Perspectives on Chimeric Antigen Receptor T-Cell Immunotherapy for Solid Tumors

Unlike hematologic cancers where the infused CAR T-cells and tumor cells co-circulate in the blood, bone marrow, and lymphatics with ample opportunity for interaction, solid tumors represent discrete foci to which infused cells must migrate in order to contact the tumor cells and engage antigen. This involves chemotaxis in response to chemokines, such as CXCL-9, -10, and -11 (37). Since many human cancers display poor T-cell infiltration (“cold tumors”) (38), it is unsurprising that migration of CAR T-cells to the TME can be inefficient. Even the i.v. infusion of large numbers of CAR T-cells does not improve the clinical outcomes when tackling solid tumors, especially in patients with bulky disease; however, these trials were using the early first generation CARs (39, 40) (Figure 1). When clinically feasible, intra-tumoral injection has been demonstrated in vivo to circumvent poor trafficking of CAR T-cells, leading to tumor regression without systemic toxicity (41). In addition to the lack of T-cell chemokines being secreted by the TME, there are other chemokines, such as CXCL12, which actively inhibit T-cell migration into the tumor through engaging CXCR4 on their cell surface (42, 43). CXCL12 is highly expressed in variety of carcinomas, including pancreatic (44, 45), ovarian (46), and breast (47). However, pharmacologically blocking CXCR4 on the T-cell surface facilitated infiltration of T-cells into a spontaneous murine model of pancreatic ductal adenocarcinoma (43). As such, the CXCR4/CXCL12 axis may represent a therapeutic target to facilitate CAR T-cell infiltration in some solid tumors. Extravasation of T-cells into the tumor from the blood can also be inefficient due to the expression of molecules such as endothelin B receptor (ET_BR) on the endothelium of the blood vessels within the tumor. High ET_BR expression results in the nitric oxide-mediated deregulation of endothelial ICAM-1 expression, which reduces T-cell adhesion and compromises their ability to extravasate (48). Several attempts have been made to improve CAR T-cell trafficking to tumors (49–51). Kershaw and colleagues have demonstrated that expression of CXCR2 improved T-cell migration in a melanoma tumor which produced CXCL1, a chemokine commonly secreted by tumor cells (52). Similarly, CCR2b, the chemokine receptor for CCL2, has been co-expressed on CAR T-cells to exploit the CCR2/CCL2 axis, which facilitates myeloid cell recruitment into the tumor. CCR2b-expressing GD2-specific CAR T-cells had a 10-fold improvement in their migration to CCL2-producing neuroblastoma and demonstrated better in vivo antitumor activity in murine models (50). A similar observation was also achieved using mesothelin-directed CAR T-cells in murine models of mesothelioma (51). Others have improved CAR T-cell trafficking and the antitumor response in vivo using CAR T-cells that express a regulatory subunit I anchoring disrupter (RIAD) peptide that inhibits the protein kinase A-mediated suppression of the TCR, which can occur in the TME (53).

There are also physical barriers that restrict T-cell infiltration within solid tumors, where extracellular matrix proteins, such as proteoglycans and collagen, can restrict entry (54, 55). To overcome this, CAR T-cells have been engineered to express heparanase, which degrades heparin sulfate proteoglycans (56). This approach enhanced CAR T-cell infiltration within the tumor, leading to improved overall survival in xenograft tumor models.

