TCR β chain-directed bispecific antibodies for the treatment of T cell cancers

Abstract

Immunotherapies such as chimeric antigen receptor (CAR) T cells and bispecific antibodies redirect healthy T cells to kill cancer cells expressing the target antigen. The pan-B cell antigen-targeting immunotherapies have been remarkably successful in treating B cell malignancies. Such therapies also result in the near-complete loss of healthy B cells, but this depletion is well tolerated by patients. Although analogous targeting of pan-T cell markers could, in theory, help control T cell cancers, the concomitant healthy T cell depletion would result in severe and unacceptable immunosuppression. Thus, therapies directed against T cell cancers require more selective targeting. Here, we describe an approach to target T cell cancers through T cell receptor (TCR) antigens. Each T cell, normal or malignant, expresses a unique TCR β chain generated from 1 of 30 TCR β chain variable gene families (TRBV1 to TRBV30). We hypothesized that bispecific antibodies targeting a single TRBV family member expressed in malignant T cells could promote killing of these cancer cells, while preserving healthy T cells that express any of the other 29 possible TRBV family members. We addressed this hypothesis by demonstrating that bispecific antibodies targeting TRBV5-5 (α-V5) or TRBV12 (α-V12) specifically lyse relevant malignant T cell lines and patient-derived T cell leukemias in vitro. Treatment with these antibodies also resulted in major tumor regressions in mouse models of human T cell cancers. This approach provides an off-the-shelf, T cell cancer selective targeting approach that preserves enough healthy T cells to maintain cellular immunity.

Fig. 1.. TRBV-specific BsAbs deplete cognate TRBV-expressing T cells while preserving most nontargeted T cells.

(A) Illustration depicting the proposed selective TRBV depletion strategy: Human T cells comprise 30 TRBV families, including TRBV1 (orange)–, TRBV5 (red)–, TRBV12 (cyan)–, TRBV20 (yellow)–, and TRBV30 (purple)–expressing cells. α-V12 binds TRBV12-expressing T cells, leading to selective killing of the TRBV12 population while sparing most of the remaining non-TRBV12 T cells. (B) α-V5, α-V12, and α-C1 BsAbs are composed of α-CD3 scFv (orange) linked with α-TRBV5–5 (red), α-TRBV12 (cyan), and α-TRBC1 (gray) scFvs, respectively. Each scFv is composed of a variable heavy (V_H) and variable light (V_L) chain. (C) Normal human T cells (1 × 10⁶) were incubated with α-C1, α-V5, or α-V12 BsAbs (0.5 ng/ml) for 17 hours, followed by counting the number of surviving T cells and flow cytometric assessment of the TRBC and TRBV distribution in surviving T cells. Data are shown as the mean viable cell count from five different normal individuals. (D and E) Flow cytometry plots showing percentage of surviving T cells from five different normal human T cell donors after α-V5 BsAb or α-V12 BsAb (D) or α-C1 BsAb (E) treatment. In (C), (D), and (E), number of human replicates, n = 5. Number of repeated experiments, N = 2.

Fig. 2.. TRBC1, TRBV5–5, or TRBV12 engagement activates T cells.

