Inclusion of 4-1BB Costimulation Enhances Selectivity and Functionality of IL13Rα2-Targeted Chimeric Antigen Receptor T Cells

Chimeric antigen receptor (CAR) T cell immunotherapy is emerging as a powerful strategy for cancer therapy; however, an important safety consideration is the potential for off-tumor recognition of normal tissue. This is particularly important as ligand-based CARs are optimized for clinical translation. Our group has developed and clinically translated an IL13(E12Y) ligand–based CAR targeting the cancer antigen IL13Rα2 for treatment of glioblastoma (GBM). There remains limited understanding of how IL13-ligand CAR design impacts the activity and selectivity for the intended tumor-associated target IL13Rα2 versus the more ubiquitous unintended target IL13Rα1. In this study, we functionally compared IL13(E12Y)-CARs incorporating different intracellular signaling domains, including first-generation CD3ζ-containing CARs (IL13ζ), second-generation 4-1BB (CD137)–containing or CD28-containing CARs (IL13-BBζ or IL13-28ζ), and third-generation CARs containing both 4-1BB and CD28 (IL13-28BBζ). In vitro coculture assays at high tumor burden establish that second-generation IL13-BBζ or IL13-28ζ outperform first-generation IL13ζ and third-generation IL13-28BBζ CAR designs, with IL13-BBζ providing superior CAR proliferation and in vivo antitumor potency in human xenograft mouse models. IL13-28ζ displayed a lower threshold for antigen recognition, resulting in higher off-target IL13Rα1 reactivity both in vitro and in vivo. Syngeneic mouse models of GBM also demonstrate safety and antitumor potency of murine IL13-BBζ CAR T cells delivered systemically after lymphodepletion. These findings support the use of IL13-BBζ CARs for greater selective recognition of IL13Rα2 over IL13Rα1, higher proliferative potential, and superior antitumor responsiveness. This study exemplifies the potential of modulating factors outside the antigen targeting domain of a CAR to improve selective tumor recognition. Significance: This study reveals how modulating CAR design outside the antigen targeting domain improves selective tumor recognition. Specifically, this work shows improved specificity, persistence, and efficacy of 4-1BB–based IL13-ligand CARs. Human clinical trials evaluating IL13-41BB-CAR T cells are ongoing, supporting the clinical significance of these findings.


Introduction
Chimeric antigen receptor (CAR) T cell therapy has achieved striking efficacy in the treatment of relapsed and refractory CD19 + B-cell malignancies (1,2), and there is tremendous interest in extending these gains into other tumor types, solid tumors in particular. An evolving area of study is the impact of CAR design -for example, affinity optimization, spacer selection, or choice 4-1BB Enhances IL13-ligand CAR Selectivity and Functionality on normal brain tissue (3)(4)(5). Our CAR designs incorporate IL13, the natural ligand for IL13Rα2, as the antigen-binding domain, with an E12Y engineered mutation that improves selectivity of the CAR for IL13Rα2 over IL13Rα1 (6)(7)(8), which is expressed more frequently on normal tissues. In addition, based on preclinical and clinical studies, we have incorporated numerous other construct modifications to optimize clinical and biological activity. These include mutations in the hinge and spacer domains that reduce interactions with Fc gamma receptors (9), as well changes in the cytoplasmic endodomain from a first-generation CD3ζ -CAR [IL13-ζ ; containing immunoreceptor tyrosinebased activation motifs (ITAMs)] to a second-generation 4-1BB-containing CAR (IL13-BBζ ; refs. 6, [10][11][12]. Finally, we have also evaluated route of delivery as an important contextual modification that greatly improves CAR T cell efficacy (6). Taken together, this body of work shows that second-generation IL13-BBζ CAR T cells are superior to first-generation CAR T cells in controlling glioblastoma (GBM) xenotransplants in mice (6), and that intraventricular delivery of IL13-BBζ CAR T cells provides improved control of multifocal disease as compared with intratumoral or intravenous delivery (6). These studies provide the foundation for several phase I clinical trials for refractory or recurrent brain tumors (NCT02208362, NCT03389230, NCT04003649, NCT04214392, NCT04510051), in which second-generation 4-1BB-containing CAR T cells are delivered intracranially to patients.
Expanding on this work, herein we investigate the molecular underpinnings of important clinical observations of the IL13-BBζ CAR constructs and identify further contextual modifications that improve IL13Rα2-CAR function in vivo. Specifically, we describe important functional differences between first-, second-, and third-generation IL13Rα2-targeted CAR constructs, identifying second-generation CARs as superior to both first-and third-generation constructs in vitro and in vivo. We show that IL13-28ζ CAR T cells are more likely than IL13-BBζ CAR T cells to cause off-target toxicity through IL13Rα1 recognition, and that IL13-BBζ CAR T cells are more effective than IL13-28ζ CAR T cells at the low effector-to-target (E:T) ratios reflective of the in vivo setting. We use protein-focused techniques to compare signaling upon antigen stimulation between these second-generation CAR constructs and find that IL13-BBζ CAR T cells activate the noncanonical NFκB pathway more effectively than IL13-28ζ CAR T cells, particularly in the setting of high antigen concentration. Finally, we show that immunocompetent syngeneic mice treated with systemic administration of IL13-BBζ CAR T cells after lymphodepleting irradiation exhibit no evidence of off-tumor toxicity and mediate potent antitumor activity in murine orthotopic glioma models. Excitingly, lymphodepleted mice that successfully clear disease after CAR T cell treatment are subsequently able to survive rechallenge with antigen-negative tumor, indicating that they develop immunologic memory to other tumor-associated antigens. These findings, taken together, indicate that IL13-BBζ CAR constructs are more suited for further clinical development than IL13-28ζ CAR constructs, and justify further clinical trials to evaluate the effects of contextual modifications such as lymphodepletion on IL13-BBζ CAR T cell efficacy.

