Rhabdoid tumors as a novel target for PSMA-directed CAR T cell therapy

Abstract

Rhabdoid tumor is an ultra-rare and highly aggressive pediatric malignancy with a poor prognosis and limited therapeutic options. To identify novel immunotherapeutic targets, transcriptomic data from the Cancer Cell Line Encyclopedia were analyzed, and we found that two rhabdoid tumor cell lines exhibit high expression of prostate-specific membrane antigen (PSMA), with levels comparable to well-established PSMA-positive prostate cancer cell lines. PSMA expression in rhabdoid tumors was subsequently validated in cell lines and in a subset of primary clinical rhabdoid tumor specimens. While PSMA-directed therapies have primarily been explored in prostate cancer, we evaluated their potential in rhabdoid tumors by employing PSMA-directed chimeric antigen receptor (CAR) T cells. These CAR T cells demonstrated potent and antigen-specific cytotoxicity against PSMA-positive rhabdoid tumor cells in vitro. In addition, the in vivo efficacy was also assessed in xenograft mouse models of non-CNS tumors, where PSMA CAR T cell treatment resulted in significant tumor regression and robust accumulation of CAR T cells within the tumor microenvironment. Together, these findings establish PSMA as a promising surface antigen beyond prostate cancer and provide preclinical evidence supporting the development of PSMA-directed therapies for this highly lethal pediatric cancer.

Graphical abstract

This study identifies prostate-specific membrane antigen (PSMA) as an unexpected immunotherapy target in a subset of malignant rhabdoid tumors. PSMA-directed CAR T cells show potent antitumor activity in preclinical models, suggesting a potential therapeutic opportunity for this ultra-rare but highly aggressive pediatric cancer.

Introduction

Rhabdoid tumor is a highly aggressive pediatric cancer primarily characterized by the presence of rhabdoid cells exhibiting large nucleoli, uncondensed chromatin, and eosinophilic cytoplasmic inclusions.¹ These tumors commonly arise in the brain, kidney, lungs, abdomen, and other soft tissues.² It predominantly affects children under the age of 3 years, though it can occasionally occur in older children and adults.^2,3 Based on the site of occurrence, rhabdoid tumors (RTs) can be categorized into three main types: tumors arising in the brain are classified as atypical teratoid/RTs (AT/RTs), tumors originating in the kidney are designated as RTs of the kidney (RTKs), and those occurring outside the brain and kidney are referred to as extrarenal or malignant RTs (MRTs). AT/RT, a central nervous system (CNS) tumor, can be further subdivided into three primary molecular subgroups based on gene expression patterns, the nature of SMARCB1 inactivation, and the extent of tumor-infiltrating lymphocytes: AT/RT-TYR (Tyrosinase), AT/RT-MYC (Myelocytomatosis), and AT/RT-SHH (Sonic hedgehog).⁴ In contrast, non-CNS tumors such as MRTs and RTKs show distinct clinical and molecular features.⁵

The survival rate for RTs is generally low. The most common treatment is multimodal combination therapy. This typically involves surgical removal of the tumor, followed by chemotherapy and radiation therapy. The median overall survival for patients is 28 months, which increases to 54 months with multimodal treatment.⁶ While this represents a significant improvement, the limited outcome indicates an urgent need for the development of novel approaches, such as targeted therapies, to improve patient survival.

RTs have a very low mutational burden, which makes them resistant to immune checkpoint inhibitor therapy. Interestingly, these tumors consistently share mutations in a single epigenetic regulator, with over 95% of cases linked to biallelic mutations in the SWI/SNF (SWItch/Sucrose Non-Fermentable)-related matrix-associated actin-dependent regulator of chromatin subfamily B member 1 (SMARCB1) gene, which encodes the integrase interactor 1 (INI1) protein.⁷ Fewer than 5% of cases involve mutations in the SMARCA4 gene.⁸ Both SMARCB1 and SMARCA4 are core subunits of the SWI/SNF complex, suggesting that dysregulation of chromatin remodeling underlies the development of RTs. However, despite extensive efforts to develop targeted therapies exploiting SMARCB1 or SMARCA4 deficiency, success has been limited so far.

