Cancer Cells Don’t Live Alone: Metabolic Communication within Tumor Microenvironments

Summary

Solid tumors reside in harsh tumor microenvironments (TMEs) together with various stromal cell types. During tumor progression and metastasis, both tumor and stromal cells undergo rapid metabolic adaptations. Tumor cells metabolically coordinate or compete with their “neighbors” to maintain biosynthetic and bioenergetic demands while escaping immunosurveillance or therapeutic interventions. Here, we provide an update on metabolic communication between tumor cells and heterogeneous stromal components in primary and metastatic TMEs, and discuss emerging strategies to target metabolic communications for improved cancer treatments.

Growing solid tumors consist of malignant cancer cells and heterogeneous stromal cell components. This Review from Simon and Li provides an updated overview of metabolic communication between tumor cells and stromal cells in primary and metastatic tumor microenvironments, and discusses emerging strategies to target metabolic interactions for improved cancer therapies.

Introduction

Growing primary solid tumors consist of malignant cancer cells harboring genetic alterations and heterogeneous non-malignant stromal cell components (Egeblad et al., 2010; Hanahan and Weinberg, 2011). Disease progression is strongly influenced by metabolic stress imposed by the local microenvironments, due to limited oxygen and nutrient supply, accumulation of metabolic waste and unfavorable pH (De Berardinis and Chandel, 2016; Gouirand et al., 2018). A majority of cancer-related deaths result from the spread of primary tumor cells to distal metastatic sites, a multi-step and highly inefficient process (Mehlen and Puisieux, 2006; Valastyan and Weinberg, 2011). To execute progression and metastasis, cancer cells undergo metabolic evolution to maximize nutrient utilization for bioenergetic and biosynthetic demands, survive harsh tumor microenvironments (TMEs) and escape immunosurveillance. This metabolic evolution, dictated by both intrinsic and extrinsic factors (Faubert et al., 2020), constitutes a hallmark of cancer (Hanahan and Weinberg, 2011). Previous studies from isolated tumors and cell lines have uncovered important cancer cell-intrinsic metabolic remodeling controlled by oncogenic signaling and oncometabolites (De Berardinis and Chandel, 2016; Gouirand et al., 2018; Pavlova and Thompson, 2016) (Vander Heiden and DeBerardinis, 2017; Lehúede et al., 2016). The field has also begun to explore metabolic communications between tumor cells and TMEs (Gupta et al., 2017; Lyssiotis and Kimmelman, 2017), and to exploit these for therapeutic interventions (Li et al., 2019). In this review, we will summarize recent key findings that improve our understanding of metabolic crosstalk mechanisms in primary and metastatic TMEs.

Metabolic Crosstalk among Solid Tumor Compartments

Significant metabolic heterogeneity exists within solid tumors, which stems largely from vascular integrity and proximity to the vasculature that create oxygen and nutrient gradients. Consequently, regional tumor cells exhibit distinct metabolic profiles. For example, in human non-small cell lung cancer (NSCLC), well-vascularized tumor subdomains utilize multiple nutrients, while less-perfused regions use glucose as their main carbon source (Hensley et al., 2016). Importantly, lactate (once considered simply metabolic waste) is preferred over glucose to fuel the tricarboxylic acid (TCA) cycle in NSCLC (Faubert et al., 2017). In this context, lactate and glucose are utilized in parallel, as lactate can be directly channeled into the TCA cycle by mitochondrial LDH activity (Chen et al., 2016). Lactate metabolism via LDHA may also impact cytosolic redox status and redirect glucose into the pentose phosphate pathway and hexosamine biosynthesis. Alternatively, lactate as a signaling molecule potentially changes intracellular signaling pathways and/or gene expression to impact glucose uptake and catabolism.

Cancer cells within tumor compartments cooperate to form metabolic “symbiosis”. The lactate shuttle is one example, where cancer cells in hypoxic regions consume glucose through anaerobic glycolysis and release lactate, lactate is then used as a fuel for TCA cycle by cancer cells in adjacent oxygenated tumor regions (Nakajima and Van Houten, 2013; Sonveaux et al., 2008). Acute hypoxia induced by anti-angiogenic therapy drives pancreatic neuroendocrine and breast cancer cells to produce excessive lactate, which is then used by cancer cells in proximity to blood vessels (Allen et al., 2016; Pisarsky et al., 2016). Similar metabolic symbiosis has also been described in lung and colon cancers, indicating that it may represent a general crosstalk pathway. The lactate shuttle likely results from differential expression of appropriate monocarboxylate transporters (MCTs): hypoxic cancer cells express high MCT4 levels that function as major lactate exporter, whereas oxygenated cancer cells express MCT1 as lactate importer (Figure 1). It is currently unclear to what extent hypoxic tumor-derived lactate contributes to oxidative metabolism in general, as well-oxygenated cancer cells can readily metabolize glucose and other circulating nutrients including lactate. Furthermore, the precise function(s) and mechanism (s) of lactate in this regard remain to be fully understood.

Figure 1. Metabolic Heterogeneity and Symbiosis among Solid Tumor Compartments.