Generation of CAR T Cells
Peripheral blood mononuclear cells (PBMC) were collected from discard apheresis kits of 5 different human donors as approved by City of Hope (COH) Internal Review Board (IRB) oversight. Central memory T cells (Tcm) were then selected from the PBMC by magnetically depleting CD14 + and CD25 + cells followed by magnetic selection of the CD62L + population, transduced with lentivirus to express either of the four CAR variants (Fig. 1A), and expanded with IL2 and IL15, all carried out as described previously (6). The CD3/28 stimulation beads (Thermo Fisher Scientific, catalog no. 11141D) were removed on day 7, successfully transduced cells were enriched on the basis of CD19t expression on days 11-15 using EasySep Human CD19 Positive Selection Kit II (Stemcell Technologies, Inc., catalog no. 17854), and the resulting CAR+ T cells were cryopreserved at day 15 or 17. Unless otherwise indicated, thawed cells were then rested overnight in media with IL2/IL15 prior to analysis or use in the assays.
Murine IL13-BBζ T cells were generated as described previously (13). Briefly, T cells isolated from mouse spleens, transduced with retrovirus to express the CAR, and expanded with IL2 and IL7. Before their use in in vivo experiments, the CD3/28 stimulation beads (Invitrogen, catalog no. 11452D) were magnetically separated from the IL13-BBζ T cells, and CAR expression was determined by flow cytometry.
All cell banks (including CAR T cells) were authenticated for the desired antigen and/or marker expression by flow cytometry and tested for Mycoplasma using the MycoAlert PLUS Mycoplasma Detecting Kit (Lonza Walkersville, Inc.), and upon thaw, maintained in culture ≤ 3 months.

AACRJournals.org
Cancer Res Commun; 3(1) January 2023 Proliferation assays used the CellTrace carboxyfluorescein diacetate succinimidyl ester (CFSE) Cell Proliferation kit (Invitrogen) to stain the T cells prior to their coculture with tumor cells at a 1:1 E:T ratio for 4 days. Cells were then stained with antibodies specific for CD45 and CD19 prior to flow cytometric analysis.
Rechallenge assays were carried out as described previously (16), with cells analyzed by flow cytometry using DAPI and antibodies specific for IL13, CD45, CD4, and CD8 (Thermo Fisher Scientific, catalog no. BDB348793) at each timepoint.
Flow cytometric evaluation of intracellular cleaved caspase-3 was carried out using the FITC Active Caspase-3 Kit (BD Biosciences) and CD19 staining after 24-hour coculture at a 1:4 E:T ratio.
For the xCELLigence-based cytotoxicity assays, PBT138 cells expressing either low or high IL13Rα2 were first allowed to settle on the xCELLigence E-plate VIEW 96 PET (ACEA Biosciences) for 30 minutes at room temperature. The plate was then placed on the xCELLigence RTCA MP instrument (ACEA Biosciences) within a 37°C incubator and run on auto for sweeps with 1-hour intervals overnight, during which the tumor cells attach onto the well bottom. Plate analysis was paused while T cells were added at a 1:10 CAR+ E:T ratio. After allowing cells to settle for 30 minutes at room temperature, the plate was placed back on the xCELLigence instrument within a 37°C incubator and analysis restarted with 1-hour interval sweeps for 4 more days. The cultures were stopped and electrical impedence data (i.e., adherent cell index data) were collected, with loss of impedence and/or cell index as a measure of cell death.

