Selective refueling of CAR T cells using ADA1 and CD26 boosts antitumor immunity

Summary

Chimeric antigen receptor (CAR) T cell therapy is hindered in solid tumor treatment due to the immunosuppressive tumor microenvironment and suboptimal T cell persistence. Current strategies do not address nutrient competition in the microenvironment. Hence, we present a metabolic refueling approach using inosine as an alternative fuel. CAR T cells were engineered to express membrane-bound CD26 and cytoplasmic adenosine deaminase 1 (ADA1), converting adenosine to inosine. Autocrine secretion of ADA1 upon CD3/CD26 stimulation activates CAR T cells, improving migration and resistance to transforming growth factor β1 suppression. Fusion of ADA1 with anti-CD3 scFv further boosts inosine production and minimizes tumor cell feeding. In mouse models of hepatocellular carcinoma and non-small cell lung cancer, metabolically refueled CAR T cells exhibit superior tumor reduction compared to unmodified CAR T cells. Overall, our study highlights the potential of selective inosine refueling to enhance CAR T therapy efficacy against solid tumors.

Graphical abstract

Highlights

• •ADA1 is conditionally secreted in response to CD3/CD26 stimulation
• •Co-overexpression of ADA1 and CD26 is essential for T cell trafficking and expansion
• •Ecto-ADA1 boosts CAR T cell expansion without impacting tumor cells
• •MR-CAR is a strategy to improve CAR T cell therapy for solid tumors

Hu et al. demonstrate that ADA1 is conditionally secreted in an autocrine manner in response to CD26 stimulation. The inclusion of a T cell engager and overexpressed CD26 boosts ADA1 capture in a membrane-proximal manner, providing inosine for T cells and minimizing feeding the tumor cells.

Introduction

Chimeric antigen receptor (CAR) T cell therapy is a promising approach that combines the specificity of monoclonal antibodies (mAbs) with the targeted biodistribution and long-term persistence of effector lymphocytes to selectively target and eliminate malignant cells. Early- and late-phase clinical trials have shown breakthrough successes in patients with hematologic malignancies treated with CAR T cells, leading to six US Food and Drug Administration (FDA)-approved CAR T cell therapies.^1,2,3,4,5,6 However, CAR T cell therapy has shown only modest results in patients with solid tumors. The limited efficacy is likely multicausal, including limited CAR T cell trafficking to solid tumors, tumor-mediated immunosuppression, and limited expansion and persistence of CAR T cells in the tumor microenvironment (TME).^7,8 Current strategies primarily use co-stimulatory molecules and, more recently, cytokines to optimize CAR T cell activation and survival.^9,10,11 However, nutritional competition between tumor cells and T cells poses a significant but underappreciated challenge to the efficacy of CAR T cells in solid tumors.^{12,13,14,15,16,17,18} Because cancer cells have high metabolic demands, the TME creates a metabolic stress environment that can affect the function of CAR T cells by limiting energy resources like glucose and producing immunosuppressive metabolites like adenosine. The majority of glucose is absorbed by tumor cells, leaving CAR T cells starved and unable to survive regardless of stimulations.

Our previous research has shown that inosine can act as a substitute carbon source to support T cell proliferation and activity in the absence of glucose.¹⁹ T cells can metabolize inosine to hypoxanthine and phosphorylated ribose through purine nucleoside phosphorylase.^19,20 The ribose component of inosine can enter into central metabolic pathways to provide energy in the form of ATP and support biosynthetic processes. In addition, inosine can modulate immune responses through the activation of A2AR and A3R in context-dependent manners. Notably, in the presence of interferon (IFN)-γ, inosine significantly boosted type 1 T helper (Th1) differentiation of naive T cells, whereas in the absence of IFN-γ, inosine inhibited this pathway.²¹ Given the ability of adoptive T cell transfer and immune checkpoint inhibitors to convert a suppressive TME into a supportive one, it is reasonable to speculate that inosine can increase the effectiveness of T cell therapy or immune checkpoint inhibitors in combating tumors. Indeed, inosine has been demonstrated to enhance tumor immunogenicity, making tumor cells more vulnerable to the cytolytic effects of immune cells.²² However, as inosine can also act as a fuel for tumor growth, targeted delivery of inosine to CAR T cells is necessary to optimize CAR T cell therapy.

Inosine is a nucleoside produced from adenosine via adenosine deaminase (ADA).²³ There are two isoenzymes of ADA present in humans, ADA1 and ADA2.^24,25,26 ADA1 is primarily a cytoplasmic protein that is ubiquitously expressed by most body cells,^23,25,27 but although ADA1 lacks a signal peptide, it has been shown to be secreted via a non-classic pathway, and ADA1 is found in human plasma.^26,28 ADA2 is the predominant isoform found in human plasma. ADA1 and ADA2 play different parts in regulating immune responses, independent of their ADA activity. ADA1 has co-stimulatory effects on T cell-mediated immunity by engaging T cells that express the ADA1 receptor CD26,^28,29,30 whereas ADA2, in a CD26-independent manner, binds to immune cells, including neutrophils, CD16⁺ monocytes, natural killer (NK) cells, B cells, and regulatory T cells (Tregs).^31,32 Both ADA1 and ADA2 are significantly increased in multiple human tumor tissues.²⁵ However, ADA1 expression positively correlates with the expression of genes associated with cell division and exhibits only a moderate positive correlation with genes associated with tumor-infiltrating lymphocytes (TILs). In contrast, ADA2 expression correlates with genes associated with immune responses in multiple types of infiltrating immune cells. These observations suggest that ADA1’s primary function is in adenosine metabolism, while ADA2 plays a more significant role in regulating immune responses. This functional disparity is aligned with the observation that ADA1 has a 100-fold higher affinity for adenosine compared to ADA2. Notably, high expression levels of ADA1 in tumor tissues have been associated with worse outcomes, while high ADA2 expression levels are linked to favorable results in a variety of cancers.²⁵ This raises questions about whether enhanced ADA1 activity in tumor tissues broadly provides an additional energy source for the growth of tumor cells. Therefore, even though CAR T cells overexpressing secreted ADA1 have shown the potential to significantly slow tumor growth in mouse models,³³ it may be crucial to prevent inosine from becoming an extra energy source for cancer cells in the treatment of human cancer.

The interaction of ADA1 with T cells through CD26 is an important step in the co-stimulation and activation of T cells. After T cells are activated, the multifunctional protein CD26 is increased on their plasma membranes, and its expression is closely controlled during T cell development.^29,30,34,35 Upon activation, CD26 recruits CARMA1 to its cytosolic domain, leading to NF-kB activation, T cell proliferation, and interleukin (IL)-2 production. CD26-mediated co-stimulation differs from CD28 co-stimulation and preferentially results in cytotoxicity through production of granzyme B, tumor necrosis factor-α, IFN-γ, and Fas ligand.³⁶ Additionally, CD26^high T cells have been shown to elicit tumor immunity against various cancers, with enhanced migration and persistence, chemokine receptor profile, cytotoxicity, resistance to apoptosis, and stemness.³⁷ However, transforming growth factor β1 (TGF-β1) can significantly downregulate CD26 expression and impair the cell activities, emphasizing the necessity to overexpress CD26 in CAR T cells to preserve their functionality.^38,39 In this study, we engineered CAR T cells with a helper vector to co-express CD26 and ADA1.CD3scFv, which facilitated the capture of ADA1 in a membrane-proximal manner. We found that this combination selectively enhanced the proliferation and cytokine production of CAR T cells without affecting tumor cells, resulting in increased cytolytic activity against cancer cells. Furthermore, co-expression of CD26 and ADA1.CD3scFv improved the persistence of CAR T cells in the TME and increased their migration toward the tumor, resulting in better tumor control. Our study highlights the potential of combining CD26 and ADA1 in CAR T cell engineering as a promising strategy to improve the efficacy of CAR T cell therapy.

Results

ADA1 conditional secretion in response to CD3/CD26 stimulation: Rational design of a helper vector optimizing ADA1 function in CAR T cells

ADA1 overexpression in CAR T cells has recently shown some positive effects in reducing tumor size in mice.^33,40,41 However, the high expression of ADA1 seen in human cancers and its link to poor survival outcomes raise questions about whether such treatments could also promote the growth of cancer cells in humans. Therefore, to harness the benefits of ADA1 in T cell therapy, it is essential to maximize its potential in boosting T cell function while avoiding inosine becoming an energy source for cancer cells. To selectively boost the immunobiology of ADA1 in CAR T cells, we designed a helper vector to metabolically refuel (MR) T cells. The MR vector drives the expression of a membrane-bound CD26 and an ADA1 protein fused to anti-CD3 scFv in the T cells to support T cell proliferation and activation (Figure 1A). Furthermore, we designed a series of constructs to overexpress secreted ADA1, cytoplasmic ADA1, or membrane-bound ADA1, and the impact of these constructs on modulating T cell function was compared (Figure 1A).

Figure 1.

Rational design and implementation of ADA1 in CAR T cells

(A) Four different versions of ADA1 were designed for use in CAR T cell therapy.