Prostate-specific membrane antigen (PSMA) is a type II transmembrane glycoprotein. It is a well-characterized cell surface protein primarily expressed in the prostate gland.⁹ In prostate cancer, PSMA is significantly overexpressed, making it a valuable diagnostic and therapeutic target.¹⁰ The FDA has approved PSMA-targeted radioligand therapy, ¹⁷⁷Lu-PSMA-617, for metastatic prostate cancer.^11,12 While PSMA-targeted therapies have been extensively studied and successfully used in prostate cancer, their application in non-prostate malignancies remains limited. In other cancers where PSMA has been proposed as a therapeutic target, such as glioblastoma, renal cell carcinoma, and lung cancer, PSMA expression is typically restricted to the tumor-associated neovasculature rather than the tumor cells themselves.^13,14,15,16 In this study, we have identified a subset of RTs that express PSMA directly on the tumor cells. Building on the established safety and efficacy of PSMA-targeted therapies in prostate cancer, we investigated the therapeutic potential of PSMA-directed chimeric antigen receptor (CAR) T cells in RTs. As a result, we demonstrate that PSMA CAR T cell therapy effectively eliminates PSMA-positive RT cells. These findings suggest that PSMA-directed immunotherapy may offer a viable treatment approach for RTs and support the broader application of PSMA-targeted therapies beyond prostate cancer.

Results

PSMA is highly expressed in RT cell lines STM91-01 and TTC709

PSMA is known to be highly expressed in most prostate cancers, making it an attractive therapeutic target for treating this disease. To investigate whether PSMA expression is present in other cancer types, we searched the Cancer Cell Line Encyclopedia (CCLE) database for cancer cell lines with high PSMA expression. As expected, prostate cancer cell lines such as LNCaP, MDAPCA2B, and VCaP ranked among the top cancer cell lines expressing PSMA. Unexpectedly, we also identified two RT cell lines, STM91-01 and TTC709, among the top five cell lines with the highest PSMA expression (Figure 1A). STM91-01 was established from a renal tumor of a pediatric patient and lacks differentiation into neural or mesenchymal lineages.¹⁷ TTC709 was derived from an extrarenal malignant RT. Both cell lines harbor the characteristic biallelic inactivation of SMARCB1.^18,19

Figure 1.

PSMA is highly expressed in rhabdoid tumor cell lines STM91-01 and TTC709

(A) PSMA mRNA levels across 1,076 CCLE cancer cell lines, shown as RPKM (reads per kilobase per million). The two MRT cell lines (STM91-01 and TTC709) are labeled in red, and four prostate cancer cell lines are labeled in blue. (B) RT-qPCR analysis of PSMA and AR expression in LNCaP, STM91-01, TTC709, and TTC1240 cells. LNCaP cells were used as the positive control and set to a relative expression level of 1.0. Each rhabdoid tumor cell line was compared individually to the negative control (TTC1240) using a Student’s t test. N = 3 biological replicates. (C) Western blot analysis of PSMA protein expression in MRT cell lines. (D) Flow cytometry analysis of LNCaP cells, with or without PSMA antibody staining, to identify PSMA-positive cells. (E) Flow cytometry analysis of STM91-01, TTC709, and TTC1240 cells using the same PSMA antibody as in (D) for the detection of PSMA-positive cells.

To validate this finding, we first assessed PSMA mRNA expression using reverse-transcription (RT)-qPCR. The result confirmed that PSMA mRNA was highly expressed in STM91-01 and TTC709 cells, while it was absent in a PSMA-negative RT cell line, TTC1240 (Figure 1B). TTC1240 was derived from an extrarenal tumor of a pediatric patient. Similar to STM91-01 and TTC709, it harbors the hallmark biallelic inactivation of SMARCB1.¹⁸ As a control, the prostate cancer-specific gene androgen receptor (AR) was expressed only in LNCaP cells and not in any of the rhabdoid cell lines (Figure 1B). Western blot analysis further confirmed PSMA expression at the protein level in STM91-01 and TTC709, but not in TTC1240 (Figure 1C). To examine whether PSMA is expressed on the cell membrane of rhabdoid cells, we examined its surface expression using flow cytometry. As a positive control, LNCaP cells showed 99.5% of cells expressing PSMA (Figure 1D). In comparison, 99.9% of STM91-01 cells and 95% of TTC709 cells were PSMA-positive. Taken together, these results demonstrate that PSMA is highly expressed as a membrane protein in a subset of in vitro-cultured RT cell lines, at levels comparable to those in prostate cancer cells.

