A Chimeric GM-CSF/IL18 Receptor to Sustain CAR T-cell Function

The adoptive transfer of chimeric antigen receptor (CAR) T cells has resulted in impressive clinical responses for a broad range of hematologic malignancies, including lymphoma, leukemia, and multiple myeloma (1). However, CAR T cells showed limited antitumor activity in patients with solid tumors (2), and their use faces many challenges, including identification of safe target antigens, improving homing to tumor sites, and enabling CAR T cells to expand and persist in the setting of chronic antigen exposure (2).

CAR T-cell activation and proliferation require a distinct set of signals, which consist of CAR-mediated antigen-specific CD3ζ activation and costimulation (e.g., CD28 or 4-1BB), resulting in cytokine production (1, 2). However, the ability of CAR T cells to produce cytokines is not sustained in repeat stimulation assays that mimic chronic antigen exposure, resulting in cell death after two or three rounds of antigen-specific stimulation (3). Several strategies are being explored to overcome this limitation. For example, CAR T cells have been genetically modified to express cytokines or delete negative regulators (2). In addition, we and others have shown that an inducible costimulatory molecule that activates MYD88, the central signaling hub for toll-like receptors and the IL1 and IL18 receptors (4), sustains CAR T-cell effector function in the setting of chronic antigen exposure (3).

T cells normally receive activation, costimulatory, and cytokine signals in a temporospatial manner. Although investigators have developed synthetic signaling circuits to mimic this by linking expression of a gene to T-cell activation (5), we wanted to develop an antigen-dependent autocrine loop with molecules that are expressed in human cells. We took advantage of GM-CSF, a cytokine that is invariably expressed after CAR T-cell activation (3), and designed a chimeric cytokine receptor (GM18) that activates MyD88. It consists of the extracellular domains of the α/b chains of the GM-CSF receptor and the transmembrane and signaling domains of the α/b chains of the IL18 receptor. We observed that expression of GM18 in CAR T cells strikingly enhanced their effector functions in the setting of chronic antigen exposure ex vivo. This translated into potent antitumor activity of CAR T cells targeting EPHA2 or HER2 in xenograft models.

To express GM18 in CAR T cells we generated a retroviral vector in which the chimeric α and β GM18 receptor chains are separated by a 2A sequence (Fig. 1A and B). To initially confirm the functionality of GM18, we transduced wild-type (WT) and MyD88 knockdown (KD) NFκB/AP-1 Ramos-Blue reporter cell lines and treated GM18-transduced or unmodified reporter cells with increasing concentrations of exogenous GM-CSF. Only GM18-transduced WT Ramos-Blue cells displayed NFκB activation in the presence of GM-CSF, demonstrating that GM18 is functional and that it signals through MyD88 (Fig. 1C).

In order to determine if GM18 improves CAR T-cell function, we initially focused on T cells genetically modified with a retroviral vector encoding an EPHA2-CD28.ζ.CAR, a 2A sequence, and truncated CD19 (EPHA2-CAR; ref. 6). EPHA2-CAR and EPHA2-CAR.GM18 T cells were generated by single transduction or co-transduction. Anti-CD19 was used to detect T cells transduced with the EPHA2-CAR vector, and CAR expression was confirmed with an anti-F(ab₂)′ showing no significant differences with regard to transduction efficiency (Supplementary Fig. S1A and S1B). Anti-CD116 (GM-CSFRα chain) was used to identify GM18-transduced T cells, because anti-CD131 (GM-CSFRβ chain) staining consistently gave a lower percentage of transduction (Supplementary Fig. S1C and S1D). Single transduction of T cells with the EPHA2-CAR vector resulted in a mean transduction efficacy of 80%, and co-transduction of T cells with the EPHA2-CAR and GM18 vectors resulted in a mean of 35.4% double-positive cells (Fig. 1D). Although the mean fluorescence intensity of GM18 varied among donors (Supplementary Fig. S1E), we sorted EPHA2-CAR and EPHA2-CAR.GM18 T cells to obtain >95% pure single- or double-transduced cell populations based on percentage positive cells for the majority of functional analyses, unless otherwise indicated. Transduced T cells contained a mixture of CD4⁺ and CD8⁺ T cells, and additional T-cell subset analysis revealed the presence of naïve, central memory, effector memory, and terminally differentiated T cells with no significant differences when GM18 was expressed (Supplementary Fig. S2A–S2C). To exclude the possibility that expression of GM18 induced nonspecific killing by CAR T cells, we performed an MTS assay using the EPHA2⁺ Ewing sarcoma cell line A673 (Supplementary Fig. S3) as target cells. T cells expressing a nonfunctional EPHA2-CAR (EPHA2-ΔCAR) served as controls (6). Although EPHA2-CAR and EPHA2-CAR.GM18 T cells induced robust killing, EPHA2-ΔCAR and EPHA2-ΔCAR.GM18 T cells did not (Fig. 1E). To demonstrate that activating GM18 by itself does not induce T-cell expansion, we cultured nontransduced (NT) and GM18 T cells with exogenous GM-CSF (10 or 100 ng/mL) or IL7/IL15. NT and GM18 T cells expanded only in the presence of IL7/IL15, indicating that GM18 does not induce nonspecific T-cell expansion (Supplementary Fig. S4A and S4B).

