Suppression of T-cell responses by tumor metabolites

Abstract

Tumor cells have developed multiple mechanisms to escape T-cell-mediated immune recognition. Recent work has revealed that the altered tumor metabolism depletes essential nutrients or leads to the accumulation of immunosuppressive metabolites in the tumor microenvironment. In this review, we discuss the suppressive activity of some metabolic key players, which are upregulated in human tumor cells, including indolamine-2,3-dioxygenase (IDO), arginase, inducible nitric oxide synthetase (iNOS), and lactate dehydrogenase (LDH)-A, on the adaptive immune system. A better understanding of the impact of metabolic alterations of tumor cells on effector T-cell functions could lead to new therapeutic strategies to improve the efficacy of cancer immunotherapy.

Introduction

Cellular immune responses mediated primarily by activated T lymphocytes play an important role in eliminating malignant tumor cells [1]. Accordingly, high frequencies of tumor-infiltrating lymphocytes (TIL) are associated with a lower risk of relapse, reduced tumor progression as well as improved overall survival in cancer patients, e.g., melanoma and colorectal cancer [2, 3]. Over the past 20 years, different strategies to stimulate antitumor immunity such as therapeutic vaccination and adoptive T-cell transfer have been extensively studied in humans. However, consistently effective immunotherapy has not yet been developed for any type of malignancy. Although some maintain that the lack of broad success with current strategies is due to the escape of tumor cells from the immunotherapies directed against them, it is also possible that current immune-based treatments are simply ineffective.

Tumors have developed numerous mechanisms to evade both innate and adaptive immunity [4], including downregulation of tumor-associated antigens (TAA) and antigen-presenting machinery. Furthermore, co-stimulatory molecules such as CD80 and CD86 are not expressed on the majority of malignant cells. On the other hand, co-inhibitory molecules like PD-L1/PD-L2 are upregulated and may induce cell death of activated T cells. Additionally, tumor cells can suppress tumor-specific cellular immune responses through the production of immunosuppressive cytokines that include transforming growth factor (TGF)-β and interleukin (IL)-10.

These escape mechanisms may lead to anergy in TILs, which is reflected by missing functionality and unresponsiveness to further stimulation with TAAs [5, 6]. Functional intolerance in TILs could be explained by dysfunction in signal transduction, e.g., by downregulation of TCR-CD3-ζ, Lck or Zap-70 [7]. During the past years, new mechanisms that may explain TILs anergy gained considerable interest. These include the altered tumor metabolism and accumulation of tumor metabolites or depletion of essential nutrients playing a crucial role for sustained activation and survival of T cells (Fig. 1). Expression of indolamine-2,3-dioxygenase (IDO) was originally described in macrophages and subsets of dendritic cells. Uyttenhove and colleagues analyzed tumor cell lines as well as a variety of human tumor samples and found a constitutive expression of functionally active IDO in a number of tumor types, including prostate, colorectal, pancreatic, and cervical carcinomas [8]. Of importance, IDO-expressing tumor cells were able to block TAA-specific T-cell proliferation. Degradation of the essential amino acid tryptophan into kynurenine by IDO was shown to suppress effector T-cell function via downregulation of TCR-CD3-ζ [9]. In addition, IDO-upregulation results in deprivation of tryptophan in the local tumor environment blocking effector T-cell differentiation and inducing apoptosis in T cells [10]. Hereby, T-cell suppression is mediated by the expression of the stress-response kinase general control non-repressed 2 (GCN2), as shown by Munn and coworkers [11]. GCN2 knockout cells are insensitive to IDO-induced T-cell anergy. Of interest, elevated serum levels of tryptophan catabolites in cancer patients have been associated with cancer progression. Accordingly, clinical data demonstrate a clear association between upregulation of IDO and low infiltration of TILs linked with an increased malignancy in colorectal cancer [12]. Thus, repression of IDO by a specific inhibitor 1-methyltryptophan or by RNA interference was able to reestablish T-cell responses in mouse models and to diminish tumor growth [8, 13].