(B) Human PBMCs (n = 3) were stimulated with 100 μM ADO, 5 μg/mL coated caveolin-1 protein (CD26), 1 μg/mL coated OKT3 mAb (CD3), combined CD3/ADO, or combined CD3/CD26. After 48 h, ecto-ADA1 was measured by ADA activity assay.

(C) HER2-MR-CAR T cells (n = 3) were stimulated with 1 μg/mL coated OKT3 mAb (CD3), 10 μg/mL coated mAb 134-2C2 (CD26), or combined CD3/CD26. After 48 h, cell culture medium was subjected to ADA1 ELISA. p = 0.0352 for CD3/CD26 vs. CD3/CD28.

(D) HER2-MR-CAR T cells (n = 3), HER2-CAR T cells, or NT cells were cultured at indicated density for 24 h. After incubation, IFN-γ was measured using ELISA. The experiments were conducted in triplicate. ∗p value for HER2-MR-CAR T cells + A549 vs. HER2-CAR T cells + A549.

(E) mRNA sequencing (n = 3) was used to analyze inflammatory cytokines, granzyme A, and granzyme B.

(F) RT-qPCR was used to measure IFN-γ (n = 3).

(G) mA.26.HER2 (n = 3) or NT cells were cultured either with or without A549 tumor cells, while mA.26.GPC3 or NT cells were cultured either with or without Huh7 cells. 24 h later, IFN-γ were measured using ELISA. p = 0.0000008 and p = 0.000236 for mA.26.HER2 and mA.26.GPC3 vs. NT. p = 0.0000003 for mA.26.HER2 vs. NT (A549). p = 0.0000001 for mA.26.GPC3 vs. NT (Huh7).

Error bars represent SEM. p values were determined by two-tailed t test.

To induce the secretion of ADA1 from human T cells, the human IL-2 signal peptide was inserted into the N terminus of ADA1.⁴² The secretion of ADA1 was evaluated through ADA1 ELISA and ADA1 enzyme activity assay in the T cell culture medium. Unexpectedly, the expression of secreted ADA1 in the cell culture medium is not detectable at 24 or 48 h. This observation suggests that the mechanism of ADA1 secretion in human T cells may differ from that of conventional proteins and may not be activated by the IL-2 signal peptide. We then asked if cellular activation might induce the secretion of ADA1 in human T cells given that lymphocytes can release ADA1 in response to stimulation with anti-CD3 antibodies.²⁸ To address this possibility, human peripheral blood mononuclear cells (PBMCs) were activated with adenosine, CD26 ligand, CD3 mAb, CD3 mAb/CD26 ligand, or CD3 mAb/adenosine using serum-free cell culture medium to eliminate ADA activity in the serum. After 48 h of culture, PBMCs were subjected to an ADA activity assay to determine ADA1 secretion (Figure 1B). The results show that ADA1 is significantly secreted in response to co-stimulation of CD3 and CD26 but moderately in response to CD3 alone. Furthermore, human HER2-CAR T cells were restimulated using CD3/CD26 mAbs, in comparison to CD3/CD28 mAbs, which are commonly used for CAR T cell preparation (Figure 1C). The results confirm that human HER2-CAR T cells significantly secrete ADA1 in response to CD3/CD26 stimulation, while no enhancement is observed with CD3/CD28 stimulation compared to CD3 alone. This indicates that the secretion of ADA1 is triggered in an autocrine manner by co-stimulation of CD3 and CD26.

We then investigated whether ADA1.CD3scFv activates human CAR T cells exclusively in the context of tumor antigen engagement conditions and not under physiological conditions. To generate MR-CAR T cells, human PBMCs were co-transduced with two RD114-pseudotyped retroviral vectors. The first MR vector encoded CD26 and ADA1.CD3scFv, while the second CAR vector encoded either a GPC3- or HER2-specific CAR (Figure S1A). Cell surface expression on the MR-CAR T cells of the CARs and CD26 was detected by flow cytometry (Figures S1B–S1D). After transduction, both HER2 and GPC3 CARs are stably expressed on the surface of human peripheral blood T cells. The expression level of CD26 on both HER2 and GPC3 MR-CAR T cells increases significantly compared to that on HER2- or GPC3-specific CAR T cells, respectively. The transcriptional expression of ADA1.CD3scFv was demonstrated by western blotting (Figure S1E).

The next step was to determine whether MR-CAR T cells were activated upon engaging tumor cells. The HER2-CAR T cells were cultured in the presence or absence of HER2-positive non-small cell lung cancer (NSCLC) A549 tumor cells at a 1:1 ratio for 24 h. The culture media were analyzed by an IFN-γ ELISA. In the absence of tumor cells, ADA1.CD3scFv in HER2-CAR T cells does not induce IFN-γ expression, but the presence of the HER2-positive NSCLC 459 cell induces significantly increased amounts of IFN-γ production by the HER2-MR-CAR T cell (Figure 1D). The results are consistent with HER2-MR-CAR T cells showing increased expression of T-bet, granzyme B, IFN-γ, and IL-2 upon CD3 antibody stimulation (Figures S2A and S2B). To assess the activation of MR-CAR T cells under physiological conditions, we used RNA sequencing to analyze the gene profiles. The results indicate that MR-CAR T cells do not exhibit an enhanced inflammatory gene profile (Figure 1E). Further, we used RT-qPCR to assess the transcriptional expression of IFN-γ in HER2-specific or GPC3-specific MR-CAR T cells cultured under physiological conditions. In these cells, ADA1.CD3scFv does not stimulate the expression of IFN-γ automatically (Figure 1F). These findings suggest that engagement with tumor cells could trigger the release of ADA1.CD3scFv, which subsequently acts as a trans-signal to activate CAR T cells in a tumor-antigen-specific manner.

ADA1 activates CD26, which is expressed on various cell types, including T cells.^29,35,37,43 Consequently, we wondered if a membrane-bound version of ADA1 could result in autocrine activation of CAR T cells through interacting with CD26. HER2-CAR T cells or GPC3-CAR T cells co-expressing membrane-bound ADA1 and CD26 were cultured with or without HER2-positive A549 NSCLC or GPC3-positive GPC3 hepatocellular carcinoma (HCC) overnight. An ELISA reveals that membrane-bound ADA1 induces the expression of IFN-γ in both HER2-CAR T cells and GPC3-CAR T cells independent of tumor stimulation, with tumor stimulation further increasing IFN-γ expression (Figure 1G). This indicates that membrane-bound ADA1 induces autocrine activation of human T cells independent of tumor antigens. Taken together, these findings demonstrate that the ADA1.CD3scFv construct optimizes ADA1 in T cell-based therapies.

Overexpression of CD26 resists TGF-β1 suppression, and the co-overexpression of ADA1 and CD26 is critical for CAR T cell trafficking and expansion

The expression of CD26 is carefully regulated during T lymphocyte development. TGF-β1 is a component of the regulatory system and inhibits CD26 expression.^38,39 We therefore examined how constitutive CD26 expression induced by the MR helper vector might change CD26 expression when TGF-β1 was present. Using flow cytometry, CD26 cell surface expression was evaluated. Human T cells that had undergone transduction express CD26 at a level that is noticeably greater than non-transduced (NT) human T cells (Figure 2A). After 48 h of in vitro incubation in the presence of 20 ng/mL TGF-β1, CD26 expression on NT T cells is downregulated by 40%, whereas MR-vector-transduced T cells defy TGF-β1-mediated suppression of CD26 expression.

Figure 2.

Overexpression of CD26 resisted TGF-β1 suppression and promoted CAR T cell mobility and proliferation

(A) Rv-CD26-transduced T cells were cultured in the presence of TGF-β1 for 48 h, and CD26 expression was detected by flow cytometry. The experiments were conducted in duplicate.

(B and C) CCR2 and CCR5 expression was determined by flow cytometry. The experiments were conducted in duplicate.

(D) The heatmap shows the expression levels of chemokine receptor genes. Triplicate samples (n = 3) were used for each group and are represented on the x axis.

(E and F) Rv-CD26-transduced T cells (n = 3) were subjected to a fluorescent migration assay and Transwell migration assay.

(G) PBMCs (N = 2) were transduced with retroviral vectors expressing HER2-CAR (HER2-CAR), ADA1.CD3scFv and HER2-CAR (ADA1.CD3scFv-HER2-CAR), CD26 and HER2-CAR (CD26-HER2-CAR), or ADA1.CD3scFv/CD26 and HER2-CAR (HER2-MRCAR) and labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE). The T cells were then cultured for 4 days and subjected to flow analysis.

(H) GPC3-MR-CAR T cells and GPC3-CAR T cells were expanded in vitro, and the cell numbers were determined at different time points using a hemocytometer. The results are presented as a growth curve. The figure indicates the p values for GPC3-MR-CAR vs. GPC3-CAR. The experiments were conducted in triplicate.

Error bars represent SEM. p values were determined by two-tailed t test.

