Harnessing metabolic dependencies in pancreatic cancers.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive disease with a 5-year survival rate of <10%. The tumour microenvironment (TME) of PDAC is characterized by excessive fibrosis and deposition of extracellular matrix, termed desmoplasia. This unique TME leads to high interstitial pressure, vascular collapse and low nutrient and oxygen diffusion. Together, these factors contribute to the unique biology and therapeutic resistance of this deadly tumour. To thrive in this hostile environment, PDAC cells adapt by using non-canonical metabolic pathways and rely on metabolic scavenging pathways such as autophagy and macropinocytosis. Here, we review the metabolic pathways that PDAC use to support their growth in the setting of an austere TME. Understanding how PDAC tumours rewire their metabolism and use scavenging pathways under environmental stressors might enable the identification of novel therapeutic approaches.

ToC Blurb

Pancreatic ductal adenocarcinoma (PDAC) has a unique tumour microenvironment. Notably, PDAC can reprogramme metabolism in response to this microenvironment. This Review discusses metabolism in pancreatic cancer, including insights into mechanisms and processes as well as the potential therapeutic applications.

Pancreatic ductal adenocarcinoma (PDAC) is a lethal disease with a 5- year survival rate of less than 10%¹. This poor prognosis is due to a number of factors, including lack of early detection methods and an intense therapeutic resistance². Over the past decades, multiple studies have identified the recurrent genetic alterations associated with PDAC development and progression^3,4. These alterations include the oncogenic activation of KRAS, which occurs in >90% of patients, and inactivation of tumor suppressor genes such as TP53, SMAD4, and CDKN2A³. Oncogenic activation of KRAS is an early event in the progression to PDAC, whereas deletion or inactivation of TP53 occurs at the later stages. Other mutations have been identified in PDAC patients; these include but are not limited to GNAS⁵, RNF43⁶, KDM6A⁷, and BRAF^8,9. The functional consequences of these mutations have been validated using genetically engineered mouse models (GEMMs) and human cell line models^10,11,12. Unfortunately, the deep understanding of PDAC genetics has not yet translated into effective therapies in this disease.

Another characteristic feature of PDAC is its distinctive tumour microenvironment (TME). The PDAC TME is composed of an abundant and dense fibrous stroma (desmoplasia), consisting of extracellular matrix (ECM) such as collagen and hyaluronic acid¹³. Indeed, it is common for desmoplasia to constitute the bulk of the tumour mass while PDAC tumour cells often make up only a small fraction in both human and animal models¹⁴. Also, a heterogeneous cellular population is present in the TME that includes cancer-associated fibroblasts (for example, pancreatic stellate cells (PSCs), a specialized type of fibroblast similar to hepatic stellate cells)¹⁵, immune cells, endothelial cells and neurons. The dense stroma promotes the collapse of the tumour vasculature, resulting in poor oxygen diffusion (hypoxia)¹⁶. Indeed, PDAC is amongst the most hypoxic of human tumours^17,18. Although there are technical challenges in assessing the absolute amounts of metabolites in the TME, the collapse of the tumour vasculature and the nutrient competition between different cell types within the TME likely lead to a decrease in nutrient availability. Indeed, attempts at measuring metabolites in PDAC compared with normal pancreas in mouse models and human PDAC, as well as in tumour interstitial fluid show varying degrees of nutrient limitation^19,20.

To survive and thrive in this hostile environment, PDAC cells can reprogramme metabolism in response to nutrient and oxygen fluctuations in the TME^21,22. Understanding how PDAC rewires metabolism could provide insights into its pathogenesis and ultimately enable the development of improved therapies. In this Review, we will emphasize three major areas of metabolism in pancreatic cancer: central carbon metabolism and fuel source utilization, nutrient scavenging mechanisms, and metabolic crosstalk in the complex TME. Lastly, we will provide an overview of the current advances in identifying and targeting metabolic dependencies in pancreatic cancers.

Genetic alterations and metabolism

Alterations in tumour metabolism have been recognized as a hallmark of cancer²³. Indeed, to support unconstrained proliferation, cancer cells have increased requirements for biomass, energy and macromolecules. Numerous studies have identified the role of oncogenic mutations in dysregulating cellular metabolism in PDAC and other malignancies²⁴. In PDAC, oncogenic KRAS is linked to an increase in glucose uptake and shunting toward anabolic pathways, glutamine reprogramming and reactive oxygen species (ROS) control^25,26. Also, to deal with scarcity or imbalance of nutrient availability, oncogenic KRAS is associated with the activation of unique metabolic scavenging pathways²⁷. In 2020, it was shown that oncogenic KRAS can also reprogramme the metabolism of PDAC by non-cell intrinsic factors, such as upregulating cytokines, and promoting immune cell infiltration that drives tumour metabolic alterations in a PDAC mouse model^28,29.

