The PI3K-AKT network at the interface of oncogenic signalling and cancer metabolism

Abstract

The altered metabolic program of cancer cells facilitates their cell autonomous proliferation and survival. In normal cells, signal transduction pathways control core cellular functions, including metabolism, to couple signals from exogenous growth factors, cytokines, or hormones to adaptive changes in cell physiology. The ubiquitous, growth factor-regulated PI3K-AKT signaling network has diverse downstream effects on cellular metabolism through either direct regulation of nutrient transporters and metabolic enzymes or the control of transcription factors that regulate the expression of key components of metabolic pathways. Aberrant activation of this signaling network is one of the most frequent events in human cancer and serves to disconnect the control of cell growth, survival, and metabolism from exogenous growth stimuli. Here, we discuss our current understanding of the molecular events controlling cellular metabolism downstream of PI3K and AKT, and how it couples two major hallmarks of cancer, growth factor independence through oncogenic signaling and metabolic reprogramming to support cell survival and proliferation.

[H1] Introduction

The phosphoinositide 3-kinase (PI3K)-AKT pathway is the most commonly activated pathway in human cancers¹. Under physiological conditions, this pathway is activated in response to insulin, growth factors, and cytokines and regulates key metabolic processes, including glucose metabolism, biosynthesis of macromolecules, and maintenance of redox balance to support both systemic metabolic homeostasis and the growth and metabolism of individual cells. Oncogenic activation of the PI3K-AKT pathway in cancer cells reprograms cellular metabolism by augmenting the activity of nutrient transporters and metabolic enzymes, thereby supporting the anabolic demands of aberrantly growing cells. Understanding how the PI3K-AKT pathway governs metabolic networks in normal cells and how this control is altered in cancer cells could reveal metabolic vulnerabilities to inform new therapeutic strategies, which is underscored by the “druggable” nature of metabolic enzymes. Here, we review the physiological functions of the PI3K-AKT network in controlling metabolic networks and the potential consequences from oncogenic PI3K-AKT signaling leading to dysregulation of these key metabolic control points in cancer cells and tumors.

Signal transduction networks are the cellular lines of communication, allowing cells to perceive and relay signals, including those from the extracellular environment, to downstream targets that suitably adapt cellular functions to maintain cell, tissue, and organismal homeostasis. The PI3K-AKT signaling network is activated downstream of receptor tyrosine kinases (RTKs), cytokine receptors, integrins, and G protein-coupled receptors (GPCRs) and plays a central role in promoting cell survival and growth². Class Ia PI3K exists as heterodimers of a catalytic subunit (p110α, β, or δ) associated with a regulatory subunit (p85α or p85β, or shorter variants thereof), while class Ib is comprised of the catalytic subunit p110γ associated with the regulatory subunit p101³. Class Ia PI3K is activated upon engagement of SH2 domains within the regulatory subunits with phospho-tyrosine residues on activated receptors (e.g., RTKs or cytokine receptors) or adaptor proteins, whereas class 1b is activated by GPCRs. Activation of PI3K at the plasma membrane stimulates phosphorylation of its phospholipid substrate phosphatidylinositol 4,5-bisphosphate (PIP₂) to produce the second messenger phosphatidylinositol 3,4,5-trisphosphate (PIP₃) (Fig. 1a). PI3K signalling is attenuated by phosphatase and tensin homolog (PTEN), which dephosphorylates PIP₃ to regenerate PIP₂. PIP₃ accumulation at the plasma membrane, and perhaps other intracellular membranes, creates docking sites to recruit downstream effector proteins that contain a subclass of pleckstrin homology (PH) domain that specifically engage this lipid species². One such protein is the serine-threonine kinase AKT (also known as protein kinase B or PKB), which upon PIP₃ binding is subsequently phosphorylated by phosphoinositide-dependent protein kinase 1 (PDK1) at T308, an event essential for kinase activation, and by mechanistic target of rapamycin (mTOR) complex 2 (mTORC2) at S473, which further increases the activity of AKT^4–6 (Fig. 1a). There are three isoforms of AKT (AKT1, AKT2, and AKT3), which are all activated in this manner. AKT1 and AKT2 are broadly expressed, with AKT2 being particularly important in insulin-responsive metabolic tissues, while AKT3 is more restricted in its tissue distribution, being highest in the brain^7,8. Active AKT phosphorylates a large and diverse array of downstream substrates, the majority of which appear to be redundantly regulated by the three AKT isoforms (hereon referred to as AKT unless an isoform-specific function is noted). AKT-mediated phosphorylation of these protein targets serves to influence a variety of cell biological functions, including cell growth, proliferation, survival and, as detailed here, metabolism⁸.

Fig. 1: The PI3K-AKT pathway and its major downstream effectors.

A. Mechanisms of AKT activation. Receptor tyrosine kinase (RTK) activation and tyrosine phosphorylation of its cytosolic domain or of scaffolding adaptors create binding sites that recruit the lipid kinase PI3K to the plasma membrane. PI3K phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP₂) to produce phosphatidylinositol 3,4,5-trisphosphate (PIP₃), which can be dephosphorylated back to PIP₂ by the lipid phosphatase PTEN. PIP₃ acts as a second messenger to recruit the serine/threonine protein kinase AKT to the plasma membrane, where it is fully activated through phosphorylation at T308 and S473 by the PDK1 and mTORC2 protein kinases, respectively. AKT signaling serves to promote cell survival, growth, and proliferation, in part, by inducing various changes to cellular metabolism. Coloured hexagons denote common points of activation and inactivation by cancer-associated mutations. “P” indicates protein phosphorylation events.

B. AKT controls cellular metabolism, in part, through three key downstream substrates: TSC2, GSK3, and the FOXO transcription factors. AKT phosphorylates and inhibits TSC2, a component of the TSC complex, to activate mTORC1 by relieving TSC complex-mediated inhibition of Rheb. S6K1 and 4EBP1 are canonical downstream targets of mTORC1, which together with other targets, serve to stimulate the processing and activation of the SREBP family of transcription factors and the mRNA translation of the MYC, HIF1α, and ATF4 transcription factors. GSK3-mediated phosphorylation of the transcription factors SREBP, MYC, NRF2, and HIF1α targets them for ubiquitination and proteasomal degradation.

PI3K, phosphoinositide 3-kinase; PTEN, phosphatase and tensin homolog; PDK1, phosphoinositide-dependent protein kinase 1; mTORC2, mechanistic target of rapamycin (mTOR) complex 2; mTORC1, mTOR complex 1; TSC2; tuberous sclerosis complex 2; GSK3, Glycogen synthase kinase 3; FOXO, forkhead box O; Rheb, Ras homolog enriched in brain; S6K1, Ribosomal protein S6 kinase 1; 4EBP1, eukaryotic translation initiation factor 4E binding protein, SREBP: Sterol regulatory element binding proteins; HIF1, Hypoxia-inducible factor 1, ATF4, activating transcription factor 4, NRF2, nuclear factor erythroid 2-related factor 2.

Genetic events leading to growth factor-independent activation of the PI3K-AKT pathway are amongst the most frequently occurring drivers of human cancer. Common alterations in cancer include 1) activating mutations in PIK3CA, the p110α catalytic subunit of PI3K, which is the most frequently mutated single oncogene found in analyses across cancer lineages¹; 2) loss of function mutations and deletions in PTEN, the second most mutated tumor suppressor gene (following p53); 3) amplification and activation of specific PI3K-activating RTKs, including EGFR and HER2; 4) amplification and gain-of-function missense mutations in one of the three isoforms of AKT^1,9–11 (Fig. 1a).

