Multiple myeloma (MM) is an incurable hematological malignancy. Chimeric antigen receptor (CAR)-expressing T cells have been demonstrated successfully in the clinic to treat B-lymphoid malignancies. However, the potential utility of antigen-specific CAR-engineered natural-killer (NK) cells to treat MM has not been explored. In this study, we determined whether CS1, a surface protein that is highly expressed on MM cells, can be targeted by CAR NK cells to treat MM. We successfully generated a viral construct of a CS1-specific CAR and expressed it in human NK cells. In vitro, CS1-CAR NK cells displayed enhanced MM cytolysis and interferon-γ (IFN-γ) production, and showed a specific CS1-dependent recognition of MM cells. Ex vivo, CS1-CAR NK cells also showed similarly enhanced activities when responding to primary MM tumor cells. More importantly, in an aggressive orthotopic MM xenograft mouse model, adoptive transfer of NK-92 cells expressing CS1-CAR efficiently suppressed the growth of human IM9 MM cells and also significantly prolonged mouse survival. Thus, CS1 represents a viable target for CAR-expressing immune cells, and autologous or allogeneic transplantation of CS1-specific CAR NK cells may be a promising strategy to treat MM.
Multiple myeloma (MM) is a B-cell malignancy characterized by the aberrant clonal expansion of plasma cells (PCs) within the bone marrow,1 with an estimated 21 700 new cases and 10 710 deaths from MM identified in the United States in 2012.2 Despite the use of proteasome inhibitors and immune-modulating drugs, which have improved overall survival,3 MM remains an incurable malignancy4 for which novel therapeutic approaches are urgently needed.
Natural-killer (NK) cells are CD56+CD3− large granular lymphocytes that can kill virally infected and transformed cells, and constitute a critical cellular subset of the innate immune system.5 Unlike cytotoxic CD8+ T lymphocytes, NK cells launch cytotoxicity against tumor cells without the requirement for prior sensitization, and can also eradicate MHC-I-negative cells.6 NK cells are safer effector cells, as they may avoid the potentially lethal complications of cytokine storms,7 tumor lysis syndrome,8 and on-target, off-tumor effects. Although NK cells have a well-known role as killers of cancer cells, and NK cell impairment has been extensively documented as crucial for progression of MM,5, 9 the means by which we might enhance NK cell-mediated anti-MM activity has been largely unexplored.
One intriguing approach for treating cancer involves the genetic modification of T cells or NK cells with chimeric antigen receptors (CARs) that directly target tumor-associated antigens. In fact, CAR T cells have been demonstrated successfully in the clinic for treatment of CD19+ acute lymphoblastic leukemia and chronic lymphocytic leukemia.8, 10, 11, 12 However, treatment of patients with CAR T cells can result in cytokine storms7, 10 and, in the setting of allogeneic transplantation, may induce graft-versus-host disease. While CAR NK cells are thought to be safer effector cells, their therapeutic use has not been explored. In myeloma, unique surface antigens have been challenging to target.13 Recent studies have suggested that the cell-surface glycoprotein, CS1, may be an ideal target for the treatment of MM.14, 15, 16 CS1 is highly, and nearly ubiquitously, expressed on MM cells, while expression remains low on NK cells, some T-cell subsets, and normal PCs, and almost undetectable on myeloid cells and the majority of healthy tissues.14 Importantly, CS1 is not expressed on hematopoietic stem cells,14 which are often used in the form of autologous stem cell transplantation for treatment of MM. While the exact role of CS1 in normal PCs is unknown, CS1 co-localizes with CD138 on the surface of polarized MM cells, and may promote MM cell adhesion, clonogenic growth and tumorigenicity via c-Maf-mediated interaction with bone marrow stromal cells.1, 16, 17 Targeting CS1 with the humanized IgG1 monoclonal antibody (mAb), Elotuzumab (formerly Huluc63), can inhibit myeloma cell adhesion to bone marrow stromal cell, induce NK cell-mediated antibody-dependent cellular cytotoxicity and promote NK cell activation without killing autologous NK cells despite low levels of CS1 expression on the surface of normal NK cells.16, 18 Importantly, Elotuzumab has already been proven to be safe in phase 1 and 2 clinical trials, and phase 3 trials are ongoing.19 This suggests that CS1-targeted cytotoxic leukocytes will not impair major immune cell subsets and hematopoietic stem cells. Using NK cells engineered to express CS1-CAR is promising as a means to augment the anti-MM capacity of NK cells.
