Claudio R. Santos, Translational Cancer Therapeutics, Cancer Research UK London Research Institute, 44 Lincoln’s Inn Fields, London WC2A 3LY, UK Fax: +44 207 269 3094 Tel: +44 207 269 3529 E-mail: firstname.lastname@example.org
Lipids form a diverse group of water-insoluble molecules that include triacylglycerides, phosphoglycerides, sterols and sphingolipids. They play several important roles at cellular and organismal levels. Fatty acids are the major building blocks for the synthesis of triacylglycerides, which are mainly used for energy storage. Phosphoglycerides, together with sterols and sphingolipids, represent the major structural components of biological membranes. Lipids can also have important roles in signalling, functioning as second messengers and as hormones. There is increasing evidence that cancer cells show specific alterations in different aspects of lipid metabolism. These alterations can affect the availability of structural lipids for the synthesis of membranes, the synthesis and degradation of lipids that contribute to energy homeostasis and the abundance of lipids with signalling functions. Changes in lipid metabolism can affect numerous cellular processes, including cell growth, proliferation, differentiation and motility. This review will examine some of the alterations in lipid metabolism that have been reported in cancer, at both cellular and organismal levels, and discuss how they contribute to different aspects of tumourigenesis.
5′ adenosine monophosphate-activated protein kinase
flavin adenine dinucleotide (hydroquinone form)
fatty acid synthase
glucose transporter 1
3-hydroxy-3-methylglutaryl coenzyme A reductase/HMGCR-CoA reductase
insulin-like growth factor 1
mammalian target of rapamycin complex I
pyruvate dehydrogenase kinase
SREBP cleavage-activating protein
sterol regulatory element-binding protein
Deregulation of lipid metabolism in cancer
Most adult mammalian cells acquire lipids from the bloodstream either as free fatty acids or complexed to proteins such as low-density lipoproteins. These lipids are obtained from dietary sources or are carbohydrate-derived fatty acids synthesized in the liver or in adipocytes, where they can also be stored in intracellular structures called lipid droplets.
De novo fatty-acid biosynthesis in the adult organism occurs mainly in the liver, adipose tissue and the lactating breast. The acetyl groups for fatty-acid biosynthesis are provided mainly by citrate, which is produced by the tricarboxylic acid (TCA) cycle (Fig. 1). The conversion of citrate into acetyl-coenzyme A (acetyl-CoA) and oxaloacetate is catalysed by adenosine triphosphate (ATP)-citrate lyase. Oxaloacetate can be converted into pyruvate by malic enzyme. This reaction generates NADPH and, along with the NADPH-producing reactions in the pentose phosphate pathway, provides the reducing power for lipid synthesis.
The committed step of fatty-acid biosynthesis requires the activation of acetyl-CoA to malonyl-CoA. This is an energy-consuming process catalysed by acetyl-CoA carboxylase (ACC). The acetyl and malonyl groups are then coupled to the acyl-carrier protein domain of the multifunctional enzyme fatty-acid synthase (FASN). Repeated condensations of acetyl groups generate basic 16-carbon saturated fatty acids (palmitic acid).
Further elongation and desaturation of newly synthesized fatty acids takes place at the cytoplasmic face of the endoplasmic reticulum membrane. Fatty acids with longer chains, such as stearic acid, are obtained through the action of a family of enzymes (‘elongation of very-long-chain fatty-acid proteins’) that add two carbons to the end of the chain in each cycle of reactions. This family comprises seven members (ELOVL1–7) with different chain lengths and saturation specificities.
Desaturation is catalysed by fatty acyl-CoA desaturases, which include the stearoyl-CoA desaturases (SCDs). SCD1 introduces a double bond in the Δ9 position of palmitic and stearic acids to produce mono-unsaturated fatty acids. Fatty acyl-CoA desaturases catalyse the synthesis of highly unsaturated fatty acids from essential polyunsaturated fatty acids, which are mainly derived from the diet. Desaturation alters the physical properties of long-chain fatty acids, including those used for the synthesis of membrane phosphoglycerides, and is an important determinant of membrane fluidity. Fatty acids can also be used for energy storage in the form of triacylglycerides. These are composed of three fatty-acid chains of different chain length and saturation bound to a glycerol molecule via ester bonds.
