Sirtuins and the metabolic hurdles in cancer

Abstract

The metabolic demands of cancer cannot be met by normal cell metabolism. Cancer cells undergo dramatic alteration of metabolic pathways in a process called reprogramming, characterized by increased nutrient uptake and re-purposing of these fuels for biosynthetic, bioenergetic or signaling pathways. Partitioning carbon sources toward growth and away from ATP production necessitates other means of generating energy for biosynthetic reactions. Additionally, cancer cell adaptations frequently leads to increased production of reactive oxygen species and lactic acid- metabolites which can be beneficial to cancer growth but also are potentially toxic and must be appropriately cleared. Sirtuins are a family of deacylases and ADP-ribosyltransferases with clear links to the regulation of cancer metabolism. Through their unique ability to integrate cellular stress and nutrient status with coordination of metabolic outputs, sirtuins are well poised to play pivotal roles in tumor metabolism. Here, we review the multi-faceted duties of sirtuins in tackling the metabolic hurdles in cancer. We focus on both beneficial and adverse effects of sirtuins in the regulation of energetic, biosynthetic and toxicity barriers faced by cancer cells.

Introduction

It is more relevant than ever to understand how metabolism influences tumor growth. Metabolic dependencies of cancer cells are increasingly being realized as promising candidates for therapeutic interventions in cancer[1–3]. A vast number of studies validate the notion that metabolic dysfunction is not just a consequence of cancer growth but rather a driving factor in tumor progression[4, 5] Indeed, altered metabolism enables tumor cells to fuel a number of processes, such as amassing a pool of biosynthetic precursors, constructing signaling molecules, generating metabolites for post-translational or epigenetic modifications, and maintaining pH and redox homeostasis[6, 7]. Furthermore, metabolic dysfunction has positioned itself at the forefront of cancer research with the recognition of the undeniable connection between increased cancer incidence and the background of obesity and metabolic dysfunction, pathologies that have reached epidemic proportions in the United States and much of the world[8–11]. It is critical to fully understand the altered metabolism of tumor cells and the pathways that might promote or oppose this metabolic dysfunction.

Sirtuins are a highly conserved family of regulatory enzymes that are well poised to play pivotal roles in tumor metabolism. The seven mammalian sirtuins (SIRT1-7) have the unique ability to integrate cellular stress response with coordination of metabolic fitness and homeostasis[12, 13]. The role of sirtuins as post-translational modifying enzymes may have originated to allow survival under stress and low nutrient conditions, and many of these roles have now been linked to growth regulation in the harsh conditions experienced by cancer cells[14–16]. In recent years, a number of studies have shown that sirtuins not only coordinate cancer cell growth and survival, but also the metabolic state of a tumor[17, 18]. There is growing interest in pinpointing metabolic regulatory nodes that can be targeted in cancer treatment and determining whether sirtuins specifically may be promising biomarkers or therapeutic targets in cancer.

Here, we review the roles of sirtuins in the metabolic hurdles of cancer. We will first overview sirtuin enzymatic activity and links to cancer incidence and severity. Then we will discuss sirtuin-mediated control of metabolic pathways with a focus on glucose metabolism, TCA cycle anaplerosis and reactive oxygen species defense in cancer. Sirtuins coordinate many other processes linked to cancer including DNA repair, metastasis, apoptosis and translation. For reviews more comprehensively assessing these roles, we refer the reader to other sources[13, 14, 19].

Sirtuin activity is linked to the metabolic state

Sirtuin enzymatic activity

SIRT1-7 are a family of deacylase and ADP-ribosyltransferases that share a conserved catalytic core domain but vary in subcellular localization and preferred substrates[20]. The differences between sirtuins lead to variations in the ultimate metabolic effect that is coordinated by each sirtuin[15]. SIRT1, 6 and 7 are primarily nuclear and regulate transcription factors and histone modifications to coordinate gene expression programs that can direct the cellular metabolic state[21]. Cytosolic functions of SIRT1 have also been identified. SIRT2 is largely cytosolic and coordinates microtubule dynamics as well as the activity of transcription factors residing outside the nucleus[22, 23]. Localization of SIRT3, 4 and 5 in the mitochondrial matrix enables these sirtuins to directly alter the activity of many metabolic enzymes[24].

Sirtuins catalyze NAD⁺-dependent deacylation or ADP-ribosylation reactions with varying degrees of substrate versatility[20]. Although most well-studied as lysine deacetylases, certain sirtuins can also remove other acyl modifications from lysine residues including propionyl, butyryl, malonyl, succinyl and the lengthy fatty-acid derived myristoyl and palmitoyl groups among others[25]. SIRT5, for example, is a strong desuccinylase, and SIRT4 was recently reported to function as a lipoamidase by removing lipoyl or biotinyl modifications from lysine residues[26]. Through these processes, sirtuins have been shown to alter substrate activity, localization, stability and protein-protein interactions[14].

Regardless of which type of substrate moiety is modified by sirtuins, a similar NAD⁺-dependent reaction mechanism proceeds. During sirtuin-catalyzed deacylation (Figure 1A), nicotinamide adenine dinucleotide (NAD⁺) nucleophilically attacks an acyl group of a substrate lysine. The resulting intermediate is cleaved to form 2′-O-acyl-ADP ribose (OAADPR) and nicotinamide, and the acyl group is removed from the lysine residue in the process. In sirtuin-catalyzed ADP-ribosylation (Figure 1B), NAD⁺ similarly attacks a substrate residue, typically an arginine residue[27], although cysteine residues are also candidate sites[28–30]. The ADP-ribose portion of NAD⁺ is transferred to the substrate residue, yielding nicotinamide as a side product[31, 32].

Figure 1. Sirtuin-catalyzed reactions.

(A) During deacylation reactions, sirtuins direct NAD⁺ to nucleophilically attack the acylated lysine residue, leading to removal of the acyl modification. NAD+ is cleaved in the process, forming nicotinamide and 2′-O-acyl ADP-ribose. Sirtuins can potentially remove diverse acyl modifications (inset) from lysine residues. (B) During ADP-ribosylation reactions, sirtuins use NAD⁺ to nucleophilically attack an arginine (shown) or cysteine residue. NAD⁺ is cleaved, resulting in release of nicotinamide and transfer of the ADP-ribose portion of NAD⁺ to the substrate residue.

