Solid cancer-directed CAR T cell therapy that attacks both tumor and immunosuppressive cells via targeting PD-L1

Abstract

Chimeric antigen receptor (CAR) T cell therapy has encountered limited success in solid tumors. The lack of dependable antigens and the immunosuppressive tumor microenvironment (TME) are major challenges. Within the TME, tumor cells along with immunosuppressive cells employ an immune-evasion mechanism that upregulates programmed death ligand 1 (PD-L1) to deactivate effector T cells; this makes PD-L1 a reliable, universal target for solid tumors. We developed a novel PD-L1 CAR (MC9999) using our humanized anti-PD-L1 monoclonal antibody, designed to simultaneously target tumor and immunosuppressive cells. The antigen-specific antitumor effects of MC9999 CAR T cells were observed consistently across four solid tumor models: breast cancer, lung cancer, melanoma, and glioblastoma multiforme (GBM). Notably, intravenous administration of MC9999 CAR T cells eradicated intracranially established LN229 GBM tumors, suggesting penetration of the blood-brain barrier. The proof-of-concept data demonstrate the cytolytic effect of MC9999 CAR T cells against immunosuppressive cells, including microglia HMC3 cells and M2 macrophages. Furthermore, MC9999 CAR T cells elicited cytotoxicity against primary tumor-associated macrophages within GBM tumors. The concept of targeting both tumor and immunosuppressive cells with MC9999 was further validated using CAR T cells derived from cancer patients. These findings establish MC9999 as a foundation for the development of effective CAR T cell therapies against solid tumors.

Keywords: CAR T cells; MT: Regular Issue; PD-L1; T cells; chimeric antigen receptor; immunotherapy; solid tumor; tumor microenvironment.

Graphical abstract

Figure 1

MC9999 CAR T cells showed antigen-specific cytotoxicity against PD-L1-expressing breast cancer cells (A and B) CAR T cell cytotoxicity was evaluated by CD107a surface expression in a degranulation assay. MC9999 CAR T cells were incubated with MDA-MB-231 PD-L1 OE or MDA-MB-231 PD-L1 KO cells at an E:T ratio of 2:1. Analysis was gated on CD4⁺CAR T cells (A) or CD8⁺CAR T cells (B); non-CAR T cells from the same donor served as negative controls. (C) MC9999 CAR T cells were co-cultured with either MDA-MB-231 PD-L1 OE or MDA-MB-231 PD-L1 KO cells. After 72 h, the supernatants were collected and examined for granzyme B by ELISA. The data were plotted as the mean ± SEM of quadruplicate sampling and are representative of three independent experiments (∗∗∗p < 0.001; ns, no significance). (D) Target cells, either MDA-MB-231 PD-L1 OE or MDA-MB-231 PD-L1 KO cells, were seeded on the electronic microtiter plates (E-plates) for 24 h. After 24 h, MC9999 CAR T cells or non-CAR T cells were added to the target cells at an E:T ratio of 40:1. CI traces were collected in triplicate every 15 min during the co-culture, and changes in impedance were normalized to the 24-h time point. The results represent three independent experiments, with a representative dataset shown.

Figure 2

MC9999 CAR T cells elicited antitumor effects in an intramammary breast cancer model (A) Female NSG mice were given an intramammary injection of luciferase-labeled MDA-MB-231 tumor cells with either PD-L1 OE or PD-L1 KO at a dose of 1.0 × 10⁶ cells per mouse. Seven days after tumor challenge, mice were randomly divided into three groups. Each group (n = 5) received a single IV infusion of one of the following: PBS, non-CAR T cells (5 × 10⁶ total T cells per mouse), or MC9999 CAR T cells (2 × 10⁶ CAR T cells out of 5 × 10⁶ total T cells from the same donor/per mouse). Weekly IVIS imaging was performed to monitor tumor progression. The representative images demonstrated the changes in tumor burden over time. The results present two independent experiments using different donor T cells for generating CAR T cells. (B–D) Three separate Kaplan-Meier survival plots—MC9999 CAR T cells (B), non-CAR T cells (C), or PBS (D)—were generated to evaluate the treatment-associated overall survival. Log rank analysis revealed that MC9999 CAR T cell treatment significantly extended the overall survival in mice challenged with MDA-MB-231 PD-L1 OE comparing to those bearing MDA-MB-231 PD-L1 KO tumors (∗∗p = 0.01). No significant differences in survival were observed in the PBS or non-CAR T cells treatment group between these two tumor models.

Figure 3

MC9999 CAR T cells exhibited antigen-specific cytotoxicity against various PD-L1 expressing solid tumors In a CD107a degranulation assay, antigen-specific cytotoxicity of MC9999 CAR T cells was assessed against three solid tumor cell lines with CD8 MC9999 CAR T cell results shown: Calu-1 lung cancer (A), SH-4 melanoma (C), and LN229 GBM (E). CD4 MC9999 CAR T cell results are included in Figure S4. The corresponding PD-L1-deficient tumor cell variants were included as negative controls. The release of granzyme B was readily detected when MC9999 CAR T cells were exposed to PD-L1-expressing target cells, including Calu-1 (B), SH-4 (D), and LN229 (F). The data, plotted as mean ± SEM of triplicate sampling, are representative of three independent experiments (n = 3) and analyzed using the multiple t test (∗∗∗p < 0.001).