(A) Illustration depicting bidirectional T cell killing by α-C1, α-V5, and α-V12 BsAb. The conventional mechanism of action of α-C1, α-V5, and α-V12 involves cross-linking the T cell–activating CD3 molecule (using α-CD3 scFv) on T cell #1 with TRBC1, TRBV5–5, or TRBV12 (using anti-TRBC1, anti-TRBV5–5, or anti-TRBV12 scFvs, collectively shown as “α-TRB”) on T cell #2, causing T cell #1–mediated killing of cell #2 (“i”). When the target cell (cell #2) is also a T cell and can be activated by cross-linking with α-C1, α-V5, or α-V12, it may be able to function as an “effector” T cell and kill T cell #1 (“ii”). (B) Cartoons of α-CD3-CD19, α-C1-CD19, α-V5-CD19, and α-V12-CD19 BsAbs, composed of anti-CD19 scFv (black) linked to anti-CD3 (orange), anti-TRBC1 (gray), anti-TRBV5–5 (red), and anti-TRBV12 (cyan) scFvs. (C and D) Normal human T cells (5 × 10⁴) were incubated with 5 × 10⁴ wild-type (WT) or CD19 knockout (CD19-KO) NALM6 B cells (expressing luciferase) with the indicated BsAbs (0.5 ng/ml) for 17 hours. IFN-γ ELISA was used to assess normal human T cell cytokine release (C), and luminescence was used to assess viable NALM6 B cells (D). (E and F) Target NALM6 B cells (expressing luciferase) (5 × 10⁴) were incubated with 5 × 10⁴ normal human T cells, or TRBV5- and TRBV12-depleted T cells along with indicated BsAbs (0.5 ng/ml) for 17 hours. IFN-γ detection was used to assess normal human T cell cytokine release (E) and luminescence was used to assess viable NALM6 B cells (F). (G and H) Target NALM6 B cells (expressing luciferase) (5 × 10⁴) were incubated with TRBV5 (“TRBV5⁺”)–enriched or TRBV12 (“TRBV12⁺”)–enriched T cells, along with indicated BsAbs (0.5 ng/ml) for 17 hours. IFN-γ detection was used to assess normal human T cell cytokine release (G) and luminescence was used to assess viable NALM6 B cells (H). In (C) to (H), bars represent means ± standard error of mean using three different normal human T cell donors, n = 3. Number of repeated experiments, N = 2. In (C) and (D), ***P < 0.001, ****P < 0.0001, by one-way ANOVA with Dunnett’s multiple comparison test. In (E) and (F), *P ≤ 0.05, **P < 0.01, ***P < 0.001. ns, not significant, by two-tailed paired t test. (G and H) ****P < 0.0001, by one-way ANOVA with Sidak multiple comparison test.

Fig. 3.. TRBV-specific BsAbs induce T cell cytokine responses against cancer cells in vitro.

(A) Normal human T cells (3.5 × 10⁴) were incubated with 3.5 × 10⁴ of the indicated target T cell cancer cell lines in the presence of α-C1, α-V5, or α-V12 (0.5 ng/ml) for 17 hours. T cell cytokine release was then assessed by IFN-γ ELISA. The surface expression of CD3, TRBC1, and TRBV is noted below each cell line. (B) A total of 5 × 10⁴ normal human T cells or TRBV5- or TRBV12-depleted normal T cells were incubated with 5 × 10⁴ Jurkat cells or HPB-ALL cells in the presence of the indicated BsAbs (0.5 ng/ml) for 17 hours. T cell cytokine release was then assessed by IFN-γ ELISA. y, yes; n, no. (C and D) Human T cells (5 × 10⁴) were incubated with 5 × 10⁴ HPB-ALL cells (C) or Jurkat T cells (D) in the presence of the indicated concentrations of α-V5 (C) or α-V12 (D) for 17 hours. T cell cytokine release was then measured with Luminex assay. The half maximal effective concentration (EC₅₀, M) for each analyte is indicated in the corresponding graphs. In (A) to (D), data shown as means ± standard error of mean from three different human T cell donors, n = 3. Number of repeated experiments, N = 2. **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant, by one-way ANOVA with Sidak multiple comparison test.

Fig. 4.. TRBV-specific BsAbs kill T cell cancer cells in vitro.

(A and B) Normal human T cells (5 × 10⁴) were incubated with 5 × 10⁴ Jurkat cells (A) or HPB-ALL cells (B) in the presence of the indicated concentrations of α-V12 (A) and α-V5 (B) for 17 hours. The Jurkat and HPB-ALL cells expressed luciferase. Luminescence was used to assess viable Jurkat and HPB-ALL cells. The EC₅₀ (M) for each BsAb is indicated in the corresponding graphs. (C) Normal human T cells (5 × 10⁴) were incubated with 5 × 10⁴ wild-type (WT) or TCR gene-disrupted (TCR-KO) Jurkat cells in the presence of the indicated BsAbs (0.5 ng/ml) for 17 hours. (D) shows the aggregate data of percentage of tumor cells in each treatment condition using T cells from three different human donors. (E) Normal human T cells (5 × 10⁴) were incubated with 5 × 10⁴ wild-type (WT) or TCR gene-disrupted (TCR-KO) HPB-ALL cells in the presence of the indicated BsAbs (0.5 ng/ml) for 17 hours. All Jurkat and HPB-ALL cells expressed GFP. Flow cytometry was then used to assess CD3 and GFP expression. (F) shows the aggregate data of percentage of tumor cells in each treatment condition using T cells from three different human donors. In (C) and (E), the numbers beside density plots indicate the percentage of surviving cells. In (A), (B), (D), and (F), data represent means ± standard error of mean using three different normal human T cell donors, n = 3. Number of repeated experiments, N = 2. ****P < 0.0001. ns, not significant, by ANOVA with Sidak multiple comparison test.