Fig. 1.

Altered tumor metabolic pathways inhibit cytotoxic T-cell (CTL) function. High glycolysis rates induce glucose (Glc) depletion as well as accumulation of lactic acid (LA) in the tumor microenvironment. In the same way, depletion of the essential amino acids tryptophan (Trp) and arginine (Arg) by an increased expression of indolamine-2,3-dioxygenase (IDO) and arginase prevents T-cell effector functions. IDO upregulation leads to an accumulation of toxic kynurenine (Kyn) derivates. An increase in nitric oxide (NO) in the tumor microenvironment is induced by high expression of nitric oxide synthase (iNOS) as well as high concentrations of adenosine (Ado) as a result of high degradation of adenosine triphosphate (ATP)

Other groups have emphasized the importance of an altered tumor arginine metabolism for the suppression of tumor-specific T-cell responses. Many tumors exhibit an increased expression of arginase and inducible nitric oxide synthetase (iNOS) leading to depletion of arginine from the tumor microenvironment [14]. Arginine depletion has been shown to inhibit protein translation by blockage of eukaryotic translation initiation factor (EIF)-2 [15]. Furthermore, iNOS leads to increased production of nitric oxide (NO), thereby promoting angiogenesis, metastasis, and immunosuppression in tumors [16]. High levels of NO block signal transduction in T cells by nitration of tyrosine and cysteine residues, thus suppressing production of IL-2 and granzyme-B via inhibition of the MAPK pathway [17]. Beside malignant cells, myeloid-derived suppressor cells (MDSCs), which are often associated with poor prognosis in cancer patients, express arginase and induce cell cycle arrest in T cells [18]. Inhibition of arginase/iNOS in both tumor cells and MDSCs was able to reconstitute effector functions of T cells, which leads to a decreased tumor growth in mouse models [19, 20]. Moreover, Rodriguez et al. postulated that arginase is induced by cyclooxygenase (COX)-2 in lung carcinoma [21]. These data may explain that increased concentrations of prostaglandin (PGE)-2 are linked to an inhibition of T-cell activation and enhanced tumorigenesis [22].

Beside tryptophan and arginine metabolites, high concentrations of adenosine could be detected in hypoxic regions of malignant tumors based on an enhanced breakdown of adenosine triphosphate (ATP). Adenosine accumulates and can trigger signaling through the A2 class of adenosine receptors, which are known to be involved in immunosuppression [23, 24]. The A2Rs are coupled with stimulatory G proteins, triggering a rise in intracellular cAMP concentrations. The increase in cAMP suppresses several effector functions of T cells [23]. Accordingly, mutant mice that lack the A2 receptor subtype exhibit improved CD8⁺ T-cell-mediated antitumor immune responses and reduced growth of experimental tumors in comparison with wild-type mice [24].

Regulatory T cells (Tregs) are crucial for the suppression of antigen-specific immune responses by activated conventional T cells. It has been demonstrated that this suppression is mediated by cAMP transported from regulatory to conventional T cells via gap junctions. Recently, it has been shown that the mechanism of cAMP accumulation in stimulated Tregs involves adenylyl cyclase-7 activation [25, 26]. Of interest, Tregs can be induced by IDO-producing acute myeloid leukemia (AML) blasts through conversion of CD4⁺ CD25⁻ into functional CD4⁺ CD25⁺ Foxp3⁺ T cells [27]. Moreover, high serum kynurenine/tryptophan ratios in AML patients correlated with poor prognosis [28].

Recent data support the idea that other metabolic pathways are dysregulated in hematologic malignancies, e.g., accumulation of 2-hydroxyglutarate (2-HG) in isocitrate dehydrogenase (IDH)-1/2-deficient leukemic blasts or an upregulated glucose metabolism in childhood pre-B acute lymphoblastic leukemia [29, 30].

