Introduction

CD19 chimeric antigen receptor (CAR) T cells offered a promising therapeutic strategy for the treatment of refractory and relapsed (R/R) B acute lymphoblastic leukemia (B-ALL).1 2 The US Food and Drug Administration has granted approval to two CD19 CAR-T products for the treatment of R/R B-ALL, tisagenlecleucel (Tisa-cel) and brexucabtagene autoleucel (Brexa-cel), which have demonstrated high rates (70–90%) of complete remissions in both pediatric and adult R/R B-ALL.3 4 However, long-term follow-up data reveal a significant challenge in enhancing treatment efficacy and extending survival against the emerging high relapse rate of approximately 50% within 1–2 years post-CAR-T therapy.5 6 The underlying factors contributing to recurrence following CD19 CAR-T cell therapy are multifactorial and not fully understood to date. Most of the relapses occurring soon after CAR-T therapy were composed of CD19+ relapse with poor persistence of CAR-T, while CD19− relapse were attributed to frameshift/missense CD19 mutations, alternative splicing of CD19 messenger RNA, hemizygous deletions spanning the CD19 locus and relapse related to lineage switch.7–10 Except for CAR-T persistence and tumor antigen modulation, evidence from both independent research groups and our experiments highlights a robust association between tumor microenvironment (TME), which actively participates in tumor progression and varies with disease status, and the efficacy of CAR-T therapy.11–13

It has been reported that CD19 CAR-T substantially modifies the TME by boosting the number and the activity of cytotoxic effector T cells. Furthermore, CAR-T cells could specifically enhance endogenous tumor-specific T-cell response.14 Our previous study revealed that endogenous T cells were triggered to produce more pro-inflammatory cytokines.12 Nevertheless, the detailed phenotypic alteration and the mechanism underlying the activation of endogenous T cells may be directly influenced by CAR-T mediated antigen-dependent tumor lysis or facilitated by antigen-presenting cell (APC) assistance in reshaped TME, which have not been well elucidated.

V-domain Ig suppressor of T-cell activation (VISTA) has been identified as an inhibitory ligand that attenuates T cell-mediated immune responses, serving as the earliest immune checkpoint (IC) regulator and non-redundant compared with programmed death-ligand 1 (PD-L1) and PD-L2.15 16 Murine tumor models demonstrated an upregulation of VISTA in TME, and VISTA monoclonal antibody (mAb) treatment impaired the suppressive character of the TME.17 Agents targeting VISTA efficiently induced T-cell activation by converting resting and exhausted cells into functional effector cells within TME.18 In line with the murine data, VISTA exhibits substantial upregulation in tumor-infiltrating immune cells across various cancer types and serves as a compensatory inhibitory pathway within the TME subsequent to IC antibody therapy in patients.19 20 In addition, VISTA exhibited high expression on APC, which regulates processes of activation and repression of the immune system with not yet completely clarified mechanisms. Both T cells and myeloid cells within TME following CAR-T cell therapy are implicated with their functionalities being associated with the efficacy of CAR-T treatment. Therefore, we hypothesized that the expression pattern of VISTA on host immune cells might determine the immune activity of TME, which would be correlated with the outcome of CAR-T cell therapy.

In the present study, we observed that downregulated VISTA expression on endogenous T cells in TME following CAR-T mediated tumor lysis promoted intrinsic T-cell activation and immune function both in hematopoietic and solid tumor models. However, in interferon (IFN)-γ signaling deficient TME, high levels of VISTA persist, leading to reduced cellular activity and toxicity of endogenous T cells, and shortened mouse survival. The expression pattern of VISTA in dendritic cell (DC) was similar to that of endogenous T cells, and blockade of VISTA signaling could enhance the proliferation and function of DC-mediated homologous T cells. These findings provide valuable insights into the impact of the TME on CAR-T therapy and its association with VISTA-mediated immune response. Also, they facilitate the development and supplementation of IC inhibitors combined with CAR-T therapy.

Materials and methodsMice

6−8 weeks-old male C57BL/6J and BALB/c mice weighing 18–24 g were purchased from Beijing Vital River Laboratory Animal Technology. IFN-γ−/− (B6.129S7Ifngtm1Ts/J) and IFN-γR1−/− (B6.129S7-Ifngr1tm1Agt/J) mice, kindly provided by Professor Yin (Jinan University), were bred in our animal facility under specific pathogen-free conditions at Xuzhou Medical University. C57BL/6N mice were provided by The Animal Center of Xuzhou Medical University. All experiments were approved by the Medical Ethics Committee of the Xuzhou Medical University (IACUC Issue No. 202 208S094).