Vascular integrity and proximity to vasculature create oxygen and nutrient gradients leading to intratumoral metabolic heterogeneity. In hypoxic regions, cancer cells increase glucose uptake and preferentially convert it to lactate; lactate is exported through MCT4, and then imported through MCT1 by cancer cells in more oxygenated regions. In addition to glucose and lactate, oxygenated cancer cells also use alternative fuels for oxidative metabolism, and potentially provide amino acids (AAs) and lipids to hypoxic cancer cells.

In a study by Pan et al, tumor cores were found to have much lower levels of amino acids than peripheral regions, including glutamine, arginine, asparagine, serine and aspartate (Pan et al., 2016). It’s tempting to speculate that peripheral tumor cells can release these amino acids for more internal cancer cells. Hypoxic cancer cells increasingly uptake exogenous unsaturated fatty acid to maintain lipid homeostasis (Ackerman et al., 2018; Young et al., 2013), and may benefit from oxygenated cancer cells undergoing fatty acid synthesis (Figure 1). Interestingly, acetate can be generated de novo from pyruvate (Liu et al., 2018a); since acetate is a bioenergetic substrate for human glioblastoma (Mashimo et al., 2014), it’s possible that well-perfused glioblastoma cells can synthesize acetate for hypoxic counterparts in this context. Alanine is one major carbon source for pancreatic ductal adenocarcinoma (PDAC) cells (Sousa et al., 2016), so oxygenated PDAC cells could provide alanine for hypoxic cells locally. Additionally, metabolic recycling of ammonia via glutamate dehydrogenase (GDH) has been shown to support breast cancer biomass (Spinelli et al., 2017), raising the possibility of utilizing ammonia by GDH-expressing cancer cells to synthesize glutamine for neighboring cancer cells with high demand of glutamine catabolism.

Depending on cancer type and local nutrient pool, it will be interesting to identify other biomass, energy and antioxidants that can be transferred within solid tumor compartments, and define the underlying cross-feeding pathways. The metabolic heterogeneity can further be uncovered through unbiased metabolic profiling and molecular characterization of tumor subdomains. Ultimately, the corresponding functional links of intratumoral metabolic crosstalk to tumor growth warrant careful investigation.

Cancer-Fibroblast Metabolic Symbiosis

Bidirectional metabolic communications between tumor cells and stromal cells contribute to tumor growth while affecting therapeutic responses (Lyssiotis and Kimmelman, 2017). Cancer-associated fibroblasts (CAFs) constitute a major stromal component that critically modulates tumor initiation, progression and metastatic dissemination (Kalluri, 2016). Chronically activated from normal fibroblasts, CAFs exhibit increased proliferation rates, survival potential and undergo metabolic reprogramming (Chaudhri et al., 2013; Du et al., 2018; Sousa et al., 2016; Zhang et al., 2015). Cancer cells and CAFs metabolically communicate through multiple mechanisms (Figure 2). A lactate shuttle, known as “reverse Warburg effect”, was described previously (Pavlides et al., 2009). Specifically, CAFs metabolize glucose through anaerobic glycolysis and export lactate, which is then taken up and utilized by oxidative cancer cells (Whitaker-Menezes et al., 2011). Indeed, intercellular contact triggers elevated glucose uptake and lactate production in prostate fibroblasts, potentially through upregulated GLUT1 and MCT4, respectively. Conversely, prostate cancer cells are reprogrammed toward aerobic metabolism, with GLUT1 downregulation and increased MCT1 expression and lactate uptake (Fiaschi et al., 2012). CAF-derived lactate may only contribute partly to cancer metabolism and growth, considering multiple lactate and other nutrient pools in vivo. Notably, the “reverse Warburg effect” does not apply to all CAF-cancer interactions, likely dictated by the cell/tissue of origin. For example, breast and colon cancer cells consume glucose and release lactate to surrounding fibroblasts (Koukourakis et al., 2006; Rattigan et al., 2012), whereas pancreatic and ovarian CAFs consume rather than release lactate (Sousa et al., 2016), which is due to low glycolytic activity (Sousa et al., 2016; Yang et al., 2016).

Figure 2. Tumor-Stroma Metabolic Communications in the TME.

Heterogenous stromal cell populations metabolically cooperate with cancer cells directly or indirectly in the tumor microenvironment. Ovarian cancer-associated fibroblasts (CAFs) provide cysteine (Cys) and reduced GSH to withstand oxidative stress. Ovarian CAFs also provide cancer cells with glutamine (Gln) and utilize cancer cell-derived glutamate (Glu) to regenerate Gln. Pancreatic CAFs provide cancer cells with alanine (Ala) through autophagy, with lysophosphatidylcholines (LPCs) to support phosphatidylcholine synthesis, and indirectly with collagen-derived proline (Pro) to support survival under nutrient limitation. Similarly, breast CAFs supply cancer cells with autophagy-derived dipeptides. Prostate and pancreatic CAF-derived exosomes provide nutrient cargo to cancer cells. Mesenchymal stem cells (MSCs) shuttle mitochondria and/or mitochondrial DNA into leukemia, lung and breast cancer cells, and consume cystine to provide leukemic cells with Cys. Tumor cell-derived Lactate (Lac) stimulates CD4⁺ T cell differentiation into regulatory T cells (Tregs), promotes tumor-associated macrophage (TAM) polarization, but inhibits natural killer (NK) cells and effector T cells (Teffs). Similarly, tumor cell-derived kynurenine (kyu) facilitates Tregs differentiation but limits function of Teffs. Murine sarcoma-derived retinoid acid (RA) promotes intratumoral monocyte differentiation toward TAMs. TAMs promote tumor growth partly by providing metabolites such as polyamines. Adipocytes provide ovarian and breast cancer cells with fatty acids (FAs), pancreatic cancer cells with Gln, and engage in an arginine (Arg) cycling pathway using citrulline (Cit) to produce nitric oxide (NO).

