Composite CD79A/CD40 co-stimulatory endodomain enhances CD19CAR-T cell proliferation and survival

Abstract

Adoptively transferred CD19 chimeric antigen receptor (CAR) T cells have led to impressive clinical outcomes in B cell malignancies. Beyond induction of remission, the persistence of CAR-T cells is required to prevent relapse and provide long-term disease control. To improve CAR-T cell function and persistence, we developed a composite co-stimulatory domain of a B cell signaling moiety, CD79A/CD40, to induce a nuclear translocating signal, NF-κB, to synergize with other T cell signals and improve CAR-T cell function. CD79A/CD40 incorporating CD19CAR-T cells (CD19.79a.40z) exhibited higher NF-κB and p38 activity upon CD19 antigen exposure compared with the CD28 or 4-1BB incorporating CD19CAR-T cells (CD19.28z and CD19.BBz). Notably, we found that CD19.79a.40z CAR-T cells continued to suppress CD19⁺ target cells throughout the co-culture assay, whereas a tendency for tumor growth was observed with CD19.28z CAR-T cells. Moreover, CD19.79a.40z CAR-T cells exhibited robust T cell proliferation after culturing with CD19⁺ target cells, regardless of exogenous interleukin-2. In terms of in vivo efficiency, CD19.79a.40z demonstrated superior anti-tumor activity and in vivo CAR-T cell proliferation compared with CD19.28z and CD19.BBz CD19CAR-T cells in Raji-inoculated mice. Our data demonstrate that the CD79A/CD40 co-stimulatory domain endows CAR-T cells with enhanced proliferative capacity and improved anti-tumor efficacy in a murine model.

Graphical abstract

Introduction of CD79A/CD40 intracellular domain into CD19CAR leads to an increase in p38 and NF-κB signals and improves in vivo persistence of CD19CAR-T cells. CD19CAR-T cells with CD79A/CD40 endodomain may convey better survival than the currently used CD19CAR-T cells with CD28 or 4-1BB endodomain.

Introduction

Adoptive immunotherapy for B cell malignancies using T cells modified to express the CD19 chimeric antigen receptor (CAR) has demonstrated promising clinical outcomes. Remission rates as high as 90% were reported in patients with relapsed/refractory acute B cell lymphoblastic leukemia/lymphoma (B-ALL), which has dismal survival outcomes with other standard therapies.1, 2, 3, 4, 5, 6, 7 Nevertheless, compared to efficacy in B-ALL, CAR-T cell efficacy in chronic lymphocytic leukemia and B cell non-Hodgkin’s lymphoma patients is modest, with complete response rates reported to be in the 30%–70% range across trials.8, 9, 10, 11, 12 To further extend the efficacy of CAR-T therapy, various approaches have been pursued, such as CAR design modification, gene transfer and expansion techniques, selection of a specific T cell subset to express CAR, and lympho-depleting regimens and CAR-T cell dosage.13, 14, 15, 16 Among these strategies, CAR structural modification has been widely investigated to maximize CAR-T cell potency.^17,18 Successful treatments in clinical trials using second-generation CD19CAR-T cells incorporating either CD28 or 4-1BB co-stimulatory domains have been reported.1, 2, 3, 4, 5, 6^,8, 9, 10, 11 Even though the initial clinical responses were impressive, in vivo CAR-T persistence after infusion, long-term efficacy, and disease-specific responses are still not satisfactory.

To enhance CAR-T cell function and persistence, in this study, we aimed to maximize CAR-T cell signaling using a novel co-stimulatory domain. Upon T cell activation, various downstream signaling pathways are stimulated to further activate and induce the nuclear translocation of transcription factors such as nuclear factor of activated T cells (NFAT), NF-κB, and activated protein-1 (AP-1) to interact and promote gene transcription, which is essential for T cell proliferation and differentiation into effector T cells.^19,20 NF-κB is essential for T cell survival as well as anti-tumor control. A previous study demonstrated that NF-κB is part of a critical signaling pathway for 4-1BB co-stimulation and contributes to increases in CAR-T cell proliferation and persistence compared to CD28.²¹ We therefore attempted to increase NF-κB signaling in CAR-T cells by introducing the B cell receptor (BCR) co-receptor, CD79A, and CD40 signaling domains. Both receptors activate a similar group of intracellular kinases and transcription factors, including NF-κB, NFAT, and AP-1.^22,23 Previous studies demonstrated that the co-engagement of BCR and CD40 receptors produces signal integration and results in the synergistic activation of B cells.24, 25, 26 Owing to the well-established synergistic effects of BCR and CD40 receptors, as well as the possibility of cooperative CD40 signaling in T cells, we developed a composite co-stimulatory domain of CD79A, which is a signal transduction molecule associated with the BCR, and CD40 to generate novel CD19CAR-T cells. As preliminary experiments, second-generation CARs containing CD79A or CD40 domains with CD3ζ were generated; however, neither enhanced cytokine production nor cell proliferation after stimulation was observed. When we combined both domains tandemly with CD3ζ, we observed robust cytokine production and proliferation. We hypothesized that the CD79A/CD40 signaling domain would cooperate to exert crucial intracellular signaling, mainly involving NF-κB, to synergize with T cell signaling and improve CAR-T cell function.

