Selective Cytotoxicity of Single and Dual Anti-CD19 and Anti-CD138 Chimeric Antigen Receptor-Natural Killer Cells against Hematologic Malignancies

Natural killer (NK) cells are part of the first line of defense that rapidly respond to malignant transformed cells. Chimeric antigen receptor- (CAR-) engineered NK cells, although are still at the preliminary stage, have been shown to be alternative to CAR-T cells, mainly due to the absence of graft-versus-host disease and safer clinical profile. Allogeneic human NK cell line NK-92 cells, equipped by CAR, are being developed for clinical applications. Herein, we designed third-generation CARs, optimized the production protocol, and generated CAR-NK-92 cells, targeting CD19 and/or CD138 antigens that employ CD28, 4-1BB, and CD3ζ signaling, with >80% CAR expression, designated as CD19-NK-92, CD138-NK-92, and dual-NK-92 cells. The generated CAR-NK-92 cells displayed high and selective cytotoxicity toward various corresponding leukemia, lymphoma, and multiple myeloma cell lines in vitro. Multitargeting approach using a mixture of CD19-NK-92 and CD138-NK-92 cells was also evaluated at various ratios to test the idea of personalized formulation to match the patients' antigen expression profile. Our data indicate that increasing the ratio of CD19-NK-92 to CD138-NK-92 could improve NK cytotoxicity in leukemia cells with a relatively higher expression of CD19 over CD138, supporting the personalized proof of concept. This information represents the basis for further in vivo studies and future progress to clinical trials.


Introduction
Chimeric antigen receptor-(CAR-) engineered T cell therapy represents a major advancement in personalized cancer therapy, providing hope to patients with relapsed and refractory diseases [1,2]. Currently, there are three CAR-T cell platforms targeting CD19 approved by the US Food and Drug Administration (FDA) for the treatment of relapsed or refractory acute lymphoblastic leukemia (ALL) in pediatric and young adults (Kymriah®), for the treatment of relapsed or refractory diffused large B-cell lymphoma (DLBCL) in adults (Kymriah® and Yescarta®), and for the treatment of relapsed or refractory mantle cell lymphoma (MCL) in adults (Tecartus™). However, one of the limitations of CAR-T cell therapy is the manufacturing of autologous CAR-T cells from patients, which is laborious, increasing the risk of production failure in clinical settings, especially from those with limited number of healthy T cells, and rendering these CAR-T cells unsuitable for patients with rapidly progressing disease. The manufacturing process could take up to 4 weeks with additional 2-week treatment-free period prior to apheresis to ensure sufficient cell numbers and viability [3].
Natural killer (NK) cells are part of the first line of defense that are rapidly activated to protect body against foreign materials and abnormal cells, including malignant transformed cells, without prior sensitization, representing an important effector cell type for cellular immunotherapy. Increasing evidence has shown that allogeneic NK cells have stronger tumor killing ability than autologous NK cells and that they do not carry the risk of inducing graft-versus-host disease (GvHD) frequently associated with allogeneic T cells [4]. However, primary NK cells have a limited life span and proliferative capacity that restrict their clinical applications. Human NK cell line NK-92 cells, which exhibit a high degree of cytotoxicity toward various cancer cells, have become the major allogeneic source of NK cells and are the only NK cell line approved by the US FDA for phase I and II clinical trials. NK-92 cells can be continuously expanded in the presence of interleukin-2 (IL-2) in a current good manufacturing practice-(cGMP-) compliant process [5].
The natural antitumor properties, ability to cultivate, and early-phase clinical safety profile of NK-92 cells make them a promising cell source for the implementation of CAR to establish off-the-shelf, universal CAR-NK cells, which have the basic framework of CAR-T cells. In comparison to CAR-T cells, a far fewer preclinical and clinical studies have been performed for CAR-NK cells, most of which aim to target and treat solid tumors. The feasibility of generation of firstand second-generation CAR-NK-92 cells targeting CD19 has been demonstrated thus far. In addition to CD19 antigen, which expresses commonly in multitude of B-cell malignancies and likely in stem cell-like subpopulation of multiple myeloma (MM) [6], CD138 (syndecan-1) has gained our attention. CD138, a member of the syndecan family involved in cell-cell and cell-matrix interaction, is a known marker of MM associated with its growth and cell proliferation [7]. The expression of CD138 has also been detected in lymphoplasmacytic lymphoma (LPL), B-cell chronic lymphocytic leukemia (B-CLL), and certain cases of acute lymphoblastic leukemia (ALL) and acute myeloblastic leukemia (AML) [8].
In the present study, we investigated the feasibility of generating third-generation CAR-NK-92 cells targeting CD19 and CD138 as well as dual CAR-NK-92 cells cotargeting CD19 and CD138 and evaluated their selective cytotoxicity in vitro toward various hematologic cancer cells with different CD19 and CD138 expression profiles. We also evaluated the cytotoxic effects of the mixture of CAR-NK-92 cells targeting CD19 and CD138 to test the idea of personalized formulation to match the patients' antigen expression profile.