In this study, we engineered human NK cells to express a CAR that was CS1 specific, and we incorporated a CD28-CD3ζ co-stimulatory signaling domain. We evaluated the anti-MM function of these cells in vitro and in an in vivo orthotopic xenograft mouse model of MM. Our results showed that the expression of the CS1-CAR could redirect NK cells to specifically and efficiently eradicate CS1-expressing MM cells, both in vitro and in vivo, and this eradication was CS1 dependent. Our data suggest that this CAR strategy may be suitable for the development of an effective NK cell-based immunotherapy as a means to treat patients with refractory or relapsed MM. In addition, in contrast to CAR T cells, CAR NK cells allow us to use allogeneic NK cell sources, which are less likely to cause and may even help to suppress graft-versus-host disease,20 while also potentiating an increase in cytotoxicity due to mismatched killer immunoglobulin-like receptors (KIRs).21
Materials and methods
Six- to eight-week-old NOD.Cg-prkdcscid IL2rgtm1Wjl/szJ (NSG) mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA). All animal work was approved by The Ohio State University Animal Care and Use Committee. Mice were monitored frequently for MM disease progression, and killed when they were moribund with the symptoms of hindlimb paralysis, lethargy, and obvious weight loss.
Generation of anti-CS1 CAR lentiviral construct
The CS1-scFv fragment, amplified from the hybridoma cell line Luc90, was fused with a sequence encoding a Myc tag immediately following the CS1-VL-encoding sequence. The fused DNA sequences were incorporated with CD28-CD3ζ that was incised from a retroviral vector. The entire CS1-scFv-myc tag-CD28-CD3ζ fragment was ligated into a lentiviral vector designated as PCDH-CMV-MCS-EF1-copGFP (PCDH, System Biosciences, Mountain View, CA, USA) to generate a PCDH-CS1-scFv-myc tag-CD28-CD3ζ (PCDH-CS1-CAR) construct.
Lentivirus production and transduction of NK cells
To produce lentivirus for infection of NK cells, 293T cells cultured in DMEM (Life Technologies, Grand Island, NY, USA) were co-transfected with the aforementioned PCDH-CS1-scFv-CD28-CD3ζ plasmid or a mock PCDH control vector together with the packaging constructs pCMV-VSVG and pCMV-dr9 using calcium phosphate transfection reagent (Promega, Madison, WI, USA). The transfection and infection procedures were modified from a previously published protocol22 and are detailed in Supplementary Information.
Generation of a U266 cell line stably expressing CS1
Human CS1 coding sequences were amplified from cDNA isolated from IM9 cells via PCR, then subcloned into a PCDH lentiviral vector to generate a PCDH-CS1 construct. Lentivirus production and infection of U266 cells were performed using the methods described above. GFP-positive cells were then sorted using an FACS Aria II cell sorter (BD Biosciences, San Jose, CA, USA).
Cells were washed with phosphate-buffered saline and directly lysed in Laemmli buffer (Bio-Rad Laboratories, Hercules, CA, USA). Lysates were electrophoretically separated on a 4–15% gradient SDS–PAGE gel (Bio-Rad Laboratories) and transferred onto a nitrocellulose membrane. Subsequent procedures were modified from a previously published protocol,22 and are detailed in Supplementary Information.
Interferon-γ release assay
Myeloma target cells were co-cultured with NK effector cells in 96-well V bottom plates for 24 h. In all, 2.5 × 105 myeloma cell line cells or 1.0 × 105 primary myeloma cells were incubated with 2.5 × 105 or 5.0 × 105 NK cells, respectively. Cell-free supernatants were assayed for interferon-γ (IFN-γ) secretion by enzyme-linked immunosorbent assay (ELISA) using a kit from R&D Systems (Minneapolis, MN, USA) according to the manufacturer’s protocol. Data depicted in figures represent mean values of triplicate wells from one of three representative experiments with similar results.