PSMA is expressed in a subset of clinical RT samples

To further evaluate PSMA expression in clinical RTs, we analyzed RNA sequencing data from 59 tumor samples representing 51 patients, retrieved from the St. Jude Genomics Platform. RT is an ultra-rare pediatric cancer, and comprehensive transcriptomic datasets for this disease are not widely available as they are for more common cancers. Of the 59 samples, 46 are AT/RTs, a subtype of MRTs localized to the CNS.²⁰ Based on their counts per million (CPM), we classified the tumors into three categories: no or low expression (CPM less than 10), moderate expression (CPM between 10 and 50), and high expression (CPM greater than 50) (Figure 2). Among these tumors, a subset of samples showed detectable PSMA expression, with three AT/RT tumor samples exhibiting particularly high levels. The highest-expressing sample, from patient SJBT032275, had a CPM of 1050.58, which is close to the PSMA expression level in LNCaP cells, measured at 1353.96 CPM based on RNA sequencing results from our recent study.²¹ Notably, LNCaP has the highest PSMA expression among all prostate cancer cell lines in the CCLE (Figure 1A), illustrating the unusually high PSMA expression in these RTs. Additionally, PSMA expression in the RT samples did not appear to correlate with tissue origin or whether the tumor was CNS-derived, as no statistical significance was observed between the distribution of AT/RTs (CNS tumors) and MRTs (non-CNS tumors) (Figure S1). Collectively, these findings suggest that a subset of clinical RTs are PSMA-positive and may be amenable to PSMA-targeted therapies.

Figure 2.

PSMA expression in rhabdoid tumor samples from various tissue origins

D1 and D2 represent tumor collections at different time points. RNA-seq data from 59 patient samples were obtained from the St. Jude Genomics Platform on DNAnexus. Gene expression data were converted to CPM for comparison across multiple samples. MRT, malignant rhabdoid tumor; NOS, not otherwise specified.

PSMA CAR T cells efficiently eliminate RT cells in vitro

We then tested whether PSMA CAR T therapy can be used to treat these PSMA-positive RTs. We have developed several single-chain variable fragments (scFvs) targeting PSMA.²² The CAR vector contains the PSMA scFv, a CD8 transmembrane domain, 4-1BB, and CD3ζ (Figure 3A). We chose to use a CAR named 4LH to explore CAR T cell activity on the rhabdoid cell lines. After lentiviral delivery of the PSMA CAR to primary T cells, western blot analysis was performed to assess PSMA CAR expression in transduced T cells. A strong CAR-specific band was detected at the expected molecular weight using the CD3ζ antibody in CAR T cells but was absent in non-transduced T cells. In contrast, endogenous CD3ζ was present at comparable levels in both cells (Figure 3B). Meanwhile, flow cytometry analysis further confirmed over 80% transduction efficiency, demonstrating robust PSMA CAR expression on the surface of CAR T cells (Figure 3C). We then used these CAR T cells in a cell cytotoxicity assay at various target-to-effector (T: E) ratios. At a T: E ratio of 1:2, PSMA CAR T cells eradicated nearly all STM91-01 cells and approximately 90% of TTC709 cells, while exhibiting minimal activity against TTC1240 cells (Figure 3D). This indicates that PSMA CAR T cells efficiently eliminate PSMA-positive MRT cells with high specificity, but not PSMA-negative cells. In contrast, non-transduced T cells showed no cytotoxicity toward any of the rhabdoid cell lines (Figure 3D). Additionally, we collected cell culture supernatants and measured IFN-γ secretion by the CAR T cells. The cytokine secretion pattern mirrored the cytotoxicity results, further confirming that T cell activation is dependent on target recognition through PSMA (Figure 3E).

