Synthetic Biology in the Engineering of CAR-T and CAR-NK Cell Therapies: Facts and Hopes

Surgery, radiotherapy, and chemotherapy have long served as the foundation of cancer therapy, but a number of malignancies have remained resistant to these three pillars of cancer treatment. Over the past decades, immunotherapy has emerged as a fourth pillar in the arsenal against cancer. By harnessing the patient's own immune system, immunotherapies such as immune checkpoint blockade (1, 2), cancer vaccines (3), and adoptive transfer of immune cells engineered to target tumor antigens, have emerged as promising treatments for malignancies that are refractory to traditional therapies. Among the different immunotherapy modalities, cell-based immunotherapy has shown particular promise against hematologic malignancies, but its application to the treatment of solid tumors remains work in progress (4). Multiple immune cell types, including T cells, natural killer (NK) cells, invariant NK T cells, macrophages, and neutrophils have been explored as potential chassis for cell-based immunotherapy, with T cells and NK cells as the most extensively evaluated effector cell types to date.

In contrast to small-molecule drugs, protein biologics, or radiation, cell-based immunotherapies are living drugs with the ability to persist, amplify, and traverse within the patient. These unique characteristics allow cell-based therapies to mount dynamic and complex immune responses against the tumor unattainable by other therapeutic modalities. The ability to specifically program the biological properties of therapeutic cells further expands the capabilities of engineered immune cells as cancer therapeutics. Synthetic biology is a growing discipline that generates biological systems with novel behaviors and functions, by assembling circuitries comprising synthetic biological components and/or naturally occurring biological parts repurposed for new applications. By focusing on the design, construction, and assembly of modular biological components, synthetic biology enables researchers to build biological circuits with programmable input processing and output parameters. In the context of cell therapies, integration of these circuits into immune cells enables development of products equipped with novel therapeutic functions to combat previously “undruggable” or untreatable diseases. In addition, synthetic biology can accelerate the acquisition of well-controlled biological datasets through the use of clear design rules, thus enabling better understanding of complex biological phenomena and facilitating rational biological engineering principles for translational applications.

Advancements in synthetic biology have been supported by new DNA synthesis and sequencing technologies that enable accurate and high-throughput design, assembly, and testing of biological circuitry (5, 6). Concurrently, gene-editing technologies such as with zinc finger nucleases, transcription activator-like effector nucleases, and clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 have expanded the synthetic-biology toolkit, enabling high-throughput gene screening, targeted gene ablation and transgene insertion, and development of more complicated biological models (7). While the first generation of FDA-approved cell-based cancer immunotherapies largely relies on the expression of a single transgene encoding a tumor-targeting receptor, the incorporation of biological circuits may further expand the safety and efficacy profile of cell therapies to address perpetually intractable diseases.

In this review, we assess the current state of synthetic biology applications in the field of cell-based immunotherapies, focusing specifically on advancements aimed to direct and repurpose T and NK cells against cancer.

NK and T cells are both lymphocytes with cytotoxic capabilities, including perforin and granzyme-mediated cytotoxicity and proinflammatory cytokine release, making them attractive candidates for cell-based antitumor therapeutics (8, 9). However, the two effector cell types also possess differential characteristics that endow each with distinct advantages (Table 1). T cells, the prototypical effectors of adaptive immunity, derive their antigen specificity from genetic recombination of the T-cell receptor, enabling highly specific recognition of foreign peptides presented on host MHC molecules to induce clonal T-cell expansion and potent cytotoxicity. Importantly, T cells can establish long-term memory and persistence, thus providing the possibility of long-term surveillance against disease reemergence (10). On the other hand, NK cells are members of the innate immune system that have a relatively short lifespan and more limited proliferative potential, and they do not naturally express receptors with a broad repertoire of antigen specificity (11, 12). Instead, NK cells are regulated by a collection of germline-encoded receptors that can either activate or suppress cytotoxicity upon ligand binding, thus enabling the elimination of virally infected, transformed, or antibody-labeled cells. Importantly, NK cells express many of the signaling molecules downstream of T-cell receptor signaling (13, 14), thus enabling the adaptation of receptors designed for T-cell therapy to the NK-cell context.

