Chimeric antigen receptors containing the OX40 signalling domain enhance the persistence of T cells even under repeated stimulation with multiple myeloma target cells.

Persistence of CAR-T cell function is associated with relapse rate after CAR-T therapy, while co-stimulatory agents are highly concerned with the persistence of CAR-T cells. In this study, we designed and constructed a series of BCMA-targeting second-generation CAR constructs containing CD28, 41BB, and OX40 molecules, respectively, to identify the costimulatory domains most favorable for persistence. The results of routine in vitro studies showed that OX40-CAR-T and 41BB-CAR-T had similar antitumor effects and were superior to CD28-CAR-T in terms of proliferation and cytotoxicity. Although difficult to distinguish by conventional functional assays, OX40-CAR-T cells exhibited greater proliferation and enhanced immune memory than 41BB-CAR-T cells with the repeated stimulation assay by BCMA-expressing target cells. In vivo studies further demonstrated that OX40-CAR-T cells had stronger proliferative activity than 41BB-CAR-T cells, which was highly consistent with the in vitro antitumor activity and proliferation results. Our study provides for the first time a scientific basis for designing OX40-CAR-T cell therapy to improve relapse in patients with MM after CAR-T treatment.

To the editor,

Multiple myeloma (MM) is the second most fatal hematologic malignancy, constituting 1–2% of neoplasms worldwide and being responsible for 2% of all cancer deaths [1]. Anti-BCMA-directed CAR-T cells treatment has achieved impressive response rate unfortunately most patients eventually relapse soon due to the poor persistence of CAR-T cells which closely related with different costimulatory molecules [2-5]; therefore, it is necessary to investigate the new costimulatory molecules to enhance this property of CAR-T cells.

Therefore, we constructed a serial of BCMA-targeted CARs containing 41BB, CD28, and OX40 co-stimulatory domain (Fig. 1A and Additional file 1: Fig. S1A), respectively, and investigated the effect on their duration of the antitumour properties. Firstly, the three groups of CAR-T cells showed comparable activation, differentiation, and apoptosis performance (Additional file 1: Fig. S1B–D and Additional file 2: Fig. S2A). However, the cytokine secretion experiment showed OX40-CAR-T cells were more prone to release the Th1 cytokines IFN-γ and TNF-α (Fig. 1B); while CD28-CAR-T cells tended to release the immunosuppressive cytokines IL-4 and IL-10 (Additional file 2: Fig. S2B and C). The exhausted markers of LAG-3, PD-1, Tim-3, and CTLA-4 inhibitory molecules of the cells were observed higher in the CD28-CAR-T cells than in the other two groups (Fig. 1C and Additional file 2: Fig S2D). With traditional experiment protocols [6], the OX40-CAR-T and 41BB-CAR-T cells showed equivalent proliferation and cytotoxicity profiles, but were significantly better than that of the CD28-CAR-T cells (Fig. 1D and E, P < 0.001 and P < 0.01, respectively). All the data indicated that the inducible co-stimulatory molecule of OX40 and 41BB might have better persistency than the primary co-stimulatory molecule of CD28 [7], whereas, the common evaluation methods could not distinguish the differences of the OX40 and 41BB of the BCMA-targeted CAR-T.

Fig. 1

Characterizations of antitumour efficacy among CD28, OX40 and 41BB based CAR-T cells. A The structure of the BCMA-CAR consists of the same variable region of the BCMA single-chain antibody, the CD8 hinge and transmembrane regions, different costimulatory molecules (from 41BB, CD28, or OX40) and CD3ζ. B Coincubation of effector cells with target 8226 cells for 24 h at a ratio of 5:1 between D10 and D15 (different days were used in different donors) and the supernatant was collected. Cytokines were detected by a human Th1/Th2/Th17 kit using flow cytometry. The qualitative analysis of the expression of IFN-γ, TNF-α was performed with Phyton 3.7 using the Matplotlib package (https://matplotlib.org/) (n = 3 donors). C The expression of the exhaustion-related markers LAG-3, TIM-3, PD-1, and CTLA-4 on T cells expressing BCMA-CAR were measured on day 7 (D7) (n = 3 donors). Data were analyzed with FlowJo software, and graphs were plotted with Phyton 3.7 using the Matplotlib package (https://matplotlib.org/). D The Cell Trace TM CFSE Cell Proliferation Kit was used to detect cell proliferation. On D13, effector T cells were stained with CFSE (CFDA-SE) dye and incubated with target K562 (negative control) and 8226 cells at a ratio of 5:1. After 5 days of incubation, CFSE fluorescence intensity was detected by flow cytometry (K562 data not shown). E Effector T cells were incubated for 24 h with target K562 (negative control) and 8226 cells at E:T ratios of 10:1, 5:1, 2.5:1, and 1:1 on D13. Cytotoxicity was determined from the amount of released LDH in the culture supernatants using an LDH kit at a wavelength of 490 nm. The figure shows the result of effector T cells incubated with the 8226 target cells (n = 3, P < 0.001 and P < 0.01, error bars denote standard deviation)

