Generation of B7-H3 CAR-T cells and functional tests in vitro
In our previous studies, B7-H3 was found to be wildly expressed in human tumor cells, and B7-H3 CAR T cells have been developed against multiple tumors [5, 30]. Although significant antitumor efficacy was observed in vitro, there remains a certain distance to achieve a cure in B7-H3 CAR T cell therapy against solid tumors. In this study, we aimed to screen drugs to enhance immunotherapy efficacy. First, second-generation B7-H3 CAR T cells were developed, which comprised a single-chain variable fragment (scFv) derived from our library [30] and included 4-1BB and CD3ζ endodomains (Additional file 1: Fig. S1A). mCherry was inserted as a tracker for detecting the expression of CAR, which was measured using fluorescence microscopy (Additional file 1: Fig. S1B) and Fluorescence-activated cell sorting (FACS) (Additional file 1: Fig. S1C). Second, the expression level of B7-H3 in a BC cell line (MDA-MB-231) and CRC cell line (HCT116) was examined by FACS using B7-H3-PE (BioLegend, Cat# 331,606) as the primary antibody (Additional file 1: Fig. S1D). Finally, we assessed the cytolytic activity of B7-H3 CAR T cells, and CD19 CAR T cells were used as a negative control in the coculture system. We found that B7-H3 CAR T cells could efficiently lyse cancer cells at an effector:target (E:T) ratio of 4:1 at 24 h but under 20% lysis at an E:T ratio of 1:4 (Additional file 1: Fig. S1E).
The schema of the screening assay
To identify compounds that promote CAR T cell antitumor activity, a small molecule library of 624 unique compounds (TOPSCIENCE, L4000) was screened. The screening process is summarized in a flow diagram (Fig. 1A). Briefly, MDA-MB-231-FFluc cells, B7-H3 CAR T cells, and MDA-MB-231-FFluc cells cocultured with B7-H3 CAR T cells at an E:T ratio of 1:4, were treated with small molecules (1 μM) for 24–48 h. The relative MDA-MB-231-FFluc cell activity was quantified by measuring luminescence. The drug-treated groups of MDA-MB-231-FFluc cells and MDA-MB-231-FFluc cocultured with B7-H3 CAR T cells were detected for firefly luminescence by a CLARIOstar plate reader (BMG LABTECH, Germany). In the drug-treated B7-H3 CAR T cell group, T cell markers (CD45RO and CD62L) were analyzed by FACS. The screening data are shown in Fig. 1 B and C. Among the three screening groups, three of our most potent hits were 3957(BML284), 6014(Picropodophyllin, PPP), 6114(JK184). After treatment with BML284/PPP/JK184 (1 μM) for 24 h, MDA-MB-231 cells exhibited shrunken morphology and the images were captured on a confocal microscope (Fig. 1D). B7-H3 CAR T cells showed an effector memory phenotype after treatment with BML284/PPP/JK184 (1 μM) for 24 h (Fig. 1E).
Identification of BML284/PPP/JK184 as potent adjutants of CAR T cells
To assess the functional relevance of the tumor cells and T cells induced by BML284/PPP/JK184 respectively, the antitumor activity of BML284/PPP/JK184-treated B7-H3 CAR T cells against cancer cells or BML284/PPP/JK184-treated cancer cells was tested in vitro. Therefore, we first evaluated the in vitro antitumor effects of BML284/PPP/JK184 on MDA-MB-231 and HCT116 cells at different concentrations. The half-maximal inhibitory concentrations (IC50) of JK184 were 78.68 nM (MDA-MB-231) and 95.56 nM (HCT116), BML284 were 293.3 nM (MDA-MB-231) and 237.9 nM (HCT116), 6014 were 29.3 nM (MDA-MB-231) and 111.9 nM (HCT116) (Additional file 1: Fig. S2A). To make the drug effects more apparent and facilitate comparison, we have tested the drugs at a concentration of 1 μM. Then, the apoptosis of MDA-MB-231 and HCT116 cells was analyzed by FACS. As shown in Fig. 2A, after treatment with 1 µM drugs (BML284/PPP/JK184) for 24 h, the percentage of early apoptosis plus late apoptosis cells was approximately 42.51%/54.32%/42.04% in MDA-MB-231 cells and 35.34%/66.83%/28.8% in HCT116 cells. The results indicated that BML284/PPP/JK184 could induce MDA-MB-231 and HCT116 cell apoptosis. We noticed that the expression of B7-H3 was not influenced with BML284/PPP/JK184 treatment (Additional file 1: Fig. S2B).