[H1] Key effectors controlling cell metabolism

While there are many downstream effectors of both PI3K and Akt that alter the function of both normal and cancer cells, we focus here on those that influence cellular metabolism. In a cell intrinsic manner, the metabolic functions of Akt serve to support its canonical functions in promoting cell survival, growth, and proliferation (Fig. 1a). AKT signaling alters metabolism either directly, through phosphorylation-mediated regulation of metabolic enzymes, or indirectly, through control of various transcription factors. Phosphorylation of metabolic enzymes allows for acute changes in the activity of metabolic pathways and the directionality of metabolic flux. Whereas, longer-term changes in cellular metabolism are often achieved through control of gene expression programs. While AKT directly phosphorylates several metabolic enzymes or regulators of nutrient transport, it also activates a few key downstream effectors that play a major role in cellular metabolic reprogramming, including mTOR complex 1 (mTORC1), glycogen synthase kinase 3 (GSK3), and members of the forkhead box O (FOXO) family of transcription factors⁸ (Fig. 1b).

mTOR is a serine/threonine kinase that functions as the catalytic subunit of two multi-protein complexes, mTORC1 and mTORC2, with distinct subunit composition, substrate specificity, and functions¹². Both mTORC1 and mTORC2 can be activated by PI3K signaling, with mTORC2 being an upstream regulator of AKT and mTORC1 being a downstream effector. The primary mechanism by which AKT activates mTORC1 is through phosphorylation of the tuberous sclerosis complex 2 (TSC2) protein, which as a component of the TSC protein complex acts as a GTPase-activating protein (GAP) to inhibit the Ras-related small G protein RHEB¹³. The AKT-mediated phosphorylation of TSC2 disrupts colocalization of the TSC complex with RHEB, thereby allowing accumulation of RHEB-GTP, which binds to and activates mTORC1¹⁴ (Fig. 1b). The AKT-mediated stimulation of mTORC1 acts in parallel to nutrient- and energy-sensing mechanisms that also control the activation state of mTORC1 and, as discussed below, serves as a key point of regulation for anabolic metabolism and cell growth¹³.

Glycogen synthase kinase 3 (GSK3) was the first identified AKT substrate and is a key regulator of cellular metabolism, established originally for its role in blocking glycogen synthesis via phosphorylation and inhibition of its namesake substrate glycogen synthase^15,16. GSK3 is active under basal conditions and is inhibited in response to growth factors and insulin via AKT-mediated phosphorylation^15,17 (Fig. 1b). GSK3 has many downstream substrates, the phosphorylation of which often exerts inhibitory control over these targets¹⁸. Among these are a several transcription factors that regulate the metabolism of both normal and cancer cells downstream of PI3K-AKT signaling. GSK3-mediated phosphorylation of these factors marks them for ubiquitination and proteasomal degradation (Fig. 1b).

Another canonical substrate of AKT is the FOXO family of transcription factors (FOXO1, FOXO3A, FOXO4), which upon phosphorylation are sequestered from the nucleus, thus preventing expression of their target genes^19,20 (Fig. 1b). Given the established FOXO gene expression program, which includes numerous suppressors of growth, proliferation, and survival, together with specific metabolic enzymes, the AKT-mediated inhibition of FOXO has been implicated in various aspects of cancer development and progression²¹.

[H1] Control of glucose metabolism

Altered glucose metabolism is perhaps the most common metabolic change distinguishing cancer cells from their cell of origin. This metabolic feature, characterized by an increased rate of glucose uptake and its glycolytic conversion to lactate even under oxygen-rich conditions, was originally described nearly one-hundred years ago by Otto Warburg²² and is referred to as aerobic glycolysis or the Warburg effect²³. Glycolysis, in addition to producing ATP, provides metabolic intermediates as substrates for metabolic pathways that branch off of glycolysis and support biosynthetic processes for production of protein, lipids, and nucleotides, required for cell growth and proliferation. Under aerobic glycolysis, the predominant diversion of pyruvate to lactate, rather than its entry into the mitochondria for oxidation, also serves a key role in redox homeostasis [G] by regenerating NAD+. How the genetic events leading to cellular transformation and cancer promote the Warburg effect is still being elucidated. However, the PI3K-AKT pathway can control several aspects of this metabolic program (Fig. 2).

Fig. 2: Direct post-translational regulation of metabolic enzymes and processes downstream of the PI3K-AKT pathway.

AKT stimulates metabolic changes that contribute to anabolic metabolism by directly phosphorylating key metabolic enzymes. AKT promotes plasma membrane localization of the glucose transporter GLUT1 and increased glucose uptake by directly phosphorylating and inhibiting TXNIP, a protein that promotes endocytosis of GLUT1. AKT signaling also promotes retention and metabolic activation of the newly acquired glucose by activating HK2, which phosphorylates glucose to generate glucose 6-phosphate, which cannot be transported out of the cell by GLUT1 and is the entry metabolite for the hexosamine pathway, the oxidative pentose phosphate pathway (PPP), and glycolysis. AKT enhances flux into glycolysis through phosphorylation and activation of PFKFB2, which produces fructose-2,6-bisphosphate, an allosteric activator of the rate-limiting glycolytic enzyme PFK1, which commits the glucose-derived carbon to glycolysis. Both the oxidative and non-oxidative PPP, which branch off of glycolytic intermediates, generate ribose-5-phosphate that serves as the sugar moiety for purine and pyrimidine nucleotides. AKT phosphorylates and activates the non-oxidative PPP enzyme TKT, thereby contributing to ribose-5-phosphate production for nucleotides. AKT also phosphorylates and increases the activity of NADK, which catalyses the phosphorylation of NAD⁺ to generate NADP⁺, a limiting substrate for the oxidative PPP, which generates two molecules of the reducing cofactor NADPH through its oxidation of glucose 6-phosphate. Downstream of AKT, mTORC1 activation acutely stimulates de novo pyrimidine synthesis through an S6K-dependent phosphorylation of the pyrimidine synthesis enzyme CAD. Pyruvate, the end product of glycolysis, is either converted to lactate by LDH, which regenerates NAD⁺ needed for sustained glycolysis, or can enter the mitochondria and TCA cycle for oxidation initiated by the pyruvate dehydrogenase (PDH) complex, the activity of which can be inhibited by pyruvate dehydrogenase kinase (PDK). AKT phosphorylates PDK and promotes its inhibition of PDH, thus favouring the LDH reaction [Note: this phosphorylation is believed to occur within the mitochondria]. AKT directly regulates lipid synthesis through phosphorylation of ACLY, which generates acetyl-CoA in cytosol from the TCA cycle-derived citrate.

TXNIP, thioredoxin-interacting protein; GLUT1, glucose transporter 1; PPP, Pentose Phosphate Pathway; PFKFB2, 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase; PFK1, phosphofructokinae 1; TKT, transketolase; NADK, NAD kinase; mTORC1, mTOR complex 1; CAD, carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase, S6K1, ribosomal protein S6 kinase 1; NAD⁺, nicotinamide adenine dinucleotide, NADP⁺, NAD⁺ phosphate, LDH, lactate dehydrogenase.