We then questioned if MR-CAR T cells would express increased amounts of chemokine receptors given that human CD26^high T cells strongly induce immune responses against numerous forms of malignancy through enhanced migration.³⁷ Using flow cytometry, the expression of CCR2 and CCR5 on T cell surfaces was examined. We find that HER2-MR-CAR T cells, when compared to HER2-CAR T cells overexpressing either ADA1.CD3scFv or CD26 alone, have considerably higher levels of CCR2 and CCR5 expression (Figures 2B and 2C). This demonstrates that on HER2-CAR T cells, the co-overexpression of CD26 and ADA1.CD3scFv increases the production of CCR2 and CCR5. Additionally, mRNA sequencing analysis proves that HER2-MR-CAR T cells have greater chemokine receptor expression profiles (Figure 2D). Using a fluorescent migration test and a Transwell migration assay, the mobility of these T cells was examined (Figures 2E and 2F). At 2, 4, and 22 h after cell culture, T cells transduced with the MR retroviral vector show improved migration. Additionally, flow analysis shows that, compared to HER2-CAR T cells overexpressing either ADA1.CD3scFv or CD26 alone, HER2-MR-CAR T cells exhibit enhanced proliferation (Figure 2G) and alleviated exhaustion (Figures S3A–S3D). Moreover, at day 23 of in vitro growth, the number of GPC3-specific MR-CAR T cells is two times higher than that of conventional GPC3-CAR T cells (Figure 2H). In conclusion, CD26 expression increases T cell mobility and proliferation while resisting TGF-β1 inhibition, demonstrating benefits of co-overexpressing CD26 and ADA1.CD3scFv in CAR T cells.

ADA1.CD3scFv enhances CAR T cell expansion preferentially without impacting tumor cells

CD26 on T cells functions as an extracellular ADA1 receptor, and its expression is elevated following T cell activation. However, ADA1 has a relatively low affinity for CD26 (K_d = 1.8⁻⁸ M)⁴⁴; thus, we conducted assessments to determine the efficacy of ADA1-CD26 binding, specifically by measuring the presence of ADA1 on the surface of CD26-positive Jurkat T cells. The results indicate that the addition of extra ADA1 does not result in increased binding to the CD26-positive Jurkat T cells, suggesting an ineffective binding of ADA1 to CD26 (Figure S4A).

We then created a CD26-positive NFAT or NF-κB luciferase reporter human Jurkat T cell line using a retroviral vector generating human CD26 in order to ascertain if the CD3scFv is required for anchoring ADA1 to CAR T cells. After being transduced with the retroviral vector expressing ADA1.CD3scFv or ADA1, human Jurkat T cells with NFAT or NF-κB luciferase reporter are subsequently grown at a high cell density to promote the secretion of ADA1.CD3scFv or ADA1. The amount of ADA1 attached to the Jurkat T cell membrane was assessed using an ADA enzyme activity assay after the Jurkat T cells had been in culture for 24 h. According to the findings, ADA1 has only approximately 40% of the activity on CD26-positive Jurkat T cells compared to ADA1.CD3scFv (Figures 3A and 3B). However, the ADA1.CD3scFv’s activity on Jurkat T cells that lacked CD26 is comparable to that of the ADA1.CD3scFv on CD26-positive Jurkat T cells. The results align with flow analysis of ADA1 on the surface of Jurkat T cells (Figure S4B) and the ADA activity assay conducted on HER2-specific MR-CAR T cells (Figure S4C). These findings demonstrate the critical role of the CD3scFv in strengthening the interaction between ADA1.CD3scFv and human T cells.

Figure 3.

ADA1.CD3scFv enhances CAR T cell expansion preferentially without impacting tumor cells

(A and B) Luciferase reporter Jurkat-Dual or Jurkat-NFAT cells either expressing CD26 or lacking CD26 were transduced with overexpressing vectors of ADA1 or ADA1.CD3scFv and cultured for 24 h. After incubation, the cells (n = 3) were subjected to an ADA activity assay. CD26-positive Jurkat T cells transduced with ADA1.CD3scFv overexpressing vector were used as the maximum value to calculate the percentage. The data are presented as the percentage of ADA1 or ADA1.CD3scFv binding with the Jurkat T cells. The figure indicates the p values for the binding of ADA1 with CD26⁺ Jurkat T cells vs. the binding of ADA1.CD3scFv with CD26⁺ Jurkat T cells.

(C) Jurkat T cells (n = 3) were transduced with retroviral vector expressing ADA1, ADA1.CD3scFv, CD26, ADA1, and CD26 or ADA1.CD3scFv and CD26, respectively. Jurkat T cells were labeled with CFSE, and cell proliferation was measured by flow analysis. The experiments were conducted in duplicate.

(D) The conditioned medium collected from the culture of HER2-MR-CAR T cells, HER2-CAR T cells, or NT human T cells was added to A549 cell cultures (n = 3) and incubated for 24 or 48 h. The numbers of tumor cells were quantified daily using a hemocytometer. p = 0.000002 for ADA1 vs. ADA1.CD3scFv at both 24 and 48 h.

(E and F) HER2-MR-CAR T cells or HER2-CAR T cells were co-cultured with A549 tumor cells in a Transwell plate for 72 h. Similarly, GPC3-MR-CAR T cells or GPC3-CAR T cells were co-cultured with Huh7 tumor cells. The CAR T cells and tumor cells were counted using a hemocytometer and are presented.

(G) 293T cells were transduced with retroviral vectors expressing either ADA1 or ADA1.CD3scFv. After 48 h of culture, the cell culture medium was collected and added to the culture of CD26-negative or CD26-positive Jurkat-NFAT T cells for 24 h. The Jurkat-NFAT cells were then subjected to luciferase activity assay. p = 0.0034 for ADA1 vs. ADA1.CD3scFv in CD26-negative Jurkat T cells. p = 0.0000003 for ADA1.CD3scFv in CD26-positive Jurkat T cells vs. CD26-negative Jurkat T cells. The experiments were conducted in triplicate.

Error bars represent SEM. p values were determined by two-tailed t test.

We next investigated the impact of ADA1.CD3scFv on T cell survival and proliferation. Our results demonstrate that human T cells transduced with Rv-ADA1.CD3scFv/CD26 exhibit enhanced proliferation compared to those transduced with Rv-ADA1/CD26, highlighting the importance of CD3scFv in promoting T cell proliferation (Figure 3C). Given the possibility that tumor cells could utilize inosine as an energy source, we also looked at whether ADA1.CD3scFv production by CAR T cells could be advantageous to tumor cells. However, we find that the culture medium from Rv-ADA1.CD3scFv-transduced T cells does not promote tumor growth, while the culture medium from Rv-ADA1-transduced T cells significantly promotes tumor growth (Figure 3D). To further examine the effect on tumor cells, we conducted a Transwell cell culture experiment using HER2-MR-CAR T cells or HER2-CAR T cells with A549, as well as GPC3-MR-CAR T cells or GPC3-CAR T cells with Huh7 cells (Figures 3E and 3F). The MR-CAR T cells proliferate more quickly than CAR T cells in both tests. Tumor cell proliferation does not differ between co-culture settings. This indicates that ADA1.CD3scFv selectively increases CAR T cell growth without affecting tumor cells. This is most likely because the ADA1 is mostly found in the cytoplasm, and once secreted, it binds to CAR T cells again and converts adenosine to inosine in the immediate vicinity of the CAR T cells. Moreover, only ADA1.CD3scFv significantly induces the activation of CD26-positive Jurkat T cells, while ADA1.CD3scFv moderately activates the CD26-negative Jurkat T cells (Figure 3G). ADA1 alone fails to activate either CD26-negative or CD26-positive Jurkat T cells. This indicates that CD3scFv is essential for the activation of CD26 signaling. Taken together, these results suggest that CD3scFv functions as a T cell engager, facilitating the anchoring of secreted ADA1 to promote T cell proliferation and activation without feeding surrounding tumor cells.

MR-CAR T cells display enhanced antitumor cytotoxicity in vitro

To compare the effectiveness of HER2-MR-CAR T cells and HER2-CAR T cells to kill tumor cells, we quantified the release of lactate dehydrogenase (LDH) in the media at various effector-to-target ratios. As a negative control, NT T cells were used as effectors. At ratios of 1:1, 5:1, 10:1, or 20:1, HER2-MR-CAR T cells or HER2-CAR T cells were co-cultured with HER2^high Calu3 (NSCLC) (Figures 4A–4C) and HER2^low A549 (NSCLC), respectively (Figures 4D–4F). After 4 h of culture at a ratio of 10:1 or above and after 18 h of culture at a ratio of 1:1 or above, the HER2-CAR T cells eliminate the HER2-positive Calu3 and A549, demonstrating the potency of the HER2-specific CAR. We find no difference in the capacity of HER2-CAR T cells to kill Calu3 tumor cells that are HER2^high compared to A549 tumor cells that are HER2^low. This shows that both HER2^high and HER2^low tumors can be killed by HER2-CAR T cells. The HER2-MR-CAR T cells demonstrate improved cytotoxicity against both Calu3 and A549 after 4 h of culture at a ratio of 10:1 or higher and after 18 h of culture at a ratio of 5:1 or higher. Additionally, HER2-MR-CAR T cells or HER2-CAR T cells were co-cultured with fresh A549 tumor cells at a 1:1 ratio, repeated for three cycles, each lasting 3 days. The results suggest that HER2-MR-CAR T cells demonstrate heightened proliferation and a reduced apoptosis rate (Figures S5A–S5C). This suggests that the expression of CD26 and ADA1.CD3scFv enhances the ability of HER2-CAR T cells to kill targeted tumor cells.