Intra-tumoral T-cells are often functionally tolerant, or suppressed, displaying reduced effector functions compared to peripheral T-cell pools, including lowered cytokine, perforin, and granzyme-B expression (57, 58). This phenomenon has also been observed for CAR T-cells (59). Two receptors that have been of particular interest in inducing this anergic T-cell state are PD-1 and CTLA-4, which represent part of a family of regulatory receptor known as checkpoint molecules, which prevent inappropriate immune activation, but can be exploited by cancer. For example the ligands for PD-1, PDL-1 and PDL-2, can be found expressed on a variety of tumor and stromal cells (60–62). Antibody-mediated blockade of immune checkpoint receptors that are expressed on the T-cell surface have shown unprecedented clinical activity in the treatment of solid tumors by sustaining endogenous antitumor immune responses, most notably in melanoma (63). The checkpoint molecules offer significant scope as therapeutic targets for combination therapy, as these molecules behave in a hierarchical structure within each TME (43, 61). Indeed, combined PD-1 and CTLA-4 blockade (using nivolumab and ipilimumab, respectively) has improved response rates in patients with advanced melanoma, compared to that of the single therapies (64). Importantly, regulatory molecules such as PDL-1 are upregulated by effector molecules of T-cell activation, such as interferon (IFN)-γ (60, 65), suggesting that the greater the T-cell activity, the more suppressive the TME becomes. Anti-HER2 CAR T-cell therapy has an enhanced antitumor response in preclinical murine models when combined with PD-1 blockade (66). The combination of immune checkpoint blockade with CAR T-cell immunotherapy is now under study in clinical trials (67). Innovations to CAR T-cells to permit insensitivity to immune checkpoints are also under investigation at the preclinical stage, involving such approaches as genetic inactivation of the PD-1 gene (68), co-expressing dominant-negative variants of the inhibitory phosphatases, such as Src homology 2 phosphatase (SHP-2), which mediate the signaling of checkpoint receptors (69), or the expression of PD-1 receptors with no signaling moiety as decoy molecules (70). Others have attempted to positively exploit the interaction by fusing the PD-1 ectodomain to the CD28 cytoplasmic tail to induce a co-stimulatory, instead of inhibitory, signal upon PDL-1 binding (71). Brentjens et al. enhanced the antitumor efficacy of CD19-specific CAR T-cell in vivo by co-expressing the activating checkpoint molecule, CD40-ligand (72). CD40L co-expression resulted in an increased CAR T-cell proliferation, Th1 cytokine secretion, and increased cytotoxicity in vivo. The stroma, as well as the tumor cells, play a fundamental role in suppressing T-cells responses in the TME. Our group has shown that tumor-associated macrophages (TAMs) can inhibit immune-mediated tumor rejection in vivo through their expression of the heme-degrading enzyme heme oxygenase-1 (HO-1) (61, 73). Carbon monoxide is one of the by-products of heme catabolism by HO-1 and has been demonstrated to be capable of suppressing T-cell proliferation, IL-2 secretion (74), and T-cell effector function (61, 75, 76). The pharmacological inhibition of this enzyme within the TME of a murine model of breast cancer resulted in a rapid restoration of a chemotherapy-elicited antitumor CD8⁺ T-cell response and immunological control of tumor growth, leading us to propose that HO-1 should also be considered as an immune checkpoint molecule (61). As HO-1 is expressed in a variety of cancers (77), it may warrant therapeutic targeting in CAR T-cell therapy. Together, these considerations present a strong rationale for combining CAR T-cell and immune checkpoint blockade therapies. Appropriate combinations of these therapies may improve CAR T-cell efficacy in the TME and support more favorable clinical response rates.