We next set out to compare the functionality of EPHA2-CAR and EPHA2-CAR.GM18 T cells using a sequential stimulation assay to mimic chronic antigen exposure. CAR T cells were stimulated every 7 days with A673 cells as long as they were viable and had killed all tumor cells by visual inspection. Before each stimulation, T cells were enumerated to determine expansion, and following 48 hours after the addition of fresh tumor cells an aliquot of culture media was collected to determine the concentration of type 1 helper T cells (T_H1) and type 1 cytotoxic T cells (T_C1; including GM-CSF, IFNγ, TNFα, and IL2); T_H2/T_C2 (including IL4, IL5, IL6, IL10, and IL13); and IL17 by multiplex analysis. EphA2-CAR T cells induced tumor killing and expanded after two or three stimulations before their effector function deteriorated. In contrast, EphA2-CAR.GM18 T cells killed and expanded for at least six stimulations (Fig. 2A). After the first stimulation, EphA2-CAR and EphA2-CAR.GM18 T cells produced high levels (>1,000 pg/mL) of GM-CSF, IFNγ, TNFα, IL13, and IL2 (only EPHA2-CAR.GM18); intermediate levels (100–1,000 pg/mL) of IL4, IL5, IL10, and IL2 (only EPHA2-CAR); and low or undetectable levels (<100 pg/mL) of IL6 and IL17α (Fig. 2B). Although EPHA2-CAR.GM18 T cells produced higher levels of cytokines, they only reached statistical significance for IL13. Cytokine expression was dependent on expression of a functional CAR, as EPHA2-ΔCAR and EphA2-ΔCAR.GM18 T cells only produced low or undetectable levels of cytokines (Fig. 2B). EPHA2-CAR.GM18 T cells continued to produce robust levels of cytokines after each stimulation. T_H1/T_C1 cytokine production decreased and T_H2/T_C2 increased between the first and sixth stimulations. However, these findings only reached statistical significance for IL5 and IL13 (Fig. 2C and D).

To provide further evidence of the observed benefit on expression of a functional GM18 receptor, we generated a nonfunctional GM18 receptor (ΔGM18) with no cytoplasmic signaling domains (Supplementary Fig. S5A). ΔGM18 did not activate the MYD88 signaling pathway, and, although it did not interfere with cytokine production of EPHA2-CAR T cells, it did not enhance their ability to expand in repeat stimulation assays (Supplementary Fig. S5B–S5E). Having a ΔGM18 receptor and ΔCAR also allowed us to address the question of whether expression of both molecules in T cells is required. CAR T cells were mixed at a ratio of 1:1 with CAR.GM18, CAR.ΔGM18, ΔCAR.GM18, or ΔCAR.ΔGM18 T cells. This admixture of T cells was then stimulated for 24 hours with recombinant protein and cultured for 7 days prior to performing FACS analysis to determine the percentage of the respective CAR T-cell populations. Although the percentage of CAR.GM18 and CAR.ΔGM18 T cells remained stable, there was a significant decline for ΔCAR.GM18 and ΔCAR.ΔGM18 T cells, demonstrating that CAR activation is critical and bystander activation very unlikely (Supplementary Fig. S6A and S6B). Finally, to explore whether the benefit of GM18 could be extended to CARs with a 4-1BB costimulatory domain, we transduced EPHA2–4-1BBζ.CAR T cells (CAR^BB T cells) with GM18 cells (Supplementary Fig. S7A). CAR^BB.GM18 T cells were functional as judged by cytokine production and had a significantly greater ability to expand than CAR^BB T cells in our repeat stimulation assay (Supplementary Fig. S7B–S7D).