Figure 4

Patient-derived primary GBM cells were targeted by MC9999 CAR T cells (A) Two patients (QNS120 and QNS712) were diagnosed with glioblastoma (grade 4) as confirmed by MRI. The red arrow highlights the tumor tissues. (B) The resected tumors were obtained from the patients to generate two patient-derived primary GBM cell lines. The PD-L1 expression on QNS120 and QNS712 tumor cells was characterized using immunostaining and analyzed with flow cytometry. (C) MC9999 CAR T cells were functionally activated by QNS120 and QNS712 tumor cells, as indicated by cell surface staining of CD107a in a degranulation assay. CD8 data are shown here with the CD4 data included in Figure S5. MDA-MB-231 PD-L1 OE and MDA-MB-231 PD-L1 KO cells were used as antigen-positive and antigen-negative controls, respectively. (D) The release of granzyme B was significantly elevated when MC9999 CAR T cells targeted QNS120 and QNS712 tumor cells. The data, plotted as mean ± SEM of triplicate sampling, are representative of three independent experiments (n = 3) and analyzed using the multiple t test (∗∗∗p < 0.001).

Figure 5

Treatment with MC9999 CAR T cells eradicated intracranially established LN229 GBM tumors (A) NSG mice were intracranially challenged with luciferase-labeled, PD-L1-OE LN229 GBM tumor cells at a dose of 0.5 × 10⁶ cells/mouse. Seven days after the tumor challenge, the mice were randomized into three groups (n = 5) and then received an IV infusion of one of the following: PBS, non-CAR T cells (5 × 10⁶ total T cells/mouse), or MC9999 CAR T cells (2 × 10⁶ CAR T cells out of 5 × 10⁶ total T cells/mouse) generated from the same donor. A second treatment dose was administrated IV on day 14. The mice were imaged weekly to track tumor progression for 150 days. The representative IVIS images illustrated the changes in tumor burdens over time. (B) A Kaplan-Meier plot was generated to compare the overall survival among the treatment groups. Log rank analysis revealed significant differences between the MC9999 CAR T cell treatment group and both control groups receiving either PBS or non-CAR T cells (∗∗p < 0.01).

Figure 6

MC9999 CAR T cells elicited cytolysis of HMC3 cells modeling tumor-associated microglias (A) Immunostaining of HMC3 microglial cells with an anti-human PD-L1 monoclonal antibody demonstrated cell surface expression of PD-L1. An isotype control antibody served as a negative control. (B) Co-incubating MC9999 CAR T cells with HMC3 cells triggered T cell degranulation, as evidenced by the cell surface detection of CD107a. (CD8 data are shown here with the CD4 data included in Figure S6.) (C) A significant release of granzyme B was detected with ELISA when MC9999 CAR T cells were co-incubated with HMC3 cells. The data, plotted as mean ± SEM of triplicate sampling, are representative of three independent experiments and analyzed using the multiple t test (∗∗∗p < 0.001). (D) The impedance-based killing assay revealed a real-time cytotoxicity exhibited by MC9999 CAR T cells against HMC3 cells. HMC3 cells were cultured for 24 h, followed by the addition of MC9999 CAR T cells, which resulted in a significant decrease in the CI, a measure of cellular impedance, within the cultured HMC3 cells. The data are representative of three independent experiments.

Figure 7

MC9999 CAR T cells target MDM-M2 that model immunosuppressive cells and patient-derived TAMs (A) The CD163⁺CD209⁺ immunophenotype of MDM-M2 was characterized and compared to that of their monocyte precursors (scatterplot). The PD-L1 expression on MDM-M2 was confirmed by immunostaining and analyzed using flow cytometry, with CD14⁺ monocytes serving as a negative control (histogram plots). (B) MC9999 CAR T cells exhibited cytotoxicity against MDM-M2 but not CD14⁺ monocytes, as determined via a CD107a degranulation assay. CD8 data are shown here with the CD4 data included in Figure S7A. (C) The release of granzyme B was significantly increased when MC9999 CAR T cells were cultured with MDM-M2. The data, plotted as mean ± SEM of triplicate sampling, are representative of three independent experiments and analyzed using the multiple t test (∗∗∗p < 0.001; ns, no significance). (D) The impedance-based killing assay demonstrated the direct killing of MDM-M2 by MC9999 CAR T cells, as evidenced by a significant decrease in the CI upon the addition of the CAR T cells to cultured MDM-M2. The data are representative of three independent experiments. (E) TAMs isolated from a surgically resected GBM tumor displayed the expected CD163⁺CD209⁺ immunophenotype upon immunostaining. (F) The CD163⁺CD209⁺ gated TAMs were highly positive for PD-L1 at the cell surface. CD14⁺ monocytes served as the negative control for both immunophenotypic characterization and PD-L1 staining. (G) Evaluation via the CD107a degranulation assay revealed that the MC9999 CAR T cells, derived from healthy donor T cells, elicited cytotoxicity against the TAMs extracted from GBM tumor. CD8 data are shown here with the CD4 data included in Figure S7B.

Figure 8

Validation of cytotoxic functionalities of GBM patient-derived MC9999 CAR T cells (A) Using peripheral blood T cells obtained from GBM patients, three batches of patient-derived MC9999 CAR T cells were generated. The cytotoxicity of these patient-derived CAR T cells was evaluated through a CD107a degranulation assay. Upon incubation with the PD-L1-expressing target cells, including QNS120 and QNS712 GBM patient-derived tumor cells, as well as HMC3 microglia cells, the CAR T cells exhibited degranulation activities, as evidenced by the presence of CD107a at the cell surface. CD8 data are shown here with the CD4 data included in Figure S9. (B) After 72 h of co-culturing patient-derived MC9999 CAR T cells with the target cells, the tissue culture supernatant was collected for quantitative analysis of T cell cytotoxic granules/cytokines, including granzyme A, granzyme B, IFN-γ, and perforin, using a customized U-PLEX Multiplex Assay. The data were plotted as mean ± SEM of triplicate sampling and analyzed using the multiple t test (∗∗p = 0.0011; ∗∗∗p < 0.001).