In response to PDAC cells, activated pancreatic stellate cells (PSCs) (a major type of pancreatic CAFs) secrete the amino acid alanine through augmented autophagy. By doing so, PSCs provide a major carbon source for PDAC cells, relieving dependency on glucose and glutamine, two essential but usually limited nutrients (Sousa et al., 2016). Collagen-derived proline promotes PDAC cell survival under nutrient limited conditions, providing another evidence of PSCs supporting PDAC metabolism and growth (Olivares et al., 2017). Interestingly, PSCs also secrete lysophosphatidylcholines (LPCs) to support phosphatidylcholine synthesis and lysophosphatidic acid (LPA) production by PDAC cells (Auciello et al., 2019). Under nutrient limitation, ovarian CAFs use branched-chain amino acids (BCAAs) and aspartate to synthesize glutamine for cancer cells; co-targeting glutamine synthetase and glutaminase in this case significantly reduces tumor growth and metastasis (Yang et al., 2016). Interestingly, ovarian CAFs induces glycogenolysis in co-cultured cancer cells, which is funneled into glycolysis, leading to increased proliferation and invasion (Curtis et al., 2019). In addition, CAFs release macromolecules to support cancer cells. As examples, lung CAFs increasingly secrete dipeptides via autophagy (Chaudhri et al., 2013), prostate and pancreatic CAFs release exosomes to deliver a spectrum of amino acids, fatty acids (FAs), and TCA cycle metabolites that are taken up and utilized by cancer cells (Richards et al., 2017; Zhao et al., 2016). Moreover, even stromal cell organelles can be transferred to cancer cells to sustain the metabolism and growth. In the study by Spees et al, mitochondrial DNA was depleted in lung cancer cells leading to an inactive electron transport chain (ETC); these cells (called rho-zero cells, þ⁰) reacquire ETC activity through mitochondria transfer from co-cultured bone marrow mesenchymal stem cells (MSCs) (Spees et al., 2006).

Cancer cells are often challenged with intense redox stress, particularly when exposed to chemotherapies (Bansal and Simon, 2018). In these contexts, CAFs modulate redox homeostasis in neighboring cancer cells. Indeed, ovarian CAFs release glutathione (GSH) and cysteine to cancer cells, thereby maintaining redox balance and sustaining chemoresistance (Wang et al., 2016). Similarly, bone marrow stromal cells import cystine and convert it to cysteine, which is then released into the microenvironment and taken up for GSH synthesis by chronic lymphoid leukemia (Zhang et al., 2012). Collectively, these findings provide important metabolic connection between cancer cells and CAFs. A crosstalk between deregulated hepatocyte metabolism and hepatic stellate cells promotes liver tumorigenesis (Li et al., 2020), adding to another layer of metabolic links between CAFs and non-tumor cells in TMEs. While CAFs engage distinct feeding strategies to support corresponding tumor metabolism and growth, cancer cell-derived metabolite(s) that drive metabolic remodeling in CAFs remain to be determined in different tissue contexts.

Immunomodulation by Tumor Cell-derived Metabolites

Evasion from immune surveillance is a hallmark of cancer (Hanahan and Weinberg, 2011). Tumor cells adopt several strategies to dampen immune system, including secreting metabolites to modulate the TME immune profiles (Figure 2) (Vinay et al., 2015). In hypoxic tumor regions, high concentrations of cancer cell-derived lactate impose pleotropic effects on immune cells. Lactate blocks monocyte and dendritic cell differentiation (Dietl et al., 2010; Fischer et al., 2007), blunts T cell activation and tumor immunosurveillance (Brand et al., 2016). On the other hand, lactate promotes differentiation and polarization of tumor-associated macrophages (TAMs) toward an M2-like phenotype with elevated expression of Arginase-1 (ARG1) and mannose receptor C type 1 (CD206). In turn, M2-like TAMs produce immunosuppressive cytokines (e.g. IL10), and metabolites like polyamines, which are essential for cell division (Colegio et al., 2014). These findings support the hypothesis that reducing lactic acid production can enhance the efficacy of anticancer immunotherapy, which remains to be fully tested in vivo.

