Introduction

Acetyl coenzyme A (acetyl-CoA) is both a central metabolic intermediate and a key signalling molecule. It mainly functions in metabolism by donating its two-carbon acetyl group, which is linked to CoA by a thioester bond. Because acetyl-CoA has a standard free energy of hydrolysis of −31.4 kJ/mol, which is similar to ATP (−30.5 kJ/mol), its acetyl group can easily be transferred to accepting molecules1. In mitochondria, acetyl-CoA is produced by glucose, lipid and amino acid catabolism, and powers the tricarboxylic acid (TCA) cycle and electron transport chain, and in the cytosol it is used for anabolism as the precursor for fatty acid and isoprenoid biosynthesis (Fig. 1). It is also the cellular substrate for protein lysine acetylation reactions, best known for regulating gene expression through nuclear histone acetylation, and it is an allosteric activator of enzymes including pyruvate carboxylase and pyruvate dehydrogenase kinase2,3. Acetyl-CoA’s central role in cell metabolism is so ancient that it is speculated to predate ATP as the cell’s main energy currency4.

Fig. 1: Compartmentalized acetyl-CoA production pathways.
figure 1

In general, mitochondrial acetyl coenzyme A (acetyl-CoA) is generated from glucose during anabolic conditions and from other nutrients, such as long-chain and medium-chain fatty acids (by β-oxidation), acetate (by acyl-CoA synthetase short-chain family member 1 (ACSS1)), amino acids and β-hydroxybutyrate during catabolic conditions, although different cell and tissue types have different nutrient preferences. Pyruvate generated from glycolysis enters the mitochondria through the mitochondrial pyruvate carrier (MPC) and undergoes oxidative decarboxylation by the pyruvate dehydrogenase complex (PDC), yielding acetyl-CoA, NADH and CO2. Fatty acids catabolized through β-oxidation are first transported into the mitochondria as acylcarnitines through the activity of carnitine palmitoyltransferase 1 (CPT1) and CPT2, which operate between the mitochondrial outer and inner membranes (not shown). Medium-chain fatty acids (MCFAs) do not require the acylcarnitine shuttle. Peroxisomes can also catabolize very long chain fatty acids (VLCFAs) to shorter-chain fatty acyl-CoA species, which are then transported to the mitochondria via the carnitine shuttle. Acetyl group carbons enter the tricarboxylic acid (TCA) cycle following a condensation reaction with oxaloacetate (OAA) catalysed by citrate synthase (CS). TCA cycle flux generates CO2 and the reducing equivalents NADH and FADH2 that drive the electron transport chain and ATP synthesis. Cytosolic acetyl-CoA is generated from citrate; citrate is exported from the mitochondria via the mitochondrial citrate carrier SLC25A1 and cleaved by cytoplasmic ATP–citrate lyase (ACLY) to make acetyl-CoA. Cytosolic citrate can also be generated through reductive carboxylation of cytoplasmic α-ketoglutarate (α-KG) by isocitrate dehydrogenase 1 (IDH1). Alternatively, cytosolic acetyl-CoA can be derived from acetate through the activity of ACSS2. Cytosolic acetyl-CoA is used for fatty acid synthesis and in the mevalonate pathway. Acetyl-CoA generated in the cytosol can diffuse into the nucleus, although ACLY, ACSS2 and the PDC have all been reported in the nucleus, where they might generate local acetyl-CoA. For example, deacetylation of histones produces acetate that can be converted to acetyl-CoA by nuclear ACSS2. ACC, acetyl-CoA carboxylase; ACAT, acetyl-CoA acetyltransferase; FASN, fatty acid synthase; FPP, farnesyl pyrophosphate; GGPP, geranylgeranyl pyrophosphate; HAT, histone acetyltransferase; HDAC, histone deacetylase; HMG-CoA, 3-hydroxy-3-methylglutaryl-CoA; HMGCR, 3-hydroxy-3-methylglutaryl-CoA reductase; HMGCS, 3-hydroxy-3-methylglutaryl-CoA synthase; IPP, isopentylphosphate; LCFA, long-chain fatty acid; ME1, malic enzyme 1; PC, pyruvate carboxylase; PPi, inorganic pyrophosphate.

Full size image

Given the multifaceted roles of acetyl-CoA in the cell, it is perhaps not surprising that it plays important roles in tumour metabolism. For example, it is well established that lipid biosynthesis contributes to tumour growth, enabling membrane production and expansion, and lipid-dependent signalling5. Accordingly, the fatty acid synthesis pathway and the mevalonate pathway are of substantial interest as therapeutic targets6,7. The acetylation of histones and other proteins in an acetyl-CoA-dependent manner has also emerged as a factor in various aspects of tumorigenesis, and might present additional opportunities for therapeutic intervention. Acetyl-CoA metabolic enzymes are frequently overexpressed in cancers, and post-translational modification of these enzymes allows their dynamic regulation in response to various signalling cues.

The goal of this Review is to discuss the roles of acetyl-CoA in cancer, primarily focusing on its functions in the cytosol and nucleus and the effects of targeting acetyl-CoA metabolic enzymes (targeting mitochondrial acetyl-CoA metabolism has been reviewed in depth elsewhere8,9). To this end, we first review major metabolic pathways involving acetyl-CoA. We then discuss how changes in acetyl-CoA production and its use as both a substrate and a metabolic signal support tumorigenesis, highlighting promising therapeutic strategies that target nuclear and cytosolic acetyl-CoA metabolic enzymes. We conclude with an outlook highlighting gaps in our understanding of acetyl-CoA regulation that require further investigation, in terms of both basic biology and cancer relevance.

Cellular roles of acetyl-CoA

Membranes are impermeable to acetyl-CoA, and it must therefore be generated within or transported into each cellular compartment in which it functions. The major pools of acetyl-CoA are present in the mitochondrial and nuclear-cytosolic compartments. The nucleus is emerging as an active metabolic compartment in which acetyl-CoA can be produced, although metabolites can also diffuse between the cytosol and the nucleus. Acetyl-CoA can additionally be generated within peroxisomes, and it can be transported from the cytosol into the endoplasmic reticulum10. To provide a backdrop for discussing acetyl-CoA metabolism in cancer, we begin here with a brief overview of acetyl-CoA functions within the mitochondria, cytosol and nucleus.

Mitochondrial acetyl-CoA

In the mitochondria, acetyl-CoA enters the TCA cycle upon condensation with oxaloacetate to produce citrate, which is oxidized to yield reducing equivalents that ultimately generate ATP through the electron transport chain and oxidative phosphorylation. Acetyl-CoA is produced in the mitochondria from several nutrient sources (Fig. 1). Pyruvate generated by glycolysis is transported into the mitochondrial matrix by the mitochondrial pyruvate carrier (MPC), where it undergoes irreversible oxidative decarboxylation by the multi-enzyme pyruvate dehydrogenase complex (PDC) to produce acetyl-CoA, NADH and CO2. Mitochondrial acetyl-CoA can also be generated by fatty acid β-oxidation, by the catabolism of certain amino acids (including the branched-chain amino acids (BCAAs) leucine and isoleucine), by the catabolism of ketone bodies such as β-hydroxybutyrate and through ATP-dependent synthesis from acetate11. The relative contribution of each of these pathways to generating mitochondrial acetyl-CoA varies with nutritional state and tissue type.

