Pooled CAR-T screening in pig-tailed macaques identifies designs with enhanced proliferation, trafficking, and persistence.

Introduction

Chimeric antigen receptor T-cell (CAR-T) therapy has transformed the treatment of B-cell malignancies, leading to the approval of 8 therapies targeting CD19 and B-cell maturation antigen (BCMA). However, tumor recurrence and patient relapse remain significant challenges, occurring in 50% to 70% of patients with diffuse large B-cell lymphoma and B-cell acute lymphoblastic leukemia treated with CD19-targeted CAR-Ts 1 , 2 , 3 , 4 , 5 , 6 and in >80% of BCMA CAR-T–treated patients with multiple myeloma. 7 , 8

The high rate of relapse highlights a critical need to improve the long-term efficacy of CAR-T therapy. Although various factors contribute to relapse, mounting evidence suggests that limited in vivo CAR-T persistence is a primary challenge. 9 Relapse of antigen-positive tumor cells is frequently associated with limited CAR-T durability, 10 , 11 detected in 33% to 78% of CD19-targeted 4 , 10 , 11 , 12 , 13 and 67% to 96% of BCMA-targeted CAR-T 7 , 14 , 15 relapse cases. Recent studies have further identified CAR-T persistence as a crucial determinant of long-term disease-free survival, 16 , 17 highlighting the need for strategies to enhance CAR-T durability.

The structure of a CAR plays a critical role in shaping CAR-T function, phenotype, and persistence. 18 CAR molecules comprise several components, including antigen recognition, hinge, transmembrane, costimulatory, and CD3ζ activation domains. The costimulatory domain is particularly influential in modulating CAR-T kinetics and durability. 19 , 20 Approved CAR-T therapies incorporate either 4-1BB or CD28 costimulatory domains, which exhibit distinct profiles: 4-1BB promotes greater persistence with slower expansion, whereas CD28 drives rapid expansion but earlier T-cell exhaustion. 21 , 22 Recent studies have identified additional domains, including Ox40, 23 , 24 , 25 B-cell activating factor receptor (BAFF-R), 26 and MyD88-CD40, 27 , 28 , 29 which have demonstrated the potential to further augment CAR-T function. Hinge and transmembrane domains also influence CAR-T activity, particularly by modulating antigen sensitivity. 30 , 31 , 32 , 33 , 34 Despite these advances, no universally superior CAR architecture has been identified. As such, direct head-to-head comparisons of CAR structures could provide critical insights into designs that improve CAR-T efficacy.

Developing and validating novel CARs through clinical trials is an exceedingly expensive, time-consuming, and resource-intensive process, with a substantial risk of failure. Moreover, commonly used xenografted mouse models frequently fail to recapitulate complex human immunology and CAR-T–mediated toxicities, including cytokine release syndrome (CRS), limiting their utility in predicting clinical outcomes. 35 , 36 To address these challenges, the nonhuman primate (NHP) model has emerged as a valuable tool for studying CAR-T therapy. Unlike immunocompromised mouse models, NHPs possess a functional immune system and can recapitulate drug-associated toxicities observed in humans. The longer life span of NHPs allows for extended safety and efficacy studies, and their larger size enables longitudinal biopsies and testing of clinical-scale manufacturing, which are not feasible in mice. 37 Our group and others have previously developed an NHP model for B-cell–targeted CAR-T therapy. 38 , 39 , 40 , 41 This model has demonstrated utility in recapitulating key clinical findings, including potent B-cell depletion in blood and tissues, and management of CAR-T–mediated toxicities. Here, we hypothesized that the NHP model could be leveraged in a pooled screening format to simultaneously evaluate multiple CAR structures and identify designs with superior activity. Our findings underscore the importance of CAR design in shaping CAR-T proliferation, persistence, and trafficking and highlight the value of immunocompetent NHP models for advancing next-generation CAR-T therapies.

Methods

CAR-T design and manufacturing

A CAR array was designed to span different combinations of hinge, transmembrane, and costimulatory domains, with a common anti-CD20 single-chain variable fragment (scFv) and CD3ζ activation domain (further details in the supplemental Methods and supplemental Table 1 , available on the Blood website). CAR-Ts were manufactured as described. 39 Briefly, CD4 + and CD8 + T cells were isolated from peripheral blood mononuclear cells (PBMCs) using bead-based selection (STEMCELL Technologies) and cultured in X-Vivo 15 (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% l -glutamine, 55-μM 2-mercaptoethanol, and 5 ng/mL recombinant interleukin-7 (IL-7) and IL-15 (PeproTech). T cells were activated with artificial antigen-presenting cells (kind gift from James Riley 42 ) at a 2:1 ratio. Three days later, T cells were transduced with lentiviral vectors with 5 μL of LentiBlast Transduction Reagent (OZ Bioscience), followed by a 7-day expansion in G-Rex24 flasks (Wilson Wolf).

Flow cytometry and chronic antigen stimulation assay

Antibodies are listed in supplemental Table 2 . All flow panels included a live/dead stain to exclude dead cells. The long-term function of CAR-Ts was evaluated through chronic stimulation with irradiated CD20-expressing K562 (K562-CD20) cells. To standardize initial conditions, CAR-T variants were analyzed for CAR expression via flow cytometry and diluted to 20% CAR + cells by adding untransduced autologous T cells. A total of 5 × 10 5 CAR-Ts were cocultured with 1 × 10 5 irradiated K562-CD20 or wild-type K562 (K562-wt) cells (effector-to-target ratio of 5:1). Every 3 to 4 days, CAR-Ts were harvested, with half being restimulated and half stained with immunophenotyping panels and analyzed by flow cytometry. The same experiment was performed with equal numbers of pooled CAR + cells from each CAR-T variant, with half being restimulated and half snap frozen for analysis via droplet digital polymerase chain reaction (ddPCR).