AKT activation has been shown to be sufficient to promote aerobic glycolysis^24,25. Expression of constitutively-active AKT results in a growth factor-independent increase in glucose uptake and glycolytic rate^24–29. This AKT-mediated induction of aerobic glycolysis can render cancer cells dependent on glucose for survival^25,29. Multiple control points in glycolysis have been found to be regulated by PI3K-AKT signaling, and these include both acute, post-translational modifications and more prolonged transcriptional effects on glucose transporters and glycolytic enzymes (Fig. 2, 3).

Fig. 3: Transcriptional control of metabolic processes downstream of AKT signaling.

AKT regulates metabolism through a number of downstream transcription factors that control the expression of genes encoding metabolic enzymes. FOXO, HIF1 and MYC regulate the expression of glucose transporters and glycolytic enzymes, and intermediates of glycolysis contribute to nucleotide and lipid synthesis. MYC and ATF4 induce the expression of enzymes contributing to nucleotide synthesis. SREBP isoforms globally induce the expression of lipogenic enzymes, as well as enzymes in the oxidative PPP, which can provide NADPH as reducing equivalents for lipid synthesis and redox. NRF2 regulates redox homeostasis. FOXO, forkhead box O; HIF1, Hypoxia-inducible factor 1; ATF4, activating transcription factor 4, SREBP: Sterol regulatory element binding proteins; NRF2, nuclear factor erythroid 2-related factor 2 and NADPH, nicotinamide adenine dinucleotide phosphate (reduced).

[H2] Direct regulation of glucose uptake and glycolysis

Glucose is a primary carbon and energy source in all organisms. The controlled uptake of glucose into cells is mediated by the glucose transporter family (GLUTs)³⁰. Among these, GLUT1 is ubiquitously expressed and most frequently elevated in cancer³¹, while GLUT4 is mainly expressed in insulin-responsive muscle and adipose tissue to function in plasma glucose clearance after food intake³². AKT promotes glucose uptake through both GLUT1 and GLUT4. Studies of insulin-stimulated glucose uptake found that AKT2 associates with vesicles containing GLUT4, and AKT2 induces trafficking of GLUT4 from these vesicles to the plasma membrane^32,33. A major mechanism underlying this regulation is the AKT -mediated phosphorylation and inhibition of TBC1D4 (also known as AS160), a GAP for Rab GTPases that promote GLUT4 trafficking^34,35. However, this mechanism appears to be specific for GLUT4 and, thus, is unlikely to play a major role in glucose uptake into cancer cells, which predominantly use GLUT1. Studies in both normal and transformed hematopoietic cells have demonstrated that the cytokine-stimulated translocation of GLUT1 to the plasma membrane is mediated through AKT activation^24,36,37. Recent studies have implicated thioredoxin-interacting protein (TXNIP) as a direct AKT substrate in the regulation of both GLUT1 and GLUT4 trafficking³⁸ (Fig. 2). TXNIP promotes endocytosis of GLUT1 and inhibits glucose uptake^39,40. AKT phosphorylates and inhibits TXNIP, resulting in a rapid increase in GLUT1 and GLUT4 at the plasma membrane and enhanced glucose uptake in various cell types (mouse embryonic fibroblasts and 3T3-L1 adipocytes) and mouse tissues (skeletal muscle, white adipose tissue, and liver)³⁸. Regulation of TXNIP by oncogenic PI3K-AKT signaling has also been suggested to enhance aerobic glycolysis in non-small cell lung cancer cell lines⁴¹, but the significance of the AKT-TXNIP-GLUT1 axis in the widely observed increase in tumor glucose uptake in vivo remains to be determined⁴².

In addition to glucose uptake, AKT controls key steps in glycolysis through phosphorylation and activation of specific glycolytic enzymes (Fig. 2). Following its transport into cells, glucose becomes activated to enter metabolic pathways through the action of hexokinases that phosphorylate glucose to form glucose-6-phosphate, which cannot be transported out of the cell. AKT activation has been found to promote hexokinase 2 (HK2) activity, at least in part, by increasing its association with voltage-dependent anion channel (VDAC) at the outer mitochondrial membrane ^27,43. This mechanism has been proposed to provide a rich source of mitochondria-derived ATP for rapid and sustained HK2-mediated glucose phosphorylation and has also been found to promote cell survival by preserving mitochondrial integrity^44–46. HK2 expression is elevated in many human cancers, including ovarian, colorectal, pancreatic, glioblastoma, and liver cancer, and its activity has been found to be critical for tumorigenesis and metastasis in various mouse models^47,48–50. HK2 has also been exploited as a metabolic liability in mouse tumor models displaying hyperactive AKT signaling, such PTEN-deficient prostate cancer, where HK2 deletion attenuates tumor growth^50–53. HK2, and other hexokinases, can be competitively inhibited by 2-deoxy-D-glucose, and HK2 has attracted a lot of interest as a potential target for cancer treatment⁵⁴, although more specific and potent inhibitors will likely be needed. The HK2 product, glucose-6-phosphate can then enter one of three metabolic pathways: the hexosamine pathway, the pentose phosphate pathway, or glycolysis.

AKT signaling indirectly stimulates the activity of phosphofructokinae 1 (PFK1), the first committed step of glycolysis. AKT phosphorylates and activates 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase (PFKFB2)⁵⁵, a key enzyme that catalyses the interconversion of fructose-6-phosphate and fructose-2,6-bisphosphate, with the latter metabolite serving as a potent allosteric activator of PFK1⁵⁶. The AKT-mediated phosphorylation of PFKFB2 increases its synthesis of fructose 2,6-bisphosphate leading to enhanced PFK1-driven glycolytic flux (Fig. 2). Akt has also been found to activate the related isoform PFKFB3 through phosphorylation of a conserved residue that it also phosphorylates in PFKFB2⁵⁷, but it is worth noting that the AKT site on PFKFB3 lacks a critical sequence feature found in other established Akt substrates⁵⁸. Fructose-2,6-bisphosphate levels have been reported to be markedly increased in mouse models of lung and mammary carcinomas compared to non-malignant cells, perhaps promoting high glycolytic flux ^59,60. However, the role of oncogenic AKT-mediated phosphorylation of PFKFB2 or PFKFB3 in the induction of aerobic glycolysis in various cancers remains to be elucidated.

PI3K signaling has also been found to potentiate glycolytic flux in an AKT-independent manner. Growth-factor stimulation of PI3K induces a Rac-dependent release of the glycolytic enzyme aldolase A from an actin-bound, low-activity state to increase glycolytic flux⁶¹. High aldolase A expression in cancer is correlated with poor patient prognosis^62,63. While the role of PI3K-stimulated aldolase A release from the actin cytoskeleton in tumorigenesis has yet to be established, studies have reported that targeting aldolase A in tumor xenograft models can attenuate tumor growth^64,65.

[H2] Transcriptional regulation of glucose uptake and glycolysis

In addition to acute regulation of glycolytic enzymes, the PI3K-AKT pathway also promotes sustained aerobic glycolysis through an increase in protein levels of glucose transporters and glycolytic enzymes mediated by control of downstream transcription factors (Fig. 3). Hypoxia-inducible factor 1 (HIF1α), which is degraded in an oxygen-dependent manner and stabilized under conditions of hypoxia, induces the expression of GLUT1 and nearly all the enzymes of glycolysis⁶⁶. As part of the adaption to hypoxia, HIF1α also activates expression of lactate dehydrogenase 1 (LDH1) and pyruvate dehydrogenase kinase 1 (PDK1), which together function to channel pyruvate to lactate and away from its oxidation into acetyl-CoA for entry into the mitochondrial tricarboxylic acid (TCA) cycle.