The metabolic by-products of sirtuin activity have potential to accumulate and influence cellular biology. At high concentrations, nicotinamide inhibits sirtuin function[33]. Work by Grubisha et al showed that pools of OAADPR generated by certain sirtuins bind and inhibit the nonselective cation channel TRPM2[34]. TRPM2 is normally activated by oxidative or nitrative stress as well as by the drug puromycin, which directly targets TRPM2[34, 35]. Activation of this channel leads to an influx of Na²⁺ and Ca²⁺ into the cytosol to induce cell death. Decreased expression of SIRT2 and SIRT3 dampened TRPM2-mediated cell death in response to puromycin. Of note, studies of the direct roles of OAADPR as well as quantification of OAADPR have been impeded by the rapid hydrolytic degradation of this molecule to ADP-ribose in cells[36]. The role of sirtuin-derived metabolites is a promising and little studied area of sirtuin biology.

Sirtuins and NAD⁺-sensitivity

Sirtuins are unique sensors of the metabolic state because their NAD⁺-dependent enzymatic activity intrinsically couples their function with the metabolic status of the cell or organism[37–41]. According to the metabolic state of the cell, the ratio of NAD toggles between varying amounts of NAD⁺ and NADH[42]. NADH is a high energy, reduced form of NAD than can donate electrons to the electron transport chain. NAD⁺ is the lower energy, oxidized counterpart required to fuel glycolysis. When the cell uses oxidative metabolism, NADH generated by the TCA cycle and glycolysis donates electrons to complex 1 of the electron transport chain (ETC). This contributes to a proton gradient that will ultimately produce ATP. Upon electron transfer, NADH is oxidized back to NAD⁺. In highly glycolytic cells with low ETC function, NAD⁺ is alternatively regenerated from NADH via lactate dehydrogenase (LDH) activity in order to sustain glycolysis. NAD⁺ can also be synthesized de novo from tryptophan and vitamin B3 derivatives, or via salvage pathways using nicotinamide or nicotinamide riboside. Thus, the NAD⁺/NADH ratio is affected by the cellular metabolic state, and changes in this ratio have potential to impact sirtuin enzymatic activity.

Several studies have linked sirtuin activity with the organismal metabolic status and cellular NAD ratio. In many tissues, the NAD⁺/NADH ratio is low during nutrient excess and high during nutrient deprivation[43, 44]. For example, in skeletal muscle and white adipose, the NAD⁺/NADH ratio is elevated during calorie restriction[45]. Due to their dependency on NAD, it is not surprising that certain sirtuins are reported to have increased activity in response to high NAD⁺ levels[46]. For example, SIRT1 in skeletal muscle and brain is activated by exercise, fasting and calorie restriction[47–49]. In contrast, low NAD⁺ levels are observed with obesity as well as old age, two factors that confer increased risk for many cancers and are also linked to decreased sirtuin activity[50, 51]. Along these lines, growing evidence suggest loss of sirtuin function plays a role in obesity- and age-associated cancers[8, 52].

It is proposed that there are tissue type and cellular compartmentalization variations in NAD⁺ and NADH levels that may lead to distinct alterations of sirtuin activity in different contexts[42]. NAD⁺ is generated by biosynthetic reactions in the mitochondria, nucleus and cytosol[53]. Generally NAD⁺ is most abundant in mitochondria, particularly in highly metabolically active tissues such as cardiac myocytes, although the distribution and relative amounts of NAD⁺ in spatially distinct compartments varies across cell types[54]. Nuclear NAD⁺ can be depleted upon DNA damage when this molecule is used as a substrate for the PARP family of enzymes to activate DNA repair pathways[55]. Inhibiting PARPs elevates NAD⁺ levels, presumably in the nucleus, and accordingly only boosts activity of nuclear SIRT1 and not mitochondrial SIRT3[56]. Upon PARP activation, the mitochondrial permeability transition pore opens to allow flow of NAD⁺ from the mitochondria to the cytosol and nucleus in order to supply necessary NAD⁺ for further PARP function. Work by Yang et al shows that upregulation of NAD⁺ biosynthesis enables mitochondrial NAD⁺ levels to be maintained at physiological levels even though cytosolic and nuclear NAD⁺ pools are depleted upon genotoxic stress[38]. Promotion of NAD⁺ biosynthesis in the mitochondria was further demonstrated to be dependent on SIRT3 and SIRT4[38].

Other regulation of sirtuins

It is important to point out that sirtuin activity is not solely dependent on NAD⁺ levels. Transcriptional, posttranslational and allosteric regulation are all important physiological modulators of sirtuin activity[57]. A major negative regulator of SIRT1 with relevance in cancer is deleted in breast cancer-1 (DBC1), a nuclear protein that functions as a tumor suppressor and is homozygously deleted in some breast cancers[58]. DBC1 inhibits SIRT1 by directly binding its catalytic domain[58, 59]. This repressive interaction is induced upon DNA damage downstream of ATM, a key mediator of the DNA damage response that acts as a kinase targeting multiple proteins including DBC1[60, 61]. Phosphorylation of DBC1 creates an additional SIRT1 binding site. Strong repression of SIRT1 in this manner stimulates apoptosis, a proper cellular response to excessive genotoxic damage in many contexts. An AMP-activated protein kinase (AMPK)-dependent pathway disrupts the DBC1-SIRT1 interaction via phosphorylation of SIRT1[62, 63]. AMPK is a cellular sensor of a low energy state that acts as a bioenergetic rheostat by phosphorylating many metabolic proteins to restore energetic homeostasis[47]. Indeed AMPK has been linked to activation of SIRT1 under low energy conditions in multiple studies and via multiple proposed mechanisms[47, 64]. SIRT1 activity is also dramatically enhanced upon phosphorylation by additional kinases in response to adrenergic signaling and stresses such as DNA damage, microtubule disruption and heat or cold shock[65, 66]. Phosphorylation at one particular site, Thr522, boosts SIRT1 activity by promoting its monomeric rather than its oligomeric, aggregation-prone state[67].