To assess the in vivo antitumor activity of EphA2-CAR.GM18 T cells, we used the A673 Ewing sarcoma xenograft model. NOD-scid IL2Rgamma^null (NSG) mice were injected with 2 × 10⁶ A673 tumor cells subcutaneously (s.c.) and on day 7 received one single intravenous (i.v.) dose of 1 × 10⁵ or 3 × 10⁵ CAR T cells (Fig. 3A; Supplementary Fig. S8). In some of the experiments, CAR T cells were genetically modified to express GFP.ffLuc (Fig. 3A). Tumor growth was monitored by caliper measurements and CAR T-cell expansion and persistence by weekly bioluminescence imaging. EphA2-CAR T cells induced complete responses (CR) in 4/14 mice (29%) at the lower cell dose and in 6/9 mice (66%) at the higher cell dose (Fig. 3B and C; Supplementary Fig. S9). In contrast, EphA2-CAR.GM18 T cells induced CRs in 14/15 mice (93%) at the lower cell dose and in 10/10 mice (100%) at the higher cell dose (Fig. 3B and C). Tumor recurrences after an initial CR occurred only in mice that had received 1 × 10⁵ EphA2-CAR T cells (two out of four mice). EphA2-CAR.GM18 T cells induced a significant survival advantage at both evaluated cell doses in comparison with untreated controls. In contrast, EPHA2-CAR T-cell therapy improved survival only at the higher (3 × 10⁵) cell dose (Fig. 3D). CAR T-cell infused mice continued to gain weight (Supplementary Fig. S10), and in long-term survivors we did not observe clinical signs of GVHD except in 2/10 mice in the 3 × 10⁵ EPHA2-CAR.GM18 T-cell group, requiring euthanasia at 9 weeks after T-cell injection.

Improved antitumor activity of EphA2-CAR.GM18 T cells at a cell dose of 1 × 10⁵ per mouse correlated with a significantly greater peak expansion in comparison with EphA2-CAR T cells (Fig. 3E and F). There was no significant difference between EphA2-CAR and EphA2-CAR.GM18 T cells at the higher cell dose with regard to peak T-cell expansion and persistence (Fig. 3E and F). We observed increased expansion of EphA2-CAR.GM18 T cells at the lower cell dose. Because EphA2-CAR.GM18 T-cell proliferation depends on antigen density (Supplementary Fig. S11A and S11B), improved expansion is most likely due to the fact that the ratio of tumors to CAR T cells is higher at the lower cell dose (i.e., an individual CAR T cell encounters more antigen). To assess the functionality of CAR T cells in long-term survivors, we rechallenged a subset of mice with A673 cells on day 100 after initial tumor cell injection (Fig. 3G). Of 19 mice rechallenged after EphA2-CAR or EphA2-CAR.GM18 T-cell therapy, only one mouse developed an EphA2-negative tumor (Fig. 3H). In contrast, tumor take was 100% for naïve mice. Rejection of tumor cells did not induce expansion of EPHA2-CAR and EPHA2-CAR.GM18 T cells as judged by weekly bioluminescence imaging, except in two mice that had received EPHA2-CAR.GM18 T cells (Supplementary Fig. S12A and S12B).

Having established that EPHA2-CAR.GM18 T cells have potent antitumor activity in vivo, we confirmed in an additional experiment that the expression of a functional CAR and GM18 is critical for the observed benefit. On day 7, A673-bearing mice received a single intravenous dose of nontransduced T cells, GM18 T cells, ΔCAR T cells, ΔCAR.GM18 T cells, or CAR.GM18 T cells; mice that received only tumor cells served as controls. Only CAR.GM18 T cells had significant antitumor activity resulting in a survival advantage (Supplementary Fig. S13A–S13C), demonstrating that the expression of a functional CAR and GM18 is critical for the observed benefit.