Figure 3.

PSMA CAR T cells effectively eradicated PSMA-positive rhabdoid cells in vitro

(A) Schematic diagram of the PSMA CAR T vector. (B) Expression of PSMA CAR in primary T cells as determined by western blot analysis using a CD3ζ antibody. The arrow marks the endogenous CD3ζ. (C) Flow cytometry analysis assessing cell surface expression of PSMA CAR on engineered CAR T cells. (D) In vitro cytotoxicity assay. Luciferase-expressing rhabdoid cells were incubated with PSMA CAR T cells or non-transduced T cells at various target-to-effector (T:E) ratios. p values were calculated by comparing each condition with TTC1240 at the corresponding T:E ratio. Group differences were evaluated using multiple unpaired t tests with FDR correction. N = 3. (E) INF-γ levels were measured by ELISA using the same experimental setup as in (D).

PSMA CAR T cells effectively eradicate PSMA-positive RTs in vivo

To further evaluate the antitumor efficacy of PSMA-targeted CAR T cells, we employed two in vivo xenograft models using luciferase-expressing PSMA-positive RT cell lines, STM91-01 and TTC709.

STM91-01 was initially propagated in vivo through subcutaneous implantation before the cell line was established,¹⁷ therefore we used a subcutaneous model to provide an appropriate context for evaluating its growth in vivo. In this model, NSG (NOD scid gamma) mice were injected with STM91-01 tumor cells and treated with either mock-transduced T cells or PSMA CAR T cells (Figure 4A), and tumor progression was monitored weekly by bioluminescence imaging. As shown in Figure 4B, mice treated with mock-transduced T cells exhibited progressively increasing tumor-associated luminescence over time. In contrast, mice receiving PSMA CAR T cells showed complete loss of luminescent signal as early as 2 weeks post-T cell injection (Figure 4C). In addition to bioluminescence imaging, caliper-based tumor volume measurements (Figure 4D) also demonstrated significant tumor regression in PSMA CAR T-treated mice compared to the mock T cell-treated group. These results indicate that PSMA CAR T cells effectively and specifically eliminate PSMA-positive STM91-01 xenografts in vivo, leading to complete tumor remission in this model.

Figure 4.

PSMA CAR T cells eradicate STM91-01 subcutaneous xenograft tumors in vivo

(A) Schematic diagram of the mouse injection and imaging timeline. (B) Bioluminescence imaging of tumor burden in mice treated with mock-transduced T cells (N = 5) or PSMA CAR T cells (N = 5). (C) Quantification of bioluminescence from (B). (D) Tumor burden assessment through tumor volume measurements. For (C) and (D), group differences at each time point were assessed using multiple t tests with false discovery rate (FDR) control and the two-stage step-up method of Benjamini, Krieger, and Yekutieli. p values were calculated by comparing PSMA CAR T cell treatment with mock T cell treatment. N = 5.

To further test PSMA CAR T cell efficacy in a disseminated disease setting, we used the TTC709 cell line in an intraperitoneal model. Derived from an extrarenal soft tissue RT, TTC709 reliably engrafts in the peritoneal cavity, making it a relevant model for advanced MRTs. Mice received an intraperitoneal injection of tumor cells followed by intravenous infusion of mock-transduced or PSMA CAR T cells (Figure 5A). Bioluminescence monitoring demonstrated a rapidly increasing tumor burden in the mock group, whereas PSMA CAR T-treated mice exhibited a sharp decline in luminescence signal over 3 weeks (Figure 5B). Quantitative analysis confirmed a significantly lower tumor burden in the CAR T group at weeks 2 and 3 (p < 0.001; Figure 5C). Survival analysis further revealed a substantial difference, as all mice in the control group died by week 5, while over 60% of PSMA CAR T-treated mice survived beyond week 7 (Figure 5D).

Figure 5.