In the case of SIRT4, activity likely does not parallel the cellular NAD⁺ level. SIRT4 plays a key role in inhibiting fat catabolism when mice are well fed, despite the low levels of NAD⁺ expected under this condition[68]. SIRT4 mRNA and protein are also more abundant in mouse tissues under fed versus fasted conditions[69, 70]. SIRT6 activity in vitro is induced by fatty acids[25]. It will be important for future studies to reveal further mechanisms by which sirtuin activity is regulated.

Sirtuins and connections to cancer

The associations between cancer metabolism and sirtuins often fall into one of two themes. First, loss of sirtuin activity may result in increased susceptibility to cancer onset. Second, an already established tumor that expresses high levels of some sirtuins may have survival advantages including resistance to chemotherapeutics.

Loss of sirtuin activity has been shown to contribute to cancer onset in some cases. The link between sirtuin loss and tumor emergence is evidenced by several models where SIRT1, SIRT2, SIRT3, SIRT4, or SIRT6 knockout mice are more prone to cancer incidence[13, 19]. Overexpression of SIRT1 in the gut epithelium was found to suppress tumorigenesis in a mouse model of colon cancer[71], however others have drawn the opposite conclusion[72, 73]. In humans, SIRT3 protein and mRNA levels are strongly decreased in breast and ovarian cancer[74]. SIRT4 expression is decreased in lung, breast, bladder and gastric cancer and specific leukemia subtypes[75, 76]. SIRT6 levels are reduced in colon carcinoma and pancreatic cancer[77]. The metabolic state maintained by sirtuins can be particularly incompatible with the onset of cancer, as discussed further below.

On the other hand, in certain established tumors it is possible sirtuins have pro-tumorigenic roles by promoting survival under the stress conditions that dominate the cancer cell state. For example, high SIRT1 expression is observed in drug-resistant cancers[78]. In numerous studies, SIRT1 levels are elevated in human cancer relative to normal tissue. In fact, maintaining SIRT1 expression appears so vital for cancer cells that there are no reported deletions of SIRT1 in cancer and only extremely rare instances of SIRT1 mutation[13]. Thus while SIRT1 may counter the onset of cancer, an established tumor can greatly benefit from ramping up SIRT1 expression to induce pro-survival pathways[79]. Expression of another nuclear sirtuin, SIRT7, promotes survival of cancer cells and maintenance of a transformed state[80]. In breast, thyroid and liver cancer, SIRT7 is up-regulated[13]. The mitochondrial sirtuin SIRT3 promotes oral squamous cell carcinoma (OSCC). In a mouse model of OSCC, downregulation of SIRT3 antagonized cancer growth by preventing apoptosis of cancer cells[81]. This report is in line with a human genetics study showing an extra germline copy of SIRT3 limits apoptosis in glioma cells and predisposes patients to brain tumors[82]. Thus, while sirtuins in many cases can suppress cancer formation, sirtuins can also benefit the growth of some established tumors depending on the cancer type, stage and accompanying mutations. A more comprehensive understanding of sirtuin functions and relevant targets in cancer may shed light on the pro- or anti-tumorigenic roles of sirtuins in particular tumor types.

Altered glucose metabolism in cancer and regulation by sirtuins

In many cancers, metabolic reprogramming is characterized by increased glucose uptake[83]. Glucose is predominantly used for biosynthetic purposes; intermediates of glycolysis are directed toward pathways that build macromolecules including nucleotides, lipids and proteins. In addition to biosynthesis, glucose contributes to ATP production, generation of signaling molecules and antioxidants, and production of lactate that can acidify the tumor microenvironment to promote migration, genetic instability and cancer cell stemness[84–87]. Upregulation of glycolysis and lactic acid production even under normoxia when mitochondria are functional is termed the ‘Warburg effect’.

Elevated glycolysis can additionally fuel ATP production, even under hypoxic conditions in cancer cells. Rapidly growing or metastatic tumors often have low oxygen due to inadequate blood supply. In the absence of sufficient oxygen, mitochondrial ATP production is limited[88]. To circumvent an energetic deficit, glycolysis can be upregulated to generate ATP via substrate level phosphorylation. The amount of ATP produced by glycolysis is only a fraction of that generated by the electron transport chain; however, greatly induced glycolysis could support bioenergetic homeostasis when oxygen is limiting[89].

HIF, PHDs and metabolic stress sensing

Hypoxia inducible factor (HIF) is a master transcriptional activator of glycolysis with strong links to cancer. HIF acts as an α/β heterodimer[90]. There are three HIFα isoforms, and of these Hif1α and 2α are the most well-studied[91]. Increased levels of HIF1α and 2α are observed in many cancers and correlate with worse prognosis[92]. While overlap exists in many HIF1α and 2α target genes, a number of genes are exclusively modulated by just one HIF isoform. In healthy cells, HIF is activated under hypoxia to promote glycolytic metabolism and other pathways that mediate cell survival under low oxygen. However, under some conditions of cellular oxidative stress and in many cancers, aberrantly activated HIF facilitates metabolic reprogramming and upregulation of glycolysis even when oxygen levels are sufficient[93]. Physiologically under normoxia, HIF transcriptional activity is limited to a low, basal level. Cytosolic HIFα is hydroxylated by oxygen-dependent prolyl hydroxylase domain (PHD) enzymes and subsequently ubiquitinated by the von Hippel-Lindau (VHL) E3 ubiqutin ligase, targeting HIFα for degradation [94].

Under hypoxia, PHD activity is inhibited by low oxygen availability and HIFα is stabilized. HIFα translocates to the nucleus and forms a heterodimer with HIFβ (also called aryl hydrocarbon nuclear receptor, ARNT) resulting in a functional transcription factor in complex with the coactivator p300/CBP[95]. This complex binds HIF-responsive elements (HRE) in promoters of target genes to activate a transcriptional program that boosts angiogenesis, erythropoiesis and glycolytic metabolism[96].