The TME is a chronic inflammatory site and contains a variety of cytokines and chemokines, which influence the immune response. The TME is often skewed toward favoring pro-tumor Th2 cytokines, such as IL-4 and IL-13, rather than antitumor Th1 cytokines like IFN-γ and tumor necrosis factor-β (78). Cytotoxic T-cell function is depressed by cytokines, such as IL-4, IL-10, and transforming growth factor (TGF)-β, which are prevalent in the TME (79, 80). TAMs are an abundant stromal cell type in solid tumors and have been shown to secrete IL-10 as well as IL-6, which also suppresses T-cell cytolytic function and proliferation (81, 82). Similarly, regulatory T-cells (Tregs; CD4⁺ CD25⁺ FoxP3⁺) produce significant quantities of suppressive cytokines, including IL-10, TGF-β (83), and IL-35 (84). T-cells, which were engineered to express a dominant-negative form of the TGF-β receptor, have been demonstrated to have augmented antitumor responses in TGF-β producing tumors (85, 86). Alternatively, IL-4 signaling has been exploited to promote CAR T-cell expansion and activity. In one such approach, the IL-4 ectodomain has been fused to the common β chain of IL-2 and IL-15 receptors in order to achieve selective ex vivo amplification of CAR transduced T-cells. This approach renders it feasible to manufacture CAR T-cells from whole blood, circumventing the need for leukapheresis (87) and has the added benefit of augmenting CAR T-cell responses against tumors containing IL-4 in the TME (88, 89). Alternatively, the IL-4 receptor ectodomain was fused to the IL-7 receptor endodomain and co-expressed in prostate stem cell antigen-specific CAR T-cells in order to transduce a T-cell proliferation signal in response to IL-4 (90). The co-expression of this receptor led to improved in vivo CAR T-cell expansion and antitumor response. IL-12 is a potent inflammatory cytokine which enhances T-cell expansion and antitumor immune responses (91, 92). In murine preclinical tumor models, CAR T-cells which constitutively express IL-12 have been demonstrated to have enhanced proliferation, in vivo expansion, cytotoxicity, and antitumor efficacy (93–95). To circumvent off-target toxicities of this approach to expedite clinical translation, inducible IL-12 expression linked to CAR engagement has also been demonstrated to improve antitumor responses (96). CAR T-cells co-expressing IL-18 (97, 98), a constitutively active IL-7 cytokine receptor (99) or a tethered form of IL-15 (IL-15 peptide fused to IL-15Rα via flexible linker) (100) have also all augmented their antitumor response in vivo. It is clear that the cytokine environment within the TME is not conducive to permitting CD8⁺ T-cell activation and effector function. However, innovative ways to modulate and positively exploit the response of the CAR T-cells, such as through hybrid receptors, or IL-12 (Figure 1) show promising preclinical data and have started to progress through to clinical trials (NCT02498912↗ and NCT01818323↗).

Tumor cells are highly metabolically active with increased glycolysis and glutaminolysis (101). These metabolic pathways result in a distinct accumulation of metabolites in the TME which can compromise CAR T-cell function. Lactate is a metabolite derived from the glycolytic-pathway that is highly produced by tumor cells and directly suppresses proliferation, cytokine production, and effector function of human cytotoxic T lymphocytes (102). Prostaglandins, which are derived from prostaglandin E2 synthase and cyclooxygenase (COX)-1/2-mediated catabolism of arachidonic acid, can also suppress T-cell function (103). In keeping with this, aspirin (a COX inhibitor), has been demonstrated to potentiate immune checkpoint therapy in improving CD8⁺ T-cell responses (104) and may warrant investigation in combination with CAR T-cell immunotherapy.

A number of amino acid-degrading enzymes that are commonly expressed in the TME can also suppress T-cell function. These include indoleamine-2,3-dioxygenase (IDO) and tryptophan-2,3-dioxygenase (TDO) which degrade tryptophan, and arginase-1 and nitric oxide synthase (NOS) which degrade l-arginine. T-cells have been demonstrated to be particularly sensitive to the depletion of these amino acids, resulting in impaired proliferation and effector function (105–107) and increased T-cell apoptosis (108, 109). As well as the physical depletion of these amino acids, the catabolites of tryptophan degradation such as l-kynurenine and 3-hydroxyanthranilic acid have also been demonstrated to be immunosuppressive (110, 111). IDO inhibitors have been demonstrated to enhance CAR T-cell efficacy (111), and these are now being evaluated in clinical trials (112, 113). Arginase activity has also been demonstrated to inhibit proliferation and cytotoxity of GD2-specific CAR T-cells (114). The degradation of l-arginine by the NOS pathway also results in the generation of reactive nitrogen species (RNS). Myeloid-derived suppressor cells (Gr-1⁺ CD11b⁺ cells) and TAMs are potent sources of reactive oxygen species (ROS), which inhibit T-cell function (115). Both ROS and RNS are believed to induce T-cell tolerance by altering the flexibility of the TCR chains, which impairs the binding and responsiveness of CD8⁺ T-cells to peptide–MHC complexes (116). It would be interesting to consider how ROS/RNS modifications might also influence the antigen binding capabilities of the CAR. ROS/RNS may also inhibit T-cell infiltration into the tumor through inactivating CCL2 by nitration (117). Metabolites generated in the tumor, once regarded as by-products, are now accepted as crucial immune-modulatory molecules in their own right. As such, the metabolic pathways that facilitate immune-regulation may require therapeutic targeting alongside CAR T-cell therapy, such as IDO, which is currently under clinical investigation.