Figure 4.

MR-CAR T cells displayed enhanced antitumor cytotoxicity in vitro

(A–F) Human PBMCs were transduced with either HER2-MR-CAR or HER2-CAR and expanded in vitro. The cytotoxic activity of HER2-MR-CAR and HER2-CAR T cells against Calu3 and A549 cells was evaluated using LDH assay. The figure indicates the p values for HER2-MR-CAR vs. HER2-CAR.

(G–J) The cytotoxic activity of GPC3-MR-CAR and GPC3-CAR T cells against HepG2 and Huh7 cells was evaluated using LDH assay. The figure indicates the p values for GPC3-MR-CAR vs. GPC3-CAR. The experiments were conducted in triplicate.

Error bars represent SEM. p values were determined by two-tailed t test.

We next investigated GPC3-MR-CAR T cells’ cytotoxicity against GPC3-positive HCC Huh7 and HepG2. GPC3-MR-CAR T cells or GPC3-CAR T cells were co-cultured with Huh7 (Figures 4G and 4H) or HepG2 (Figures 4I and 4J) at ratios of 1:1 or 20:1. The GPC3-CAR T cells kill both Huh7 and HepG2, which are GPC3 positive, after 4 h of culture at a ratio of 20:1 and after 18 h of culture at a ratio of 1:1, indicating the effectiveness of the GPC3-specific CAR. The GPC3-MR-CAR T cells display enhanced cytotoxicity against both Huh7 and HepG2 after 4 h of culture at a ratio of 20:1 and after 18 h of culture at a ratio of 1:1. To explore the effects of PD-L1 inhibition on GPC3-MR-CAR T cell proliferation and effector functions, we assessed the expression of PD-1 on the surface of both GPC3-MR-CAR T cells and GPC3-CAR T cells (Figure S6A). Interestingly, both cell types exhibit increased PD-1 expression. However, while PD-L1 significantly suppresses the proliferation of GPC3-CAR T cells, GPC3-MR-CAR T cells show significant rescue in their proliferation (Figure S6B). Additionally, PD-L1 does not inhibit the expression of T-bet, granzyme B, IL-2, IFN-γ, and TNF-α in GPC3-CAR T cells, whereas GPC3-MR-CAR T cells exhibit enhanced expression levels of these effector molecules (Figures S6C–S6J). These results suggest that the expression of CD26 and ADA1.CD3scFv enhances the ability of GPC3-CAR T cells to kill targeted tumor cells and confers resistance to PD-L1 inhibition.

We next asked whether the enhanced cytotoxicity of HER2-MR-CAR T cells could result in T cell exhaustion. To this end, we conducted mRNA profiling and find that HER2-specific MR-CAR T cells exhibit a lower level of T cell exhaustion markers compared to HER2-CAR T cells (Figure S7). Furthermore, mRNA gene expression analysis reveals that HER2-specific MR-CAR T cells displayed increased NOTCH signaling and enhanced DNA repair mechanisms and complement activation compared to HER2-CAR T cells (Figures S8A and S8B; Table S1). Notch signaling is known to regulate cytokine and chemokine expression and cell migration and promotes cell survival through antiapoptotic protein expression (Bcl-2, Bcl-xL, Mcl-1) and activation of the PI3K/Akt and MAPK signaling pathways involved in cell survival and proliferation.^{45,46,47,48,49} The intracellular complement system is involved in metabolic refueling necessary for effector responses.⁵⁰ In addition, HER2-MR-CAR T cells have a reduced inflammatory response (i.e., TGF-β1 and IFN-α) compared to HER2-CAR T cells, as well as reduced levels of pro-inflammatory cytokines such as IL-6, IFN-γ, TNF-α, and IL-2. This may be beneficial in preventing unwanted immune responses and promoting a more controlled immune response against tumor cells. HER2-CAR T cells show an increased expression of genes related to glycolysis, reactive oxygen species, oxidative phosphorylation, ultraviolet response, hypoxia, and apoptosis. This suggests that these cells may experience cellular stress and could be at risk of damage or death due to the potentially harmful effects of these processes. HER2-CAR T cells also show an upregulation of genes related to MTORC1, P53, and KRAS, which may suggest a potential for T cell exhaustion. To assess whether other immune cells will interfere with MR-CAR effects, HER2-MR-CAR T cells or HER2-CAR T cells were co-cultured with primary PBMCs at a ratio of 1:1. After 48 h, flow analysis was conducted to assess exhaustion markers on CAR T cells. Unexpectedly, the results reveal that PBMCs reduce the exhaustion of CAR T cells (Figures S9A and S9B) and promote CAR T cell expansion (Figure S9C), likely due to accessory stimulation. In conclusion, HER2-MR-CAR T cells exhibit reduced exhaustion and display advanced cell division phenotypes.

MR-CAR T cells have enhanced antitumor activities in multiple xenograft mouse models

We hypothesized that HER2-specific or GPC3-specific MR-CAR T cells may inhibit HER2-positive or GPC3-positive tumor growth more effectively than control CAR T cells in xenograft animal models based on the above-described in vitro results. Using a subcutaneous A549 tumor model, we first looked into the anticancer effects of HER2-specific CAR T cells. 2 × 10⁶ A549 cells were injected subcutaneously into the right flank of NSG mice to create the tumors. The mice received 2 × 10⁶ HER2-MR-CAR T cells, 2 × 10⁶ HER2- CAR T cells, or phosphate-buffered saline (PBS), respectively, once the tumor size reached an average of 4–6 mm in diameter. Compared to control animals, HER2-CAR T cells slightly reduce the development of the A549 tumor. In contrast, mice that received HER2-MR-CAR T cells demonstrate a considerably greater inhibition of A549 tumor growth compared to mice that received HER2-CAR T cells or PBS (Figure S10A). The effectiveness of HER2-MR-CAR T cells in treating HER2^high human NSCLC Calu3 xenograft mice model was next assessed. Tumors treated with HER2-MR-CAR T cells show a trend toward reduced size when compared to unmodified HER2-CAR T cells after a single dosage of 2 × 10⁶ CAR T cells (Figures S10B and S10C). These findings indicate that HER2-MR-CAR T cells boost antitumor effectiveness without causing any obvious adverse effects.

Then, by increasing the dose, we investigated the anticancer effects of HER2-MR-CAR T cells. 5 × 10⁶ HER2-MR-CAR T cells, 10 × 10⁶ HER2-MR-CAR T cells, or two doses of 2 × 10⁶ HER2-MR-CAR T cells were given to groups of mice in subcutaneous A549 animal models. As a control, HER2-CAR T cells or PBS was used. The findings demonstrate that HER2-MR-CAR T cells are superior to HER2-CAR T cells at inhibiting tumor growth at the same dose (Figures 5A–5C). Additionally, compared to 5 × 10⁶ HER2-MR-CAR T cells, 10 × 10⁶ HER2-MR-CAR T cells or two doses of 2 × 10⁶ HER2-MR-CAR T cells show an improved reduction of tumor growth. In mice receiving either therapy, there was no indication that their body weight had changed. These findings imply that HER2-MR-CAR T cells’ anticancer activities can be further enhanced by increasing the dose or undergoing more treatments without manifesting any obvious negative side effects.

Figure 5.

MR-CAR T cells demonstrated antitumor activity in xenograft mouse models

(A–C) A549 tumor-bearing mice (n = 5) were treated with a single dose of either 5 × 10⁶ or 1 × 10⁷ or two doses of 2 × 10⁶ (at 1 week intervals) HER2-MR-CAR T cells, HER2-CAR T cells, or PBS. Tumor size and mouse body weight were monitored every 2–3 days. Data represent mean ± SD (n = 5).

(D–F) At week 0, mice (n = 10) were intercostally injected with 2 × 10⁶ A549-luc tumor cells. One week after tumor implantation, the mice were administered either 2 × 10⁶ T cells or PBS through the tail vein. Tumor development was monitored weekly using bioluminescence in vivo imaging (D). Mean photon count with SDs of mice groups is shown at the indicated time points (E). p = 0.04513, 0.01437, or 0.002137 for MR-CAR vs. ADA1.CD3scFv-HER2-CAR at weeks 9, 10, or 11 individually. Mouse survival was monitored (F). p = 0.0177 for HER2-MR-CAR vs. HER2-CAR in survival.