It was noted, over 50 years ago, that neoplastic tissues are able to synthesize lipids  in a manner similar to embryonic tissues (in this issue of FEBS, Altam & Dang discuss the similarities between the metabolism of tumour cells and that of normal proliferating cells, using T cells as an example). In 1996, Kuhajda and colleagues showed that OA-519, a prognostic marker in breast cancer, corresponds to FASN . Since then, several studies have shown that tumour cells reactivate de novo lipid synthesis ( and references therein). Some cancers, including breast and prostate [4–6], show increased expression of FASN, which suggests that fatty-acid synthesis plays an important role in cancer pathogenesis . Furthermore, it has been shown that ATP-citrate lyase is required for cell transformation in vitro and for tumour formation in vivo [7,8], and chemical inhibition of ACC induces growth arrest and apoptosis in prostate cancer cells . The long-chain fatty-acid elongase ELOVL7 was shown to be overexpressed in prostate cancer and required for prostate cancer-cell growth, possibly because of its role in the synthesis of steroids, such as androgens , and although overexpression of elongases has not been reported in other tumours, it is worth noting that overexpression of oncogenic Ras has been shown to increase the levels of very-long fatty-acid chains [11,12], suggesting that they may play a yet-unknown role in transformation.
Another important biosynthetic process within lipid metabolism is the mevalonate pathway, which facilitates the synthesis of cholesterol (Fig. 1). The first steps of cholesterol biosynthesis involve the condensation of acetyl-CoA with acetoacetyl-CoA to form 3-hydroxy-3-methylglutaryl (HMG)-CoA. The reduction of HMG-CoA to mevalonate by HMG-CoA reductase (HMGCR) represents the rate-limiting reaction of the cholesterol synthesis pathway and is highly regulated. HMGCR is an endoplasmic reticulum-transmembrane protein and its stability is regulated by a sterol-sensing-domain that mediates its degradation under saturating sterol levels . Cholesterol is an important component of biological membranes as it modulates the fluidity of the lipid bilayer and also forms detergent-resistant microdomains called lipid rafts that coordinate the activation of some signal-transduction pathways . The cholesterol biosynthesis pathway also generates intermediates required for the isoprenylation of small GTPases, including the farnesylation of Ras and the geranyl-geranylation of Rho . Finally, sterols have an important role in organismal development as they form the structural backbone for the synthesis of steroid hormones.
Accumulation of cholesterol has been reported in prostate cancer  and deregulation of the mevalonate pathway has been associated with transformation . Interestingly, HMGCR is the target for a class of cholesterol-lowering drugs known as statins. Statins show antiproliferative activity in several cancer-cell lines, with the described effects ranging from cell cycle arrest (e.g. in breast cancer cells [18,19]) to apoptosis (e.g. in acute myeloid leukaemia ). Statins have also been shown to increase the sensitivity of colorectal cancer cells to chemotherapeutic agents through induction of epigenetic reprogramming , and indeed combination of statins with chemotherapy has shown promising results in clinical trials of patients with acute myeloid leukaemia  and hepatocellular carcinoma , among others. As millions of patients throughout the world are treated with statins to lower cholesterol, this has raised the question of whether its use may be associated with a decreased incidence of cancer. A multitude of epidemiological analyses, mainly retrospective, have been published in the last few years but a conclusive answer is yet to be found. The effect of statins on cancer incidence seems to be highly dependent on the tumour type and class of statins used. For example, several meta-analyses found either no effect or a small, nonsignificant trend towards a protective effect of statins against colon cancer [24–26]. Other studies found a significant protective effect in hepatocellular carcinoma  or in patients treated with a lipophilic statin after being diagnosed with breast cancer . There are currently a number of ongoing prospective studies that should help to finally resolve this question.