In vivo NHP CAR-T studies

Animals received cyclophosphamide (40 mg/kg) on days –7 and –6 before CAR-T infusion. To ensure equal pooling of CAR variants in infusion products, CAR-T variants were analyzed for CAR expression by flow cytometry, and equal numbers of CAR + variants were pooled to achieve a final dose of 9 × 10 6 CAR + cells per kg ( supplemental Table 3 ). Irradiated K562-CD20 cells were infused at 2.5 × 10 7 cells per kg on days 21 and 35 after CAR-T infusion for antigen restimulation. 38 , 39 Blood was collected longitudinally until necropsy. Sampling in spleen, lymph nodes, cerebrospinal fluid (CSF), bone marrow, and bronchoalveolar lavage was performed at –2, 2, and 4 weeks after CAR-T infusion. Tissues were processed for downstream analyses using established protocols. 39 Details on CRS management and immunohistochemistry are presented in the supplemental Methods.

ddPCR assay

ddPCR was used to quantify CAR variant abundance. Unique primer/probe sets were designed and validated for specific detection ( supplemental Table 4 ). Genomic DNA (gDNA) was extracted (Qiagen), diluted to ∼100 ng/μL, and divided across 6 ddPCR assays, with 1 to 10 μL of gDNA per reaction. Each multiplex assay targeted a subset of CAR variants (2- to 6-plex). To normalize these data, 1:10 diluted gDNA was used to detect a housekeeping gene (macaque RNAse-P p30 subunit) 43 and T-cell receptor D, a gene that is excised during T-cell receptor gene rearrangement and is therefore only present in non-T cells. 44 The ddPCR reactions were performed using the Droplet Digital PCR System (Bio-Rad) with ddPCR Multiplex Supermix reagents. Negative controls (untransduced T cells and H 2 O) and single-plex CAR-Ts were routinely included to ensure the specificity of multiplexed assays. Approximately 20 000 droplets were generated per reaction. Samples with >3 ddPCR-positive droplets were considered positive. Further information about ddPCR normalization and sensitivity measurements are presented in the supplemental Methods.

Results

Design and characterization of a combinatorial CD20-targeted CAR array

To systematically assess how CAR structure affects CAR-T activity, we designed an array of CD20-directed CARs spanning all combinations of established hinge (CD8 and CD28), transmembrane (CD8 and CD28), and costimulatory domains (4-1BB, CD28, BAFF-R, MyD88-CD40, and Ox40; Figure 1 A). For clarity, we refer to each CAR construct by a variant number (variant 1 [V1] to V20) and group them throughout the article according to their costimulatory domain: 4-1BB (V1-V4), CD28 (V5-V8), BAFF-R (V9-V12), MyD88-CD40 (V13-V16), and Ox40 (V17-V20; Table 1 ). To enable direct comparison, we retained the same CD20 scFv and CD3ζ activation domains across all designs 38 , 39 and included a coexpressed truncated epidermal growth factor receptor (EGFRt) to detect CAR-transduced cells. 45 We manufactured the CAR-T array using established methods, 38 , 39 including isolation and activation of CD4 + and CD8 + cells from PBMCs, transduction with lentiviral vectors, and expansion in flasks that maximize gas and nutrient convection ( Figure 1 B).

All CAR variants were successfully manufactured ex vivo, with MyD88-CD40 (V13-V16) and BAFF-R CAR-Ts (V9-V12) exhibiting enhanced expansion during manufacturing ( Figure 2 A). CAR-T variants were phenotyped by flow cytometry after manufacturing ( supplemental Figure 1 ), with each variant demonstrating robust CAR expression ( Figure 2 B), balanced CD4:CD8 ratios ( Figure 2 C), and a predominant central memory phenotype (CD28 + CD95 + ; Figure 2 D). 38 Among these, BAFF-R and MyD88-CD40 CAR-Ts displayed higher CD4:CD8 ratios and a greater proportion of central memory T cells than 4-1BB (V1-V4) or CD28 CAR-Ts (V5-V8). Notably, MyD88-CD40 CAR-Ts showed significantly increased expression of CD25 ( Figure 2 E; P < .0005) and HLA-DR activation markers ( Figure 2 F; P < .00005), 46 suggesting this domain confers an hyperactivation phenotype. These data validate our arrayed anti-CD20 CAR-T manufacturing and highlight CAR-dependent phenotypes that emerge during ex vivo manufacturing.

CAR-T variants display distinct functional profiles ex vivo

We next extended our ex vivo experiments to explore the impact of CAR architecture on CAR-T function, including proliferative, cytokine, and cytolytic responses. To measure CAR-specific proliferation, we quantified the dilution of CellTrace Violet dye ( Figure 3 A). We observed robust division of CAR + cells when stimulated with CD20-expressing K562 target cells (K562-CD20), with minimal bystander proliferation in the CAR – fraction ( Figure 3 B; P < .00005). Notably, MyD88-CD40 CAR-Ts uniquely exhibited significant tonic signaling, as indicated by antigen-independent proliferation of CAR + cells ( Figure 3 C; P < .00005). These cells displayed phenotypic features of tonic signaling, 47 including elevated CD25 and HLA-DR, as well as reduced expression of exhaustion markers programmed cell death protein 1 (PD-1) and T-cell immunoreceptor with Ig and ITIM domain (TIGIT; Figure 3 D). In parallel, we evaluated transcription factor activity using a Jurkat triple parameter reporter cell line, 48 which showed that CD28-based CARs (V5-V8) exhibited significantly higher AP-1, NFAT, and NF-κB activation under both antigen-positive and antigen-negative stimulation ( supplemental Figure 2 A-C), in contrast to MyD88-CD40 CARs, which did not exhibit measurable tonic signaling despite clear evidence of tonic proliferation in primary T cells ( Figure 3 C). After K562-CD20 stimulation, all CAR variants exhibited antigen-specific cytokine secretion ( supplemental Figure 3 A-G), with distinct patterns across costimulatory domain groups. Most notably, CD20-stimulated MyD88-CD40 CAR-Ts produced minimal IL-6 ( Figure 3 E), significantly reduced IL-10 ( Figure 3 F; P < .005), and increased IL-13 ( Figure 3 G; P < .005), a cytokine less characterized in CAR-T biology but previously linked to MyD88-CD40 signaling. 28 Assays measuring the specific lysis of target cells over time ( Figure 3 H) confirmed antigen-specific cytolysis of LLCMK2-CD20 target cells across all CAR variants, with minimal nonspecific killing of wild-type LLCMK2 cells ( Figure 3 I; P < .00005 ). Interestingly, MyD88-CD40 CAR-Ts also displayed modest to significantly enhanced antigen-dependent cytotoxicity ( supplemental Figure 3 H). Finally, we applied principal component analysis (PCA) to integrate data from short-term functional assays ( Figure 3 B-I). PCA further highlighted the distinctiveness of MyD88-CD40 CAR-Ts, which clustered separately from other variants ( Figure 3 J). Collectively, these findings underscore the influence of costimulatory domains on CAR-T function, with MyD88-CD40 CARs distinguished by tonic signaling, unique cytokine profiles, and potent cytolytic activity.