Cytosolic acetyl-CoA

Cytosolic acetyl-CoA is used for de novo biosynthesis of lipids, which are critical for building cellular membranes, storing energy and generating signalling metabolites such as diacylglycerols and phosphatidylinositol 3,4,5-trisphosphate. These processes are often upregulated in tumour cells12. It is also necessary for cytosolic protein acetylation and the generation of acetylated metabolites such as UDP N-acetylglucosamine, a nucleotide sugar that is crucial for glycosylation13. For use in de novo lipogenesis (DNL), cytosolic acetyl-CoA is irreversibly converted to malonyl-CoA by acetyl-CoA carboxylase 1 (ACC1) and ACC2 (encoded by the genes ACACA and ACACB, respectively) (Fig. 1). Malonyl-CoA also inhibits fatty acid oxidation, thus preventing simultaneous synthesis and degradation of fatty acids. Cytosolic acetyl-CoA can also be used to form mevalonate, the precursor for sterols and isoprenoids, by the consecutive actions of acetyl-CoA acetyltransferase (ACAT), 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) synthase (HMGCS) and HMG-CoA reductase (HMGCR), with the last step consuming cytosolic NADPH (Fig. 1).

When cytosolic acetyl-CoA is needed, mitochondrial citrate is exported to the cytosol, where it is cleaved by ATP–citrate lyase (ACLY) in an ATP-consuming reaction that generates oxaloacetate and acetyl-CoA14,15 (Fig. 1). Citrate can also be produced through the reductive carboxylation of α-ketoglutarate by the mitochondrial or cytosolic forms of isocitrate dehydrogenase (IDH2 and IDH1, respectively), a pathway that is particularly active in cells with mitochondrial defects or in hypoxia16,17,18. Finally, citrate may be obtained from the circulation through the plasma membrane citrate carrier SLC13A5, which is highly expressed in the liver and structurally distinct from the mitochondrial citrate transporter. In liver cancer cells, citrate uptake was shown to protect against nutrient and oxygen stress by feeding the TCA cycle and lipid synthesis19.

A major route to cytosolic acetyl-CoA that does not involve citrate is through the activity of acyl-CoA synthetase short-chain family member 2 (ACSS2), which converts acetate to acetyl-CoA at the cost of one ATP molecule (Fig. 1). Acetate can be imported from the circulation, with one of its main sources being the gut microbiota, or it can be generated within cells by histone deacetylation reactions (discussed later), by hydrolysis of acetylated metabolites or directly from pyruvate in highly glycolytic cells20,21. Notably, two other acetyl-CoA synthetase isoforms localize to the mitochondrial matrix, ACSS1 and ACSS3; the main difference between these enzymes is that ACSS3 prefers propionate as its substrate22. ACSS2 expression increases upon genetic deletion or inhibition of ACLY in certain contexts — for example, in cultured cells, in the liver during fructose consumption and in white adipose tissue — which switches the major source of acetyl-CoA from citrate to acetate to maintain DNL and histone acetylation23,24. Such compensatory interplay between ACLY and ACSS2 could potentially impact responses to inhibitors of these enzymes.

Nuclear acetyl-CoA

In its role as a signalling metabolite, acetyl-CoA is best known as the substrate for protein lysine acetylation reactions. Post-translational lysine acetylation occurs throughout the cell on metabolic enzymes, signalling enzymes and transcription factors25, although it is currently most appreciated in cancer for its role in nuclear histone acetylation, which is highly sensitive to acetyl-CoA availability26,27,28 (Fig. 1). Acetyl-CoA can enter the nucleus by diffusion through the nuclear pore; however, accumulating evidence suggests that acetyl-CoA is also produced directly within the nucleus11 (Fig. 1). Indeed, the predominantly cytosolic enzyme ACLY is also present in the nucleus26, where it has been shown to promote histone acetylation near sites of DNA double-strand breaks, enabling the recruitment of BRCA1 and break repair — a role that is lost if ACLY is confined to the cytosol by tagging it with a nuclear export signal29. ACSS2 is also both cytosolic and nuclear, and in some cell types, such as neurons, and contexts, such as nutrient deprivation, is predominantly nuclear30,31,32,33. Nuclear ACSS2 regenerates acetyl-CoA from acetate produced by histone deacetylation reactions, enabling acetyl-CoA recycling within the nucleus31. Such recycling might enable the maintenance of histone and transcription factor acetylation when levels of glucose-derived acetyl-CoA are low. An emerging concept is that certain high-stoichiometry histone acetylation sites such as histone H3 K23 might serve as acetyl-CoA reservoirs that can be accessed to allow histone acetylation at other sites that are key for gene regulation34,35 (Box 1). It is unclear whether these acetyl-CoA reservoirs can also be accessed to feed other metabolic pathways during temporary nutrient deprivation, as is frequently encountered in the tumour microenvironment (TME).

The PDC is another potential source of nuclear acetyl-CoA. While canonically localized to mitochondria, the PDC has also been reported to translocate as a complex from mitochondria to the nucleus in response to certain conditions, including mitochondrial stresses, growth factor signalling and specific developmental cues, where it regulates histone acetylation36,37,38 (Fig. 1). Although it is surprising that such a large complex (up to 10 MDa) could translocate to the nucleus, a recent study has delineated a mechanism that involves mitochondrial tethering to the nucleus and entry of the PDC into the nucleus independently of nuclear pores39. Finally, carnitine acetyltransferase (CRAT), a mitochondrial enzyme that buffers mitochondrial acetyl-CoA levels through acetylcarnitine synthesis and export, has also been reported to be present in the nucleus to support histone acetylation by regenerating acetyl-CoA from acetylcarnitine40, although the roles and regulation of this pathway remain little studied.

Acetyl-CoA enzyme regulation in cancer

The production and use of acetyl-CoA is dynamically controlled through the regulation of metabolic enzymes at both the transcriptional level and the post-translational level. The wealth of data discussed in the following section indicates that the abundance and activity of enzymes that produce and use acetyl-CoA are altered by oncogenic signalling mechanisms, transcriptional regulation and tumour microenvironmental stresses to promote tumour growth (Fig. 2).