(G and H) A murine HCC xenograft model was established in NSG mice (n = 5) by subcutaneous inoculation of 2 × 10⁶ Huh7 tumor cells on the right flank. When the average tumor size reached 4–6 mm in diameter, experimental mice were treated with a single dose of 2 × 10⁶ GPC3-MR-CAR T cells, 2 × 10⁶ GPC3-CAR T cells, or PBS. Tumor size and mouse body weight were monitored every 2–3 days. Data represent mean ± SD (n = 5). p = 0.0209 and p = 0.00191 for GPC3-MR-CAR T cells vs. GPC3-CAR T cells on days 9 and 13 respectively. No difference in body weight was observed among groups.

Error bars represent SEM. p values were determined by two-tailed t test.

The antitumor effects of HER2-MR-CAR T cells were subsequently investigated in an orthotopic model. NSG mice were intercostally injected with A549-luciferase tumor cells, followed by treatment with either HER2-MR-CAR T cells or control T cells. In vivo tumor imaging results demonstrate that HER2-MR-CAR T cells more effectively inhibit tumor growth (Figures 5D and 5E) and prolong mouse survival (Figure 5F) compared to control HER2-CAR T cells. We also sought to replicate the improved antitumor effects of MR-CAR T cell with GPC3-specific CARs in a mouse model of GPC3-positive HCC. To establish the tumors, 2 × 10⁶ Huh7 cells were inoculated subcutaneously into the right flank of NSG mice. Once the tumor size reached an average of 4–6 mm in diameter, the mice were treated with either 2 × 10⁶ GPC3-MR-CAR T cells, 2 × 10⁶ GPC3-CAR T cells, or PBS delivered via the tail veil. In comparison to mice treated with GPC3-CAR T cells or PBS, tumor progression in the GPC3-MR-CAR T cell-treated mice is considerably reduced (Figure 5G). The mice receiving either form of CAR T cell therapy showed no indication that their body weight has changed (Figure 5H). These findings imply that GPC3-specific MR-CAR T cells improve antitumor effectiveness without causing any significant side effects. In conclusion, our research indicates that in preclinical mouse models, MR-CAR increases the effectiveness of CAR T cell therapy.

MR-CAR T cells retain their capability to proliferate, migrate, and lyse tumor cells in the TME

Subsequently, we examined the T cell responses within the TME. A549 tumor-bearing mice received 10 × 10⁶ of HER2-MR-CAR T cells, HER2-CAR T cells, or NT T cells. At specific times after T cell administration, tumor tissues were dissected, and single-cell suspensions were prepared. The concentration of inosine in the single-cell washout was determined using an inosine assay (Figure 6A). The results show that the concentration of inosine significantly increases in the HER2-MR-CAR group compared to that in HER2-CAR or NT group, while there is no difference in inosine concentration between the HER2-CAR and NT group. This indicates that ADA1.CD3scFv is expressed in the TME and effectively converts adenosine to inosine in the TME.

Figure 6.

MR-CAR T cells retained their capability to proliferate, migrate, and lyse tumor cells in the tumor microenvironment (TME)

NSG mice were subcutaneously inoculated with 2 × 10⁶ A549 tumor cells on the right flank. When the average tumor size reached 200–300 mm³, experimental mice (n = 10) were treated with a single dose of 1 × 10⁷ HER2-MR-CAR T cells, HER2-CAR T cells, or NT T cells.

(A) Seven days after treatment, tumor tissues were dissected, and a single-cell suspension was subjected to an ADA activity assay to measure inosine concentrations. ∗p = 0.00847 for HER2-MR-CAR vs. HER2-CAR.

(B) After 7 days of treatment, tumor-infiltrating CD3⁺ cells were sorted using flow cytometry. The number of sorted cells was quantified and is presented (n = 3). ∗p = 0.000177 for HER2-MR-CAR vs. HER2-CAR.

(C and D) Single-cell suspensions from tumor tissues were stained and analyzed by flow cytometry.

(E and F) Sorted CD3⁺ cells were co-cultured with A549 tumor cells at an effector-to-target (E:T) ratio of 1:1 overnight. IFN-γ was determined by ELISA. To determine the tumor-killing capacity, an LDH assay was performed.

(G) Sorted CD3⁺ cells were co-cultured with A549 tumor cells in a Transwell culture plate to assess the migration capacity of the CD3⁺ cells. ∗p = 0.00053 for HER2-MR-CAR vs. HER2-CAR.

(H) Sorted CD3⁺ cells were labeled with CFSE and then co-cultured with A549 tumor cells at an E:T ratio of 1:1 for 48 h, followed by flow analysis to determine their proliferation.

Error bars represent SEM. p values were determined by two-tailed t test.

Mice treated with the HER2-MR-CAR T cells have a higher number of tumor-infiltrating T cells compared to those receiving HER2-CAR T cells or NT T cells (Figure 6B). A higher number of CD8⁺ TILs is observed in the tumor tissue of mice receiving HER2-MR-CAR T cells (Figures 6C and 6D). Furthermore, the treatment with HER2-MR-CAR T cells results in an increased number of CD3⁺ TILs expressing CCR5, IFN-γ, granzyme B, and perforin compared to HER2-CAR T cell treatment. In contrast, HER2-CAR T cells show no difference compared to NT T cells, suggesting a significant immunosuppression of their immunological activity in the TME. As indicated by the increased number of tumor-infiltrating T cells and the expression of key activation markers, these results suggest that the HER2-MR-CAR T cells are more effective at trafficking to and lysing the tumor cells compared to the HER2-CAR T cells.

Finally, we looked into whether the tumor-infiltrating CAR T cells still had an enhanced capacity to migrate, proliferate, and lyse tumor cells. HER2-positive A549 tumor cells were co-cultured for 24 h with the sorted HER2-MR-CAR T cells, HER2-CAR T cells, or NT T cells at a ratio of 5:1. The culture media were then collected for quantitating IFN-γ and LDH as indicators of the T cell’s activation and cytotoxicity, respectively (Figures 6E and 6F). The results show that HER2-MR-CAR T cells exhibit heightened activation and cytotoxicity against A549 cells compared to HER2-CAR T cells and NT T cells. Additionally, the sorted HER2-MR-CAR T cells show significantly enhanced migration capability compared to HER2-CAR T cells and NT T cells when assessed using a Transwell migration assay (Figure 6G). Moreover, the proliferation capability of HER2-MR-CAR T cells is significantly stronger compared to that of HER2-CAR T cells and NT T cells as assessed by staining with carboxyfluorescein diacetate succinimidyl ester and analyzing their proliferation using flow analysis (Figure 6H). In conclusion, these findings suggest that HER2-MR-CAR T cells retain their capability to proliferate, migrate, and lyse tumor cells in the TME.

Discussion

We here report on a CAR T cell therapy that metabolically refuels T cells with inosine using a helper vector that encodes CD26 and ADA1 fused to anti-CD3 scFv. Our MR-CAR T cell therapy has the potential to address the challenges of current CAR T cell therapy in treating solid tumors by overcoming the immunosuppressive mechanisms present in the TME. To achieve this, our MR-CAR T cell strategy implements three innovative mechanisms: (1) ADA1 mediates the conversion of adenosine to inosine, which overcomes adenosine-mediated immunosuppression of CAR T cells; (2) inosine selectively promotes CAR T cell proliferation in the nutrient-deprived TME without feeding tumor cells; and (3) the overexpression of CD26 provides optimal co-stimulatory signals to CAR T cells, improving their mobility and enhancing antitumor effects.

First, adenosine signaling is an important immuno-metabolic checkpoint in tumors.^23,51 Ecto-nucleotidases, such as CD39, CD73, CD38, CD203a, ALP, and PAP, can generate adenosine from ATP or NAD+ that accumulate in the TME.^{20,52,53,54,55,56} Adenosine enables tumor cells to evade immune surveillance by suppressing the function of various protective immune cells, such as T cells, dendritic cells, NK cells, macrophages, and neutrophils, while promoting the activity of immunosuppressive cells, such as myeloid-derived suppressor cells and Tregs.^{32,54,57,58,59,60,61,62,63,64} Adenosine also activates cancer-associated fibroblasts and induces the formation of new blood vessels.^60,65 Many drugs have been discovered, including small molecules or monoclonal antibodies targeting CD73 and CD39 to limit adenosine production or A2AR and A2BR to inhibit adenosine binding to immune cells.^51,58 Their antitumor effectiveness has been shown in preclinical investigations, both by themselves and in conjunction with other immunotherapies, such as immune checkpoint inhibitors and adoptive cell transfer. However, early-phase clinical studies have failed to demonstrate adequate antitumor effects. Existing adenosine-targeted therapies only shut down one particular adenosine pathway while leaving the others unaffected, whereas multiple ecto-nucleotidases contribute to extracellular adenosine production, and adenosine binds to multiple receptors, such as A2AR, A2BR, and A3R, to suppress antitumor immunity. ADA1 offers an alternate strategy to current adenosine-target-specific medicines for combating adenosine-mediated immunosuppression by converting it to inosine.

