Pre-clinical validation of a pan-cancer CAR-T cell immunotherapy targeting nfP2X7

Immunotherapy has transformed cancer treatment by enabling a patient’s immune system to recognize and eliminate tumor cells. CAR-T cell therapy, which merges antibody-derived antigen recognition with T-cell cytotoxic signaling in an MHC-independent manner, has achieved high response rates in hematologic malignancies (e.g., CD19 CAR-T in B-ALL, CLL, and NHL). Despite this success, translating CAR-T to solid tumors has been challenging due to tumor heterogeneity, an immunosuppressive tumor microenvironment, insufficient trafficking and persistence, and the scarcity of truly tumor-specific antigens shared across multiple solid cancers. P2X7 is an ATP-gated ion channel that, in its wild-type form, can either support cell survival or trigger cell death depending on ATP exposure. A non-functional conformer, nfP2X7, presents a unique extracellular epitope exposed on many solid tumors but is undetectable on healthy tissues, suggesting a potential broad and tumor-specific CAR target. This study evaluates nfP2X7 as a pan-cancer target and develops a second-generation nfP2X7-targeted CAR-T, optimizing its hinge length and assessing its in vitro and in vivo anti-tumor efficacy.

Clinical trials of CD19 CAR-T cells have shown complete response rates up to ~80–93% in B-ALL and durable remissions, but a meta-analysis reports only ~9% overall response for CAR-T in solid tumors, highlighting translational barriers. Antigen selection is critical; on-target off-tumor toxicity remains a concern for many antigens. P2X7 has a dichotomous function: brief ATP exposure opens a cation channel supporting survival, whereas prolonged exposure forms a large pore leading to cell death. A distinct non-functional conformation (nfP2X7) exposes a cryptic extracellular epitope found on diverse cancers (breast, ovary, brain, prostate, skin, bowel, cervix, lung, pancreas, stomach) and not on healthy cells, including T cells adjacent to tumors. Antibodies to this epitope have shown tumor detection and clinical safety in topical therapy for basal cell carcinoma. Prior work indicates nfP2X7 confers a selective advantage by preventing pore-mediated apoptosis in ATP-rich tumor microenvironments, and siRNA knockdown of nfP2X7 increases apoptosis in prostate cancer cells. These observations motivate targeting nfP2X7 with CAR-T cells as a cancer-specific, potentially pan-tumor antigen.

CAR design and optimization: A second-generation human CAR was constructed using a binding domain derived from a published peptide binder to nfP2X7, with an IgG4-based hinge, CD28 transmembrane, 4-1BB costimulatory, and CD3ζ signaling domains. Three hinge lengths (short, medium [IgG4 CH3-CH2], long) with CH2 modifications to reduce Fcγ receptor interactions were tested. A truncated EGFR (EGFRt) reporter linked via T2A enabled tracking of CAR expression. T-cell manufacturing: Human CD3+ T cells (CD4+ and CD8+ subsets purified) from healthy donors were lentivirally transduced and expanded for ~14 days. Cytokine supplements were optimized: initial IL-2-only cultures yielded suboptimal expansion and more terminal differentiation; adding IL-7 and IL-15 favored naive/central memory phenotypes (Tn, Tcm, Tscm) and improved yields, becoming the standard cocktail. Immunophenotyping: Flow cytometry assessed maturation (CD45RA, CD45RO, CD62L/‘CD26L’, CCR7), activation (CD27, CD28, CD95, CXCR3), co-inhibitory receptors (PD-1, CTLA-4, LAG-3, TIM-3, CD39), and cytotoxic molecules (IFN-γ, TNF-α, IL-2, Granzyme B, Perforin, CD107a). CAR binding specificity was examined using a biotin-labeled nfP2X7 peptide mimetic alongside EGFRt staining. In vitro cytotoxicity: Two assays were used. (1) Standard 4 h 51Cr-release: CAR-T cells were co-cultured with target lines at E:T ratios (30:1, 10:1, 1:1), including K562, K562-OKT3 (positive control), MDA-MB-231, U87, M21, SK-N-DZ. (2) Bright-Glo luciferase assay (16 h) against luciferase-expressing cancer cells at E:T ratios (100:1, 32:1, 10:1), with cytotoxicity calculated from luminescence reduction. Real-time kinetics were measured using xCELLigence impedance assays with MDA-MB-231, PC3, OVCAR3, and time-lapse confocal imaging with GFP+ MDA-MB-231-LM2 targets. Off-target assessment: Healthy donor PBMCs were transduced to express luciferase and tested for cytotoxicity by nfP2X7 CAR-T at up to 10:1 E:T. On-target validation: CRISPR/Cas9 editing of P2X7 in PC3 cells generated knockout clones (confirmed by PCR/Sanger sequencing); cytotoxicity of nfP2X7 CAR-T was compared between PC3 KO and non-targeting control cells. In vivo studies: NSG mice received human tumor xenografts (MDA-MB-231-LM2 in mammary fat pad; PC3 in flank). After tumor establishment (~12–14 days), mice were infused intravenously with 2 × 10^6 nfP2X7-M CAR-T or untransduced T cells. Tumor growth was monitored; intratumoral T-cell infiltration was assessed by anti-human CD3 IHC on FFPE sections. Intravital multiphoton microscopy visualized CAR-T trafficking, vascular adhesion, and transmigration in MDA-MB-231 tumors following tail vein infusion of CellTracker Orange-labeled CAR-T cells with dextran vascular labeling. Statistics: Paired t-tests and two-way ANOVA with Bonferroni post-tests were used; data are presented as mean ± SEM.

