Advances into understanding metabolites as signaling molecules in cancer progression

Abstract

Recent years have seen a great expansion in our knowledge of the roles that metabolites play in cellular signaling. Structural data has provided crucial insights into mechanisms through which amino acids are sensed. New nutrient-coupled protein and RNA modifications have been identified and characterized. A growing list of functions have been ascribed to metabolic regulation of modifications such as acetylation, methylation, and glycosylation. A current challenge lies in developing an integrated understanding of the roles that metabolic signaling mechanisms play in physiology and disease, which will inform the design of strategies to target such mechanisms. In this brief article, we review recent advances in metabolic signaling through post-translational modification during cancer progression, in order to provide a framework for understanding signaling roles of metabolites in the context of cancer biology and illuminate areas for future investigation.

Introduction

Cellular metabolism functions to break down nutrients to generate smaller molecules and release energy (catabolism), as well as to synthesize larger molecules from smaller building blocks (anabolism). Certain metabolites generated during these processes can also exert signaling functions to modulate cellular activities in a manner responsive to nutritional status or metabolic shifts. In this brief article, our goal is to provide an update on advances from the last two years in our understanding of the roles of metabolites in signaling via post-translational modification, with a focus on cancer biology. Other diverse aspects of nutrient signaling have been covered in recent reviews [1–8]. Throughout the article, we highlight developing concepts that are reinforced by recent findings, including that: 1) nuclear metabolite production plays crucial roles in chromatin regulation; 2) systemic metabolism modulates tumor cell metabolic signaling; 3) acetyl-CoA and α-ketoglutarate (αKG) have emerged as important signaling molecules involved in multiple processes throughout tumorigenesis; and 4) metabolic signaling within the tumor microenvironment can impact therapeutic responses. We first examine recent findings in metabolic signaling during tumor initiation and primary tumor growth, and then in cancer progression and metastasis. We also discuss metabolic signaling in other cell types in the tumor microenvironment, and finally, the role of nuclear metabolite production in chromatin regulation. We conclude with perspectives on emerging concepts in metabolic signaling and how they can be explored in the context of cancer.

Metabolic Signaling Mechanisms in Tumor Initiation and Growth

The multistep process of cancer development begins with tumor initiation, in which cells acquire the requisite traits that allow them to form tumors, involving oncogene activation and tumor suppressor silencing. As cellular metabolism is remodeled during tumorigenesis to facilitate survival and anabolic growth [9,10], metabolites also exert signaling functions that contribute to tumor formation. A canonical example is seen in cancer-associated mutations in the genes encoding the metabolic enzymes isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2), which result in aberrant production of the oncometabolite, D-2-hydroxyglutarate (D-2-HG) [11,12]. D-2-HG acts as a competitive inhibitor of oxoglutarate-dependent dioxygenases [13], including epigenetic enzymes such as TET methylcytosine oxidases and JmjC histone demethylases. IDH mutations result in genome-wide epigenetic dysregulation, linked to malignant transformation [14]. Beyond D-2-HG, several other metabolites, including αKG, acetyl-CoA, and UDP-GlcNAc, have emerged as mediators of early tumorigenesis, at least in part via signaling mechanisms.

By way of its regulation of oxoglutarate-dependent dioxygenases, the ratio of αKG to succinate, a product and inhibitor of these enzymes, has emerged as a key node for modulating cell phenotypes, including stem cell pluripotency and macrophage polarization [14,15]. αKG has also been recently linked to p53-mediated tumor suppression. Using a mouse model of pancreatic cancer with a doxycycline-inducible shRNA targeting p53, p53 re-activation in tumors was found to induce differentiation and suppress tumor growth, at least in part via control of αKG accumulation [15]. Restoration of p53 function boosted the αKG:succinate ratio, and genetic and pharmacological augmentation of this ratio was sufficient to promote tumor differentiation. Conversely, high levels of succinate interfered with p53-dependent tumor suppression. Accumulation of 5-hydroxymethylcytosine (5hmc), indicative of elevated TET activity, correlated with the αKG:succinate ratio (Fig. 1). Though correlative, promotion of TET activity is a plausible mechanism of tumor suppression, since reduced 5hmc as a consequence of impaired TET function is frequently observed in cancers. TET enzymes are also thought to be a crucial target of both D-2-HG and its stereoisomer L-2-HG [13,16–18], as well as of succinate and fumarate, which accumulate in SDH and FH deficient tumors [1,19] (Fig. 1). Accumulation of L-2-HG in cancer cells can result from promiscuous activity of lactate dehydrogenase and malate dehydrogenase under hypoxia [20,21] or from loss of L-2-HG dehydrogenase function [22].