Fig. 2: Regulation of acetyl-CoA metabolic enzymes.
figure 2

a, Transcriptional and post-translational regulation of acetyl coenzyme A (acetyl-CoA) metabolic enzymes in normal cells. ATP–citrate lyase (ACLY) cleaves citrate into oxaloacetate and acetyl-CoA. The acetylation of ACLY blocks its ubiquitylation, preventing degradation and enhancing stability, while phosphorylation by AKT is thought to enhance its activity. Both acetyl-CoA carboxylase 1 (ACC1) and ACC2 convert acetyl-CoA to malonyl-CoA, which is a substrate for de novo lipogenesis and an allosteric inhibitor of the mitochondrial enzyme carnitine palmitoyltransferase 1A (CPT1A) — an enzyme that catalyses the rate-limiting step of fatty acid oxidation (FAO). ACC1 is inhibited upon phosphorylation by AMPK, whereas ACC2 hydroxylation by PHD3 activates the enzyme. ACSS2 produces acetyl-CoA from acetate, and its acetylation is inhibitory, whereas phosphorylation promotes its stability. Acetyl groups are removed from acyl-CoA synthetase short-chain family member 2 (ACSS2) by sirtuin 1 (SIRT1). b, The impact of oncogenic, microenvironmental and systemic metabolic factors on the regulation of acetyl-CoA metabolic enzymes. Red signals are associated with oncogenic signalling and high nutrient availability and promotion of fatty acid synthesis from acetyl-CoA; green signals are associated with high lipid availability and promotion of FAO to generate acetyl-CoA and/or might be linked to obesity-related cancers; blue signals are associated with nutrient stress-inducing adaptation within the tumour microenvironment, which might induce FAO or fatty acid synthesis depending on the context; for example, energetic stress leading to AMPK activation promotes FAO, while lipid limitation, hypoxia or low pH can trigger fatty acid synthesis through steroid regulatory element-binding protein (SREBP) transcription factors. Each of these adaptations might support tumour growth and facilitate cell survival within the tumour microenvironment. α-KG, α-ketoglutarate; CPT1A, carnitine palmitoyltransferase 1A; CUL3, cullin 3; KLHL25, Kelch-like family member 25; mTORC1, mTOR complex 1; OAA, oxaloacetate; OGT, O-linked N-acetylglucosamine transferase; PHD3, prolyl hydroxylase 3; TCA tricarboxylic acid; Ub, ubiquitin.

Full size image

Regulation of acetyl-CoA metabolic enzymes by transcription factors

The genes encoding ACLY, ACSS2, ACC1 and ACC2 are often co-regulated, along with other lipogenesis genes, by sterol regulatory element-binding protein (SREBP) transcription factors. SREBPs are classically activated by sterol depletion, and are also activated in cancer by AKT–mTOR complex 1 signalling in response to oncogenic and growth factor signalling41,42,43 (Fig. 2). SREBP transcription factors can also cooperate with MYC to activate lipogenic gene expression in MYC-driven cancers, promoting dependence on DNL44. Moreover, tumour microenvironmental factors such as hypoxia and low pH have each been documented to drive the SREBP2-dependent upregulation of ACSS2 expression and acetate-dependent DNL45,46. Interestingly, in an obesity-associated multiple myeloma model, it was reported that adipocyte-derived angiotensin II promotes SREBP-dependent ACSS2 expression in myeloma cells, which in turn drives acetyl-CoA production and the acetylation and activation of the transcription factor interferon regulatory factor 4 (IRF4), to which myeloma cells are addicted47.

Transcription factors other than SREBPs have also been implicated in controlling the expression of acetyl-CoA metabolic enzymes in cancer. ACLY and ACACA are co-regulated by the nuclear receptor PPARγ to promote increased DNL in hepatocellular carcinoma (HCC)48. Further, NRF2–a transcription factor that regulates antioxidant and other cellular stress response mechanisms that is activated in several cancers – was found to regulate ACSS2 expression in oesophageal squamous cell carcinoma49. In terms of clinical correlates with expression, high ACSS2 expression is associated with low survival in several cancer types, including cervical cancer, renal cell carcinoma, triple-negative breast cancer, and grade II and grade III gliomas45,47,50,51,52. Conversely, low ACSS2 expression was found to correlate with aggressive phenotypes and low survival in HCC and colorectal cancer53,54. ACLY expression is upregulated across numerous cancer types, and high ACLY expression is associated with lower survival in ovarian and breast cancers, acute myeloid leukaemia and HCC55,56,57,58.

Regulation of acetyl-CoA enzymes by post-translational modifications

Post-translational modifications dynamically modulate acetyl-CoA metabolic enzymes in response to a variety of cues, including oncogenic signalling pathways (Fig. 2). ACLY has several described phosphorylation sites, of which S455, located within the enzyme’s disordered loop, is the most well studied. S455 can be phosphorylated by AKT, protein kinase A (PKA) and branched-chain ketoacid dehydrogenase kinase (BCKDK)59,60,61, suggesting that it is a critical node of metabolic regulation. S455 phosphorylation has been reported to enhance enzyme activity62, although it is worth noting that some prior studies found little change in the kinetic parameters of ACLY following S455 phosphorylation63,64,65,66. AKT signalling to ACLY occurs downstream of oncogenic and growth factor-mediated signalling, and has been shown to promote histone acetylation, aligning with an activating function of this modification27,29,67,68,69,70. Consistent with the presence of an AKT–ACLY–histone acetylation axis, global histone acetylation in human prostate tumours and gliomas correlates positively with phosphorylated AKT S473 levels27.

Regulatory pathways governing ACLY post-translational modifications other than phosphorylated S455 might also have roles in cancer. For example, ACLY tyrosine phosphorylation by the tyrosine-protein kinase LYN was recently reported in acute myeloid leukaemia cells and implicated in promoting lipid synthesis and histone acetylation71. Further, ACLY S447 and ACLY S451 are phosphorylated by glycogen synthase kinase 3 (GSK3), which functions in various cellular processes and is of interest as a therapeutic target in cancer72, although a function for these modifications has not yet been described73. The acetylation of ACLY at K540, K546 and K554, which increases under high-glucose conditions, has been shown to promote protein stability, lipid synthesis and tumour growth by blocking the ubiquitylation of ACLY by the ubiquitin ligase UBR4 (ref. 74). ACLY is also ubiquitylated at the same sites by cullin 3 (CUL3) and its adaptor protein KLHL25, and levels of CUL3 negatively correlate with ACLY levels in lung cancer75. Collectively, these studies show that ACLY is subjected to extensive post-translational mechanisms, at least some of which exhibit aberrant regulation in cancer.

ACSS2 acetylation and phosphorylation are regulated by cell growth and survival pathways often dysregulated in cancer. The NAD-dependent deacetylase SIRT1, which shows increased activity during cell stress, activates ACSS2 by deacetylating ACSS2 K66176. Tumour-associated stresses such as low levels of nutrients and hypoxia also promote ACSS2 nuclear localization31,33, and it has been shown that AMPK-dependent phosphorylation of ACSS2 S659 exposes a nuclear localization sequence in the protein33. Further, ACSS2 is regulated downstream of O-linked N-acetylglucosamine transferase (OGT), an enzyme that mediates the O-linked N-acetylglucosamine post-translational modification, which is increased in many tumours and implicated in promoting tumour growth77. OGT promotes CDK5-dependent phosphorylation of ACSS2 at S267, resulting in its stabilization and subsequent stimulation of acetate-driven DNL78.