Chronic antigen stimulation highlights phenotypic and functional differences among CAR-T variants

To investigate the influence of CAR design on in vivo persistence, 21 , 22 we next compared the long-term function of CAR-T variants in an arrayed ex vivo chronic antigen stimulation assay. CAR-Ts were repeatedly exposed to K562-CD20 cells to evaluate CAR-T durability amid exhaustive conditions ( Figure 4 A). 49 , 50 , 51 , 52 All variants demonstrated robust expansion, marked by increased EGFRt expression across successive stimulations ( Figure 4 B), with CD28 CAR-Ts exhibiting greater cumulative expansion ( Figure 4 C). Consistent with previous reports, 27 , 28 , 53 only MyD88-CD40 CARs retained lower PD-1 ( Figure 4 D; P < .0005) and TIGIT ( Figure 4 E; P < .005) after chronic stimulation, suggesting this domain confers superior resistance to exhaustion. MyD88-CD40 CAR-Ts also produced less IL-10 ( Figure 4 F) and substantially more IL-13 ( Figure 4 G; P < .00005). Interestingly, although all CAR-T variants demonstrated antigen-specific proliferation ( Figure 4 B) and interferon gamma (IFN-γ) secretion ( Figure 4 H) after chronic stimulation, 4-1BB CAR-Ts failed to retain antigen-specific cytotoxicity ( Figure 4 I), suggesting that chronic stimulation may induce a state of partial dysfunction unique to 4-1BB CAR-Ts in macaques. PCA of chronic stimulation data further highlighted the distinct clustering of MyD88-CD40 and 4-1BB CARs ( Figure 4 J), underscoring the role of costimulatory domains in shaping CAR-T exhaustion, cytokine secretion, and cytotoxicity during prolonged antigen exposure.

Pooled chronic stimulation ex vivo reveals donor-specific proliferation of CAR-T variants

Before in vivo experiments, in which CAR variants were pooled and infused into single recipient NHPs, we developed a competitive ex vivo stimulation assay in which pooled CAR-T variants were serially stimulated with K562-CD20, as shown in Figure 4 A. Because flow cytometry cannot distinguish unique CAR variants, we turned to CAR quantitation at the DNA level using ddPCR. Our variant-specific ddPCR assay was designed to enable precise quantification of individual CAR variants within pooled samples. Primer-probe sets were designed for individual CAR variants and validated to ensure variant-specific amplification ( supplemental Figure 4 A) and sensitivity to single-copy thresholds ( supplemental Figure 4 B). Donor-specific expansion patterns emerged, with 4-1BB or CD28 CARs preferentially expanding ( Figure 5 ). Notably, V7 (CD8 hinge, CD28 transmembrane, and CD28 costimulatory domain) showed significant enrichment in 2 donors, whereas BAFF-R or MyD88-CD40 CARs exhibited no expansion across all donors, despite demonstrating robust proliferation in arrayed chronic stimulation assays ( Figure 4 B). We next performed a corresponding competitive in vivo study using the same biological donors to determine whether these ex vivo findings were recapitulated in vivo.

Pooled CAR-Ts expand in vivo, induce B-cell aplasia, and exhibit distinct inflammatory profiles

After extensive ex vivo characterization, we next evaluated our CAR array in a cohort of 3 pig-tailed macaques. We pooled CAR-T variants in equal proportions ( supplemental Figure 5 ) and infused into autologous, conditioned NHPs ( Figure 6 A-B). 39 Each animal received a dose of 9 × 10 6 CAR + cells per kg, similar to previous studies. 38 , 39 To align with our ex vivo data and previous methods, 38 we provided supplemental antigen in the form of irradiated K562-CD20 cells that were administered at 3 and 5 weeks after CAR infusion. CAR-T proliferation, trafficking, and persistence were evaluated in blood and tissues using flow cytometry ( supplemental Figure 6 ) and ddPCR for up to 102 days after infusion.

Total CAR-Ts expanded significantly after infusion, constituting 30% to 96% of peripheral CD3 + T cells, with peak levels observed between 7 to 11 days after infusion ( Figure 6 C-E). CAR-T expansion coincided with rapid B-cell depletion in blood ( Figure 6 C-E) and substantial B-cell follicle loss in lymph nodes and spleen, as confirmed by immunohistochemistry ( Figures 6 F-G). The K562-CD20 boost was well tolerated in all animals, with no evidence of immune perturbation despite their xenogeneic origin. In Z21020 , this boost was followed by a modest secondary CAR-T expansion at day 25 ( Figure 6 C). Z22040 exhibited extreme CAR-T expansion within the first week (92%-96% of CD3 + T cells) and developed immune effector cell–associated neurotoxicity syndrome (ICANS)–like symptoms despite intensive supportive care ( supplemental Table 3 ), requiring an end of study at day 9. Importantly, integration site analysis of infusion products and PBMCs confirmed highly polyclonal expansion (7036-17 501 unique clones) of CAR-Ts in all NHPs ( supplemental Table 5 ). For Z21020 and Z21093 , B cells recovered between 30 and 40 days after infusion, although levels remained low through study completion ( Figure 6 G). Z21020 and Z21093 developed gut infections at 64 and 29 days after infusion, respectively, which did not respond to antibiotic treatment and were associated with progressive weight loss and clinical decline. Based on clinical signs and low levels of CAR + cells at these time points, these animals were taken off study at days 92 and 102. Clinical and histologic findings at necropsy revealed only subclinical ( Z21020 ) and mild-to-moderate colonic pathology ( Z21093 ; supplemental Table 6 ).