B-ALL mice model establishment and evaluation

The first-generation B-ALL cells were generated by infecting bone marrow cells with a retrovirus encoding P210-NGFR. Mice were examined and sacrificed in case of prostration, tousled hair, severe weakness, and paralysis of both lower limbs. B-ALL cells were identified as CD11b−CD19+NGFR+. B-ALL cells were amplified and purified in vivo for three generations in C57BL/6J mice receiving 4.5 Gy 60Co irradiations. By the third generation of B-ALL cells, the percentages of CD11b−CD19+NGFR+ cell were 82.5%±13% and 92.1%±3.9% in spleen (SP) and bone marrow (BM), respectively. In this study, the B-ALL mice model was established by intravenous injection of 0.5–2×105 of the third-generation B-ALL cells in recipients after 4 Gy 60Co irradiation.

CD19 CAR-T and hCAIX CAR-T cell generation

The CAR construct and retrovirus package were described in the previous study.12 The CD19 CAR construct consists of the anti-murine CD19 single-chain fragment variable (scFv) domain, the transmembrane and intracellular domains of CD28, and the CD3z intracellular domain. The hCAIX CAR is composed of anti-human CAIX scFv, 4-1BB co-stimulatory domain, and CD3z intracellular domain. Additionally, the retroviral vector encodes green fluorescent protein (GFP) to evaluate transduction efficiency and identify CAR-T cells. In brief, splenic CD3+ T cells from male C57BL/6, IFN-γ−/− B6 and BALB/c mice were isolated and purified using EasySep negative selection reagents according to the manufacturer’s instructions (STEMCELL Technologies). Purified T cells (2×106 /mL) were activated in 24-well plate pre-coated with anti-CD3 mAb (2 µg/mL; clone17A2; BioGems), supplemented with soluble anti-CD28 mAb (2 ug/mL; clone37.51; BioGems) and hIL-2 (100 U/mL, PeproTech). Spin infections were performed 24 hours after T-cell activation using CD19 CAR or hCAIX CAR retroviral particles. Infected T cells were cultured for an additional 2 days in the presence of hIL-2 and the percentage of GFP-positive CAR-T cells was analyzed using flow cytometry.

CAR-T therapy and re-challenge experiment in B-ALL mice

B-ALL models were established in C57BL/6J, IFN-γ−/− and IFN-γR1−/− mice, respectively. Three days later, the mice were randomly allocated into groups and administered a single dose of either wild-type (WT) or IFN-γ−/− CD19 CAR-T cells (0.5–1×106/mouse) via tail vein injection, while activated syngeneic T cell were given as the control group. The day of CAR-T cell administration was considered as day 0. Daily observations and recordings were conducted to monitor survival rates, body weight changes, and clinical score throughout the experiment duration. The tumor burden analysis was performed using flow cytometry. Mice with long-term survival after CD19 CAR-T therapy were re-challenged with 1×105 B-ALL cells per mouse. Prior to tumor cell re-challenge, the complete elimination of tumors was confirmed throughflow cytometry.

Murine lymphoma model establishment and therapy

Murine lymphoma A20 cell line, preserved in our laboratory and cultured in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% fetal bovin serum (FBS), was evaluated for CD19 expression by flow cytometry. BALB/c mice were exposed to sublethal irradiation (3.5 Gy 60Co) and subsequently received a subcutaneous injection of CD19+A20 cells (10×106 /mouse) into the subepidermal layer on the right underarm region, thereby establishing B-cell lymphoma models after a 4-hour interval. A20 cell seeded mice were randomly allocated to receive either activated T cell or CD19 CAR-T cell therapy after 8 days. The day of CAR-T cell administration was designated as day 0, and tumor size and body weight changes were monitored every 2 days.

Murine melanoma model establishment and therapy

The B16F10 cell line overexpressing human CAIX antigen was engineered and used to establish a subcutaneous transplantation melanoma model in C57BL/6N mice. On day 0, 100 µL of 2×105 B16F10 cells were injected into the subepidermal layer on the right side of the dorsal region. On day 9, marrow clearance was achieved by pretreating with 200 mg/kg cyclophosphamide. On day 10, tumor-bearing mice were randomly divided into two groups named as untreated control group and the CAR-T cell treated group (5×106 cells/mouse). After a period of 30 days from tumor seeding, the mice were euthanized for subsequent analysis. Tumors, draining lymph nodes (dLN), and spleens were dissected and processed into single-cell suspensions for flow cytometry analysis of endogenous T-cell quantity and functionality.