- CAR expression and phenotype: All hinge variants transduced efficiently. EGFRt+ frequencies (representative): CD4+ nfP2X7-S ~99.2%, nfP2X7-M ~92.0%, nfP2X7-L ~85.7; CD8+ nfP2X7-S ~92.0%, nfP2X7-M ~93.1%, nfP2X7-L ~70.9. Expanded CAR-T populations were enriched for naive, central memory, and stem cell memory phenotypes with high activation marker expression and low co-inhibitory receptor expression pre-infusion. - Hinge optimization: Only the medium-length (IgG4 CH3-CH2) hinge (nfP2X7-M) consistently yielded superior cytokine release and cytotoxicity in vitro compared with short or long hinges, despite similar CAR surface expression levels. - In vitro cytotoxicity (selected data): In 51Cr assays, nfP2X7-M CD8+ CAR-T showed dose-dependent killing: MDA-MB-231 ~43%, U87 ~25%, M21 ~21% (representative). In luciferase assays at E:T 10:1, cytotoxicity above UT controls included: DU145 41.55%, BT-549 52.22%, MDA-MB-231 43.45%, SH-SY5Y 70.89%, BE(2)M17 72.11%, HCT116 60.45%, LIM1215 53.11%, RD 27.82%, ASPC-1 20.22%, Namalwa 72.11%, OVCAR5 70.45%, OVCAR3 51.77%, UMSCC-1 48.55%, PC9 46.89%, NCI-H460 38.67%, C32 41.76%, SK-MEL-28 4.66%, SK-MEL-5 7.78%. Overall, nfP2X7-M CAR-T were reactive against 22/24 cancer cell lines tested in vitro. - Kinetics: Real-time impedance assays showed robust, durable killing of MDA-MB-231, PC3, and OVCAR3 with near-complete tumor cell elimination by ~70 h; time-lapse imaging confirmed direct engagement and killing by CAR-T but not by UT T cells. - On-target dependency: CRISPR P2X7 knockout in PC3 reduced nfP2X7-M CAR-T-mediated cytotoxicity by ~40–60% versus non-targeting controls, while UT T cells showed low background killing, indicating antigen-dependent activity. - Off-target safety in vitro: No detectable cytotoxicity against healthy donor PBMCs even at E:T 10:1, despite PBMC expression of wild-type P2X7, indicating specificity for the nfP2X7 conformer. - In vivo efficacy and trafficking: In NSG xenografts, nfP2X7-M CAR-T significantly inhibited tumor growth in MDA-MB-231 and PC3 models (tumor size reduction, p < 0.0001). Intravital microscopy visualized CAR-T adhesion to tumor vasculature and transmigration within 2 h post-infusion. Ex vivo, higher intratumoral human CD3+ T-cell frequencies were recovered from CAR-T-treated tumors with an increased CD4:CD8 ratio (~3.41 vs ~1.11 in UT). In one PC3 study, 100% survival was observed up to 90 days post-inoculation in nfP2X7-M CAR-T-treated mice.

This work identifies nfP2X7 as a broadly expressed, cancer-specific epitope that can be targeted by CAR-T cells with minimal recognition of the wild-type P2X7 on healthy cells. Optimizing the extracellular hinge length was crucial, with the medium IgG4 CH3-CH2 spacer producing superior activation, cytokine release, and cytotoxicity, consistent with the importance of CAR geometry and immunological synapse formation. The nfP2X7-M CAR-T cells demonstrated robust in vitro killing across diverse tumor lineages and validated antigen dependence via CRISPR knockout experiments. Lack of cytotoxicity toward PBMCs supports specificity for the non-functional conformation rather than wild-type P2X7. In vivo, nfP2X7 CAR-T trafficked to tumors, infiltrated the microenvironment, and significantly inhibited growth in breast (MDA-MB-231) and prostate (PC3) xenografts, with strong intratumoral T-cell accumulation and a predominance of CD4+ subsets after recovery. Differences in efficacy between models likely reflect antigen density/heterogeneity and microenvironmental constraints, with potential antigen loss contributing to reduced sensitivity in some tumors. The manufacturing process that preserves naive and central memory phenotypes using IL-7/IL-15 likely supported persistence and function. Collectively, these findings support nfP2X7 as a promising pan-cancer CAR-T target for solid tumors, warranting clinical translation with patient selection based on tumor nfP2X7 expression.

nfP2X7-targeted CAR-T cells display broad, antigen-specific cytotoxicity against multiple solid tumor types in vitro and significant anti-tumor efficacy in breast and prostate cancer xenograft models, with evidence of tumor trafficking, infiltration, and on-target dependency. Medium hinge length optimization was key to potency, and the manufacturing approach enriched for favorable T-cell phenotypes. These preclinical data position nfP2X7 as a viable pan-cancer CAR-T target with limited risk of off-tumor effects on healthy cells expressing wild-type P2X7. Future work should include development and application of high-specificity nfP2X7 immunohistochemistry/flow diagnostics for patient selection, evaluation in additional solid tumor models, assessment of durability and resistance mechanisms (e.g., antigen loss), and combination strategies (e.g., checkpoint blockade, standard therapies) to enhance clinical efficacy.

The study is pre-clinical and relies on immunodeficient mouse xenograft models that do not recapitulate an intact human immune system or full tumor microenvironment complexity. Some in vitro and in vivo results showed variable sensitivity among tumor lines, suggesting antigen heterogeneity and expression thresholds may limit universality. Real-time tumor monitoring was not performed across all experiments, and sample sizes for some in vivo analyses were limited. Potential adaptive resistance (e.g., antigen downregulation) was not comprehensively characterized. Long-term persistence and safety (including cytokine release, neurotoxicity, on-target off-tumor in non-hematopoietic tissues) require clinical evaluation. Patient selection will depend on reliable detection of nfP2X7 expression, necessitating validated diagnostic reagents.