AKT has also been proposed to phosphorylate and increase the activity of PDK1 during hypoxia, thus further facilitating a switch towards glycolysis to support cancer cell proliferation under these conditions⁶⁷ (Fig. 2). However, it is unknown how activated AKT in the cytosol gains access to PDK1 in the mitochondria and whether AKT exerts this direct regulation of PDK1 in response to oncogenic PI3K signaling. Importantly, several studies have found that the activation of mTORC1 downstream of AKT signaling in cell and tumor models, including prostate cancer, leads to an increase in HIF1α protein levels, expression of gene targets, and increased glucose uptake and glycolytic conversion to lactate, even under normal oxygen concentrations (i.e., normoxia)^68–71. HIF1α remains unstable under such conditions, but its protein levels increase due to enhanced, mTORC1-regulated translation of the HIF1A mRNA ^68,72,73. It is worth noting that hypoxia is a much stronger inducer of HIF1α, and the PI3K-mTOR pathway is not involved in this oxygen-mediated regulation of HIF1α protein stability. The importance of HIF1α as a downstream effector of AKT signaling in cancer development and progression is unclear, and one study found this transcription factor to be dispensable for AKT-driven tumor growth in a hepatoma xenograft model⁷⁴. However, the normoxic upregulation of the HIF1α program downstream of PI3K-mTOR signaling is likely to contribute to the induction of aerobic glycolysis in cancer cells and may provide metabolic flexibility [G] in growing tumors that facilitates regional adaptation to nutrient and oxygen fluctuations.

The transcription factor MYC induces expression of numerous genes that promote cell growth and proliferation, and is one of the most frequently altered oncogenes in human cancers⁷⁵. MYC induces expression of the major glucose transporters and most glycolytic enzymes and can drive aerobic glycolysis⁷⁵. In contrast to HIF1α, MYC activation does not appear to lead to preferential directing of glycolysis-derived pyruvate toward lactate and away from mitochondrial respiration. Instead, many MYC gene targets function to enhance mitochondrial metabolism⁷⁵. MYC activity is generally dictated by its abundance, and it is increased downstream of PI3K-AKT signalling through a combination of transcriptional, translational, and posttranslational mechanisms⁷⁵. For example, mTORC1 signaling enhances the translation of MYC^76,77, while AKT promotes MYC stabilization by inhibiting GSK3, which phosphorylates MYC and targets it for proteasomal degradation^78–80. The FOXO transcription factors have been found to suppress the expression of glycolytic genes in some settings through an antagonistic effect on MYC function^81,82–84. Activation of AKT relieves the inhibitory effects of FOXO transcription factors on glycolytic enzymes and MYC, thereby further enhancing glycolysis⁸³. However, it is important to note that MYC activation downstream of the PI3K-AKT pathway is likely to be context dependent in cancer. For instance, PI3K inhibitors have failed to reduce MYC levels in various cancer models, including colorectal cancer, acute myeloid leukemia, and multiple myeloma, likely due to dominant regulatory inputs from the RAS-ERK pathway to MYC that are active in many cancer settings^85–87. Thus, the vast metabolic program downstream of MYC should not be universally equated to PI3K-Akt signalling.

This collective work demonstrates the central role that the PI3K-AKT pathway plays in promoting glucose uptake and glycolysis under both physiological conditions and in cancer. In the setting of oncogenic PI3K-AKT signaling, these combined regulatory mechanisms are likely to drive the constitutive induction of aerobic glycolysis. This common feature of cancer cells facilitates metabolic flux into pathways that branch off of glycolysis and contribute to the synthesis of cellular macromolecules²³, and these biosynthetic processes are also further controlled downstream of Akt. While mitochondrial metabolism through the TCA cycle has also emerged as a process supporting the energetic and biosynthetic demands of proliferating cancer cells^88,89, defined functions of the PI3K-AKT pathway in direct control of the TCA cycle have not been established. It is possible that under conditions of aerobic glycolysis, the PI3K-AKT pathway might promote anaplerotic metabolism [G] to sustain TCA cycle flux, for instance, by promoting glutaminolysis [G] via MYC activation^77,90.

[H1] Control of anabolic metabolism

Proliferating cells need to double their protein, lipid, and nucleotide content with each cell division. In order to meet this biosynthetic demand, cells stimulate anabolic processes [G] to drive the production of these macromolecules. Here, we discuss the role of AKT, mTORC1, and MYC in inducing a PI3K-driven anabolic program in cancer.

[H2] De novo lipid synthesis

Aberrant activation of lipid biosynthesis is a common feature of cancer cells⁹¹. While the majority of cells in our body rely on the uptake of fatty acids and lipoproteins from the bloodstream to fulfill their lipid requirements, cancer cells activate de novo lipid biosynthesis to facilitate the generation of cellular membranes and support their increased growth and proliferation⁹². Both sterols and fatty acids are synthesized from cytosolic acetyl-CoA, which is produced from the tricarboxylic acid (TCA) cycle intermediate citrate via ATP-citrate lyase (ACLY) or from acetate via acetyl-CoA synthetase. The PI3K-AKT pathway induces de novo lipid synthesis through both post-translational and transcriptional mechanisms.

AKT can initiate de novo lipid synthesis by directly phosphorylating ACLY⁹³, thereby increasing its activity⁹⁴ and boosting the production of cytosolic acetyl-CoA to be used for sterol and fatty acid synthesis, as well as protein acetylation reactions (Fig. 2). The AKT-ACLY axis has been reported to promote tumor growth and to globally influence histone acetylation^95,96. In addition to its oncogenic regulation through the PI3K-AKT pathway, ACLY is frequently overexpressed in various human cancers, and inhibition of ACLY diminishes cancer cell proliferation both in vitro and in vivo, making this enzyme a potentially attractive target for cancer therapy^97–99. Owing to its key role in lipid biogenesis, inhibitors of ACLY, originally developed for metabolic disorders such as hypercholesterolemia and type-2 diabetes^100,101, are being considered as potential anti-cancer drugs^102,103.

AKT signaling also promotes de novo lipid synthesis through activation of the SREBP family of transcription factors (SREBP1a, SREBP1c, and SREBP2), which induce the expression of nearly all enzymes of fatty acid and sterol synthesis, including ACLY^104,105. The SREBPs exist as inactive endoplasmic reticulum (ER) transmembrane proteins that must traffic to the Golgi for proteolytic processing to be activated. SREBP processing releases the N-terminal portion of the protein that serves as the mature active form, which then translocates to the nucleus to initiate transcription at sterol response elements contained in the promoters of lipogenic genes and those involved in NADPH production required to support lipid synthesis. AKT activates SREBP through at least two downstream branches. mTORC1 has been shown to stimulate the processing and nuclear translocation of SREBPs through several different proposed mechanisms, leading to the induction of lipogenic gene expression and an increase in de novo lipid synthesis in the liver in response to insulin or in cancer cells downstream of oncogenic PI3K or Ras^68,105–108. Once processed, the mature active form of SREBP is targeted for ubiquitin-dependent degradation by GSK3-mediated phosphorylation^109,110. Thus, AKT signaling can stimulate the processing of SREBP through mTORC1 activation and promote stability of the processed active SREBP by inhibiting GSK3. Recently, MYC was also reported to cooperate with SREBP to induce lipogenesis and promote cancer growth¹¹¹. Interestingly, many of the mRNAs induced by SREBP are regulated by the serine/arginine-rich (SR) protein family of splicing factors. The splicing of these mRNAs encoding lipogenic enzymes has been found to be stimulated by mTORC1 signaling through its downstream target S6K, which phosphorylates and activates SR protein kinase 2 (SRPK2) leading to subsequent phosphorylation and activation of the SR proteins and enhanced splicing¹¹². Many factors and enzymes involved in de novo lipid synthesis, including the SREBPs, SRPK2, and the lipogenic enzymes induced by the SREBPs have been found to be elevated in diverse cancer lineages, and these remain potential targets of interest for cancer therapy^91,108,113.

