BCMA- and CST6-specific CAR T cells lyse multiple myeloma cells and suppress murine osteolytic lesions

Abstract

We have previously demonstrated that cystatin E/M (CST6), which is elevated in a subset of patients with multiple myeloma (MM) lacking osteolytic lesions (OLs), suppresses MM bone disease by blocking osteoclast differentiation and function. CST6 is a secreted type 2 cystatin, a cysteine protease inhibitor that regulates lysosomal cysteine proteases and the asparaginyl endopeptidase legumain. Here, we developed B cell maturation antigen (BCMA) CST6 chimeric antigen receptor T cells (CAR-T cells), which lysed MM cells and released CST6 proteins. Our in vitro studies show that these CAR-T cells suppressed the differentiation and formation of tartrate-resistant acid phosphatase-positive (TRAP+) osteoclasts. Using xenografted MM mice, bioluminescence images showed that both BCMA-CAR-T and BCMA-CST6-CAR-T cells inhibited MM growth to a similar extent. Reconstructed micro-computed tomography images revealed that BCMA-CST6-CAR-T cells, but not BCMA-CAR-T cells, prevented MM-induced bone damage and decreased osteoclast numbers. Our results provide a CAR-T strategy that targets tumor cells directly and delivers an inhibitor of bone resorption.

Keywords: Bone disease; Cancer immunotherapy; Hematology; Oncology.

Figure 1. Construction and characteristics of BCMA–CST6–CAR-T cells.

(A) Generation of BCMA-CST6-CAR constructs. The BCMA-CST6-CAR vector was constructed by BCMA-specific scFvs, a safety switch (RQR8) in the hinge region, and a 4-1BB coactivation domain with CD3ζ. A P2A self-cleaving peptide was inserted between the CAR vectors and CST6. (B) The BCMA-CAR vector contained BCMA-scFv, a safety switch, and a 4-1BB coactivation domain with CD3ζ. (C) The MOCK-CAR vector only contained a safety switch and a 4-1BB coactivation domain with CD3ζ. All vectors contained SFFV promoters to drive expression of targeted gene(s) in CAR T cells. (D) Schematic representation of the BCMA–CST6–CAR-T structure. (E) Lentiviruses combined with the RetroNectin transfection technology were engineered with CD3⁺ T cells derived from healthy donor PBMCs. CAR-T cells were detected with RQR8-specific anti-CD34 antibodies by flow cytometry. CD34⁺ rates were 29.6% for BCMA–CST6–CAR-T cells. CD34⁺ cells were enriched to 75.6% for BCMA–CST6–CAR-T cells sorted with anti-CD34 microbeads. CD34⁺ cells were enriched to 76.9% for sorted BCMA–CAR-T cells. (F) Flow cytometric analyses revealed CD4⁺/CD8⁺ ratios of MOCK–CAR-T cells, BCMA–CAR-T cells, and BCMA–CST6–CAR-T cells (n = 3, representative result from 3 independent experiments). (G) Bar plots represent the frequency of CAR-T cells gated on CD4⁺ and CD8⁺ T cells (n = 3). Data represent the mean ± SD. One-way ANOVA was used for the statistical analysis; NS = P > 0.05. LTR, long terminal repeat.

Figure 2. BCMA–CST6–CAR-T cells efficiently kill MM cell lines and release cytokines in vitro.

(A) Flow cytometric analysis of BCMA expression in MM1.S, OPM2, and H929 cell lines. BCMA was highly expressed in MM1.S, OPM2, and H929 cell lines. (B–D) In vitro cytolytic activity of BCMA–CST6–CAR-T cells. CAR-T cells were added to MM1.S (B), OMP2 (C), and H929 (D) cell lines at an E/T ratio of 1:5 to 5:1. After 24 hours of coculturing, cytolytic activity was measured. Percentage of  lysis = (experimental lysis − spontaneous lysis)/(maximal lysis − spontaneous lysis) × 100%. n = 5. (E) Expression of the T cell activation marker CD69 was detected at an E/T ratio of 5:1 after 24 hours of coculturing (n = 5). (F) IL-2 concentrations in supernatants were detected at an E/T ratio of 5:1 after 24 hours of coculturing (n = 5). (G) IFN-γ concentrations in supernatants were detected at an E/T ratio of 5:1 after 24 hours of coculturing (n = 5). (H) CST6 concentrations in supernatants were detected at an E/T ratio of 5:1 after 24 hours of coculturing (n = 5). Data represent the mean ± SD. ***P < 0.001, by 1-way ANOVA; NS = P > 0.05.

Figure 3. BCMA–CST6–CAR-T cells selectively kill primary human MM cells in vitro.