Much like sirtuins, PHDs are perfectly poised to elicit metabolic alterations in response to changing nutrient availability or stress. PHDs are a family of α-ketoglutarate dependent dioxygenases that hydroxylate proline residues of target proteins[97, 98]. There are three main mammalian PHDs (also called egg laying defective nine, or Egln, proteins in reference to their originally described function in egg laying by C. elegans)[99]. During catalysis, PHDs transfer one atom of molecular oxygen to a proline residue of a substrate protein, resulting in prolyl-hydroxylation. The other oxygen atom is transferred to α-ketoglutarate which is subsequently dexarboxylated to form carbon dioxide and succinate[97]. PHD catalytic activity is sensitive to several key molecules that can be viewed as indicators of the cellular metabolic state: oxygen, ROS and specific TCA cycle intermediates[91]. In response to changing levels of these inputs, PHDs have been shown to instigate HIF-driven metabolic changes that restore homeostasis and redox balance.

Because the PHD catalytic mechanism requires molecular oxygen, a drop in intracellular oxygen levels can decrease PHD activity and consequently stabilize HIF[88]. Two PHD family members, PHD1 and 2, are quite sensitive to subtle changes in cellular oxygen levels due to their weak affinity for oxygen. The K_m for oxygen is only slightly higher the normal oxygen concentration in the cell. This suggests PHD1 and 2 normally operate at sub-optimal conditions and any drop in oxygen can potentially make PHDs much less active[96]. Oxygen sensing by PHD1 and 2 situates these enzymes as integral components in the HIF-driven transcriptional response to low intracellular oxygen. PHD activity is frequently repressed in tumors that have become hypoxic due to excessive oxygen consumption or insufficient blood supply[100].

The use of α-ketoglutarate as a co-substrate makes PHDs sensitive to TCA cycle imbalances that are observed in some cancers. Thus, PHD activity and HIF stability can be regulated even under normoxic conditions. At high concentrations, succinate and fumarate competitively inhibit the PHD binding site that is normally occupied by the structurally similar molecule α-ketoglutarate[101, 102]. In some tumors, deficiencies in the TCA cycle enzymes succinate dehydrogenase and fumarate hydratase lead to a build-up of succinate and fumarate, respectively[103]. The overabundance of these metabolites inhibits PHD function and is linked to HIF-driven metabolic reprogramming in cancer.

PHD enzymatic activity is intrinsically sensitive to redox status due to the requirement for reduced iron in the catalytic site. The PHD catalytic domain contains a conserved triad of two histidines and one aspartate that coordinate iron[104]. To enable oxygen binding at this catalytic site, iron must be maintained in the reduced (Fe²⁺) state, a function achieved by the antioxidants ascorbate (vitamin C) or glutathione[105, 106]. Under high ROS, these antioxidant molecules may be depleted leading to oxidation of the catalytic iron and inhibition of PHD activity. In this way, the HIF response can also be turned on downstream of increased ROS, a common scenario in cancer[107].

Modulation of HIF by sirtuins

Several studies have shown that the stress- and nutrient-sensing pathways that coordinate HIF activity also intersect with sirtuins at numerous nodes. For example, multiple sirtuins oppose HIF-driven metabolic rewiring (Figure 2). Elaborate control mechanisms enforced by SIRT1, SIRT2, SIRT3, SIRT6 and SIRT7 counter HIF activity to keep glucose metabolism in check[13].

Figure 2. Sirtuin-mediated repression of the basal HIF response and glycolytic metabolism during normoxia.

(A) Sirtuins obstruct HIF-mediated reprogramming of cell metabolism under normoxia. HIF activity is restricted under normoxia due to proteasomal degradation of HIF1α, which is triggered by PHD family members and p-VHL ubiquitin ligase. Degradation limits movement of HIF1α to the nucleus. In mitochondria, SIRT3 limits ROS levels, thus helping maintain PHD function so that HIF1α can be degraded. In the cytosol, SIRT2 represses basal HIF1α by direct deacetylation. HIF1α that does enter the nucleus is deacetylated and repressed by SIRT1. SIRT6 inhibits HIF1α in the nucleus via direct binding to HIF1α at hypoxia responsive elements (HRE) on gene promoters, preventing formation of a functional transcription factor complex. SIRT6 further represses HIF-regulated genes by deacetylating Histone H3K9 at glycolytic gene promoters. (B) Loss of sirtuin activity, such as in the transition to cancer or in genetic mouse knockout models, leads to activation of HIF-mediated glycolytic metabolism even under normoxia. In the absence of SIRT3, ROS levels increase and inhibit PHD-catalyzed hydroxylation of HIF1α. Also, in the absence of specific sirtuins, HIF1α is hyperacetylated and more readily moves to the nucleus and forms a functional transcription factor in complex with HIF1β and p300. HIF= hypoxia inducible factor. PHD= prolyl hydroxylase domain. VHL= von Hippel-Lindau E3 ubiquitin ligase. OH= prolyl hydroxylation. Ac= lysine acetylation. Ub= ubiquitination. ROS= reactive oxygen species.

In the nucleus, there is complex and considerable interplay between SIRT1 and HIF in line with the role of SIRT1 as a promoter of oxidative metabolism and mitochondrial function (Figure 2A). Under normoxia, SIRT1 inhibits the basal HIF response by promoting stability of the VHL transcript to drive degradation of HIFα[50]. SIRT1 further inactivates HIF1α in the nucleus by removing an acetyl group that is key to the interaction between HIF1α and p300[108]. Under hypoxia, the gradual drop in NAD⁺ decreases SIRT1 function and contributes to HIF activation[108], which the authors of this study propose synergizes with intratumoral PHD inhibition to maximize HIF-driven glycolysis. Despite declining NAD⁺, Dioum et al show residually active SIRT1 can deacetylate and activate HIF2α under hypoxia, helping drive a switch toward isoform-specific target genes (discussed further below)[109]. Interestingly, SIRT1 is a HIF1 and 2 target gene[110]. It is possible that upregulation of SIRT1 expression under hypoxia serves to build a pool of SIRT1 that can rapidly dampen the HIF signal as soon as adequate oxygen is achieved. A similar rationale has been suggested to explain why some PHD family members are HIF target genes[104].

Also in the nucleus, SIRT6 represses HIF transcriptional activity to limit glycolysis in cancer (Figure 2A) [111]. First, SIRT6 deacetylates histone H3K9 on the promoter of HIF target genes, aiding in gene silencing. SIRT6 also directly interacts with and inhibits HIF1α on HRE of glycolytic genes. SIRT6 knockout in mouse embryonic fibroblasts (MEFs) boosts expression of key glycolytic enzymes including pyruvate dehydrogenase kinase 1 (PDK1). Elevated PDK1 was additionally shown to be vital for the transformation phenotype of SIRT6 KO MEFs[77]. The authors of this study further show that glycolysis is increased upon SIRT6 conditional knockout in an APC^min/+ mouse model of colon cancer and is linked with increased tumor incidence[77].