[H2] Nucleotide synthesis

Nucleotides, comprised of purines and pyrimidines, are essential building blocks for the synthesis of nucleic acids (RNA and DNA), among other cellular functions¹¹⁴. In contrast to normal, quiescent cells, cancer cells stimulate robust de novo synthesis of nucleotides to accommodate the nucleic acid synthesis required for cell growth and proliferation^115,116. The de novo nucleotide synthesis pathways require coordinated input from multiple metabolic pathways, including the pentose phosphate pathway (PPP) and the serine and glycine synthesis pathway (both of which branch off of glycolysis), aspartate synthesis from the TCA cycle intermediate oxaloacetate, one carbon metabolism, and glutamine uptake, to supply the ribose sugar and necessary atoms to form the corresponding pyrimidine and purine bases (Fig. 4a). Thus, it is not surprising that AKT signaling appears to regulate nucleotide synthesis through multiple parallel mechanisms affecting these metabolic inputs.

Fig. 4: Regulation of nucleotide metabolism downstream of the AKT-mTORC1 pathway.

A. Left to right: Schematic of purine and pyrimidine nucleotides indicating the donors of carbon and nitrogen atoms that form nucleotides. The small molecules are color coded according to the contributing metabolic pathways from which they are commonly derived, shown schematically to the right.

B. Transcriptional and post-translational mechanisms contributing to de novo nucleotide synthesis downstream of the AKT-mTORC1 pathway. Left to right: AKT-mediated phosphorylation of the non-oxidative PPP enzyme TKT and SREBP-mediated regulation of the oxidative PPP enhance the production of ribose-5-phosphate, which can then be used for nucleotide synthesis by conversion to phospho-ribosyl pyrophosphate through the enzyme PRPS, the levels of which are elevated upon MYC activation. MYC promotes glutamine uptake and stimulates expression of several genes encoding enzymes of both purine and pyrimidine synthesis pathways. In addition to being downstream of mTORC1, SREBP and MYC can also be stabilized via AKT-mediated inhibition of GSK3. ATF4 activation downstream of mTORC1 contributes to enhanced purine synthesis through the induction of serine biosynthesis enzymes and the mitochondrial tetrahydrofolate cycle enzyme MTHFD2, thus supplying both glycine and one-carbon formyl units. mTORC1 acutely stimulates pyrimidine synthesis through the S6K1-mediated phosphorylation of the pyrimidine synthesis enzyme CAD. The newly synthesized purine and pyrimidines are used for the synthesis of RNA, predominantly rRNA for ribosome biogenesis, and DNA in proliferating cells.

NADPH, nicotinamide adenine dinucleotide phosphate (reduced), PPP, Pentose Phosphate Pathway; TKT, transketolase, SREBP: Sterol regulatory element binding proteins; PRPS, phosphoribosyl pyrophosphate synthase; MTHFD2, methylenetetrahydrofolate dehydrogenase 2, ATF4, activating transcription factor 4; mTORC1, mTOR complex 1.

The PI3K-AKT-mTORC1 network promotes glucose carbon flux into both the oxidative and non-oxidative branches of the PPP, thus producing ribose for nucleotide synthesis. As an alternative to entering glycolysis, glucose 6-phosphate, can enter the oxidative PPP through the action of glucose-6-phosphate dehydrogenase (G6PD) to be irreversibly oxidized to produce ribose-5-phosphate. Downstream of AKT signaling, mTORC1 activation enhances oxidative PPP flux, at least in part, via activation of SREBP and its transcriptional induction of G6PD expression⁶⁸. AKT has also been found to directly activate transketolase (TKT), a key enzyme in the non-oxidative PPP¹¹⁷ (Fig. 4b). Treatment of breast cancer cells with PI3K inhibitors was found to disproportionally reduce glucose flux through the non-oxidative PPP and result in nucleotide depletion and DNA damage¹¹⁸, suggesting a higher dependence on the non-oxidative PPP in this setting. Enzymes in both branches of the PPP, including G6PD and TKT, are overexpressed in cancer, and their inhibition attenuates cell proliferation in various settings, including colorectal, breast, lung, and liver cancer cells^119–121. More research is required to develop potent and selective inhibitors of key PPP enzymes and to identify the cancer contexts in which PPP inhibition is a vulnerability¹²².

AKT exerts transcriptional control of nucleotide synthesis, in part, through its regulation of MYC. MYC drives expression of an array of metabolic genes involved in supplying metabolite precursors for nucleotide synthesis, including glutamine, and directly induces the expression of many of the enzymes of the pyrimidine and purine synthesis pathways^123,124 (Fig. 4b). MYC controls both the transcription and translation of phosphoribosyl pyrophosphate (PRPP) synthase 2 (PRPS2), which catalyzes the reaction that commits ribose-5-phosphate to nucleotide synthesis by producing PRPP, and PRPS2 has been found to be a critical effector of MYC-mediated tumorigenesis^125,126. Glutamine serves as the major nitrogen source for the synthesis of both pyrimidine and purine bases, and MYC promotes glutamine uptake by inducing the expression of the glutamine transporters SLC1A5 and large amino acid transporter 1 (LAT1), which is a heterodimer composed of SLC7A5 and SLC3A2^90,127 (Fig. 4b).

As a downstream effector of PI3K-AKT signaling, mTORC1 has also emerged as a major driver of de novo nucleotide synthesis, which it regulates through both posttranslational and transcriptional mechanisms (Fig. 2, 4b). Growth factor signaling to mTORC1 acutely stimulates pyrimidine synthesis through the S6K1-mediated phosphorylation and activation of the first and rate-limiting enzyme in this pathway, carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase (CAD), which catalyses the first three steps of pyrimidine biosynthesis^128,129 (Fig. 4b). While this phosphorylation is not required for basal CAD activity, it is required for PI3K-AKT signaling to increase metabolic flux through this pathway and increase pyrimidine synthesis. With more delayed kinetics, mTORC1 signaling also induces de novo purine synthesis through transcriptional mechanisms involving MYC, SREBP and ATF4, which stimulate the expression of specific metabolic enzymes in the purine synthesis pathway or pathways that feed metabolites into purine synthesis¹³⁰. For instance, mTORC1 was found to activate ATF4, in a manner distinct from its canonical activation as part of the integrated stress response, resulting in transcriptional induction of the enzymes catalyzing serine synthesis and its conversion through the mitochondrial tetrahydrofolate (mTHF) cycle to formate, which provides one carbon units essential for building the purine ring¹³⁰ (Fig 4a,b).

It is interesting to note that in growing cells, including cancer cells, there is a robust increase in ribosome biogenesis. Over half of the mass of the ribosome is comprised of rRNA, which constitutes more than 80% of total cellular RNA. Thus, ribosome biogenesis underlying cell growth places a substantial increased demand on the cell for more nucleotides to support rRNA synthesis. It is well established that mTORC1 and MYC serve as key drivers of ribosome biogenesis in both normal and cancer cells^131,132, thereby providing the logic for their parallel induction of nucleotide synthesis. Indeed, the newly synthesized pyrimidine and purine nucleotides produced in response to mTORC1 activation can be traced into rRNA^128,130.