Next, to gain better insight into the effect of BML284/PPP/JK184 on T cells, we measured the T cell survival rate after treatment with BML284/PPP/JK184 at different concentrations. We observed that lower concentration of BML284/PPP/JK184 (1 µM) showed non-significant influence on T cell survival (Additional file 1: Fig. S2C). In addition to the analysis of phenotypic changes in activated T cells treatment with BML284/PPP/JK184 (1 µM), we also examined the expression levels of the canonical T cell activation and exhaustion markers such as CD69, CD25, PD-1, and LAG-3 using FACS. Flow cytometric analysis showed that BML284/PPP/JK184 treatment did not cause any significant change in the expression of the T cell markers CD25 and CD69 (Additional file 1: Fig. S2D). Interestingly, the expression of PD1 and LAG3 was declined after treatment with JK184 in B7-H3 CAR T cells and T cells (Additional file 1: Fig. S2E and S2F). Moreover, we also tested the phenotypic changes in naive T cells treated with BML284/PPP/JK184 (1 µM), and the results were similar to that of activated T cells (data not shown).
In the coculture assay, to reflect the synergistic killing effect of BML284/PPP/JK184, we set a low effector to target (E:T) ratio of 1:1 and observed a significant killing effect of B7-H3 CAR T cells against MDA-MB-231 and HCT116 cells when combined with BML284/PPP/JK184 (1 µM) after coincubation for 24 h. In addition, the secretion of perforin and granzyme B by T cells was also analyzed with FACS (Fig. 2B). B7-H3 CAR T cells with or without drugs mediated antitumor activity, yet the secretion level of perforin or granzyme B increased dramatically only in the BML284/PPP/JK184 combination group. The residual tumor cells were estimated with crystal violet staining, and the results are presented in Additional file 1: Fig. S2G. Moreover, we monitored the cytotoxicity of multiple treatment groups on MDA-MB-231 and HCT116 cells using the Real-Time Cellular Analysis (RTCA) system (Fig. 2C). Similarly, the antitumor effects of the BML284/PPP/JK184+B7-H3 CAR T cell combination treatment group were dramatically enhanced compared to those of the B7-H3 CAR T cell (p < 0.0001, p < 0.0001, p < 0.0001, respectively), BML284/PPP/JK184+CD19 CAR T cell (p < 0.0001, p < 0.0001, p < 0.0001, respectively) and CD19 CAR T cell (p < 0.0001, p < 0.0001, p < 0.0001, respectively) groups. To partially mimic the compactness of solid tumors, we established a three-dimensional (3D) spheroid model of HCT116 cells. Calcein/PI staining showed that BML284/PPP/JK184 induced HCT116 spheroid death (Fig. 2D). In Fig. 2E, B7-H3CAR T cells were primarily stained with CFSE, and HCT116-mCherry cells were employed. It was directly observed that more B7-H3 CAR T cells were deposited in the peripheral rim of tumor spheroid regions and jointly attacked the tumor spheroid in the combination treatment group. In summary, 3957/6014/JK184 combined with B7-H3 CAR T cells has a better antitumor effect than BML284/PPP/JK184 or B7-H3 CAR T cells alone.
The mechanism by which BML284/PPP/JK184 acts on MDA-MB-231 cells
To further explore the molecular mechanism involved in the effect of BML284/PPP/JK184 in MDA-MB-231 cells, RNA-sequencing assays were performed. As expected, the RNA sequencing of MDA-MB-231 cells treated with BML284/PPP/JK184 identified 494/419/493 highly upregulated and 1081/1008/1097 significantly downregulated genes. In Fig. 3 A, B and E, the RNA sequencing data were shown by heat maps, which displayed different gene expression profiles among samples after treatment with control or BML284/PPP/JK184. Meanwhile, we applied gene set enrichment analysis (GSEA) to identify the changed phenotype of MDA-MB-231 cells. GSEA demonstrated that both extracellular matrix-receptor (ECM-receptor) interaction pathway and gap junction pathway were significantly enriched in cells treated by BML284 or PPP (Fig. 3C, D). In agreement with above results, in JK184-treated cells, we found reduced expression of ECM-receptor interaction pathway genes and a lower ECM-receptor interaction output score compared to control cells (Fig. 3F). Astonishingly, based on the GSEA enrichment analysis, the MAPK and TGFβ pathways which play critical roles in cell development were found to be significantly enriched in MDA-MB 231 cells treated by JK184 (Fig. 3F). JK184 activated more signaling pathways after acting on cells than the first two drugs did (p = 0.7048, p = 0.3179, respectively). Therefore, how JK184 regulated the concomitant activation of these pathways which drive tumor cells to be sensitively recognized and killed by T cells is interesting.