Figure 1. Metabolites regulate TET methylcytosine oxidases to impact tumor formation.

Metabolically-linked alterations in TET activity have been identified as consequences of both oncogene activation and tumor suppressor silencing. TET enzymes, which belong to the family of 2-oxoglutarate-dependent dioxygenases that rely on αKG as a cosubstrate, mediate the oxidation of 5-methylcytosine (5mC) to yield 5-hydroxymethylcytosine (5hmC) and subsequently 5-fC and 5-caC, to facilitate removal of DNA methylation. Oncogenic mutations in isocitrate dehydrogenase (IDH) lead to neomorphic activity and production of the oncometabolite D-2-hydroxyglutarate (D-2-HG). The stereoisomer L-2-HG can also be generated under hypoxia or accumulate upon loss of L-2-HG dehydrogenase activity. Succinate dehydrogenase (SDH) and fumarate hydratase (FH) loss of function drive accumulation of succinate and fumarate, respectively. 2-HG, succinate, and fumarate all inhibit the activity of TET enzymes. Conversely, increasing the αKG: succinate ratio, which has been identified as a component of the p53 tumor suppressive program[15], is associated with promotion of TET activity.

Other metabolites, including acetyl-CoA and pyruvate, have also been linked to tumor initiation. Acetyl-CoA was found to play a role in promoting pancreatic tumorigenesis [23]. Acetyl-CoA and global histone acetylation levels are elevated in murine pancreatic acinar cells harboring KRAS mutations, and genetic deletion of Acly suppressed acinar-to-ductal metaplasia in vitro and impaired tumor formation in vivo (Fig. 2). Pyruvate metabolism is intimately coupled to stem cell function [24,25], and inactivation of the mitochondrial pyruvate carrier was found to enhance intestinal tumor formation and promote gene expression signatures associated with stemness in adenomas[26]. Thus, metabolic reprogramming can facilitate tumor development, linked to regulation of chromatin modification and gene expression.

Figure 2. Acetyl-CoA mediates signaling functions throughout cancer progression.

The signaling functions of metabolites have been implicated in multiple steps throughout tumorigenesis, including tumor initiation, metastasis, and in interaction with other cell types (immune, cancer-associated stroma). Here, acetyl-CoA is shown as a representative example. Nuclear-cytosolic acetyl-CoA is generated from mitochondria-derived citrate via ATP-citrate lyase (ACLY) or acetate via acyl-CoA synthetase short-chain family member 2 (ACSS2). This pool of acetyl-CoA feeds into multiple biosynthetic pathways and also serves as the sole substrate for acetylation reactions, mediated by lysine acetyl-transferases (KATs). ACLY-dependent elevations in acetyl-CoA and global histone acetylation were observed in KRAS mutant pancreatic acinar cells[23]. Accumulations in acetyl-CoA are also implicated in regulation of metastasis. In two models of cancer, inhibition of acetyl-CoA consuming enzymes acyl-CoA thioester 12 (ACOT12) and acetyl-CoA carboxylase (ACC) increased the expression or activity of EMT transcription factors Twist2 and Smad2 via histone or transcription factor acetylation, respectively[33,34]. Lastly, control of acetyl-CoA production in T cells, partly under the control of the transcription factor Bhlhe40, and histone acetylation are implicated in regulating IFNγ production and antitumor immunity[54,55,81].

Beyond the roles of cell autonomous metabolic alterations in cancer, epidemiological data identifies obesity as a risk factor for several cancers [27,28]. New evidence in mice suggests that hyperglycemia may increase DNA damage, resulting in KRAS mutations, in pancreatic cells [29]. This was attributed to a reduction in deoxyribonucleotides (dNTPs) due to glucose-sensitive O-GlcNAcylation and subsequent destabilization of the ribonucleotide reductase (RNR) complex, which generates dNTPs from their ribonucleotide counterparts. The O-GlcNAc modification depends on synthesis of UDP-N-acetyl-glucosamine (UDP-GlcNAc) through the hexosamine biosynthesis pathway and the subsequent transfer to the substrate protein by O-GlcNAc transferase (OGT) [30]. Although the determinants of OGT substrate specificity remain poorly understood, O-GlcNAcylation of specific enzymes has been proposed as a physiological mechanism to fine-tune metabolic regulatory pathways [31]. However, pathologic elevations of this modification, as observed under diabetic hyperglycemia, can have detrimental consequences, including the promotion of carcinogenesis. Interestingly, insulin stimulation of breast cancer cells was also associated with increased evidence of DNA damage [32]. Together, these emerging data suggest that increased mutational burden may be a contributing factor to the increased cancer incidence associated with hyperglycemia and hyperinsulinemia.