Results

Generation of CD19CAR-T cells and the NF-κB signaling assay

To examine whether the CD79A/CD40 co-stimulatory domain provides additional signals to CAR-T cells, we developed a CD19CAR construct that incorporated the composite CD79A/CD40 co-stimulatory domain (CD19.79a.40z) into the original CD19CAR backbone. The second-generation CD19CAR with the CD28 co-stimulatory domain (CD19.28z) was used as a control (Figure 1A). We attempted to confirm whether the co-stimulatory domain could generate a significant signal upon antigen stimulation using Jurkat-based triple parameter T cell reporter (Jurkat-TPR) cells, in which the activity of NF-κB, NFAT, and AP-1 could be measured.^27,28 However, AP-1 signaling (mCherry) was not evaluable in our institute due to the lack of an appropriate laser. Notably, CD19.79a.40z CAR-transduced Jurkat-TPR cells generated significantly higher NF-κB and NFAT signaling following CD19⁺ target cell stimulation compared to CD19.28z CAR (Figure 1B). To introduce CD19CAR genes into human primary T cells, we used the sequence of events described in Figure 1C. We successfully introduced CD19.79a.40z CAR and CD19.28z CAR genes into human CD3⁺ cells with similar transduction efficacy (range, 27.0%–40.1% and 17.0%–43.3%, respectively) and achieved >97% purity (range, 96.9%–99.3%) after tEGFR selection (Figures 1D and 1E). Notably, compared to CD19.28z CAR-T cells, the expansion of CD19.79a.40z CAR-T cells after EBV-LCL stimulation was significantly increased (Figure S1A). For the CD4 and CD8 subsets of CAR-T cells after transduction and expansion, we found no differences between the co-stimulatory domains (Figures S1B and S1C).

Figure 1.

Generation of CD19CAR-T cell and transcription factor assay

(A) Schematic of CD19CAR constructs. Either a CD28 (CD28z) or a CD79A/CD40 (CD79a.40z) co-stimulatory domain was fused into anti-CD19 scFv-H/CH2CH3-CD28TM, followed by CD3ζ and tEGFR. Ctrl, control; IgG4 Fc-derived spacer of hinge (H)-CH2-CH3; tEGFR, truncated EGFR; TM, transmembrane domain; VH, heavy-chain variable fragment; VL, light-chain variable fragment. (B) Activation of Jurkat-triple parameter reporter (TPR) cells transduced with different CAR constructs and stimulated with K562, CD19-K562, or Raji cells at a 1:1 ratio for 24 h. Activation of NF-κB and NFAT was measured with flow cytometry. ΔNF-κB or NFAT mean fluorescence intensity (MFI) was calculated as the difference between each datum and the mean of stimulated/untransduced Jurkat-TPR cells. Assays were done in triplicate, and the data are presented as the means ± SDs (Student’s t test; ∗∗p < 0.01, ∗∗∗∗p < 0.0001). (C) Experimental schematic of CD19CAR-T cell generation and analysis of CD19CAR-T cells. Stim, stimulation; Tdx, transduction. (D) Representative flow plots of transduced primary human T cells stained for tEGFR expression after 10 days of activation before (top) and after selection (bottom). The numbers are percentages of tEGFR⁺ cells of lymphocyte gate. (E) Transduction efficacy and purification of CD19CAR-T cells from 5 different donors. The data are the means ± SEMs (Student’s t test).

CD19.79a.40z CAR exhibited higher IL-2 secretion and robust CAR-T cell proliferation upon CD19⁺ target cell stimulation in vitro

To determine the cytokine secretion capacity upon antigen exposure, we incubated CAR-T cells with CD19⁺ target cells and stained the cells for intracellular interferon-γ (IFN-γ) and interleukin-2 (IL-2). In CD4⁺ (as CD8⁻) fraction, CD19.28z CAR showed a higher proportion of IFN-γ responders after CD19-K562 cell stimulation, whereas IL-2 responders were more frequently observed in CD19.79a.40z CAR. In CD8⁺ fraction, the percentage of IFN-γ⁺ cells was higher in CD19.28z CAR as in CD4⁺ fraction; however, the percentages of IL-2-secreting cells were similar in CD19.28z CAR and CD19.79a.40z CAR (Figures 2A and 2B). Both types of CD19CAR-T cells produced similar amounts of IFN-γ, IL-2, and tumor necrosis factor-α (TNF-α) after co-culturing with CD19-K562 cells, if we did not segregate CD4⁺ and CD8⁺ cells (Figure 2C). We observed a higher mean IL-2 concentration in CD19.79a.40z CAR-T cells compared to that in CD19.28z CAR-T, although the difference was not statistically significant. We next assessed whether CARs harboring CD79a.40z signaling domains could enhance CAR-T cell proliferation. Notably, robust CAR-T cell expansion was observed for CD19.79a.40z CAR, and a higher number of viable cells was maintained after 2 weeks of culture, regardless of IL-2 supplementation (Figures 2D and 2E). The marked differences in cell proliferation were statistically significant after culturing for 5 and 10 days without and with exogenous IL-2, respectively. CD19.79a.40z CAR-T cells exhibited a greater CAR-T cell proliferation upon antigen encounter, which may be partially due to higher IL-2 secretion capacity compared to CD19.28z CAR-T cells.

Figure 2.

Cytokine secretion, T cell proliferation, and cytotoxic capacity upon CD19⁺ target cell stimulation

(A) Representative flow plots of intracellular cytokine staining. The numbers indicate the percentages of cytokine positive fraction. The percentage of cytokine producing cells was calculated separately in CD4 and CD8 fractions. (B) Intracellular cytokine staining for IFN-γ and IL-2 in CD4 and CD8. T cells were stimulated with CD19-K562 cells at an E:T ratio of 1:2 for 4 h, and then permeabilized and stained for intracellular IFN-γ and IL-2. CD4⁺ (as CD8⁻) and CD8⁺ fractions were analyzed separately. (C) Measurement of IFN-γ, IL-2, and TNF-α secretion. T cells and γ-irradiated CD19-K562 cells were co-cultured at a 1:1 ratio for 16 h, and culture supernatants were analyzed using an ELISA. (D and E) T cell proliferation assay. T cells were stimulated once with γ-irradiated CD19-K562 cells at a 1:1 ratio and cultured (D) without IL-2 supplementation or (E) with IL-2 supplementation (50 IU/mL). T cell expansion was measured by counting viable cells. Arrows mark the day of CD19-K562 cell stimulation. (F) Cytotoxicity was measured with the ⁵¹Cr release assay. T cells were incubated with ⁵¹Cr-labeled K562, CD19-K562, primary Ph⁻ B-ALL, or primary Ph⁺ B-ALL cells at various E:T ratios for 4 h and analyzed for specific cytolysis. (G) Prolonged co-culture assay. T cells and CellTrace Violet-labeled CD19-K562 cells were co-cultured at E:T ratios of 1:8 (left) and 1:16 (right) for a total of 9 days without exogenous IL-2. The percentages of effector and residual target cells were assessed with flow cytometry at the indicated time points. All of the data were pooled from 3 different donors and are presented as the means ± SEMs. One-way ANOVA for (B), (C), and (F); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; 2-way ANOVA for (D) and (E); ∗∗∗∗p < 0.0001; Student’s t test for (G); ∗p < 0.05.