Bead-bound Receptor Stimulation Assay
Protein G Dynabeads (Invitrogen) were washed and resuspended in PBS with 0.02% Tween-20 (PBS-T) and loaded with recombinant human IL13Rα2-Fc chimera (R&D Systems, catalog no. 614INS; 2.1 μg per 3.9 × 10 7 beads) by incubation for 10 minutes at room temperature. After washing in PBS-T, the IL13Rα2-Fc-bound beads were resuspended at 3.90 × 10 7 beads/100 μL of warmed media. CAR+ T cells (Jurkat-derived or human donor-derived) were plated at 2 × 10 5 cells/well on a 96-well U-bottom plate with or without beads at a 195:1 bead-to-cell ratio and incubated for 16 (Jurkat T cells) or 12 (human donor-derived T cells) hours at 37°C. After incubation, T cells were harvested, washed with ice-cold PBS, and resuspended in ice-cold lysis buffer containing a working dilution of Halt Phosphatase Inhibitor Single-Use Cocktail (Thermo Fisher Scientific). After 30 minutes of incubation on ice, lysates were centrifuged at 17,200 × g for 20 minutes at 4°C, and supernatants were collected and either frozen at −80°C or immediately analyzed by Western blot analysis. In brief, after determining lysate protein concentration by Bradford protein assay, equal proportions of protein were combined with Laemmli buffer (Bio-Rad) and DTT (Sigma-Aldrich) and boiled at 95°C for 5 minutes.
Protein was loaded into a 7.5% TGX gel (Bio-Rad) using a Mini-PROTEAN Tetra Cell (Bio-Rad) and transferred to 0.2 μm nitrocellulose (Prometheus). Membranes were incubated in blocking buffer for 1 hour at room temperature, washed in TBS with 0.05% Tween-20 (TBST), and then incubated overnight at 4°C with 1:1,000 dilution of primary antibody-either anti-p52/p100 (Millipore Sigma, catalog no. 05361) or anti-β-Actin (Cell Signaling Technology, catalog no. 3700). Membranes were then washed and incubated in blocking buffer containing horseradish peroxidase-linked horse anti-mouse (1:5,000; Cell Signaling Technology, catalog no. 7076) for 45 minutes at room temperature. After washing with TBST, membranes were imaged on a ChemiDoc Imaging System (Bio-Rad) with SuperSignal Chemiluminescent Substrate (Thermo Fisher Scientific).
AACRJournals.org Cancer Res Commun; 3(1) January 2023 Supernatants were then evaluated for IFNγ levels using the Legend Max ELISA kit with precoated plates (human IFNγ; BioLegend). Surface phenotype of cells that had been removed from the well was determined by flow cytometry using fluorochrome-conjugated antibodies specific for 4-1BB or CD69.

Mouse Studies
All mouse experiments were approved by the City of Hope Institute Animal Care and Use Committee. For xenograft models, on day 0, either ffLuc+ Groups of mice were monitored for intracranial tumor engraftment by noninvasive optical imaging as described previously (8) using a Lago-X (Spectral Instruments Imaging), or for subcutaneous tumor size using calipers. Where indicated, survival was monitored with euthanasia applied according to the American Veterinary Medical Association Guidelines.

Statistical Analysis
Student t tests, ANOVA tests, and log-rank (Mantel-Cox) tests were used as indicated and respective P values are provided in each figure legend.

Data Availability
Data were generated by the authors and either included in the article or available on request.