In the final set of experiments, we wanted to establish that the benefit of GM18 expression was not limited to only EPHA2-CAR T cells. We focused on T cells expressing a second-generation HER2-CAR with a CD28.ζ signaling domain (HER2-CAR), which had been evaluated in preclinical as well as clinical studies (7–9). We generated HER2-CAR and HER2-CAR.GM18 T cells by retroviral transduction (Fig. 4A), and phenotypic analysis revealed no significant differences between CAR T-cell populations (Supplementary Fig. S14A–S14E). The in vitro experiments were conducted with unsorted (N = 3) and sorted (N = 1) T cells, whereas the in vivo experiments were conducted with sorted cells. We performed our standard restimulation assay using HER2⁺ LM7 osteosarcoma cells as targets, with the only exception that exogenous IL15 (5 ng/mL) was added with each restimulation. HER2-CAR.GM18 T cells expanded significantly more in comparison with HER2-CAR T cells within four stimulations (Fig. 4B and C). After the first stimulation, HER2-CAR and HER2-CAR.GM18 T cells produced high levels (>1,000 pg/mL) of GM-CSF, IFNγ, IL13, and TNFα (only HER2-CAR.GM18); intermediate levels (100–1,000 pg/mL) of IL4, IL5, and TNFα (only HER2-CAR); and low or undetectable levels (<100 pg/mL) of IL2, IL5, IL6, and IL17α (Fig. 4D). HER2-CAR.GM18 T cells produced higher levels of cytokines than HER2-CAR T cells, but this reached statistical significance only for IL13. HER2-CAR.GM18 T cells, as observed for EPHA2-CAR.GM18 T cells, sustained cytokine stimulation with each stimulation (Fig. 4E).

To determine if GM18 also endowed HER2-CAR T cells with enhanced antitumor activity in vivo, we utilized an established osteosarcoma model where LM7.GFP.ffLuc cells were injected intraperitoneally (i.p.) into NSG mice followed by one single intravenous CAR T-cell injection on day 7 (Fig. 4F). We evaluated two cell doses (1 × 10⁵ and 3 × 10⁵ T cells) at which HER2-CAR T cells were ineffective. In contrast, HER2-CAR.GM18 T cells had potent antitumor activity at both cell doses, resulting in a significant survival advantage (Fig. 4G–I; Supplementary Figs. S15 and S16).

Here we describe the development and characterization of an autocrine loop that links T-cell activation to the MyD88 signaling pathway, which plays a critical role in innate immunity. To create a functional loop, we designed a chimeric cytokine receptor, GM18, that (i) binds GM-CSF, a cytokine secreted upon CAR T-cell activation; and (ii) signals through the IL18 receptor. CAR T cells expressing GM18 recognized tumor cells in an antigen-dependent manner and had improved effector function in vitro in the setting of chronic antigen exposure in comparison with unmodified CAR T cells. This translated into potent antitumor activity in two solid tumor xenograft models targeting EPHA2 or HER2.

Several approaches are being developed to improve the antitumor activity of CAR T cells, including the expression of transcription factors, cytokines, constitutively active cytokine receptors, inducible costimulatory molecules, or chimeric cytokine or switch receptors (10–12). In addition, gene editing approaches are actively being pursued to delete negative regulators in CAR T cells such as PD-1 (13) or Regnase-1 (14). In most studies thus far, expression of the introduced gene is not linked to T-cell activation, with the exception of studies that have utilized the nuclear factor of activated T cells (NFAT) promoter to drive transgene expression (15–17). Although effective in preclinical studies, the NFAT promoter was not sufficient to control IL12 production by genetically modified tumor-infiltrating lymphocytes in early-phase clinical testing, resulting in systemic toxicities (16). Other regulatory systems have been developed that take advantage of small synthetic molecules or molecules that are present in the tumor microenvironment (18). Here, we decided to develop an autocrine loop to support T-cell effector function in which a constitutively expressed molecule is triggered by GM-CSF, which is consistently expressed upon CAR T-cell activation (3). Because we and other investigators had previously shown that activating MyD88 in CAR T cells improves their effector function (3, 19, 20), we decided to link GM-CSF secretion to MyD88 signaling. Although MyD88 is the central signaling hub for toll-like receptors, it is also the critical downstream signaling molecule for IL1 and IL18 receptors within T cells (4). We selected the IL18 receptor for our autocrine loop and designed a chimeric GM-CSF receptor with the transmembrane and signaling domains of IL18R. Our design was based on a chimeric GM-CSF/IL2 receptor that enabled cytotoxic T-cell clones to grow in vitro without exogenous cytokines in an activation-dependent manner (21).

