Facts and Hopes of Chimeric Antigen Receptor-Redirected NK T Cells.

Introduction
Chimeric antigen receptor (CAR)-engineered T cells have achieved remarkable success as a breakthrough cancer immunotherapy for patients with hematologic malignancies. However, progress in solid tumors remains limited due to key challenges including a scarcity of targetable tumor-associated antigens, physical barriers to tumor infiltration, and the immunosuppressive tumor microenvironment (TME) (1–3). Additionally, the prohibitive cost of manufacturing autologous CAR-T cells (CAR-Ts) has hindered their broader clinical application. In this context, innate immune cells such as natural killer cells (NKs) (4,5), γδT cells (6,7), and invariant natural killer T cells (NKTs) (8,9) are emerging as promising alternative platforms. These cells possess intrinsic antitumor properties and offer “off-the-shelf” potential due to their limited alloreactivity.
NKTs are evolutionarily conserved innate lymphocytes that significantly differ from conventional CD4 and CD8 T cells (Table 1). NKTs are a small fraction of the circulating immune cells (on average less than 0.1% of peripheral blood mononuclear cells), and are characterized by the unique expression an invariant T cell receptor alpha chain (iTCR) (10). In contrast to conventional αβTCRs expressed by CD4 and CD8 T cells, iTCR is not restricted by the Major Histocompatibility Complex (MHC), but recognizes α-galactosylceramide (αGalCer) and other glycolipids presented by the monomorphic CD1d molecule (11). They play a critical role in mediating protective immune responses against several pathogens (12). Further, an increased frequency of NKTs in the peripheral blood and at tumor sites has been shown to correlate with better clinical outcomes in several human malignancies, suggesting that NKTs participate in the antitumor immune response (13–15). In models of neuroblastoma (NB), NKTs migrate to tumors in response to certain chemokines such as CCL2 and CCL20 (14,16). Tumor-infiltrating NKTs also co-localize with CD1d-expressing tumor-associated macrophages (17), which they eliminate or polarize to the M1-like phenotype thereby remodeling the TME to be less immunosuppressive (17–19). Similarly, NKTs have been shown to block myeloid-derived suppressor cells (20–23), and donor-derived NKTs have also been reported to mitigate the occurrence of graft versus host disease (GvHD) in the allogeneic stem cell transplant setting (24–26). These intrinsic properties have motivated our groups and others to exploit this unique cell subset for the development of CAR-NKT cell products that target both specific tumor antigens and elements of the TME and can also be used as “off-the-shelf” cell therapy products. This article provides a comprehensive overview of the latest pre-clinical and clinical advances in CAR-NKT engineering for the treatment of cancer and future directions in the field.

Human CAR-NKTs from pre-clinical models to phase I clinical studies
CAR-redirected human NKTs were first studied in the context of NB models using a CAR that targets the ganglioside GD2 (GD2.CAR) (27). This study provided proof-of-concept that human NKTs can be transduced, expanded to clinical scale, and redirected to eliminate GD2+ NB cells while maintaining the cytotoxic function of the native iTCR (Figure 1). Additionally, GD2.CAR-NKTs were found to traffic more effectively to tumor sites than GD2.CAR-Ts in a xenogeneic NB mouse model (27). Similarly, human NKTs expressing a CD19.CAR mediated cytotoxicity against both CD19+ lymphoma cell lines and CD1d+ lymphoma cell lines (28,29). Through dual targeting of tumor antigen and CD1d, CD19.CAR-NKTs have also been shown to eliminate CD19+CD1d+ primary mantle cell lymphoma and marginal zone lymphoma cells more effectively than CD19.CAR T cells (29). More recently, human NKTs have been engineered to express CARs that target multiple myeloma antigens such as BCMA and CD38. Redirected human NKTs demonstrated anti-tumor activity when cultured with bone marrow mononuclear cells isolated from multiple myeloma patients, with one patient sample also showing activity against CD1d+ malignant cells (30). CAR-NKTs have also been developed to target clonal TCRVβ chains expressed by T cell malignancies. Engineered CAR-NKTs effectively lysed T cell lymphoma and leukemia cells in peripheral blood mononuclear cells from two patients while sparing the remaining normal T cells (31). Beyond CAR-NKTs generated from peripheral blood cells, there is growing interest in genetically manipulating hematopoietic stem cells to differentiate into CAR-NKTs. Specifically, CD34+ cells isolated from cord blood units have been engineered to express the iTCR and a CAR of interest followed by an extended culturing process that leads to NKT differentiation (32). This system has been used to generate “off-the-shelf” NKTs expressing CARs specific for BCMA, CD19, CD33, GD2, and GPC3, all of which have demonstrated anti-tumor activity in vitro and in tumor-bearing immunodeficient mouse models (32).

