or are deficient in NK cell function or numbers. In a single study, NK cell depletion abolished the HuLuc63-mediated antitumor effect (Fig. 4D). Similarly, there was no statistically significant difference between HuLuc63 and isotype control on antitumor activity in the NK cell–deficient NOD-SCID/IL-2Rγ knockout mice (Fig. 4D; ref. 39).

We have identified CS1, a member of the signaling lymphocyte activating–molecule family of cell surface receptors, as a potential novel target for the treatment of multiple myeloma. We also generated a humanized anti-CS1 mAb, HuLuc63 that recognizes the extracellular region of CS1 and targets myeloma cells for ADCC. The goals of this study were to examine in detail the expression of CS1, to characterize the cellular binding properties of HuLuc63, and examine the potential of HuLuc63 as a therapeutic antibody candidate for the treatment of multiple myeloma.

Our gene expression profiling data showed that CS1 was highly prevalent in multiple myeloma. The expression did not correlate with any of the molecular classifications in multiple myeloma (24), suggesting that this target is not restricted to any particular multiple myeloma subgroup. A separate gene expression study, done in parallel, was in agreement with our results that CS1 was highly expressed and prevalent in multiple myeloma (32). In addition, multiple myeloma patients that had been treated with either Total Therapy 2 (40) or Total Therapy 3, continued to express CS1 at the time of relapse, indicating that significant chemotherapeutic regimen do not markedly affect the expression of CS1 in the myeloma cells. Using IHC, we also showed high-level expression in all cases of multiple myeloma and plasmacytomas. In addition, cases of secondary plasma cell leukemia were also positive. Thus, plasma cell malignancy in bone marrow, tissue, and blood all seemed to express high levels of CS1.

Examination of a series of B-cell lymphomas also showed that CS1 expression was not observed at any appreciable degree in mature B-cell neoplasms without plasmacytic differentiation (small lymphocytic lymphoma, mantle cell lymphoma, follicular lymphoma, diffuse large B-cell lymphoma, and Burkitt lymphoma). Cases of lymphoplasmacytic lymphoma showed CS1 expression in the neoplastic plasmacytic components, whereas small mature B-cells lacked CS1. Marginal zone lymphomas, which may also show plasmacytic differentiation, weakly expressed CS1 Thus, even in neoplastic lymphoid tumors, CS1 was seen only in terminally differentiated plasma cells. Interestingly, a subset of T-cell lymphomas also expressed CS1, possibly reflecting derivation from a normal counterpart expressing CS1 or perhaps an activated state. Using this method, we were unable to show significant expression in small lymphocytes in B- and T-cell zones of lymphoid tissues, possibly due to decreased sensitivity of this assay compared with flow cytometry.

The expression of CS1 in malignant plasma cells, combined with the lack of expression in the parenchyma of any major organs, makes CS1 an attractive target for the treatment of multiple myeloma. In comparison, other potential antibody targets for multiple myeloma treatment such as CD138 (syndecan-1) and CD40 are expressed in other tissues (41–49). The normal tissue profile of CS1 suggests that there does not seem to be major concern for tissue toxicity and/or a sink for an antibody-directed therapy. Importantly, hematopoietic stem cells, which are often used in the treatment of multiple myeloma in the form of autologous stem cell transplantation, do not express CS1, allowing the possibility of combining a potential HuLuc63 treatment with autologous stem cell transplantation.

In a study exploring potential efficacy, we showed that HuLuc63 can elicit allogeneic and autologous NK cell–mediated ADCC toward primary myeloma cells from multiple myeloma patients. In a parallel study, HuLuc63 elicited ADCC using PBMC effector cells but does not elicit complement-dependent cytotoxicity (32). A role for ADCC was supported by our in vivo xenograft studies in which OPM2 tumor growth inhibition with HuLuc63 treatment was dependent on Fc–Fc-receptor interaction, as well as the presence of functional NK cells. HuLuc63 was also reported to have direct effects on the survival of primary myeloma cells in vitro, presumably by inhibiting interaction with bone marrow stromal cells (32). However, a direct effect on the in vitro survival and/or proliferation of OPM2 and L363 myleoma cells was not observed in our study (data not shown). Therefore, although other potential mechanisms of action cannot be ruled out, our data supports the hypothesis that the primary mechanism of action of HuLuc63 involves NK-mediated cell killing.

In summary, CS1 is expressed at high levels in benign and neoplastic plasma cells derived from multiple myeloma patients. Importantly, CS1 is expressed in multiple myeloma samples from extensively treated patients on currently approved therapies. HuLuc63, a humanized anti-CS1 mAb, mediates in vitro and in vivo antimyeloma activity and, therefore, shows promise as a therapeutic agent. Currently, HuLuc63 is in a phase 1 safety trial for the treatment of relapsed/refractory multiple myeloma.

B. Balasa., A. Draksharapu, B.P. Shum, M. Huseni, D. Powers, A. Nanisetti, Y. Zhang, A.G. Rice, A.V. Abbema, M. Wong, G. Liu, M. Dillon, S. Chen, S. Rhodes, F. Fuh, N. Tsurushita, S. Kumar, V. Vexler, D.E.H. Afar, and M.B. Williams are current or past employees and shareholders of PDL BioPharma, Inc.

We thank Nicholas Landolfi for his guidance and support, in whose laboratory the discovery of CS1 as a candidate therapeutic target was conducted; J. Yun Tso who initiated the collaboration with the Cleveland Clinic and whose suggestion that CS1 may be a potential target for the treatment of multiple myeloma, led to the work presented in this paper; Gary Starling, Debbie Law, and Renee Kosslak (PDL BioPharma) for critical review of the manuscript; the nursing and support staff of the Myeloma Program at the Cleveland Clinic for their help in procuring blood and bone marrow samples; and Katie Stone and Priyangi Malaviarachchi are acknowledged for excellent technical assistance.