EPHA2-CAR.GM18 or HER2-CAR.GM18 T cells produced higher amounts of cytokines after the first exposure to tumor cells than their unmodified counterparts. However, this reached statistical significance only for IL13. IL13 is a T_H2/T_C2 cytokine that is implicated in dampening antitumor responses and directly promoting growth of IL13Rα2-positive tumors (22). Yet, CAR T cells coexpressed IL13 with T_H1 cytokines such as IFNγ and TNFα after activation, which most likely counteracts any of its potential immunosuppressive effects or skewing toward a T_H2/T_C2 phenotype. Indeed, CAR.GM18 T cells outperformed CAR T cells in repeated stimulation assays as judged by their ability to expand and produce cytokines. These data suggest that the increased durability and potency of CAR.GM18 T cells may outweigh any of the potential protumorigenic effects of IL13. Nevertheless, studies in additional animal models are needed to confirm our findings. With repeat stimulations, there was no significant decrease in GM-CSF production by CAR.GM18 T cells, indicating that our designed autocrine loop remained intact. Importantly, there was also no increase, arguing against a self-amplifying autocrine loop, which would raise safety concerns. We observed an increase in IL5 secretion over repeat stimulations of EPHA2-CAR.GM18 T cells. Increased IL5 has been associated with eosinophil-mediated immune modulation and metastasis in some solid tumor models, thus warranting further investigation in immune-competent models (23). Finally, exogenous GM-CSF without T-cell activation or activated bystander CAR T cells did not induce CAR.GM18 T-cell proliferation, making it very unlikely that GM-CSF, which can be produced by endogenous cells in the tumor microenvironment (24), could induce nonspecific CAR T-cell expansion. Of interest, EPHA2-CAR.GM18 T cells required no addition of exogenous IL15 to observe a benefit in vitro, whereas HER2-CAR.GM18 T cells did. This is most likely due to differences in the strength of CAR-mediated T-cell activation, as EPHA2-CAR.GM18 T cells secreted higher amounts of T_C1/T_H1 cytokines after initial tumor cell stimulation in comparison with HER2-CAR.GM18 T cells, although this reached statistical significance only for GM-CSF (Supplementary Fig. S17).

Although high CAR T-cell doses (5 × 10⁶ to 10 × 10⁶ CAR T cells per mouse) have been effective in numerous preclinical solid tumor models, clinical activity against solid tumors has been limited (1, 2). We therefore focused on evaluating low doses of CAR T cells (1 × 10⁵ or 3 × 10⁵ per mouse) to determine if transgenic expression of GM18 improves their effector function. EPHA2-CAR.GM18 and HER2-CAR.GM18 T cells outperformed EPHA2- and HER2-CAR T cells in xenograft models and induced complete remission in >90% of animals at a cell dose of 1 × 10⁵ T cells per mouse at which unmodified EPHA2-CAR or HER2-CAR T cells were ineffective. In one of our models, we assessed T-cell expansion and demonstrated that improved expansion and persistence of CAR.GM18 T cells correlated with improved antitumor activity. At an effective CAR.GM18 T-cell dose of 1 × 10⁵ T cells per mouse, T cells persisted long term, did not induce xenogeneic GVHD, and rejected a second tumor challenge 3 months after the initial injection. Collectively, this indicates that modification of CAR T cells with GM18 does not induce antigen-independent CAR T-cell expansion in vivo. In contrast, constitutive, transgenic expression of IL18 in CAR T cells improves their effector function but also induces weight loss in some mouse models and antigen-independent T-cell function (15, 25). These studies highlight that it is critical to link the MyD88 signaling pathway to T-cell activation, which we accomplished with the designed GM18 receptor. Other investigators have used the NFAT promoter to link IL18 expression to T-cell activation (15); however, as discussed above, this promoter failed to control IL12 production by tumor-infiltrating lymphocytes in early-phase clinical testing (16). In our study, we demonstrated the benefit of expressing GM18 in human CAR T cells. In the future, our findings have to be confirmed in immune-competent animal models especially because there is no biological cross-reactivity between human and murine GM-CSF. Likewise, immune competent models will also allow us to perform additional studies focused, for example, on T-cell metabolism and safety.

In conclusion, we designed a chimeric cytokine receptor, GM18, to establish an autocrine loop that links T-cell activation to the MyD88 signaling pathway. Expression of GM18 in CAR T cells strikingly improved their effector function, resulting in potent antitumor activity. Thus, hijacking cytokines, such as GM-CSF, that are secreted by T cells in an antigen-dependent manner, for therapeutic benefit, presents a viable option for improving current T-cell therapy approaches.