Role of signaling endodomains in CAR-NKTs.
The incorporation of co-stimulatory endodomains such as CD28 and 4–1BB into CAR constructs has been crucial to the clinical success of CAR-Ts in patients with hematologic malignancies (33). The role of these endodomains in shaping the activity of CAR-NKTs has been studied in multiple contexts. For example, GD2.CAR-NKTs encoding CD28 or 4–1BB (second-generation) or both CD28 and 4–1BB in tandem (third-generation) produced high levels of IFNγ (27). NKTs expressing both constructs demonstrated potent anti-tumor activity in an in vivo NB xenotransplant model, and the third-generation construct mediated durable tumor control after repeat dosing. In the context of multiple myeloma, NKTs expressing a CD38.CAR with the same second- and third-generation co-stimulatory endodomain arrangements showed a similar in vitro Th1-biased cytokine profile (30). In lymphoma models, NKTs expressing a third-generation CD19.CAR with CD28 and OX40 proliferated better than NKTs with a CD28-only second-generation construct after in vitro stimulation (29). Overall, these data support the inclusion of co-stimulatory endodomains in CAR constructs to promote a prominent Th1 phenotype in human CAR-NKTs.

Role of cytokine co-expression in CAR-NKTs.
Early in the course of pre-clinical studies, CAR-Ts and CAR-NKs cells co-expressing cytokines like IL-15 were shown to be superior to cells without cytokine co-expression in terms of in vivo persistence and antitumor activity (34–36) (Figure 1). This has also been demonstrated clinically in multiple trials; for example, IL-15 co-expressing CAR-NKs derived from cord blood units showed clinical activity in a phase I/II clinical study in patients with relapsed B cell malignancies (4), and GPC3.CAR-Ts with IL-15 co-expression mediated responses in children with solid tumors while GPC3.CAR-Ts alone did not (37). In human NKTs, IL-15 co-expression induces more robust proliferation and promotes a Th1-like cytokine profile (38) while also protecting against hypoxia in the TME of xenogeneic NB mice (16). GD2.CAR-NKTs with CD28 co-stimulation and IL-15 co-expression produced higher levels of IFNγ, persisted better after chronic in vitro stimulation, and demonstrated superior persistence and anti-tumor activity in a xenogeneic NB model compared to NKTs expressing the CAR alone (39). Beyond IL-15, other cytokines have been shown to enhance CAR-NKT functionality through transgenic co-expression or incorporation into culture conditions during in vitro CAR-NKT expansion. For example, Ngai et al showed that supplementing CAR-NKT growth medium with IL-21 boosts NKT in vitro cytotoxicity and in vivo anti-tumor activity (40). Liu et al showed that CAR-NKTs co-expressing IL-21 persisted better than CAR-NKTs without the cytokine in a xenogeneic renal tumor model, though improvement in anti-tumor activity in this model was modest (41). O’Neal et al demonstrated that co-administering recombinant human IL-7 with an extended half-life (rhIL-7-hyFc) with CAR-NKTs increased their persistence and anti-tumor activity (42). Finally, co-expression of the pro-inflammatory cytokine IL-12 has been shown to polarize human NKTs to polyfunctional Th1 cells characterized by long-term persistence while decreasing expression of exhaustion markers (43). Co-expression of secreted or membrane-bound IL-12 in CAR-NKTs enhanced in vivo antitumor activity in xenogeneic tumor models beyond levels observed in cells co-expressing IL-15 (43). In our most recent study, we found that IL-18 co-expression in CAR-NKTs broadly reprograms NKT cell metabolism and boosts CAR-NKT antitumor activity in vivo without toxicity.