As expected, the RNA sequencing of MDA-MB-231 cells treated with JK184 identified the enriched pathways, such as breast cancer, Hedgehog signaling, and apoptosis, which were indicated by using the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis (Additional file 1: Fig. S3A). The enrichment results were consistent with previous studies, and we found that some highly expressed genes of the Hedgehog signaling pathway were inhibited after JK184 treatment, such as SMO, PTCH1, GLI1, GLI2, GLI3, and SUFU (Additional file 1: Fig. S3B). The expression of these genes was also confirmed by real-time PCR (Additional file 1: Fig. S3C).
The mechanism by which JK184 acts on T cells
To the best of our knowledge, Hedgehog signaling pathway inhibitors have been developed for the treatment of cancer, but few studies have investigated the effects of Hedgehog inhibitors on immune cells or their potential to be combined with ICB-based immunotherapy. Interestingly, a recent study highlighted the relevance of Hedgehog signaling in immune cells, which described a novel immunometabolic function of Hedgehog signaling on macrophages [35]. Here, to detect transcriptomic changes in T cells under JK184 (1 µM) treatment, we performed mRNA profiling. Next, the RNA sequencing data were shown using a heat map (Fig. 4A), and a total of 761 genes were significantly changed, including 510 upregulated genes and 251 downregulated genes. From the KEGG analysis, the top 25 enrichment pathways are shown in Fig. 4B. Strangely, cytokine-cytokine receptor interactions and the MAPK signaling pathway were enriched, but the Hedgehog signaling pathway was not found. Meanwhile, we applied gene set enrichment analysis (GSEA) to identify the changed phenotype of T cells. A similar description of the gene sets, including upregulated MAPK signaling pathway-related genes, toll-like receptor cascade-related genes, and calcium signaling pathway genes, is provided in Fig. 4E. Interestingly, the analysis of differentially expressed genes showed that JUN, GZMB, IFNG, IL12RB2 and BCL6 were upregulated, whereas genes such as FOXP3 were downregulated (Fig. 4C), and the result was confirmed by real-time PCR in T cells treated with JK184 (Fig. 4D). Thus, JK184 induces tumor cell death through the inhibition of the Hedgehog signaling pathway and the concomitant activation of pathways. However, JK184 may increase the activity of T cells and influence critical components of the cytotoxic machinery in T cells by regulating the MAPK signaling pathway.
Enhancement by JK184 of the antitumor activity of B7-H3 CAR T cells in vivo
According to the preceding results, JK184 could improve the activity of T cells and promote antitumor effect of B7-H3 CAR T cells in vitro. Next, we tested the efficacy of the combinatorial therapy in vivo using xenograft NSG mouse models. Here, we adopted two xenograft tumor models. First, 2 × 106 MDA-MB-231-FFluc or HCT116-FFluc cells were subcutaneously injected into the right flank of mice. 7 days after tumor inoculation, mice were injected intravenously with 1 × 107 B7-H3 CAR T cells on days 7 and 14 respectively, and 10 consecutive days of JK184 treatment (Fig. 5A). Tumor burden was monitored by in vivo bioluminescence imaging (BLI) every 3 days, and tumor average radiance (p/s/cm2/sr) was calculated using Living Image. As shown in Figs. 5C, 4F, although B7-H3 CAR T cells treatment or JK184 treatment alone mediated significant regression of xenografts compared with the control, it was more effective, even completely regressing, in the combinatorial therapy. The overall survival of MDA-MB-231-FFluc or HCT116-FFluc tumor-bearing mice was significantly prolonged in the treatment groups, including B7-H3 CAR-T cells treatment, JK184 treatment and combinatorial therapy (Fig. 5B and G). Animal weights did not significantly change during the course of treatment (Additional file 1: Fig. S4A and S4B). Mouse-bearing tumors from MDA-MB-231-FFluc cells were sacrificed on day 20 after inoculation. To evaluate the presence and frequency of B7-H3 CAR T cells infiltrating the tumor, IHC of CD3 staining was performed. The results demonstrated that more CAR T cells had infiltrated into tumors in the combination group. Meanwhile, IHC of cleaved caspase3 staining predicated that the tumor cells underwent apoptosis in each group, and more apoptotic tumor cells were observed in the combination therapy group (Fig. 5H).