To date, T-cell therapy development has led the way in oncology applications, culminating in the FDA approval of six autologous chimeric antigen receptor (CAR)-T cell products for B-cell hematologic malignancies and hundreds of active clinical trials against various malignancies. However, challenges accompanying autologous cell therapy, such as the risk of manufacturing failure and disease progression by patients awaiting cell manufacturing (15), have motivated research into allogeneic alternatives such as NK cells (Table 2). Unlike T cells, NK cells do not recognize MHC-presented antigens, thus avoiding the potential for GVHD that can be triggered by allogeneic adoptive T-cell therapy. This unique property positions NK-cell therapies for “off-the-self” use without the need to knock out endogenous receptors. Furthermore, unlike T cells, allogeneic NK cells can be generated from scalable sources such as cord blood, immortalized cell lines, and induced pluripotent stem cells (iPSC), which can dramatically decrease the cost of adoptive cell therapy (16–18). Finally, clinical evidence to date suggests NK-cell therapies can achieve antitumor efficacy without triggering severe cytokine release syndrome, a serious and commonly observed side effect of T-cell–based therapies (17).

Early clinical trials evaluating the adoptive transfer of NK cells relied on their innate ability to identify transformed cells for antitumor efficacy, leading to modest outcomes (19, 20). However, advances in synthetic biology and the remarkable efficacy of CAR-T cells have provided a template for engineering more targeted and potent NK-cell therapies. Early development of synthetic activating receptors in NK cells utilized prototypical CARs developed for T cells and validated the ability to redirect NK-cell cytotoxicity in an antigen-specific manner (21). Similarly, the earliest CAR-NK cell clinical trials utilized prototypical, CD19-targeting CARs (NCT00995137, NCT01974479). While the recent successes with CAR-T cells have reinvigorated the field of cellular therapy, more sophisticated strategies are needed to broaden its therapeutic outlook and synthetic biology is expected to be key in facilitating the engineering of these next-generation therapies.

CARs are synthetic receptors that enable immune cells to recognize and initiate antigen-specific cytotoxic responses (22). The adoptive transfer of CAR-T cells has demonstrated remarkable clinical efficacy in treating various B-cell hematologic malignancies, becoming the first genetically modified cell therapy to receive FDA approval in 2017 (23). CARs are transmembrane proteins comprising four major components: an extracellular antigen-binding domain, extracellular spacer domain, transmembrane domain, and intracellular signaling domain (Fig. 1A). This architecture allows for transduction of an extracellular antigen-binding event into an intracellular signaling cascade, resulting in downstream cell activation and subsequent target-cell killing. First-generation CARs, which only utilize CD3ζ as an intracellular signaling domain, failed to elicit potent antitumor activity in clinical trials (24, 25). In response, one or two intracellular costimulatory domains were incorporated in second- and third-generation CARs, respectively, to enhance cytokine production, proliferation, and in vivo persistence (26–28). Although the early development of CAR-NK cells directly adopted CARs developed for T cells, the modular design of CARs has also allowed development of novel CARs containing alternative transmembrane and signaling domains that may be better suited for NK-cell function. For example, “NK-CARs” containing an NKG2D transmembrane domain and a 2B4 costimulatory domain, both NK cell–related proteins, have shown promising antitumor function in preclinical settings. In comparison with T cells expressing prototypical “T-CARs” comprising CD28 transmembrane and CD28 plus 4-1BB costimulatory domains, iPSC-derived NK cells expressing NK-CARs achieved stronger efficacy and reduced off-tumor in a mouse xenograft model of ovarian cancer (29). These results suggest that tailoring of CAR components for optimal signaling in NK cells may further increase the efficacy of CAR-NK cell therapies for solid tumors.

Unlike antibodies and small-molecule drugs, cell-based therapies use living cells that can proliferate, persist, and circulate within the body, thereby mounting dynamic and complex immune responses against target cells. However, the dynamism and potency of cell-based therapies also pose unique safety challenges in the clinic, with the potential for severe toxicities such as cytokine release syndrome and immune effector cell–associated neurotoxicity syndrome (30). A potential strategy to reduce unintended toxicity is to implement regulatory devices that can control either the expression or the function of CAR proteins, and consequently regulate the activity of CAR-expressing cells.