Role of transcription factors in CAR-NKTs.
In human CAR-Ts, the impact of modulating transcription factors associated with T cell differentiation and exhaustion has been well-studied (44). In murine NKTs, transcription factors such as PLZF, Gata3, RORγ-t, and T-bet have been shown to be essential for growth and differentiation (45). The impact of manipulating specific transcription factors in CAR-NKTs to modulate persistence, exhaustion, and metabolic fitness remains an area of active interest. We showed that CD62L expression in human NKTs characterizes a central-memory-like phenotype associated with superior persistence and anti-tumor activity in vivo (28) (Figure 1). We then identified the lymphoid enhancer-binding factor 1 (LEF1) as a transcriptional driver of this central-memory phenotype in NKTs and showed that LEF1 co-expression in CAR-NKTs enhances in vitro proliferation and in vivo anti-tumor activity (46) (Figure 1). Of note, we also observed that IL-12 plays a critical role in influencing the NKT phenotype by increasing CD62L expression through upregulation of the transcription factor FOXO1 and by generating Th1-polarized cells with memory properties (43). To identify additional regulators of NKT functional fitness, we recently developed a limited CRISPR/Cas9 mutagenesis screen through which we identified PRDM1 as a negative regulator of CAR-NKT memory differentiation and effector function (47). Further, using patient data from the GINAKIT2 trial evaluating GD2.CAR-NKTs co-expressing IL-15 in relapsed/refractory NB, we discovered that the BTG anti-proliferation factor 1 (BTG1) drives hyporesponsiveness in exhausted NKTs and T cells and that knocking down BTG1 boosts GD2.CAR-NKT cell anti-tumor activity (9). Thus, there is a growing body of evidence showing that CAR-NKT antitumor activity and persistence can be enhanced by modulating the expression of transcriptional master regulators.

CAR-NKTs and early-stage clinical trials.
Adoptive transfers of ex vivo expanded either autologous or allogeneic NKTs without genetic manipulation have been explored in multiple early phase clinical studies in a variety of human malignancies (Table 2)(48–58). These studies demonstrated safety of cellular products either alone or in combination with other treatment modalities. However, objective clinical responses are limited. Based on compelling preclinical data with NKTs engineered to express a CAR, we and others initiated clinical trials to evaluate the use of CAR-NKTs in human subjects in both autologous and allogeneic “off-the-shelf” settings (Table 2)(8,9,59,60). The development of clinical grade CAR-NKT products as well as the design of the Phase I clinical studies followed regulatory approval procedure previously developed for CAR-T cell products. GINAKIT2 is a dose-escalation phase I clinical study of autologous GD2.CAR-NKTs co-expressing IL-15 on which 12 pediatric patients with relapsed/refractory NB have been treated. Patients undergo lymphodepletion with cyclophosphamide and fludarabine. We have shown that GD2.CAR-NKTs are well tolerated with adverse events limited to mostly grade 1 and one case of grade 2 cytokine release syndrome. Interim results have yielded a 25% objective response rate (3/12) including two partial responses and one complete response, with two additional patients showing evidence of anti-tumor activity (9). The clinical study continues to enroll patients, and we recently reported the occurrence of a lethal adverse event in a patient who received 3×108 CAR-NKT/m2 (61). Lethal toxicity was associated with severe hyperleukocytosis and multi-organ deterioration attributed to leukostasis and extensive vascular occlusion confirmed by post-mortem autopsy. The observed hyperleukocytosis was not associated with insertional mutagenesis or any known genetic predisposition and was likely of multifactorial origin (61). Despite the small patient sample size, GINAKIT2 correlative studies identified a significant correlation between CAR-NKT expansion/persistence in vivo and clinical response. The study has also confirmed the critical role of NKT central-memory differentiation associated with stemness in predicting CAR-NKT expansion/persistence. Specifically, CAR-NKT expansion/persistence correlated with the percentage of CD62L expressing cells in the infused product (9). In the allogeneic “off-the-shelf” setting, we initiated the phase I ANCHOR trial (NCT03774654) in which allogeneic CD19.CAR-NKTs are administered to patients with relapsed/refractory B cell malignancies. We generated five lots of allogeneic CD19.CAR-NKTs co-expressing IL-15 and a small hairpin (sh)RNA targeting β2-microglobulin and CD74 to downregulate the MHC. To date, nine patients have been infused on three dose levels (1×107, 3×107, 1×108 CAR-NKTs/m2) without evidence of GvHD, and three patients have achieved complete responses (60). Additionally, there are currently several ongoing trials testing CD70.CAR-NKTs in patients with multiple types of solid tumors (NCT06394622, NCT06182735, NCT06728189), but safety and efficacy data from these studies have not yet been reported.

