CAR-T therapy has achieved great success in the fight against hematopoietic malignancies, however, its application in the treatment of solid tumor continues to meet with major challenges. Firstly, it is very difficult to identify the ideal tumor specific targets, and owing to the heterogeneity of solid tumor, targeting one antigen is unlikely to eliminate all the tumor cells7,8. Secondly, unlike liquid tumors, extensive infiltration of CAR-T and TCR-T cells into solid tumors, especially parenchyma, is hampered by many physical barriers29. Thirdly, it is still unclear how to sustain the survival and function of adoptive transferred T cells inside the tumor microenvironment. Research has shown that tumor infiltrated lymphocytes (TILs) tend to be hypofunctional due to the hypoxic conditions and upregulation of immune inhibitory molecules, such as PD-1 and CTLA-4, as well as suppression from the myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs)30–32. Therefore, an ideal cell therapy product targeting solid tumors may overcome at least some, if not all, of these hurdles.
Great efforts have been made in recent years to battle these challenges. For example, the combination of CAR-T and the blockade of PD1/PD-L1 has been shown to improve the mesoCAR-T therapy in solid tumor11. Similarly, it has been reported that CAR-T engineered to secrete checkpoint inhibitor antibodies or CAR-T expressing chimeric PD1/CD28 which switches inhibiting signal to costimulatory signal could improve anti-tumor efficacy in solid tumor settings33–37. Interestingly, PD-L1 is abundantly expressed on many tumors under inflammatory conditions, serving as a tumor associated antigen. Indeed, in recent years, robust anti-solid tumor efficacy in xenograft models has been recorded by targeting PD-L1, with either PD-L1 CAR or dominant-negative form of PD125,26. Based on these studies, we aimed to explore whether PD-L1 can be efficiently targeted by the TCR-T-based platform to mitigate the PD-1/PD-L1 mediated inhibition of TIL and to lyse tumor cells, ultimately eliminating solid tumors.
For this purpose, we constructed a new type of chimeric TCR by combining the advantages of both CAR-T and TCR-T to target PD-L1. The anti-PDL1 scFv was derived from Socazolimab, an anti-PDL1 antibody discovered by Sorrento Therapeutics and in development by Sorrento Therapeutics’ Licensing Partner, Lee’s Pharmaceutical and are now under Phase 3 study in China. We chose TCRγδ as the prototype to build the synthetic T cell receptor for several reasons. Firstly, the engagement of TCRγδ does not rely on additional co-stimulating signals to induce a strong proliferative response as compared to TCRαβ38. Secondly, since the endogenous TCRα chain is disrupted by gene editing occurring at the TRAC locus, a synthetic TCR using α chain could possibly pair with the endogenous TCRβ chain to form non-functional receptors. Therefore, the use of TCRγδ for constructing synthetic receptors could avoid the possibility of mismatching with endogenous TCRβ chain.
Our study revealed that δ-TCRγδ with anti-PD-L1 scFv fused to the δ chain of TCRγδ heterodimer could efficiently target PD-L1-expressing tumor cells with a similar potency to anti-PD-L1 CAR in both in vitro and in vivo solid tumor models (Fig. 1 and Fig. 2A). We found that even in xenograft model, such as melanoma where CAR-T failed to function at all even with substantial T cell expansion, δ-TCRγδ-T cells could completely suppress the tumor growth (data not shown), suggesting that the novel synthetic TCR could target a wider range of solid tumors than conventional CAR.
Furthermore, we discovered that when anti-PD-L1 scFv was fused to the TCRγ of γδ heterodimer, the resultant receptor γ-TCRγδ demonstrated even stronger antitumor activities than δ-TCRγδ, and exhibited great expansion capacity (Fig. 3). γ-TCRγδ-T cells also demonstrated excellent persistency as evidenced by the re-challenge study (Fig. 4D and Fig. 4E). Interestingly, γ-TCRγδ and δ-TCRγδ exhibits distinct requirement for associating with CD3 subunits, which could at least partially explain the observed difference in tumor killing and T cell expansion capacity. For a typical TCR, the antigen recognition and binding always involve two heterodimeric chains, either αβ or γδ, as well as the associated CD3 subunits. In the case of δ-TCRγδ where anti-PD-L1 scFv was tagged onto the delta chain, the TCR-CD3 complex was readily detected on the cell surface by flow cytometry (Fig. 3B). In contrast, when scFv was fused to the TCRγ to form γ-TCRγδ, CD3 was not detected at the cell surface, implying that the fusion of scFv to different chains could have distinct effect on the structure or conformation of the entire receptor complex, especially regarding the association with CD3 subunits28. Given the robust antitumor activity of γ-TCRγδ (Fig. 3 and Fig. 4), the constant region of TCRδ appears to serve merely as an accessory chain within the synthetic TCRγδ heterodimer. Indeed, the notion was further supported by the observation that when delta chain was deleted from γ-TCRγδ, the resultant synthetic receptor, γ-TCRγ remained fully functional in terms of lysing tumor cells and inducing T cell expansion (Fig. 5). Nevertheless, the exact biological significance of delta chain in our synthetic anti-PD-L1 scFv GDT and the signaling transduction of γ-TCRγ warrant further exploration.
To solve this mystery, we speculated that the gene expression profiles in γ-TCRγ-T and δ-TCRγδ-T cells could be different. However, we did not identify much of a difference regarding the gene expression after antigen stimulation. These results suggest that the outcome of T cell activation is similar, regardless of whether it is activated in the form of TCRγδ heterdimer-CD3 or single TCRγ without CD3. However, we noticed that several sets of genes are upregulated in the γ-TCRγ-T group, such as CCR5, KLF2, S1PR1, which are related to the chemotaxis and memory formation, possibly accounting for the better access to the tumor and subsequent greater anti-tumor efficacy. Th/Tc17 CAR T cells have been shown with enhanced in vivo persistence and augmented tumor immunity39,40. Interestingly, we also noticed that cytokine IL-17A and IL-17F were upregulated in the γ-TCRγ-T group, providing another mechanism for better anti-tumor activity of γ-TCRγ-T cells.
It is worth noting that PD-L1 is extensively expressed in many types of tumors, as well as in normal cells, implying possible on-target off-tumor toxicity. Hence, to alleviate potential systemic toxicity and to improve the safety profile of the novel anti-PD-L1 scFv TCR platform, it is advisable to integrate a suicide gene into the construct.
In summary, we developed a new set of PD-L1-targeting, γδ-based TCR, which could function without associating with CD3. The novel synthetic receptors, γ-TCRγδ and γ-TCRγ, demonstrated superior antitumor activities both PD-L1 in vitro and in in vivo solid tumor models, such as NSCLC and melanoma, supporting the strategy of inhibiting the PD1/PD-L1 axis to treat solid tumors. Furthermore, we believe that the innovative single chain synthetic TCR, γ-TCRγ could represent a new platform of cancer immunotherapy that awaits thorough investigation in future studies.