To contextualize these findings, we compared outcomes from our pooled CAR array cohort to a previously published historical cohort of pig-tailed macaques that received V1 CAR-Ts under highly comparable conditions, including the same cyclophosphamide (Cytoxan) preconditioning, CAR-T manufacturing, and dose (9 × 10 6 CAR + cells per kg). 39 Intriguingly, inflammatory biomarkers in serum were distinct between these cohorts. Unlike the historical V1 cohort, which exhibited a significant serum IL-6 spike after CAR-T infusion, the pooled CAR array cohort showed no IL-6 elevation ( Figure 6 H) but had markedly higher ferritin levels ( Figure 6 I). C-Reactive protein levels were comparable ( Figure 6 J). Peak plasma cytokines measured during acute CAR expansion showed decreased IFN-γ and tumor necrosis factor α and elevated IL-2 in the CAR array cohort ( supplemental Figure 7 A). In longitudinal analyses, we observed a modest increase in IL-2 coinciding with MyD88-CD40 expansion ( supplemental Figure 8 ), although overall cytokine trends were variable. We performed lumbar punctures to collect CSF during acute CAR expansion and observed increased IL-18 in CAR array NHPs compared to the historical V1 cohort, suggesting a unique neuroinflammatory response ( supplemental Figure 7 B). We also assessed postinfusion PBMCs to compare CAR-T phenotypes across cohorts. Although observed differences did not reach statistical significance, at early time points, CAR-Ts from the CAR array cohort trended toward greater expansion of CD8 + -dominant, activated (HLA-DR + ), and proliferative (Ki-67 + ) populations, with lower TIGIT expression suggestive of reduced exhaustion ( supplemental Figure 9 ). Collectively, the pooled CAR-T array demonstrated robust CAR-T expansion, durable B-cell depletion, and distinct inflammatory dynamics.

MyD88-CD40 CARs preferentially expand, traffic, and persist in vivo

We surmised that the divergent inflammatory profiles in CAR array animals relative to our historical V1 cohort and the increased toxicity observed in Z22040 were driven by distinct CAR variant(s) within the array. Consistent with this hypothesis, longitudinal variant-specific ddPCR revealed significant enrichment of MyD88-CD40 CAR-Ts (V13-V16) in PBMCs across all time points and NHPs ( Figure 7 A-C). Similarly, MyD88-CD40 CAR-Ts were strikingly enriched in lymph nodes, spleen, bone marrow, bronchoalveolar lavage, and CSF samples collected within 2 weeks of infusion ( Figure 7 D; P < .005). Among these, V14 (CD28 hinge/transmembrane and MyD88-CD40 costimulation) was most abundant during early expansion ( Figure 7 E) and clustered distinctly from other MyD88-CD40 variants via PCA ( Figure 7 F). 4-1BB and Ox40 CARs (V17-V20) showed some early expansion and modest tissue trafficking but did not persist beyond 2 weeks. Neither CD28 nor BAFF-R CARs showed appreciable expansion in blood or tissues.

MyD88-CD40 CARs persisted in blood and tissues until necropsy ( supplemental Figure 10 A-B). At necropsy, V13, V14, and V15 showed significant enrichment across various tissues ( supplemental Figure 10 C), and PCA of necropsy data revealed these variants clustered distinctly from each other and the rest of the CAR array ( supplemental Figure 10 D). Together, our NHP study demonstrates that MyD88-CD40 domains drive superior CAR-T expansion, persistence, and tissue trafficking in vivo, significantly outperforming conventional 4-1BB and CD28 designs.

Discussion

Although CAR-T therapy has redefined the treatment landscape of numerous B-cell malignancies, achieving long-term efficacy remains a challenge, in part due to suboptimal CAR-T persistence. Addressing this limitation will be paramount for unlocking the full potential of CAR-T therapies, not only in the context of B-cell malignancies but also for expanding their application to solid tumors, infectious disease, and autoimmunity. 39 , 54 We developed a novel in vivo CAR competition assay in NHPs to simultaneously evaluate multiple CAR designs in an immunocompetent background. We identified specific CAR architectures, notably the MyD88-CD40 costimulatory domain, that drive enhanced CAR-T proliferation, persistence, and tissue trafficking.

In primary cell–based assays, MyD88-CD40 CAR-Ts exhibited 4 key features: (1) enhanced activation, (2) tonic signaling, (3) distinct cytokine secretion, and (4) resistance to exhaustion after chronic antigen stimulation. These characteristics likely supported the robust in vivo proliferation of MyD88-CD40 CARs. Importantly, a recent pooled CAR-T screen also identified MyD88-CD40 as a superior design in a B-cell cancer model, as observed in our NHP study, but not in a solid tumor model, 55 suggesting that the efficacy of MyD88-CD40 may vary with tumor type and microenvironment. This group similarly linked MyD88-CD40 to a hyperactivation and tonic signaling phenotype, reinforcing the reproducibility of our findings.

Although our screen was designed to systematically evaluate the influence of hinge, transmembrane, and costimulatory domains, PCA analyses across ex vivo and in vivo data sets consistently showed that CAR variants clustered primarily by costimulatory domain, suggesting that costimulation was the main driver of phenotypic and functional differences. One notable exception was V14 (CD28 hinge/transmembrane and MyD88-CD40 costimulation). V14 was significantly enriched during early time points after infusion and clustered separately from other costimulatory groups and MyD88-CD40 variants via PCA. This finding aligns with prior studies reporting that CD28 hinge and transmembrane domains enhance CAR activity and antigen sensitivity, which may have contributed to V14’s early expansion. 32 Notably, this effect was not observed in 4-1BB CAR-Ts, which also expanded early but did not show preferential enrichment of CD28 hinge/transmembrane variants, suggesting that this enhancement may be specific to MyD88-CD40 signaling.

The impact of tonic signaling on CAR-T efficacy remains controversial. Many groups have linked tonic signaling to scFv stability and characterized it as a detrimental feature associated with excessive activation, terminal differentiation, and exhaustion. 56 , 57 Others have noted that modest tonic signaling may improve CAR-T persistence and antitumor function. 58 , 59 In our study, we maintained a consistent anti-CD20 scFv, allowing for us to attribute the tonic signaling primarily to intracellular domains. Notably, the tonic signaling we observed in MyD88-CD40 CARs did not lead to exhaustion; instead, these CARs displayed a robust resistance to exhaustion. Although they exhibited antigen-independent proliferation, MyD88-CD40 CARs did not display antigen-independent cytotoxicity or cytokine release, suggesting that tonic signaling promoted expansion without triggering full effector function.