The regulation of gene expression or protein activity using small-molecule drugs has a long history in mammalian synthetic biology (31, 32), and drug-regulatable platforms have been used to enable on-demand cessation of CAR signaling activity without permanently ablating the engineered cell population (Fig. 1B). For example, ON switches can be engineered by splitting the CAR protein into two nonfunctional domains such that CARs are in the inactive OFF state by default, and expression can be turned on through the administration of a dimerizing drug (33–35). Alternatively, OFF switches can be engineered such that CARs fused to degradation domains are in a default ON state, but administration of small-molecule drugs can induce CAR proteasomal degradation to turn expression off (36, 37). Although drug-regulatable expression systems provide increased flexibility and control, common challenges for these switch designs include: “leaky” (i.e., high baseline) activity in the OFF state, reduced CAR expression and potency compared with constitutive systems, and poor dynamic range between the ON and OFF states. Various engineering strategies are under active evaluation to address these obstacles. A recently reported drug-regulatable platform termed signal neutralization by an inhibitable protease, or synthetic intramembrane proteolysis (SNIP), has been shown to have a tight OFF state, improved dynamic range and improved potency compared with constitutively active CARs in multiple hematologic and solid tumor models (38). Carefully designed clinical trials will be required to evaluate the tunability of drug-regulated CAR designs in human patients, where additional variables such as cell quality, tumor burden, and drug pharmacokinetics could individually and jointly impact the behavior of engineered cells.

In addition to toxicities associated with overly active immune responses, off-tumor toxicity presents an additional challenge to cell-based cancer immunotherapy. The lack of targetable tumor-restricted antigens necessitates the targeting of tumor-associated antigens (TAA) that are not exclusively expressed on tumor cells, thus exposing healthy tissues that express the same antigen to “on-target off-tumor” toxicity. For example, CD19 CAR-T cell and CD19 CAR-NK cell therapies both result in B-cell aplasia in responding patients, as the CD19-targted immune cells simultaneously eliminate CD19-expressing healthy and malignant B cells (17, 39). B-cell aplasia is a clinically manageable condition (40), but analogous on-target, off-tumor toxicities against other antigens such as HER2, mesothelin, and carcinoembryonic antigen have led to early trial terminations and patient fatalities (41, 42).

Tumor-targeting specificity can be improved by biological circuitry that computes AND-gated or AND-NOT–gated Boolean logic, which requires the engagement of multiple TAAs in a specific combination before triggering target-cell killing (Fig. 1C). An AND-gate CAR requires that two or more antigens are present to trigger CAR signaling. One way to achieve this is to distribute the CD3ζ activation domain and costimulatory domain (e.g., CD28 or 4-1BB) of a typical second-generation CAR into two different receptors, one first-generation CAR containing only the CD3ζ chain, and a chimeric costimulatory receptor (CCR) containing only the costimulatory domain (43). Both receptors must be bound to their respective ligands to trigger full-intensity T-cell responses against the target cell. Another approach to achieving AND-gate logic requires sequential detection of multiple antigens, such as with the synthetic Notch (synNotch) or SNIP receptor systems (44–46). These designs require two gene-expression cassettes. The first cassette constitutively expresses a synthetic receptor containing a transcription factor that is cleaved and released when the receptor binds its ligand (antigen A). The released transcription factor translocates to the nucleus and drives the inducible expression of a conventional CAR, which subsequently targets antigen B. Such AND-gated designs have been shown to prevent killing of off-tumor targets that express only one antigen but not the other. However, AND-gate designs remain capable of off-tumor killing if healthy tissue expressing the CAR-targeted antigen is colocalized with the tumor cells, as shown in the case of a synNotch-controlled ROR1 CAR-T cell therapy that triggered severe toxicity in the Raji lymphoma model due to simultaneous destruction of ROR1-expressing lymphoma and healthy tissue in the bone marrow (47).

As an alternative strategy, AND-NOT–gate CARs have also been shown to increase target specificity. AND-NOT gates require both the presence of antigen A and the absence of antigen B to trigger cell activation (48, 49). This is accomplished by coexpressing a conventional, activating CAR (aCAR) targeting antigen A with an inhibitory CAR (iCAR) targeting antigen B. iCAR signaling triggered by antigen B overrides aCAR signaling, thus inhibiting cell activation when antigen B is present. By using an aCAR that targets a TAA and an iCAR that targets HLA-A2, this AND-NOT–gate strategy has been applied to target tumor cells that have downregulated MHC expression, thus simultaneously increasing tumor specificity while addressing a potential immune escape mechanism (49, 50).