An orthotopic MM mouse model and in vivo treatment of MM-bearing mice and bioluminescence imaging
IM9 cells were retrovirally transduced with Pinco-pGL3-luc/GFP virus expressing firefly luciferase as previously described.24 GFP-positive cells were sorted using an FACS Aria II cell sorter (BD Biosciences), and were designated as ‘IM9-Luc’ cells. Then, six- to eight-week-old male NSG mice were intravenously (i.v.) injected with 0.5 × 106 IM9-Luc MM cells in 400 μl of phosphate-buffered saline via tail vein on day 0 to establish a xenograft orthotopic MM model. Beginning on day 7, the mice were i.v. injected with 5 × 106 effector cells, that is, CS1-CAR NK-92 cells or mock-transduced control cells, in 400 μl of phosphate-buffered saline once every 5 days (five times in total). Four weeks after IM9-Luc inoculation, the mice were intraperitoneally (i.p.) infused with D-luciferin (150 mg/kg body weight; Gold Biotechnology, St. Louis, MO, USA), anesthetized with isoflurane, and imaged using an In Vivo Imaging System (IVIS-100, Perkin-Elmer, Waltham, MA, USA) with the Living Image software (Perkin-Elmer).
Spinal vertebrae were fixed in 10% buffered formalin phosphate and decalcified in saturated EDTA, and then embedded in paraffin. Five-micron thick sections were stained with hematoxylin and eosin (H&E) for histological examination. The sections were immunostained for identification of human MM cells with mouse anti-human CD138 mAb (1:50 dilution; Thermo-Scientific, Waltham, MA, USA) following standard immunohistochemistry staining procedures. Horseradish peroxidase-conjugated anti-mouse IgG was used as a secondary antibody, followed by a peroxidase enzymatic reaction.
Unpaired Student’s t test was utilized to compare two independent groups for continuous end points if normally distributed. One-way ANOVA was used when three or more independent groups were compared. For non-normally distributed end points, such as in vivo bioluminescence intensity, a Kruskal–Wallis test was utilized to compare the median of NK-92-CS1-CAR to NK-92-EV and phosphate-buffered saline groups. For survival data, Kaplan–Meier curves were plotted and compared using a log-rank test. All tests are two-sided. P-values were adjusted for multiple comparisons by the Bonferroni method. A P-value of <0.05 is considered as statistically significant.
Generation of NK-92 and NKL NK cells expressing CS1-CAR
We generated a specific CS1-CAR construct with a PCDH lentiviral vector backbone, sequentially containing a signal peptide (SP), a heavy chain variable region (VH), a linker, a light chain variable region (VL), a Myc tag, a hinge, CD28 and CD3ζ (Figure 1a). NK-92 and NKL NK cell lines were transduced with the CAR construct and then sorted for expression of GFP, a marker expressed by the vector. Western blotting of the sorted cells demonstrated that CS1-CAR was successfully introduced and expressed, as evidenced by the expression of the chimeric CS1-scFv receptor containing CD3ζ in both NK-92 and NKL cell lines transduced with the CAR construct rather than with the control vector (Figure 1b). Moreover, a flow cytometric analysis after anti-Myc Ab surface staining indicated that CS1-CAR was expressed on the surface of both NK-92 and NKL cells transduced with the CS1-CAR construct (Figure 1c).
CS1-CAR-modified NK cells more effectively eradicate CS1+ MM cells, but not CS1− cells, in vitro in comparison with mock-transduced NK cells
After generating the CS1-CAR NK cells, we determined whether they selectively kill CS1+ better than CS1− MM cells. For this purpose, we first confirmed that IM9 and L363 MM cell lines constitutively expressed CS1 protein on their surface, while expression of CS1 was negligible in U266 MM cells (Figure 2a). Next, a 4-h chromium-51 release assay indicated that, compared with mock-transduced NK-92 cells, NK-92 cells transduced with CS1-CAR were significantly enhanced in their ability to kill CS1+ IM9 and L363 cells (Figures 2b and c, left panels). Similar data were observed in experiments repeated using NKL cells transduced with CS1-CAR (Figures 2b and c, right panels). However, both the CS1-CAR- and mock-transduced NK-92 or NKL cells were similar in their low levels of cytotoxicity against CS1− U266 myeloma cells (Figure 2d). In addition, forced expression of CS1-CAR did not induce obvious apoptosis in NK-92 or NKL cells as determined by analyses of 7AAD/Annexin V-staining using flow cytometry (Supplementary Figure 1), suggesting that CS1-CAR expression did not cause cytotoxicity to the NK-92 or NKL cells themselves. Similarly, we also found that CS1-CAR expression in purified primary human NK cells augmented their cytotoxicity against CS1+ IM9 myeloma cells (data not shown).