A673 (Ewing sarcoma) was purchased from the American Type Culture Collection (ATCC), and the LM7 (osteosarcoma) cell line was provided by Dr. Eugenie Kleinerman (MD Anderson Cancer Center, Houston, TX). Cell lines were authenticated by the ATCC human short-tandem repeat profiling cell authentication service and routinely checked for mycoplasma by the MycoAlert Mycoplasma Detection Kit (Lonza). The generation of LM7 cells, genetically modified to express an enhanced green fluorescent protein firefly luciferase molecule (LM7.GFP.ffLuc), was previously described (7). Once thawed, cell lines were kept in culture for a maximum of 3 months before a new reference vial was thawed. Cell lines were maintained and expanded in Dulbecco's Modified Eagle Medium (GE Healthcare Life Sciences HyClone Laboratories) supplemented with 10% fetal bovine serum (FBS; GE Healthcare Life Sciences HyClone Laboratories) and 2 mmol/L Glutamax (Invitrogen).

The generation of SFG retroviral vectors encoding EPHA2-CAR-2A-tCD19, EPHA2-ΔCAR-2A-tCD19, EPHA2-CAR-2A-tCD19 with 4-1BB costimulatory domain (CAR^BB), or HER2-CAR have been described previously (6, 7). The SFG retroviral vector encoding GM18 was generated by synthesizing gene fragments (Thermo Fisher Scientific) and In-Fusion cloning (Takara Bio). It consists of (i) the GM-CSFRβ isoform 2 extracellular domain ending with amino acids MW; (ii) the transmembrane domain and intracellular domain of the IL-18Rβ chain starting with amino acids GV (omitting the second V); (iii) a T2A sequence; (iv) the GM-CSFRα extracellular domain ending with amino acids DG; and (v) the transmembrane domain and intracellular domain of the IL18Rα chain starting with amino acids MI. A nonsignaling GM18 construct (ΔGM18) was generated by In-Fusion cloning to truncate the intracellular domains of IL18Rβ and IL18Rα to 10 amino acids (ending with amino acids IE for IL18Rβ and amino acids YR for IL18Rα). The sequences of the final constructs were verified by sequencing (Hartwell Center, St. Jude Children's Research Hospital). RD114-pseudotyped retroviral particles were generated by transient transfection of 293T cells as previously described (6). Supernatants were collected after 48 hours, filtered, and snap-frozen.

Human peripheral blood mononuclear cells (PBMC) were obtained from whole blood of healthy donors under an institutional review board–approved protocol at St. Jude Children's Research Hospital, after written informed consent was obtained in accordance with the tenets of the Declaration of Helsinki or from deidentified donor pheresis products of the St. Jude Blood Donor Center. Retroviral transduced T cells were generated as previously described (6). Briefly, PBMCs were stimulated on plates coated with anti-CD3 and anti-CD28 for 48 hours. Recombinant human IL7 (10 ng/mL, Peprotech) and IL15 (5 ng/mL, Peprotech) were added 24 hours after initial stimulation and were maintained in culture until functional studies were performed. Cells were then seeded onto retronectin-coated (Clontech) plates with retroviral particles for 2 to 4 days for transduction. Nontransduced T cells were prepared similarly, except that no retrovirus was included in the retronectin wells. To generate CAR.GM18 and CAR.ΔGM18 T cells, T cells were co-transduced with both retroviral particles in the same well. For generation of EPHA2-CAR-GFP.ffLuc and EPHA2-CAR.GM18-GFP.ffLuc T cells, activated T cells were first transduced with CAR or CAR +GM18 for 24 hours and then transferred to GFP.ffLuc retrovirus-containing retronectin-coated plates for 3 to 4 days. T cells were then expanded and sorted for functional analysis for 7 to 10 days after transduction. For all in vitro and in vivo experiments in which the effector function of CAR and CAR.GM18 T cells was compared, T cells from the same donor were transduced with (i) CAR-encoding retroviral particles or (ii) a mixture of CAR- and GM18-encoding retroviral particles. All T cells were cultured with RPMI-1640 supplemented with 10% FBS and 2 mmol/L Glutamax (complete RPMI).

The FACSCanto II instrument (BD Biosciences) was used for acquisition and FlowJo v10 (FlowJo) was used for analysis of flow cytometry data. Samples were washed with and stained in PBS (Lonza) with 1% FBS. For all experiments, known negatives (e.g., NT T cells) served as gating controls. eBioscience Fixable viability dyes (Invitrogen) were used to exclude dead cells from analysis. T cells were evaluated for CAR expression 5–10 days post transduction using anti-CD19 (J3–119, Beckman Coulter; SJ25C1, BD Biosciences) or antihuman immunoglobulin G (IgG) F(ab′)₂ fragment-specific antibody (Jackson ImmunoResearch) for EPHA2-CAR. Anti-mouse IgG F(ab′)₂ fragment-specific antibody (Jackson ImmunoResearch) was used to detect HER2-CAR expression. GM18 expression was analyzed by staining with anti-CD116 (4H1, BioLegend). CAR T-cell phenotype was established using the following antibodies: CD4 (SK3, BD Biosciences), CD8 (HIT8a, BD Biosciences), CD45RA (HI100, BD Biosciences), and CCR7 (G043H7, BioLegend; 150503, BD Biosciences). The beta chain of GM18 was detected with anti-CD131 (3D7, BD Biosciences). Cell-surface EPHA2 expression was detected using anti-EPHA2 (371805, R&D Systems).

