Pan-carcinoma sialyl-Tn-targeting expands CAR therapy to solid tumors

Abstract

Accurate identification of tumor-specific markers is vital for developing chimeric antigen receptor (CAR)-based therapies. While cell surface antigens are seldom cancer-restricted, their post-translational modifications (PTMs), particularly aberrant carbohydrate structures, offer attractive alternatives. Among these, the sialyl-Tn (STn) antigen stands out for its prevalent presence in various epithelial tumors. Although monoclonal antibodies (mAbs) against STn have been developed, their clinical application has been hindered by concerns regarding specificity. Herein, we describe AM52.1, a mAb with unprecedented specificity for STn and lack of reactivity with healthy tissues. The single-chain variable fragment (scFv) of AM52.1 was assembled into a second-generation CAR scaffold. AM52.1CAR T cells efficiently targeted STn-expressing cancer cell lines and patient-derived organoids (PDOs), while sparing STn-negative cells. In further preclinical models, AM52.1CAR T cells robustly controlled gastric and tubo-ovarian tumors, as well as colorectal cancer mucinous peritoneal metastases, highlighting their strong therapeutic potential for targeting and managing complex solid tumors.

Keywords: CAR T cells; antibodies; cancer; glycobiology; glycosylation; immunotherapy; sialyl-Tn; solid tumors.

Graphical abstract

Figure 1

AM52.1 antibody shows specific binding to sialyl-Tn (A) Schematic illustration of the predominant glycan structures in ovine submaxillary mucin (OSM) and its desialylated form, Asialo-OSM (A-OSM). (B) Binding of AM mAbs to varying concentrations of OSM and A-OSM, measured by indirect enzyme-linked immunosorbent assay (ELISA). B72.3, CC49, and 1E3 antibodies were included as controls. (C) Binding specificity of AM mAbs to various sialyl-Tn (STn)-related O-glycan structures, as measured by glycan microarray analysis; 10 μg/mL of each antibody were used. B72.3 mAb served as a control for the assay. Refer to Figure S1A for a comprehensive list of the analyzed glycan structures. (D) Gastrointestinal cancer wild-type (WT) and core 1 β1,3-galactosyltransferase (C1GalT1) knockout (KO) cells were stained for flow cytometry using the AM mAbs, and well-established anti-STn and anti-Tn antibodies as controls. Histograms represent the cell populations stained with the annotated antibodies. (E) Cell lines of different origins (lymphoma, osteosarcoma, melanoma, glioblastoma, prostate, colorectal, gastric, ovarian, breast, and oral cancer) were stained with AM52.1. Representative flow cytometry plots of isotype control and AM52.1 staining are shown. (F) AM51.1 and AM52.1 antibodies, alongside with the positive control B72.3, were used to stain tissue from gastric, colon, pancreatic, prostate, and ovarian carcinomas. Sections show representative staining of tumor tissue with the annotated antibodies. Scale bar corresponds to 200 μm. (G) Graphs depicting the percentage of carcinoma sections exhibiting positive staining for each antibody tested and the corresponding H-score for each tissue type. (H) Healthy tissues (heart, kidney, liver, spleen, lung, thyroid, lip, vocal cords, cervix, and endometrium) were stained with the AM antibodies and the B72.3 positive control. Scale bar corresponds to 100 μm. (I) Pie chart representing cell population distribution in human peripheral blood mononuclear cells (PBMCs) from healthy donors (n = 5) stained with the AM mAbs. (J) Histograms depicting the percentage distribution of various immune cell populations stained with AM51.1 and AM52.1. Data are represented as mean ± standard deviation (SD).

Figure 2

Sequence and structural analysis indicating a sialyl-Tn-restricted binding of the AM52.1-based single-chain variable fragment (A) Multiple sequence alignment (MSA) for the single-chain variable fragment (scFv) region derived from various anti-sialyl-Tn (STn) monoclonal antibodies (mAbs). The aligned light and heavy chain sequences were visualized using the Clustal Color Scheme in Jalview, grouping residues based on their similarity. The complementarity-determining regions (CDR) regions were mapped on the alignment and annotated using the ImMunoGeneTics (IMGT) database numbering scheme. (B) Percentage of sequence identity calculated for the light and heavy chains of the scFv derived from the investigated antibodies, using MSA. (C) Front and top view of the AM52.1-based scFv structure with the CDR regions mapped (CDR1: violet; CDR2: yellow; CDR3: red). The antibody surface is illustrated using a white-colored volume mesh. (D) The AM52.1-based scFv (cyan color) was structurally aligned with the scFv derived from the other anti-STn mAbs (gray color). Root-mean-square deviation (RMSD) values are given below the structure for each pair.