CS1-CAR-modified NK cells secrete more IFN-γ than mock-transduced NK cells do after exposure to CS1+ MM cells
The signaling domain of the CD28 co-stimulatory molecule, which we included in our CAR construct, may enhance activation after recognition of the CS1 scFv with the CS1 antigen on the surface of MM cells. Therefore, the inclusion of this signaling domain may have the capacity to activate NK cells not only to have higher cytotoxicity, but also to produce more IFN-γ, the latter of which is also important for tumor surveillance and activation of CD8+ T cells and macrophages.25, 26, 27 To test this, CS1-CAR-modified or control-engineered effector NK cells were either cultured alone or co-cultured with CS1+ myeloma cells including the IM9 and L363 MM cell lines. After 24 h, the IFN-γ production was measured by ELISA. As shown in Figure 3, both CS1-CAR-modified and mock-transduced NK-92 or NKL cells spontaneously produced low or negligible levels of IFN-γ when incubated alone. Co-culture with CS1+ MM tumor cells (IM9 or L363) induced IFN-γ in both CS1-CAR and mock-transduced NK-92 or NKL cell lines; however, significantly higher levels of IFN-γ were produced by CAR-modified NK-92 or NKL cells than by mock-transduced NK-92 (Figures 3a and b, left panels) or NKL cells (Figures 3a and b, right panels). When co-cultured with the CS1− MM cell line, U266, both mock-transduced and CS1-CAR-transduced NK-92 cells but not the transduced NKL cells produced higher levels of IFN-γ than corresponding cells that had not been co-cultured with U266 cells (Figure 3c). This suggests that a unique interaction between NK cell receptors on NK-92 cells and their ligands on U266 cells may induce CS1-independent IFN-γ production by NK-92 cells. Moreover, CS1-CAR-transduced NK-92 and NKL cells failed to produce more IFN-γ than corresponding mock-transduced NK-92 and NKL cells when they were co-cultured with U266 cells (Figure 3c). These results are in agreement with the aforementioned cytotoxicity data, and together indicate that modification with CS1-CAR can dramatically enhance NK cell effector functions, in terms of both cytotoxicity and IFN-γ production, in response to CS1+ but not to CS1− myeloma cells.
Enforced CS1 expression in U266 cells enhances cytotoxicity and IFN-γ production of NK-92-CS1-CAR cells
We next explored whether this enhanced activity of CS1-CAR NK cells relies on CS1 antigen expression on MM cells. Our aforementioned observation—that the introduction of CS1-CAR conferred NK-92 cells with increased cytotoxic activity and enhanced IFN-γ production in response to CS1+ myeloma cells, but not to CS1− U266 myeloma cells—prompted us to investigate whether CS1 overexpression in U266 cells is sufficient to change the sensitivity of U266 cells to NK-92-CS1-CAR cells. For this purpose, we ectopically expressed CS1 in U266 cells by lentiviral infection. Flow cytometric analysis confirmed that CS1 protein was successfully expressed on the surface of the U266-CS1 cells (Figure 4a). Chromium-51 release assay indicated that, when compared with mock-transduced NK-92 cells, there was a significant increase in the cytotoxic activity of CS1-CAR-transduced NK-92 cells toward U266 cells overexpressing CS1 (Figure 4b). Likewise, compared with parallel co-cultures containing mock-transduced NK-92 cells, NK-92-CS1-CAR cells co-cultured with U266 cells overexpressing CS1 secreted significantly higher levels of IFN-γ (Figure 4c). However, consistent with data in Figures 2d and 3c, there was no difference in cytotoxicity and IFN-γ secretion between NK-92-CS1-CAR cells and mock-transduced NK-92 cells when they were incubated with U266 cells transduced with an empty vector control (Figures 4b and c). These results suggested that the increased recognition and killing of myeloma cells by NK-92-CS1-CAR cells occurs in a CS1-dependent manner.