For functional studies, cells were sorted on a BD FACS Aria III cell sorter (BD Biosciences). Cells were stained for anti-CD19-APC (J3-119, Beckman Coulter; EPHA2-CAR) or anti-mouse IgG F(ab′)₂ fragment-specific AlexaFluor 647 (HER2-CAR) plus anti-CD116-PE (4H1, BioLegend, or hGM-CSFR-M1, BD Biosciences). DAPI (4′,6-diamidino-2-phenylindole, Thermo Fisher) was used as a viability indicator. Cells were rested 48 to 72 hours in RPMI containing 20% FBS, 25 μg/mL gentamicin (Gibco), 1× penicillin–streptomycin (Gibco), and 1.5 μg/mL amphotericin B (Gibco) prior to functional assays.

Ramos-Blue (WT) and Ramos-Blue KD-MyD (MyD88 knockdown, KD) NFκB/AP-1 reporter B lymphocytes (InvivoGen) were transduced with GM18 or DGM18 or were not transduced (NT) and were expanded per manufacturer instructions. Then, 4 × 10⁵ NT or GM18 Ramos-Blue cells were seeded in 96-well plates in Iscove's Modified Dulbecco's Medium (Gibco) with 10% FBS (GE Healthcare Life Sciences HyClone Laboratories) and stimulated with exogenous recombinant human GM-CSF (R&D Systems) at titrated concentrations for 24 hours. Supernatant was collected and incubated with QUANTI-Blue solution (InvivoGen) for 2 hours. Colorimetric changes were then detected via absorbance at 640 nm using an Infinite 200 Pro MPlex plate reader (Tecan).

1 × 10⁶ (or 5 × 10⁵) T cells were cocultured in complete RPMI with 5 × 10⁵ (or 1 × 10⁵) tumor cells in a 24-well or 48-well tissue culture-treated plate, respectively. IL15 (5 ng/mL) was added to the HER2-CAR T-cell experiments. Cells were fed with fresh complete RPMI at 48 and 120 hours after coculture. After 7 days, T cells were harvested, counted, and replated at the same ratio with fresh tumor cells as long as they had killed tumor cells, as determined by microscopic inspection.

Supernatants from repeat stimulation assays were collected 48 hours after each new stimulation and frozen at −80°C. Cytokines and chemokines were then quantified using a MILLIPLEX MAP Human Cytokine/Chemokine Magnetic Bead Panel kit (EMD Millipore) on a FLEXMAP 3D System (Luminex).

Eight days after transduction, NT or GM18 T cells were plated with recombinant human IL7 (10 ng/mL, Peprotech) and IL15 (5 ng/mL, Peprotech), recombinant human GM-CSF (100 ng/mL or 10 ng/mL, R&D Systems), or no exogenous cytokines for 5 days. Cells were then enumerated, and the frequency of GM18⁺ cells was determined by flow cytometry.

A CellTiter96 AQueous One Solution Cell Proliferation Assay (Promega) was utilized to assess CAR T-cell cytotoxicity. In a tissue culture–treated 96-well plate, 19,000 A673 cells were cocultured with serial dilutions of CAR T cells. Media only and tumor cells alone served as controls. Each condition was plated in technical triplicates. After 24 hours, the media and T cells were removed by gently pipetting up and down to avoid disrupting adherent tumor cells. CellTiter96 AQueous One Solution Reagent (phenazine ethosulfate) in complete RPMI was added to each well and incubated at 37°C for 3 hours. The absorbance at 492 nm was measured using an Infinite 200 Pro MPlex plate reader (Tecan) to quantify the viable cells in each well. Percent live tumor cells were determined by the following formula: (sample-media only)/(tumor alone-media only) × 100.

All animal experiments were approved by the St. Jude Children's Research Hospital Institutional Animal Care and Use Committee. Xenograft experiments were performed with 7- to 10-week-old NSG mice obtained from St. Jude Children's Research Hospital NSG colony.