Construction of CAR Vectors.
Third-generation CARs targeting CD19 and CD138 with CD3ζ signaling domain and costimulatory domains CD28 and 4-1BB, designated as anti-CD19 CAR and anti-CD138 CAR, respectively, were constructed by Creative Biolabs (Shirley, NY). Specifically, the full length of signal peptide, anti-CD19 scFv from FMC63, and anti-CD138 scFV from indatuximab with CD8 hinge, CD28, 4-1BB, and CD3ζ were synthesized and subcloned into lentiviral vector Lenti-EF1a-AT-Free. The inserts were confirmed by Sanger sequencing, and the structures of CAR vectors are schematically illustrated in Figure 1(a).

Lentiviral
Production. Lentiviral production was performed using HEK293T packaging cells in conjunction with pCMV.dR8.2 dvpr lentiviral packaging and pCMV-VSV-G envelope plasmids (Addgene #8454 and #8455) [13]. Briefly, HEK293T cells were transfected with anti-CD19 CAR or anti-CD138 CAR, pCMV.dR8.2 dvpr, and pCMV-VSV-G plasmids at the ratio of 12 : 5: 1 using Lipofectamine 3000 (Thermo Fisher Scientific) and were checked for the CAR expression by flow cytometry at 48 h posttransfection to ascertain the mammalian expression of the constructed CAR vectors. The lentiviral particles were harvested and pooled at approximately 24 and 48 h posttransfection and were concentrated using Amicon Ultra-15 centrifugal filters (Merck Millipore, Tullagreen, Ireland).

Evaluation of CAR Expression by Flow
Cytometry. CAR expression was determined by flow cytometry based on Fab fragments as well as target antigens. For anti-Fab staining, cells were incubated with FITC-(fluorescein-) conjugated anti-mouse-IgG, F(ab ′ )2 fragment antibody (F(ab) ′ 2-FITC; Jackson ImmunoResearch, West Grove, PA) for 30 min at 4°C and analyzed using a FACScalibur flow cytometer (BD Biosciences, San Jose, CA USA).
For target antigen-based detection, cells were incubated with 10 μg/mL rh CD19 (20-291) protein with His tag to the C-terminus (Abcam, Cambridge, UK) or rh CD138 (18-254) protein with His tag to the N terminus (Abcam) for 1 h at 4°C, followed by an incubation with FITCconjugated anti-His tag antibody (His tag-FITC; Abcam) for 15 min at room temperature and flow cytometric analysis. Cells that were incubated with His tag-FITC, but not with rh CD19 or rh CD138, were used as a basal control. The percentage of cells that expressed CAR could be calculated from the subtraction of FITC-positive cells in basal control from those with target proteins.           Journal of Immunology Research the percentage of cells having condensed chromatin and/or fragmented nuclei. The apoptotic index was calculated as the percentage of cells with apoptotic nuclei over the total number of PKH67-positive cells.
2.10. Statistical Analysis. The data represent means ± s:d: from three or more independent experiments. Statistical analysis was performed by two-sided Student's t-test or one-way ANOVA followed by a Bonferroni posttest at a significance level of p < 0:05.