ACC enzymes are also highly regulated by post-translational modification. ACC1 and ACC2 are phosphorylated by AMPK during nutrient deprivation-induced energy stress or in response to excess fatty acid availability, inhibiting the conversion of acetyl-CoA to malonyl-CoA and thereby attenuating lipid synthesis and stimulating fat oxidation79,80. Mice expressing mutant forms of ACC1 and ACC2 and that are resistant to phosphorylation and inhibition by AMPK owing to alanine mutations at its phosphorylation sites exhibit elevated hepatic DNL and enhanced hepatic tumour formation when challenged with the chemical carcinogen diethylnitrosamine81,82. ACC2 is also hydroxylated and activated by the 2-oxoglutarate-dependent prolyl hydroxylase PHD3 to suppress fatty acid oxidation83, and low PHD3 expression has been shown to promote dependence on fatty acid oxidation in acute myeloid leukaemia cell lines83. Finally, ACC enzymes are allosterically activated by citrate and inhibited by fatty acyl-CoA esters84.

Acetyl-CoA pathways in tumorigenesis

Alterations in acetyl-CoA metabolism have been shown to contribute to tumorigenesis through several pathways, including the mevalonate pathway, DNL and protein acetylation. Here we discuss the evidence for the importance of each pathway in cancer and how the dysregulation of acetyl-CoA metabolism might enforce or support elevated pathway activity.

Lipid, sterol and isoprenoid synthesis

Substantial preclinical evidence points to the mevalonate pathway as a vulnerability in cancers dependent on this pathway7. Specific cancer genetic alterations have been shown to promote mevalonate pathway dependence, including TP53 loss in HCC, TP53 mutation in breast cancer and t(4;14) chromosomal translocation in multiple myeloma85,86,87. Synthesis of the mevalonate pathway intermediate HMG-CoA is extremely sensitive to cytosolic acetyl-CoA production88,89, and the anticancer effects of targeting acetyl-CoA metabolism might be mediated through the mevalonate pathway in some contexts. For example, in a KRAS-G12D-driven murine model of pancreatic cancer, deletion of Acly in the pancreas impedes tumour formation. Acinar-to-ductal metaplasia, a wound healing response co-opted by mutant KRAS to promote tumour formation90, is inhibited in ex vivo assays by treatment with statins — which inhibit the conversion of HMG-CoA to mevalonate — or ACLY deficiency, and cholesterol supplementation rescues acinar-to-ductal metaplasia in statin-treated cells67. Statin treatment is also antiproliferative in pancreatic cancer cell lines, an effect that is reversible on addition of mevalonate or the diterpenoid mevalonate pathway intermediate geranylgeranyl pyrophosphate67,91.

Although statins are generally insufficient as single-agent anticancer therapeutics, a number of combination strategies have been proposed to enhance their efficacy7. For example, statin treatment can drive the activation of a feedback loop involving compensatory activation of SREBP2 and upregulation of the statin target HMGCR, and blocking this response has been shown to enhance statin-induced apoptosis and suppress xenograft tumour growth92,93. Statins have been used in different combination therapies in clinical trials, with some studies reporting benefits94. The potential for statins as anticancer agents is discussed in depth in recent reviews7,94. Further work is needed to identify optimal contexts for statin use in cancer and understand the extent to which targeting acetyl-CoA metabolic enzymes could mirror or improve on statin effects.

Fatty acids, either de novo synthesized or imported from the circulation, have many potential roles in cancer cells, including as structural components of membranes, as signalling molecules and as fuel for energy production. DNL is upregulated across many cancer types and has been of substantial interest as a therapeutic target6,12. Inhibiting the conversion of glucose or acetate to fatty acids through targeting ACLY or ACSS2, respectively, reduces the growth of multiple types of cancer in mice45,50,95,96,97,98. Similarly, targeting fatty acid synthase (FASN) exerts anticancer effects in some preclinical models6, and the FASN inhibitor TVB-2640 is currently being tested as an anticancer therapy in clinical trials (NCT03808558, NCT02223247, NCT02980029, NCT03032484, NCT03179904 and NCT05118776)99,100,101,102,103,104.

The use of acetyl-CoA in both the DNL pathway and the mevalonate pathway, and thus the enzymes that produce it, might have key roles in defending against oxidative stress and ferroptosis in cancer cells (Fig. 3). Ferroptosis is an iron-mediated mechanism of cell death driven by the reactive oxygen species-dependent peroxidation of polyunsaturated fatty acids (PUFAs) in cell membranes105. Several ferroptosis defence mechanisms are active in cancer cells106, including the GPX4 glutathione peroxidase pathway, which catalyses the reduction of lipid hydroperoxides, and the coenzyme Q (CoQ) oxidoreductase fibroblast-specific protein 1 (FSP1)107. CoQ — also known as ubiquinol in its fully reduced form (CoQH2) — is also thought to inhibit ferroptosis, in addition to its well-known mitochondrial function as an electron carrier in the electron transport chain. CoQ inhibits ferroptosis through the reduction of toxic lipid peroxide intermediates, and FSP1 and dihydroorotate dehydrogenase (DHODH) regenerate ubiquinol by promoting the NADH-dependent and flavin mononucleotide (FMNH2)-dependent reduction of its oxidized form, respectively. CoQ contains a redox-active head group connected to an isoprenoid tail that is synthesized from acetyl-CoA, and therefore inhibitors of acetyl-CoA synthesis may affect CoQ availability and function in ferroptosis defence107,108. Notably, statin treatment, which inhibits the mevalonate pathway, was shown to reduce CoQ abundance and trigger the compensatory upregulation of NRF2 in pancreatic cancer cells, and targeting this compensatory pathway in conjunction with statins induced oxidative stress and cell death91. The accumulation of the triterpenoid squalene, which branches from the mevalonate pathway to produce cholesterol and other sterols, has also been shown to protect against ferroptosis in lymphomas with loss of squalene monooxygenase109. In a parallel acetyl-CoA-driven pathway, de novo synthesized saturated and monounsaturated fatty acids can replace reactive oxygen species-sensitive PUFAs in membranes to reduce the overall susceptibility of the membrane to lipid peroxidation and thus reduce the susceptibility of cells to ferroptosis110,111. Thus, the studies described above suggest that strategies aimed at inhibiting acetyl-CoA use in lipid and isoprenoid synthesis might promote sensitivity to ferroptosis in cancer cells. For an in-depth discussion of ferroptosis and its roles in cancer, the reader is referred to recent reviews106,112.