Greater CAR-T cell proliferation of CD19.79a.40z CAR sustained anti-tumor efficacy in a co-culture assay

We hypothesized that the superior proliferative capacity of CD19.79a.40z CAR would result in improved anti-tumor cytotoxicity. The cytotoxicity of CD19.79a.40z CAR-T and CD19.28z CAR-T cells against CD19-K562 cells and two primary B-ALL cell lines was similar. However, cytolysis against K562 cells was significantly higher with CD19.28z CAR-T cells, which was possibly caused by higher natural killer cell activity with CD19.28z CAR-T cells (Figure 2F). Next, we performed a co-culture assay to examine the integrated activities of anti-tumor cytotoxicity as well as the potential T cell expansion of CD19CAR-T cells. We co-cultured T cells with CellTrace Violet (CTV)-labeled CD19-K562 cells at various effector to target (E:T) ratios. At high E:T ratios, both CD19.79a.40z CAR-T and CD19.28z CAR-T cells completely eradicated target cells after 5 days, and the suppression continued until the end of the culture (Figures S2A and S2B). The T cell proliferation assay showed a marked difference in CAR-T cell expansion after 5 days of culture (Figure 2D). Therefore, we extended the co-culture time to a total of 9 days and further decreased the E:T ratio. We found that CD19.28z CAR-T cells rapidly killed the target cells in the first few days. However, a tendency for target cell outgrowth was observed after 1 week of culture, whereas CD19.79a.40z CAR-T cells continued to eradicate CD19-K562 cells throughout the culture period (Figures 2G and S2C). These results indicate that greater proliferation and persistence of CD19.79a.40z CAR-T cells result in superior anti-tumor efficacy compared to CD19.28z CAR-T cells.

Structural modification in the hinge domain preserved the proliferative capacity

Because previous studies demonstrated that CAR expressing short hinge (SH) has superior anti-tumor activity compared to long hinge (LH) that contains CH2CH3 in vivo,²⁹ we modified the CD19CAR structure by removing CH2CH3 from the hinge domain and left only 12 amino acids (aa) in the hinge (Figure 3A). In addition, previous studies demonstrated superior CAR-T cell persistence in vivo established by CD19CAR with the 4-1BB co-stimulatory domain.^4,8,30 Therefore, we then further compared our construct with the 4-1BB signaling domain. We successfully transduced CD19CAR.SH into human CD3⁺ cells with slightly higher transduction efficacy in CD19.79a.40z.SH CAR (Figure 3B). Both CD19.79a.40z.SH CAR-T cells and CD19.BBz.SH CAR-T cells robustly expanded upon stimulation compared to CD19.28z.SH CAR-T cells (Figure 3C). We next assessed T cell signaling using CD19CAR.SH-transduced Jurkat-TPR cells. We observed markedly higher NF-κB and NFAT activity in CD19.79a.40z.SH CAR and CD19.BBz.SH CAR compared to CD19.28z.SH CAR, whereas relatively small but significant differences were observed between CD19.79a.40z.SH CAR and CD19.BBz.SH CAR (Figure 3D).

Figure 3.

CD19CAR short hinge (SH) T cell generation and in vitro assays

(A) Schematic of CD19CAR SH constructs. A 4-1BB (BBz.SH), CD28 (CD28z.SH), or CD79A/CD40 (CD79a.40z.SH) co-stimulatory domain was fused into anti-CD19scFv-H-CD28TM followed by CD3ζ and tEGFR. IgG4 Fc-derived spacer of H only (12 aa); tEGFR, truncated EGFR; TM, transmembrane domain; VH, heavy-chain variable fragment; VL, light-chain variable fragment. (B) Transduction efficacy and purification of CD19CAR SH T cells. (C) Ex vivo expansion of CD19CAR SH T cells stimulated with γ-irradiated EBV-LCL at a 1:7 responder:stimulator ratio for 10 days. Data were pooled from 3 different donors and are shown as the means ± SEMs here and in (B) (Student’s t test for B; 2-way ANOVA for C; ∗∗∗∗p < 0.0001). (D) Activation of Jurkat-TPR cells transduced with different CAR SH constructs and stimulated with K562, CD19-K562, or Raji cells at a 1:1 ratio for 24 h. Activation of NF-κB (left) or NFAT (right) was measured with flow cytometry. ΔNF-κB or NFAT MFI was calculated as the difference between each datum and the mean of stimulated/untransduced Jurkat-TPR cells. Assays were done in triplicate, and the data are presented as the means ± SDs (Student’s t test; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001). (E) Intracellular cytokine staining for IFN-γ and IL-2 in CD4 and CD8. T cells were stimulated with CD19-K562 cells at an E:T ratio of 1:2 for 4 h, and then permeabilized and stained for intracellular IFN-γ and IL-2. CD4⁺ (as CD8⁻) and CD8⁺ fractions were analyzed separately. (F and G) T cell proliferation assay. T cells were stimulated once with γ-irradiated CD19-K562 cells at a 1:1 ratio and cultured (F) without IL-2 supplementation or (G) with IL-2 supplementation (50 IU/mL). Arrows mark the day of CD19-K562 cell stimulation. (H) Non-apoptotic cell fraction of CAR-T and tEGFR-T cells after 24 h co-culture with K562 or CD19-K562 cells. Data were pooled from 3 different donors and are presented as the means ± SEMs in (E)–(H). One-way ANOVA for (E) and (H); 2-way ANOVA for (F) and (G); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To validate CD19CAR.SH functions, we performed an intracellular cytokine assay and found a similar proportion of IFN-γ responders among the co-stimulatory domains. IL-2 responders were increased in CD19.79a.40z.SH CAR compared to CD19.CD28z CAR in the CD4⁺ fraction, although the difference was not observed in the CD8⁺ fraction (Figure 3E). Remarkably, both LH and SH CD19.79a.40z CAR exhibited significantly higher CAR-T cell proliferation compared to CD19.BBz.SH CAR or CD19.28z CAR, regardless of exogenous IL-2 (Figures 3F and 3G). Moreover, we did not observe a significant difference in T cell growth between LH and SH structures of each of the tested CAR constructs. The hinge domain modification of the CD19.79a.40z CAR maintained higher signaling capacity compared with CD19.28z, but showed lower capacity than CD19.BBz CAR, while preserving its advantage in CAR-T cell proliferation compared to CD19.28z CAR and CD19.BBz CAR.