Inhibitory effects of lactic acid on T cells

Many years ago, Otto Warburg found that cancer cells primarily rely on glycolysis for energy production, a phenomenon known as aerobic glycolysis or “Warburg effect” [31]. The “Warburg effect” is based on the upregulation of glycolytic enzymes such as pyruvate kinase, hexokinase, and lactate dehydrogenase (LDH) as well as an increased expression of glucose transporters (GLUT) (Fig. 2a, b). Accordingly, tumor cells are characterized by an increased uptake of glucose, and the positron emission tomography (PET) exploits this feature for tumor diagnostics.

Fig. 2.

Schematic diagram of glycolysis in non-malignant cells and tumor cells. Glucose is transported into the cell by glucose transporters, phosphorylated by hexokinase (HK), and degraded to pyruvate by glycolysis. Pyruvate is either applied to the tricarbonic acid cycle (TCA) by pyruvate dehydrogenase (PDH) or catabolized to lactate by lactate dehydrogenase (LDH)-A, which is transported to the extracellular space by monocarboxylate transporters. a In normal high glucose-containing tissues, oxidative phosphorylation serves as the main energy source and cytotoxic T cells (CTL) exhibit normal effector functions, e.g., secretion of cytokines and granules. b Activation of receptor tyrosine kinases (RTK), hypoxia-inducible factor (HIF), and mammalian target of rapamycin (mTOR) leads to an upregulation of aerobic glycolysis as well as pyruvate dehydrogenase kinase (PDK) in tumor cells inducing depletion of glucose and accumulation of lactic acid, thereby suppressing CTL functions. c Pharmacologic compounds, such as 2-deoxyglucose (2-DG), dichloroacetate (DCA), and tyrosine kinase inhibitors (TKI), impede glycolytic key proteins and may therefore restore CTL functions

The altered tumor metabolism is under the control of hypoxia as well as oncogenes and tumor suppressor genes. Expression of the myc oncogene occurs in about 30% of human cancers and leads to upregulation of glycolytic enzymes like LDHA [32]. Oncogenic myc also collaborates with hypoxia inducible factor (HIF), which is stabilized in response to hypoxia and induces the transcription of more than 70 genes e.g., GLUT-1 and LDH [33]. Moreover, genetic alteration or loss of p53, one of the most frequently mutated genes in cancer, also leads to decreased oxygen consumption and increased lactate production [34].

Glucose is metabolized via glycolysis, and its end product lactate is secreted in cotransport with protons. Mueller-Klieser and colleagues described a correlation between high lactate levels in the primary lesion of different tumor entities with the incidence of distant metastases as well as a reduced overall survival in cancer patients [35]. This hypothesis is supported by our own data showing a positive correlation between lactate serum levels and tumor burden in cancer patients [36]. Therefore, it seems likely that the glycolytic phenotype of tumor cells plays an important role for tumor progression and metastatic disease.

Several in vitro studies have addressed the immunomodulatory properties of high lactate levels. Lactic acid concentrations used for these experiments mimic the intratumoral micromilieu as intratumoral analyses revealed concentrations of up to 40 mM lactate inside the tumor tissue [37]. We and others have shown that tumor-derived lactic acid strongly inhibits the differentiation of monocytes from dendritic cells [38, 39]. Shime et al. demonstrated that lactic acid regulates transcription and secretion of IL-23 in human monocytes/macrophages, a tumor-promoting cytokine that is involved in the generation of Th17 cells [40]. Furthermore, lactate showed positive as well as negative effects on T-cell mediated immune responses [36, 41, 42]. Droge et al. studied the effect of lactate on murine T-cell populations and found that lactate was able to suppress the cytotoxic T-cell response in vivo, whereas activation of cytotoxic T cells (CTL) in vitro was augmented [41, 43]. In contrast, Feder-Mengus described an inhibitory effect of lactic acid on human CTL in melanoma spheroid cocultures [42]. In line with these data, we have demonstrated that lactic acid suppresses the proliferation and cytokine production of human TAA-specific T cells up to 95% and the cytotoxic activity for up to 50% [36]. CTL-infiltrating lactic acid-producing multicellular tumor spheroids also revealed a reduced cytokine production compared with the controls, whereas pre-treatment of tumor spheroids with an inhibitor of lactic acid production could partially abrogate this inhibitory effect. Of interest, a recovery period of 24 h in medium without lactic acid restored cytokine production suggesting that lactic acid-induced T-cell suppression is reversible. Therefore, a therapeutic rescue of CTL from lactic acid-mediated suppression seems conceivable and suggests that selective targeting of glycolysis in tumor cells may restore immune cell activation and effector function toward tumor cells.