Despite significant potential advantages, AND-gated and AND-NOT–gated CAR designs also face important caveats. First, each of the circuits described above require the expression of multiple transgenes, and the increased genetic payload size can significantly reduce the efficiency of transgene integration (51). Second, early synthetic-biology attempts at making computation circuits, exemplified by the synNotch system, often utilized nonhuman components such as viral transcription factors, and the potential immunogenicity of such designs presents a significant barrier to clinical translation. Finally, by increasing the number of antigens that must be present or absent in a specific combination, the system increases the probability for antigen escape, as the tumor now only needs to alter one of multiple antigens’ expression pattern to avoid detection.

Although tumor specificity—that is, ability to specifically recognize tumor cells and not healthy tissue—is a critical consideration in therapy design due to its safety implications, the flip side of the coin—that is, inability to recognize all tumor cells by targeting a single antigen—also presents a key challenge in cell-based therapy. Antigen escape can arise from either natural tumor heterogeneity or antigen downregulation in response to selective pressure imposed by therapy. For example, a substantial fraction of patients with B-cell malignancies relapse after CD19 CAR-T cell therapy, and up to 94% of relapsing patients have CD19-negative tumors (52–54). Antigen escape poses an even greater challenge in solid tumors such as glioblastoma multiforme (GBM) due to intrinsic heterogeneity in antigen expression (55). Consequently, antigen loss has been observed as a major mechanism of CAR-T cell treatment resistance in patients with GBM (56, 57). Several methodologies have been proposed to address antigen escape, including incorporating tandem bispecific CARs or multiple CARs targeting different antigens into a single cell (OR-gate Boolean logic), or simultaneously or sequentially administering multiple cell products that target different antigens. In most head-to-head comparisons, tandem bispecific CARs have exhibited greater antitumor efficacies compared with coexpressing two CARs in a cell or administering a pooled T-cell population (58–60). In OR-gate Boolean logic, therapeutic cells are engineered to recognize two or more antigens, and the presence of any recognizable antigen would trigger cell-mediated toxicity, thus requiring the tumor to lose all recognizable antigens to successfully evade detection. For example, tandem single-chain variable fragment (scFv) bispecific CARs have demonstrated reduced tumor relapse and superior antitumor activity against B-cell malignancies susceptible to CD19 antigen loss (61, 62). Early success with CD19/CD20- and CD19/CD22-targeting bispecific CARs have led to the initiation of numerous ongoing clinical trials (NCT04007029, NCT04700319, NCT03241940). Multi-antigen targeting has also been extended to preclinical and clinical studies in other malignancies, including multiple myeloma (e.g., NCT04162353; ref. 60) and GBM (58). It should be noted that, depending on the choice of target antigens, OR-gated CAR designs have the potential to exacerbate off-tumor toxicity, as a wider range of healthy tissue may become recognizable to engineered cells that target multiple antigens.

To date, designs that enable regulated or logic-gated CAR activities have largely been demonstrated in the T-cell setting. However, such designs may become increasingly important in CAR-NK cell engineering as this treatment modality expands beyond B-cell malignancies and into indications with greater toxicity potential or more heterogeneous antigen-expression profiles (63). Despite the largely safe clinical profile of NK-cell therapies reported to date (17, 64), biological designs that enable regulated CAR activity and logic-gated signal computation remain useful resources as CAR-NK cells advance to a broader array of disease indications.

To date, much of the synthetic biology efforts in cell-based therapies have focused on the engineering of CAR proteins or circuitries that revolve around CAR proteins. However, numerous biological pathways work in concert to impact the antitumor activities of T cells and NK cells, thus providing a wide variety of engineering targets that can potentially enhance efficacy. To date, solid tumors remain intractable challenges for cell-based therapies due to various immunosuppressive and immune-evading mechanisms unique to the tumor microenvironment (TME), such as overexpression of inhibitory ligands and cytokines, hypoxia, a poor nutrient profile, and dysfunctional immune or stromal cells (65). To address this, CAR-expressing immune cells have been “armored” with cytokines, chemokines, receptors, or other transgenic molecules to enhance T-cell cytotoxicity, persistence, or tumor infiltration, or to remodel the TME to a more proinflammatory state that favors antitumor activity (ref. 66; Fig. 1D). CAR-T cells have been engineered to coexpress cytokines such as IL12 (67), IL7 (68), and IL15 (69), as well as payloads such as bispecific T-cell engagers (70, 71) and bacterial virulence factors (72) to improve targeting of solid tumors with heterogeneous antigen expression. Many armored CAR-T cells are currently being evaluated in the clinic (Table 3), and promising preliminary results have begun to emerge. In a phase-I clinical trial (NCT03198546), 6 patients with liver, pancreatic, or ovarian cancer were treated with mesothelin- or GPC3-targeting CAR-T cells armored with IL7 and CCL19. A preliminary report notes one complete response and one partial response among the treated patients, with fever and fatigue as common side effects but “no grade 2–4 adverse events or major complications” reported (73). Although the number of patients treated with armored CAR-T cells to date is relatively small, the potential to achieve response without severe toxicity in patients with solid tumors is highly encouraging.