To investigate whether the higher proliferation observed in CD19.79a.40z.SH CAR was a reflection of reduced activation-induced cell death, we performed annexin V/propidium iodide (PI) assay. After 24 h of stimulation, we observed similar apoptotic cell frequencies among CD19CAR-T cells with 3 different intracellular domains (ICDs) (Figure 3H).

CD79A/CD40 ICD mediated greater intracellular signals through p38 and NF-κB

To further investigate the signaling alteration after CD19 ligation, we performed the intracellular phospho-flow analysis using the CD19CAR-T cells with 3 different ICDs. At 30 min post CD19-K562 stimulation, the CAR-T cells rapidly upregulated levels of all phospho-proteins examined (Figure 4A). At 30 min, CD19.79a.40z.SH CAR demonstrated the greater signaling in phospho-p38 and phospho-NF-κB compared to two other constructs, whereas CD19.CD28z.SH CAR showed the higher signaling in phospho-Erk (Figures 4B–4D). CD79A/CD40 and 4-1BB ICDs showed similarly higher signaling in NF-κB at 30 min. At 240 min, similar levels of signaling were observed in phospho-Erk among all of the constructs (Figure 4C). In phospho-p38 and phospho-NF-κB, we observed sustained signals in CD19.79a.40z.SH CAR compared to two other constructs (Figures 4B and 4D).

Figure 4.

Analysis of intracellular signaling after stimulation through the CAR

(A) Representative flow plots of intracellular phospho-specific staining of p38, Erk, and NF-κB. CAR-T cells and tEGFR-T cells were stimulated with CD19-K562 cells at a 1:5 ratio for 30 min. Data are representative of 3 independent experiments with 3 donors. (B–D) MFI data at indicated time after stimulation of phospho-p38 (B), phospho-Erk (C), and phospho-NF-κB (D). Each dot represents the results of one donor. Error bars shows means the means ± SEMs. Significance levels are shown only for the comparisons with significance; 1-way ANOVA; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 in (B)–(D).

CD19.79a.40z CAR sustained higher CAR-T cell proliferation without changing the CAR-T cell fate and exhaustion phenotypes

Next, we hypothesized that the CD79A/CD40 ICD may not mediate proper T cell differentiation signals because it has been artificially produced as a composite domain fused to a B cell signaling moiety. We therefore sought to investigate the maximal capability of CAR-T cell growth following multiple weekly stimulations as well as CAR-T cell fate and exhaustion phenotypes (Figure 5A). Although all CAR-T groups exhibited equal levels of vigorous expansion following 3 consecutive weekly CD19-K562 stimulations, after the fourth week of stimulation, CD19.79a.40z.SH CAR-T cells proliferated at a significantly higher rate than the other 2 types of CAR-T cells (Figure 5B). To investigate changes in T cell differentiation, we performed staining for CD45RA and CD62L for classification into CD45RA⁺ CD62L⁺ naive T (T_N), CD45RA⁻ CD62L⁺ central memory T (T_CM), CD45RA⁻ CD62L⁻ effector memory T (T_EM), and CD45RA⁺ CD62L⁻ effector memory re-expressing CD45RA T (T_EMRA) cell subsets. Following multiple weekly stimulations, the T_CM subset of CD19.79a.40z CAR-T cells significantly decreased after the second and third stimulations, and the T_EM subset significantly increased after the third stimulation, compared to other co-stimulatory domains (Figures 5C and S3A). However, no significant differences were found at the end of the culture (Figure 5C). In addition, the expression of T cell exhaustion markers, including programmed cell death 1 (PD-1), cytotoxic T lymphocyte-associated protein 4 (CTLA-4), T cell immunoglobulin and mucin domain containing-3 (TIM-3), and lymphocyte-activation gene 3 (LAG3), did not change significantly in CD19-K562 cell-stimulated CD19CAR-T cells despite repetitive stimulation except early upregulation of LAG3 in CD79a.40z CAR (Figures 5D and S3B). CD19.79a.40z.SHCAR sustained higher CAR-T cell proliferation with indistinct phenotypes compared to CD19.28z.SH CAR and CD19.BBz.SH CAR following repetitive stimulation.

Figure 5.

CD19CAR SH T cell generation, T cell proliferation, and T cell phenotype assays

(A) Schematic of repetitive antigen stimulation and T cell phenotype assays. The CAR-T cells were stimulated weekly with γ-irradiated CD19-K562 cells at a 1:1 ratio for 4 consecutive weeks and cultured with IL-2 supplementation (50 IU/mL). Then, the differences in T cell growth, T cell differentiation, and exhaustion phenotypes were assessed before and after weekly stimulation. (B) Fold expansion during the repetitive T cell proliferation assay. (C) T cell subset heatmap to determine CD45RA⁺ CD62L⁺ naive T (T_N), CD45RA⁻ CD62L⁺ central memory T (T_CM), CD45RA⁻ CD62L⁻ effector memory T (T_EM) cells, and CD45RA⁺ CD62L⁻ effector memory re-expressing CD45RA T (T_EMRA) cells before and after stimulation over time. Percentages of total cells are shown. (D) T cell exhaustion heatmap to determine PD-1⁺, CTLA-4⁺, TIM-3⁺, and LAG3⁺ T cells before and after simulation over time. Percentages of positive cells for indicated molecules are shown. Data were pooled from 3 different donors and are shown as the means ± SEMs in (B) (2-way ANOVA; ∗∗∗∗p < 0.0001). Heatmap data were pooled from 3 different donors and are shown as the mean for (C) and (D) (1-way ANOVA at each time point; ∗p < 0.05, ∗∗p < 0.01).