Increased tryptophan catabolism by cancer cells produces kynurenine, a ligand of endogenous aryl hydrocarbon receptor (AHR) (Opitz et al., 2011). Activation of a kynurenine-AHR signaling axis in CD4⁺ T cells favors their differentiation into immunosuppressive regulatory T cells (Mezrich et al., 2010; Munn et al., 1999; Nguyen et al., 2010). In a similar study, tumor-repopulating cells transfer kynurenine to induce PD-1 expression in CD8⁺ T cells, contributing to impaired effector functions (Liu et al., 2018b).

Murine sarcoma cells produce retinoic acid (RA) to polarize intratumoral monocyte differentiation toward TAMs and away from dendritic cells, resulting in immune suppression and tumor growth (Devalaraja et al., 2020). Additionally, lipid accumulation in tumour infiltrating myeloid cells, including myeloid-derived suppressor cells (MDSCs) and TAMs, has been shown to promote metabolic reprogramming and skew these immune cells towards immunosuppressive phenotypes (Al-Khami et al., 2017; Li et al., 2019; Niu et al., 2017). These lipids may come partly from neighboring cancer cells with enhanced fatty acid synthesis.

Together, tumor cell-derived metabolites create favorable immune microenvironment for disease progression. Some of these metabolites also function as signaling molecules, as discussed below. To identify more immunomodulatory metabolites, comprehensive metabolic profiling of cancer cell secretome followed by functional tests on specific immune cell subsets would be helpful.

Metabolic Communications between Tumor Cells and Adipocytes

Obesity is closely linked to increased risk and malignance of many types of cancer (Donohoe et al., 2017; Lengyel et al., 2018). It’s now appreciated that adipocytes and adipose tissue (AT) directly mediate some protumorigenic effects of obesity (Cao, 2019) (Figure 2). For example, adipocytes serve as a source of extracellular lipids for cancer cells (Woolthuis et al., 2016). Indeed, co-cultured adipocytes undergo lipolysis and provide FAs to increase breast cancer cell proliferation (Hoy et al., 2017). Human omental adipocytes also induce co-cultured ovarian cancer cells to overexpress CD36 (also known as FA translocase), which is responsible for enhanced FA uptake, cholesterol and lipid droplet (LD) accumulation, as well as tumor growth (Ladanyi et al., 2018). In these contexts, cancer cell-derived trigger(s) of adipocyte lipolysis need to be carefully characterized. Under hypoxia, cancer cells increase extracellular lipid utilization for bioenergetic and biosynthetic demands, as well as to maintain membrane homeostasis (Ackerman et al., 2018; Young et al., 2013). Specifically, HIF-1α induces the expression of FABP (fatty acid binding protein) 3 and 7 for lipid uptake, and adipophilin, a LD structural protein (Bensaad et al., 2014); HIF-2α promotes LD coat protein PLIN2 expression for lipid storage (Qiu et al., 2016). Given the fact that CAFs can deliver fatty acids to cancer cells via exosomes (Richards et al., 2017; Zhao et al., 2016), and the hypothesis that albumin-bound lipids can be taken up via micropinocytosis (Recouvreux and Commisso, 2017), it’s quite possible that these endocytosis pathways are also involved in lipid transfer from adipocytes to cancer cells.

In addition to adipocytes, ovarian adipose stromal cells have also been shown to metabolically communicate with cancer cells, through arginine metabolism. In this regard, ovarian cancer cells metabolize arginine to produce nitric oxide and citrulline through inducible nitric oxide synthetase (iNOS). The nitric oxide promotes glycolysis and cancer cell proliferation, while citrulline is released and captured by stromal adipocytes to convert back to arginine; the arginine is further excreted into extracellular spaces and utilized by cancer cells, forming a symbiotic metabolic loop (Rizi et al., 2015).

Nutrient Competition between Cancer Cells and Immune Cells

To maximize nutrient utilization, cancer cells must compete with stromal cells for limited substrates, especially under metabolic stress (De Berardinis and Chandel, 2016; Nakazawa et al., 2016; Pavlova and Thompson, 2016). We focus on emerging cancer-immune cell nutrient competition (Figure 3), due to metabolic flexibility connected to functional activity of immune cell subsets (Buck et al., 2017; Kedia-Mehta and Finlay, 2019).

Figure 3. Nutrient Competition between Cancer Cells and Immune Cells.

Increased glucose (Glu), arginine (Arg), tryptophan (Trp), serine (Ser) and methionine (Met) uptake and catabolism by cancer cells directly limits their availability to effector T cells (Teffs). Tryptophan (Trp) catabolism by cancer cells and tumor-associated macrophages (TAMs) produces immunosuppressive kynurenine (kyu) to facilitate Tregs differentiation. Glucose (Glu) catabolism by cancer cells produces lactate (Lac) to promote macrophage polarization into TAMs.