Warburg phenotype in renal cell carcinoma results in low CD8⁺ T-cell infiltration

Of interest, metabolomic profiling of the serum of patients with renal cell carcinoma (RCC) revealed low levels of glucose and high levels of lactate [44]. This may be linked to the finding that clear cell carcinoma is often associated with gene mutations most notably in the von Hippel Lindau (VHL) gene [45]. Loss in pVHL results in stabilization of HIF1α, inducing a hypoxia-like phenotype in RCC. In our studies, immunohistochemical analyses of tissue microarrays (TMA) of 249 RCC patients revealed an upregulation of GLUT-1 in clear cell carcinoma [46]. Our data are supported by others demonstrating an association of high GLUT-1 expression with HIF1α and a trend toward shorter survival in patients with high GLUT-1 expression [47]. Additionally, we observed a significantly higher expression of LDH5 in RCC tumor tissues compared with the corresponding normal kidney tissue [46]. LDH5, the isoenzyme of lactate dehydrogenase with the highest efficiency for the transformation of pyruvate to lactate, is not only overexpressed in many tumors but also linked to an aggressive phenotype [48]. In line with these results, upregulation of the LDHA isoenzyme but also other enzymes involved in glycolysis such as aldolase, pyruvate kinase M2, and glycerine-aldehyde-3-phosphate indicates a global shift from respiration to glycolysis in this tumor type.

RCC, especially of the clear cell type, is considered as a highly immunogenic tumor showing a significant rate of spontaneous regression and sustained response to interferon and IL-2-based immunotherapy in a subgroup of patients. Surprisingly, there is no clear positive correlation between the numbers of tumor-infiltrating lymphocytes and prognosis in RCC patients [49], which is in contrast to reports demonstrating a significant improved survival in colorectal cancer patients with a high TIL rate [3]. TILs in RCC are mainly CD3⁺ T cells with a predominance of CD8⁺ over CD4⁺ T-cells. Van den Hove and co-workers described that CD8⁺ TILs are highly activated, but display poor cytolytic activity against tumor cells ex vivo [50].

Our data revealed a high infiltration of CD3⁺ CD8⁺ T cells, but also Foxp3⁺ T regulatory cells in RCC. Furthermore, expression of effector molecules such as granzyme-B and perforin was rather low in the tumor compared with normal kidney tissues indicating that RCCs are infiltrated by functionally inactive T cells [46]. In line with these results, Attig et al. described a simultaneous infiltration of effector and suppressor T cells into RCC [51]. Interestingly, when T-cell infiltration in clear cell carcinoma was correlated with different grades of GLUT-1 and LDH5 expression in the tumor, we found that high expression of both GLUT-1 and LDH5 significantly correlated with a low infiltration rate of CD3⁺ T cells. Our data indicate that high expression of the glucose transporter GLUT-1 and LDH5 by tumor cells may suppress CD3⁺ T-cell infiltration and/or survival in the tumor environment. High LDH5 expression could result in high lactate levels at the tumor site. As lactic acid has been shown to induce apoptosis and decreased effector functions in T cells in vitro, this could be a possible explanation for the lower numbers of CD3⁺ T cells in LDH5-expressing RCC [36]. Furthermore, highly proliferating tumor cells but also activated T cells rely on glucose metabolism to provide energy for proliferation and effector functions and may therefore compete for glucose in the tumor microenvironment. Accordingly, activation of T cells leads to accelerated glucose uptake and glycolysis [52] and is critical for the survival as well as the production of IFN-γ. Thus, an accelerated glucose metabolism of tumor cells may be of crucial importance for the survival and function of tumor-infiltrating effector T cells.