The unique cytokine profile associated with MyD88-CD40 CAR-Ts offers key insights into their mechanisms of action and implications for next-generation CAR-T therapies. IL-6 is a hallmark of CAR-T expansion and CRS observed in patients and in previous NHP studies with 4-1BB CARs. 38 , 39 , 60 Interestingly, MyD88-CD40 CAR-T expansion was not associated with elevated peripheral IL-6, potentially due to lower systemic IFN-γ and tumor necrosis factor α, which activate myeloid cells, the primary drivers of IL-6 during CAR-T expansion. 61 As novel CAR designs move toward the clinic, our findings highlight the need to consider domain-specific inflammatory profiles. MyD88-CD40 CAR-Ts secreted significantly reduced IL-10 ex vivo. Given IL-10’s potent anti-inflammatory properties, 62 its diminished secretion may have facilitated the robust MyD88-CD40 expansion observed in vivo. These results align with emerging interest in combining CAR-Ts with IL-10 blockade, which has been shown to reinvigorate CAR-T activity in preclinical models. 63 , 64 Interestingly, MyD88-CD40 CAR-Ts secreted significantly increased IL-13 ex vivo, a cytokine associated with Th2-mediated immunity and allergic inflammation. 59 A similar IL-13–dominant type II signature was recently reported in clinical CAR-T products from long-term responders. 65 Together, these findings point to a potential role for IL-13 in promoting durable responses and suggest that IL-13 secretion could be harnessed in armored CAR-T designs to enhance persistence.

Our pooled ex vivo stimulation assays, which were intended to simulate the pooled in vivo competition assay, revealed superior expansion of CD28 and 4-1BB CARs in a donor-specific manner, as well as almost immediate depletion of MyD88-CD40 CARs. Importantly, these results were not replicated in vivo, in which we saw striking expansion of MyD88-CD40 CAR-Ts and near-complete absence of CD28 CAR-Ts. This discrepancy may potentially be attributed to paracrine signaling factors and resource competition unique to the ex vivo pooled setting, which may impose selective pressures that amplify fitness differences between CAR designs. Our findings caution against overreliance on artificial ex vivo pooled screens and emphasize the importance of evaluating CAR-T function in physiologically relevant preclinical models.

We also validated the value of the NHP model in identifying acute toxicities, including those associated with MyD88-CD40. In our study, 1 NHP experienced uncontrolled CAR-T expansion with clinical features of CRS and ICANS, including tremors, altered consciousness, and seizure activity. This animal had the highest baseline B-cell levels, suggesting that elevated antigen burden may have driven excessive CAR-T expansion and toxicity. Similar toxicities have been observed clinically. In an early trial evaluating prostate stem cell antigen-specific CAR-Ts incorporating an inducible MyD88-CD40 switch for prostate and pancreatic cancer, 2 treatment-related deaths and other dose-limiting toxicities, including grade 4 CRS and ICANS, led to trial discontinuation, despite evidence of biological activity. 29 Although most preclinical studies using MyD88-CD40 CARs have not reported significant toxicity, 1 murine study described cachexia and inflammation after treatment, which was resolved by incorporating an inducible caspase-9 safety switch to deplete CAR-Ts. 27 Collectively, our results underscore the importance of immunocompetent NHP models in recapitulating serious CAR-mediated toxicities observed in the clinic and highlight the potent and inflammatory nature of MyD88-CD40 CARs. Future studies should focus on optimizing dosing and developing strategies to balance MyD88-CD40 potency with safety, such as incorporating cis-acting safety switches. 27

In summary, our study highlights the power of the NHP model as a platform for pooled screening of CAR designs, enabling the identification of superior architectures such as MyD88-CD40. The unique features associated with MyD88-CD40, including hyperactivation, tonic signaling, distinct cytokine secretion, and resistance to exhaustion, highlight key attributes that drive MyD88-CD40 potency and provide valuable insights that could be leveraged in the design of next-generation therapies.

Conflict-of-interest disclosure: H.-P.K. reports consultancy fees from and ownership interests in Rocket Pharmaceuticals, Homology Medicines, Vor Biopharma, and Ensoma, Inc; and is a member of the scientific advisory board of Umoja Biopharma. The remaining authors declare no competing financial interests.

Acknowledgments

The authors thank Helen Crawford for assistance in preparing the manuscript. All nonhuman primate–related work was conducted at the Washington National Primate Research Center (WaNPRC), with additional support from Solomon Wangari, Britni Curtis, Joel Ahrens, and Naoto Iwayama for tissue collection and laparoscopic procedures. The authors also thank the Fred Hutchinson Cancer Center Vector Production core, including Logan Hargis, Zach Burger, and Megha Gupta, for their work on lentiviral vector production; the WaNPRC Virology and Immunology core, led by Sandra Dross, for cell subset analysis; and the University of Washington Laboratory Medicine Research Testing Service, including Chihiro Morishima and Jacquie Chee. The authors are grateful to Michael Jensen, James Riley, and Brian Till for providing critical reagents.

This work was supported by grants from the National Institute of Health (NIH)/National Institute of Allergy and Infectious Diseases (R01 AI167004 and R01 AI170214 [C.W.P.], UM1 AI164570 [F.D.B., H.-P.K., and C.W.P.], and U19 HL156247 [H.-P.K.]), NIH/Office of Research Infrastructure Programs (P51 OD010425 and U42 OD011123 [WaNPRC]). This work used shared resources from the Fred Hutchinson Cancer Center/Seattle Children’s Cancer Consortium (P30 CA015704) including the Flow Cytometry Core Facility (RRID:SCR_022613), Experimental Histopathology Core Facility (RRID:SCR_022612), Immune Monitoring Core Facility (RRID:SCR_022615), and the Cell Manipulation Tools Core Vector Production (National Institute of Diabetes and Digestive and Kidney Disease Cooperative Center of Excellence in Hematology U54 DK106829 [H.-P.K.]). H.-P.K. also received support from the Fred Hutchinson Stephanus Endowed Chair for Cell and Gene Therapy.