Glucose abundance has strong impacts on cellular metabolism and growth (Lyssiotis and Kimmelman, 2017; Renner et al., 2017). Glucose utilization is increased in tumor cells, resulting in low extracellular levels and metabolically restrictive environments for infiltrating immune cells. For instance, glucose consumption by melanoma cells limits glucose availability for T cells, leading to decreased abundance of glycolytic intermediate phosphoenolpyruvate (PEP). PEP modulates T cell receptor-mediated Ca2⁺-NFAT signaling and effector functions by repressing sarco/ER Ca2⁺-ATPase (SERCA) activity. Consequently, defective antitumor T cell responses allow for melanoma growth (Ho et al., 2015). Similarly, glucose consumption by murine sarcoma cells metabolically restricts T cells and limits their effector functions (Chang et al., 2015). In these contexts, antitumor T cell responses can be restored by blocking glucose uptake in cancer cells and redirecting more glucose to infiltrating T cells. Indeed, increased tumor glycolysis is associated with immune resistance to adoptive T cell therapy in melanoma, whereas inhibition of glycolysis enhances T cell-mediated antitumor immunity (Cascone et al., 2018).

Increased dependency of extracellular arginine has been observed in cancer (Patil et al., 2016). Several types of cancer also lack the urea cycle enzyme argininosuccinate synthetase 1 (ASS1), rendering them unable to synthesize endogenous arginine and exclusively dependent on exogenous supply (Keshet et al., 2018; Ochocki et al., 2018). Due to high levels of iNOS and ARG1 expression, arginine can also be rapidly catabolized by MDSCs and macrophages (Mondanelli et al., 2017). In these regards, arginine availability profoundly impacts effector T cells (Fletcher et al., 2015; Lamas et al., 2012). Indeed, arginine sufficiency induces global metabolic changes and promotes the generation of central memory-like cells, whereas arginine limitation conversely dampens T cell effector functions through direct amino acid deprivation (Geiger et al., 2016). Currently, replenishing arginine and preventing arginine degradation are attractive strategies to reinvigorate T cell effector function (Li et al., 2019).

Similar to arginine, tryptophan can be depleted by tumor cells and macrophages through increased uptake and catabolism. Indoleamine 2,3-dioxygenase (IDO), the first and rate-limiting enzyme of tryptophan catabolism through the kynurenine pathway, is usually highly expressed by tumor cells and macrophages (Munn et al., 1999; Platten et al., 2012). Tryptophan catabolism further produces immunosuppressive metabolite kynurenine, promoting regulatory T cell differentiation (Mondanelli et al., 2017; Nguyen et al., 2010). Consequently, inhibition of tryptophan catabolism together with immune checkpoint blockade is being tested in several ongoing clinical trials (Li et al., 2019).

Serine is a non-essential amino acid that can be taken up or synthesized de novo through the serine synthesis pathway (SSP) (Locasale, 2013; Yang and Vousden, 2016). Certain cancers such as breast cancer and melanoma show gene amplification of the SSP enzymes and depend on SSP for survival even under serine-fed conditions (Mattaini et al., 2016; Pacold et al., 2016). Many other cancer cells selectively consume exogenous serine, which is converted to intracellular glycine and one-carbon units for building nucleotides and maintaining mitochondrial redox homeostasis (Yang et al., 2020; Ye et al., 2014). Importantly, extracellular serine is required for optimal T cell expansion and effector functions. This occurs in a manner where serine supplies glycine and one-carbon units for de novo nucleotide biosynthesis in proliferating T cells (Ma et al., 2017).

Methionine is an essential amino acid that participates in protein synthesis and also produces S-adenosyl-L-methionine (SAM) for all methylation reactions. MAT2A, the enzyme involved in methionine catabolism, has been identified as an oncogene overexpressed in cancer (Ramani and Lu, 2017). Methionine is rapidly taken up by activated T cells and serves as the major substrate to maintain the SAM pools. Methionine restriction reduces histone H3K4 trimethylation and expression of key genes involved in Th17 cell proliferation and cytokine production (Roy et al., 2020). Thus, methionine consumption by cancer cells potentially affects T cell activation and differentiation.

Overall, advantageous consumption of essential nutrients by cancer cells directly limits the availability to tumor-killing immune cells, especially the cytotoxic T cells, and also produces immunosuppressive metabolites (e.g. lactate and kynurenine), collectively leading to impaired antitumor immunity (Figure 3). In addition to rapid proliferation, cancer cells outcompete by overexpressing transporters for nutrient uptake and enzymes for nutrient catabolism, which are largely controlled by oncogenic signaling pathways and/or oncometabolites (Pavlova and Thompson, 2016).

Metabolites Mediate Intercellular Crosstalk in TMEs

Metabolites were traditionally considered as metabolic intermediates or end products involved in bioenergetics and macromolecule biosynthesis, but recent years have witnessed a great expansion in delineating their actions to regulate signal transduction and gene expression, through both cell autonomous and non-autonomous mechanisms (Haas et al., 2016; Liu and Wellen, 2020). Cell autonomous mechanisms include direct interactions between metabolite and protein component of signaling pathways, modulating metabolite sensor pathways, modifications of protein stability/activity, and regulation of epigenome and epitranscriptome (Haas et al., 2016; Liu and Wellen, 2020). In a non-autonomous manner, metabolites released into the TME can signal to neighboring cells to mediate intercellular crosstalk. Here we highlight some metabolites that exhibit extracellular or new intracellular actions as signaling molecules.