Modeling CAR-NKTs in the tumor microenvironment
It is generally accepted that the intrinsic immunosuppressive characteristics of the TME constitute a significant obstacle to exploiting the full potential of tumor-specific adoptive cell therapies. However, the complexity of the TME is poorly recapitulated in commonly used immunodeficient mouse models, which provide an accurate context for evaluating the effector function and persistence of engineered human immune cells but do not support interrogation of how adoptively transferred cells interact with an intact immune system or elements of the TME. A more accurate approach to studying the potential interactions of CAR-engineered immune cells with the TME is represented by the development of syngeneic tumor models. A study utilizing an allogeneic murine B cell lymphoma model, for example, demonstrated that murine CD19.CAR-NKTs showed superior anti-tumor activity compared to CAR-Ts through cross-priming of host CD8+ T cells (62). In another study, murine NKTs engineered to express a tumor-specific αβTCR outperformed αβTCR-T cells, demonstrating robust control of multiple solid tumor types due to better tumor infiltration, cross-priming of host T cells, and CD1d-dependent modulation of the TME (63). Similarly, we recently reported that CAR-NKTs are superior to CAR-Ts in syngeneic models of melanoma, colon, and ovarian cancer (64). We showed that this can be explained mechanistically as CAR-NKTs directly target tumor cells and CD1d+ pro-tumoral macrophages while sparing M1-like macrophages and that they transactivate endogenous T cell responses via enhanced epitope spreading mediated by CD1d+ dendritic cell activation (64) (Figure 1). Another study assessed the antitumor effects of CAR-Ts, CAR-NKTs, CAR-NKs, and macrophages in a syngeneic glioma mouse model (65). CAR-NKTs and CAR-Ts showed similar antitumor effects, partly attributed to the “immune-cold” characteristics of this glioma model. Interestingly, the combination of CAR-Ts and CAR-NKTs enhanced tumor control, but mechanistic studies are needed to better dissect the crosstalk between these cells in the context of “immune-cold” tumors. Beyond syngeneic murine tumor models, results from a canine animal model used to evaluate unedited allogeneic NKTs showed that enrichment of central-memory signature and telomerase-related gene expression in NKT products infused into animals represent favorable biomarkers correlating with superior persistence in MHC-mismatched recipients (66). This canine platform offers a valuable alternative to murine syngeneic models for evaluating engineered NKTs in a translational context. These studies underscore the importance of using immunocompetent animal models for preclinical evaluation of ex vivo-expanded and engineered NKTs. Further, the indirect antitumor effects mediated by CAR-NKTs such as modulation of the TME and activation of endogenous immune responses highlight their potential for clinical applications in solid tumors.

Combination of CAR-NKTs with other agents
While adoptive transfer of ex vivo-engineered CAR-Ts, CAR-NKTs, and CAR-NKs has generated clinical activity in patients, the limited durability of these responses is of concern and combination therapies may be needed to improve their length.

Combination of CAR-NKTs with immune checkpoint blockade.
Chronic antigen stimulation causes CAR-T cell dysfunction and exhaustion, leading to therapy failure in patients (67–69). To counteract this effect, CAR-Ts are being combined with immune checkpoint inhibitors in both pre-clinical models and early-phase trials (70–73). We observed that GD2.CAR-NKTs isolated from GINAKIT2 patients express PD1 and TIM3, suggesting that adoptively transferred CAR-NKTs may undergo exhaustion (9). Our syngeneic tumor models recapitulated this observation showing that CAR-NKTs upregulate PD1 in the TME, especially in the context of high tumor burden (64). In this model, combination with PD-1 blockade enhanced the antitumor effects of CAR-NKTs (64). Targeting PD-L1 has also been proven to promote Th1 cytokine release, boost cytotoxicity, and enhance antitumor immunity via activation of NKTs (74). These studies provide a strong rationale for combining CAR-NKTs with immune checkpoint inhibitors. However, additional studies are needed to evaluate the potential utility of other available checkpoint inhibitors that block LAG3, TIM3, and TIGIT to determine the optimal agent to synergize with CAR-NKTs in patients.