(A) Experimental workflow for the process of CAR-T cell–mediated cytotoxicity against primary human MM cells. Samples from 3 patients were collected. CAR-T cells were generated using the donors’ peripheral blood T cells and directed against their own BMMCs. (B) Flow cytometric analysis was used to identify primary human CD138⁺ MM cells from 3 patient samples with the following different treatments for 24 hours: PBS, MOCK–CAR-T cells, BCMA–CAR-T cells, and BCMA–CST6–CAR-T cells. (C) Percentage of the subpopulation of human CD138⁺ MM cells decreased in 3 of 3 primary MM samples after BCMA–CST6–CAR-T treatment. *P < 0.05, by 1-way ANOVA; NS = P > 0.05.

Figure 4. BCMA–CST6–CAR-T cells suppress osteoclast differentiation.

(A) Experimental workflow for the detection of TRAP⁺ osteoclasts. (B) CST6 concentrations in supernatants were detected at E/T ratios of 1:5 to 5:1 after 24 hours of coculturing (n = 5). (C) CAR-T cells were incubated with MM1.S cells at ratios of 1:5 to 5:1 for 24 hours, and CM were collected and added into RAW 264.7 cells with RANKL. On day 7, osteoclasts were stained with TRAP solution (n = 5, representative result from 5 independent experiments). Scale bars: 200 μm. (D) Bar plots present quantifications of the TRAP⁺ area (n = 5). (E) Human CD14⁺ monocytes sorted from bone marrow of patients with MM were differentiated into osteoclasts with M-CSF and RANKL for 7 days. CM were added to human CD14⁺ monocyte culture media (50%) for another 7 days. On day 14, osteoclasts were stained with TRAP solution (n = 5, representative result from 5 independent experiments). Scale bars: 200 μm. (F) Bar plots represent the number of TRAP⁺ osteoclasts derived from human CD14⁺ monocytes per view (n = 5). Data represent the mean ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA; NS = P > 0.05.

Figure 5. BCMA–CST6–CAR-T cells suppress MM1.S cell growth in vivo.

(A) Schematic of the experimental plan. NSG mice were administered 1.5 × 10⁶ MM cells via i.v. injection. On day seven, 1.5 × 10⁶ CAR-T cells were administered after injection of MM cells. Myeloma progression was monitored until the mice developed hind limb paralysis. (B) Tumor burden was evaluated by bioluminescence imaging of MM1.S cell–bearing mice treated with MOCK–CAR-T, BCMA–CAR-T, or BCMA–CST6–CAR-T cells. (C) Quantitative analysis of bioluminescence intensity (n = 5). (D) Mouse serum levels of CST6 protein detected by ELISA (n = 5). (E) Mouse serum levels of calcium detected by ELISA (n = 5). (F) Mouse serum levels of PTHrP detected by ELISA (n = 5). (G) Kaplan-Meier disease progression analysis of CAR-T treatment in NSG models (n = 5). Data represent the mean ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA (C–F) and log-rank test (G); NS = P > 0.05.

Figure 6. Validation of BCMA–CST6–CAR-T cell suppression of MM growth in OPM2 and H929 MM cell lines in vivo.

(A) Tumor burden was evaluated by bioluminescence imaging of OPM2 cell–bearing mice treated with MOCK–CAR-T cells, BCMA–CAR-T cells, or BCMA–CST6–CAR-T cells. (B) Quantitative analysis of bioluminescence intensity in the OPM2 xenograft model (n = 5). (C–E) Mouse serum levels of CST6 (C), calcium (D), and PTHrP (E) detected by ELISA in the OPM2 xenograft model (n = 5). (F) Tumor burden was evaluated by bioluminescence imaging of H929 cell–bearing mice treated with MOCK–CAR-T cells, BCMA–CAR-T cells, or BCMA–CST6–CAR-T cells (n = 5). (G) Quantitative analysis of bioluminescence intensity in the H929 xenograft model (n = 5). (H–J) Mouse serum levels of CST6 (H), calcium (I), and PTHrP (J) detected by ELISA in the H929 xenograft model (n = 5). Data represent the mean ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA; NS = P > 0.05.

Figure 7. BCMA–CST6–CAR-T cells inhibit MM cell–induced bone resorption in vivo.

(A) Reconstructed μCT images of tibia sagittal sections show bone lytic lesions (indicated with arrows) and trabecular architecture in the MM1.S xenograft model (n = 5, representative result from 5 mice). (B–E) Number of bone lytic lesions on the right medial tibia surface according to the trabecular bone parameters BV/TV (B); Tb.Th (C); BMD (D); and Tb.Sp (E) in the MM1.S xenograft model (n = 5). (F) TRAP staining showed osteoclasts (indicated with arrows) in tibias derived from the MM1.S xenograft model (n = 5, representative result from 5 mice). Scale bar: 50 μm. (G) Histomorphometric analyses of the number of TRAP-stained osteoclasts per view (n = 5). Data represent the mean ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA; NS = P > 0.05.

Figure 8. Schematic representation of the role of BCMA–CST6–CAR-T cells in MM growth and focal lesions.

BCMA–CST6–CAR-T cells effectively target MM tumor cells in vitro and in vivo and secrete CST6 protein, thereby inhibiting osteoclastogenesis and bone resorption.