Triacylglycerides and cholesterylesters are stored in lipid droplets, highly ordered intracellular structures that are formed from the endoplasmic reticulum through a budding process . They are composed of a phospholipid and sterol monolayer containing specific proteins and a core of nonpolar lipids. Cancer cells seem to contain increased numbers of lipid droplets compared with normal tissue. This was observed in colon adenocarcinomas or intestinal epithelial cells after transformation with H-ras V12 . The same study demonstrated that cyclooxygenase 2 and prostaglandin synthase localize to lipid droplets in cancer cells, suggesting that these structures could be involved in regulating cancer pathogenesis .
Triacylglycerides provide a reservoir of fatty acids that can be mobilized for energy generation through the action of a series of lipases, such as hormone-sensitive lipase, adipose triglyceride lipase and monoacylglycerol lipase (MAGL). Cytoplasmic free-fatty acids are then coupled to CoA and the acyl chain is transferred to carnitine by carnitine acyltransferase to be transported into the mitochondrial matrix (Fig. 1). After entering the mitochondrial matrix, acyl chains are recoupled to CoA and degraded by repeated rounds of oxidation and hydration. This process is known as β-oxidation and produces NADH and flavin adenine dinucleotide (hydroquinone form) (FADH2) as well as acetyl-CoA, which can enter the TCA cycle to be completely oxidized. The mode of regulation of β-oxidation ensures that lipid synthesis and degradation are mutually exclusive. The activity of carnitine acyltransferase is inhibited by malonyl-CoA, produced by ACC during fatty-acid biosynthesis. The two isoforms of ACC – ACACA and ACACB – differ in their ability to be activated by citrate , and ACACB is thought to be the main isoform responsible for inhibiting β-oxidation. ACC is also a target for the 5′ adenosine monophosphate-activated protein kinase (AMPK), a heterotrimeric protein kinase that is activated in response to low energy levels and inhibits energy-consuming processes while promoting energy production . AMPK phosphorylates and inhibits both ACACA and ACACB, resulting in the inhibition of fatty-acid synthesis and the induction of β-oxidation .
There are a number of studies that link β-oxidation with cancer. Fatty-acid oxidation is a dominant pathway for energy generation in prostate cancer  and enhanced mitochondrial β-oxidation of fatty acids has been linked to tumour promotion in pancreatic cancer . Inhibition of β-oxidation induces apoptosis in leukaemia cells and in glioblastoma cells [36,37]. Pharmacological activation of β-oxidation can also rescue the glucose dependency of Akt-transformed cells , suggesting that this pathway can provide important metabolites for cancer-cell survival.
Activation of oncogenic pathways stimulates lipid synthesis
Most enzymes involved in fatty-acid and cholesterol biosynthesis are regulated by the sterol regulatory element-binding proteins (SREBPs) [39,40] (Fig. 1). SREBPs are transcription factors of the helix-loop-helix leucine zipper family. They are translated as 125-kDa precursors that are inserted into the endoplasmic reticulum membrane where they are bound by the SREBP cleavage-activating protein (SCAP). Three SREBP isoforms – SREBP1a, SREBP1c and SREBP2 – have been identified in mammalian cells . SREBP1a and SREBP1c are generated by alternative splicing and vary in their expression levels across different tissues, with SREBP1a being the most abundant isoform in most cultured cell lines . Although there is overlap between their target genes, SREBP1 mainly regulates fatty acid, phospholipid and triacylglycerol synthesis, while SREBP2 controls the expression of cholesterol-synthesis genes .