Metabolites Regulate Metastasis

Metabolites in the Transcriptional Regulation of EMT

Metastasis is a complex process that requires cells to be equipped with the ability to disseminate from primary tumors, survive in and extravasate from circulation, and ultimately colonize in distant sites. New studies demonstrate that the signaling roles of metabolites in metastasis may be multiple and context-dependent.

Two recent studies highlighted mechanistically distinct roles for acetyl-CoA in facilitating the transcriptional regulation of epithelial-mesenchymal transition (EMT) (Fig. 2). In investigating the adipokine leptin as a putative link between obesity and breast cancer pathogenesis, leptin-mediated activation of EMT was found to depend on AMPK-dependent phosphorylation of acetyl-CoA carboxylase (ACC), the rate limiting enzyme in fatty acid synthesis [33]. This inhibition of ACC led to an accumulation of acetyl-CoA and enhanced acetylation of the transcription factor Smad2, promoting invasiveness and activating EMT. A second mechanism through which acetyl-CoA promotes EMT was reported in hepatocellular carcinoma [34]. ACOT12, a predominantly cytosolic acetyl-CoA hydrolase which is expressed in healthy liver [35], is downregulated in tumors. Acetyl-CoA levels increased as a consequence of ACOT12 suppression, accompanied by elevated histone H3 acetylation at the Twist2 locus and expression of the encoded EMT transcription factor. The mechanisms by which such specificity in gene regulation is achieved remain to be further elucidated, although both of these studies align with a developing model that nuclear-cytosolic acetyl-CoA abundance modulates expression of specific sets of genes by controlling the levels or function of particular transcription factors [1,36,37].

Metabolites Define the Metastatic Niche

The final steps in the metastatic cascade involve colonization of cancer cells in a distant organ, which requires both remodeling of the extracellular matrix (ECM) and reprograming of other cell types within the local microenvironment [38]. Recent advances have revealed that nutrient availability and metabolic signaling may play a role in both of these processes, and thus help to define the metastatic niche.

In breast cancer cells, pyruvate uptake and metabolism can facilitate ECM remodeling at metastatic lung lesions by affecting collagen hydroxylation [39]. Stable collagen synthesis involves hydroxylation of proline residues on procollagen strands by the αKG-dependent collagen prolyl-4-hydroxylase (P4HA). Inhibition of pyruvate uptake in cancer cells reduced collagen hydroxylation, as well as lung metastases in vivo [39]. This effect was attributed to a decrease in production of αKG through alanine aminotransferase (ALT), which converts pyruvate and glutamate to αKG and alanine (Fig. 3). Remarkably, in vivo supplementation of αKG was sufficient to rescue collagen hydroxylation and metastatic burden in the lung. Notably, HIF1α, which is also prolyl hydroxylated via oxoglutarate-dependent enzymes to target HIF for degradation [40], transcriptionally regulates P4HA1 expression [41]. The activity of P4HA1 in consuming αKG for collagen hydroxylation has been proposed to limit HIF hydroxylation, resulting in its stabilization, and thereby promoting stemness and chemoresistance in breast cancer [42]. Thus, the availability of nutrients such as pyruvate at sites of metastasis may dictate regulation of and interplay between αKG-dependent enzymes, impacting ECM remodeling and cancer progression (Fig. 3). The notion that enzymes that consume αKG can restrict its availability to other enzymes is also supported by evidence in acute myeloid leukemia that overexpression of branched chain amino acid transaminase 1 (BCAT1), a reversible enzyme which transfers the α-amino group from BCAAs to αKG, depletes αKG pools, resulting in HIF1α stabilization and reduced TET activity [43]. As discussed, growing evidence supports the notion that metabolic signaling contributes to metastasis, though an outstanding question is whether distant signals from the primary tumor are involved in metabolically priming the metastatic niche. Although relatively little is known about this, glucose availability at sites of metastases has been proposed to be affected by secreted factors from the primary tumor [44].

Figure 3. Nutrient availability and αKG regulate aspects of metastatic biology in breast Cancer.