To investigate the durability discrepancies between OX40 and 41BB co-stimulatory molecules of BCMA-targeted CAR-T, we designed a novel in vitro approach that imitates the killing process of tumor cells in the body by repeatedly stimulating CAR-T cells with target cells to induce their exhaustion. Surprisingly, we found that T effect memory (Tem) cells of BCMA-targeted OX40-CAR-T accounted for 60.0% (around two times) than that seen in the BCMA-targeted 41BB-CAR-T group (Fig. 2A and Additional file 3: Fig S3A), which is consistent with previous reported data of CD30-targeted OX40-CAR-T cells showed long-term immune memory [8]. And the percentage of CAR+ cells remarkably increased in the BCMA-targeted OX40-CAR-T cells group, reaching almost 80% (Fig. 2B), which indicated that in the case of consecutive antigen exposure, BCMA-targeted OX40-CAR-T cells proliferate rapidly. Finally, the data from in vitro culture experiment demonstrated that introducing OX40 as a costimulatory molecule had a crucial role in maintaining a high number of CAR+ cells (Fig. 2C and Additional file 2: Fig S2E) and reduced the incidence of loss of CAR+ cells (Additional file 2: Fig S2F).

Fig. 2

OX40-CAR-T cells are the least exhausted and have superior persistence. A Effector cells were subjected to three consecutive repeated stimulations with 8226 target cells (CAR+ cells and 8226 target cells at a ratio of 1:1; 3-day interval was used for each stimulation), changes in CAR+ cell subtypes were determined with flow cytometry (n = 3). B CAR+ cells from different experimental groups after repeated stimulation. As described in A, FlowJo was used for data analysis and presentation (n = 3). C The average copy number of the CAR gene in single cells was detected by qPCR technology on day 7 (D7), D14, and D21 after the cells were activated with anti-CD3/CD28 antibodies (n = 3, P < 0.001 and P < 0.01, respectively. error bars denote standard deviation). D Schematic outline of the mouse model experiment (n = 5). All the Methods and Materials were described in the Additional file 4. E Tumour progression was monitored by IVIS imaging. In order to scientifically show the small gap between different CAR-T, the scales are normalized for PBS and NC group with 1 × 105 ~ 1 × 106, and 4-1BB, CD28 and OX40 CAR-T group with 1 × 104 ~ 1 × 105. F Tumour progression was monitored by total flux of each mice. G The proportion of CAR-T in WBC (white blood cell) were detected from mice PB by flow cytometry on day 7. ***P ≤ 0.001, NS no significant. H Representative GSEA of DNA repair pathways and MSigDB hallmark gene set for two different BCMA-targeted CAR-T cells (OX40-CAR-T cells and 41BB-CAR-T cells). I Representative GSEA of the oxidative phosphorylation pathway and MSigDB hallmark gene sets for different BCMA-targeted CAR-T cells

We observed the similar phenomenon in our in vivo study with the repeatedly stimulating BCMA CAR-T (Fig. 2D). The more lower fluorescence intensity (Fig. 2E, F) reflected the superior efficacy of BCMA CAR-T, more importantly, 4-1BB and OX40 CAR-T cells seems to have a more lasting anti-tumour effect, compared to two mice of CD28 CAR-T are about to relapse(Fig. 2E). BCMA-targeted OX40-CAR-T cells could proliferate rapidly in vivo (Fig. 2G). This is consistent with the results of in vitro study.

To further explore the mechanism that OX40-CAR-T cells displayed unique persistence advantages, transcriptomic analysis of each group of BCMA-targeted CAR-T cells were performed and found that the increased gene expression profiles of proteins known to strengthen DNA repair (which enabled improved BCMA-targeted OX40-CAR-T survival activity) (Fig. 2H) and metabolism (which enabled enhanced proliferation and immune memory) (Fig. 2I and Additional file 3: Fig S3B) in the OX40-CAR-T group. At the same time, the enrichment of the TNF-α and IFN-γ signalling pathways explains the effective killing power of the BCMA-targeted OX40-CAR-T cells (Additional file 3: Fig S3C). These findings have not been reported elsewhere before.

In conclusion, our study demonstrated for the first time that OX40-mediated BCMA-targeted CAR-T showed more durable antitumor activity than 41BB-mediated CAR-T cells upon repeated stimulation of BCMA-expressing target cells. Our findings not only provide a scientific basis for designing novel BCMA-targeted CAR-T cells for MM to gain more durable anti MM activities, but also provide valuable data for improving the anti-tumor persistence and reducing recurrence after CAR-T cell therapy.

Additional file 1: Figure S1. Evaluation of transduction efficiency, activation, differentiation abilities of BCMA-CAR-T cells.

Additional file 2: Figure S2. Comparison of apoptosis, cytokine inducing abilities and persistence of BCMA-CAR-T cells.

Additional file 3: Figure S3. Representative Flow cytometry analysis about CAR-T cells subtypes and Gene set enrichment analysis results.

Additional file 4. Detailed materials and methods as well as experimental procedures.