The activity of SREBPs is tightly regulated by the concentration of intracellular sterols. When sterol levels are low, the SREBP/SCAP complex can associate with COPII-coated vesicles and translocate to the Golgi where a two-step proteolytic cleavage releases the 65-kDa N-terminal transcriptionally active fragment. This mature protein can then enter the nucleus and regulate transcription by binding to sterol-regulatory elements within the promoter regions of SREBP target genes . When cellular cholesterol concentrations reach saturating levels, association of the SREBP/SCAP complex with COPII is inhibited as a result of binding of the insulin-induced gene, and the complex is retained in the endoplasmic reticulum . This classic model of sterol-dependent regulation applies mainly to SREBP2 and has been termed ‘regulated intramembrane proteolysis’ . It is conserved between flies and mammals. However, SREBP processing in Drosophila is regulated by phosphatidylcholine and phosphatidylethanolamine rather than by sterols . Interestingly, it has recently been shown that depletion of phosphatidylcholine in mammalian cells leads to nuclear accumulation of SREBP1, but not of SREBP2, even in the presence of cholesterol and through a SCAP-independent mechanism , suggesting that phospholipid levels may be the main regulators of SREBP1.
In addition to the regulation by proteolysis, the activity of SREBP transcription factors is modulated by their interaction with transcriptional co-activators such as p300 or cyclic adenosine monophosphate response element-binding-binding protein . SREBP can also associate with the activator-recruited co-factor/mediator complex to activate specific target genes . Furthermore, SREBPs carry a cdc4 phospho-degron motif and can be phosphorylated by glycogen synthase kinase 3, resulting in polyubiquitination and degradation of the mature protein [50, 51].
The Phosphoinositide 3-kinase/Akt/PKB (protein kinase B) signalling pathway is frequently activated in human cancer . Insulin stimulates lipid synthesis and ACC activity in liver and adipose tissue  and Akt can phosphorylate ATP-citrate lyase  and activate the expression of several genes involved in cholesterol and fatty-acid biosynthesis . One important downstream effector of Akt is mammalian target of rapamycin complex I (mTORC1), a multiprotein kinase involved in the regulation of several metabolic processes, including protein synthesis . The activity of mTORC1 is also regulated by specific amino acids (reviewed by Richard F. Lamb in this issue of FEBS). Interestingly, mTORC1 activity is required for the nuclear accumulation of mature SREBP1 in response to Akt activation , and transcriptional profiling of cells deficient for the tuberous sclerosis complex 1 or 2 genes, two negative regulators of mTORC1, revealed that SREBP is an important component of the metabolic regulatory network downstream of this signalling axis . mTORC1 also regulates the expression of SREB1F and is required for the stimulation of lipogenesis in the liver . SREBP function is also essential for Akt-dependent regulation of cell size, both in mammalian cells and in the developing wing of Drosophila melanogaster , suggesting that the Akt/mTORC signalling axis regulates protein and lipid synthesis in a concerted manner during cell growth.
SREBP function is also downstream of several tumour-suppressor pathways. AMPK, which is downstream of the liver kinase B1 tumour suppressor, can directly phosphorylate SREBP, thereby preventing its proteolytic activation . Loss of the retinoblastoma protein promotes the expression of genes involved in the isoprenylation of N-Ras through induction of SREBP1 and SREBP2 . Furthermore, mutant tumor protein p53 associates with SREBP at the promoters of genes within the mevalonate pathway and increases their expression . This hyperactivation disrupts tissue architecture and promotes the formation of breast cancer, placing SREBP-dependent lipogenesis at the core of the transformation process.
SREBP1 and SREBP2 are overexpressed in a number of cancers . SREBP1 is activated by aberrant epidermal growth factor receptor signalling in human glioblastoma multiforme, albeit independently of mTORC1 , and SREBP1-dependent induction of low-density lipoprotein receptor expression is crucial for the survival of these cancers . These findings demonstrate that activation of SREBP is an important function of oncogenic signalling pathways in cancer.