In metastatic breast cancer cells, αKG metabolism participates in extracellular matrix remodeling (ECM), stemness and chemoresistance. Pyruvate uptake promotes production of αKG through alanine aminotransferase (ALT). αKG is required by collagen prolyl-4-hydroxylase (P4HA) to hydroxylate procollagen for stable collagen synthesis in the process of ECM remodeling[39]. This shunting of αKG toward P4HA1 also limits HIF hydroxylation, resulting in stabilization and transcription of genes involved in chemoresistance and stemness[42].

Cancer-associated stromal cells also play a critical role in defining the metastatic niche. Proteomic analysis of matched primary vs metastatic ovarian cancer samples revealed a role for the metabolic enzyme nicotinamide N-methyltransferase (NNMT) in regulating gene expression in cancer associated fibroblasts (CAFs) [45]. NNMT, which is upregulated in the stroma of metastatic sites, depletes the universal methyl donor, S-adenosyl methionine (SAM), which it metabolizes to S-adenosyl homocysteine and a metabolically inactive product, 1-methylnicotinamide. High NNMT in cancer cells suppresses histone and DNA methylation, promoting altered gene expression and in glioblastoma, stem cell maintenance and tumor growth [46,47] Likewise, in CAFs, NNMT expression promoted hypomethylation of histones and DNA, as well as expression of cytokines and extracellular matrix components linked to metastasis [45]. Thus, metabolic signaling mechanisms can contribute to metastasis not only through direct roles in cancer cells but also within other cell types present in tumors.

Epigenetic Modifications Link Metabolites and Immune Function

Metabolic signals contribute to immune cell function and impact anticancer immunity, leading to efforts to target immune cell metabolism and to leverage cellular metabolism to optimize cancer immunotherapy [48]. As such, a critical issue in considering the roles of metabolic signals in cancer biology is how they play out within and between different cell types in the tumor microenvironment.

Acetyl-CoA metabolism and its links with histone acetylation represents a crucial node in T cell functioning. Building on prior work demonstrating roles for glucose and acetate metabolism and metabolic control of histone acetylation in regulating IFNγ production in T cells [49,50], the transcription factor Bhlhe40 was identified as a transcriptional regulator of mitochondrial metabolism in CD8+ T cells. Bhlhe40 was found to promote histone acetylation and IFNγ production, presumably via mitochondrial citrate export and ACLY-dependent acetyl-CoA production. Consistently, in a separate study, both the malate-aspartate shuttle and citrate export for acetyl-CoA synthesis were found to promote IFNγ production in CD4+ T_H1 cells [51]. Notably, acetate, which can bypass the requirement for mitochondrial metabolism for acetyl-CoA production [52,53], could augment the production of IFNγ [54–56] in a manner that was further enhanced by histone deacetylase inhibition [55]. Conversely, silencing of ACSS2 in T cells suppressed IFNγ levels and impaired tumor-clearing [54]. ACSS2 has been proposed as a cancer therapeutic target [57], and its role in promoting anti-cancer immunity may be a caveat to targeting this enzyme. Interestingly, these pathways also appear to be relevant for maintenance of T cell stemness. High extracellular potassium, which is often present in tumors due to release of necrotic cell contents, impairs nutrient transport into T cells and results in a state of functional starvation that includes low acetyl-CoA and histone acetylation levels. Although these conditions limited expression of effector genes such as IFNγ, they also preserved T cell stemness and capacity for anti-tumor response [56], highlighting a nuanced role for nutrient regulation of T cell function in cancer.

Metabolic regulation of αKG-dependent enzymes is also an important determinant of T cell function. Glutaminase inhibition suppresses αKG levels in T cells, increasing global histone methylation and altering gene expression to promote T_H1 specification [58]. T_H17 differentiation was simultaneously impeded [58]. Analogously, complex III inhibition raised levels of succinate and 2-HG, promoting hypermethylation and interfering with the suppressive function of regulatory T cells [59]. Encouragingly, in vivo, inhibition of glutamine metabolism was found to impair cancer cell metabolic programs that promote an immunosuppressive environment, while also enhancing antitumor immune responses [60]. Glutamine metabolic blockade was linked to an upregulation of pyruvate carboxylase-dependent anaplerosis and acetate utilization in CD8+ T cells but not in cancer cells [60], pointing to the potential to optimize metabolic therapeutic strategies taking into account effects on both cancer cells and tumor-associated immune cells.

