Two for one: targeting BCMA and CD19 in B-cell malignancies with off-the-shelf dual-CAR NK-92 cells

Genetic engineering of T cells with a chimeric antigen receptor (CAR) has demonstrated high clinical activity in several B-cell malignancies [1]. This has led to the recent regulatory approval of several CAR-T-cell products, either targeting CD19 or B-cell maturation antigen (BCMA) [2‐7]. Due to its ubiquitous expression on malignant B-cells, CD19 was an evident choice as a target antigen for B-cell leukemia and lymphoma. Depending on the cell product, overall response rates and complete responses range from 52 to 85% and 40 to 59%, respectively [1, 7]. These impressive results sparked the development of CARs for a myriad of other malignancies and led to the recent wave of BCMA-targeted CARs under clinical investigation for multiple myeloma (MM). BCMA is highly expressed on myeloma cells, whereas its presence on healthy cells is restricted to lower levels on mature B-cells and plasma cells [8]. Its expression profile, combined with its vital role in the proliferation, survival and drug resistance of MM cells makes BCMA an excellent target for MM.

Despite remarkable initial response rates mediated by CD19- and BCMA CAR-T cells in different hematological malignancies, a majority of treated patients ultimately relapsed [1, 9]. A considerable fraction of these relapse cases can be attributed to the selection and outgrowth of a small tumor cell population with downregulated or absent target antigen expression [10‐13]. Subsequent administration of CAR-T cells targeting other cell surface antigens, such as CD20 and CD22 in B-cell leukemia, can reinstate remission [14‐16].

Due to the developmental relationship between B-cells and plasma cells, it comes as no surprise that some overlap exists between CD19 and BCMA expression. In this regard, several groups confirmed the presence of BCMA antigen on a subset of CD19⁺ cells from B-cell leukemia and lymphoma patient samples, as recently reviewed by Dogan et al. [17]. Similarly, a small, less differentiated BCMA⁺CD19⁺ MM subpopulation was recently described, which was shown to be chemoresistant and to have cancer stem cell-like properties [18, 19]. Therefore, combinatorial approaches employing chemotherapeutics together with CD19- and BCMA-targeted CAR therapies are already under investigation in the context of MM [8, 20, 21].

Combinatorial multi-target therapies in the clinic are currently relying on simultaneous or sequential administration of the different CAR-T-cell products [8, 22‐25]. However, co-administration of two single-CAR-T-cell products could lead to one product outcompeting the other, indicating that a single effector cell carrying multiple CARs would be more desirable [22]. There are, however, several drawbacks related to the production of single-antigen targeted autologous CAR-T cells [26]. First, the quality of autologous, patient-derived T cells is generally poor due to the detrimental effects of prior treatments [26, 27]. Second, CAR-T-cell production is already associated with a substantial price tag [28], further reducing financial accessibility when adding a second production run. Third, current manufacturing processes are lengthy, making them unsuitable for patients with aggressive disease. The use of allogeneic T cells derived from healthy donors could represent a solution but this field is still in its infancy ([29‐32] and ClinicalTrials.gov identifier: NCT04142619) and their applicability is beset by the need for additional genetic modifications in order to reduce the risk of graft-versus-host disease (GvHD) [33]. Another population of effector cells that is gaining attention as an alternative for T cells is the natural killer (NK) cell. NK cells are of particular interest due to their innate anti-tumor capacity mediated through their activating receptors, their favorable cytokine profile and the lack of GvHD [33]. However, primary NK cells generally face the same issues for clinical application as T cells, namely their limited ex vivo expansion capacity and population heterogeneity, in addition to their considerable resistance to genetic modification [34]. In contrast, the allogeneic NK-92 cell line provides a continuously expanding, homogeneous and easily engineerable off-the-shelf source of NK cells that is increasingly used in the clinic. This is exemplified by the fact that the NK-92 cell line has obtained FDA investigational new drug application status and by the growing number of clinical trials using (CAR-modified) NK-92 cells [33].

Here, we present a dual-targeting strategy with NK-92 co-expressing two complete CD19- and BCMA-specific CARs. We demonstrate that simultaneous transfection of multiple CAR-encoding mRNAs is feasible and results in high dual-CAR expression. Dual-CAR NK-92 cells efficiently recognize and eliminate single- and double-positive target cells, including primary tumor cells, even at low effector to target ratios. Furthermore, we confirm that dual-CAR NK-92 cells maintain their functionality after gamma-irradiation, which supports their off-the-shelf clinical applicability.