Second, we have shown in earlier studies that inosine can serve as an alternative energy source to support T cell proliferation and function in the absence of glucose.¹⁹ Cancer cells with high glycolytic activity may deplete glucose, leading to a shortage of glucose in T cells. To enhance T cell-based therapies, supplementing T cells with inosine has been proposed. Nevertheless, tumor cells also have the ability to utilize inosine. Hence, it is imperative to develop a strategy to specifically deliver inosine to CAR T cells while avoiding tumor cells. In this regard, we overexpressed ADA1 in the cytoplasm of CAR T cells, which increased the generation of intracellular inosine, thereby boosting the growth of CAR T cells. Furthermore, upon conditional release in the tumor environment, the anti-CD3 scFv will engage ADA1 and the CAR T cells. The ecto-ADA1 on CAR T cells will convert adenosine in the T cell surroundings to inosine, thus overcoming adenosine-mediated immunosuppression and increasing inosine concentration around CAR T cells. Moreover, the CD3scFv will bind to endogenous CD3⁺ T cells, resulting in bystander effects.

We explored different options for using ADA1 in CAR T cell therapy, including overexpressing cytoplasmic ADA1,⁴⁰ secreted ADA1,³³ and membrane-bound ADA1.⁴¹ However, we find that overexpressing secreted ADA1 is not feasible since its secretion is not inducible, even with the addition of an IL-2 signal peptide. Qu et al. reported only a marginal increase in ADA1 secretion upon CAR T cell induction in response to the IL-2 signal peptide, which could be due to the stress condition present in their cell culture system.³³ Overexpressing the wild-type cytoplasmic ADA1 in CAR T cells could also be problematic since its conditional secretion could provide inosine as an energy source to tumor cells. While ADA1 is a ubiquitously expressed cytoplasmic protein, it can be secreted in response to CD3/CD26 stimulation, and the secretion is mediated by a non-classical pathway that does not require a signal peptide. This is similar to a subset of proteins, including IL-1α, IL1-β, IL-18, IL-33, IL-36α, IL-37, and IL-38, which lack a signal peptide and do not follow the classical endoplasmic reticulum-to-Golgi pathway of secretion.^66,67 Instead, these proteins are likely to transport out of the cell via exocytosis of preterminal endocytic vesicles during inflammatory or stress conditions. Indeed, elevated ADA1 expression in human tumors has been detected and linked to poor survival outcome.²⁵ Furthermore, our data demonstrate that permanent membrane-bound ADA1 can induce autocrine activation in a tumor-antigen-independent manner, likely due to CD26 engagement. Hence, anchoring ADA1 to T cells through CD3scFv and CD26 is the most efficient way to use inosine in CAR T cells. Cytoplasmic ADA1 converts intracellular adenosine to inosine and, upon secretion, produces inosine in a CAR T cell membrane-proximal manner, which avoids feeding tumor cells.

Third, our strategy overexpresses CD26 on the surface of CAR T cells, as the CD26 expression level on the unmodified CAR T cells is relatively low and can be further downregulated by TGF-β1, a cytokine found in significant quantities in various tumor tissues.^{68,69,70,71,72,73,74} Thus, we develop a strategy to overexpress CD26 on the surface of CAR T cells, which sustains co-stimulation and membrane-proximal ADA1 capture even in TGF-β1-rich environment and provides a survival advantage for these engineered therapeutic cells.

In preclinical NSCLC models, HER2-specific MR-CAR T cells exhibit enhanced antitumor activity. HER2 is overexpressed in certain types of cancers, including NSCLC, breast, gastric, and ovarian cancers. Although two HER2-targeted antibody-drug conjugates (ADCs), DS-8201 and T-DM1, have been approved by the FDA,^75,76,77 they have serious toxicities, such as grade 3+ adverse events, treatment discontinuation, interstitial lung disease or pneumonitis, and even death. Additionally, HER2-ADCs have shown limited efficacy in some patients with advanced HER2-positive cancers, likely due to a lack of effective internalization and intracellular delivery of the drug payload, as well as the restricted ability of these ADCs to cross the intact blood-brain barrier. Resistance to HER2-ADCs can develop over time, as the cancer cells evolve to evade the effects of the drug payload. This can limit the long-term effectiveness of HER2-ADCs and result in disease progression. Thus, an HER2-specific MR-CAR T cell strategy may offer a safer and more effective alternative for patients with HER2-positive cancers.^78,79,80

In summary, this study explores the use of a novel approach involving T cell-anchoring ADA1 and overexpressed CD26 in CAR T cell therapy. The significance lies in addressing the critical issue of low glucose levels within solid tumors, which is often overlooked by current strategies. Our strategy aims to provide inosine as an alternative carbon source to T cells, given the inadequate availability of glucose. This innovative approach holds immense promise for overcoming the challenges associated with solid tumor CAR T cell therapy. Importantly, recent research by Klysz et al. underscores the transformative impact of ADA overexpression in CAR T cells.⁴¹ ADA’s conversion of adenosine into inosine not only triggers stemness induction but also significantly enhances CAR T cell functionality. Furthermore, exposure to inosine not only bolsters CAR T cell function but also promotes stemness characteristics. Mechanistically, inosine induces profound metabolic reprogramming, including reduced glycolysis, increased mitochondrial and glycolytic capacity, heightened glutaminolysis and polyamine synthesis, and a shift in the epigenome toward enhanced stemness. These findings underscore the pivotal role of inosine as a potent modulator of CAR T cell metabolism and epigenetic stemness programming, offering a transformative platform for enhancing CAR T cell therapy efficacy, particularly in the context of solid tumors where glucose availability is limited. Additionally, our strategy holds promise for a diverse array of T cell-based therapies under investigation, spanning autologous CAR T cell therapy, allogeneic CAR T cell therapy, TIL therapy, and in vivo CAR therapies.