CD19.79a.40z.SH CAR demonstrated superior in vivo efficacy in Raji-bearing murine model

To evaluate the potential anti-tumor efficacy in vivo, we studied Raji-bearing NOD-SCID common-γchain (NSG) knockout (KO) mice treated with a low CD19CAR-T cell dosage of different co-stimulatory domains (Figure 6A). CD19.79a.40z.SH CAR-T cells exhibited the better tumor suppression activity, and the difference became significant after 21 days of CAR-T cell infusion (Figures 6B and 6C). Although not all mice were necessarily dissected to confirm the cause of death, the mice that were dissected usually had tumor masses in the liver. Some of the mice with bioluminescence around the brain without a strong bioluminescence in their body were suggestive of death from central nervous system lesions, because we sometimes observed limb-paralyzed mice.³¹ In the Raji-inoculated lymphoma model, the survival was significantly higher in mice treated with CD79a.40z.SH CAR compared to the other treatment arms (Figure 6D).

Figure 6.

In vivo efficacy of CD19.79a.40z.SH CAR-T cells in Raji-inoculated mice model

(A) Schematic of the in vivo experiments. NOD-SCID common-γ chain KO (NSG) mice were inoculated with 0.5 × 10⁶ Raji/ffluc-GFP cells via tail vein injection on day 0, followed by injection of either 1 × 10⁶ tEGFR- or CD19CAR SH-transduced T cells 1 week later. Tumor burden was assessed using bioluminescence imaging (BLI) at the indicated time points. (B) Representative BLI of Raji-inoculated mice treated with CD19CAR SH or control T cells over time. (C) Tumor burden (average radiance was calculated as the sum of the radiance from each pixel inside the region of interest divided by the number of pixels or super pixels) of mice treated with tEGFR- or CD19CAR SH-transduced T cells at the indicated time points. Data are presented as the means ± SEMs (2-way ANOVA; ∗p < 0.05, ∗∗p < 0.01) and pooled from 3 independent experiments using 3 different donors (CD19CAR SH T cells: n = 7–8 per group; tEGFR-T: n = 4). (D) Kaplan-Meier analysis of survival of Raji/ffluc-bearing NSG mice (Gehan-Breslow-Wilcoxon test; ∗p < 0.05, ∗∗∗∗p < 0.0001).

CD19.79a.40z.SH CAR-T showed enhanced in vivo CAR-T cell proliferation and persistence in a NALM-6-bearing murine model

To evaluate in vivo efficacy in another model, we studied NALM-6-bearing NSG mice treated with the same low CD19CAR-T cell dosage to evaluate the efficacy and functional limits of different co-stimulatory domains (Figure 7A). CD19.79a.40z.SH CAR-T cells profoundly suppressed tumor cell growth in NALM-6-bearing mice for 4 weeks after tumor inoculation compared to other CAR-T groups (Figures 7B and 7C). For long-term assessment, we observed the distinct kinetics of tumor eradication in mice treated with each type of CAR-T cells. CD19.79a.40z.SH CAR-T cells and CD19.BBz.SH CAR-T cells exhibited uniform and effective tumor suppression in the first few weeks following adoptive transfer. Nevertheless, mice treated with CD19.BBz.SH CAR-T did not achieve complete tumor eradication, leading to increased tumor recurrence and decreased survival compared to CD19.79a.40z.SH CAR-T cells (Figures 7D and 7E). Treatment with CD19.28z.SH CAR-T cells yielded inconsistent results and was associated with early tumor progression in a subset of treated mice (Figures 7D and 7E). However, complete tumor eradication was ultimately observed in a small subset of mice treated with CD19.28z.SH CAR-T cells, similar to those seen in CD19.79a.40z.SH CAR-T cells in long-term follow-up (Figures 7D and 7E). CD19.79a.40z.SH CAR-T cells slightly prolonged survival in the early phase, leading to small but significant improvement in overall survival throughout the observation period compared to CD19.BBz.SH CAR (Figure 7E). To investigate CAR-T cell expansion in vivo, we assessed the frequency of transferred T cells in mouse peripheral blood 1 week after T cell transfer. We observed higher frequencies of CD19.79a.40z.SH CAR-T cells in vivo, which may simultaneously correlate with the potent anti-tumor efficacy of these cells compared to CAR-T cells harboring constructs with other co-stimulatory domains (Figure 7F). We did not observe any adverse reactions during the treatment period. These results indicated greater in vivo CAR-T cell proliferation and persistence of CD19.79a.40z.SH CAR, which correlated with superior anti-tumor activity compared to CD19.79a.28z.SH CAR and CD19.BBz.SH CAR.

Figure 7.

CD19.79a.40z.SH CAR-T cells profoundly suppress tumor cell growth and mediate greater in vivo T cell proliferation

(A) Schematic of the in vivo experiments. NSG mice were inoculated with 0.5 × 10⁶ NALM-6/ffluc-GFP cells via tail vein injection on day 0, followed by injection of either 1 × 10⁶ tEGFR- or CD19CAR SH-transduced T cells 1 week later. Tumor burden was assessed using BLI at the indicated time points. (B) Representative BLI of NALM-6-inoculated mice treated with CD19CAR SH or control T cells over time. (C) Tumor burden (average radiance) of mice treated with tEGFR-transduced T cells, and each CD19CAR SH-transduced T cell at the indicated time points. Data are presented as the means ± SEMs (2-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001) and pooled from 3 independent experiments using 3 different donors (CD19CAR SH T cells: n = 13 per group; tEGFR-T: n = 7). (D) Tumor burden (average radiance) of individual mice treated with tEGFR-, or CAR-T cells over time. Summary of 3 independent experiments using 3 different donors (CD19CAR SH T cells: n = 13 per group; tEGFR-T: n = 7). (E) Kaplan-Meier analysis of survival of NALM-6/ffluc-bearing NSG mice treated with 1 × 10⁶ tEGFR- or CD19CAR SH-transduced T cells (Gehan-Breslow-Wilcoxon test; ∗p < 0.05, ∗∗∗∗p < 0.0001). (F) Percentage of circulating human T cells in peripheral blood of mice 7 days after T cell transfer. Data are presented as the means ± SEMs (1-way ANOVA; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