The antitumor activity of JK184 in combination with ICB
Generally, cell line xenograft models were used to target human tumor studies, but these immunodeficient models could not mimic the interplay of the endogenous immune system, and such xenograft models failed to recapitulate endogenous immune activation and detect severe toxicity in normal tissues. Here, we tested the antitumor activity of JK184 using 4T1 and MC38 mouse models. Mouse anti-PD1 or anti-CTLA4 monoclonal antibodies were used as combination reagents. The schemes of the 4T1 and MC38 mouse treatments are shown in Fig. 6A and D. Briefly, 1 × 105 4T1 or MC38 cells were subcutaneously injected into Balb/c or C57BL/6 mice. Then, tumor-bearing mice were divided into four groups: control (isotype-matched control antibody), ICB treatment, JK184 treatment, and ICB + JK184 treatment. Starting on day 7 and continuing through day 17, mice received JK184 treatment. On days 7 and 10, 100 µg anti-PD1 or anti-CTLA4 antibody was intraperitoneally injected into each mouse. Tumor volume was measured using a Vernier caliper every 3 days beginning on day 7. As shown in Fig. 6B, E and F, JK184 and JK184 combined with ICB mediated significant regression of 4T1 or MC38 tumors compared with the control, and the antitumor activity of combination therapy was superior to that of ICB or JK184 alone. In the 4T1 mouse model, although tumor cells displayed few responses to the anti-PD1 antibody treatment, it had made great regression of tumor burden in the anti-PD1 combination with JK184 therapy (p < 0.001) (Fig. 6B). Furthermore, the combination therapy significantly retarded the growth of the tumor and prolonged the survival of mice (Fig. 6C). In the MC38 mouse model, JK184, anti-PD1 or anti-CTLA4 antibody treatment inhibited tumor growth, but the combination treatment almost achieved complete tumor regression in some mice (Fig. 6E and F), and no mice died during the 60-day survival observation (Fig. 6G and H). The tumor volume calculation results and the corresponding solid tumor graph are shown in Fig. 6I, J, and the animal weights in the three models were measured (Additional file 1: Fig. S4C, S4D and S4E).
JK184 combined with ICB promoted T cell infiltration and reshaped the TME
To explore the effect of JK184 combined with PD1 in MC38 tumor-bearing mice, flow cytometry assays were used to analyze the tumor immune microenvironment. First, the number of tumor-infiltrating lymphocytes (TILs) was collected, and both PD1 and JK184 increased the density of TILs compared to control (p < 0.05). The combination treatment group had a significantly increased density of TILs compared to the control (p < 0.001) (Fig. 7A). In addition, the ratio of T cells in all viable cells was analyzed compared to the control, JK184 (p < 0.01), PD1 (p < 0.01), and JK184+PD1 (p < 0.0001) groups (Fig. 7B). Apart from the number of T cells, we further examined the cytotoxic activity of T cells and observed that JK184+PD1 simultaneously enhanced it. The number of CD8+T cells was elevated after JK184 (p < 0.01), PD1 (p < 0.01) or JK184+PD1 (p < 0.0001) treatment (Fig. 7C). Moreover, the densities of granzyme B+and IFN-γ+T cells were enormously elevated after JK184 + PD1 treatment relative to the control (p < 0.0001, p < 0.0001, respectively), and both were increased, which was statistically significant in the JK184 or PD1 treatment group (Fig. 7D and E).
On the other hand, we focused on the analysis of regulatory T (Treg) cells and myeloid-derived suppressor cells (MDSCs), which are two leading components of the immune-suppressive tumor microenvironment. CD4+FOXP3+Tregs are a subset of T cells with immunosuppressive properties. As shown in Figs. 7F and G, there was a significant reduction in the proportions of Tregs in the CD4+T cells and MDSCs in the CD45+cells in the treatment groups. The percentages of Tregs and MDSCs were significantly decreased compared with the control group: JK184 (p < 0.01 and p < 0.01, respectively), PD1 (p < 0.01 and p < 0.01, respectively), and JK184+PD1 (p < 0.0001 and p < 0.0001, respectively). Furthermore, JK184+PD1 regulated the polarization of macrophages. Although there was no significant difference in the proportions of the total macrophages between the treatment group and control group, the ratio of M1-like macrophages (M1) to M2-like macrophages (M2) increased compared with the control group in the JK184 (p < 0.01), PD1 (p < 0.01), and JK184+PD1 (p < 0.0001) groups (Additional file 1: Fig. S5). Our results demonstrated that JK184 could promote T cell infiltration, reshape the TME, and combine with PD1 to enhance the antitumor immune response.