Limitations of the study

The limitations of our study include the examination of the potential adverse effects of adenosine on tumors. While we have demonstrated in an in vitro cell culture system that secreted ADA1.CD3scFv is exclusively localized on the surface of CAR T cells, with no detectable presence in the cell culture medium and no apparent effects on promoting tumor cell growth, it is important to note that in vitro cell culture may not fully replicate the complex interactions observed in vivo within the TME, where T cells and tumor cells may have closer cell-to-cell contact. Therefore, further investigation is needed to determine whether the secreted ADA1.CD3scFv could potentially bind bystander T cells or tumor cells in the surrounding environment. Additionally, since CD26 is also expressed on tumor cells, the disparity in expression levels between tumor cells and tumor-infiltrating T cells, as well as the ratio of these cells, may influence the effects of ADA1 on immune cells or tumor cells. Thus, it is imperative to explore the expression levels of CD26 on tumor cells compared to CAR T cells in future studies.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Alexa Fluor 647 Goat Anti-Mouse IgG, F(ab')₂ fragment specific	Jackson ImmunoResearch	Cat# 115-605-006; RRID: AB_2338903
Brilliant Violet 785 anti-human CD3 Antibody	BioLegend	Cat# 317329; RRID: AB_ 2563507
Pacific Blue anti-human CD8 Antibody	BioLegend	Cat# 344717; RRID: AB_10551438
FITC anti-human CD195 (CCR5) Antibody	BioLegend	Cat# 313705; RRID: AB_345305
APC anti-human/mouse Granzyme B Recombinant Antibody	BioLegend	Cat# 372203; RRID: AB_2687028
Brilliant Violet 711™ anti-human Perforin Antibody	BioLegend	Cat# 308129; RRID: AB_2687189
Brilliant Violet 785™ anti-T-bet Antibody	BioLegend	Cat# 644835; RRID: AB_2721566
Brilliant Violet 570™ anti-human IFN-γ Antibody	BioLegend	Cat# 502534; RRID: AB_2563880
PE anti-human/mouse Granzyme B Recombinant Antibody	BioLegend	Cat# 372208; RRID: AB_2687032
PE anti-human CD3 Antibody	BioLegend	Cat# 300407; RRID: AB_314061
APC anti-human CD279 (PD-1) Antibody	BioLegend	Cat# 329908; RRID: AB_940475
Pacific Blue anti-human CD223 (LAG-3) Antibody	BioLegend	Cat# 369341 RRID: AB_2910415
BUV395 Mouse Anti-Human CCR2 Antibody	BD Biosciences	Cat# 747854; RRID: AB_ 2872316
FITC Mouse Anti-Human CD26 Antibody	BD Biosciences	Cat# 555436; RRID: AB_395829
Alexa Fluor 647 Mouse Anti-Human PD-1 Antibody	BD Biosciences	Cat# 566851; RRID: AB_2869905
BV711 Mouse Anti-Human IL-2 Antibody	BD Biosciences	Cat# 563946; RRID: AB_2738501
PE-CF594 Mouse Anti-Human IFN-γ Antibody	BD Biosciences	Cat# 562392; RRID: AB_11153859
PE Mouse Anti-Human TIM-3 (CD366)	BD Biosciences	Cat# 565570; RRID: AB_2716866
Alexa Fluor 700 anti-human TIGIT Antibody	R&D Systems	Cat# FAB7898N
PE adenosine deaminase Antibody	Santa Cruz	Cat# sc-28346; RRID: AB_626634
Anti-human CD3 Antibody	Miltenyi Biotec	Cat# 130-093-387; RRID: AB_1036144
Anti-human CD28 Antibody	BD Biosciences	Cat# 567117; RRID: AB_2916451
Anti-human ADA antibody	Sigma-Aldrich	Cat# HPA001399; RRID: AB_1078099
Chemicals, peptides, and recombinant proteins
Recombinant Human ErbB2/Her2 Fc His Alexa Fluor 647 Protein	Biotechne	Cat# AFR1129
Recombinant Human IL-2	PeproTech	Cat# 200-02
Halt Protease and Phosphatase Inhibitor Cocktail (100X)	ThermoFisher	Cat# 78840
NuPAGE 4 to 12%, Bis-Tris, 1.0–1.5 mm, Mini Protein Gels	ThermoFisher	Cat# NP0322
Luna Universal qPCR Master Mix	NEB	Cat# M3003
Critical commercial assays
Human IFN-gamma Quantikine ELISA Kit	R&D Systems	Cat# DIF50C
Human ADA ELISA Kit	Invitrogen	Cat# EH9RB
LDH Cytotoxicity WST Assay	ENZO life science	Cat# ENZ-KIT157
Adenosine Deaminase Assay Kit	GenWay Biotech	Cat# GWB-BQK080
Inosine Assay	Cell Biolabs	Cat# MET5092
Cell Migration/Chemotaxis Assay Kit (24-well, 5 μm)	Abcam	Cat# ab235696
CFSE Cell Division Tracker Kit	BioLegend	Cat# 423801
ApoScreen Annexin V Apoptosis Kit	SouthernBiotech	Cat# 10010-09
iScript™ Advanced cDNA Synthesis Kit, 100 × 20 μL rxns	Biorad	Cat# 1725038
RNeasy Plus Mini Kit	Qiagen	Cat# 74134
Deposited data
GPC3 CAR Expressed in T cell (replicate 1)	This paper	GEO: GSM8169900
GPC3 CAR Expressed in T cell (replicate 2)	This paper	GEO: GSM8169901
GPC3 CAR Expressed in T cell (replicate 3)	This paper	GEO: GSM8169902
Metabolic refueling modified GPC3 CAR expressed in T cell (replicate 1)	This paper	GEO: GSM8169903
Metabolic refueling modified GPC3 CAR expressed in T cell (replicate 2)	This paper	GEO: GSM8169904
Metabolic refueling modified GPC3 CAR expressed in T cell (replicate 3)	This paper	GEO: GSM8169905
HER2 CAR Expressed in T cell (replicate 1)	This paper	GEO: GSM8169906
HER2 CAR Expressed in T cell (replicate 2)	This paper	GEO: GSM8169907
HER2 CAR Expressed in T cell (replicate 3)	This paper	GEO: GSM8169908
Metabolic refueling modified HER2 CAR expressed in T cell (replicate 1)	This paper	GEO: GSM8169909
Metabolic refueling modified HER2 CAR expressed in T cell (replicate 2)	This paper	GEO: GSM8169910
Metabolic refueling modified HER2 CAR expressed in T cell (replicate 3)	This paper	GEO: GSM8169911
Non-transduced T cell (replicate 1)	This paper	GEO: GSM8169912
Non-transduced T cell (replicate 2)	This paper	GEO: GSM8169913
Non-transduced T cell (replicate 3)	This paper	GEO: GSM8169914
GPC3 CAR Expressed in T cell (replicate 1)	This paper	GEO: GSM8169900
GPC3 CAR Expressed in T cell (replicate 2)	This paper	GEO: GSM8169901
Experimental models: Cell lines
293T cells	ATCC	Cat# CRL-3216
A549 cells	ATCC	Cat# CCL-185
Calu3 cells	ATCC	Cat# HTB-55
HepG2 cells	ATCC	Cat# HB-8065
Huh-7 cells	Gift from Dr. Andras Heczey	N/A
Jurkat-Lucia NFAT cells	Invitrogen	Cat# jktl-nfat
Jurkat-Dual cells	Invitrogen	Cat# jktd-isnf
Human Peripheral Blood Mononuclear cells	STEMCELL	Cat# 70025
Experimental models: Organisms/strains
NOD.Cg-PrkdcSCID Il2rgtm1Wjl/SzJ (NSG) mice	The Jackson Laboratory	RRID: IMSR_JAX:005557
Oligonucleotides
IFN gamma forward primer	This paper	GAGTGTGGAGACCATCAAGGA
IFN gamma reverse primer	This paper	TGTATTGCTTTGCGTTGGAC
GAPDH forward primer	This paper	GTCTCCTCTGACTTCAACAGCG
GAPDH reverse primer	This paper	ACCACCCTGTTGCTGTAGCCAA
Recombinant DNA
GPC3-CAR	Gift from Dr. Andras Heczey	N/A
HER2-CAR	This paper	N/A
cytoplasmic ADA1	This paper	N/A
secreted ADA1	This paper	N/A
permanent integral membrane bound ADA1	This paper	N/A
ADA1.CD3scFv	This paper	N/A
CD26	This paper	N/A
ADA1.CD3scFv/CD26	This paper	N/A
Software and algorithms
FlowJo 10.0	FlowJo, LLC	https://www.flowjo.com/
GSEA	Broad Institute	https://www.gsea-msigdb.org/gsea/index.jsp
BioRender	BioRender	https://www.biorender.com/
Prism 10	GraphPad	https://www.graphpad.com

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact Xiaotong Song (xsong@tamu.edu).

Materials availability

All unique reagents generated in this study are listed in the key resources table and available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability

• •RNA sequencing data have been deposited at NCBI GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Cell lines and cell culture

The cell lines 293T, A549, Calu3, and HepG2 were obtained from the American Type Culture Collection. The cell line Huh-7 was a kind gift from Dr. Andras Heczey (Texas Children’s Hospital, Baylor College of Medicine, Houston, TX). These cell lines were cultured in DMEM (Corning) supplemented with 10% fetal bovine serum (FBS, Gibco). Jurkat-Lucia NFAT Cells (Invivogen, jktl-nfat) and Jurkat-Dual Cells (Invivogen, jktd-isnf) were used as reporter cell lines and maintained as per the instructions provided.

Mice and in vivo studies

6–8-week-old NOD.Cg-PrkdcSCID Il2rgtm1Wjl/SzJ (NSG) mice were purchased from The Jackson Laboratory and maintained at the PAR Facility of Texas A&M Institute of Biosciences and Technology and all animal experiments were approved by the Institutional Animal Care and Use Committee of Texas A&M University College of Medicine, Institute of Biosciences and Technology. Tumor cells or CAR T cells were diluted in 100 μL normal saline and were injected via indicated routes.

Subcutaneous tumor model: Briefly, 6- to 10-week-old NSG mice were injected subcutaneously with different tumor cell lines respectively to generate tumor xenograft models. Tumor volume was determined by caliper measurement (LxW²/2). Once tumors reached 4–6 mm in diameter, indicated number of CAR T cells and MR-CAR T cells were injected intravenously (i.v.) via tail vein. Mice were assessed daily, and tumor sizes were measured three times a week. For ex vivo analysis of tumor infiltrating lymphocytes, CAR T cells were injected when tumors reached an average size of 200–300 mm³. At the indicated time points, tumors were dissected, and tumor infiltrating lymphocytes were analyzed.

Orthoptic lung cancer model: At week 0, mice were intercostally injected with 2x10e6 A549-luc tumor cells. One week after tumor implantation (week 1), the mice were administered either 2x10e6 T cells or PBS through the tail vein. Tumor development was monitored weekly using bioluminescence in vivo imaging. In vivo imaging analyses were conducted at indicated time points, and the mouse survival was also monitored.

Method details

Generation of retroviral constructs

The Moloney murine leukemia virus derived SFG retroviral vector backbone, which has been clinically validated, was generously provided by Dr. Andras Heczey of Baylor College of Medicine.^40,41 The retroviral vector was used to encode four distinct ADA1 constructs, including cytoplasmic ADA1, secreted ADA1, permanent integral membrane bound ADA1, or ADA1 fusion protein with anti-human CD3 scFv (ADA1.CD3scFv). The genes were synthesized by Genscript and subcloned into SFG gamma-retroviral vectors. To express integral membrane-bound ADA1, a human CD28 transmembrane domain was added to the C-terminal of ADA1. To express ADA1.CD3scFv, the ADA1 gene was fused with an anti-human CD3 scFv and was separated by a (G4S)3 linker. The anti-human CD3 scFv was obtained from the OKT3 clone.