In Vitro Effector Function of IL13(E12Y)-CAR Variants with Low Tumor Burden is Independent of Costimulatory Domain
We have previously described the preclinical generation and clinical evaluation  1B). In particular, the ratio of CAR to CD19t marker was markedly different between constructs (Fig. 1C) Fig. 1D). Each of the IL13(E12Y)-CAR T cell variants were also able to efficiently kill IL13Rα2-expressing tumor cells in a 48-hour coculture at E:T ratios of 1:4 ( Fig. 1E), although the third-generation IL13-28BBζ CAR displayed slightly reduced killing potency against the GBM line PBT030-2. In addition, all CAR variants robustly proliferated after 4 days of coculture with the HT1080-Rα2 and PBT030-2 stimulator lines compared with MOCK T-cell controls (Fig. 1F). IL13-28BBζ T cells, however, did not proliferate as well as the other CAR T cell variants, which may relate to the slightly suboptimal killing of PBT030-2 that had been observed with this construct (Fig. 1E). We also noted that IL13-28ζ appeared to direct higher levels of killing of the IL13Rα2-negative HT1080 cells, and these CAR T cells exhibited higher levels proliferation upon HT1080 stimulation ( Fig. 1E and F). Overall, these studies indicate that IL13Rα2-dependent activation and effector function of IL13(E12Y)-CARs was not strongly influenced by either CAR expression levels or the costimulatory domain.

IL13-BBζ CAR T Cells Outperform Other CAR Variants In Vitro at High Tumor Burden
It was surprising to us that initial in vitro studies showed little difference in effector activity of the IL13(E12Y)-CAR T cell variants (Fig. 1), even for IL13-ζ , which lacked costimulatory signaling. As a result, we next examined cocultures with PBT030-2 cells at E:T ratios that were decreased to a more challenging ratio of 1:20 to evaluate the recursive killing potential of CAR variants ( Fig. 2A-D). In this 14-day in vitro stress test, second-generation IL13-BBζ and IL13-28ζ outperformed both IL13-ζ and IL13-28BBζ , demonstrating the requirement for optimal costimulation for recursive killing potency. All IL13(E12Y)-CAR T cell variants showed some level of tumor killing at this low E:T ratio of 1:20, suggesting their capacity to recursively kill. IL13-BBζ CAR T cells, however, exhibited clear superiority with regards to tumor killing and CAR T cell persistence and proliferation, with significant numbers of CAR-positive T cells detected in cocultures after tumor cell elimination ( Fig. 2B and C). To further compare the recursive killing potential of IL13(E12Y)-CAR T cell variants, a tumor rechallenge assay was performed with the addition of GBM cells every other day  efficient tumor elimination and CAR T cell proliferation following repetitive tumor addition (Fig 2D).
To understand whether the differences in persistence may be due to CARmediated activation-induced cell death (AICD), we evaluated signaling in CAR T cells after antigen stimulation. In CAR T cells cocultured with PBT030-2 (E:T 1:4) for 24 hours, cleaved caspase-3 was most highly expressed in IL13-28BBζ CAR T cells, whereas IL13-BBζ CAR T cells had the lowest levels, suggesting that IL13-BBζ CAR T cells were the least inclined toward AICD (Fig. 3A). It has previously been shown that noncanonical NFκB signaling is upregulated in 4-1BB-containing CAR T cells, and that this leads to increased survival and proliferation (19). First using a Jurkat CAR T model system, we found that IL13-BBζ CAR T cells stimulated for 16 hours with antigen-coated beads upregulated noncanonical NFκB pathway signaling much more than IL13-28ζ CAR T cells as evidenced by an increase in p100 processing ( Fig. 3B and C). Similarly, in CAR T cells generated from a normal human donor, the noncanonical NFκB pathway activation was increased in IL13-BBζ CAR T cells relative to IL13-28ζ CAR T cells after 12 hours of stimulation ( Fig. 3D and E). Taken together, these findings suggest that the pronounced functional superiority of IL13-BBζ CAR T cells observed at low E:T ratios might be a result of improved CAR T cell proliferation and survival through augmented noncanonical NFκB signaling and consequent decreased caspase-3 activity.