Endogenous T-cell isolation and tumor response assessment

Endogenous T cells labeled with GFP−CD3+ markers were isolated from splenic cells using flow cytometry from B-ALL B6 mice treated with CD19 CAR-T (trained T) or untreated (untrained T) on days 7–10. For ex vivo experiments, the purified trained and untrained endogenous T cells were co-cultured with B-ALL cells at an Effector:Target ratio of 1:1 for 24 hours. Similarly, endogenous untrained or trained T cells by CD19 CAR-T were isolated from A20 BALB/c mice, and subsequently co-cultured ex vivo with CD19+ A20 or CD19−A20 cells, respectively. The tumor proportions were quantified using flow cytometry. Additionally, these isolated trained and untrained T cells (4×105/mouse) were adoptively transferred into two groups of B-ALL mice on day 0, respectively. The proportion of B-ALL cells in peripheral blood was detected on day 16, and the survival rate of mice was recorded. The quantity, phenotype characteristics, and cytokine production of T cells in adoptively transferred mice were detected by flow cytometry.

PCR

The total genomic DNA extraction of untrained/trained endogenous T cells and CD19 CAR-T cells was performed using a DNA isolation kit (TIANGEN). Two pairs of primers were designed to amplify different sizes of CD19 scFv fragments: Forward 5’-AAGAACAAGGCCACCCTGAC-3’; Reverse 5’-GTAGTAGCCGCCGTAGATGC-3’, and Forward 5’- TGAAGCTGTCCTGCAAGGTG-3’; Reverse 5’- CAGGGTGGCCTTGTTCTTGA-3’. Primers for the EGFP fragments (Forward 5’-ACGACGGCAACTACAAGACC-3’; Reverse 5’-GCCTTGATGCCGTTCTTCTG-3’) were used to verify its consistency with scFv. Amplification was carried out using 2×Hot Start Taq PCR Mix (TIANGEN), with an addition of 200 ng of extracted DNA to the final volume of the PCR mixture (25 µL). Subsequently, the amplified DNA product underwent electrophoresis analysis on a gel containing agarose at a concentration of 2%, followed by visualization using the TANON MINI Space 2000 Multifunctional Gel Image Analysis System (Tanon Science & Technology).

Western blotting

For western blotting, antibodies specific to anti-VISTA, anti-STAT3, and anti-STAT3 (pY705) were purchased from Cell Signaling Technology. Antibodies specific to anti-STAT1 and anti-STAT1(pY701) were purchased from BD Bioscience. Antibodies specific to β-actin were purchased from Proteintech. All primary antibodies were used at a dilution of 1:2,000. HRP-conjugated secondary antibodies were diluted at 1:10,000 (Proteintech). B-ALL cells co-cultured with naïve T cells, activated T cells, and CAR-T cells for 6 hours, respectively, were lysed in RIPA buffer (VICMED) supplemented with Protease Inhibitor (100×; VICMED) and Phosphatase Inhibitor Cocktail (100×; Proteintech) at the indicated time points. Total lysates were cleared by spinning at 12,000 rpm for 10 min. 25 mg of lysates were subjected to SDS-PAGE and western blotting. Proteins were detected using Western ECL Substrate (UUBIO) and Tanon 5200 Chemiluminescent Imaging System (Tanon Science & Technology).

Cytotoxicity assays

B-ALL cells were co-cultured with WT or IFN-γ−/− CD19 CAR-T cells as effectors at a 1:1 ratio in 48-well plates for in vitro killing experiments. Anti-CD3 and anti-CD28 pre-stimulated T cells were used as controls. The B-ALL cell alone group was included to assess cell growth in the medium. Flow cytometry was employed to monitor the lysis of B-ALL cells after 6, 12, 24, and 48 hours.

Bone marrow-derived dendritic cell induction

BM cells were isolated from the femurs, tibias, and sciatic bone of female C57BL/6 J mice aged 6– weeks. The cells were cultured in RPMI-1640 medium supplemented with 10% FBS, 1% penicillin/streptomycin, and 20 ng/mL rmGM-CSF (STEMCELL Technologies) for 3 days. On day three, all non-adherent cells were removed and fresh supplemented medium was added for an additional 3 day culture. Immature dendritic cells (DCs) were harvested on day 6. Immature DCs were stimulated with 100 ng/mL lip0polysaccharides (LPS) in a complete medium for a duration of 24 hours to induce mature DCs. Flow cytometry analysis was performed to identify bone marrow-derived dendritic cell (BMDC) as CD11b+ CD11c+.