Phenotypic characterization of NK-92-CS1-CAR cells
We next investigated whether the expression of a CS1-specific CAR could change the NK cell phenotype. We employed flow cytometry to compare expression of antigens on the surface of CS1-CAR-transduced and mock-transduced NK-92 cells, following culture in the presence or absence of IM9 myeloma cells. As shown in Figure 5a, we observed that there was no difference between CS1-CAR- and mock-transduced NK-92 cells, whether cultured in the presence or absence of IM9 cells, in the expression of NK cell receptors including NKp30, NKp46, NKG2C and NKG2D. We also assessed expression of the NK cell activation markers, CD6928 and HLA-DR.29, 30 We found that recognition of IM9 cells did not elicit CD69 expression on mock-transduced NK-92 cells, yet induced a moderate, but significant, increase in CD69 expression on the surface of CS1-CAR-transduced NK-92 cells (Figure 5a). Interestingly, co-incubation with IM9 cells caused a dramatic increase in the expression of HLA-DR in both CS1-CAR-transduced and mock-transduced NK-92 cells. In the absence of IM9 target cells, there was no obvious difference in HLA-DR expression between CS1-CAR-transduced and mock-transduced NK-92 cells; however, upon stimulation with IM9 cells, the expression of HLA-DR became significantly higher in NK-92-CS1-CAR cells than in mock-transduced NK-92 cells. Thus, the increase in the activation markers, especially HLA-DR, expressed on NK-92-CS1-CAR cells may have occurred in connection with the enhanced cytotoxicity and IFN-γ production by these cells when they are exposed to CS1+ MM cells. Using intracellular staining, we also observed that, when compared with mock-transduced NK cells, NK-92-CS1-CAR cells had significantly higher levels of perforin and granzyme B expression, even in the absence of MM tumor cells (Figures 5b and c). This is consistent with a previous report regarding the elevated expression of granzyme B in CAR T cells,31 and also consistent with the fact that perforin and granzyme B expression are generally correlated with cytotoxic activity of NK cells.32
CS1-CAR-transduced NK-92 cells more effectively recognize and kill NK-resistant primary MM cells ex vivo
To make our findings more clinically relevant, we investigated whether CS1-CAR-modified NK-92 cells also harbored enhanced cytolytic activity and IFN-γ production when recognizing primary MM cells ex vivo. Flow cytometry was used to assess surface expression of CS1 on primary CD138+ magnetic bead-selected MM cells from MM patients (Figure 6a). In accordance with the previous report, showing that CS1 protein was highly expressed on CD138 magnetic bead-purified MM patient cells,14, 15 we observed that CS1 protein was indeed uniformly expressed on the surface of primary MM cells (Figure 6a). By chromium-51 release assay, we found that primary myeloma cells freshly isolated from MM patients were highly resistant to NK-92 cell-mediated lysis even at E:T ratios as high as 40:1 and 20:1; however, this resistance could be partially overcome by NK-92 cell expression of CS1-CAR, which resulted in a dramatic increase in eradication of primary myeloma cells (Figure 6b). In line with the cytotoxicity result, after 24 h co-culture with primary myeloma cells, CS1-CAR-transduced NK-92 cells also secreted significantly higher levels of IFN-γ than mock-transduced NK-92 cells (Figure 6c).
CS1-CAR-transduced NK-92 cells inhibit MM tumor growth and prolong survival of tumor-bearing mice in an orthotopic xenograft MM model
To further address the potential therapeutic application of NK-92-CS1-CAR cells, we examined their antitumor activity in IM9-xenografted NSG mice. We generated an IM9 cell line expressing firefly luciferase (IM9-Luc) by retrovirally transducing IM9 cells with virus expressing firefly luciferase, then performing GFP-based cell sorting. The expression of full-length firefly luciferase mRNA was confirmed by RT-PCR (Supplementary Figure 2a). Like typical myeloma cells, IM9-Luc cells expressed CD138 protein on their surface (Supplementary Figure 2b). In agreement with a previous report,33 we observed that IM9-Luc-bearing NSG mice displayed disseminated disease, manifested by hindlimb paralysis and motor ataxia. Histological examination of spinal vertebrae in a mouse displaying hindlimb paralysis showed the presence of numerous tumor cells and osteolytic lesions in bone tissue (Figure 7a). Immunohistochemical staining with human-specific anti-CD138 antibody further confirmed the presence of tumor cells (Figure 7b). Bioluminescence imaging was used to monitor the IM9-Luc tumor growth. As shown in Figures 7c and d, and in agreement with the in vitro cytotoxicity data, comparing the mice who later received injections with mock-transduced control cells, IM9-Luc tumors were significantly suppressed in mice who instead later were administered NK-92-CS1-CAR cells. Moreover, treatment with NK-92-CS1-CAR cells significantly prolonged the survival of mice bearing IM9-Luc tumors as compared with treatment with the mock-transduced NK-92 control cells (Figure 7e). Of note, when NK-92-CS1-CAR cells or mock-transduced NK-92 cells were similarly administered, but without i.v. injection of IM9-Luc cells, mice did not develop disseminated disease or die (data not shown).