Fig. 3: Acetyl-CoA pathways have roles in ferroptosis protection.
figure 3

Ferroptosis is a form of regulated cell death driven by the iron-dependent phospholipid peroxidation of membrane polyunsaturated fatty acids (PUFAs) to produce toxic phospholipid peroxides (PLOOH). If phospholipid peroxides are not neutralized by ferroptosis defence mechanisms, they can propagate and damage cell membranes, inducing cell death. Many cancer cells, owing to their unique metabolic properties and propensity for producing high concentrations of lipid peroxide-generating reactive oxygen species, must engage ferroptosis defence mechanisms to survive. Therefore, ferroptosis may be a targetable vulnerability in many tumour types. Several pathways protect against ferroptosis by reducing lipid peroxides. The major defence pathway is mediated by the selenoenzyme glutathione peroxidase 4 (GPX4), which uses glutathione (GSH) to reduce and detoxify phospholipid peroxides. Additional protection is mediated by squalene and the antioxidant function of ubiquinol (CoQH2), both of which are produced from acetyl coenzyme A (acetyl-CoA) via the mevalonate pathway; ferroptosis suppressor protein 1 (FSP1) and dihydroorotate dehydrogenase (DHODH) have been show to function in this context by reducing coenzyme Q (CoQ) to ubiquinol (CoQH2) at different locations in the cell. Saturated fatty acids (SFAs) and monounsaturated fatty acids (MUFAs) are resistant to oxidation, and thus replacing PUFAs with de novo synthesized SFAs and MUFAs that are made from acetyl-CoA precursors might reduce ferroptosis sensitivity. Long-chain acyl-CoA synthetase 3 (ACSL3) converts long-chain fatty acids into acyl-CoA esters. Ferroptosis regulators for which inhibitors are being considered are indicated with an asterisk. Currently, the most widely studied ferroptosis inducers inhibit either solute carrier family 7 member 11 (SLC7A11), which controls cystine uptake and glutathione production (for example, the inhibitor erastin and its analogues) or GPX4 (such as RSL3, ML162 and ML210)106. DNL, de novo lipogenesis; GSR, glutathione disulfide reductase; HMGCR, 3-hydroxy-3-methylglutaryl-CoA reductase. FMN, flavin mononucleotide; FMNH2, reduced flavin mononucleotide.

Full size image

Protein acetylation

The use of acetyl-CoA as the substrate for the acetylation of histones and other proteins is emerging as an important contributor to tumorigenesis113. Oncogenic metabolic reprogramming and exogenous cues such as growth factors, adipokines and microenvironmental stimuli can increase acetyl-CoA availability through effects on metabolic enzyme expression or activity, and changes in acetyl-CoA availability have been linked to the regulation of gene expression in several studies114,115. A key question is how the regulation of specific genes is achieved by changes in the availability of acetyl-CoA, with two predominant mechanisms being implicated: the regulation of transcription factors (for example, by acetylation) and the compartmentalization of acetyl-CoA production within the nucleus to regulate histone acetylation at specific genomic loci13. Recent work has demonstrated that the profile of short-chain and medium-chain acyl-CoA molecules is distinct between the nucleus and the cytosol89, highlighting the potential importance of compartmentalized acetyl-CoA production.

The transcriptional regulation of DNL has been linked to acetyl-CoA availability and metabolically sensitive histone acetylation in cancer cells, suggesting that an increase in the amount of lipogenic acetyl-CoA in the cytosol is coordinated with its ability to act as a gene regulation signal through histone acetylation (Fig. 4). For example, acetate availability and presumably its conversion to acetyl-CoA was shown to promote the expression of the DNL enzymes ACC1 and FASN in HCC cells, correlating with increased histone acetylation at the promoters of the corresponding genes, while simultaneously feeding fatty acid synthesis directly116. Similarly, in a study in prostate cancer cells, acetyl-CoA production by a nuclear PDC promoted histone acetylation at SREBP target genes; mitochondrial PDC was simultaneously needed in this model to supply citrate to be exported to the cytosol for production of lipogenic acetyl-CoA by ACLY, thereby allowing concurrent regulation of lipogenic gene expression by nuclear PDC and production of cytosolic acetyl-CoA as a lipogenic substrate38. ACLY-dependent histone acetylation downstream of lactate metabolism in patient-derived glioblastoma cells was also recently implicated in regulation of gene expression related to mitochondrial metabolism117.

Fig. 4: Metabolic regulation of histone acetylation impacts the phenotypes of both cancer cells and non-malignant cells in tumours.
figure 4

In addition to its direct roles in metabolism, acetyl coenzyme A (acetyl-CoA) is used for the modification of proteins through acetylation. Nutrient-sensitive histone acetylation has been linked to the regulation of gene expression in different cell types, potentially impacting tumour progression. In cancer cells, acetyl-CoA availability for histone acetylation has been linked to gene expression related to lipid synthesis, proliferation and invasive properties. In macrophages, polarization towards the immune-suppressive M2 phenotype is facilitated by ATP–citrate lyase (ACLY) correlating with histone acetylation and regulation of genes such as ARG1. In T cells, production of interferon-γ (IFNγ) is responsive to acetyl-CoA production, impacting antitumour immunity. The net effect of targeting acetyl-CoA metabolic enzymes might depend on the specific effects on multiple cell types in the tumour microenvironment.

Full size image

The control of cell migration, epithelial–mesenchymal transition (EMT) and metastasis has also been linked to nutrient-sensitive acetyl-CoA production through protein acetylation. For example, leptin-dependent AMPK activation and ACC1 phosphorylation was shown to trigger acetyl-CoA accumulation and the acetylation of SMAD2, promoting EMT in human and murine breast cancer cells118. In human HCC cell lines, loss of the acetyl-CoA hydrolase ACOT12 boosted acetyl-CoA levels and stimulated expression of the transcription factor TWIST2 to promote EMT, correlating with increased histone acetylation at the TWIST2 locus119. Finally, high acetyl-CoA abundance has been associated with elevated H3 K27 acetylation at genes associated with cell adhesion and migration in human glioblastoma cells, in a manner linked to the Ca2+-dependent activation of the transcription factor NFAT1120. These data show that mechanisms increasing acetyl-CoA availability are associated with the regulation of transcription factors to promote cancer progression in different contexts.

Acetyl-CoA in cells of the TME

The roles of acetyl-CoA metabolism in cells within the TME are beginning to be elucidated and will be important to consider in any strategy that targets acetyl-CoA metabolic enzymes. Thus, this section discusses emerging concepts in acetyl-CoA biology in macrophages and T cells, which are key components of the TME.

Acetyl-CoA metabolism in macrophages

Macrophages in the TME, also known as tumour-associated macrophages (TAMs), can take on different phenotypes depending on exogenous stimuli, which can allow them to either promote or oppose tumour growth. Although it is now appreciated that TAM phenotypes are complex and incompletely described by conventional M1 and M2 macrophage phenotypes121, M2-like macrophage polarization is generally associated with immune suppression, and ACLY has been shown to facilitate acquisition of the M2 phenotype through both inhibitor and gene knockout experiments69,122,123,124 (Fig. 4). In a mouse model of tumour growth in which bone marrow-derived macrophages exposed to M2-polarizing stimuli were co-injected with Lewis lung carcinoma cells, wild-type bone marrow-derived macrophages promoted tumour growth whereas ACLY-deficient bone marrow-derived macrophages did not123. In another study, however, the growth of tumours implanted into mice with myeloid-specific Acly knockout was equivalent to that in wild-type mice, despite differences in TAM phenotypes124. Finally, activation of macrophages by CpG DNA, which leads to a macrophage phenotype distinct from either M1 polarization or M2 polarization, stimulates the phagocytosis of tumour cells and suppresses tumour growth in a manner suppressed by inhibitors of fatty acid oxidation and ACLY125. Overall, these data suggest that ACLY might facilitate different functions of macrophages and highlight that a more comprehensive understanding of the role of ACLY in macrophages within the TME is needed.