Nuclear SIRT7 in cell culture has been shown to inhibit the HIF response by decreasing HIF1α and 2α protein levels[112]. The mechanism remains to be elucidated, but appears independent of SIRT7 catalytic activity as well as PHD- and proteasome-mediated degradation pathways. Of note, SIRT2 deacetylates and represses HIF1α in the cytosol[113]. However, this interaction is less extensively studied.

In the mitochondria, SIRT3 promotes mitochondrial metabolism and limits HIF-driven glycolysis by two mechanisms: deacetylation and coordination of ROS signaling (Figure 2A). SIRT3 boosts mitochondrial metabolism by deacetylating and activating enzymes involved the TCA cycle and fatty acid oxidation including SDH, LCAD, GDH and IDH2[15]. In tandem, SIRT3 limits glycolytic metabolism by coordinating a multipronged strategy to decrease ROS (discussed further below) and reduce HIF function. By repressing ROS, SIRT3 promotes PHD activity and HIF degradation. SIRT3 loss dramatically boosts ROS, which is proposed to deactivate PHD family members and consequently stabilize HIF1α (Figure 2B) [74, 114]. Indeed, in SIRT3 knockout MEFs, increased ROS promotes a HIF-mediated transition to the Warburg effect[74].

Thus, loss of sirtuin function has been shown in many cases to shift the cell toward glycolytic metabolism in a process that is amenable to transformation (Figure 2B). In cancer cell lines and sirtuin knockout mouse models, low expression of SIRT1, 3 and 6 correlates with increased levels of HIF1 target genes including the glucose transporter GLUT1 as well as enzymes in glycolysis and lactate production, including HK, PDK1, PGK1 and LDHA[50, 74, 111].

Additional glycolytic regulation by sirtuins

Aside from HIF regulation, sirtuins can modulate glycolysis in other ways. SIRT1 has been shown to regulate glycolytic gene expression downstream of the transcriptional activator MYC, although there are conflicting reports concerning the direction of this modulation[115–118]. Loss of SIRT3 in cancer cells additionally boosts glycolytic metabolism via pyruvate dehydrogenase (PDH), the major enzyme that channels glucose-derived pyruvate into the TCA cycle. SIRT3 deacetylates and activates the PDH catalytic subunit E1a, which directs pyruvate to the TCA cycle[119, 120]. In the absence of SIRT3, pyruvate entry into the TCA cycle is blocked. This is thought to enable glycolytic intermediates to be redirected toward biosynthetic pathways and lactate production. Overexpression of a constitutively deacetylated mimetic of PDH in MCF7 breast cancer cells resulted in a less transformed phenotype in soft agar assays, while a constitutively acetylated mimetic- which mimics loss of SIRT3- had a more highly transformed phenotype[120].

Sirtuin regulation of gluconeogenesis in cancer

Gluconeogenesis is a lesser-studied arm of glucose metabolism in cancer. A number of sirtuins coordinate gluconeogenesis, and the links to cancer are just beginning to be explored[15]. Physiologically, glucose production occurs in the liver and to a small extent the kidneys in order to maintain blood glucose between meals. However, recent studies demonstrate cancer cells may operate at least portions of the gluconeogenic pathway. Leithner et al found that lung cancer cell lines and tissue samples overexpress the mitochondrial isoform of PEPCK (PCK2), a gluconeogenic gene[121]. Metabolic tracing demonstrated that lung cancer cells convert lactate to pyruvate, then to oxaloacetate via pyruvate carboxylase, and then to phosphoenolpyruvate via PCK2. This pathway generates glycolytic intermediates in cancer cells deprived of glucose. Of note, SIRT6 represses gluconeogenic gene expression in concert with p53, the histone acetyltransferase GCN5, the transcription factor FOXO1 and the transcriptional coactivator PGC-1α[122, 123]. Zhang et al observed that the tumor suppressor p53 boosts SIRT6 expression in order to limit gluconeogenesis, a finding observed in both liver and colon cancer cell lines[123]. The authors propose that suppressing gluconeogenesis via SIRT6 may be an additional anti-neoplastic function of p53. Future studies are needed to explore whether regulation of gluconeogenesis by other sirtuins is meaningful for cancer metabolism.

Alternate fuel sources in cancer and coordination by sirtuins

Beyond altered glucose utilization, many cancers additionally or alternatively display addiction to fatty acids or amino acids such as glutamine. An intensely studied use of these alternate fuels is anaplerosis, the process of refilling the TCA cycle. Anaplerotic pathways provide alternative entry sites to generate TCA cycle intermediates, which are often used for anabolic and bioenergetic purposes in cancer[124]. In many healthy tissues, PHD is the major enzyme that channels glucose-derived pyruvate into the TCA cycle[125]. PDH converts pyruvate to acetyl-CoA, and then acetyl-CoA condenses with oxaloacetate to form citrate. However, PDH activity is often limited in cancer; additionally, roadblocks at other steps in the TCA cycle or shunting of metabolites toward biosynthetic pathways can limit production of oxaloacetate, which is needed to fuel subsequent rounds of the TCA cycle[124]. In cancer, the TCA cycle can be refueled by alternate pathways including glutaminolysis, reverse TCA cycling and fatty acid oxidation.

Regulation of the glutamine anaplerosis by sirtuins

A major driver of glutamine metabolism in cancer is the MYC family of transcriptional activators including MYC (c-MYC), L-MYC and N-MYC. This family is known to ubiquitously amplify expression of most genes undergoing transcription, and MYC target sequences have been identified on 30% of all genes[126]. In tumors, aberrant upregulation of specific gene sets by MYC is linked to growth advantages in cancer cells. Specifically, MYC boosts glutamine metabolism by increasing expression of glutamine transporters as well as enzymes involved in directing glutamine toward the TCA cycle, including glutaminase (GLS)[127]. Many sirtuins have been linked to regulation of MYC, but SIRT6 has most strongly been shown to coordinate glutamine metabolism via MYC[77]. By deacetylating H3K56 residues at MYC target gene promoters, SIRT6 suppresses MYC transcription activity specifically toward genes involved in glutamine as well as glucose metabolism. Accordingly, SIRT6 KO MEFS show increased glutamine uptake, an tumor advantage that may promote the increased tumor growth of SIRT6 knockout MEF xenografts models compared to wildtype[77]. SIRT7 also suppresses MYC, while SIRT1 boosts MYC activity, and future studies may reveal the relevance of these interactions on glutamine metabolism in cancer[126].