Lactate exhibits pleotropic effects on various cell types in the TMEs, most of which are attributed to indirect mechanisms, such as affecting local pH, modulating cellular metabolism and/or redox status (Baltazar et al., 2020; de la Cruz-López et al., 2019). Recent studies provide new insights into how lactate directly regulates signal transduction and gene expression (Figure 4). In the study by Lee et al, lactate binds to and stabilizes NDRG3 (NDRG family member 3, a PHD2/VHL substrate), and mediates hypoxia-induced activation of the Raf-ERK pathway essential for angiogenesis and cell growth (Lee et al., 2015). These findings uncover a lactate-induced mechanism for cellular adaptation to hypoxia, which is different from those controlled by HIFs (Lee et al., 2020a). Glycolysis-derived lactate inhibits retinoic-acid inducible gene I (RIG-I)-like receptor (RLR) signaling by directly binding to a mitochondrial antiviral-signaling protein (MAVS) transmembrane domain and preventing MAVS aggregation, thereby limiting type I IFN response (Zhang et al., 2019b). More recently, lactate was shown to promote a new histone posttranslational modification termed lactylation in macrophages. Specifically, glucose-derived lactate produces lactyl-CoA that contributes a lactyl group to histone protein lysine tails through acetyltransferase enzyme p300, activating wound-healing genes and resulting in an M2-like phenotype (Zhang et al., 2019a).

Figure 4. Emerging Functions of Lactate in Regulating Signal Transduction and Gene Expression.

(1) Hypoxia-induced lactate binds to NRDG3 and prevents it from pVHL-dependent degradation; stabilized NDRG3 protein binds c-Raf to mediate activation of the Raf-ERK pathway. (2) Binding to lactate interrupts MAVS mitochondrial localization, RIG-I and MAVS interaction, subsequent MAVS aggregation, and attenuates downstream TBK1-IRF3 signaling. (3) Lactate accumulation produces lactyl-coA for histone lactylation that contributes to target gene expression in macrophages.

Accumulation of nucleoside adenosine has been detected in TMEs. Adenosine is generated in tumors and regulatory T cells through ectonucleotidases (CD39 and CD73) that convert extracellular adenosine triphosphate (ATP) to adenosine. Adenosine binds to surface adenosine 2A receptor (A2AR) on cytotoxic T cells and NK cells, inhibiting antitumor immunity (Haskó et al., 2008). Accordingly, A2AR blockade has been shown as an immunotherapy for treatment-refractory renal cell cancer (Fong et al., 2020).

As described above (Figure 2), kynurenine and RA are two metabolites derived from cancer cells and function in immune cells; kynurenine promotes regulatory T cell differentiation and upregulates PD-1 in CD8⁺ T cells by AHR signaling, while RA polarizes intratumoral monocyte differentiation toward TAMs via RAR signaling (Devalaraja et al., 2020).

(R)-2-hydroxyglutarate (R-2-HG) and succinate are oncometabolites that act as epigenetic modifiers by interfering with α-ketoglutarate (α-KG)-dependent dioxygenases (Kaelin and McKnight, 2013; Liu and Wellen, 2020; Lu and Thompson, 2012). Apart from intracellular functions, they have recently been shown to mediate intercellular crosstalks. Interestingly, IDH1-mutant glioma-derived R-2-HG is taken up by T cells and perturbs NFAT transcriptional activity and polyamine biosynthesis, resulting in T cell activity suppression. In this context, antitumor immunity is improved by inhibition of the neomorphic enzymatic function of mutant IDH1 (Bunse et al., 2018). In another study, cancer cells release succinate and activate the succinate receptor (SUCNR1)-PI3K-HIF-1α axis to polarize macrophages into TAMs, functionally promoting tumor metastasis (Wu et al., 2020). These findings uncover non-tumor cell-autonomous roles of oncometabolites in shaping the TMEs.

Another notable metabolite is itaconate, which is synthesized from cis-aconitate in the TCA cycle specifically in activated macrophages (O’Neill and Artyomov, 2019). Itaconate was shown to mediate crosstalk between macrophage metabolism and peritoneal tumor growth in murine models. Specifically, tumor cells elicit metabolic reprogramming and itaconate accumulation in peritoneal tissue-resident macrophages, where inhibition of itaconate synthesis impairs MAPK activation and growth of tumors (Weiss et al., 2018). Future studies are required to further understand the functions and mechanisms of itaconate in TMEs.

Other metabolites including amino acids and fatty acids can be linked to intracellular signaling and gene expression. For example, amino acid abundance affects sensors like mTORC1 and GCN2-ATF4 pathways (Goberdhan et al., 2016; Saxton and Sabatini, 2017; Ye et al., 2010), and/or impact downstream metabolites like α-KG, ac-CoA and SAM that interplay with epigenetics (Haas et al., 2016; Liu and Wellen, 2020). Fatty acids are known to directly regulate gene expression by binding to specific transcription factors (Jump et al., 2013). Therefore, these metabolites can also function as signaling molecules through multiple mechanisms, depending on specific TME contexts.