T-cell proliferation and function

To detect the effect of VISTA expression of DCs on T-cell proliferation, T cells were labeled with CFSE (carboxyfluorescein diacetate, succinimidyl ester) (BioLegend) at a final concentration of 10 mM for 25 min at 4°C. Subsequently, they were pre-activated with plate-bound anti-CD3 (clone17A2; BioGems) at a concentration of 2 µg/mL for 24 hours under a condition of 37°C, 5% CO2. Following this, the T cells were co-cultured with mature or immature DCs at a ratio of 5:1 for 3 days. NSC622608 (Topscience) as a specific inhibitor of VISTA was added or not at a concentration of 5 µM to the co-culture system. The proliferation rate of T cells was assessed by measuring CFSE intensity using flow cytometry. To assess T-cell function, pre-activated T cells were co-cultured with mature or immature DCs at a ratio of 5:1 for 2 days. Brefeldin A (10 µg/mL, Invitrogen) was added to the medium during the last 6 hours of incubation to inhibit protein transport. Subsequently, flow cytometry was employed to measure T-cell cytokines including IFN-γ, tumor necrosis factor (TNF)-α, and granzyme B.

Flow cytometry and antibodies

Single-cell suspensions were obtained from the BM, spleen, and liver. Prior to flow cytometry analysis, staining was performed using the following monoclonal antibodies: CD3 FITC (fluorescein isothiocyanate) (clone17A2; BioLegend), CD3 BV650 (clone17A2; BioLegend), CD3ε PE/Cy5 (clone145-2C11; Elabscience), CD4 PE/Cyanine7 (cloneGK1.5; BioLegend), CD4 Alex Flour700 (cloneRM4-5; BioLegend); CD8a APC (allophycocyanin) (clone53-6.7; BioLegend), CD8 APC/Cyanine7 (clone53-6.7; BioLegend); CD11b PB (Pacific Blue) (cloneM1/70; BioLegend), F4/80 BV605 (cloneBM8; BioLegend), MHC-II AF700 (cloneM5/114.15.2; BioLegend), CD19 PE-cf594 (clone6D5; BioLegend), CD19 BV510 (clone6D5; BioLegend), CD11c PE-cf594 (cloneN418; BioLegend), IFN-γ PE/Cyanine7 (cloneXMG1.2; BioLegend), TNF-α APC (cloneMP6-XT22; BioLegend); Granzyme B eFlour450 (cloneM5/114.15.2; Thermo Fisher Scientific); CD62L PB (cloneMEL-1; BD Biosciences), CD44 APC/eFlour780 (cloneIM7; BD Biosciences); PD-1 PE/Cyanine7 (clone RMP1-30; BioLegend), TIGIT (Vstm3) PE (clone1G9; BioLegend); NGFR (CD271) APC (cloneME204; BioLegend), VISTA (PD-1H) PE (cloneMIH63; BioLegend), CD69 BV510 (cloneH1.2F3; BioLegend), and CD20 APC/Cyanine7 (clone SA275A11; BioLegend). For intracellular cytokine staining, cells were stimulated with PMA (50 ng/mL, Sigma) and Ionomycin (750 ng/mL, Sigma) in the presence of Brefeldin A (10 µg/mL, Invitrogen) at 37°C for 5 hours. Analyses were performed using a BD LSRFortessa Fortessa Flow Cytometer (RRID: SCR_019601), and the data was analyzed using FlowJo V.10.8.1 (BD Biosciences).

Statistical analysis

Statistical analyses and graphs were performed using GraphPad Prism V.9 software (GraphPad Prism, San Diego, RRID: SCR_002798). The data exhibited normal distribution and were presented as mean±SD or mean±SEM. Student’s t-test was employed for comparing two groups with homogeneous variance, while non-parametric tests were used for unequal variance. One-way analysis of variance was conducted for multiple group comparisons. Survival rates were assessed using the log-rank test. Correlation analysis was carried out and graphs were generated using R V.4.2.1 software. After analyzing the correlation between variables across multiple data sets, trends in each data set were visualized using gglot2 (3.3.6) and ggtext (0.1). Spearman’s rank correlation coefficient was used to assess the relationship between two variables. A p value of <0.05 was considered statistically significant.

ResultsVISTA expression was downregulated on endogenous T cells in B-ALL mice after CD19 CAR-T cell treatment

We took advantage of an immunocompetent and syngeneic model receiving C57BL/6 mice-derived BCR-ABL P210+ B-ALL cells to assess CD19 CAR-T function (figure 1A). Significant longer survival was investigated in mice treated with CD19 CAR-T cells than in T cells injected control mice (figure 1B). Percentages of B-ALL cells that were labeled by CD19 and NGFR were dramatically reduced after CAR-T cell administration (figure 1C). In figure 1D VISTA level was almost undetectable gating on B-ALL, which was consistent with the previous reports that B cells were negative for VISTA expression.15 The percentages of endogenous CD4+ and CD8+ T cells were consistently observed higher in CAR-T-treated mice, as anticipated (figure 1E).12 Compared with levels of VISTA on endogenous CD4+ T and CD8+ T cells in CAR-T untreated mice, downregulated VISTA expression was demonstrated in BM, SP, and liver after CAR-T injection (figure 1F). These data suggested that the reduced VISTA might deactivate the immunosuppressive function of endogenous T cells to assist CAR-T cell-induced tumor clearance.