Results
3.1. Construction of Anti-CD19 CAR and Anti-CD138 CAR and Validation for CAR Expression in HEK293T Cells. We constructed the third-generation CARs to target CD19 and CD138 antigens, which both harbor an anti-CD19 or anti-CD139 scFv fragment consisting of heavy and light chains, linked to human CD28, 4-1BB, and CD3ζ signaling domains via a CD8 hinge region with a signal peptide sequence (MALPVTALLLPLALLLHAARP) (Figure 1(a)). Notably, these CAR designs and constructs can also be used to generate anti-CD19 or anti-CD138 CAR-T cells. VSV-Gpseudotyped lentiviral vector particles were produced using HEK293T cells and used for transduction of NK-92 cells.
Due to the relatively large size of CAR vectors (~9 kb) and the limited efficiency of gene transfer into NK cells when compared with other human cells, including T cells, we first evaluated the CAR expression in anti-CD19-and anti-CD138-transfected HEK293T cells, designated as CD19-HEK293T and CD138-HEK293T cells, in comparison to their wild-type (WT) counterpart. The results show that scFv expression on the surface of CD19-HEK293T and CD138-HEK293T cells, as evaluated by flow cytometry using anti-F(ab ′ )2 antibody, was approximately 40% for both CARs ( Figure 1(b)) and that the produced lentiviral particles from these cells were able to transduce NK-92 cells into CD19-NK-92 and CD138-NK-92 cells with the transduction efficiency of approximately 10% (Figure 1(c)). Alternatively, the specific anti-CD19 and anti-CD138 CAR expression, as evaluated by their binding activities to corresponding antigens fused with His tag, yielded similar results to those of anti-F(ab′)2 staining (Figure 1(d)), thus confirming the practical CAR constructs and the reliable detection methods.

Expression of CD19 and CD138 Antigens in Various
Hematologic Cancer Cells. To test the cytotoxicity and specificity of established CD19-NK-92 and CD138-NK-92 cells, we first assessed the expression of CD19 and CD138 surface antigens in various hematologic cancer cell lines to identify representative target cells with differential CD19 and CD138 expression profiles. Figure 2(c) shows that while the majority (>98%) of human CML-derived K562 cells were negative for CD19 and CD138, human ALL-derived REH cells were shown to be positive for both antigens (>98% CD19; >70% CD138). Human BL-derived Raji cells were identified as CD19-positive cells alone (>99% CD19; <1% CD138), and human MM-derived H929 and RPMI-8226      Figure 5(c)).

FACS Enrichment of Dual CD19/CD138-NK-92 Cells and Its Selective Cytotoxicity against Various Hematologic Cancer
Cells. Certain cases of hematologic cancer cells, particularly LPL, B-CLL, ALL, and AML, might coexpress CD19 and CD138 [8]. We therefore generated dual-NK-92 cells cotargeting both CD19 and CD138 by stepwise transduction of anti-CD19 CAR into CD138-NK-92 cells and FACS enrichment based on CD19 antigen detection (Figure 6(a)) and investigated whether it would improve the killing ability against targets with both antigens. Figure 6 To ascertain the specificity of dual-NK-92 cells, we performed its cytotoxicity toward various hematologic cancer cells with different CD19 and CD138 profiles.

