A single-chain antibody generation system yielding CAR-T cells with superior antitumor function

Abstract

Cancer immunotherapy using T cells redirected with chimeric antigen receptor (CAR) has shown a lot of promise. We have established a single-chain antibody (scFv) generation system in which scFv library-expressing CAR-T cells can be screened appropriately based on their antitumor functions. A variable region library containing the variable and J regions of the human immunoglobulin light or heavy chain was fused with the variable region of a heavy or light chain encoded by an existing tumor-specific antibody to generate a new scFv library. Then, scFv library-expressing CAR-T cells were generated and stimulated with target cells to concentrate the antigen-specific population. Using this system, target-specific recognition of CAR-T cells appeared to be finely tuned by selecting a new variable region. Importantly, we have demonstrated that the newly optimized scFv-expressing CAR-T cells had better proliferation capacity and durable phenotypes, enabling superior reactivity against advanced tumors in vivo in comparison with the original CAR-T cells. Therefore, the optimization of an scFv is needed to maximize the in vivo antitumor functions of CAR-T cells. This system may allow us to adjust an immunological synapse formed by an scFv expressed by CAR-T cells and a target antigen, representing an ideal form of CAR-T-cell immunotherapy.

Fig. 1. Identification of antitumor scFvs optimal for CAR-T cells using a T-cell-based scFv generation system.

Each of the steps and times required for preparation of the CAR library, generation of CAR-library T cells, antigen stimulation of CAR-library T cells, and isolation of new scFvs are summarized. a Variable region libraries derived from the human immunoglobulin light chain (hL, hK) or heavy chain (hH) were prepared. These libraries were fused with the variable region of an existing antitumor antibody (VH, e.g. 3M4E5-H, 35-G01-H; VL, e.g. 3M4E5-L, 35-G01-L, indicated in blue and red, respectively) to generate antitumor scFv libraries. Then, a CAR library encoding the scFv library followed by the transmembrane/intracellular domains of a CAR (CD28ζ) construct was prepared. b A CAR library was retrovirally transduced into human peripheral blood T cells to generate antitumor CAR-library T cells. ScFv-optimized (dark red), scFv-suboptimal (light red), and non-reactive (gray) CAR-T cells can be included among CAR-library T cells. c Antitumor CAR-library T cells were stimulated with tumor cells expressing a target antigen, such as A2/NY-ESO-1₁₅₇ and CD19, to concentrate the scFvs for delivery of activation signals to peripheral blood T cells via a CAR construct. d After isolation of antigen-specific CAR-T cells by flow cytometry using A2/NY-ESO-1₁₅₇ tetramer and soluble CD19 dimer, cDNA was synthesized and the primary structures of new scFvs that would be optimal for CAR-T cells were determined.

Fig. 2. Target reactivity of CAR-T cells expressing new scFvs identified from A2/NY-ESO-1₁₅₇ CAR-library T cells.

a A2/NY-ESO-1₁₅₇-specific scFv libraries possessing the variable and J regions of each human kappa chain (hK), lambda chain (hL) or heavy chain (hH) library derived from four different donors along with the A2/NY-ESO-1₁₅₇-specific 3M4E5-H or 3M4E5-L, indicated as hK/3M4E5-H, hL/3M4E5-H, and hH/3M4E5-L, were individually generated. A second-generation CAR library encoding each A2/NY-ESO-1₁₅₇-specific scFv library together with CD28ζ was transduced into peripheral blood T cells. A2/NY-ESO-1₁₅₇ CAR-library T cells similarly stimulated three times were stained with 20 μg/mL A2/NY-ESO-1₁₅₇ or A2/HIV Gag₇₇ tetramer. Representative stainings of A2/NY-ESO-1₁₅₇ CAR-library CD8⁺ T cells (hL/3M4E5-H and hH/3M4E5-L) derived from donors 1 and 2, and CD4⁺ T cells (hL/3M4E5-H) derived from donors 3 and 4 are shown. Each percentage of A2/NY-ESO-1₁₅₇ tetramer-positive cells among CAR-library T cells is indicated. b A2/NY-ESO-1₁₅₇ tetramer positivity in CAR-library CD8⁺ T cells and CD4⁺ T cells is summarized. Each dot represents each library source derived from the different donors. The original 3M4E5 CAR-T cells and control T cells were prepared as a positive and a negative control, respectively. Two-tailed Mann–Whitney test was performed to compare two different groups. *p < 0.05. c Control T cells, or hL/3M4E5-H CAR-library CD8⁺ T cells (donor 1) were incubated with T2 cells pulsed with 10 μg/mL NY-ESO-1₁₅₇ peptide or HIV Gag₇₇ peptide. Multiple cytokine production by these CAR-T cells was measured by intracellular cytokine staining via flow cytometry. TNFα⁺IFNγ⁺IL2⁺ cells in ΔNGFR⁺CD8⁺ T cells are shown. The experiments were performed in triplicate, and error bars show the SD. Welch’s t test (two-sided) was performed for comparison. *p < 0.05. d, e Newly identified A2/NY-ESO-1₁₅₇ CARs were individually reconstituted in Jurkat 76 cells. Jurkat 76/CAR transfectants expressing each different VL (3M4E5-L, L1, L73, L88, L66, or L80) or VH (H73 or H1) chain paired with 3M4E5-H or 3M4E5-L were stained with 5 μg/mL A2/NY-ESO-1₁₅₇ or A2/HIV Gag₇₇ tetramer. Staining of control Jurkat 76 cells is also shown as a negative control (d). Jurkat 76/CAR transfectants were cocultured with T2 cells loaded with 10 μg/mL NY-ESO-1₁₅₇ peptide or HIV Gag₇₇ peptide. CD69 upregulation of Jurkat 76/CAR transfectants was measured by flow cytometry. The experiments were performed in triplicate, and error bars depict the SD (e). f Jurkat 76/CAR transfectants were stained with graded concentrations of A2/NY-ESO-1₁₅₇ tetramer (top). They were also incubated with T2 cells pulsed with graded concentrations of A2/NY-ESO-1₁₅₇ peptide (bottom). Percentage maximal staining of each indicated transfectant was calculated relative to the staining pattern of each transfectant with 10 μg/mL A2/NY-ESO-1₁₅₇ tetramer as 100% (top). Percentage maximal reactivity of each transfectant was measured relative to the CD69 upregulation of each transfectant for 10 μg/mL NY-ESO-1₁₅₇ peptide loaded onto T2 cells as 100% (bottom). g Structural avidity, shown as EC₅₀ values in μg/mL, was calculated as the concentration of A2/NY-ESO-1₁₅₇ tetramer required to achieve 50% of the maximal staining. Functional avidity, expressed as EC₅₀ values in μg/mL, was estimated as the concentration of NY-ESO-1₁₅₇ peptide required to obtain 50% of the maximal reactivity. Each dot represents each Jurkat 76/CAR transfectant. The nonparametric Spearman correlation coefficient was calculated. r and p values are shown.