Hypoxia and lipid metabolism
Solid tumours frequently present hypoxic areas as a consequence of an increase in tumour volume that outgrows the capacity of its vascular network. Low oxygen availability leads to the activation of the hypoxia-inducible factors (HIFs), two heterodimeric transcription factors composed of an α-subunit (HIF1-α or HIF2-α), and a β-subunit. Under normoxia, HIF-1α and HIF-α are targeted by the oxygen-sensitive prolyl-hydroxylases and are marked for degradation by Von Hippel–Lindau (VHL) tumour suppressor-dependent ubiquitination . Mutations in VHL frequently occur in renal cell carcinomas and promote a pseudo-hypoxic state that leads to the stabilization of HIF1α and HIF2α, even in the presence of oxygen . HIFs can also be activated by oncogenic pathways  and by loss of p53 . Interestingly, metabolic activity can also contribute to HIF activity. Inactivating mutations in the TCA cycle enzymes fumarate hydratase or succinate dehydrogenase lead to the accumulation of succinate, which blocks the activity of prolyl-hydroxylases and results in the accumulation of HIF1α [69,70].
HIF activation not only promotes angiogenesis by inducing expression of the vascular endothelial growth factor , it also drives adaptation to the hypoxic environment through a metabolic switch to anaerobic energy production. HIF induces the expression of the glucose transporter 1 (GLUT1)  and several glycolytic enzymes . HIF also prevents the entry of pyruvate into the TCA cycle by inducing the expression of pyruvate dehydrogenase kinase 1 (PDK1), a kinase that phosphorylates and inhibits pyruvate dehydrogenase , thereby preventing glucose-derived lipid synthesis. However, it was demonstrated that HIF1 induces the expression of FASN in human breast-cancer cell lines and that FASN expression is increased in hypoxic tumour regions . Because the flow of carbon from glucose to fatty acids is attenuated by hypoxia, other carbon sources are required to support fatty-acid synthesis under these conditions. Indeed, acetyl-CoA synthetase 2, the bidirectional enzyme catalysing the synthesis of acetyl-CoA from cytoplasmic acetate, is induced by hypoxia and promotes cancer-cell survival under these conditions . More recently, three independent studies showed that glutamine becomes the major carbon source for lipid synthesis in the absence of functional mitochondria. These studies found that isocitrate dehydrogenase-1 can produce cytoplasmic citrate by reductive carboxylation of glutamine-derived α-ketoglutarate. This metabolic activity was found to be active in cancer cells with defective mitochondria  and under hypoxia [78,79].
The inhibitory effect of hypoxia on β-oxidation has been documented in different tissues. Ischaemia causes reduced β-oxidation in the heart by preventing the oxidation of NADH and FADH2 , and exposure of macrophages to hypoxic conditions results in enhanced storage of triacylglycerides . HIF1 has been reported to promote lipid accumulation through induction of the hypoxia-inducible protein 2, a protein involved in the deposition of neutral lipids into lipid droplets . HIF1 also promotes the uptake of free fatty-acids and the production of triacylglycerol in liver and adipose tissue through the induction of the peroxisome proliferator-activated receptor γ . Recently, it emerged that HIF2 is responsible for the changes in lipid metabolism observed upon loss of VHL in the liver . Liver-specific deletion of VHL in mice resulted in steatosis accompanied by increased lipid droplet formation and a reduction in the expression of β-oxidation genes. The same study showed that HIF2α also inhibits the expression of SREBP1c and its target genes in the liver . Interestingly, clear-cell renal carcinomas, which are characterized by loss of VHL and stabilization of HIF1, also show frequent accumulation of lipids . While the exact role of lipid droplets in supporting cancer-cell survival and/or tumour progression is not fully understood, it is possible that enhanced storage of triacylglycerides could be beneficial during conditions of intermittent hypoxia as they may be used as a readily available fuel source after reoxygenation.