Likewise, NK cells have been engineered to express armors that exploit their biological characteristics to further improve therapeutic outcomes. In a landmark clinical trial, cord-blood NK cells were engineered to express a second-generation CD19 CAR plus IL15 to enhance survival, as well as inducible caspase 9 as a safety switch (17). The therapy was efficacious and well tolerated in a large cohort of lymphodepleted patients with relapsed or refractory CD19-postive malignancies, with a 64% complete response rate and CAR-NK cells remaining detectable in peripheral blood for >12 months. This persistence level contrasts with earlier adoptive NK-cell transfer regimens, where NK-cell persistence was limited to days or weeks without cytokine supplementation (19, 74). Armoring strategies designed to promote the proliferation and persistence of therapeutic cells may be particularly critical in the allogeneic setting, where the durability of response has been a major concern. Whereas autologous CAR-T cells have been shown to persist in treated patients for a decade (75), the durability of response in patients treated with allogeneic CAR-T and CAR-NK cell is too early to fully assess as clinical datapoints remain relatively low in number. The fact that NK cells do not establish long-term memory like T cells, and that allogeneic CAR-T cells remain vulnerable to immune rejection despite genetic knockout of endogenous T-cell receptor and/or MHC molecules (76–79), render engineering strategies that can artificially booster cell persistence of particular interest.

In addition to armoring T cells and NK cells with cytokines to drive proliferation, one could also combine therapeutic cells with other forms of immunotherapy to further enhance the immune response. Preclinical studies have suggested that combination of CAR-T cell therapy with anti-PD-1 immune checkpoint blockade could promote the functional persistence of CAR-T cells (80, 81). However, early clinical results reported to date have not yet demonstrated definitive advantage of combining CAR-T cell therapy with checkpoint blockade (82, 83). Pursuing a different modality of combination therapy, an ongoing phase-I clinical trial by BioNTech (NCT04503278) is evaluating the effect of an mRNA vaccine designed to target claudin-6 (CLDN-6) expression to dendritic cells, which can subsequently boost the activation and proliferation of anti-CLDN-6 CAR-T cells. Early results from the trial suggest promising efficacy and safety (84), and the clinical experience generated from such trials will prove highly valuable as the field explores different combination strategies to tackle solid tumors.

Yet another strategy of armoring immune cells is to equip the cells with receptors that can either abrogate immunosuppression or convert suppressive signals in the TME into stimulatory ones for the engineered cells. For example, TGFβ is a potent inhibitory cytokine in the TME. To overcome this obstacle, researchers have evaluated strategies to ablate TGFβ signaling, such as CRISPR/Cas9-mediated knockout of the endogenous TGFβ receptor chain II (NCT04976218). Alternatively, expression of the dominant-negative TGFβ receptor (DNR), which is a truncated TGFβ receptor chain II that lacks the intracellular domain, is also under investigation. The DNR has been coexpressed with tumor-targeting receptors in both T cells (85) and NK cells (86), and preclinical data indicate the DNR can robustly abrogate endogenous TGFβ signaling by both serving as a sink for TGFβ binding and by poisoning the heterodimeric TGFβ receptor complex to abolish signaling. A phase-I clinical trial (NCT03089203) treated patients with prostate cancer using T cells that coexpress the DNR with a prostate-specific membrane antigen (PSMA)-targeting CAR (87). In this trial, the best response was stable disease, achieved by five of 13 patients. Of note, the only patient to receive the highest dose in the trial (1–3 × 10⁸ m⁻² CAR-T cells) experienced clonal CAR-T cell expansion and a 98% reduction in serum PSA levels, but rapidly developed grade-4 CRS and died because of complications from sepsis and multimodal immunosuppression. The underlying mechanisms that drove the dramatic clonal T-cell expansion and severe toxicity in this individual patient remain unresolved at this time, but this study highlights the need to further explore both the potential and the risks associated with armoring CAR-T cells to overcome immunosuppression.