Compartmentalized Metabolism and Chromatin Modifications in Cancer

As links between metabolism and chromatin modification continue to emerge in the context of cancer progression, a central question has been the role of metabolic compartmentalization [61]. The role of compartmentalization is an obvious consideration for TCA cycle metabolites with nuclear signaling functions, such as αKG and succinate. Evidence has also emerged that nucleus and cytosol may be at least partially distinct metabolic compartments [62], even though small molecules can diffuse through nuclear pores. Empirical evidence for the nucleus as metabolically distinct from cytosol came from study of NAD+-dependent poly ADP-ribosylation (PARylation) during adipocyte differentiation. While whole cell NAD+ levels remained unchanged, a reduction in nuclear NAD+ was necessary for induction of an adipogenic transcription program [63]. Compartmentalized metabolite measurements were achieved using fluorescent biosensors, allowing monitoring of NAD+ levels in living cells [63].

Prior work has established the nuclear localization of multiple metabolic enzymes that produce acetyl-CoA (ACLY, ACSS2, PDC) [64–66] or are components of methionine or folate cycles, involved in SAM production (MAT2A, MTHFD1, SESAME complex) [67–69]. Recent studies have provided new insight into the potential pathways that may facilitate acetyl-CoA production within the nucleus. One study which investigated how histone acetyltransferase 1 (HAT1) coordinates growth-factor and nutrient-stimulated histone production proposed that the deacetylation of newly synthesized H4 histones after their nuclear translocation provides a local source of acetate for acetylation of promoter regions, including those encoding histone H4 genes [70] (Fig. 4). Another study investigating the functional significance of nuclear glycogen, which had been observed histologically in lung cancer cells, proposed that nuclear glycogen metabolism represents a nuclear source of acetyl-CoA [71]. Leveraging an isolated nuclei system, glycogen synthesis from glucose-6-phosphate and breakdown to pyruvate were found to occur within nuclei, with nuclear glycogenolysis contributing acetyl groups for histone acetylation (Fig. 4). Although this study did not document the mechanism of acetyl-CoA production from pyruvate, presumably this would require either pyruvate conversion to acetyl-CoA by a nuclear PDC [72] or generation of acetate from pyruvate [73] and subsequent acetyl-CoA production by ACSS2. The advantages of nuclear glycogen synthesis and breakdown to supply acetyl-CoA for histone acetylation versus more straightforward synthesis from citrate or acetate are not yet clear, though could potentially serve to maintain histone acetylation in the face of fluctuating nutrient availability. Nuclear glycogen accumulation may also provide other unknown benefits to the cancer cell.

Figure 4. Novel mechanisms toward the production of nuclear acetyl-CoA.

ACLY, ACSS2, and PDC have all been found within the nucleus for local and context-specific generation of acetyl-CoA. Breakdown of nuclear glycogen into pyruvate has been identified as a novel source of acetyl-CoA for histone acetylation [71]. Nuclear-generated pyruvate could in principle contribute to histone acetylation through PDC-dependent production of acetyl-CoA [72] or through non-enzymatic conversion of pyruvate to acetate and subsequent acetyl-CoA generation by ACSS2[73]. Nuclear transport and deacetylation of newly synthesized H4 histones has also been proposed as a source of acetate for acetylation of histones at promoters of histone H4 genes [70].

Conclusions and Emerging Questions

Within the past two years, several intriguing themes have emerged from the advances surrounding the signaling function of metabolites in cancer, as we discuss above, and the findings have equally raised many questions moving forward. One key question surrounds sources of particular metabolites in different contexts, both at the cellular and systemic level. For example, recent evidence points to ethanol metabolism as a source of acetate in the regulation of brain histone acetylation [74]. In addition to metabolites in circulation and produced intracellularly, metabolic crosstalk between cells within the tumor microenvironment likely participates in metabolic signaling between cancer cells, immune cells, and fibroblasts. A second important question surrounds the roles of the ever-growing list of acylations and other post-translational modifications that are metabolically sensitive [75]. For example, lysine lactylation, derived from lactate, was recently reported and may be mediated either enzymatically, by one or more acyltransferase, or non-enzymatically from lactoyl-glutathione [76,77]. Since lactate is produced abundantly and can be both exported and imported by cells, the modification offers a potential mechanism of crosstalk between cells in the tumor microenvironment. Additionally, newly identified cysteine modifications that are sensitive to the abundance of certain electrophilic metabolites have been described as metabolic signaling mediators of anti-oxidant and anti-inflammatory responses [78–80]. Ongoing research into the roles of metabolites as signaling molecules will continue to enhance our molecular understanding of how metabolites interface with signaling pathways and with chromatin to tune cellular responses in accordance with metabolic state.