Whole-body lipid metabolism and cancer
Certain changes in lipid metabolism can contribute to the predisposition of obese patients to cancer development. Obesity is associated with an increased disease risk for several cancer types. Current estimations are that 20% of all tumours and 50% of endometrial and oesophageal cancers can be attributed to obesity . Obesity contributes to increased cancer risk mainly by causing acquired insulin resistance. The accumulation of lipids in muscle and liver lead to increased availability of intracellular diacylglycerol and ceramide, which impairs insulin signalling and inhibits insulin-induced glucose uptake (reviewed in ). This leads to increased secretion of insulin by pancreatic beta cells, and enhances the availability of insulin-like growth factor 1 (IGF1) through reduced production of insulin-like growth factor-binding proteins 1 and 2  (Fig. 2A). Insulin and IGF1 are both pro-tumourigenic growth factors that stimulate proliferation and can protect cells from apoptosis (reviewed in ). In addition, insulin resistance is also induced by chronic low-grade inflammation . Mouse models of diet-induced or genetically induced obesity have shown that the development of hepatocellular carcinoma is dependent on the production of inflammatory cytokines .
In contrast, dietary restriction is thought to have an anti-tumourigenic effect. However, the extent of this may depend on the tissue of origin and the genetic background of the cancer cells. A study has show that human cancer-cell lines that display constitutive activation of the PI3K pathway are resistant to the growth-inhibitory effects of dietary restriction in mouse xenografts . Dietary restriction reduces the levels of circulating insulin and IGF1  and it is likely that constitutive activation of PI3K renders tumours independent of their growth-promoting effect.
Alteration in lipid metabolism can also be a consequence of cancer development as part of a disease known as cancer cachexia (Fig. 2B). Cancer cachexia is a wasting syndrome associated with extreme weight loss and physical decline that is frequently observed in cancer patients and leads to considerable morbidity. Cachexia is characterized by loss of skeletal muscle, with or without loss of adipose tissue, and can be associated with anorexia, inflammation and insulin resistance . Cachectic patients show metabolic alterations that include elevated carbohydrate utilization, protein degradation and reduction in fat stores. The reduction in fat stores is believed to be mainly caused by increased lipolysis in adipose tissue, rather than by a reduction in lipid biogenesis [95,96], and leads to a reduction in adipocyte cell volume but not of adipocyte cell number .
One of the mechanisms causing increased lipolysis in patients with cachexia is the enhanced expression of the hormone-sensitive lipase in adipocytes . Increased expression of the cell death-inducing DNA fragmentation factor-alpha-like effector A has also been observed in adipose tissue of cachectic patients . It attenuates glucose oxidation in vitro by blocking the entry of pyruvate into the mitochondria through induction of PDK1 and PDK4 . This promotes the oxidation of fatty acids and could contribute to the loss of adipose tissue. This inhibition of pyruvate dehydrogenase may explain why glucose administration does not suppress fatty-acid oxidation in cachectic patients .
While the mechanism that leads to increased lipolysis in the adipose tissue of cancer patients has been partially unravelled, the crucial link that dictates how tumours induce these changes in adipocytes remains elusive. However, the metabolic changes elicited during cachexia can promote tumour growth by fuelling the metabolism of cancer cells. Increased levels of circulating free fatty-acids, monoacylglycerides and diacylglycerides have been observed in cachectic ovarian cancer patients . Glycerol molecules released during the degradation of triacylglycerides can be used for gluconeogenesis by the liver, while free fatty-acids may provide the tumour with energetic or biosynthetic substrates or signalling molecules .
Lipid metabolism contributes to the transformed phenotype of cancer cells
Owing to the diversity of their biological roles, lipids contribute to several aspects of tumour biology, such as growth, energy and redox homeostasis, as well as to the dissemination of cancer cells to form distant metastases. Some of the potential functions of altered lipid metabolism in cancer cells are discussed below (see also Fig. 3).
Cell growth and proliferation
The high proliferation of cancer cells requires large amounts of lipids as building blocks for biological membranes. Lipid synthesis has been shown to be required for cell growth, following treatment with interleukin 3 or in response to activation of Akt in cultured mammalian cells [7,57]. Interestingly, SREBP function was also required for the maintenance of cell size and organ size in D. melanogaster , indicating that the importance of lipogenesis for growth is conserved. SREBP activity is regulated during mitosis, suggesting that the expression of lipogenic genes is required for cell-cycle progression .