Metabolic Adaptations and Communications in Metastatic TMEs

Metastasis is the leading cause of death in cancer patients (Mehlen and Puisieux, 2006; Valastyan and Weinberg, 2011), yet metabolic mechanisms controlling this highly inefficient and multi-step process are only beginning to be explored. Due to organic-specific features including structural/cellular composition, metabolic/immune profiles and nutrient availability, primary tumor cells accordingly undergo metabolic adaptions and communications in circulation and metastatic TMEs (Doglioni et al., 2019; Elia et al., 2018; Schild et al., 2018) (Figure 5).

Figure 5. Organ-Specific Metabolic Communication in Metastatic TME.

Brain metastases use acetate, glutamine, branched-chain amino acids (BCAAs) and polyunsaturated fatty acids (FAs) from astrocytes to fuel growth. In addition to glucose, liver metastases produce creatine kinase, brain-type B (CKB) to phosphorylate hepatocyte-derived creatine, and then import phosphocreatine for energy metabolism. Lung metastases preferentially metabolize pyruvate to create a collagen-rich niche which potentially provides proline (Pro) for metastatic growth. Lung metastases also release miR-122-containing exosomes to limit glucose access to fibroblasts. Bone metastases release serine (Ser) and lactate (Lac) to promote osteoclast differentiation and create an osteolytic niche for tumor growth. Adipocytes within the omentum potentially provide fatty acids (FAs) to fuel metastatic growth. Bile acids (BAs) accumulation in the lymph node facilitates metastatic tumor growth by activating the YAP-fatty acid oxidization (FAO) axis.

Metastatic tumor cells must survive the circulation, where oxygen/nutrient availability and cell-matrix interaction are significantly altered. Indeed, circulating melanoma cells experience significant oxidative stress that is a profound barrier for successful metastasis (Piskounova et al., 2015). To shed light on how circulating tumor cells (CTCs) survive, stress conditions have been modeled in vitro and several important metabolic mechanisms have been uncovered. Matrix-deprived metastatic fibrosarcoma cells prevent ROS-induced anoikis by promoting activating transcription factor 4 (ATF4) and nuclear factor-erythroid 2-related factor 2 (NRF2)-induced heme oxygenase 1 (HO-1) expression (Dey et al., 2015). Importantly, stabilization of BACH1 downstream of the NRF2-HO1 axis triggers metabolic remodeling and facilitates glycolysis-dependent metastasis of lung cancer cells (Lignitto et al., 2019; Wiel et al., 2019). The energy sensor AMP-activated protein kinase (AMPK) is activated in matrix-detached cancer cells to maintain NAPDH homeostasis, by decreasing NADPH consumption in fatty-acid synthesis and increasing NADPH generation via fatty-acid oxidation (FAO) (Jeon et al., 2012). Moreover, cancer cells combine mitochondrial oxidative decarboxylation with cytosolic reductive carboxylation to maintain redox balance and anchorage-independent growth (Jiang et al., 2016). Interestingly, matrix detachment allows cancer cells to form cell clusters and “hypoxia”, driving HIF-1α-mediated mitophagy to clear damaged mitochondria and limit ROS (Labuschagne et al., 2019). Additionally, CTCs increase PGC1α-dependent mitochondrial biogenesis to maintain redox and energetic homeostasis (Lebleu et al., 2014). Together, CTCs can employ several favorable metabolic programs to survive for distal metastasis.

Due to the brain’s unique metabolic rewiring capacity in response to varying nutrient availability, brain metastases display a remarkable metabolic adaptation by utilizing locally abundant non-glucose substrates, including acetate (Mashimo et al., 2014), glutamine and BCAAs (Chen et al., 2015). Additionally, polyunsaturated fatty acids released from astrocytes can activate peroxisome proliferator-activated receptor γ (PPARγ) and enhance brain metastatic cancer cell proliferation, while systemic administration of PPARγ antagonists significantly reduces brain metastasis in vivo (Zou et al., 2019). Compared with patient-matched extracranial metastases, melanoma brain metastases exhibit enriched OXPHOS that can be therapeutically targeted with IACS-010759, a new complex I inhibitor (Fischer et al., 2019; Molina et al., 2018).

The liver is an integral metabolic organ for whole body energy balance, where metabolic zonation confers metastatic breast tumors the ability to preferentially engage in glycolytic metabolism in hypoxic regions. Here, breast cancer cells upregulate HIF-1α-dependent pyruvate dehydrogenase kinase 1 (PDK1) to limit mitochondrial activity and favor a glycolytic phenotype (Dupuy et al., 2015). Metastatic colorectal cancer cells secret creatine kinase, brain-type B (CKB) to phosphorylate extracellular creatine released from hepatocytes, and then import phosphocreatine to replenish intracellular ATP pools (Loo et al., 2015).