Discussion

In the present study, we generated an innovative CAR by incorporating a composite co-stimulatory domain consisting of CD79A/CD40 into the CD19CAR backbone and compared this new CAR with the well-established CD28 and 4-1BB co-stimulatory domains. Using the CAR-transduced Jurkat-TPR cell system and CAR-transduced primary T cells, we demonstrated that CD79a.40z CAR-T cells exhibited increased NF-κB and p38 activity and enhanced in vitro proliferation upon CD19 stimulation compared to CD28z and BBz CAR-T cells. Owing to the prominent T cell growth, CD19.79a.40z CAR-T cells continuously suppressed tumor cells until the end of the co-culture period despite low E:T ratios in the co-culture assay. Moreover, CD19.79a.40z CAR-T cells profoundly suppressed tumor cell growth in NALM-6 or Raji-bearing NSG mice for a few weeks after tumor inoculation, leading to long-term effective tumor clearance and prolonged overall survival compared to other treatment arms. Although we observed the better results of BBz.CAR-T cells in some in vitro experiments, the survival of mice treated with BBz.CAR-T was inferior in mice experiments. We think that the reason why CD19.BBz CAR-T cells could not demonstrate comparable in vivo tumor control with CD19.79a.40z CAR is mainly due to the low CAR-T dose. Because the 4-1BBz ICD is known to contribute to relatively slow and consistent proliferation after stimulation, this experimental setting may have resulted in a major disadvantage for CD19.BBz CAR-T cells, particularly in the NALM6-inoculated leukemia model.

We demonstrated enhanced and sustained signaling through NF-κB and p38 in the CD19CAR-T cells with the CD79A/CD40 co-stimulatory domain. Accordingly, the advantage in greater T cell proliferation may be due to a mechanism that involves selective NF-κB signaling enhancement by the CD79A/CD40 co-stimulatory domain. Canonical and non-canonical NF-κB signaling pathways are critical for 4-1BB and contribute to greater CAR-T cell proliferation compared to CD28, as shown in previous studies.^21,32 Consequently, this strong signaling affected CAR-T cell expansion and persistence in the present study, although in terms of cytokine secretion and cytotoxicity, the CD79A/CD40 module was comparable to the other two co-stimulatory domains. The significant CAR-T cell proliferation established by the CD79A/CD40 co-stimulatory domain emphasized the advantage of the B cell-derived signaling effect on CAR-T cells after antigen engagement. This is the second CAR using the B cell signaling domains. The previous CAR-Ts incorporating MyD88/CD40 signaling domains have been reported and also show better proliferation upon antigen stimulation and further retains a less differentiated state.^33,34 The introduction of these signal domains may have affected CAR-T cell metabolism.³⁵ Strategies to incorporate B cell signaling into the development of CAR-T may still be open to further investigation.

We demonstrated that the introduction of CD79A/CD40 into CAR-T cells resulted in an increase in both NF-κB and NFAT signaling. NFAT activity drives T cell anergy and exhaustion when solely activated,36, 37, 38 whereas the balanced activation of NFAT and its transcriptional partner, AP-1, promotes T cell activation and skews the exhaustion programs.^19,37, 38, 39 Thus, the high NFAT signals induced by the CD79A/CD40 module may accelerate the exhaustion process of T cells. Acquiring more knowledge about the signals that are more favorable to T cell survival may allow us to exploit them in CAR-based therapeutic strategies. Several reports have described the synthetic production of signals,^40,41 and thus, we may be able to optimize T cell fate by using such a strategy, and ultimately improve CAR-T cell function.

We further attempted to define another plausible factor that influences the increase in CAR-T cell growth by evaluating T cell phenotypic changes after antigen encounter. Because CD79A and CD40 are both derived from B cell signaling molecules, we hypothesized that their distinct B cell signaling would be able to cooperate with T cell signaling while not substantially inducing T cell exhaustion. Despite multiple rounds of antigen stimulation, CD19.79a.40z CAR-T cells sustained the capability to generate high numbers of CAR-T cells with an insignificant change in T cell fate and exhaustion phenotypes compared to control CAR-T cells. B cell-derived ectopic signals did not alter T cell phenotypic changes after antigen encounter.

Another factor that could account for the differences between CD19.79a.40z CAR-T cells and CAR-T cells harboring two other co-stimulatory domains is the kinetics of tumor eradication, which was described in a previous study by Zhao et al.³⁰ They investigated the distinct tumoricidal profiles between CD19.28z CAR-T and CD19.BBz CAR-T cells by lowering the T cell doses in a xenograft model. Notably, CD19.28z CAR-T cells showed greater anti-tumor activity, which resulted in rapid tumor control. However, CD19.BBz CAR-T cells gradually eradicated tumors and eventually achieved similar tumor control according to the greater CAR-T cell persistence. In this study, CD19.28z CAR-T cells showed rapid tumor cell killing in a co-culture assay, similar to an instant but inconsistent tumor eradication pattern in NALM-6-bearing leukemia mice treated with low-dose CAR-T cells, which led to the early recurrence of tumor outgrowth and decreased survival outcomes. CD19.79a.40z CAR-T cells exhibited continuous cytotoxic activity in the co-culture assay, and their greater T cell proliferation and persistence profoundly cleared tumor cells and delayed tumor recurrence in vitro and in vivo. Because we observed that CD19.79a.40z CAR-T conveyed more favorable survival outcomes in the Raji-inoculated lymphoma model compared to that in the NALM-6 leukemia model, a relatively slow-starting, long-lasting response such as that produced by CD79a.40z CAR-T may be more appropriate for the treatment of lymphoma. In contrast, CAR-Ts such as CD28zCAR-T, which may cause a quick-start response, even if it is short-lived, may be more appropriate for the treatment of leukemia. The CD79a.40z CAR-T cells have two ICDs. Therefore, it is also interesting to compare it with the third-generation CAR-T, which has 2 ICDs, CD28 and 4-1BB.⁴²