Overexpression of SCD, a target gene of SREBP, has been observed in oncogene-transformed cells  and in several human cancers [105,106]. SCD is associated with genetic predisposition to cancer in mice , and is required for cell transformation in vitro  and for the growth of prostate cancer cells in vivo . Stable silencing of SCD or disruption of the Scd gene in mice, blocks lipid synthesis and increases β-oxidation through activation of AMPK [109,110]. Inhibition of SCD1 with a chemical inhibitor blocks cell-cycle progression and induces cell death in lung-cancer cells .
The importance of membrane synthesis in cancer cells has been highlighted by the observation that the expression and activity of choline kinase, an enzyme required for the synthesis of phosphatidylcholine and phosphatidylethanolamine (the major phospholipids found in cellular membranes) is increased in tumours from several tissues and correlates with poor prognosis [112–114]. Choline kinase has oncogenic activity when overexpressed, suggesting that the synthesis of phospholipids is rate limiting for transformation [115,116].
While there is compelling evidence for the requirement of de novo lipid synthesis for cancer-cell proliferation, it remains unexplained why this enhanced lipid demand cannot be met by the uptake of lipids from the bloodstream. Therefore, it is reasonable to hypothesize that the process of lipid synthesis itself may contribute to the tumourigenic phenotype. Cancer cells use large amounts of glucose for energetic and biosynthetic purposes , resulting in a high rate of lactate production and secretion. This requires the activation of mechanisms that equilibrate the intracellular pH and can lead to the acidification of the tumour microenvironment . It is possible that one of the roles played by lipid synthesis, at least in some cancer cells and conditions, is as a carbon sink to sequester excess pyruvate and avoid lactate production while still maintaining a high glycolytic rate. Furthermore, it may also contribute to redox balance. Hypoxia-tolerant organisms use NADP+, produced during lipid synthesis, as an electron acceptor when oxygen is not available  and it has been proposed that hypoxic cancer cells may follow a similar strategy . Additionally, in hypoxic cells, lipid synthesis-derived NADP+ could also help to increase the availability of cytoplasmic NAD+ required to maintain glycolysis. Ward & Thompson  have recently proposed that a putative mitochondria-cytosolic NADPH shuttle may exist, in which cytosolic NADP+ is used by isocitrate dehydrogenase-1 to produce α-ketoglutarate. This metabolite can then be transported to the mitochondria where the recently described inverse reaction catalysed by isocitrate dehydrogenase-2 [77–79] converts it back to isocitrate with concomitant production of NADP+. When oxygen is not available to maintain flux through the electron transport chain, the ratio of mitochondrial NADH/NAD+ increases and therefore the excess NADH could then be used by the mitochondrial nicotinamide nucleotide transhydrogenase  to transfer a proton to NAPD+ and generate NAD+ that can be made available to glycolysis through the malate-aspartate or the glycerol phosphate shuttles. Lipid synthesis would thus contribute both to redox balance between the cytoplasm and the mitochondria and to maximize glycolysis.
Resistance to oxidative stress
Recent evidence suggests that de novo lipid biosynthesis in cancer cells can increase their resistance to oxidative stress. Mammalian cells are inefficient at synthesizing polyunsaturated fatty acids because they lack an Δ3 desaturase. Consequently, a high rate of de novo lipid synthesis in tumours increases the relative amount of saturated and monounsaturated fatty acids, compared with those obtained through diet . Polyunsaturated acyl-chains are more susceptible to peroxidation. It has been shown that inhibition of lipid synthesis renders cancer cells susceptible to cell death induced by oxidative stress or chemotherapeutic agents . This intriguing observation requires further investigation but suggests that inhibition of lipid synthesis could be used to increase the effect of chemotherapy.