Figure 1

Decreased VISTA expression on endogenous T cells in B-ALL mice after CAR-T cell treatment. (A) The experiment design. (B) Survival of mice bearing B-ALL tumors following T cells or CAR-T cell treatment. (C) Percentages of B-ALL cells within BM and SP on day 23. Each dot represents one mouse. (D) The representative flow cytometry data showing the proportion and VISTA expression of B-ALL cells in BM and SP. (E) Percentages of endogenous CD4+ T and CD8+ T cells of BM and SP. (F) The fluorescence intensities of VISTA on endogenous CD4+ and CD8+ T cells in BM, SP and liver. The independent experiment was replicated three times. Data were mean±SD or mean±SEM, and unpaired student’s t-test was used. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. B-ALL, B acute lymphoblastic leukemia; BM, bone marrow; CAR, chimeric antigen receptor; FMO, fluorescence minus one; SP, spleen; VISTA, V-domain Ig suppressor of T-cell activation.

Reduction of VISTA on syngeneic T cells was not achieved solely through CAR-T cell-induced tumor lysis

It aims to clarify if the variation of VISTA is directly caused by the killing effect of CAR-T, or if it is dependent on the presence of TME. We next examined the expression of VISTA on syngeneic-activated T cells co-cultured with CAR-T and B-ALL in vitro. Flow cytometry analysis was performed as the gating strategy shown in figure 2A. The CD19 CAR-T cells were distinguished based on GFP expression, while the GFP− cells were further segregated by CD11b expression to differentiate between myeloid and lymphocytic populations. Then the GFP− syngeneic T cells and NGFR+ B-ALL cells were identified using surface markers CD4, CD8 and CD19, respectively. The expression of VISTA was assessed in GFP+ CAR-T cells and GFP− syngeneic T cells (figure 2A). The death rate of B-ALL exceeded 90% after co-culturing for 6 hours, indicating that the CAR-T cells induced a potent and rapid tumor lysis compared with activated T cells (figure 2B). The results in figure 2C demonstrated that activated T cells exhibited a modest production of IFN-γ and TNF-α, which were associated with tumor cell death or senescence,21 22 whereas significantly elevated levels were observed in the CAR-T groups. Furthermore, no difference in VISTA expression was observed between the activated T group and syngeneic GFP− T cells in terms of tumor lysis induced by GFP+ CAR-T cells (figure 2D). The similar pattern of VISTA expression of activated T cells in the presence or absence of tumor lysis in vitro might reflect non-optimal culture conditions that fail to mimic the complicated TME in vivo. In other words, the in vitro experiments confirmed that the VISTA expression on homologous activated T cells was not solely dependent on CAR-T cytotoxicity, but TME was essential for the remodeling of endogenous T cells.

Figure 2

CAR-T cell-induced tumor lysis was not sufficient to reduce VISTA expression on syngeneic activated T cell. (A) Gating strategy of flow cytometry. (B) Percentages of B-ALL cell death in vitro. B-ALL cell alone or co-cultured with purified T-cell stimulated with anti-CD3 and anti-CD28 (Tactivated) in vitro were used as the control and T-cell groups, respectively. (C) The production of cytokines was evaluated in the co-culture cellular system for 6 hours. Naïve T cells were included as a control group without antibody activation (Tunactivated). (D) The representative flow cytometry histograms of VISTA levels. The T cells lacking CAR chimerism in the CAR-T groups were designated as GFP− T cells. Three independent experiments were repeated. B-ALL, B acute lymphoblastic leukemia; CAR, chimeric antigen receptor; FMO, fluorescence minus one; GFP, green fluorescent protein; VISTA, V-domain Ig suppressor of T-cell activation.