Our findings have highlighted that CAR-T cell persistence alone may not be sufficient to enhance in vivo anti-tumor efficacy and survival outcomes. The combination of intense tumoricidal activity and T cell persistence is therefore required for further effective CAR-T cell therapy. From our perspective, the ideal co-stimulatory molecules should activate various downstream signaling pathways to facilitate different aspects of CAR-T cell function, promote memory T cell formation, and, ultimately, prevent T cell exhaustion.

In summary, we successfully generated a CD19.79a.40z CAR, which demonstrated strong p38, NF-κB and NFAT signaling upon antigen stimulation. In addition, we confirmed the possibility of incorporating B cell co-stimulatory signaling molecules into a CAR structure, which enhanced CAR-T cell proliferation and persistence compared to other conventional signaling molecules.

Materials and Methods

Cell lines

K562, Raji, and NALM-6 cell lines were maintained in our laboratory. The authenticity of all cell lines is routinely validated by examining their immunophenotypes using flow cytometry. Cells are cultured for a maximum of 2 months before use. A CD19⁺ Epstein-Barr virus-transformed lymphoblastoid cell line (EBV-LCL) was used as a source of feeder cells for T cell culture. Cell lines were cultured in RPMI 1640 medium containing 10% fetal bovine serum, 0.8 mM l-glutamine, and 1% penicillin-streptomycin. K562 cells were retrovirally transduced to express truncated CD19 (CD19-K562), as described elsewhere.⁴³ Raji and NALM-6 cells were lentivirally transduced with the GFP-ffluc gene and sorted to a purity of >99%.

Human subjects

The research protocols were approved by the institutional review board of Nagoya University Graduate School of Medicine. Peripheral blood mononuclear cells (PBMCs) were obtained from healthy volunteer donors, and primary B-ALL cells were obtained from the bone marrow (BM) of two B-ALL patients. Written informed consent was obtained from each donor and patient in accordance with the Declaration of Helsinki.

Viral vector construction

CD19CAR, which includes CD28 (CD28z) and the composite CD79A/CD40 co-stimulatory domain (CD79a.40z), was generated by fusing the co-stimulatory domain with the anti-CD19 single-chain variable fragment (scFv)-hinge-CH2CH3-CD28 transmembrane domain followed by the CD3ζ ICD. The CD19-scFv was based on the clone FMC63 for all CAR constructs in this study.⁴⁴ The CD79a.CD40 ICD was created by connecting aa position 166–226 of CD79A (GenBank: NM_001783) and aa position 216–277 of CD40 (GenBank: NM_001250) with a 3-aa GGG linker. CD19CAR constructs were then fused with a truncated version of the epidermal growth factor receptor (tEGFR) that lacked EGF binding and intracellular signaling domains downstream of the self-cleaving T2A sequence. tEGFR is used as a transduction and selection marker by biotinylated Erbitux monoclonal antibody (mAb) (Bristol-Myers Squibb, New York, NY, USA).45, 46, 47 The CAR constructs were then subcloned into the LZRS-pBMN-Z vector, and the sequence was verified by direct sequencing. Gammaretroviral supernatants were produced using the Phoenix-Ampho system (Orbigen, San Diego, CA, USA). To improve CAR-T cell potency, we modified the CD19CAR structure to express a SH (12-aa sequence), which incorporated CD79A/CD40, CD28, or 4-1BB co-stimulatory domains.²⁹ The sequences of the second-generation CAR-T cells, including CD19.28z CAR and CD19.BBz CAR, were the same with previously reported and clinically applied.^29,48

Generation, expansion, and selection of CD19CAR-T cells

PBMCs were obtained from healthy donors and isolated by centrifugation of whole blood using Ficoll-Paque (GE Healthcare, Chicago, IL, USA). CD3⁺ or CD8⁺ lymphocytes were then purified with immunomagnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and activated with anti-CD3/CD28 beads (Invitrogen, Carlsbad, CA, USA). Retroviral transduction was done on days 3 and 4 using recombinant human retronectin fragment-coated plates (Retronectin, Takara Bio, Otsu, Japan) and centrifugation at 2,100 rpm for 1 h at 32°C. Transduced T cells were expanded in RPMI 1640 medium containing 10% human serum, 0.8 mM l-glutamine, 1% penicillin-streptomycin, and 0.5 μM 2-mercaptoethanol, which was supplemented with 50 IU/mL recombinant human IL-2. On day 10, CAR⁺ T cells were enriched using biotin-conjugated anti-EGFR mAb and streptavidin beads (Miltenyi Biotec). The enriched CAR⁺ cells were subsequently expanded by culturing with γ-irradiated EBV-LCL at a 1:7 responder:stimulator ratio, and supplemented with 50 IU/mL IL-2 on days 1, 4, and 7.⁴³ After 10 days, the expanded CAR-T cells were harvested and used for downstream experiments. Regarding T cell phenotypes, the CAR-T cells were stimulated weekly with γ-irradiated CD19-K562 cells at a 1:1 ratio for 4 consecutive weeks. For the mouse experiments, the enriched CAR⁺ cells were stimulated a second time with anti-CD3/CD28 beads at a 1:1 ratio for 7–10 days before use. Untransduced T cells were cultured in parallel according to the same procedures used for transduced T cells, except for the addition of virus.