As primary respiration organ in humans, the lungs are exposed to oxidative stress from high levels of oxygen and toxic compounds. Accordingly, lung metastases metabolically withstand oxidative damage, for example, through enhanced mitochondrial biogenesis by PGC1α upregulation (Andrzejewski et al., 2017), or through enhanced FAO by overexpressing aldo-keto reductase AKR1B10 (van Weverwijk et al., 2019). Consistently, fluorouracil-labeled RNA sequencing (Flura-seq) was developed to uncover specific oxidative stress and anti-oxidant gene signatures in murine breast cancer-derived lung micrometastases (Basnet et al., 2019). Intriguingly, due to the particular pyruvate availability in the lungs, metastatic tumors upregulate pyruvate carboxylase (PC) to utilize pyruvate over glutamine to fuel the TCA cycle (Christen et al., 2016); pyruvate metabolism through prolyl-4-hydroxylase (P4HA) creates a collagen-rich niche to support breast cancer-derived lung metastasis (Elia et al., 2019). Since proline catabolism critically supports lung metastases formation (Elia et al., 2017), this collagen-rich niche may also contribute to proline homeostasis in tumors. Moreover, breast tumors also release miR-122-containing exosomes to suppress glucose metabolism in resident lung fibroblasts, increasing the glucose availability for metastatic seeding (Fong et al., 2015).

Bone is a common metastatic site for prostate and breast cancers, where the latter release serine and lactate to promote differentiation and metabolic competency of osteoclasts and form osteolytic metastatic niches (Lemma et al., 2017). Ovarian cancer preferentially metastasizes to omentum, a metabolic organ mainly composed of adipocytes. Co-cultured ovarian cancer cells stimulate lipolysis in adipocytes to produce lipids that are utilized as energy source by cancer cells (Ladanyi et al., 2018; Nieman et al., 2011), suggesting a metabolic mechanism for omental metastatic growth. Finally, metastatic tumor growth in lymph nodes (LNs) relies on FAO driven by transcriptional coactivator yes-associated protein (YAP), which is activated by bile acids accumulated in LNs. Importantly, pharmacological inhibition of FAO suppresses LN metastasis in mice (Lee et al., 2019).

Targeting Metabolic Communications for Cancer Treatment

To exploit metabolic communications for therapeutic intervention, much attention has focused on cancer and immune cells, as deregulated cancer metabolism shapes TMEs that impose metabolic stress on tumor-infiltrating lymphocytes, resulting in local immunosuppression and defective tumor surveillance. Therefore, multiple strategies have been avidly explored to target metabolic pathways that enhance antitumor immunity, which are subjects of several recent reviews (Buck et al., 2017; Li et al., 2019; Reina-Campos et al., 2017). Generally, these include targeting cancer-specific metabolic vulnerabilities, limiting immunosuppression and boosting effector functions by enhancing metabolic fitness in tumor specific immune cells. Ideally, a metabolic approach would synergize with immunotherapy to selectively and sustainably eliminate tumor cells. While significant progress has been made in this field, it should be noted that tumor metabolism is dictated by both intrinsic and extrinsic factors (Faubert et al., 2020). Tissue of origin and local microenvironment influence metabolic fitness and rewiring in cancer (Biancur et al., 2017; Davidson et al., 2016; Lee et al., 2020b; Mayers et al., 2016), while certain genetic or epigenetic modifiers and tumor plasticity confer differential sensitivity to metabolic interventions (Li et al., 2015; Lissanu Deribe et al., 2018; Shackelford et al., 2013). Therefore, in the future it’s necessary to stratify certain types of cancer into molecular and metabolic subtypes that are most suited to precision treatment. Along with this, cancer cells within single tumors undergo metabolic remodeling during progression and metastasis, as well as in response to treatment, which results in metabolic reprograming in TMEs, leading to different immune microenvironments. Similarly, immune cells can also develop strategies to adapt to treatments that target metabolism. Understanding these compensatory mechanisms will provide helpful insights to overcome potential resistance to metabolic interventions. Overall, targeting metabolic communication in TMEs should consider tumor type, stage, location and potential compensatory mechanisms.

Concluding Remarks

We present here recent findings on how tumor cells metabolically communicate with stromal cells predominantly in the primary TME. In comparison, metabolic adaptions and communications in the metastatic TME are less clear, partly attributed to metabolic plasticity in metastatic tumors together with complexity of organ-specific local microenvironment. Consequently, it will be imperative to dissect stage-specific metabolic communications along the metastatic cascade, such as those dictating CTC metabolism and survival (Tasdogan et al., 2020), pre-metastatic niche formation (Doglioni et al., 2019), and maintenance of established macro-metastatic microenvironment. Given the complexity of TMEs, the experimental models should not operate in isolation and instead should include both tumor cells and stromal cells, preferentially under conditions that can recapitulate certain metabolic aspects of TME, such as hypoxia and nutrient limitation. For this, experimental systems such as three-dimensional culture, organoid cultures, and co-cultures under stress conditions (hypoxia, low serum, low nutrients), will be helpful. Finally, the mechanisms identified from in vitro systems must be validated in clinically-relevant in vivo models. As we continue to explore the metabolic communications, particularly in the metastatic TME, we will be able to identify more actionable metabolic vulnerabilities, discover metabolic drugs with improved specificity and efficacy, and more precisely target metabolic communications as a single agent or combination therapy. Ultimately, these preclinical findings will be evaluated in clinic and benefit the patients.