Augmented endogenous T-cell responses within the TME following CAR-T cell therapy in solid tumor models

To ascertain the generalizability of endogenous T-cell activation within TME subsequent to CD19 CAR-T cell administration, we evaluated murine lymphoma established by subcutaneously seeded A20 cells. The administration of CD19 CAR-T cells led to a significant regression of lymphoma in mice (figure 3A). Notably, there was an increase in the number of endogenous T cells, as depicted in figure 3B. Furthermore, the activation marker CD69 exhibited higher expression on endogenous T cells within the tumor and SP, while VISTA levels were reduced and PD-1 expression showed various changes (figure 3C). Additionally, there was an elevation in tumor-killing-associated cytokines secreted by endogenous T cells (figure 3D). Furthermore, to investigate the impact of diverse CAR-T cell on endogenous T-cell function within TME, we employed hCAIX CAR-T therapy in murine hCAIX+ melanoma models. Consistent findings revealed a significant increase in endogenous T-cell populations within the tumor and draining LN (figure 3E). The activation of endogenous T cells was indicated by the upregulation of CD69, while VISTA and PD-1 exhibited downregulation in both tumor and SP samples (figure 3F). IFN-γ and granzyme B production increased in tumor and dLN-derived T cells (figure 3G). In summary, the above data demonstrated that CAR-T therapy elicits the activation of endogenous T cells within the TME, regardless of the specific type of CAR-T used, encompassing both hematologic malignancies and solid tumors.

Figure 3

Enhanced endogenous T-cell responses within the tumor microenvironment following CAR-T cell therapy in solid tumor models. The A20 lymphoma mice were subcutaneously inoculated on −8 days and subsequently treated with or without CD19 CAR-T cells as day 0 (n=9). (A) Tumor size, weight and volume in CAR-T treated and untreated lymphoma mice. (B) The percentages of endogenous T-cell infiltrated in tumor tissue and spleen on day 11. (C) The representative flow cytometry histograms of CD69, VISTA, and PD-1 of endogenous T cells from lymphoma mice. (D) Cytokine production of endogenous T cells in tumor and spleen of lymphoma mice with or without CD19 CAR-T cell therapy. (E–G) exhibited the data from hCAIX CAR-T treated and untreated melanoma murine model for 30 days (n=6). (E) The percentages of endogenous T cells infiltrated in the tumor, in the draining LN and spleen were evaluated, respectively. (F) Expression of CD69, VISTA, and PD-1 were detected in endogenous T cells. (G) The secretion of cytokines in endogenous T cells from tumor tissue, draining LN and spleen were exhibited. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001. CAR, chimeric antigen receptor; dLN, draining lymph nodes; FMO, fluorescence minus one; PD-1, programmed cell death protein-1; SP, spleen; VISTA, V-domain Ig suppressor of T-cell activation.

Endogenous T cell-derived IFN-γ maintains the therapeutic effect of CD19 CAR-T and regulates VISTA expression in TME

The production of IFN-γ by CAR-T cells and the responsiveness of host immune cells to IFN-γ was found to be critical for remodeling tumor immune landscape, promoting a more activated and less suppressive TME.23 Therefore, IFN-γ−/− CAR-T cells were employed to investigate the relationship between VISTA and CAR-T efficacy in the presence or absence of IFN-γ signaling. The transfection efficiency was evaluated in both WT and IFN-γ−/− CAR-T cells, with no significant statistical difference observed (online supplemental figure 1A). In vitro lysis of B-ALL was detected in a co-cultured system involving IFN-γ−/− CAR-T or WT CAR-T cells, while tumor alone or activated T cells were used as controls, respectively. Surprisingly, IFN-γ−/− CAR-T exhibited comparable cytotoxicity to WT CAR-T cells in vitro (figure 4A). To validate the tumor clearance efficacy in vivo, we employed the same B-ALL murine models established in WT and IFN-γ−/− B6 mice for treatment with either WT or IFN-γ−/− CAR-T cells. As depicted in figure 4B, all recipient mice treated with either WT or IFN-γ−/− CAR-T survived. In comparison to untreated and T-cell control groups, there was no discernible benefit observed after administration of IFN-γ−/− CAR-T cells in IFN-γ−/− mice. Body weight change and clinical score confirmed that there were no differences between mice treated with WT or IFN-γ−/− CAR-T cells. However, IFN-γ−/− recipients displayed lower body weight and higher clinical score (figure 4C–D). These findings suggest that the efficacy of CAR-T cells is not solely reliant on their production of IFN-γ, and it is plausible that IFN-γ secreted by cells within the TME may serve as a compensatory factor. Consequently, we investigated endogenous T cells as a crucial cellular source of IFN-γ. The results demonstrated that T cells in the WT CAR-T mice exhibited the highest level of IFN-γ production. Although the IFN-γ levels were comparatively lower in the IFN-γ−/− CAR-T group compared with the WT group, they still remained significantly higher than those observed in the IFN-γ−/− recipient group (figure 4E). The VISTA fluorescence intensity on endogenous CD4+ and CD8+ T cells was found to be increased in IFN-γ−/− recipients, but not in WT mice following IFN-γ−/− CAR-T treatment (figure 4F). In line with these findings, the downstream proteins pSTAT1 and pSTAT3 of IFN-γ signaling exhibited a gradual increase in naïve T cells, activated T cells, and CAR-T cells, concomitant with a decrease in VISTA expression; however, no significant changes were observed in total STAT1 and STAT3 levels (online supplemental figure 1B). Additionally, lower levels of granzyme B and TNF-α were observed in both WT and IFN-γ−/− recipients injected with IFN-γ−/− CAR-T cells (figure 4G). The above data suggests that the regulation of IFN-γ on the TME may exert a significant influence on the expression of VISTA and the functionality of endogenous T cells, thereby highlighting their dual role as both compensatory sources of IFN-γ for CAR-T cells and valuable allies in tumor clearance.