Authorship

Contribution: L.H.M., H.-P.K., K.R.J., and C.W.P. conceived and designed the study; L.H.M. manufactured the chimeric antigen receptor T-cell (CAR-T) products; L.H.M., E.J.C., and S.M.D. performed functional assays on CAR-T products; V.N., M.H., S.H., C.L., K.C., C.W., and E.W. provided care for nonhuman primates (NHPs); R.D.M. performed necropsy analyses on NHPs; L.H.M., E.J.C., S.M.D., and T.E. collected and processed longitudinal samples and performed flow cytometry assays, which were analyzed by L.H.M.; L.H.M. extracted genomic DNA; H.Z. and L.S. performed droplet digital polymerase chain reaction assays, which were analyzed by L.H.M.; C.E.S. performed and analyzed immunohistochemistry; C.B., J.K.E., A.M.R., and F.D.B. performed and analyzed integration site analyses; L.H.M. and C.W.P. wrote the manuscript, which was reviewed by all authors.

Footnotes

All data supporting the findings of this study are available within the article and its supplemental files, or upon request from the corresponding authors, Hans-Peter Kiem ( hkiem@fredhutch.org ) and Christopher W. Peterson ( cwpeters@fredhutch.org ).

The online version of this article contains a data supplement.

There is a Blood Commentary on this article in this issue.

The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Contributor Information

Hans-Peter Kiem, Email: hkiem@fredhutch.org.

Christopher W. Peterson, Email: cwpeters@fredhutch.org.

Supplementary Material

Supplemental Methods, References, Tables, and Figures

References

1. Neelapu SS, Locke FL, Bartlett NL, et al. Axicabtagene ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma. N Engl J Med. 2017;377(26):2531–2544. doi: 10.1056/NEJMoa1707447.

2. Schuster SJ, Bishop MR, Tam CS, et al. Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma. N Engl J Med. 2019;380(1):45–56. doi: 10.1056/NEJMoa1804980.

3. Locke FL, Miklos DB, Jacobson CA, et al. Axicabtagene ciloleucel as second-line therapy for large B-cell lymphoma. N Engl J Med. 2022;386(7):640–654. doi: 10.1056/NEJMoa2116133.

4. Maude SL, Laetsch TW, Buechner J, et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N Engl J Med. 2018;378(5):439–448. doi: 10.1056/NEJMoa1709866.

5. Shah BD, Ghobadi A, Oluwole OO, et al. KTE-X19 for relapsed or refractory adult B-cell acute lymphoblastic leukaemia: phase 2 results of the single-arm, open-label, multicentre ZUMA-3 study. Lancet. 2021;398(10299):491–502. doi: 10.1016/S0140-6736(21)01222-8.

6. Roddie C, Sandhu KS, Tholouli E, et al. Obecabtagene autoleucel in adults with B-cell acute lymphoblastic leukemia. N Engl J Med. 2024;391(23):2219–2230. doi: 10.1056/NEJMoa2406526.

7. Munshi NC, Anderson LD, Jr., Shah N, et al. Idecabtagene vicleucel in relapsed and refractory multiple myeloma. N Engl J Med. 2021;384(8):705–716. doi: 10.1056/NEJMoa2024850.

8. Rodriguez-Otero P, Ailawadhi S, Arnulf B, et al. Ide-cel or standard regimens in relapsed and refractory multiple myeloma. N Engl J Med. 2023;388(11):1002–1014. doi: 10.1056/NEJMoa2213614.

9. Shah NN, Fry TJ. Mechanisms of resistance to CAR T cell therapy. Nat Rev Clin Oncol. 2019;16(6):372–385. doi: 10.1038/s41571-019-0184-6.

10. Gardner RA, Finney O, Annesley C, et al. Intent-to-treat leukemia remission by CD19 CAR T cells of defined formulation and dose in children and young adults. Blood. 2017;129(25):3322–3331. doi: 10.1182/blood-2017-02-769208.

11. Myers RM, Li Y, Barz Leahy A, et al. Humanized CD19-targeted chimeric antigen receptor (CAR) T cells in CAR-naive and CAR-exposed children and young adults with relapsed or refractory acute lymphoblastic leukemia. J Clin Oncol. 2021;39(27):3044–3055. doi: 10.1200/JCO.20.03458.

12. Lamble AJ, Myers RM, Taraseviciute A, et al. Preinfusion factors impacting relapse immunophenotype following CD19 CAR T cells. Blood Adv. 2023;7(4):575–585. doi: 10.1182/bloodadvances.2022007423.

13. Schultz LM, Eaton A, Baggott C, et al. Outcomes after nonresponse and relapse post-tisagenlecleucel in children, adolescents, and young adults with B-cell acute lymphoblastic leukemia. J Clin Oncol. 2023;41(2):354–363. doi: 10.1200/JCO.22.01076.

14. Brudno JN, Maric I, Hartman SD, et al. T cells genetically modified to express an anti-B-cell maturation antigen chimeric antigen receptor cause remissions of poor-prognosis relapsed multiple myeloma. J Clin Oncol. 2018;36(22):2267–2280. doi: 10.1200/JCO.2018.77.8084.

15. Shi M, Wang J, Huang H, et al. Bispecific CAR T cell therapy targeting BCMA and CD19 in relapsed/refractory multiple myeloma: a phase I/II trial. Nat Commun. 2024;15(1):3371. doi: 10.1038/s41467-024-47801-8.

16. Wittibschlager V, Bacher U, Seipel K, et al. CAR T-cell persistence correlates with improved outcome in patients with B-cell lymphoma. Int J Mol Sci. 2023;24(6) doi: 10.3390/ijms24065688.

17. Shiqi L, Jiasi Z, Lvzhe C, et al. Durable remission related to CAR-T persistence in R/R B-ALL and long-term persistence potential of prime CAR-T. Mol Ther Oncolytics. 2023;29:107–117. doi: 10.1016/j.omto.2023.04.003.

18. Jayaraman J, Mellody MP, Hou AJ, et al. CAR-T design: elements and their synergistic function. EBioMedicine. 2020;58 doi: 10.1016/j.ebiom.2020.102931.