PSMA CAR T cells effectively eliminate TTC709 intraperitoneal tumors in vivo

(A) Schematic diagram outlining the mouse injection and imaging timeline. (B) Bioluminescence imaging showing tumor burden in mice treated with either mock-transduced T cells (N = 7) or PSMA CAR T cells (N = 8). (C) Quantification of bioluminescence as shown in (B). (D) Kaplan-Meier survival analysis of mice treated with mock-transduced T cells or PSMA CAR T cells. Kaplan-Meier curves were compared using the log-rank (Mantel-Cox) test (log-rank p = 0.0002).

Collectively, these results demonstrate that PSMA CAR T cells robustly and specifically eliminate PSMA-positive RTs across multiple in vivo models, leading to significant tumor regression and prolonged survival.

PSMA CAR T cells infiltrate RTs in vivo

Next, we examined whether PSMA CAR T cells could infiltrate tumors following treatment. STM91-01 subcutaneous tumors were harvested 6 days after T cell injection and analyzed by immunohistochemistry (IHC). Hematoxylin and eosin (H&E) staining revealed similar overall tumor architecture in both mock and CAR T-treated groups; however, we observed cells with smaller, darker nuclei in the CAR T-treated tumors, consistent with the morphology of infiltrating T cells (Figure 6A, top). To confirm this, we performed IHC using an anti-human CD3 antibody and detected robust infiltration of human CD3⁺ T cells in PSMA CAR T-treated tumors, whereas CD3 staining was minimal in tumors treated with mock-transduced T cells (Figure 6A, middle). As a control, PSMA expression remained strong and uniform across both groups, indicating that CAR T cell activity did not lead to antigen downregulation (Figure 6A, bottom). Quantification of positively stained cells further confirmed a significant increase in CD3⁺ cell counts in the PSMA CAR T group, while PSMA levels were unchanged between groups (Figure 6B). These findings indicate that PSMA CAR T cells can efficiently infiltrate RTs, supporting their specificity and sustained functional activity in vivo.

Figure 6.

Immunohistochemical analysis of PSMA CAR T cells in rhabdoid tumors

(A) 8- to 10-week-old NSG mice were subcutaneously injected with 5 × 10⁶ STM91-01 cells. 7 days later, mice received either mock T cell or PSMA CAR T cell treatment. 6 days post-treatment, tumors were harvested for H&E staining and IHC staining with anti-CD3 and anti-PSMA antibodies. Scale bars, 100 μm. (B) Quantification of positively stained cells in the PSMA and CD3 IHC sections, respectively. The number of positive cells was quantified using the particle analysis algorithm and normalized to the baseline count from mock T sections. N = 3 tumors per group.

Discussion

RTs are highly aggressive pediatric cancers with limited treatment options and poor clinical outcomes. In this study, we identified PSMA as a potential therapeutic target in a subset of MRTs. Using PSMA-targeted CAR T cells, we demonstrated potent and selective antitumor activity against PSMA-positive rhabdoid cell lines in vitro and in two in vivo models, including both localized and disseminated tumors. We observed that CAR T cells successfully infiltrated and eliminated tumor cells without evidence of antigen escape. These findings provide a strong rationale for developing PSMA-directed therapies as a potential treatment strategy for children with PSMA-positive RTs.

The discovery also provides a dual opportunity for both targeted therapy and non-invasive tumor detection. Clinically approved PSMA-targeted positron emission tomography (PET) tracers, such as ⁶⁸Ga-PSMA-11 and ¹⁸F-DCFPyL, can be used to detect PSMA-expressing tumors, including those in the CNS.²³ This suggests that they may also be used for identifying PSMA-positive AT/RTs. These diagnostic approaches enable precise patient selection, ensuring that PSMA-targeted therapies, such as CAR T cell therapy, are applied only to patients likely to benefit. Moreover, the ability to visualize PSMA-positive lesions non-invasively allows for longitudinal monitoring of disease burden and therapeutic response. Therefore, combining PSMA-targeted immunotherapy with PET imaging represents a personalized and specific treatment approach for RTs. In addition, the identification of PSMA expression in a subset of RTs suggests that radioligand therapy, such as ¹⁷⁷Lu-PSMA-617, is also a viable therapeutic option. However, its efficacy can be limited by the physical half-life of the radionuclide and reduced effectiveness in tumors with low antigen density. In contrast, PSMA-targeted CAR T cell therapy may offer better sensitivity, as CAR T cells can expand, persist in vivo, and actively traffic to tumor sites, leading to sustained antitumor effects. Further investigation will be necessary to determine the relative benefits of these therapeutic strategies and to explore the potential for their complementary use.