Figure 4

The production of IFN-γ within tumor microenvironment exerts an influence on the VISTA expression and functionality of endogenous T cells. (A) Evaluation of the killing effect of WT and IFN-γ−/− CAR-T cells in vitro. Two independent experiments were repeated. (B–D). The survival rate, body weight changes and clinical score were recorded in recipients treated with WT or IFN-γ−/− CAR-T cells. (E) The representative flow cytometry of IFN-γ production and statistical analysis of endogenous CD8+ T cells in BM. (F) The fluorescence intensities of VISTA on endogenous CD4+ and CD8+ T cells of BM. (G) The representative flow cytometry plots and statistical analysis showed secretion of TNF-α and granzyme B from CD8+ T cells. The independent in vivo experiment was repeated twice. Data were mean±SD or mean±SEM, and one-way analysis of variance was used to do statistical analysis. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. BM, bone marrow; CAR, chimeric antigen receptor; IFN, interferon; TNF, tumor necrosis factor; VISTA, V-domain Ig suppressor of T-cell activation; WT, wild-type.

Characteristics of endogenous T cells within TME following CAR-T therapy

To elucidate the value of endogenous T cells trained by CD19 CAR-T, we assessed the cytotoxicity of isolated GFP−CD3+ T cells through ex vivo co-culture with B-ALL and adoptive transfer to B-ALL mice, respectively. The tumor-clearing capacity of CD19 CAR-T-trained endogenous T cells was more pronounced compared with untrained T cells, as evidenced by both the co-culture system (figure 5A) and adoptive transfer experiments (figure 5B). To exclude the potential contamination of CAR-T cell-induced tumor lysis, we evaluated the purification of endogenous T cells and amplified fragments of CD19 scFv; however, no amplification of scFv was observed in trained endogenous T cells (online supplemental figure 2A,B). Furthermore, ex vivo co-cultures were performed using untrained and trained endogenous T cells derived from CD19 CAR-T treated A20 mice with CD19+ A20 or CD19− A20 cells, respectively. Consistent with findings in B-ALL B6 mice, trained T cells demonstrated tumor cytotoxicity, whereas CD19 CAR-T exhibited no impact on CD19− A20 cells (online supplemental figure 2C,D). Importantly, CD19 CAR-T trained endogenous T cells injected into B-ALL mice confer significant benefits (figure 5C). Moreover, significantly higher levels of CD69 were observed in the BM and SP of B-ALL mice that received trained endogenous T cells, while lower levels of VISTA and PD-1 were detected (figure 5D). These findings indicate the enhanced tumor reactivity of endogenous T cells following CD19 CAR-T therapy, characterized by heightened expression of active markers and increased cytokine production (online supplemental figure 3A). Furthermore, based on the surface markers CD44 and CD62L, endogenous T cells exhibited three distinct subsets: effector cells (CD44+CD62L−), memory cells (CD44+CD62L+), and naïve-like cells (CD44−CD62L+) in online supplemental figure 3B and figure 5E. Effector cells of CD4+ T predominated with no observed differences among the groups in the BM. Within the spleen, reduction in IFN-γ levels from CAR-T cells resulted in decreased percentages of effector and memory T cells, while an increase was observed in the naïve-like CD4+ T cell (figure 5E). In mice treated with IFN-γ−/− CAR-T, a decrease of CD8+ memory T subset in BM was evident, and this reduction was more pronounced within the TME completely lacking IFN-γ signaling. Additionally, in IFN-γ−/− recipients there was a decrease in splenic effector CD8+ T cell accompanied by an increased proportion of naïve-like T subsets (figure 5E). Endogenous T cells were further investigated on B-ALL re-challenge in WT and IFN-γ−/− recipients following CAR-T treatment. The experimental design was depicted in online supplemental figure 4A. In CAR-T treated mice, longer survival, alleviated clinical manifestations and delayed disease occurrence were observed compared with the newly established B-ALL model (figure 5F). However, no difference was presented in CAR-T proportion among groups (online supplemental figure 4B). A higher proportion of memory and quiescent naïve-like endogenous CD4+ T cells were found in the BM of CAR-T treated mice with long-term remission, while most endogenous CD8+ T cells exhibited memory phenotypes. On tumor re-challenge, elevated levels of effector CD4+ and CD8+ T cells with higher cytokine secretion were induced in both wild-type and IFN-γ−/− recipients, independent of IFN-γ production by CAR-T (figure 5G and online supplemental figure 4C). The VISTA expression level was highest in memory CD4+ T cells and successively decreased in naïve-like and effector CD4+ T cells, while no significant difference was observed among CD8+ T-cell subsets (figure 5H). Figure 5I depicted the comprehensive profiles of endogenous T cells across groups, illustrating a close association between VISTA expression patterns and the composition and immune activity of T-cell subpopulations.