Fig. 3. Sufficient antitumor reactivity and minimized cross-reactivity of fine-tuned scFv-expressing A2/NY-ESO-1₁₅₇ CAR-T cells.

a Clone L1, or the original 3M4E5-L second generation (CD28ζ) CAR was transduced into peripheral blood T cells. Control T cells and these transfectants were stained with 20 μg/mL A2/NY-ESO-1₁₅₇ tetramer or A2/HIV Gag₇₇ tetramer. Representative dot plots of both CD8⁺ T cells and CD4⁺ T cells are shown. b, c CAR-T cells generated as above were incubated with the indicated peptide-pulsed T2 cells (b), K562/A2 cells, or K562/A2/NY-ESO-1 cells (c), and their cytokine production was measured by intracellular cytokine assays. The experiments were performed in triplicate, and error bars depict the SD. d L1 CAR or original 3M4E5-L CAR CD8⁺ T cells and CD4⁺ T cells were incubated with T2 cells pulsed with 9 different alanine-substituted peptides at 10 μg/mL. IL2 production (%) was measured by intracellular cytokine assays. Each response to alanine-substituted peptides was compared with the response to the original NY-ESO-1₁₅₇ peptide. The experiments were conducted in triplicate, and error bars depict the SD. Welch’s t test (two-sided) was performed for comparison. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Note that control T cells were generated by transduction of the ΔNGFR gene into human peripheral blood T cells. Gene-modified T cells with similar transduction efficiency were prepared, and then ΔNGFR-positive cells were gated and analyzed (a–d).

Fig. 4. Target reactivity of CAR-T cells expressing new scFvs isolated from CD19 CAR-library T cells.

a CD69 upregulation of Jurkat 76/35-G01-LH CAR cells and control Jurkat 76 cells in response to K562, K562/CD19, and Raji cells was measured by flow cytometry. The experiments were performed in triplicate, and error bars show the SD. b A construct of the soluble CD19 (sCD19) dimer is shown (top). Jurkat 76/CAR transfectants and control Jurkat 76 cells were stained with 40 μg/mL sCD19 dimer or PE-anti-his mAb alone (bottom). c CD19-specific scFv libraries possessing each human kappa chain (hK), lambda chain (hL) library derived from two different donors along with 35-G01-H, indicated as hL/35-G01-H and hK/35-G01-H, were individually generated. A second-generation CAR library encoding each CD19-specific scFv library together with CD28ζ was transduced into peripheral blood T cells. CD19 CAR-library T cells similarly stimulated three times were stained with 40 μg/mL sCD19 dimer or PE-anti-his mAb alone. Representative stainings of CD19 CAR-library CD8⁺ T cells (hL/35-G01-H and hK/35-G01-H) derived from donors 1 and 2, and CD4⁺ T cells (hL/35-G01-H) derived from the same donors are depicted. Each percentage of sCD19 dimer-positive cells among CAR-library T cells is shown. d sCD19 dimer positivity in CAR-library (hK/35-G01-H, hL/35-G01-H, and hH/35-G01-L) CD8⁺ T cells and CD4⁺ T cells is summarized. Each dot represents each library source derived from the different donors. The original 35-G01 CAR-T cells and control T cells were prepared as a positive and a negative control, respectively. e Newly isolated CD19 CARs were individually reconstituted in Jurkat 76 cells. Jurkat 76/CAR transfectants expressing each different VL (L16, L7, L17, L4, K4, K5, K9, and K6) chain paired with 35-G01-H were stained with 40 μg/mL sCD19 dimer or PE-anti-his mAb alone. L and K indicate lambda and kappa chain, respectively. f Jurkat 76/CAR transfectants were incubated with the indicated target cells, and their CD69 upregulation was measured by flow cytometry. The experiments were performed in triplicate, and error bars show the SD. g Jurkat 76/CAR transfectants were stained with graded concentrations of sCD19 dimer. Percentage maximal staining was calculated relative to the staining pattern of each transfectant with 40 μg/mL sCD19 dimer as 100%.