While inhibitors of enzymes involved in nucleotide synthesis were the first chemotherapeutics introduced in the 1940s¹³³ and have become part of established cancer therapy regiments, the factors that dictate vulnerability to such inhibitors remain poorly understood. As immune cells, like cancer cells, upregulate de novo nucleotide synthesis to support cell proliferation upon activation, several immunosuppressants that target nucleotide synthesis pathways have been developed for clinical use. The growth of cancer cells in vitro and tumor models in vivo with uncontrolled mTORC1 or MYC activation has been found to be very sensitive to inhibitors of the enzyme inosine monophosphate dehydrogenase (IMPDH), such as mizoribine, which is a widely used immunosuppressant in Asia^134,135. Interestingly, IMPDH is required for the synthesis of guanylates, which are disproportionately represented in pre-rRNA (37%). Therefore, the process of rRNA synthesis more rapidly depletes guanylates in cells with active ribosome biogenesis, thereby depriving cells of nucleotides for DNA synthesis, resulting in replication stress and apoptosis¹³⁴. Thus, it is possible that such agents, despite their immunosuppressive effects, might be effective and selective anti-tumor agents in settings with strong increases in ribosome biogenesis, such as those with active mTORC1 or MYC. Interestingly, widely used chemotherapeutic agents that target nucleotide synthesis, including methotrexate and 6-mercatopurine, have been found to inhibit mTORC1 signaling through a mechanism involving depletion of purine, but not pyrimidine, nucleotides^136,137. Whether some of the anti-cancer effects of these agents are due to inhibition of mTORC1 remains to be determined.

[H2] Protein synthesis

The growth program of PI3K-AKT signaling involves a robust increase in protein synthesis, with much of this driven by mTORC1 activation. mTORC1 activation leads to an increase in the protein synthesis capacity of cells through multiple mechanisms¹³ (Box 1). As a complement to the control of translation by mTORC1, MYC transcriptionally induces the expression of multiple components of the protein synthesis machinery, including ribosomal proteins and translation initiation factors¹³². MYC can directly stimulate Pol-I and Pol-III to transcribe rDNA and also promotes the expression of rRNA processing enzymes required to produce the mature forms that are assembled into ribosomes¹³². Interestingly cancers with high MYC are characterized by elevated rates of protein synthesis and ribosome biogenesis, and inhibition of protein synthesis has been shown to confer synthetic lethality in these cancers^138,139. Future studies are needed to understand whether cancer cells with oncogenic PI3K signaling share this vulnerability.

Box 1. Protein synthesis downstream of mTORC1.

Protein synthesis is a highly nutrient-and energy-costly process that is induced downstream of AKT signalling through mTORC1¹³. Much of this regulation occurs through phosphorylation of its two best established substrates, S6K and eukaryotic translation initiation factor 4E (eIF4E) binding protein (4E-BP). The mTORC1-mediated phosphorylation of 4E-BP triggers its release from eIF4E at the 5’-cap of mRNAs, allowing assembly of the translation initiation complex¹⁸⁹. This regulation has been found to be particularly important for the translation of mRNAs that have 5’-terminal oligopyrimidine (5’-TOP) or 5’-TOP-like sequences at the 5’-end of their 5’-UTRs, which encompass transcripts encoding the translation machinery, including ribosome proteins (rProteins) and translation factors^190,191 (Box Figure). Oncogenic activation of mTORC1 is believed to enhance the rate of protein synthesis largely through the 4E-BP1-eIF4E axis, to support cancer cell growth¹³⁸. Phosphorylation of another mTORC1 substrate, La-related protein 1 (LARP1), has also been implicated in the selective induction of translation of 5’-TOP mRNAs downstream of mTORC1^192–194. In addition to mTORC1, both AKT and S6K have been reported to phosphorylate LARP1 and relieve its inhibitory effect on translation by dissociating it from 5’-UTRs¹⁹². A key target of S6K in the induction of translation is eIF4B, which upon phosphorylation forms an active heterodimer with the RNA helicase eIF4A, thereby enhancing the unwinding of 5’-UTRs with complex secondary structures^77,195,196. Interestingly, translation of the MYC mRNA is particularly sensitive to the S6K-mediated phosphorylation of eIF4B⁷⁷. Finally, mTORC1 signaling enhances rDNA transcription through both Pol-I and Pol-III, thus producing the rRNA needed for assembly with newly translated rProteins into ribosomes¹³¹.

[H1] PI3K-AKT signaling and redox homeostasis

One consequence of the metabolic changes underlying cancer cell proliferation is an increase in the production of reactive oxygen species (ROS). Accumulating evidence suggest that moderate levels of ROS support cancer cell growth, proliferation, and survival, whereas too much ROS is detrimental to the macromolecular constituents of the cell¹⁴⁰. In addition to promoting anabolic metabolism and cell growth, the PI3K-AKT pathway also regulates multiple metabolic processes that modulate ROS levels.

[H2] Production of cellular reducing power

NADPH, a pyridine dinucleotide cofactor, is a key reservoir of electrons required for reductive biosynthesis and defence against oxidative stress (Fig. 5). Cancer cells upregulate NADPH-producing pathways to support an increased anabolic demand and to enhance their antioxidant capacity^122,141,142. Recent studies have addressed the routes of consumption and production of NADPH using quantitative flux analysis [G]^143–145. These analyses revealed that the majority of cytosolic NADPH is consumed by anabolic metabolism and particularly by fatty acid synthesis¹⁴⁴ (Fig. 5). Indeed, de novo synthesis of a single molecule of one of the most abundant fatty acids in human plasma, palmitate, by the multifunctional enzyme fatty acid synthase (FASN) requires 14 molecules of NADPH. Ribonucleotide reductase (RNR) requires NADPH to catalyse the reduction of ribonucleotide 5’-diphosphates (NDPs) to deoxyribonucleotides diphosphates (dNDP), which are subsequently converted to the dNTPs required for DNA replication¹⁴⁶. Additionally, NADPH is required for the activity of dihydrofolate reductase to produce the tetrahydrofolate for the reactions of one carbon metabolism. NADPH is also required for synthesis of the amino acid proline, which has emerged as an important process for cancer cell growth and survival¹⁴⁷. There are several metabolic enzymes that can reduce NADP⁺ to replenish NADPH, the levels of which can be influenced by PI3K-AKT signaling (Fig. 5). The majority of the cytosolic NADPH pool is derived from reactions within the oxidative branch of PPP¹⁴⁵, which can be stimulated downstream of AKT signaling through the mTORC1-SREBP axis⁶⁸ (Fig. 5). SREBP also promotes the expression of malic enzyme 1 (ME1), which can contribute to the cytosolic NADPH pool¹⁴⁸. Interestingly, both G6PD and ME1 expression are reciprocally repressed by the tumor suppressor p53, and their induction contributes to tumor growth in mouse models^149,150. Finally, SREBP¹⁵¹, as well as the FOXO transcription factors¹⁵², have been found to control the expression of isocitrate dehydrogenase 1 (IDH1), another cytosolic enzyme that can produce NADPH. The relative importance of these three major sources of cytosolic NADPH to the oncogenic effects of PI3K-AKT signaling requires further research.