IL13-28ζ CAR T Cell Activation is More Sensitive to Lower Levels of IL13Rα2 Antigen
To evaluate further the impact of antigen density on the functional activity of IL13(E12Y)-CAR T cells, we challenged CAR variants with plate-bound IL13Rα2 at different concentrations (Fig. 4). The IL13-BBζ CAR T cells, which functioned well at the low E:T ratios in long-term cocultures (Fig. 2), released IFNγ levels in a manner similar to the first-and third-generation variants (Fig. 4A). This contrasts with the IL13-28ζ CAR T cells, which produced detectable IFNγ at antigen concentrations that were 4-fold lower than that observed with any of the other variants (Fig. 4A). Analysis of the activation markers 4-1BB and CD69 also suggested that the IL13-28ζ CAR T cells had a lower IL13Rα2-induced activation threshold compared with the IL13-BBζ CAR T cells (Fig. 4B).
To understand how differences in antigen density impact tumor killing, we established a GBM line with low and high IL13Rα2 levels (Fig. 4C). At low antigen density, IL13-28ζ CAR T cells demonstrated more rapid and efficient tumor killing than IL13-BBζ CAR T cells (Fig. 4C). Consistent with our observations in Fig. 1, T cell killing was comparable at higher antigen density levels. Together, these results indicate that while IL13-BBζ CAR T cells have superior persistence and proliferation, IL13-28ζ CAR T cells may exhibit higher sensitivity to lower levels of IL13Rα2 antigen.

IL13-BBζ CAR T Cells Exhibit Superior IL13Rα2-directed Antitumor Efficacy In Vivo
We hypothesized that superior persistence and antitumor efficacy of IL13-BBζ CAR T cells observed at the low E:T ratios in long-term cocultures would translate to improved disease control in vivo. To test this, we established brain tumor xenografts using ffLuc-expressing PBT030-2, treated the mice intracranially with the IL13Rα2-CAR T cell variants, and then followed tumor size over time using biophotonic imaging (Fig. 5). While antitumor efficacy could be seen in mice treated with each of the variants, the most significant effects on tumor burden (Fig. 5B) and overall survival (Fig. 5C) were observed with the IL13-BBζ CAR T cells. Indeed, mice treated with IL13-BBζ CAR T cells exhibited long-term tumor-free survival in >60% of mice after 152 days (Fig. 5D).

IL13-28ζ CAR T Cells Exhibit Off-target Activity Against IL13Rα1
To further evaluate differences in these IL13Rα2-CAR T cell variants, we examined off-target effects toward IL13Rα1. IL13 binds both IL13Rα1 and IL13Rα2, and while the IL13(E12Y) mutein was originally designed to increase selectivity for IL13Rα2 (7), some groups have reported IL13Rα1 cross-reactivity in similar IL13(E12Y)-based CARs containing a CD28 costimulatory domain (20,21).

Systemic Delivery of Murine IL13-BBζ CAR T Cells for Treatment of GBM is Both Safe and Effective with Preconditioning Lymphodepletion
The above studies comparing human IL13(E12Y)-CAR design support the selective IL13Rα2 targeting of IL13-BBζ CAR T cells. Because in vitro coculture assays and xenograft mouse models likely do not reveal the full potential for off-tumor toxicities given the cross-species differences in IL13-ligand binding, we developed a fully murine IL13(E12Y)-CAR to further study the safety and specific tumor targeting of IL13-BBζ CAR T cells in a syngeneic, immunocompetent mouse brain tumor model ( Fig. 7A; ref. 13). C57BL/6 splenocytes were engineered to produce muIL13-BBζ CAR T cells that exhibited equivalent numbers of CD4 + and CD8 + subsets with mixed undifferentiated and differentiated  (24), and colorectal cancer (25), we set out to utilize this fully murine platform to preclinically assess the safety and efficacy of intravenously delivered muIL13-BBζ CAR T cells. Furthermore, because systemic intravenous administration of CAR T cells often incorporates preconditioning with lymphodepletion and exacerbates CAR T cell-mediated toxicities, we also wanted to evaluate the effect of lymphodepletion on both the safety and antitumor potency of the muIL13-BBζ CAR T cells in this model. Using irradiation, we were able to confirm lymphodepletion by flow cytometric analysis of the blood (Fig. 7B). Biophotonic imaging then revealed that the best antitumor efficacy was observed in mice treated with both irradiation and muIL13-BBζ CAR T cells (Fig. 7C). Overall survival was improved with muIL13-BBζ CAR T cells alone, but the tumors ultimately progressed, and complete cures were only observed upon treatment with both irradiation and muIL13-BBζ CAR T cells (Fig. 7D). Furthermore, when the cured mice were rechallenged with intracranial administration of parental KLuc cells (i.e., not expressing muIL13Rα2), 60% survival was again observed, demonstrating that intravenous administration of CAR T cells can promote endogenous antitumor immunity, even in the setting of prior lymphodepletion ( Fig. 7E). This is consistent with a study recently published by our group establishing that locoregional delivery of IL13-BBζ CAR T cells, in the absence of lymphodepletion, can reshape the tumor microenvironment and elicit endogenous antitumor immune responses (13). Importantly, throughout this study, the mice were monitored for any obvious signs of distress or general toxicity, and those treated with the muIL13-BBζ CAR T cells did not exhibit any signs of therapy-associated adverse events-they did not exhibit weight loss and were bright, alert, and reactive until affected by their tumor burden.