Acetyl-CoA metabolism in T cells

Substantial evidence points to crucial roles for acetyl-CoA metabolism in T cell biology. The activation of T cells increases their glucose use, a process that is required to increase production of the cytokine interferon-γ (IFNγ) through the regulation of its transcription, in a manner correlating with increased histone acetylation, as well as its translation126,127,128 (Fig. 4). Because IFNγ production is thus sensitive to glucose availability, this establishes a competition for glucose between T cells and tumour cells in the microenvironment that might restrict T cell function (although this notion was recently challenged by in vivo studies using [18F]fluorodeoxyglucose positron emission tomography tracers that demonstrate preferential uptake of glucose into CD45+ immune cells over cancer cells129). The suppression of T cell IFNγ production by glucose restriction in vitro can be rescued by acetate supplementation in an ACSS2-dependent manner130, consistent with acetyl-CoA being a critical mediator of glucose-dependent gene regulation in T cells. In alignment, IL-12 stimulation boosts acetyl-CoA production and IFNγ production in T cells exposed to tumour-conditioned medium, and the treatment of CD8+ T cells with IL-12 or high concentrations of pyruvate increased the antitumour activity of these cells, in an ACLY-dependent manner, upon adoptive transfer into tumour-bearing mice128. In addition to regulation of histone acetylation and gene expression, conditional CD8+ T cell-specific Acaca deletion showed that these cells also require acetyl-CoA use for DNL to proliferate131. Acetyl-CoA metabolic enzymes are therefore emerging as critical regulators of immune cell function in the TME. The contribution of acetyl-CoA metabolism in other cells of the TME, such as fibroblasts, adipocytes and B cells, remains understudied.

Tissue-specific acetyl-CoA regulation

A key question of growing importance in cancer biology research is how tumour metabolism is influenced by the metabolism of its cell or tissue of origin132. Thus, there is a strong rationale for understanding normal acetyl-CoA regulation in cell and tissue lineages, especially those from which cancer cells arise. Recent studies have yielded critical information regarding the tissue-specific requirements for enzymes involved in acetyl-CoA metabolism, which is important for informing on how pharmacological inhibitors of these pathways could impact metabolic homeostasis. First, we focus on data from the liver and pancreas, as difficult-to-treat tumours arise in these tissues and studies using genetic models have revealed substantial information on acetyl-CoA metabolism at these sites. We then discuss adipose tissue, also because of the availability of studies of acetyl-CoA metabolism and because adipose tissue activity can impact tumour growth both locally and remotely.

Liver

Non-alcoholic fatty liver disease (NAFLD) has become a leading risk factor for HCC, which currently accounts for the fourth-highest number of cancer deaths worldwide. The incidence of HCC has been rising in recent decades, and the percentage of HCC cases linked to NAFLD has risen dramatically in particular (for example, the incidence of HCC cases arising from NAFLD rose from 10% to 35% between 2000 and 2017 in the UK)133. NAFLD has become a widespread condition affecting about a quarter of the world’s population and the majority of people with obesity or type 2 diabetes mellitus133, and thus understanding mechanisms linking obesity and NAFLD to HCC is of high clinical importance (Box 2).

One metabolic commonality between NAFLD and HCC that supports tumour growth is elevated DNL48,134,135,136,137. Fructose, a potent stimulator of hepatic DNL and a major component of the modern diet138, promotes NAFLD and can potentiate HCC tumour growth in mice139. Defining sources of lipogenic acetyl-CoA and their contributions to DNL could therefore inform strategies to prevent or treat NAFLD and HCC. As ACLY links carbohydrate and lipid metabolism through the conversion of citrate to acetyl-CoA (Fig. 1), it might be anticipated that loss of hepatic ACLY would protect against toxic lipid accumulation, particularly driven by diets high in fructose. However, knockout of Acly in hepatocytes had no effect on hepatic triglyceride levels or on rates of DNL as assessed by D2O tracing, either under standard chow-fed conditions or when mice were given sweetened drinking water containing a 50% glucose and 50% fructose mixture24. This result was attributed to the upregulation of ACSS2 following Acly loss in the context of high fructose consumption, and the conversion of gut microbiota-derived acetate to acetyl-CoA by ACSS2 in the liver24. Notably, stable-isotope tracing in conjunction with silencing of Acss2 or treatment with antibiotics showed that ACSS2 is required for fructose-dependent DNL if the fructose is consumed rapidly as a bolus, as fructose reached and was metabolized directly to acetate by the gut microbiota in this context. Gradual consumption of fructose limited its direct access to the microbiota and led to flexible use of acetyl-CoA generated in the liver through either ACLY or ACSS2, such that ACLY and ACSS2 had to be targeted simultaneously to suppress DNL24. These findings are consistent with the findings of recent stable-isotope infusion studies in mice, which demonstrated that the liver predominantly uses lactate — which would presumably enter the lipogenic acetyl-CoA pool through ACLY — and acetate — which would depend on ACSS2 — to supply acetyl-CoA for fatty acid synthesis140. Intriguingly, in mice made obese through a high-fat diet, Acly deficiency surprisingly resulted in elevated DNL and increased levels of hepatic lipids compared to wild-type mice, which was attributed to upregulation of SREBP1c141. These studies thus suggest no benefit or even adverse effects with respect to hepatic lipid accumulation upon targeting ACLY.

Nevertheless, targeting ACLY has been found to be beneficial in reducing the levels of hepatic lipids in other studies, likely owing to use of different dietary models. In a mouse model of non-alcoholic steatohepatitis (NASH) involving high-fat, high-fructose feeding and thermoneutral conditions, mice with hepatocyte-specific knockout of Acly or treated with the ACLY inhibitor bempedoic acid showed reduced hepatic lipid accumulation, suggesting that ACLY plays a dominant role in mediating fructose-dependent lipid accumulation in this model142. Interestingly, the effect of bempedoic acid on hepatic lipids was stronger than that of Acly deletion — potentially because of the activity of bempedoic acid in cells other than hepatocytes such as hepatic stellate cells142 or the effects of bempedoic acid independent of ACLY — and further work is needed to more comprehensively define the mechanisms through which bempedoic acid reduces NASH. Consistent with this study, genetically obese db/db mice (db is also known as Lepr) have reduced lipid accumulation in the liver upon Acly silencing143. Thus, studies using different models and dietary contexts have reached different conclusions about the role of ACLY in hepatic lipid accumulation, suggesting a complex interplay. Further work is needed to understand the mechanisms through which diet, the microbiota and other environmental factors, such as temperature, impact the dependence of DNL in the liver on ACLY versus ACSS2.