Fig. 5: AKT signaling and control of NADPH production and consumption.

NADPH serves as a major electron donor for reductive biosynthesis and defence against ROS, reactions that yield its oxidized form, NADP⁺. In the cytosol, NADPH can be regenerated from NADP⁺ through reactions involving two enzymes of the oxidative PPP (G6PD and PGD), isocitrate dehydrogenase 1 (IDH1) and malic enzyme (ME). Downstream of AKT and mTORC1, SREBP induces expression of these enzymes to enhance NADPH production. AKT-mediated phosphorylation of NADK serves to boost NADP⁺ abundance to further enhance the production of NADPH through these redox reactions. Many cellular reactions consume a large quantity of NADPH as reducing power. NADPH is required for key reactions in the biosynthesis of macromolecules, including fatty acid synthase (FASN) to produce palmitate, ribonucleotide reductase (RNR) to convert ribonucleotide diphosphates (NDPs) into deoxyribonucleotide diphosphates (dNDPs), dihydrofolate reductase (DHFR) for tetrahydrofolate synthesis, and the proline synthesis enzymes pyrroline-5-carboxylate (P5C) synthase (P5CS) and pyrroline-5-carboxylate reductase (P5CR). NADPH also serves as essential reducing power for antioxidant enzymes including glutathione reductase (GR), which reduces the oxidized glutathione disulfide (GSSG) to glutathione (GSH), and thioredoxin reductase (TrxR), which transfers electrons from NADPH to reduce thioredoxin for subsequent reduction of oxidized cysteines.

ROS, reactive oxygen species; PPP, pentose phosphate pathway; NADPH, nicotinamide adenine dinucleotide phosphate (reduced); G6PD, glucose 6-phosphate dehydrogenase; PGD, phosphogluconate dehydrogenase; SREBP: Sterol regulatory element binding proteins; NADK, NAD kinase.

The size of the cellular pool of NADP⁺ and NADPH available for interconversion through reduction and oxidation (redox) reactions is determined, in part, by the activity of NAD kinase (NADK). NADK catalyzes the phosphorylation of NAD⁺ to produce NADP⁺, which is the rate-limiting substrate for NADPH-producing enzymes. In response to growth factors, or in a growth factor-independent manner in cancer cells, AKT directly phosphorylates and acutely stimulates an increase in NADK activity, resulting in increased production of NADP⁺ and, subsequently, NADPH¹⁵³ (Fig. 2 and Fig. 5). The AKT-mediated phosphorylation of NADK facilitates the anchorage-independent growth of cancer cells¹⁵³. This finding might help explain the key role of PI3K signaling in the survival of cells detached from the extracellular matrix, which requires a robust anti-oxidant response¹⁵⁴. The stimulated increase in NADK activity is also likely to serve as a key part of the broader anabolic program induced by the PI3K-AKT pathway, providing reducing cofactors in high demand for the biosynthetic processes underlying cell growth. Loss of NADK attenuates tumor growth in xenograft models^155,156, and an activating mutation in NADK, identified in pancreatic ductal adenocarcinoma, was found to increase tumor growth in a xenograft model¹⁵⁵. Future research is needed to define the role of NADK as a downstream target of AKT signaling in cancer development and progression and whether this enzyme is a viable metabolic target for cancer therapies.

[H2] Oxidative stress response

Activation of the PI3K-AKT pathway stimulates both ROS producing and scavenging mechanisms that vary between cellular settings. For instance, in neutrophils and macrophages, activation of PI3K-AKT signaling stimulates ROS production through regulation of the p47phox component of the NADPH oxidase NOX, thereby generating the respiratory burst required to eliminate extracellular and phagocytosed pathogens^157,158. In addition, AKT directly phosphorylates and activates endothelial nitric oxide (NO) synthase (eNOS), another NADPH oxidase, which produces NO in endothelial cells to control vascular tone but can also generate ROS^159,160. However, whether these or other specialized mechanisms downstream of PI3K-AKT signaling contribute to ROS production within cancer cells is unknown. Instead, most of the evidence to date, support an anti-oxidant role for PI3K-AKT signaling in cancer.

Elevated ROS, including superoxide (O₂⁻) and hydrogen peroxide (H₂O₂), can oxidize and damage lipids, protein, and nucleic acids and have detrimental consequences on cell growth and survival¹⁴⁰. Cells have multiple enzymes and systems to neutralize ROS, including superoxide dismutase (SOD), which converts O₂^- to H₂O₂, catalase, which reduces H₂O₂ into water, and the peroxiredoxin (Prx)/thioredoxin (Trx) and glutathione peroxidase (GPx)/glutathione (GSH) antioxidant systems, which utilize NADPH to reduce H₂O₂ into water and to repair macromolecules oxidized by exposure to ROS¹⁴⁰. It is interesting to note that the FOXO transcription factors, which are negatively regulated by AKT, can also function to detoxify ROS through the induction of several ROS scavenging systems^161–163. However, more research is necessary to understand the significance of the PI3K-AKT-FOXO circuit in redox control in the context of cancer.

Consistent with PI3K signaling playing an important role in the cellular response to ROS, an increase in ROS levels can activate the pathway through various mechanisms. H₂O₂ can influence cell signaling events by oxidizing cysteine residues on proteins, including the catalytic cysteine of protein and lipid phosphatases. These ROS-sensitive phosphatases include negative regulators of PI3K signaling, such as protein tyrosine phosphatase 1B, protein phosphatase 2A, and PTEN^164–166, the inhibitory oxidation of which can activate AKT in response to a rise in H₂O₂^165,167 (Fig. 6). Thus, together with downstream mechanisms to mitigate ROS levels, the PI3K-AKT pathway can serve as part of an adaptive oxidative stress response pathway.

Figure 6: Interplay between ROS and the PI3K-AKT pathway.

ROS in the form of hydrogen peroxide can activate the PI3K-AKT pathway through inactivation of protein phosphatases, including PTP1B, which attenuates the activity of insulin receptor and insulin receptor substrate (IRS), and PP2A, which normally dephosphorylates T308 on AKT, thus leading to enhanced AKT phosphorylation and activation in response to ROS. ROS also inhibits the lipid phosphatase PTEN, resulting in accumulation of PIP₃ and activation of AKT. AKT activation can facilitate the adaptation to ROS by activating the NRF2 transcription factor, which stimulates numerous enzymes to mount an antioxidant response. Downstream of AKT, GSK3 phosphorylates NRF2 and marks it for ubiquitination by the β-TRCP E3 ligase, leading to subsequent proteasomal degradation. Thus, through its inhibition of GSK3, AKT signaling can stabilize NRF2. NRF2 is stabilized and activated by ROS via cysteine oxidation of the E3 ligase Keap1, thereby disrupting its binding of NRF2. AKT also directly phosphorylates and inhibits the cystine-glutamate antiporter xCT, which transports cystine into cells that, upon NADPH-dependent reduction to cysteine, can be used to produce glutathione for ROS neutralization. xCT and the enzymes of glutathione synthesis are encoded by gene targets of NRF2 as part of its anti-oxidant response.

ROS, Reactive oxygen species; PTP1B, protein tyrosine phosphatase 1B; PP2A, protein phosphatase 2A; PTEN, phosphatase and tensin homolog; PDK1, phosphoinositide-dependent protein kinase 1; NRF2, nuclear factor erythroid 2-related factor 2; GSK3, Glycogen synthase kinase 3; NADPH, nicotinamide adenine dinucleotide phosphate (reduced), Keap1, kelch-Like ECH-associated protein 1; β-TRCP, β-transducin repeat containing protein.