Discussion
Optimization of CAR T cell therapy for solid tumors faces many challenges; in particular, poor CAR T cell persistence, tumor antigen heterogeneity, antigen escape, and off-tumor targeting present barriers to both the safety and curative potential of CAR T cell therapy. It is increasingly clear that successful therapies will need to combine both rational receptor design and contextual modifications to the tumor microenvironment and immune system. This work with  cells exhibited IL13Rα-dependent degranulation, activation marker expression, cytotoxicity, proliferation, and in vivo antitumor activity, which is consistent with previous reports evaluating CD28-containing IL13-ligand CAR designs (21). Taken together, these findings argue against using IL13-28ζ for selective IL13Rα2-targeting of tumors in favor of IL13-BBζ CAR T cells.
Other studies have shown CD28-containing CARs to exhibit more robust signaling and lower thresholds of activation than CARs with 4-1BB domains. Indeed, similar to our findings, the effect of the costimulatory domain on differential recognition of low antigen density levels has been reported with CARs targeting CD19 (26,27), HER2 (28), PSCA (29), and ROR1 (26). This study extends these findings by showing that costimulatory domains can also impact recognition based on affinity differences between the CAR and its target antigens [i.e., the affinity of IL13(E12Y)-CAR for IL13Rα1 vs. IL13Rα2]. Specifically, the CD28 domain increased the off-target recognition of the low-affinity To further evaluate the safety of IL13-BBζ CAR T cells, we developed a fully murine IL13(E12Y)-CAR to better assess off-tumor toxicities in immunocompetent mouse models of GBM, and administered the CAR T cells systemically with and without lymphodepletion. Lymphodepletion is known to be required for optimal efficacy with hematologic malignancy-targeted CAR T cells (30,31) and adoptive therapy of other solid tumors (32); and augments the potency and toxicity profiles of CAR T cell therapy. We have previously shown, both preclinically and clinically, that locoregional delivery of IL13Rα2-CAR T cells was safe and effective at eliminating malignant brain tumors (6,10). We extend that work here, and, similar to a previous study by Suryadevara and colleagues using lymphodepletion (33), demonstrate that the use of lymphodepleting AACRJournals.org Cancer Res Commun; 3(1) January 2023 radiation before intravenous delivery of muIL13-BBζ CAR T cells also significantly enhances in vivo antitumor efficacy and survival, with complete cures only observed upon treatment with both irradiation and muIL13-BBζ CAR T cells. Importantly, systemic treatment with muIL13-BBζ CAR T cells did not result in any signs of therapy-associated adverse events. Furthermore, when the cured mice were rechallenged, prolonged survival was again observed, suggesting that the establishment of immunological memory in these mice was not compromised by lymphodepleting radiation. Our work contributes to a growing body of evidence supporting the use of preconditioning lymphodepletion as a valuable adjunct for solid tumor-directed CAR T cell therapies. In addition, these studies open the door to investigations combining lymphodepletion and locoregional CAR T cell delivery, studies which are currently ongoing in our group.
Ultimately, this study supports several general principles of CAR design, including improved specificity, persistence, and efficacy of 4-1BB-based second-generation CARs, in part, through increased noncanonical NFκB signaling and decreased caspase-3 activity. These differences are most pronounced at lower effector to target ratios, which may help to explain why 4-1BB-based CARs are often clinically superior to CD28-based CARs for solid tumors. In addition, these findings extend our previous work demonstrating that changes to the context of CAR T cell therapy substantially affect their efficacy; locoregional delivery and lymphodepletion independently augment the ability of IL13-BBζ CAR T cells to control tumors. Taken together, these data justify novel clinical trials that combine IL13-BBζ CAR T cells, locoregional delivery, and lymphodepletion for solid tumor therapy.