SIRT4 coordinates glutamine anaplerosis in an alternate manner. SIRT4 shuts down an access point that directs glutamine into the TCA cycle (Figure 3A). In cancer cells, the ability of SIRT4 to restrict the supply of this alternative fuel limits tumorigenesis[75]. SIRT4 impedes glutamine anaplerosis by ADP-ribosylating and inhibiting glutamate dehydrogenase (GDH)[28]. To direct glutamine toward the TCA cycle, glutamine can be converted to glutamate by the glutaminase family of enzymes. Glutamate is then converted to the TCA cycle intermediate α-ketoglutarate by mitochondrial GDH[128] or alternatively by transaminases. In cancer cells, DNA damage dramatically induces SIRT4 expression by an unknown mechanism, and glutamine anaplerosis is inhibited[75]. Consequently, TCA cycle intermediates are depleted and via mechanisms yet to be elucidated the cell cycle stalls. This SIRT4-mediated metabolic pause allows time for DNA repair before the cell proceeds through the cell cycle. In the absence of SIRT4, glutamine anaplerosis remains activated even during DNA damage. DNA damage persists and cellular proliferation and transformative properties are increased, possibly due to newly occurring DNA mutations.

Figure 3. Sirtuin-mediated regulation of TCA cycle anaplerosis.

(A) SIRT4 leads to repression of glutamine anaplerosis. SIRT4 expression is strongly induced in response to various cellular stresses including dysfunctional mTORC1 and DNA damage. The connections between mTORC, DNA damage and SIRT4 are unclear. In cancer cells, induction of SIRT4 limits entry of glutamine into the TCA cycle, in part by ADP-ribosylating and inhibiting GDH. (B) SIRT1 promotes glutamine anaplerosis and reductive carboxylation during chronic acidosis. Chronic acidosis can occur in cancer cells due to a metabolic switch to increased glycolysis and lactate production. Shunting pyruvate toward lactate can deplete the TCA cycle, driving the need for anaplerosis by other fuels. SIRT1 is activated by chronic acidosis. Under this condition, SIRT1 deacetylates HIFα. Deacetylation inhibits HIF1α but activates HIF2α leading to expression of a specific subset of target genes that promote glutamine anaplerosis and reductive carboxylation. SLC1A5 is induced to increased glutamine import into mitochondria. GLS1 is induced to convert glutamine to glutamate. IDH1 is induced to redirect α-ketoglutarate towards isocitrate via a reductive carboxylation. GDH= glutamate dehydrogenase. GLS1= glutaminase isoform 1. IDH1= isocitrate dehydrogenase 1.

In related studies, Csibi et al found that mammalian Target of Rapamycin Complex 1 (mTORC1), a serine/threonine kinase that drives cellular nutrient uptake and proliferation, inhibits SIRT4-mediated repression of anaplerosis (Figure 3A). mTORC1 represses SIRT4 expression by inhibiting CREB2, a transcription factor that induces SIRT4[76]. In MEFs with hyperactivated mTORC1, SIRT4 expression was decreased, thus activating glutamine metabolism and anaplerosis. High mTORC1 signaling commonly occurs in cancer and is proposed to benefit cancer cell growth and survival by increasing glutaminolysis via SIRT4 repression. In future studies, it will be interesting to examine whether mTOR signaling converges on DNA damage responses via SIRT4 (Figure 3).

Sirtuins and reverse TCA cycling

Although the TCA cycle had long been thought to operate in one direction, it is now known that at least a portion of the TCA cycle operates in reverse in some cancer cells, particularly in response to redox stress, impaired respiration or hypoxia[129–131]. Forward TCA cycling requires pyruvate-derived acetyl CoA to condense with oxaloacetate and form citrate. Under harsh conditions, such as proliferating cells in hypoxia, pyruvate is directed almost entirely toward lactate[132]. Acetyl CoA may be sufficiently depleted such that forward TCA cycling is limited. In this case, an alternate pathway is needed to produce citrate. Citrate is especially essential for cancer cells because it is a key building block for fatty acid synthesis[132]. In cancer cells, glutamine-derived α-ketoglutarate can undergo reductive carboxylation catalyzed by isocitrate dehydrogenase (IDH) 1 or 2 to generate citrate[133, 134].

A recent study shows SIRT1 promotes reductive carboxylation (Figure 3B)[135]. In diverse cancer cell lines, prolonged acidosis (pH 6.5), which mimics extensive lactate production, upregulates genes important for reverse TCA cycle flux in a SIRT1-dependent manner. Mechanistically, under low pH SIRT1 deacetylates HIF1α and 2α. Deacetylation inhibits HIF1α but activates HIF2α. By activating HIF2α, SIRT1 triggers expression of key target genes including the glutamine transporter SLC1A5, the mitochondrial glutaminase isoform 1 (GLS1) and IDH1. Thus, the SIRT1/HIF2α axis promotes a metabolic shift to reductive glutamine metabolism in order to maintain levels of TCA cycle intermediates under the harsh conditions experienced by cancer cells.