Interestingly, analysis of clinical samples revealed that high PSMA expression was frequently observed in AT/RTs. While this pattern may be partially influenced by the limited availability of extracranial RT specimens, the consistent detection of PSMA in AT/RTs suggests a biologically relevant association. PSMA, also known as glutamate carboxypeptidase II, is expressed in the nervous system, particularly on the surface of glial cells such as astrocytes and Schwann cells.²⁴ It catalyzes the hydrolysis of the neuropeptide N-acetylaspartylglutamate (NAAG) into N-acetylaspartate and glutamate and plays an important role in modulating glutamatergic neurotransmission.²⁵ NAAG is the most abundant neuropeptide in the mammalian brain, and PSMA knockout mice exhibit altered NAAG levels, changes in brain lipid composition, and disruptions in neural metabolism.^26,27 Therefore, the finding of PSMA expression in AT/RTs may reflect the tumor’s origin from neural progenitor or glial-like cells and suggests that these tumors may aberrantly activate developmental or tissue-specific pathways. Similarly, PSMA is detected in kidney tissue, and STM91-01, an RT cell line derived from lung metastases of a primary kidney tumor, may reflect the tumor’s tissue of origin or represent a subset of PSMA-positive renal RTs. Although only non-CNS tumor models were used in this study, CAR T efficacy will be evaluated in CNS tumor cell lines or patient-derived xenograft (PDX) mouse models once they become available in the future.

Furthermore, a key driver of RT is the loss of SMARCB1, a core component of the SWI/SNF chromatin remodeling complex. SMARCB1 loss leads to widespread epigenetic dysregulation and the inappropriate activation of genes that are normally silenced in differentiated cells.^28,29 For instance, SMARCB1-deficient RTs show upregulation of the Shh signaling pathway and overexpression of GLI1, a key downstream effector.^4,30 Many AT/RT tumors also display elevated levels of EZH2, a histone methyltransferase that mediates gene repression via H3K27 trimethylation (H3K27me3).^31,32 This altered epigenetic landscape may contribute to the aberrant re-expression of developmentally restricted genes such as PSMA. Further studies will be needed to determine whether EZH2 or other epigenetic mechanisms directly regulate PSMA expression in RTs.

Collectively, the identification of PSMA expression in a subset of RTs reveals a promising therapeutic target in a disease with limited treatment options. By demonstrating that PSMA-directed CAR T cells selectively eliminate PSMA-positive rhabdoid cells both in vitro and in vivo, this study provides preclinical evidence supporting the feasibility of PSMA-targeted immunotherapy. Future studies using orthotopic kidney or brain models will assess the therapeutic potential of CAR T cells in relevant tissue-specific microenvironments. Overall, these findings establish a foundation for an antigen-specific therapeutic strategy in RTs and offer hope for improving outcomes in this highly lethal pediatric cancer.

Materials and methods

Cell lines

STM91-01 and TTC1240 were kindly provided by Dr. Timothy Triche at Children’s Hospital of Los Angeles. TTC709 was kindly provided by Dr. Charles Roberts at St. Jude Children’s Research Hospital and Dr. Bernard Weissman from UNC-Chapel Hill. HEK293T cells were obtained from the American Type Culture Collection (ATCC). STM91-01, TTC709, and TTC1240 were cultured in RPMI 1640 medium (Gibco, 11875093). Stable cell lines of STM91-01, TTC709, and TTC1240 expressing luciferase were generated by lentiviral transduction and maintained under blasticidin (Gibco, A1113903) antibiotic selection of 5 μg/mL. HEK293T cells were cultured in DMEM medium (Gibco, 11965092). All media were supplemented with 10% fetal bovine serum (FBS) (R&D system, S11150) and maintained at 37°C in a 5% CO2 incubator.