Fig. 5. Antitumor reactivity, proliferation capacity, and durable phenotypes of scFv-optimized CD19 CAR-T cells.

a, b Clone L17, or the original 35-G01-L CD19-specific second generation (CD28ζ) CAR was transduced into peripheral blood T cells, which were then stained with 40 μg/mL sCD19 dimer or PE-anti-his mAb alone, and representative dot plots are shown. Each percentage of sCD19 dimer-positive cells in ΔNGFR⁺ cells is displayed (a). L17, and 35-G01-L CD19 CAR-T cells were incubated with the indicated target cells, and their cytokine production was measured by intracellular cytokine assays. The experiments were performed in triplicate, and error bars show the SD (b). Control T cells were prepared by transduction of the ΔNGFR gene into human peripheral blood T cells. Gene-modified T cells with similar transduction efficiency were obtained, and then ΔNGFR-positive cells were gated and analyzed. c The cytotoxicity of established and isolated CD19 CAR-T cells against target cells was examined by ⁵¹Cr-release assays. The experiments were performed in triplicate, and error bars depict the SD. d, e L17, and 35-G01-L CAR-T cells established from four different donors were isolated, labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE), and stimulated with irradiated Raji cells at an E/T ratio of 1:1. After 3 days of incubation, the mean fluorescence intensity (MFI) of CFSE-labeled CAR CD8⁺ T cells and CD4⁺ T cells was examined. The MFI values for each of the donor-derived CFSE-labeled CAR-T cells in the presence of K562 cells instead of Raji cells were utilized as a control. Histograms for CFSE-CAR CD8⁺ T cells and CD4⁺ T cells (donor 2) stimulated with Raji cells (filled lines), or K562 cells (dotted lines), are shown. Red represents the histograms of L17 CAR-T cells, and blue those of 35-G01-L CAR-T cells (left). The percentage of proliferation was calculated as: (control MFI−experimental MFI)/(control MFI) × 100 (%), and is summarized (right) (d). CD45RA⁺CD62L⁺CCR7⁺, CD45RA^-CD62L⁺CCR7⁺, and CD45RA⁺CD62L^-PD1⁺ populations among individual CAR CD8⁺ T cells and CD4⁺ T cells before and after stimulation with Raji cells are also summarized (e). Each dot depicts the percentage of the indicated population among CAR-T cells generated from each donor. Differences between L17 and 35-G01-L CAR-T cells were statistically assessed by paired t test (two-sided). *p < 0.05; **p < 0.01; ***p < 0.001; n.s., not significant.

Fig. 6. Superior in vivo antitumor functions mediated by scFv-optimized CD19 CAR-T cells.

a Five hundred thousand Raji cells were intravenously injected into irradiated NOD/Shi-scid IL2rgamma(null) mice. After engraftment of the Raji cells, 2.0 × 10⁶ L17, 35-G01-L CAR-T cells, or control T cells generated from the same donor were intravenously injected on Day 7. Overall survival of the mice (n = 5 mice per group) is depicted in the form of Kaplan–Meier curves, and survival was compared among the groups. Log-rank (Mantel–Cox) test was performed, and p values are also shown. b One million Raji cells were subcutaneously inoculated into irradiated NOD/Shi-scid IL2rgamma(null) mice, then 2.0 × 10⁶ L17 or 35-G01-L CAR-T cells, which can express SLR, were intravenously injected on Day 7. Photon images of mice obtained after CAR-T-cell injection are shown (n = 4 mice per group). The small white rectangles are magnified and shown in the upper left corner for each mouse. c Tumor sizes (mm³) (left) and photon counts normalized by each photon count obtained from 1.0 × 10⁵ L17 or 35-G01-L CAR-T cells (right) are shown. The arrow inside the left-hand graph indicates the time at which CAR-T cells were injected. Error bars show the SD. Two-way ANOVA with Sidak’s correction was conducted for multiple comparisons. *p < 0.05; **p < 0.01; n.s., not significant. d Mice shown in (b) were sacrificed on Day 12 after CAR-T injection, and tumor tissues isolated from the right flanks were assessed histologically. Representative immunostained sections (hematoxylin and eosin (HE); CD20 for tumor cells; CD8 and CD4 for CAR-T cells) obtained from tissues in a representative mouse from each group are shown. The white rectangles (left, ×40) are displayed as magnified images (right, ×100). The scale bars in the pictures represent 100 μm.