Figure 5

Tumor reactivity and phenotype characteristics of endogenous T cells. Endogenous T cells named as trained and untrained T cells, respectively, were isolated from CD19 CAR-T treated and untreated B-ALL mice on days 7–10. (A) Isolated endogenous T cells were co-cultured with B-ALL ex vivo for 24 hours and the percentages of B-ALL were detected. (B) Isolated trained and untrained T cells were adoptive transferred to B-ALL mice on day 0, and the burden of B-ALL in PBMC were detected on day 16. (C) The survival rate of B-ALL mice following adoptive transferred trained and untrained endogenous T cells. (D) The representative histograms and statistical analysis of MFI of surface markers CD69, VISTA, and PD-1 on T cells from endogenous T-cell adoptive transferred mice. (E). The compositions of the endogenous CD4+ and CD8+ T subsets in WT and IFN-γ−/− recipients receiving WT or IFN-γ−/− CAR-T cells, respectively. (F) Survival rate, the clinical score and disease occurrence of B-ALL cell re-challenged CAR-T Treated host were recorded. Hosts treated with CAR-T cells without re-challenge sustained remission and observation was concluded on day 42. (G) Representative flow cytometry plots showed the endogenous CD4+ and CD8+ T cells subsets in mice following re-challenge. (H) The levels of VISTA on endogenous CD4+ and CD8+ T subsets. (I) Heatmap analysis of the phenotypic characteristics of endogenous CD4+ and CD8+ T cells in mice following re-challenge. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. B-ALL, B acute lymphoblastic leukemia; BM, bone marrow; CAR, chimeric antigen receptor; FMO, fluorescence minus one; IFN, interferon; MFI, median fluorescence intensity; PBMC, peripheral blood mononuclear cell; cell; PD-1, programmed cell death protein-1; SP, spleen; VISTA, V-domain Ig suppressor of T-cell activation; WT, wild-type; .

The subpopulations of VISTA and PD-1 exhibit independent and distinct expression patterns

The IC molecular detection results revealed that PD-1 and T cell immunoreceptor with Ig and ITIM domains (TIGIT) were highly expressed on cells of IFN-γ–/– recipients with the shortest survival, whereas their expression levels were low on endogenous T cells of WT recipients after remission (figure 6A–B). In conjunction with the diverse VISTA expression patterns depicted in figure 4A, these findings suggest a diminished inhibitory function of ICs on endogenous T cells during the early remissive stage of CAR-T therapy. Whereas, in the absence of IFN-γ within the TME, endogenous T cells exhibit an immunosuppressive state regulated by ICs. Furthermore, prolonged remissive mice from WT CAR-T treated host showed low PD-1 expression and high VISTA expression on endogenous T cells, suggesting a resting state of them (figure 6C–D). However, the expression of PD-1 on endogenous T cells in CAR-T treated hosts was once again induced to be upregulated on re-exposure to B-ALL, while the level of VISTA exhibited varying degrees of decrease across all groups. Compared with the IFN-γ–/– CAR-T group, the reduction of VISTA on T cells in the WT CAR-T group was more pronounced, indicating a stronger immune activity of endogenous T cells (figure 6C–D). Then, analysis of T-cell populations using t-distributed stochastic neibor embedding (t-SNE) in IFN-γR1–/– recipients confirmed that VISTA and PD-1 were distributed among distinct subpopulations of T cells (figure 6E). Subpopulation localization analysis revealed a negative correlation between VISTA and CD44 expression, as well as a positive correlation between VISTA and CD62L expression. Conversely, PD-1 showed a positive correlation with CD44 expression and a negative correlation with CD44 expression (figure 6F).