PSMA CAR T cell production

Buffy coats from healthy donors were purchased from the Gulf Coast Regional Blood Center to isolate PBMCs using density gradient centrifugation. Briefly, buffy coats were diluted with PBS containing 2% FBS and loaded onto 50 mL SepMate tubes (STEMCELL Technologies, 85450) containing density gradient medium Lymphoprep (STEMCELL Technologies, 18061). PBMCs were separated by centrifugation. The EasySep Human T Cell Isolation Kit (STEMCELL Technologies, 100–0204) was used to isolate primary T cells from freshly obtained PBMCs, following the manufacturer’s protocol. T cell transduction was performed using the TransDux Max Transduction Kit (SBI, LV860A-1), following the manufacturer’s protocol. The lentivirus-transduced cells were monitored daily and expanded in RPMI 1640 medium supplemented with 10% heat-inactivated FBS. IL-2 (100 U/mL) was added every other day.

Western blot analysis

Total protein was extracted from cells using a lysis buffer containing 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), and 0.1% NP-40. Protein concentrations were measured using the bicinchoninic acid protein assay (BCA) (Thermo Fisher Scientific, 23225). Equal amounts of protein were denatured in 4× Laemmli buffer (Bio-Rad, 1610747) and separated on a 4%–20% SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) gel (Bio-Rad, 4561091). Proteins were transferred onto a nitrocellulose membrane (Bio-Rad, 1704158), which was then blocked using Invitrogen’s membrane-blocking solution (000105). Membranes were incubated with the respective primary antibodies overnight, followed by incubation with appropriate secondary antibodies for 1 h. Signal detection was performed using the ChemiDoc Imaging System (Bio-Rad). The anti-PSMA antibody was developed in-house. Other antibodies used for western blot analysis include anti-CD3ζ-horseradish peroxidase (HRP) (Santa Cruz Biotechnology, sc-528127), anti-actin-Alexa 488 (Santa Cruz Biotechnology, sc-516606), and anti-mouse secondary antibody (Cell signaling 7076S).

PSMA expression quantification in clinical tumors

RNA sequencing data (raw transcript read counts) from 59 tumor samples collected at diagnosis (from 51 patients) were downloaded from the St. Jude Genomics Platform on DNAnexus. To evaluate PSMA expression, raw counts were normalized to CPM using the formula: CPM = (raw read count/total reads per sample) × 10⁶. Expression levels were then categorized into three groups based on commonly accepted thresholds: no or low expression (CPM <10), moderate expression (CPM between 10 and 50), and high expression (CPM >50).³³

Flow cytometry analysis

Cells were suspended in fluorescence-activated cell sorting (FACS) buffer and incubated with a biotin-labeled recombinant PSMA probe. The recombinant PSMA protein was purified from a baculoviral expression system and labeled using a biotinylation kit. To detect transduced PSMA CAR T cells, allophycocyanin (APC)-conjugated streptavidin (BD Biosciences, 554067) was used in conjunction with the biotinylated PSMA probe.

Cytotoxicity assay

RT cells were transduced with a lentiviral vector expressing luciferase and selected using blasticidin. Cells were seeded at a density of 1 × 10⁴ cells per well in a 96-well plate and incubated for 24 h. PSMA CAR T cells or control T cells were then added at varying T: E ratios and co-cultured for an additional 24 h. After incubation, supernatants were collected for IFN-γ quantification using an enzyme-linked immunosorbent assay (ELISA). To assess cytotoxicity, One-Glo luciferase reagent (Promega, E6110) was added to each well, and bioluminescence was measured using the Synergy Neo2 Multi-Mode Reader (BioTek).