Fig. 5.. TRBV-specific BsAb kills patient-derived T-ALL cells in vitro.

(A) Flow cytometric analysis of two T-ALL patient samples with circulating lymphoblasts expressing TRBV12. Numbers adjacent to the plots indicate the percentage of CD3⁺ cells that express TRBV12. (B) Normal human T cells (5 × 10⁴) were cocultured with 5 × 10⁴ patient-derived T-ALL target cells (from patient 1 and patient 2) in the presence of the indicated BsAbs (0.5 ng/ml) for 17 hours. T cell cytokine release was assessed by measurement of IFN-γ in the supernatant. (C) Normal human T cells (5 × 10⁴) were cocultured with 5 × 10⁴ patient-derived T-ALL target cells (from patient 1) in the presence of the indicated BsAbs (0.5 ng/ml) for 17 hours. T cell activation and exhaustion markers were assessed by flow cytometry. In (B) and (C), bars represent means ± standard error of mean from three technical replicates, n = 3. (D) Flow cytometry histogram of HLA-A3–stained normal human T cells and patient-derived T-ALL malignant cells. (E and F) Normal human T cells (5 × 10⁴) were cocultured with 5 × 10⁴ patient-derived T-ALL target cells in the presence of α-CD19 or α-V12 BsAbs (0.5 ng/ml) for 17 hours. Flow cytometric analysis of HLA-A3 and CD3 was then performed. Numbers adjacent to the plots indicate the numbers of cells counted by flow cytometry in a representative experiment (E), with data from three technical replicates, n = 3, shown in (F). ***P < 0.001, ****P < 0.0001 by one-way ANOVA with Sidak’s multiple comparison test.

Fig. 6.. TRBV-specific BsAbs specifically kill cancer cells in vivo.

(A) Timeline of in vivo tumor experiments. (B and C) NSG mice were intravenously injected with 5 × 10⁶ normal human T cells and 5 × 10⁶ WT or TCR-KO Jurkat cells (B), or WT or TCR-KO HPB-ALL cells (C). All Jurkat and HPB-ALL cells expressed luciferase and GFP. Intraperitoneal pumps containing 100 μg of α-CD19, α-V12, or α-V5 BsAb were placed in the animals 4 days after cell injection, and bioluminescence imaging (BLI) was performed on the indicated days. BLI data representative of one of two independent experiments with six NSG mice in each group are shown in (B) and (C). Number of animals in two experiments, n = 5 and n = 6. Number of independent experiments, N = 2. (D) Combined radiance value from two independent experiments with a total of 11 NSG mice in each group was measured on the indicated days. (E and F) Circulating cancer cell and T cell counts were assessed from six different NSG mice for each cancer cell type. Flow cytometry on mouse blood collected on day 19 was used to detect circulating WT Jurkat or HPB-ALL cells (CD3⁺, GFP⁺, top right quadrant), circulating TCR-KO Jurkat or HPB-ALL cells (CD3⁻, GFP⁺, bottom right quadrant) or circulating normal human T cells (CD3⁺, GFP⁻, top left quadrant) after the indicated treatments. In (E), flow cytometry data are representative of one of six NSG mice in each group. In (F), data combined from six NSG mice in each group are shown as means ± standard error of mean. (G and H) Kaplan-Meier survival curves of WT or TCR-KO Jurkat (G) or HPB-ALL (H)–bearing NSG mice after various treatments. Survival data were aggregated from two independent experiments with a total of 11 NSG mice in each group. Median overall survival is reported beside the survival curves. Jurkat WT/α-CD19 versus Jurkat WT/α-V12 hazard ratio (HR) = 0.18, ****P < 0.0001, log-rank (Mantel-Cox) test. HPB-ALL WT/α-CD19 versus HPB-ALL WT/α-V5, HR = 0.19, ****P = 0.0001, log-rank (Mantel-Cox) test. In (D) and (F), *P ≤ 0.05, ****P < 0.0001 by one-way ANOVA with Sidak’s multiple comparison test.