Discussion
In the present study, we generated three new third-  Table S1). In real-life scenario, CAR-NK-92 cells against one or two specific antigens will not suit all patients. The idea of multitargeting approach using a combination of different CAR-NK-92 cells, or CAR-NK cells from other sources, at various ratios to match the target antigen expression has been raised and preliminarily tested. With the optimized CAR design and production protocols, a library of CAR-NK-92 cells targeting a wide variety of tumor antigens could be created to be applied as  13 Journal of Immunology Research allogeneic off-the-shelf products in the future and to be mixed as personalized formulation for individual patients.
Regarding the CAR design, the costimulatory domains of CAR structure used in the clinical trials of CAR-NK therapy to date are T cell costimulatory molecules [4,14]. We herein used CD28 and 4-1BB as costimulatory domains as they were reported to enhance the proliferation, survival, and cytotoxicity of CAR-T cells [15,16]. For CAR gene transfer, lentiviral transduction was chosen because of a lower genotoxicity and insertional mutagenesis when compared to retroviral transduction [17]. The major advantage of CAR-NK cells over CAR-T cells is the capability to produce graft-versus-tumor effect without inducing GvHD in allogeneic setting even in HLA-mismatched recipients, the complication of which restricts the use of allogeneic CAR-T cells [4,18]. Additionally, CAR-NK cells are potentially safer with respect to clinical implications as they induce a low incidence of cytokine storms and neurotoxicity [14,19].
Majority of current clinical trials with CAR-NK cells use NK-92 cell line. Parental NK-92 cells were derived from NHL patient with severe chronic active EBV infection [20,21], and thus, CAR-NK-92 cells need to be irradiated before clinical application to avoid the secondary tumorigenesis and potential EB virus susceptibility. In the first, small phase I clinical trials, the repeated infusion of γ-irradiated (10 Gy) CAR-NK-92 cells targeting CD33 at doses up to 5 × 10 9 cells/patient was reported to be safe in three relapsed/refractory AML patients with no substantial adverse effects, although the clinical efficacy was not found to be significant [22], suggesting a good safety profile of these CAR-NK-92 cells similar to their parental NK-92 cells. Preclinical titration experiments demonstrated that a γ-irradiation at 10 Gy was sufficient to abrogate proliferation of CAR-NK-cells, without affecting their in vitro cytotoxicity for at least 48 h or reducing in vivo antitumor activity [23,24]. It is worthy of note that the limited lifespan of CAR-NK-92 cells in vivo after irradiation could lower the risk of on-target, off-tumor toxicity.
Antigen escape from a downregulation or even loss of target antigen is one of the resistance mechanisms to CAR-T cell therapy [25]. Due to the antigen-unrestricted killing activity of NK cells, CAR-NK-92 cells can still be activated by their natural NK activating receptors even if the targeted antigens are lost. This CAR-independent mechanism of killing could also benefit to tumors with heterogeneous expression of the CAR target antigens. However, one of the drawbacks of CAR-NK-92 cells is the lack of CD16 (FcγRIII) expression [4]; thereby, they could not eliminate tumor cells through CD16-mediated antibodydependent cell-mediated cytotoxicity (ADCC). Hence, CAR-NK-92 cells exert its tumor killing mainly through (i) granulemediated cytotoxicity via perforin and granzymes; (ii) death receptor-mediated apoptosis, involving FasL and TRAIL; and (iii) the secretion of proinflammatory cytokines and chemokines, e.g., TNF-α, IFN-Ɣ, and CCL3 [4,18,26,27].
Multitargeting approach is a potential strategy to expand the effectiveness of this novel immunotherapy, as it might lower the resistance and relapse rate of cancer. In MM, a decrease in CD138 expression has been observed during the course of clinical treatment in some patients and in clonogenic MM cells [28,29], which are thought to be responsible for MM progression. A small subset of CD138-negative and CD19-positive MM cells are believed to be stem-like cells, also known as MM progenitor cells. It was reported that up to 4% of CD19-positive MM cells were detected in bone marrow aspirates of 103 MM patients [30]. CAR-NK-92 cells targeting CD19 and CD138 may exhibit improved clinical efficacy in relapsed and/or refractory MM. Notably, the coexpression of CD19 and CD138 might be detected in certain cases of LPL, B-CLL, ALL, and AML [8], as example in human ALLderived REH cells herein, and hence, CAR-NK-92 cells targeting CD19 and CD138 may benefit those of hematologic malignancies as well.
Several multitargeting CAR-T cell therapies in clinical trials include (i) pooled CAR-T cells by a mixture of two different CAR-T cells targeting different antigens; (ii) dual CAR-T cells targeting two different antigens in a single CAR-T cells; (iii) tandem CAR-T cells, where two antigen binding domains are connected in tandem in a single CAR; and (iv) trivalent CAR-T cells targeting three different antigens in a single CAR-T cells [31]. However, there are not many studies on the multitargeting CAR-NK cells. With the feasibility for off-the-shelf manufacturing, we believe that a library of CAR-NK-92 cells or CAR-NK cells from other sources, including donor-derived peripheral blood mononuclear cells (PBMC), umbilical cord blood, and those derived from CD34-positive hematopoietic stem/progenitor cells (HSPCs) or induced pluripotent stem cells (iPSCs), could be created. For the concept of individual patients' optimized therapy, surface antigens of patient's tumors will be detected and spontaneous mixture of multiple CAR-NK cells at the right combinations will be formulated to match the antigen expression profiles. Additionally, the cytotoxicity of those CAR-NK formulations can be first evaluated in patient's primary cells in vitro in order to predict the best response. A sequential administration of different CAR-NK cells, termed cocktail immunotherapy, might be applied as in the case of CAR-T cells [32].

Conclusions
In summary, our findings reveal a promising CAR design and production protocol for single-and multiantigen-targeted CAR-NK-92 cells against various hematologic malignancies. The generated CAR-NK-92 cells targeting CD19 and/or CD138 antigens that employ CD28, 4-1BB, and CD3ζ signaling displayed high and selective cytotoxicity against established leukemia, lymphoma, and MM cells in vitro. Further studies should focus on the generation of the library of CAR-NK-92 cells targeting various novel tumor antigens, using current good manufacturing practice-(cGMP-) compliant transduction and expansion methodologies. This information represents the basis for further in vivo studies and future progress to clinical trials.

Data Availability
The data used to support the findings of this study are included within the article.