In contrast to whole-body Acly-knockout mice, which are not viable as ACLY is required for embryonic development144, whole-body Acss2-knockout mice are viable and phenotypically normal when fed a laboratory chow diet32, and have reduced tumour burden in a liver cancer model50. These findings suggest a promising therapeutic window for ACSS2 inhibitors. Acss2-knockout mice are resistant to obesity and hepatic steatosis when fed a high-fat diet, which is associated with lipid metabolic reprogramming in several tissues, including the liver32. Tissue-specific knockout models will be useful in elucidating the direct roles of ACSS2 in cancer cells versus anticancer effects exerted by changes in the systemic metabolism or other non-malignant cell types.

ACC is necessary for committing acetyl-CoA generated by either ACLY or ACSS2 to DNL. Although whole-body Acaca-knockout mice die embryonically145, mice with liver-specific knockout of Acaca are viable and do not have impaired malonyl-CoA production or DNL in the liver owing to the compensatory upregulation of ACC2, suggesting that malonyl-CoA pools can be used flexibly146. Liver-specific deletion of both Acaca and Acacb led to reduced levels of liver triglycerides, but elevated levels of plasma triglycerides, in mice fed chow, high-fat and Western diets; this differential regulation of liver and plasma triglycerides was also observed in rodents and humans given a liver-targeted ACC inhibitor147. Mechanistically, the elevation in the levels of plasma triglycerides was found to be caused by a reduction in the synthesis of PUFAs from essential fatty acids — a process dependent on malonyl-CoA — which triggered the upregulation of SREBP1c expression and the secretion of very low density lipoproteins147. Other studies have similarly reported reduced hepatic lipid levels following ACC inhibition or treatment with antisense oligonucleotides targeting ACC1 and ACC2 in rodent models81,148,149, although one study reported the opposite150. Reciprocally, loss of the ability to suppress ACC by phosphorylation led to hepatic lipid accumulation82. Altogether, the preponderance of evidence supports a model in which inhibiting ACC reduces the levels of hepatic lipids.

Elevated DNL in HCC and the association of HCC with NAFLD and NASH make targeting ACC attractive for this cancer, although different studies investigating this have reached different conclusions. One study of mice exposed to the carcinogen diethylnitrosamine showed that mice lacking Acaca and Acacb had elevated tumour formation over wild-type mice, mechanistically implicating the upregulation of antioxidant defences151. In another study, however, diethylnitrosamine-induced liver tumorigenesis was enhanced by ACC activation and reduced by ACC inhibition81, aligning with a pro-tumour role for DNL. More work is needed to understand the mechanisms underlying these different results. Cumulatively, the data point to ACC-dependent DNL as important in both experimental models and humans for hepatic lipid accumulation, making it a target deserving further investigation for its potential in combatting HCC.

Pancreas

A unique feature of the pancreas is that acetyl-CoA pools in acinar cells are heavily derived from the BCAA leucine67, which produces three molecules of acetyl-CoA when catabolized152 (Fig. 1). Infusion of 13C-labelled BCAAs in mice has revealed that the TCA cycle in the pancreas is heavily fed by BCAAs153. This is interesting as acinar cells are a potential cell type of origin for pancreatic ductal adenocarcinoma and plasma BCAA levels are reportedly elevated in individuals who develop pancreatic cancer, even years preceding their initial diagnosis154. KRAS mutation, which is observed in nearly all cases of human pancreatic ductal adenocarcinoma, drives an increase in acinar cell acetyl-CoA abundance in an ACLY-dependent manner, and consistently, Acly deletion in mice reduces tumour formation without impacting normal pancreatic endocrine or exocrine function67. Similarly, pancreas-specific deletion of BCAT2, a protein that mediates the transamination of BCAAs preceding their mitochondrial catabolism, also protects against pancreatic ductal adenocarcinoma155, further suggesting a role for pancreatic BCAA utilization in early stages of pancreatic tumorigenesis. Although ACLY and BCAT2 have emerged as metabolic factors that are needed for efficient tumour formation, the molecular mechanisms through which BCAA and acetyl-CoA metabolism impact pancreatic tumorigenesis largely remain to be defined.

Adipose tissues

In general, adipose tissues come in two varieties: energy-storing white adipose tissue and thermogenic brown adipose tissue, both of which secrete endocrine signals that influence systemic metabolism156. Adipose tissues have roles in tumour growth through several mechanisms, including providing fatty acids to cancer cells as an energy source, secreting growth-promoting adipokines, modulating systemic insulin sensitivity and glucose homeostasis157, and contributing to the TME through tumour-induced dedifferentiation and remodelling into myofibroblasts and macrophage-like cells158. Interestingly, the activation of brown adipose tissue was recently shown to suppress tumour growth in mice by increasing the uptake of glucose in this tissue and limiting glucose availability to tumours159. That study also showed that cold exposure in a human patient reduced tumour uptake of [18F]fluorodeoxyglucose159. Therapeutically, stimulating brown fat is under intense investigation as an anti-obesity strategy, and these new data suggest an unexpected link between non-shivering thermogenesis and tumour growth that might have therapeutic implications.

Recent studies have identified key roles for acetyl-CoA metabolism in the functioning of both brown adipose tissue and white adipose tissue. In an in vitro model of brown adipocyte differentiation, ACLY was shown to be required for differentiation of precursors into mature brown adipocytes as well as lipogenesis in vitro68. This correlated with defective histone acetylation, and both differentiation and lipogenesis were partially rescued by acetate supplementation68. Interestingly, conditional knockout of Acly in fully differentiated, mature brown adipocytes results in the adoption of a whitened phenotype in these cells and abnormal lipid accumulation in vivo68. Deleting Acly in all adipocytes resulted in systemic impairments in glucose handling, with mice exhibiting reduced white adipose mass, accumulation of hepatic lipids and the development of insulin resistance when challenged with high-carbohydrate diets160. This is consistent with other evidence that adipose DNL plays important roles in whole-body insulin sensitivity161. Whether adipocyte ACLY’s roles in regulating systemic metabolism impact tumour progression is unclear, although cancer cells can locally reprogramme adipocytes in the TME, including repressing the lipogenic programme and/or differentiated phenotype, to promote tumour progression158,162,163.