Genetic manipulation of cytotoxic immune cells, including T cells and NK cells, to express a CAR that is specific for a tumor-associated antigen has emerged as a promising strategy for treating hematological malignancies.8, 10, 11, 12 Identification of suitable target antigens is a prerequisite for developing CAR-expressing immune cell therapies against malignancies, yet this has been the biggest hurdle for developing CAR T or NK cells.13 Substantial progress has been made recently for the development of CAR immune cell therapy against B-cell lineage malignancies, among which CD19-directed CAR T cells have been demonstrated to induce prolonged remission in advanced B-cell acute lymphoblastic leukemia and chronic lymphocytic leukemia.8, 10, 11, 34 Unfortunately, CD19-directed CAR T cells cannot be utilized to treat MM patients because over 95% of MM patients lack the expression of CD19 on their tumor cells.35 Although many other surface antigens, such as CD40, CD56, CD138 and CD74 can also be expressed on MM cells, the clinical utility of each of these antigens is very limited given their lack of specificity for MM cells or variability between patients.15 Both CD38 and B-cell maturation antigen were recently identified as promising immunotherapeutic targets in MM.36, 37 B-cell maturation antigen was found to be expressed by normal PCs and subsets of mature B cells,38, 39, 40 and CD38 has widespread expression on hematopoietic and non-hematopoietic tissues as well as on hematopoietic progenitor cells,41 despite the relative safety of naked CD38 antibodies in phase 1 trials.42 With a restricted pattern of expression in normal tissues, CS1 is highly and uniformly expressed on MM cells from nearly all patients.14 CS1 has low expression on NK cells and subsets of T cells and is almost undetectable on myeloid cells and hematopoietic stem cells. Moreover, CS1 expression is also maintained on MM cells of patients even after disease relapse.14 In addition, mounting evidence has suggested that CS1-specific immunotherapy can target neoplastic cells without inducing major damage to normal cells, including immune cells like T and NK cells in MM patients.43 On the basis of all these findings, it is tempting to speculate that CS1 may be an ideal antigen for targeting CAR NK cells against MM. In the present study, we have demonstrated for the first time that manipulating human NK cells to express a CS1-specific CAR can dramatically enhance their killing of CS1+ myeloma cell lines and primary myeloma cells, and can also augment their secretion of IFN-γ in vitro. More importantly, NK-92-CS1-CAR cells more efficiently eradicate human IM9 tumors established in NSG mice, resulting in improved overall survival of these IM9-bearing mice. All of this evidence strongly corroborates our hypothesis that CS1 represents an appropriate tumor antigen target for CAR NK cells against MM.
In contrast to first-generation CARs, which typically only bear the intracellular domain from CD3ζ, second-generation CARs, which have already been widely used for generating and studying CAR-expressing T or NK cells, usually incorporate CD28 or 4-1BB (CD137) as a co-stimulatory signal,44, 45, 46 and both of these co-stimulatory signals have been shown functional in CAR T cells.8, 10, 11, 12 In this study, we showed that NK cells expressing a CS1-specific CAR containing CD28 and CD3ζ signaling moieties display enhanced anti-MM activity, both in vitro and in vivo. Consistent with our study, others have demonstrated that primary NK cells grafted with a HER-2-specific CAR harboring CD28 and CD3ζ signaling moieties specifically recognized and efficiently eradicated HER-2-positive carcinoma.45 Regarding 4-1BB-CD3ζ CAR in NK cells, it has been reported that primary NK cells grafted with a CD19-specific CAR harboring 4-1BB-CD3ζ displays increased cytokine production and cytotoxic activity toward leukemic cells.46 Therefore, it appears that the tumor antigen-specific CARs carrying either CD28-CD3ζ or 4-1BB-CD3ζ signaling moieties can efficiently redirect NK cells to specifically target and kill tumors cells expressing the corresponding tumor antigens, yet a direct comparison is needed to address which one is superior.