As described above, AKT signaling increases cellular reducing power (NADPH) available for anti-oxidant responses (Fig. 5). However, the PI3K-AKT pathway also contributes to ROS detoxification through sustained activation of NRF2 (nuclear factor erythroid 2-related factor 2), a transcription factor that controls the expression of numerous genes involved in the antioxidant response, including the enzymes of glutathione synthesis and function, the thioredoxin system, NADPH regeneration, and ROS detoxification¹⁶⁸. NRF2 is cellular sensor of both oxidative stress and growth factor signaling through mechanisms that control its protein stability. NRF2 is targeted for rapid proteasome-mediated degradation through two distinct E3 ubiquitin ligases, Keap1 and β-TrCP (Fig. 6). Oxidative stress directly inhibits the binding of Keap1 to NRF2, leading to increased abundance of NRF2^169,170. Activation of the PI3K pathway also results in stabilization and activation of NRF2. AKT inhibits GSK3, which directly phosphorylates NRF2 and targets it for degradation by β-TrCP¹⁷¹ (Fig. 6). In addition, AKT has been proposed to phosphorylate and stabilize p21^Cip1/WAF1, allowing it to compete with KEAP1 for NRF2 binding, thus leading to NRF2 stabilization and activation of anti-oxidant gene targets^172,173.

Accumulating evidence indicates that NRF2 confers an advantage for aggressive cancer cell survival and proliferation by upregulating metabolic and antioxidant pathways¹⁷⁴. Recently, the PI3K-AKT-NRF2 axis was reported to contribute to both anabolic metabolism and ROS detoxification through control of metabolic genes involved in NADPH regeneration¹⁷⁵. Activation of AKT signaling through loss of PTEN induces NRF2 target genes to support proliferation and tumorigenesis^175,176. Moreover, oncogenic activation of the PI3K-AKT pathway in breast cancer induces an NRF2-dependent transcriptional program, which enhances glutathione biosynthesis to support tumor growth and confer resistance to oxidative stress¹⁷⁷. Interestingly, inhibition of glutathione biosynthesis was found to synergize with the chemotherapeutic agent cisplatin and induce tumor regression in PI3K-driven breast cancer models¹⁷⁷, suggesting that the anti-oxidant response downstream of PI3K-AKT signaling represents a targetable metabolic vulnerability.

NRF2 also enhances glutathione synthesis through transcriptional induction of the cystine-glutamate antiporter xCT (or SLC7A11)¹⁷⁸, which imports cystine molecules (oxidized dimers of cysteine) that are subsequently reduced to cysteine, an essential substrate for glutathione synthesis. Paradoxically, xCT function has been shown to be attenuated downstream of growth factor-stimulated and oncogenic PI3K signaling^179,180. AKT was found to directly phosphorylate xCT and decrease its cystine transport activity¹⁷⁹, thereby rendering cells with oncogenic PI3K signaling dependent on endogenous cysteine synthesis (Fig. 6). Since cystine reduction to cysteine requires NADPH¹⁸¹, acute inhibition of xCT by AKT could serve to preserve NADPH for use in lipid synthesis downstream of AKT-mediated activation of ACLY. Inhibition of the xCT antiporter could also serve to spare glutamate and glutamine nitrogen to be used for nitrogen-dependent synthesis of amino acids and nucleotides. Perturbations in the balance between anti-oxidant and biosynthetic activities of cancer cells imposed by oncogenic PI3K-AKT signaling is a particularly interesting and active area of investigation with potential to reveal new therapeutic approaches.

[H1] Clinical perspective

The PI3K-AKT pathway has emerged as one of the most frequently activated drivers of human cancer, making it a prime candidate for therapeutic intervention. Specific inhibitors targeting both PI3K and AKT have been developed as cancer therapies, with most trials demonstrating limited therapeutic benefit as single agents¹⁸². Due to the critical role of PI3K-AKT signaling in insulin-responsive glucose uptake into tissues, such as skeletal muscle, pan-PI3K inhibitors inevitably cause hyperglycemia, with the consequent hyperinsulinemia being shown to overcome pathway inhibition and reactivate PI3K signaling in tumors¹⁸³. Dietary interventions, such as a ketogenic diet, have been found to alleviate this hyperglycemia and hyperinsulinemia and improve responses to these inhibitors in mouse cancer models¹⁸³. Furthermore, resistance to PI3K inhibitors can arise due to redundant regulation of key downstream effectors such as mTORC1¹⁸⁴. Importantly, patient stratification for oncogenic PIK3CA mutations, which are found in approximately 40% of ER+, HER2- breast cancers, together with use of a p110α-selective PI3K inhibitor (BYL719, trade named Piqray) yielded improved clinical responses when used in combination with an estrogen receptor antagonist¹⁸⁵. Preclinical and clinical studies continue to focus on understanding the tumor response to PI3K inhibitors and to identify both combination therapies and unique vulnerabilities arising from uncontrolled PI3K signaling in distinct cancer settings. As discussed above for nucleotide synthesis, pharmacological targeting of specific metabolic enzymes and pathways induced downstream of PI3K-AKT signalling in cancer may offer effective alternative therapies to PI3K inhibitors for cancer treatment.

[H1] Conclusion

Research in the past two decades has defined how aberrant activation of the PI3K signaling pathway drives tumorigenesis, at least in part, through the control of metabolism. Future research aimed at refining our molecular map of the critical regulatory nodes connecting the oncogenic PI3K signaling network to metabolic networks in different cancers will help reveal metabolic dependencies and novel therapeutic strategies. One example of this is the finding that PI3K inhibitors can deplete intracellular nucleotide pools and cause DNA replication stress, which when combined with inhibitors of poly ADP-ribose polymerase (PARP), an enzyme involved in DNA repair, increases the anti-cancer efficacy of PI3K inhibitors in tumor models and a subset of cancer patients^{118,186–188}. Metabolic enzymes are inherently druggable and offer a wealth of new targets to which specific pharmacological inhibitors can be developed, as new metabolic outcomes and vulnerabilities are identified. Finally, while we focused here on the cancer cell intrinsic regulation of metabolism, it is clear that the physiological features of the tissue of origin for a given tumor and the stromal cell milieu, the nutritional and metabolic status of the host, and distinct metabolic niches of sites of distant metastases will all differentially influence cancer cell metabolism and metabolic dependencies as they relate to cancers with oncogenic PI3K-AKT signaling.

Glossary:

redox homeostasis

Maintaining proper levels of cellular NAD(P)+ and NAD(P)H for metabolic reduction and oxidation (redox) reactions

metabolic flexibility

The ability of a cell to adapt its metabolism in response to changing environmental conditions, such as nutrient and energy availability

anaplerotic metabolism (or anaplerosis)

Metabolic reactions that replenish TCA cycle intermediates used for biosynthetic processes

glutaminolysis

The two-step removal of the amide and amine nitrogens from glutamine to produce the TCA cycle intermediate α-ketoglutarate, reactions that can serve as one form of anaplerosis

anabolic processes

Metabolic processes and pathways that utilize nutrients and ATP to generate macromolecules such as proteins, lipids and nucleotides

quantitative flux analysis

Measurement of the rate of consumption and production of metabolites in specific metabolic pathways, often achieved through the tracing of stable isotope-labelled nutrients and quantification via mass spectrometry