Sirtuins and subcellular metabolic crosstalk

A fertile area for future sirtuin research is the possible link between cancer and transcriptional control of mitochondrial biogenesis and oxidative metabolism. Peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) is a master transcriptional coactivator of mitochondrial metabolism and oxidative phosphorylation[136]. It is known that increased mitochondrial biogenesis and respiration driven by PGC-1α promote tumor cell invasion and metastasis[137]. Further, SIRT1 and 3 can promote mitochondrial metabolism by turning on PGC-1α mediated transcriptional programs[138], however the direct connections between sirtuins, mitochondrial biogenesis and cancer metabolism are unclear. Under fasting SIRT1 deacetylates and activates PGC-1α[139]. Additionally, SIRT3 enhances expression of PGC-1α and is essential for turning on PGC-1α induced mitochondrial biogenesis[140, 141]. Thereby, these sirtuins promote efficient energy production during nutrient deprivation. By favoring mitochondrial metabolism over glycolytic metabolism, it is reasonable to hypothesize that, on one hand, sirtuins hinder the metabolic changes needed for cellular transformation. However, functional mitochondria are still vital for an established cancer cell population to grow and metastasize[142], and therefore sirtuin-induced upregulation of mitochondrial metabolism may be one reason why sirtuin expression can benefit many existing cancers.

Sirtuin-driven mitochondrial programs for ROS homeostasis

The role of ROS in cancer is complex. High ROS have been linked to cancer incidence in numerous studies[143–145]. ROS production can be increased by stalling or inefficiencies in the electron transport chain. ROS have both adverse and beneficial consequences on cancer cells, which may in part be determined by the stage of tumor progression or specific cancer-associated mutations[146]. First, at excessive levels ROS can damage cellular machinery including proteins, lipids, DNA and RNA. By causing genetic damage, ROS may have mutagenic and pro-tumorigenic capacities. ROS also serve as important signaling molecules that can drive growth and cell division in cancer[147–149]. For example, ROS have been shown to stimulate the PI3K signaling pathway, a major driver of cell growth that is hyperactivated in many cancers[146]. ROS activates PI3K by inhibiting the major negative regulator of this pathway, PTEN[150]. Mechanistically, cytosolic ROS oxidizes the catalytic cysteine residue of PTEN and promotes formation of a disulfide bond to block PTEN function and enable unrestrained PI3K activity[151]. ROS additionally boosts activity of AKT, a mediator of the PI3K pathway, by blocking activity of a phosphatase that otherwise inhibits AKT[152].

While in some contexts ROS boosts cellular transformation and cancer cell growth, high ROS has been shown in other contexts to act as a pro-apoptotic signal instructing cancer cells to die[153, 154]. Accelerated metabolism in cancer often generates high ROS. Because ROS can reach toxic concentrations, adaptive mechanisms must be put it place by cancer cells to restore ROS homeostasis and allow survival. Therefore, many cancer cells up-regulate antioxidant pathways that endow tumors with additional stress protection. In this way, antioxidant programs may actually promote cancer progression of established tumors[155–157].

Coordination of ROS by nuclear and cytosolic sirtuins

Several sirtuins have major roles in preventing excessive, damaging levels of ROS (Figure 4). In the nucleus, SIRT1 promotes ROS defense via deacetylation of key transcriptional regulators of stress resistance including p53, forkhead homeobox type O (FOXO) proteins, PGC-1α, heat shock factor protein 1 (HSF1) and nuclear erythroid factor 2-related factor 2 (NRF2)[19, 158]. These targets increase antioxidant defenses, limit ROS production and drive apoptosis in the event of uncontrollable ROS. For example, in response to oxidative stress or nutrient deprivation, SIRT1 deacetylates FOXO[159, 160]. This promotes nuclear retention of FOXO where it can turn on oxidative stress resistance genes such as those encoding as mitochondrial superoxide dismutase 2 (SOD2), catalase and Bim. SIRT2 can also deacetylate cytosolic FOXO, causing it to move to and remain in the nucleus. By promoting the pro-apoptotic factor Bim, SIRT2 was shown to promote cell death when cells are under severe stress[161]- an appropriate response that can benefit survival of the surrounding tissue. When PGC-1α is deacetylated by SIRT1, this directs PGC-1α toward a specific gene set that promotes increased mitochondrial biogenesis and oxidative capacity to limit ROS production[162].

Figure 4. Roles of sirtuins in ROS defense.

Sirtuins coordinate a multi-faceted regimen to limit ROS. In the nucleus, SIRT1 deacetylates a number of transcriptional regulators to boost gene expression programs that increase antioxidant defenses, limit ROS production and drive apoptosis in the event of uncontrollable ROS. In the cytosol, SIRT2 deacetylates FOXO to drive its nuclear import and activity. SIRT2 also deacetylates and activates the glycolytic enzyme PGAM. Activated PGAM boosts conversion of 3-phosphoglycerate to 2-phoshoglycerate. PGAM promotes production of the antioxidant molecule NADPH, whereas a buildup of 3-phosphoglycerate would otherwise inhibit NADPH synthesis. In mitochondria, SIRT3 boosts oxidative capacity and limits ROS production by deacetylation and activation of subunits in all five electron transport chain complexes, including SDHA which dually serves as a TCA cycle enzyme that generates FADH₂. SIRT3 interacts with FOXO3a to elevate expression of ROS defense pathways both in the nucleus and mitochondria. SIRT3 also drives antioxidant strategies to clear ROS. Deacetylation and activation of IDH2 generates the antioxidant molecule NADPH. Finally, SIRT3 deacetylates and activates SOD2, an enzyme that clears superoxide. PGAM= phosphoglycerate mutase. SDHA= succinate dehydrogenase A. IDH2= isocitrate dehydrogenase 2.

In the cytosol, SIRT2 also combats ROS by promoting generation of the antioxidant molecule NADPH (Figure 4). In response to H₂O₂ treatment in cell culture, SIRT2 deacetylates and activates phosphoglycerate mutase (PGAM)[163]. PGAM is a glycolytic enzyme that converts 3-phopshoglycerate to 2-phosphoglycerate. In the absence of SIRT2, PGAM is less active and levels of 3-phosphoglycerate increase. Not only is glycolysis impeded, but also high amounts of 3-phosphoglycerate inhibit 6-phosphogluconate dehydrogenase (6PGD), an enzyme in the pentose phosphate pathway that produces NADPH. Thus SIRT2 is needed to activate PGAM and help maintain NADPH synthesis for use in clearing ROS.