Mice received s.c. injection of 2 × 10⁶ A673 cells in Matrigel (Corning) in the right flank. On day 7, mice received a single i.v. dose of 1 × 10⁵ or 3 × 10⁵ EPHA2-CAR or EPHA2-CAR.GM18 T cells via tail-vein injection. Tumor growth was measured by weekly caliper measurements. Mice were euthanized when they met physical euthanasia criteria (significant weight loss, signs of distress), when the tumor burden reached 20% of total body mass (≥4,000 mm³), or when recommended by veterinary staff. For rechallenge experiments, mice received an additional s.c. injection of 2 × 10⁶ A673 cells in the left flank between days 102 and 104 after the initial tumor cell injection.

Mice were injected i.p. with 1 × 10⁶ LM7.GFP.ffLuc cells, and on day 7 received a single i.v. dose of 1 × 10⁵ or 3 × 10⁵ HER2-CAR or HER2-CAR.GM18 T cells via tail-vein injection. Mice were euthanized when they reached our bioluminescence flux endpoint of 1 × 10¹⁰ for 2 consecutive weeks and/or the above-mentioned general euthanasia criteria.

Mice were injected i.p. with 150 mg/kg of d-luciferin 5 to 10 minutes before imaging, anesthetized with isoflurane, and imaged with a Xenogen IVIS-200 imaging system. The photons emitted from the luciferase-expressing tumor cells were quantified using Living Image software (Caliper Life Sciences). Mice were imaged once per week to track either T cells (EPHA2-CAR T-cell experiments) or LM7 tumor burden (HER2-CAR T-cell experiments).

For all experiments, the number of biological replicates and statistical analysis used are described in the figure legends. For comparisons between two groups, a two-tailed t-test was used. For comparisons of three or more groups, values were log transformed as needed and analyzed by analysis of variance (ANOVA) with Dunnett or Tukey post-test. Survival was assessed by the log-rank test with Bonferroni adjustment for multiple comparisons. Bioluminescence imaging data were analyzed using either ANOVA or area under the curve. Statistical analyses were conducted with GraphPad Prism software.

S. Lange received grants from the National Institutes of Health, Alex's Lemonade Stand Foundation, National Cancer Institute, and American Lebanese Syrian Associated Charities during the conduct of the study; she also has a pending patent for Chimeric GMCSF-IL18 Receptor. L.G.L. Sand reported personal fees from Gadeta BV outside the submitted work; she also has a pending patent for Chimeric GMCSF-IL18 Receptor. S.L. Patil received grants from the National Institutes of Health, Alex's Lemonade Stand Foundation, National Cancer Institute, and American Lebanese Syrian Associated Charities during the conduct of the study. D. Langfitt received grants from the National Institutes of Health, Alex's Lemonade Stand Foundation, National Cancer Institute, and American Lebanese Syrian Associated Charities during the conduct of the study. S. Gottschalk received grants from the National Institutes of Health, National Cancer Institute, Alex's Lemonade Stand Foundation, and American Lebanese Syrian Associated Charities during the conduct of the study; received personal fees from TESSA Therapeutics and Immatics outside the submitted work; is on the scientific advisory board of Tidal; and has a pending patent for Chimeric GMCSF-IL18 Receptor. No potential conflicts of interest were disclosed by the other authors.

S. Lange: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing. L.G.L. Sand: Conceptualization, resources, data curation, formal analysis, investigation, methodology, writing–review and editing. M. Bell: Data curation, formal analysis, investigation, methodology, writing–review and editing. S.L. Patil: Data curation, formal analysis. D. Langfitt: Data curation, investigation, methodology, writing–review and editing. S. Gottschalk: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing.

The authors thank Krista Millican and Amanda George (St. Jude Animal Resource Center) and Rebecca Thorne (St. Jude Center for In Vivo Imaging and Therapeutics) for assistance with the in vivo mouse studies. The schematic shown in Figure 1B was created with BioRender (Biorender.com), for which we have a license. The work was supported by the Alex's Lemonade Stand Foundation and the American Lebanese Syrian Associated Charities. Animal imaging was performed by the St. Jude Center for In Vivo Imaging and Therapeutics, which is supported in part by grants from the National Institutes of Health (P01CA096832 and R50CA211481). Cellular images were acquired at the St. Jude Children's Research Hospital Cell and Tissue Imaging Center, which is supported by St. Jude and a grant from the National Cancer Institute (P30 CA021765). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.