CAR-T-cell therapy has demonstrated immense therapeutic potential in hematological malignancies, including B-ALL, B-cell lymphoma and MM, even in patients in whom all standard treatment options have been exhausted. Yet, cancer cells exploit several escape mechanisms to thwart this potent immunotherapy, eventually leading to relapse in a large fraction of patients. Two main types of relapses after CAR-T-cell therapy can be discerned, one being the result of downregulation or complete loss of the target antigen on the tumor cell surface resulting in an inability of the CAR-T cells to recognize the tumor cells (antigen-negative relapses) and the other being the result of the loss of (functional) CAR-T cells; in this case, antigen expression is retained, hence the term antigen-positive relapses. There is now an intensive search for strategies to overcome the problem of antigen-negative as well as antigen-positive relapses. One way to tackle the problem of relapse due to antigen loss is to target more than one antigen [14‐16, 44, 45], whereas the second problem could be addressed by producing and re-administering a second batch of CAR-T cells. However, the latter solution is cumbersome given the complex logistics and lengthy duration of autologous CAR-T-cell manufacturing, which is sometimes not possible due to the low numbers or low quality of patient-derived T cells. This problem could in turn be addressed by using an unlimited, off-the-shelf source of functionally active effector cells, such as the NK-92 cell line.

Here, we tackled the aforementioned root causes of relapse following CAR-T-cell therapy by developing a dual-CAR NK-92 cellular therapeutic targeted towards CD19 and BCMA. The reasons to select these two particular targets are obvious; CD19 and BCMA are highly relevant in the context of CAR-T-cell therapy for B-cell hematological malignancies, and, up until now, only CD19- and BCMA-targeted CAR-T-cell products have received regulatory approval. Furthermore, CD19 and BCMA are expressed across different stages in the development of B-cells to mature plasma cells. Hence, dual targeting of CD19 and BCMA offers the prospect of a broadly applicable therapeutic product that can be used in a spectrum of B-cell hematological malignancies ranging from B-cell leukemia and lymphoma to multiple myeloma. Some antigens are co-expressed on cells at the same developmental stage. This is, for example, the case for CD19 and CD20 on B cells; dual-CD19/CD20 CAR-T cells are now under active investigation in B-cell leukemia and lymphoma [46‐49]. Similarly, for MM, combinatorial approaches with BCMA and other plasma cell surface antigens such as CD38 are under development [45, 50‐54]. In vivo models investigating these dual CAR strategies consistently showed superior tumor clearance and prevention of antigen escape, offering the prospect for deeper and more durable clinical responses [44, 48, 53, 55].

Very recently, Luanpitpong et al. described a dual-CAR NK-92 approach similar to ours, but using a different combination of antigens (CD19 and CD138) and using lentiviral transduction as CAR loading strategy [56]. Here, the two fully human CAR constructs were introduced in the NK-92 by means of mRNA electroporation. In contrast to lentiviral transduction, mRNA electroporation is a rapid, simple, relatively low-cost and highly efficient method for gene transfer in human cells. Boissel et al. previously applied the mRNA electroporation technology for introduction of a single CD19 CAR in NK-92 cells, reaching transfection efficiencies of approximately 50% [57, 58]. CD19 CAR mRNA electrotransfection of primary human NK cells yielded comparable results [59, 60]. In this study, we confirmed that mRNA electroporation is a suitable method for CAR loading of NK-92 cells, with either the CD19-CAR and BCMA-CAR being expressed at high levels. In addition, for the first time, we demonstrated that this technology can be used in the NK-92 therapeutic cell source to simultaneously introduce two different CAR constructs without hampering the expression of either CAR molecule. Despite its obvious advantages, such as potentially reducing the duration of severe side effects in treated patients, the temporary CAR expression following mRNA electroporation could imply a need for repeated administration of the therapeutic cell product [61]. However, in the NK-92 model, permanent gene expression is not an inherent requirement, since NK-92 cells have a relatively short lifespan after administration [27]. In this regard, it is of critical importance to carefully consider the origin of the extracellular antigen-recognition domains. As repeated administration of CAR products containing murine-derived components can cause immunization and anaphylaxis, severely limiting safety and therapeutic efficacy [62], we have opted for the use of two fully-human CARs.

For clinical application, proliferation of NK-92 cells needs to be halted prior to infusion to avoid NK-92 cell engraftment in vivo, which could lead to the development of NK cell lymphoma. Although alternative methods are being examined [63], this “inactivation” step is most commonly performed by gamma-irradiation. We confirmed that 10 Gy gamma-irradiation effectively blocks NK-92 cell proliferation and leads to a gradual decrease in cell viability down to zero over the course of 1 week, paralleling the CAR expression kinetics after mRNA electroporation. Importantly, gamma-irradiation did not affect the functionality of the (dual) CAR mRNA-electroporated NK-92 cells, underlining the potential clinical applicability of the proposed therapeutic cell-based product.