Targeting acetyl-CoA metabolic enzymes

ACLY

Interest in ACLY as a therapeutic target has existed for decades, beginning with the identification of (−)-hydroxycitrate (HC), an ACLY inhibitor that was first extracted from the tropical fruit Garcinia cambogia in the 1960s164. HC is available over the counter as a dietary supplement and has attracted interest as a weight loss agent on the basis of favourable metabolic effects in both rodents and humans165,166. However, a randomized controlled trial failed to find effects of HC in promoting weight loss or fat loss167. HC has also been investigated as a calorie-restriction mimetic due to its ability to promote autophagy, which occurs downstream of the depletion of acetyl-CoA and acetyllysine on HC treatment168,169. Caloric restriction has been shown to reduce tumour growth170,171,172, suggesting the potential for calorie-restriction mimetics as cancer treatments. One study showed HC increased immunogenic chemotherapy efficacy in a fibrosarcoma subcutaneous allograft model in an autophagy-dependent manner173 as the effectiveness of such chemotherapy regimens depends on tumour cell autophagy and ATP release from the tumour cell for immune cell recruitment to the tumour174. It should be noted that that context is likely to be critical for these effects as autophagy also contributes to immune evasion and substantial evidence indicates that inhibiting autophagy can exert anticancer effects175,176.

ACLY inhibitors other than HC have been shown to suppress cancer cell proliferation and tumour growth in mice75,96,177,178. Evidence for the efficacy of ACLY inhibitors to date has mainly been derived from xenograft studies in mice. The ACLY inhibitor SB-204990179 was shown to suppress the growth of grafted human and murine tumours75,96. Another inhibitor, BMS-303141, suppressed HepG2 xenograft tumour growth when combined with sorafenib177. However, these inhibitors, as well as other inhibitors evaluated in preclinical studies focused on metabolic diseases, have not progressed to clinical trials, at least in part due to poor bioavailability and low target specificity180. The only ACLY inhibitor currently in clinical use is bempedoic acid (Nexletol), which is well tolerated and has been approved by the US Food and Drug Administration approved for treatment of familial hypercholesterolaemia and cardiovascular disease106. Bempedoic acid is a prodrug that is converted to the active molecule bempedoyl-CoA by the hepatic enzyme ACSVL1, and thus acts specifically in the liver181. The expression of ACSVL1 in a subset of human HCC tumours182 suggests that bempedoic acid could potentially be used against HCC. Promisingly, a recent study reported suppression of tumour growth by bempedoic acid treatment in a carcinogen-induced mouse model of HCC, which was enhanced by combining bempedoic acid with anti-PDL1 therapy183. Bempedoic acid or ACLY silencing was also found to suppress metastasis in a colorectal cancer model184. More work is needed to understand whether the effects of bempedoic acid and other ACLY inhibitors on tumour growth occur through ACLY-dependent or ACLY-independent pathways; indeed, bempedoic acid is also known to activate AMPK185, which could well contribute to its anticancer effects186,187,188.

In sum, accumulating preclinical evidence supports ACLY as an attractive target for cancer treatment. The recent solving of the ACLY homotetramer structure by three separate groups opens exciting new possibilities for inhibiting this enzyme with small molecules189,190,191. One of these studies reported a novel ACLY inhibitor (NDI-091143) that acts allosterically189, although effects in cells or mice were not reported. In addition to the development of improved inhibitors, determining the necessity of ACLY for the normal functioning of cells and tissues of the adult mammal is needed to help identify potential toxic effects. The generation of inducible whole-body knockout models could help identify potential side effects associated with systemic ACLY inhibition in adults.

ACSS2

ACSS2 has also attracted interest as a therapeutic target, especially because Acss2-knockout mice are viable, suggesting a favourable therapeutic window32. ACSS2 expression is elevated in several cancer types, and ACSS2 loss or inhibition shows anticancer effects in mouse models of HCC, breast cancer and multiple myeloma45,47,50,98. These studies have implicated the roles of ACSS2 in regulating lipid synthesis and gene expression in promoting tumour growth. The ACSS2 inhibitor MTB-9655 has recently entered phase I clinical trials in patients with advanced solid tumours (NCT04990739).

ACC

ACC inhibition suppresses DNL, making it an attractive therapeutic target. Inhibiting ACC has anticancer effects in several malignancies, including those of the lung, liver and lymphoid tissue81,192,193. Several ACC inhibitors, including firsocostat, PF-05221304 and MK-4074, have entered clinical trials for NAFLD or NASH6,194. A possible caveat to targeting ACC, however, is that procancer effects have been reported in certain contexts. For example, ACC inhibition was shown to increase acetyl-CoA availability, leading to the acetylation of SMAD2 to promote EMT in a breast cancer model118. It may also impede certain therapeutic responses; for example, AMPK-dependent ACC inactivation was shown to reduce the synthesis of PUFA-containing lipids, promoting resistance to ferroptosis195. Moreover, ACC2 inhibition due to PHD3 loss or suppression promotes tumour growth in acute myeloid leukaemia and obesity-linked colon cancer models83,196. Thus, ACC inhibition holds the potential to suppress tumorigenesis, although it might not do so in every context, as noted above. For a comprehensive discussion of lipogenesis inhibitors, we refer the reader to an excellent recent review6.

Conclusions and perspectives

Acetyl-CoA metabolism has been a subject of therapeutic interest for decades, yet its mysteries are still being uncovered. As we have discussed throughout this Review, acetyl-CoA production and utilization are dependent on various factors, including nutrient availability, metabolite signals, hormonal and growth factor cues, and cell type. The available evidence shows that the inhibition of acetyl-CoA metabolic enzymes is differentially effective depending on these factors, and our current understanding of how nutrition might be optimally combined with acetyl-CoA pathway inhibitors is limited. Moreover, the contexts in which acetyl-CoA metabolic enzymes can compensate for one another are poorly defined. Further defining key fates of acetyl-CoA that support tumour growth and potential compensatory mechanisms and identifying synthetic lethal interactions using genetic or chemical screening approaches will be helpful in uncovering new targetable vulnerabilities. Finally, the roles of acetyl-CoA metabolism in non-malignant cell types, including those of the TME and those that regulate systemic metabolism, require continued investigation to better understand how they impact tumour growth.

Gaps still exist in our understanding of basic acetyl-CoA biology. Armed with a variety of new approaches to study localized metabolite pools197, researchers are just beginning to probe how different perturbations impact acetyl-CoA metabolism in different cellular compartments. In particular, the significance of acetyl-CoA pools in the endoplasmic reticulum and peroxisome remains poorly defined, although recent evidence suggests these carry out important biological functions10,198,199. Metabolic regulation has been observed for other acetyl metabolites, such as N-acetylaspartate, N-acetylcysteine and N-acetylglutamate200,201, as well as for RNA acetylation (N4-acetylcytidine)202, and these topics are highly interesting areas for further study. How acetyl-CoA-dependent enzymes interact with other acyl-CoAs is also emerging as another potentially important node of metabolic regulation as many acetyl-CoA-dependent enzymes also interact with CoA and short-chain and long-chain acyl-CoA molecules202,203, which may result in either the competitive inhibition of these enzymes or the use of these metabolites as alternative substrates (for example, some lysine acetyltransferases can mediate acylations beyond acetylation22). The field is now poised to make rapid progress in further elucidating the basic biological roles of acetyl-CoA and to hopefully translate these advances towards strategies to help patients with cancer.