Combination with CD1d “regulators.”
The level of CD1d expression in target cells affects how NKTs recognize CD1d-expressing cells and mediate cytotoxicity (64,75). CD1d is expressed by monocytes, macrophages, dendritic cells, B lymphocytes, and thymocytes (76,77). Epithelial cells, adipocytes, and vascular smooth muscle cells have also been reported to express CD1d (78). Some human malignancies such as multiple myeloma, glioblastoma, medulloblastoma and lymphoma (79–82) are reported to express CD1d, and CD1d expression in tumor cells can be epigenetically modulated by histone deacetylase inhibitors or retinoic acid (29,83). Based on this evidence, pharmacologic agents that induce CD1d upregulation in tumor cells may promote dual-targeting of tumor cells by CAR-NKTs, reducing the risk of tumor escape due to antigen loss. For example, the RARα ligand all-trans retinoic acid has been reported to induce CD1d upregulation in chronic lymphocytic B cell leukemia (29,84). This highlights the potential of using CD1d modulators to enhance the effectiveness of CAR-NKT-based immunotherapies.

Combination with NKT cell agonists.
We showed that in vivo expansion and persistence of adoptively transferred CAR-NKTs correlate with anti-tumor activity in NB patients (9). Therefore, strategies to boost CAR-NKT in vivo persistence using NKT-specific antigens could potentially improve anti-tumor activity. Administration of antigen presenting cells pulsed with the NKT ligand αGalCer in patients with non-small cell lung cancer was shown to be safe and increased the number of circulating NKTs (85). Similarly, irradiated tumor cells or microparticles used to deliver αGalCer increased circulating NKT numbers (86,87). However, αGalCer monotherapy and administration of dendritic cells pulsed with αGalCer have generated limited anti-tumor activity in patients (51,88,89). Combining αGalCer-loaded microparticles with CAR-NKTs or αβTCR-engineered NKTs was shown to result in superior antitumor activity compared to engineered NKTs alone (20,64). Synthetic NKT ligands including ABX196 and 7DW8–5 have also been shown to safely promote NKT activation and expansion and are currently being investigated in phase I clinical studies (90,91). This evidence underscores the potential benefits of combining CAR-NKTs with NKT agonists to enhance persistence and therapeutic efficacy.

Combination with other agents.
In addition to pharmacologic agents, other immunomodulatory agents with broader immune effects have been shown to directly or indirectly improve NKT cell functions. The immunomodulatory drug lenalidomide increases NKT expansion in combination with αGalCer in both healthy donors and patients with multiple myeloma (92,93). Armed oncolytic viruses that release IL-21 have been shown to synergize with NKTs in a humanized NSG mouse model of small cell lung cancer (94). Additionally, oncolytic viruses targeting PGE2 have been shown to reduce immunosuppression in the TME, in turn improving the trafficking of NKTs to tumor sites (95). Collectively, this evidence indicates that CAR-NKTs can be combined with multiple therapeutic modalities to safely enhance their therapeutic effects.

Conclusions and future directions
NKTs have unique properties that can be harnessed for cellular therapy applications, giving them an edge over conventional T cells. In particular, their ability to target CD1d+ myeloid-derived suppressor cells and to transactivate CD8+ T cell responses situate NKTs as promising candidates for solid tumor cellular immunotherapy, given that modulation of the TME and epitope spreading are central to the elimination of solid tumors. CAR-NKTs have produced promising clinical results in ongoing phase 1 trials, but more work is needed to improve their persistence in patients and the durability of responses. While IL-15 co-expression was confirmed to be important for CAR-NKT expansion and persistence in patients, other cytokines have shown even greater potential in pre-clinical studies and could be tested in the clinical setting. Therapeutic combinations with other agents that can counter CAR-NKT exhaustion and boost expansion could be rapidly tested in the clinic with existing CAR-NKTs co-expressing IL-15. Finally, since NKTs naturally do not cause alloreactivity, they represent an appealing platform for allogeneic “off-the-shelf” products, and early clinical data seem to support their safety profile. Further, the unique possibility of stimulating NKTs in vivo using NKT agonists could be used to enhance persistence in both autologous and allogeneic settings. In conclusion, NKTs are a versatile cellular immunotherapy platform with innate antitumor properties and minimal alloreactivity.