Clinical perspectives

Induction of tumor-specific T-cell responses provides the basis for successful immunotherapeutic approaches against cancer. One reason for the failure of spontaneously or induced antitumor immunity may be the altered tumor metabolism leading to accumulation of suppressive metabolites or depletion of nutrients required for T-cell effector functions. Modification of tumor-specific T cells or inhibition of crucial metabolic pathways in cancer represents an interesting tool to improve the efficacy of cancer immunotherapy.

One possible approach would consist of the modification of tumor-specific T cells by genetic means to make them resistant to suppressive tumor metabolites thus enhancing T-cell reactivity. Knockdown of GCN2 using siRNA technology in tumor-reactive T cells has been shown to render T cells resistant to IDO-induced anergy [11].

Another approach directly targets the tumor environment: inhibition of key metabolic enzymes such as IDO, iNOS, and arginase may reconstitute antitumoral T-cell activity [8, 19]. Our group has demonstrated that inhibition of LDHA by oxamic acid in tumor spheroids could recover T-cell effector functions in vitro [36]. In preclinical and clinical studies, multiple pharmacological modulators of tumor metabolism are under investigation. As increased glycolysis in cancer is regulated by changes in signaling pathways such as Akt, mTOR, and transcription factors such as HIF1 preclinical studies aim to target these metabolic pathways (Fig. 2c). Temsirolimus, a specific inhibitor of mTOR, improved survival among patients with metastatic renal cell carcinoma and a poor prognosis [53]. PX-478, an inhibitor of HIF1α and HIF-1 transcription factor activity, has marked antitumor activity in vitro and in vivo, which correlates negatively with HIF-1α levels [54]. A novel HIF-1 inhibitor, LW6, promotes proteasomal degradation of HIF1α via upregulation of VHL in colon cancer lines [55]. In addition, there are immunologic approaches to target some of the downstream effectors of HIF. Bevacizumab, a monoclonal antibody that binds to circulating VEGF protein, produced a significant prolongation of time to disease progression compared with placebo in patients with treatment-refractory metastatic RCC in a small randomized trial [56]. Further studies aim to inhibit key glycolytic enzymes, e.g., hexokinase, using 2-deoxyglucose (2-DG) or lonidamine. Eberhart et al. found 2-DG to increase the apoptotic effect of glucocorticoids on ALL cell lines, thereby allowing a dose reduction of the latter to sublethal concentrations [57]. Moreover, 2-DG was administered in order to optimize radiotherapy in malignant glioma patients [58]. Hereby, transient side effects such as hypoglycemia as well as restlessness were observed in those patients that received higher doses of 300 mg 2-DG/kg BW. Another promising drug is dichloroacetate (DCA), an inhibitor of pyruvate dehydrogenase kinase (PDK), which facilitates the conversion of pyruvate to lactate. DCA led to reduction in tumor growth in xenograft models and showed remarkable anticancer effects in preclinical studies [59]. Of particular interest, tyrosine kinase inhibitors, e.g., imatinib, are highly efficient for treatment of various cancer entities. Imatinib treatment decreased the activity of glycolytic enzymes in leukemia cells, leading to the suppression of aerobic glycolysis [60].

In summary, research on tumor metabolism and its influence on adaptive immune responses may represent the basis for innovative anticancer therapies. The development of drugs that specifically modulate the altered tumor metabolism may not only target cancer cells but may also result in an improved tumor-specific T-cell response.