19. Cappell KM, Kochenderfer JN. A comparison of chimeric antigen receptors containing CD28 versus 4-1BB costimulatory domains. Nat Rev Clin Oncol. 2021;18(11):715–727. doi: 10.1038/s41571-021-00530-z.

20. Selli ME, Landmann JH, Terekhova M, et al. Costimulatory domains direct distinct fates of CAR-driven T-cell dysfunction. Blood. 2023;141(26):3153–3165. doi: 10.1182/blood.2023020100.

21. Ying Z, He T, Wang X, et al. Parallel comparison of 4-1BB or CD28 co-stimulated CD19-targeted CAR-T cells for B cell non-Hodgkin's lymphoma. Mol Ther Oncolytics. 2019;15:60–68. doi: 10.1016/j.omto.2019.08.002.

22. Zhao X, Yang J, Zhang X, et al. Efficacy and safety of CD28- or 4-1BB-based CD19 CAR-T cells in B cell acute lymphoblastic leukemia. Mol Ther Oncolytics. 2020;18:272–281. doi: 10.1016/j.omto.2020.06.016.

23. Guercio M, Orlando D, Di Cecca S, et al. CD28.OX40 co-stimulatory combination is associated with long in vivo persistence and high activity of CAR.CD30 T-cells. Haematologica. 2021;106(4):987–999. doi: 10.3324/haematol.2019.231183.

24. Zhang H, Li F, Cao J, et al. A chimeric antigen receptor with antigen-independent OX40 signaling mediates potent antitumor activity. Sci Transl Med. 2021;13(578) doi: 10.1126/scitranslmed.aba7308.

25. Tan J, Jia Y, Zhou M, et al. Chimeric antigen receptors containing the OX40 signalling domain enhance the persistence of T cells even under repeated stimulation with multiple myeloma target cells. J Hematol Oncol. 2022;15(1):39. doi: 10.1186/s13045-022-01244-0.

26. Goodman DB, Azimi CS, Kearns K, et al. Pooled screening of CAR T cells identifies diverse immune signaling domains for next-generation immunotherapies. Sci Transl Med. 2022;14(670) doi: 10.1126/scitranslmed.abm1463.

27. Collinson-Pautz MR, Chang WC, Lu A, et al. Constitutively active MyD88/CD40 costimulation enhances expansion and efficacy of chimeric antigen receptor T cells targeting hematological malignancies. Leukemia. 2019;33(9):2195–2207. doi: 10.1038/s41375-019-0417-9.

28. Prinzing B, Schreiner P, Bell M, Fan Y, Krenciute G, Gottschalk S. MyD88/CD40 signaling retains CAR T cells in a less differentiated state. JCI Insight. 2020;5(21) doi: 10.1172/jci.insight.136093.

29. Stein MN, Dumbrava EE, Teply BA, et al. PSCA-targeted BPX-601 CAR T cells with pharmacological activation by rimiducid in metastatic pancreatic and prostate cancer: a phase 1 dose escalation trial. Nat Commun. 2024;15(1) doi: 10.1038/s41467-024-53220-6.

30. Hudecek M, Sommermeyer D, Kosasih PL, et al. The nonsignaling extracellular spacer domain of chimeric antigen receptors is decisive for in vivo antitumor activity. Cancer Immunol Res. 2015;3(2):125–135. doi: 10.1158/2326-6066.CIR-14-0127.

31. Jonnalagadda M, Mardiros A, Urak R, et al. Chimeric antigen receptors with mutated IgG4 Fc spacer avoid fc receptor binding and improve T cell persistence and antitumor efficacy. Mol Ther. 2015;23(4):757–768. doi: 10.1038/mt.2014.208.

32. Majzner RG, Rietberg SP, Sotillo E, et al. Tuning the antigen density requirement for CAR T-cell activity. Cancer Discov. 2020;10(5):702–723. doi: 10.1158/2159-8290.CD-19-0945.

33. Fujiwara K, Tsunei A, Kusabuka H, Ogaki E, Tachibana M, Okada N. Hinge and transmembrane domains of chimeric antigen receptor regulate receptor expression and signaling threshold. Cells. 2020;9(5) doi: 10.3390/cells9051182.

34. Li N, Quan A, Li D, et al. The IgG4 hinge with CD28 transmembrane domain improves V(H)H-based CAR T cells targeting a membrane-distal epitope of GPC1 in pancreatic cancer. Nat Commun. 2023;14(1):1986. doi: 10.1038/s41467-023-37616-4.

35. Seok J, Warren HS, Cuenca AG, et al. Genomic responses in mouse models poorly mimic human inflammatory diseases. Proc Natl Acad Sci U S A. 2013;110(9):3507–3512. doi: 10.1073/pnas.1222878110.

36. Parekh C, Crooks GM. Critical differences in hematopoiesis and lymphoid development between humans and mice. J Clin Immunol. 2013;33(4):711–715. doi: 10.1007/s10875-012-9844-3.

37. Maynard LH, Humbert O, Peterson CW, Kiem HP. Genome editing in large animal models. Mol Ther. 2021;29(11):3140–3152. doi: 10.1016/j.ymthe.2021.09.026.

38. Taraseviciute A, Tkachev V, Ponce R, et al. Chimeric antigen receptor T cell-mediated neurotoxicity in nonhuman primates. Cancer Discov. 2018;8(6):750–763. doi: 10.1158/2159-8290.CD-17-1368.

39. Bui JK, Starke CE, Poole NH, et al. CD20 CAR T cells safely and reversibly ablate B cell follicles in a non-human primate model of HIV persistence. Mol Ther. 2024;32(5):1238–1251. doi: 10.1016/j.ymthe.2024.02.030.

40. Nicolai CJ, Parker MH, Qin J, et al. In vivo CAR T-cell generation in nonhuman primates using lentiviral vectors displaying a multidomain fusion ligand. Blood. 2024;144(9):977–987. doi: 10.1182/blood.2024024523.

41. Kaminski J, Fleming RA, Alvarez-Calderon F, et al. B-cell-directed CAR T-cell therapy activates CD8+ cytotoxic CARneg bystander T cells in patients and nonhuman primates. Blood. 2024;144(1):46–60. doi: 10.1182/blood.2023022717.