Regulation of ROS by mitochondrial sirtuins

Mitochondrial SIRT3 is a major player in cellular antioxidant strategies. SIRT3 coordinates a multi-faceted post-translational program to reduce ROS (Figure 4). First, SIRT3 limits ROS production by promoting efficient electron flow through the ETC. SIRT3 deacetylates and activates specific subunits in all five ETC complexes. SIRT3 activates Complex I[164] and succinate dehydrogenase A (SDH-A) in Complex II[165, 166], components of the ETC where electrons are initially donated. SDH-A dually functions as a TCA cycle enzyme that oxidizes succinate to fumarate while converting FAD to FADH₂. Electrons from FADH₂ are directly fed into the SDHB subunit of Complex II and then on through the ETC. By activating Complex I and II, SIRT3 enables NADH and FADH₂ to more readily contribute electrons to the ETC. SIRT3 activates Complex III[167, 168] and IV[43, 51] to further promote efficient electron flow and generation of a proton gradient. Activation of Complex V (ATP synthase) boosts ATP production[168, 169]. By activating all ETC components, SIRT3 increases mitochondrial oxidative capacity, prevents ETC stalling and limits ROS production. Of note, mitochondrial SIRT5 suppresses SDH activity, possibly via desuccinylation, but the relevance to cancer is not known[170]. In cancer cells grown in galactose- a condition that increases the dependency on oxidative phosphorylation- SIRT3 has been shown to boost the electron transport chain in an additional way. SIRT3 deacetylates cyclophilin D leading to release of hexokinase 2 (HK2) from mitochondria[171]. HK2 in mitochondria is linked to increased reliance on glycolytic metabolism, but released HK2 stimulates oxidative phosphorylation.

In addition to limiting ROS production, SIRT3 also promotes ROS clearance. SIRT3 deacetylates SOD2, a key mitochondrial enzyme in antioxidant defense which initiates ROS detoxification by converting superoxide to H₂O₂[172, 173]. Mice heterozygous for SOD2 have more DNA damage and 100% increased cancer incidence than wildtype controls[174]. To complete ROS clearance, H₂O₂ is reduced to water by the antioxidant molecule glutathione. SIRT3 indirectly boosts levels of glutathione. By deacetylating and activating the TCA cycle enzyme IDH2, SIRT3 promotes conversion of isocitrate to α-ketoglutarate in a reaction that simultaneously produces NADPH[175]. NADPH is a reducing agent with antioxidant functions that generates reduced glutathione. Finally, SIRT3 has been shown to boost transcription of antioxidant enzymes by interacting with mitochondrial FOXO3a and boosting its affinity for antioxidant gene promoters in the nucleus and oxidative phosphorylation subunit gene promoters in mitochondria[167, 176, 177]. Through these multiple mechanisms, SIRT3 boosts cellular oxidative capacity, decreases ETC stalling and promotes antioxidant defenses.

Recent studies suggest SIRT3 is not the only mitochondrial sirtuin that coordinates ROS; SIRT5 also limits ROS by at least two mechanisms. SIRT5 desuccinylates and activates SOD1, the largely cytosolic isoform of SOD that is also present at low amounts in mitochondria[178]. SIRT5 additionally boosts transcription of NRF2, consequently promoting gene expression programs important for maintaining redox homeostasis[179].

Through regulation of ROS, mitochondrial sirtuins play a critical yet complex role in cancer progression[180]. By decreasing ROS, SIRT3 has the capacity to limit tumorigenesis. In healthy cells SIRT3 decreases ROS, maintains PHD activity and represses of HIF, as described above[74, 114]. Consequently SIRT3 represses the transition to the Warburg Effect, thus restricting a metabolic pathway that is quite often vital to neoplastic transformation[181]. SIRT3 may further impede cancer onset by limiting DNA damage caused by ROS[167]. Additionally, in pancreatic cancer cell lines, SIRT3 overexpression was shown to decrease ROS and suppress proliferation through coordination of iron metabolism[182]. By limiting ROS, SIRT3 repressed redox-sensitive iron responsive proteins (IRPs), thus downregulating an iron-related gene set that includes the transferrin receptor. In the absence of SIRT3, increased levels of transferrin receptor correlated with high intracellular iron. Loss of SIRT3 conferred a growth advantage to pancreatic cancer cells, at least in part due to the abundance of iron, an essential cofactor in DNA synthesis.

While reducing ROS helps SIRT3 act as a tumor suppressor in certain cases, established tumors may benefit from maintaining SIRT3 function to limit ROS and avoid apoptosis. High SIRT3 expression was observed in oral squamous cell carcinoma cell lines and human samples[81]. In cardiomyocytes, overexpression of SIRT3 conferred resistance to genotoxic and oxidative stress-inducing agents including camptothecin and H₂O₂[183]. Thus, it is tempting to speculate that in an already established tumor, high SIRT3 may promote cancer cell viability by limiting ROS, a pro-apoptotic signal that would otherwise instruct cancer cells to die. Future studies are needed to reveal the range of cancer contexts in which ROS is either beneficial or harmful and in which sirtuins have tumor-suppressive or tumor-promoting functions.

Conclusion

Our knowledge of the metabolic consequences of the seven sirtuins in distinct cancer contexts is generating an increasingly clear view of the metabolic susceptibilities that can potentially be targeted in cancer treatment. Along these lines, several diverse classes of sirtuin inhibitors including Tenovins are currently being tested for anti-cancer properties in animal models and in the clinic[19, 184, 185]. It is overwhelmingly clear that sirtuins are major players in cancer cell metabolism. Many more connections between sirtuins and cancer metabolism likely remain to be discovered based on the numerous substrates, both known and unknown, that have not yet been tested for a role in cancer. Additionally, we are rapidly learning of the importance of new and diverse metabolic pathways involved in cancer. A number of sirtuins coordinate lipogenesis and fatty acid oxidation, both pathways recently highlighted for involvement in tumor metabolism. The role of sirtuins in these processes in cancer is not yet known. Initial successes in targeting the metabolic dependencies of cancer cells, such as inhibiting mutant IDH2 in the treatment of acute myelogenous leukemia[186], set a precedent for impeding metabolic pathways in the future. Sirtuin research will surely aid in this pursuit. Sirtuin research has revealed interconnected metabolic and signaling pathways that drive tumor biology and could be specifically targeted in cancer treatment. Additionally, sirtuin expression could be considered for its utility as a metabolic biomarker to potentially designate the most promising therapeutic approach in a new era of metabolically-based precision medicine.