Taking it one step further, researchers have engineered synthetic receptors that convert TGFβ binding into immunostimulatory signaling. For example, Burga and colleagues fused the DNR to DNAX-activation protein 12 (DAP12), and demonstrated that NK cells bearing the fusion protein demonstrated improved efficacy and persistence compared with the DNR in mice bearing TGFβ-secreting neuroblastoma xenografts (88). As another example, Chang and colleagues developed CARs that can respond to soluble antigens and demonstrated the ability to activate T cells in response to TGFβ through the expression of a TGFβ-binding CAR (89). CARs responsive to additional soluble antigens can potentially broaden this signal-conversion strategy to overcome a variety of immunosuppressive cytokines in the TME (90).

Finally, endogenous receptors can also be modified to enhance effector cell function. For example, Zhu and colleagues demonstrated that the NK cell–activating receptor FcγRIIIa (CD16a), which is responsible for NK cell–mediated antibody-dependent cellular cytotoxicity, can be mutated to prevent proteolytic cleavage by ADAM17 (91). iPSC-derived NK cells expressing the mutated form of CD16a significantly outperformed unmodified peripheral blood NK cells in controlling Raji xenografts when coadministered with an anti-CD20 antibody (rituximab) in a repeated-dosing study. An NK-cell product (FT596) containing both the “NK-CAR” and optimized CD16a, along with membrane-tethered IL15, is currently under clinical evaluation. Preliminary results from this trial are encouraging: 20 patients with relapsed and/or refractory B-cell lymphoma were treated with escalating doses of 20–300 million engineered NK cells, with or without cotreatment with rituximab, resulting in >50% objective response rate among 17 efficacy-evaluable patients, including seven complete responses (16). Two patients who achieved a complete response had previously relapsed after CAR-T cell therapy, highlighting the potential for this therapy to serve as an alternative to CAR-T cells. This remarkable outcome underscores the utility of systematically investigating and incorporating synthetic payloads rooted in biological relevance, a methodology that may lead to next-generation cellular therapies equipped to overcome currently intractable malignancies.

As the synthetic-biology toolkit continues to expand, the generation of increasingly sophisticated biological circuits has become possible in the engineering of cell-based therapies for cancer. Exciting preclinical and clinical data for NK-cell–based and T-cell–based therapies are now emerging, demonstrating higher specificity, safety, and durability. Increasing complexity in biological design is frequently coupled with increasing complexity in manufacturing as well as vulnerability to unintended consequences due to system cross-talk or component failures that compromise overall system function. Therefore, the implementation of multi-layered biological circuits in the cell-based therapy context must be done in a judicious manner to maximize robustness and minimize unnecessary complexity. Nevertheless, the wide array of novel biological functions made accessible by synthetic biology approaches offer potential solutions to many roadblocks currently limiting the broad application cell-based immunotherapies, and provides avenues to extend these therapies to more patients with advanced cancers.

J.D. Clubb reports grants from NIH during the conduct of the study. Y.Y. Chen reports grants from NIH, National Science Foundation, Mark Foundation for Cancer Research, Parker Institute for Cancer Immunotherapy, Jean and Stephan Kaplan, and Cancer Research Institute during the conduct of the study as well as personal fees from ImmPACT Bio, Catamaran Bio, Notch Therapeutics, Prime Medicine, Sonoma Biotherapeutics, Waypoint Bio, and Pluto Immunotherapeutics outside the submitted work; in addition, Y.Y. Chen has a patent for US 11,160,833 B2 issued and licensed to ImmPACT Bio, a patent for US 11,253,546 B2 issued and licensed to ImmPACT Bio, and a patent for US 11,014,980 B2 issued and licensed to ImmPACT Bio. No disclosures were reported by the other author.

J.D. Clubb is supported by the UCLA Tumor Immunology Training Grant (USHHS Ruth L. Kirschstein Institutional National Research Service Award # T32CA009120). T.A. Gao is supported by the National Science Foundation Graduate Research Fellowship Program (fellowship to T.A. Gao) and the Mark Foundation for Cancer Research (18-029-ELA, grant to Y.Y. Chen). Y.Y. Chen is supported by the Parker Institute for Cancer Immunotherapy, Jean and Stephan Kaplan, and Cancer Research Institute (CRI2701).