42. Suhoski MM, Golovina TN, Aqui NA, et al. Engineering artificial antigen-presenting cells to express a diverse array of co-stimulatory molecules. Mol Ther. 2007;15(5):981–988. doi: 10.1038/mt.sj.6300134.

43. Levy CN, Hughes SM, Roychoudhury P, et al. A highly multiplexed droplet digital PCR assay to measure the intact HIV-1 proviral reservoir. Cell Rep Med. 2021;2(4) doi: 10.1016/j.xcrm.2021.100243.

44. Zoutman WH, Nell RJ, Versluis M, et al. Accurate quantification of T cells by measuring loss of germline T-cell receptor loci with generic single duplex droplet digital PCR assays. J Mol Diagn. 2017;19(2):236–243. doi: 10.1016/j.jmoldx.2016.10.006.

45. Wang X, Chang WC, Wong CW, et al. A transgene-encoded cell surface polypeptide for selection, in vivo tracking, and ablation of engineered cells. Blood. 2011;118(5):1255–1263. doi: 10.1182/blood-2011-02-337360.

46. Rea IM, McNerlan SE, Alexander HD. CD69, CD25, and HLA-DR activation antigen expression on CD3+ lymphocytes and relationship to serum TNF-alpha, IFN-gamma, and sIL-2R levels in aging. Exp Gerontol. 1999;34(1):79–93. doi: 10.1016/s0531-5565(98)00058-8.

47. Swiech K, Malmegrim KCR, Picanço-Castro V. 1st ed. Humana; 2020. Chimeric Antigen Receptor T Cells - Development and Production.

48. Rosskopf S, Leitner J, Paster W, et al. A Jurkat 76 based triple parameter reporter system to evaluate TCR functions and adoptive T cell strategies. Oncotarget. 2018;9(25):17608–17619. doi: 10.18632/oncotarget.24807.

49. Fraietta JA, Lacey SF, Orlando EJ, et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat Med. 2018;24(5):563–571. doi: 10.1038/s41591-018-0010-1.

50. Deng Q, Han G, Puebla-Osorio N, et al. Characteristics of anti-CD19 CAR T cell infusion products associated with efficacy and toxicity in patients with large B cell lymphomas. Nat Med. 2020;26(12):1878–1887. doi: 10.1038/s41591-020-1061-7.

51. Kouro T, Himuro H, Sasada T. Exhaustion of CAR T cells: potential causes and solutions. J Transl Med. 2022;20(1):239. doi: 10.1186/s12967-022-03442-3.

52. Selli ME, Landmann JH, Arveseth C, Singh N. Inducing T cell dysfunction by chronic stimulation of CAR-engineered T cells targeting cancer cells in suspension cultures. STAR Protoc. 2023;4(1) doi: 10.1016/j.xpro.2022.101954.

53. Salter AI, Ivey RG, Kennedy JJ, et al. Phosphoproteomic analysis of chimeric antigen receptor signaling reveals kinetic and quantitative differences that affect cell function. Sci Signal. 2018;11(544) doi: 10.1126/scisignal.aat6753.

54. Maldini CR, Ellis GI, Riley JL. CAR T cells for infection, autoimmunity and allotransplantation. Nat Rev Immunol. 2018;18(10):605–616. doi: 10.1038/s41577-018-0042-2.

55. Perez CR, Garmilla A, Nilsson A, et al. Library-based single-cell analysis of CAR signaling reveals drivers of in vivo persistence. Cell Syst. 2025;16(5) doi: 10.1016/j.cels.2025.101260.

56. Long AH, Haso WM, Shern JF, et al. 4-1BB costimulation ameliorates T cell exhaustion induced by tonic signaling of chimeric antigen receptors. Nat Med. 2015;21(6):581–590. doi: 10.1038/nm.3838.

57. Landoni E, Fucá G, Wang J, et al. Modifications to the framework regions eliminate chimeric antigen receptor tonic signaling. Cancer Immunol Res. 2021;9(4):441–453. doi: 10.1158/2326-6066.CIR-20-0451.

58. Singh N, Frey NV, Engels B, et al. Antigen-independent activation enhances the efficacy of 4-1BB-costimulated CD22 CAR T cells. Nat Med. 2021;27(5):842–850. doi: 10.1038/s41591-021-01326-5.

59. Chen J, Qiu S, Li W, et al. Tuning charge density of chimeric antigen receptor optimizes tonic signaling and CAR-T cell fitness. Cell Res. 2023;33(5):341–354. doi: 10.1038/s41422-023-00789-0.

60. Cosenza M, Sacchi S, Pozzi S. Cytokine release syndrome associated with T-cell-based therapies for hematological malignancies: pathophysiology, clinical presentation, and treatment. Int J Mol Sci. 2021;22(14) doi: 10.3390/ijms22147652.

61. Norelli M, Camisa B, Barbiera G, et al. Monocyte-derived IL-1 and IL-6 are differentially required for cytokine-release syndrome and neurotoxicity due to CAR T cells. Nat Med. 2018;24(6):739–748. doi: 10.1038/s41591-018-0036-4.

62. Carlini V, Noonan DM, Abdalalem E, et al. The multifaceted nature of IL-10: regulation, role in immunological homeostasis and its relevance to cancer, COVID-19 and post-COVID conditions. Front Immunol. 2023;14 doi: 10.3389/fimmu.2023.1161067.

63. Batchu RB, Gruzdyn OV, Mahmud EM, et al. Inhibition of interleukin-10 in the tumor microenvironment can restore mesothelin chimeric antigen receptor T cell activity in pancreatic cancer in vitro. Surgery. 2018;163(3):627–632. doi: 10.1016/j.surg.2017.10.056.

64. Sullivan KM, Jiang X, Guha P, et al. Blockade of interleukin 10 potentiates antitumour immune function in human colorectal cancer liver metastases. Gut. 2023;72(2):325–337. doi: 10.1136/gutjnl-2021-325808.

65. Bai Z, Feng B, McClory SE, et al. Single-cell CAR T atlas reveals type 2 function in 8-year leukaemia remission. Nature. 2024;634(8034):702–711. doi: 10.1038/s41586-024-07762-w.

Associated Data

Supplementary Materials

Supplemental Methods, References, Tables, and Figures