To generate ADA1.CD3scFv/CD26 vector (MR vector), codon-optimized genes encoding CD26, T2A, and ADA1.CD3scFv were synthesized by Genscript and subcloned into SFG gamma-retroviral vectors. To generate HER2-CAR, codon-optimized genes encoding anti-human HER2 scFv derived from the FRP5 clone, CD28 transmembrane domain, CD28 endodomain, and CD3ζ were synthesized by Genscript and subcloned into SFG gamma-retroviral vector. This HER2-specific CAR is identical to the second-generation HER2 CAR, which has been elevated in several clinical studies.^42,43,44 The GPC3-CAR retroviral vector encoding anti-human GPC3 scFv derived from the GC33 clone, CD28 transmembrane domain, 4-1BB endodomain, and CD3ζ was also generously provided by Dr. Andras Heczey from Baylor College of Medicine.^40,41

Retrovirus production and transduction of primary T cells

Retroviral supernatants were produced by transient transfection of HEK 293T cells with retroviral vector-containing plasmids of GPC3-CAR construct, HER2-CAR construct, or MR vector. RDF plasmid encoding the RD114 envelope and PegPam3 plasmid encoding the MoMLV gag-pol were used to generate retroviral supernatants as previously described.^40,41 To generate CAR T cells, human peripheral blood mononuclear cells purchased from STEMCELL (70025) were stimulated by OKT-3 (Miltenyi Biotec, 130-093-387) and CD28 (BD Pharmingen, 567117) mAb-coated plates for 48 h in RPMI-1640 (Corning) with 10% FBS (Gibco) and 100 U/ml IL-2 (PeproTech, 200-02). After 48h of stimulation, cells were transduced on Retronectin (Takara, T100A)-coated and retroviral particle-loaded plates and after 48h cells were removed, washed, and cultured in IL-2 containing complete RPMI1640 media for further expansion.

Flow cytometry

Anti-F(ab)2 Alexa Fluor 647-conjugated antibody (Jackson ImmunoResearch, 115-605-006) was used to detect GPC3-CAR expression. Alexa flour 647 conjugated HER2 protein (Biotechne, AFR1129) was used to detect HER2 expression. The antibodies used for T cell phenotyping and cytokine production analyses are: CD3-BV785 (BioLegend, 317329), CD8-Pacific Blue (BioLegend, 344717), CCR2-BUV395 (BD, 747854), CCR5-FITC (BioLegend, 313705), CD26-FITC (BD, 555436), ADA1-PE (Santa Cruz, sc-28346), PD-1-Alexa flour 647 (BD, 566851), PD-1-APC (BioLegend, 329908), LAG-3-Pacific Blue (BioLegend, 369341), TIM-3-PE(BD Biosciences, 565570), TIGIT-AF647(R&D Systems, FAB7898N), IL2-BV711 (BD #563946), IFN gama-PE-CF594 (BD, 562392), GZMB-APC (BioLegend, 372203), Perforin-BV711 (BioLegend, 08129), T-bet-BV785 (BioLegend, 644835), IFN-γ-BV570 (BioLegend, 502534) and GZMB-PE (BioLegend, 372208). Intracellular cytokine staining was performed using a Transcription Factor Buffer Set (BD Biosciences, 562574). Flow cytometry assessment was performed on ZE5 Cell Analyzer (BIO-RAD). Results were analyzed with FlowJo software (TreeStar).

Western blot

Cells were washed with PBS and lysed using cell lysis buffer containing Protease and Phosphatase Inhibitor Cocktail (ThermoFisher, 78440). 30 μg of quantified proteins were separated on Bis-Tris pre-cast SDS-PAGE mini-gel (ThermoFisher, NP0322) and transferred to polyvinylidene fluoride membrane using a dry blotting system (ThermoFisher). After blocking the membranes were incubated overnight with desired ADA1 primary antibody (Sigma Aldrich, HPA001399) and then incubate 1 h with secondary antibody. Protein bands were visualized using a ChemiDoc XRS imager system (Bio-Rad).

qPCR

HER2-CAR, HER2-MR-CAR, GPC3-CAR, and GPC3-MR-CAR was cultured, and mRNA was gathered using RNeasy Plus mini kit (Qiagen, 74134) according to the manufacturer’s instruction for qPCR. Briefly, 1000ng of RNA was used as a template for cDNA synthesis in 10μL reaction volume (Bio-Rad, 1725038). qPCR was performed using Luna universal QPCR mix (NEB, M3003) in the CFX96 Real-Time PCR system with a C1000 Thermal Cycler (Bio-Rad). Results are represented as fold change above control after normalization to GAPDH.

ELISA assay

To measure IFN gamma and ADA1 concentrations, the Human IFN gamma ELISA kit (R&D Systems, DIF50C) or human ADA1 ELISA kit (Invitrogen, EH9RB) was used according to the manufacturer’s instructions. HER2-CAR or HER2-MR-CAR cells were cultured in presence or absence of A549 at 1:1 ratio. Cell Culture supernatants were collected, centrifuged, and frozen until the assay.

LDH assay

Cytotoxicity of GPC3-CAR T cells and HER2-CAR T cells were assessed and compared to GPC3-MR-CAR T cells and HER2-MR-CAR T cells using LDH assay. The GPC3-CAR and GPC3-MR-CAR T cells were co-cultured with HepG2 or Huh7 cells, and the HER2-CAR and HER2-MR-CAR T cells were co-cultured with A549 or Calu3 cells at varying E:T ratios. Cell culture supernatants were collected and measured using LDH Cytotoxicity WST Assay (ENZO, ENZ-KIT157) according to the manufacturer’s instructions.

ADA activity determination

Cells were cultured at indicated cell density for 24 h. Then cell culture supernatants and cells were collected for determination of ADA activity. ADA activity was measured using Adenosine Deaminase Assay Kit (GenWay Biotech, GWB-BQK080) according to the manufacturer’s instructions.

Inosine assay

Supernatant of equal amount of tumor tissue homogenization from Non-transduced, HER2-CAR and HER2-MR-CAR group from ex vivo analysis was used to perform Inosine assay according to Cell BioLabs Kit (MET5092) protocol and the inosine concentration was calculated from the standard curve.

Migration assay

Transwell inserts (Falcon, 5um) were used in 24 well tissue culture plate. 4x10e5 Non transduced, HER2-CAR and HER2-MRCAR cells were placed in the top well of a transwell plate (Cell Migration/Chemotaxis Assay Kit, ab235696). 235 μl of A549 cancer cells supernatant was placed in bottom well as chemoattractant. Supernatant from A549 cancer cells were collected 24 h after plating. The ability of cells to migrate assayed by both fluorescence activity and cell counting as per KIT protocol.

CFSE assay

Cells were stained with Carboxyfluorescein diacetate succinimidyl ester (CFSE) Cell Division Tracking Kit (BioLegend, 423801), and then cultured in complete RPMI1640 with IL-2 for 3 days. On day 3, cells were harvested and the CFSE fluorescent dilution was analyzed by flow cytometry.

Repeat stimulation stress test

In vitro expansion and persistence were assessed by repeatedly coculturing CAR T cells with fresh Huh-7 tumor cells every 3 to 4 days at a 1:1 ratio. At the end of each coculture interval, Annexin V staining was performed to assess CAR T cell viability 2 days after each stimulation using the ApoScreen Annexin V Apoptosis Kit (SouthernBiotech, 10010-09) according to the manufacturer’s instructions.

Cell sorting

Tumor tissues were collected from mice and filtered through 70 μm cell strainers (CELLTREAT Scientific Products, 229483) for single-cell suspensions. The filtered cells were then stained with CD3-PE (BioLegend, 300407) for Flow cytometry sorting (BD FACSAria Fusion).

RNA sequencing

RNA-seq was performed in triplicate for each experimental group. Six days after transduction RNA was isolated using the RNeasy Plus mini kit (Qiagen). RNA was checked for quality control and then sequenced by Genewiz form Azenta Life Sciences. After getting the Fastq file reads were mapped to the human reference genome (hg38) using STAR (v2.7.2b) RNAseq alignment tool. Transcript levels were quantified to the reference genome using a Bayesian approach. Normalization was done using counts per million (CPM) method. Differential expression was done using DESeq2 (v3.5) with default parameters. The normalized counts of each gene were log2 scaled and ranked by their fold changes values between paired samples. A ranked list of genes with log2[FC] values was used in pre-ranked GSEA analysis. Hallmark genesets were utilized as background genesets database in GSEA. Broad institute’s standalone GSEA (version 4.0.3) software was used to perform enrichment analysis. The RNA sequencing data has been deposited to GEO with the code GSE262447. Accession numbers for individual samples are included in the key resources table.

Quantification and statistical analysis

Descriptive analysis was performed to summarize data for cytokine production and CAR expression on T cells. Response variables were transformed, if necessary, to achieve normality. Unpaired two-tailed Student’s t test was used to assess differences between two experimental groups. Mouse survival was analyzed using the log rank test in Prism 10. Graphs that incorporate error bars present the mean as the central value, while the error bars represent the standard error of the mean (SEM). A p value < 0.05 was considered statistically significant. Since these experiments were exploratory in nature, there was no estimation to determine the appropriate sample size. Therefore, we relied on traditional sample sizes of greater than 5 for our animal studies.

Published: April 29, 2024