CAR T cells have been effectively used for treatment of refractory chronic lymphocytic leukemia and acute lymphoblastic leukemia.8, 10, 11, 12 One advantage of CAR NK cells as opposed to CAR T cells is that CAR T cells can induce a cytokine storm in the clinic,7, 10 while NK cells might be safer effector cells due to the lack of clonal expansion; on the other hand, this might limit efficacy. Another advantage of CAR NK cells versus CAR T cells is that CAR NK cells may be used in the setting of allogeneic transplantation, enhancing an all too often weak graft-versus-tumor effect without inducing graft-versus-host disease.20 We speculate that co-administration of CAR NK cells, either at the time of allograft or post-transplant, could enhance the graft-versus-tumor effect. Moreover, allogeneic NK cells should be more potent than autologous NK cells for lysis of MM cells due to mismatched KIR effects.21, 47 It is noteworthy that normal CD34+ hematopoietic stem cells, which are often utilized at the time of autologous stem cell transplant, do not express CS1, suggesting an opportunity for CAR NK cell therapy at a time when there is minimal residual disease.14
It is generally believed that the initial control of malignant PCs by NK cells is often attenuated in the inexorable progression of MM.5 This suggests that it is critical to infuse NK cells that are more potent, such as CAR NK cells. Of note, adoptive NK cell immunotherapy for MM has already been evaluated in several promising studies, which have fostered ongoing interest.48, 49 However, adoptive immunotherapy with primary NK cells is often complicated by the difficulties in expansion of these cells. This barrier might be overcome by the recent success of an effective NK cell expansion strategy, which utilizes 4-1BB ligand.50, 51 Another option for overcoming these limitations is to use established NK cell lines.52, 53 The large-scale expansion of NK cell lines under good manufacturing practice (GMP) conditions is easier, more cost effective, and makes these cells more readily available for use in clinical adoptive therapy.54 The NK-92 cell line, one of the cell lines that we used in the present study, seems to be the best choice due to two reasons. First, NK-92 cells have been proven generally safe with mild and transient toxicities, and expansion of these cells is feasible.55 Second, NK-92 cells have displayed appreciable in vitro and in vivo anti-cancer activity in a variety of malignancies including MM.53, 56 In fact, infusion of NK-92 cells is being tested in hematological malignancies including MM (http://www.clinicaltrials.gov).
In the treatment of cancer such as MM, a single agent may not work as effectively as combination therapy. This could imply that other therapeutic approaches designed to synergize with CS1-CAR-directed NK cell therapy may be needed to achieve the best possible MM-killing outcome. Examples of these approaches include but are not limited to: JAK inhibitors that can increase the susceptibility of MM cells to NK cell-mediated cytotoxicity;57 the anti-CS1 mAb, Elotuzumab that can enhance the antibody-dependent cellular cytotoxicity, natural cytotoxicity, and IFN-γ production of NK cells;16 the pan-KIR antibody, IPH2101, that can enhance NK cell activity through blocking the KIR inhibitory function;58 and even circulating or exogenous microRNAs that can activate NK cells through Toll-like receptor signaling.24
In conclusion, CS1 is a promising target for the use of CAR NK cells for MM treatment. We have successfully generated a CAR that specifically recognizes CS1, which is uniformly expressed on the surface of MM cells. We also present evidence that NK cells armed with this CS1-specific CAR can more efficiently and specifically recognize and eradicate myeloma cells both in vitro and in vivo. Our study may help pave the way toward the clinical application of CS1-CAR-modified immune cells, either used alone or in combination with other approaches, for the treatment of MM.
The project was supported in part by Multiple Myeloma Opportunities for Research and Education (MMORE), and grants from the National Institutes of Health (CA155521 and 1R41OD018403 to JY and CA095426 and CA068458 to MAC), a 2012 scientific research grant from the National Blood Foundation (JY), an institutional research grant (IRG-67-003-47) from the American Cancer Society (JY), and The Ohio State University Comprehensive Cancer Center Pelotonia grant (JY).
JC designed research, performed experiments, and wrote the paper; YD, SH, YP, HM and LY performed experiments; DMB, KG, XH, SMD, XZ and MAC designed the research; TH reviewed and edited the manuscript; JZ analyzed the data; CCH devised the concept, designed the research and reviewed the manuscript; JY devised the concept, designed the research, supervised the study, and wrote the paper.
About this article
Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)
British Journal of Cancer (2019)