Figure 3

AM52.1-derived CAR T cells reveal specific cytotoxicity against sialyl-Tn-expressing cancer cell lines (A) Illustrative schematic of CAR construct design. (B and C) (B) Jurkat and (C) primary T cells were stably transduced with CAR constructs, and CAR expression was assessed through quantification of murine fragment antigen-binding (mFab) region and CD34 expression. n = 2–3 for the Jurkat and n = 5–7 for the healthy donors. (D) Degranulation activity of CD4⁺ and CD8⁺ CAR T cells co-cultured with sialyl-Tn (STn)-negative and STn-positive target cells at an effector:target (E:T) ratio of 1:2 for 24 h. n = 3 healthy donors. (E) CAR T cell-specific cytotoxicity assessed by a bioluminescence-based killing assay. CAR T cells were co-cultured with cell lines expressing various STn levels at the indicated E:T ratios. Data correspond to the 12-h time point. n = 3, n = 5, and n = 2 healthy donors for Caov-3, SNU-16, and OVCAR-3 and OV-90, respectively. (F) Cytokine secretion following 24-h co-culture of CAR T cells with STn-expressing and non-expressing cells. The E:T ratio used was of 1:1. n = 3 healthy donors. Data are represented as mean ± standard deviation (SD). Statistical significance was determined using paired t test for (A) and (B), and two-way ANOVA for (E). ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Figure 4

AM52.1CAR T cells are effective and specific in a preclinical model of patient-derived organoids (A) sialyl-Tn (STn) expression evaluation using CC49 and AM52.1 mAbs in five patient-derived organoids (PDOs). Scale bars correspond to 100 and 50 μm, as specified. (B) Schematic outlining the experimental design followed to establish the co-culture of the PDOs with the immune cells. (C) Viability of a STn-negative (PDO2) and a STn-positive (PDO5) PDOs upon co-culture with specified CAR T cells at different effector:target (E:T) ratios. Statistical comparison reflects differences at the same E:T ratio between the different organoids. n = 1 healthy donor. (D) Immunofluorescent analysis of cleaved caspase-3 (CC3) expression in the selected PDOs after co-culture with indicated CAR T cells at an E:T ratio of 10:1. Scale bars correspond to 50 and 10 μm, as specified. (E) Viability of PDOs derived from normal gastric mucosa (PDO1) and gastric tumor tissue (PDO2-5) with different levels of STn expression following co-culture with specified CAR T cells. The data correspond to the E:T ratio of 10:1. n = 3 healthy donors. Data are represented as mean ± standard deviation (SD). Statistical significance was determined using paired t test. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, and ∗∗∗∗p ≤ 0.0001.

Figure 5

AM52.1CAR T cells efficiently control tumor growth in xenograft mouse models (A, E, and I) (A) OV-90, (E) OVCAR-3, and (I) SNU-16. Experimental scheme of the xenograft mouse models established. (B, F, and J) (B) OV-90, (F) OVCAR-3, and (J) SNU-16. Representative images of mice and bioluminescent signals at the indicated time points throughout the experiment. (C, G, and K) (C) OV-90, (G) OVCAR-3, and (K) SNU-16. Quantitative plots showing the bioluminescent signal (p/s) for each mouse over the course of the experiment. (D, H, and L) (D) OV-90, (H) OVCAR-3, and (L) SNU-16. Kaplan-Meier survival curves. Statistical analysis of survival curves was performed using the log rank (Mantel-Cox) test. ∗p ≤ 0.05 and ∗∗p ≤ 0.01.

Figure 6

AM52.1CAR T cells control complex peritoneal metastases from colorectal cancer patient-derived xenograft model (A) Experimental scheme of the patient-derived xenograft (PDX) model established. (B) Images of two representative anesthetized mice 13 days post-treatment, either untreated or treated with AM52.1CAR T cells. (C) Kaplan-Meier survival curves for the PDX model. (D) Magnetic resonance imaging (MRI) images of three representative mice at week following CAR T cell injection. (E) Quantification of mucinous tissue in untreated mice (n = 4). (F) MRI images of three representative mice at week 2 post-injection of Mock T cells. (G) Quantification of mucinous tissue in Mock T cell-treated mice (n = 6). (H) MRI images of three representative mice at week 2 post-injection of AM52.1CAR T cell-treated mice. (I) Quantification of mucinous tissue in AM52.1CAR T cell-treated mice (n = 6). Statistical analysis of survival curves was performed using the log rank (Mantel-Cox) test. ∗p ≤ 0.05 and ∗∗p ≤ 0.01.