Another potential advantage of using NK cells over the gold-standard CAR-T cells is the natural anti-tumoral activity of NK cells. However, this intrinsic cytotoxic capacity is largely dependent on exogenous activation stimuli, such as IL-2 [64]. IL-2 is an essential cytokine for NK cell growth and is, therefore, indispensable during NK-92 cell culture. IL-2 administration in humans can cause severe toxicity and can lead to regulatory T-cell activation, which is an undesired effect in the context of cancer immunotherapy due to their counterproductive inhibitory effect on cytotoxic lymphocytes [64]. Therefore, to pave the way towards clinical application, we omitted the supplementation of IL-2 during the last 24 h of NK-92 culture. As expected, this led to an almost complete abrogation of the natural cytotoxicity of NK-92 towards the NK-sensitive tumor cell line K562 [64‐66]. In addition, as exemplified here both in the Daudi lymphoma cell line model as well as in the primary patient samples, some B-cell hematological malignancies are largely resistant to NK cell lysis [67]. Here, we show that CAR engineering of NK-92 cells overcomes the IL-2 dependence and restores their anti-tumor cytolytic activity. Corroborating the results of other dual-targeted CAR products [56, 68, 69], dual-CAR NK-92 were at least equally effective as their single-CAR counterparts in eliminating single and dual antigen expressing target cells, effectively reducing the probability of antigen escape. Interestingly, dual-CAR NK-92 display higher cytotoxicity towards CD19⁻BCMA⁺ cells compared to BCMA-CAR NK-92. The reason for this discrepancy, remains to be elucidated but the increased frequency of BCMA CAR⁺ cells in the dual-CAR NK-92 population compared to the BCMA-CAR NK-92 provides a likely explanation for the heightened lytic activity against BCMA⁺ target cells in the dual-CAR NK-92 conditions. Moreover, it was recently reported that dual CD19- and BCMA-CAR-T cells were able to completely ablate regulatory B-cells from the bone marrow of MM patients, contributing to a favorable environment for clearance of myeloma cells in the bone marrow [25]. Hence, in addition to their direct anti-tumor activity, our dual-CAR NK-92 cells could also play an important role in reshaping the tumor microenvironment.

The use of dual-CAR NK-92 cells as presented in this study contains some limitations. First, supplementating the NK-92 culture medium with animal serum instead of human serum limits the clinical translational potential. To the best of our knowledge, there have been no comparative studies between the two, leaving uncertainty on whether the serum source affects CAR NK-92 performance. Of interest, one study on serum-free NK-92 culture reports no significant difference in viability, proliferation, receptor expression levels, or perforin and granzyme levels, but a significantly decreased degranulation and cytotoxic potential in vitro which could be partly recovered after the addition of serum [70]. Second, our follow-up period of NK-92 cell viability and proliferation after irradiation was limited to 7 days. However, others have reported on the complete abrogation of NK-92 expansion for more than 30 days using the same irradiation protocol as described here [42, 71]. In our view, this provides sufficient proof for the safe clinical application of dual-CAR NK-92. Finally, we did not directly investigate potential “off-tumor” effects in our work. As discussed above, so far, NK-92 clinical studies have not revealed any toxicity towards normal cells or tissues. Moreover, given that “on-target/off-tumor” toxicities of current CD19- and BCMA-CAR-T products, such as hypogammaglobulinemia following B-cell depletion, are well described and manageable [1, 9], it seems unlikely that combinatorial targeting of these antigens will result in any unanticipated off-tumor side effects. Moreover, in the event such toxicities occur with our dual-CAR NK-92 approach, repeated administration of the cells can be terminated, benefiting from their limited persistence after irradiation and the transient nature of the CAR-encoding mRNA.

In conclusion, we demonstrate the efficient generation of dual CD19- and BCMA-CAR expressing NK-92 cells. These dual-CAR NK-92 cells have potent and specific cytotoxic activity towards CD19⁺ and BCMA⁺ tumor cell lines, as well as primary B-ALL and MM cells. Gamma-irradiation was confirmed to be adequate to avoid proliferation and persistence of the NK-92 cells while maintaining functionality. Future work should be conducted to validate our findings in vivo before proceeding to early phase clinical testing to evaluate safety and feasibility. However, in the long term, we envision that our CD19/BCMA-targeted dual-CAR NK-92 can be used as an off-the-shelf therapeutic for B-cell leukemia, lymphoma and myeloma patients to address the current problems of antigen-negative and antigen-positive relapses after autologous single CAR-T administration.