Resistance to energy stress
While most tumours exhibit a high rate of glucose uptake, which contributes to support both their energetic and biosynthetic requirements , some tumour types exhibit increased dependence on lipid oxidation as their main energy source. One such example is prostate tumours, which generally display a low rate of glucose utilization [125,126], show increased uptake of fatty acids, such as palmitate , and overexpression of some β-oxidation enzymes . This may be caused by the specialized metabolism of prostate epithelial cells, which secrete large amounts of citrate into the prostatic fluid. During transformation, prostate cancer cells reactivate the TCA cycle and increase the oxidation of citrate . Moreover, it has also been shown that β-oxidation is required for the proliferation and survival of leukaemia cells .
Activation of β-oxidation may be crucial to support cancer-cell viability during periods of energy stress. Constitutive activation of the PI3K/Akt pathway sensitizes haematopoietic cells to withdrawal of glucose or growth factors . However, activation of β-oxidation is sufficient to maintain cell viability under these conditions . β-oxidation has also been shown to contribute to ATP production and to resistance to oxidative stress in glioblastoma cells, by providing substrates for NAPDH and glutathione production, allowing cells to remove reactive oxygen species .
Signalling functions of lipids
Increased fatty acid and cholesterol biosynthesis, as well as the mobilization of free fatty-acids from triacylglycerides, may lead to an increase in the levels of lipids with a signalling function that can contribute to different aspects of tumourigenesis.
Cholesterol is an important component of cholesterol-rich microdomains, called lipid rafts, which coordinate the activation of receptor-mediated signal-transduction pathways . In addition, an intermediate of cholesterol synthesis, farnesyl-pyrophosphate, is required for protein prenylation. Several proteins with important signalling functions are modified by the addition of an isoprenoid chain. Farnesylation is important for the activity of Ras and Rheb proteins, while geranyl-geranylation is required for Rho, Rac and cdc42 activity . It has been shown that inactivation of the retinoblastoma tumour suppressor causes senescence by increasing the prenylation of N-Ras through the E2 transcription factor-dependent activation of SREBP .
Lipids also form the structural basis of paracrine hormones and growth factors, including prostaglandins, leukotrienes, lyophosphatidic acid (LPA) or steroid hormones. Prostaglandins and leukotrienes are derived from the 20-carbon-unit arachidonic acid, produced from phosphoglycerides by the action of phospholipases A2 and C. The synthesis of prostaglandin involves the enzyme cyclooxygenase 2, which has been implicated in inflammation and tumour/stroma interactions that promote tumour growth, neovascularization and metastatic spread . LPA is a water-soluble phospholipid composed of glycerol, a single fatty-acid chain and a phosphate group. LPA stimulates cell proliferation, survival and migration through the regulation of G-protein-coupled receptors. Aberrant production of LPA can contribute to cancer initiation and progression .
A recent study has shown that the modulation of the levels of free fatty-acids can affect lipid hormone synthesis. MAGL is overexpressed in aggressive cancer-cell lines and in advanced ovarian tumours, resulting in elevated levels of free fatty-acids, LPA and prostaglandin. Inhibition of MAGL reduced cell migration, invasion and survival. However, the negative effect of MAGL inhibition on tumour growth in vivo was abolished when mice were kept on a high-fat diet, suggesting that dietary lipids can affect tumour-promoting signalling processes .
The last few decades of work have started to reveal the importance of lipids for cancer biology. Novel diagnostic techniques, such as acetate-based positron emission tomography, are already providing new insights into lipid metabolism in tumours, and inhibitors of FASN are considered as promising anticancer agents that have been shown to be effective in vitro and in xenograft models . However, the complex interplay between oncogenic signalling and lipid metabolism, and the large spectrum of lipid functions at both cellular and organismal levels, highlight the importance of a more detailed understanding of the alterations to lipid metabolism in cancer. Targeting lipid metabolism may also offer novel therapeutic strategies for cancer treatment.