Transcription factor assay

To assess NF-κB and NFAT signal transduction, we used the Jurkat-TPR cells that contain response elements for NF-κB and NFAT, which drive the expression of the fluorescent proteins cyan fluorescent protein and enhanced green fluorescent protein (EGFP), respectively.^27,28 AP-1 signaling was not evaluable in our institute due to lack of an appropriate laser and filter system. CD19CAR constructs were transduced into Jurkat-TPR cells and purified to express CD19CAR at >95%. Untransduced or CD19CAR-transduced Jurkat-TPR cells were then plated and mixed with γ-irradiated K562, CD19-K562, or Raji cells at a 1:1 E:T ratio for 24 h in triplicate cultures. Following 24 h of stimulation, Jurkat-TPR cells were harvested and analyzed for expression of the fluorescent protein of interest using flow cytometry.

Immunophenotyping

All of the samples were analyzed with flow cytometry on a FACSAria II and FACSCanto II (BD Biosciences, San Jose, CA, USA), and the data were analyzed using FlowJo software (Tree Star, Ashland, OR, USA). Human T cells were stained with combinations of the following mAbs conjugated to fluorophores: CD3, CD8, CD45RA, CD62L, IL-2, IFN-γ, PD-1, LAG3 (BD Biosciences), mCD45, CTLA-4, and TIM-3 (BioLegend, San Diego, CA, USA). Biotinylated Erbitux and streptavidin-phycoerythrin (BD Biosciences) were used to identify tEGFR⁺ cells.

Intracellular cytokine staining and cytokine secretion assay

Untransduced or CD19CAR-T cells were plated at an E:T ratio of 1:2 with either K562 or CD19-K562 cells in the presence of brefeldin A (Sigma-Aldrich, St. Louis, MO, USA) for 4 h. The stimulated cells were then fixed and permeabilized with a cell fixation/permeabilization kit (BD Biosciences) and stained with anti-IFN-γ or IL-2 mAb to detect intracellular cytokines. For the cytokine secretion assay, untransduced or CD19CAR-T cells were co-cultured at a 1:1 ratio with γ-irradiated CD19-K562 cells for 16 h. Supernatants from co-cultures were collected, and the IL-2, IFN-γ, and TNF-α concentration was measured with an ELISA (BD Biosciences).

Annexin V/PI assay

To identify apoptotic cells derived from CAR-T cells and K562/CD19-K562 differentially, K562/CD19-K562 cells were stained by CTV (Thermo Fisher Scientific, Waltham, MA, USA). CAR-T cells or tEGFR-T cells were co-cultured in a 1:1 mixture with K562 or CD19-K562 cell lines and then stained using the APC annexin V apoptosis detection kit with PI (BioLegend) according to the manufacturer’s protocol at 24 h post stimulation.

⁵¹Chromium (Cr) release assay and co-culture assay

Cytotoxic activity of CD19CAR-T cells against target cells was determined using a standard ⁵¹Cr release assay. Target cells were labeled for 2 h with ⁵¹Cr (PerkinElmer, Waltham, MA, USA), washed twice, dispensed at 2 × 10³ cells per well into triplicate cultures in 96-well round-bottom plates, and incubated for 4 h at 37°C with untransduced or CD19CAR-T cells at various E:T ratios. Supernatants were obtained and analyzed for radioactivity. The percentage of specific cytolysis was calculated using a standard formula ([experimental − spontaneous release]/[maximum load − spontaneous release] × 100%) and expressed as the mean of triplicate samples. To determine the integrated activities of cytotoxicity and growth of CAR-T cells upon encountering the target cells, untransduced or CD19CAR-T cells were co-cultured with CTV-labeled CD19-K562 cells (Thermo Fisher Scientific) at various E:T ratios without IL-2 supplementation. The percentages of effector cells and residual target cells were assessed with flow cytometry at the indicated time points up to 9 days.

CAR-T cell proliferation assay

Untransduced or CD19CAR-T cells were stimulated once with γ-irradiated CD19-K562 cells at a 1:1 ratio and cultured in culture medium with or without exogenous IL-2. T cell expansion was determined by counting viable cells after 72 h and up to 2 weeks.

Intracellular phospho-flow analysis

CAR-T cells or tEGFR-T cells and γ-irradiated CD19-K562 were mixed at a 1:5 ratio, spun down briefly, and incubated at 37°C for 30 min and 240 min for phospho-p38, phospho-Erk, and phospho-NF-κB. Cells were then fixed by the addition of 2% formaldehyde at 37°C for 10 min, permeabilized in ice-cold 90% methanol, and left on ice for 30 min. For staining, the following phospho-specific Abs were used: p38 (pT180/pY182, clone D3F9), Erk1/2 (pT202/pY204, clone D13.14.4E), and NF-κB p65 (pS536, clone 93H1) (all unconjugated; Cell Signaling Technology, Danvers, MA, USA), and donkey anti-rabbit IgG-Alexa Fluor 647 (secondary Ab; BioLegend).

NALM-6 and Raji xenograft model and bioluminescence imaging (BLI)

All of the murine experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee of Nagoya University Graduate School of Medicine. Six- to 8-week-old male NSG mice were intravenously inoculated with 0.5 × 10⁶ NALM-6-ffluc cells or Raji-ffluc cells via the tail vein. One week later, the mice were treated with tEGFR-transduced or CD19CAR-T cells. BLI was performed to assess tumor progression at the indicated time points using the IVIS Spectrum System (Caliper Life Science, Waltham, MA, USA). For the survival analysis, death was used as an endpoint. Peripheral blood was obtained by tail vein bleeding, and erythrocytes were removed using erythrocyte lysis buffer (QIAGEN GmBH, Hilden, Germany).

Statistical analysis

All of the experimental data are presented as the mean ± SEM or standard deviation (SD). The differences among results were evaluated with the Student’s t test, one-way ANOVA, or two-way ANOVA with Bonferroni’s or Tukey’s post-test correction when appropriate. Survival times were analyzed with the Kaplan-Meier method and a Gehan-Breslow-Wilcoxon test. Differences were considered statistically significant when p < 0.05. Statistical analysis was performed using GraphPad Prism version 8.1.2 software (GraphPad Software, La Jolla, CA, USA).