ELISA

Supernatants collected from the in vitro cytotoxicity assay were used to quantify IFN-γ secretion by T cells or CAR T cells using the Human IFN-γ DuoSet ELISA kit (R&D Systems, DY285B). A 10% BSA solution was used for both blocking and reagent dilution. 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (BD OptEIA, 555214) was added for color development. All washing steps were carried out using the 50TS Microplate Washer (BioTek) with wash buffer consisting of 1× PBS and 0.5% Tween 20. Absorbance was measured at 450 nm using the Synergy Neo2 Multi-Mode Reader (BioTek).

Xenograft mouse models

All animal studies were ethically performed following protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Houston. 8- to 10-week-old NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice were selected for the study. Group sizes were based on our in vivo studies evaluating PSMA CAR T cell efficacy in prostate cancer models, where similar group sizes consistently yielded reproducible and statistically significant results. Mice were randomly assigned to treatment and control groups using simple randomization. For the subcutaneous tumor model, each group was subcutaneously injected with 5 × 10⁶ luciferase-expressing STM 91-01 cells in the right flank. For the intraperitoneal tumor model, 5 × 10⁶ luciferase-expressing TTC709 cells were injected into the lower left quadrant of each mouse. Mice were monitored daily for tumor growth. 7 days post-injection, mice were randomly grouped into control and experimental groups. Each group was injected intravenously with 1 × 10⁷ cells of either vector-transduced T cells or anti-PSMA CAR T cells. Luciferase-based bioluminescence imaging was performed with an IVIS (in vivo Imaging System) (PerkinElmer). D-luciferin (150 mg/kg) was injected subcutaneously for the subcutaneous tumor model or intraperitoneally for the intraperitoneal tumor model, and luciferase images were obtained after 5 min of exposure. Bioluminescence intensity was measured as total counts for pixels within the region of interest (ROI). The injected luciferase dose and ROI area were kept constant during measurements to reduce variability. Tumor volume was also measured using calipers and calculated using the formula: (length × width²)/2. No data points were excluded. To minimize potential confounders, all mice were housed under identical conditions, and tumor measurements and imaging were performed in a consistent order across groups.

IHC

STM 91-01 cells (5×10⁶) were subcutaneously injected into the right flank of mice. After 1 week, mice were randomly divided into two groups, with each group receiving either CAR T cells (1×10⁷) or empty vector-transduced T cells intravenously. Six days post-injection, the mice were euthanized following the ACO (Animal Care and Operations) protocol, and the subcutaneous tumors were excised. Tumors were fixed overnight in 4% paraformaldehyde (PFA) and then stored in 70% ethanol at 4°C until processed for IHC. Tumor sections were stained with H&E and with antibodies specific for PSMA and CD3. Images were captured using a Citation 5 imager (BioTek). For cell count normalization, ImageJ (Version 1.54) was used by thresholding antigen-specific staining intensity. The number of positive cells was quantified using the particle analysis algorithm and normalized to the baseline count from mock T cell sections.³⁴

Statistical analysis

Data in this study were analyzed using Prism 10.4.2 (GraphPad, San Diego, CA, USA). The sample size was set to a minimum of three independent experiments (biological replicates) and experimental findings were consistently reproducible. The statistical tests used for each figure, the comparison groups for each p-value, the number of biological replicates (N), and the references for data normalization are provided in the figure legends. Error bars represent standard mean unless otherwise stated. Differences were considered statistically significant at p < 0.05, with p values individually marked on the graphs. “NS” indicates not significant when p > 0.05.

Data and code availability

All experimental data generated or analyzed during this study are available from the corresponding authors upon reasonable request.

Acknowledgments

We thank other current and former members of the Feng lab for their advice and helpful discussions. This work was supported by a seed grant from the University of Houston, United States.

Author contributions

Q.F. and B.H. conceived the idea and designed the experiments. A.M., T.Z., J.W., T.N., C.Y., A.B., L.H., and X.J. conducted the experiments. A.M. and T.Z. analyzed the data. R.V. and M.M.J. provided consultation. Q.F., A.M., and T.Z. wrote the manuscript, and all authors contributed to the writing and provided feedback.

Declaration of interests

B.H. is an inventor on a patent related to the PSMA CAR T cells used in this study. Q.F. is an immediate family member of this inventor. The remaining authors declare no competing interests.