Importantly, other mutations, amplifications, or deletions that occur concurrently with oncogenic KRAS also alter PDAC metabolism in distinct ways. These include mutations in the G-protein αs (such as GNAS), TP53, PTEN and MYC. GNAS-activating mutations drive a shift in the utilization of lipids and increase fatty acid oxidation, promoting PDAC tumour progression³⁰. The transcription factor MYC is considered a master regulator of many metabolic processes in many cancers including PDAC^31,32,33. Multiple studies have established that MYC is an important component of PDAC progression in model systems through both extrinsic and intrinsic factors^34,35,36. The tumor suppressor p53, which is frequently mutated in PDAC development, controls cellular metabolism by inducing the expression of different transcriptional programmes such as the autophagy network³⁷ and by controlling ROS through the induction of the p53-target TIGAR (TP53-induced glycolysis and apoptosis regulator)^38,39. Additionally, it has been demonstrated that restoration of wild-type p53 in a PDAC mouse model induces accumulation of α-ketoglutarate, leading to increased chromatin accessibility that ultimately lead to tumour suppression⁴⁰. Lastly, the tumour suppressor PTEN, a lipid phosphatase, regulates the activity of the phosphoinositide-3-kinase (PI3K)–AKT pathway, which is involved in many metabolic processes including glucose metabolism, redox control and de novo lipid synthesis⁴¹. PTEN expression is commonly lost in PDAC, leading to hyperactive PI3K–AKT signalling that cooperates with mutant KRAS to promote tumour initiation and progression in multiple PDAC models^42,43,44,45. In summary, oncogenic KRAS in concert with other genetic and/or genomic events has a critical role in driving metabolic shifts in PDAC tumours.

Fuel source utilization

Glucose.

The glucose molecule stores high levels of chemical energy and is a major source of energy in the cell^46,24. During glycolysis, glucose undergoes a series of reactions that end up in the production of energy in the form of ATP and two molecules of pyruvate. Pyruvate can be transported to the mitochondria to feed into the tricarboxylic acid (TCA) cycle (described in the next section) or converted to lactate in the cytosol. Indeed, Otto Warburg in the 1920s observed that cancer cells tend to undergo glycolysis even in the presence of oxygen, producing high levels of lactate⁴⁷. Importantly, glucose can be diverted from glycolysis into side branches that produce key metabolites needed for biomass, these include: the pentose phosphate pathway, which is responsible for the production of the cofactor nicotinamide adenine dinucleotide phosphate (NADPH), as well as for the production of ribose 5-phosphate, an essential intermediate of ribose synthesis⁴⁸; and the hexosamine biosynthetic pathway, responsible for the synthesis of uridine diphosphate N-acetyl glucosamine (UDP-GlcNAc), an important substrate for protein glycosylation⁴⁹. Glycolytic intermediates can also be utilized as precursors for the serine–glycine–one-carbon metabolism pathway, supporting nucleotide synthesis, methylation reactions, and lipid synthesis^50,51.

As glucose metabolism has so many roles in cellular metabolism, it is not surprising that oncogenic signalling (such as KRAS or MYC) drives increased uptake and downstream utilization of glucose to support tumour growth. Through an integrated metabolomics and transcriptomic analysis of a mouse model of PDAC, it was demonstrated that oncogenic KRAS^G12D enhances glucose utilization in PDAC through the transcriptional upregulation of glucose transporter GLUT1 and the enzymes hexokinase 1, hexokinase 2, phosphofructokinase 1, and lactate dehydrogenase A²⁵. The transcriptional upregulation of these genes leads to an overall increase in glycolytic flux. Additionally, KRAS^G12D increased the expression of glutamine-fructose-6-phosphate transaminase 1, the first and rate-limiting step of the hexosamine biosynthetic pathway, and increased flux through this pathway²⁵. It was also shown that KRAS^G12D increased flux of glucose into the non-oxidative arm of pentose phosphate pathway by increasing the expression of the rate-limiting enzymes ribulose-5-phosphate-3-epimerase and ribose-5-phosphate isomerase. Importantly, genetic inhibition of the hexosamine biosynthetic pathway or pentose phosphate pathway in vitro and in vivo reduced PDAC growth²⁵. Further studies both in vitro and in vivo systems demonstrated these KRAS-driven metabolic reprogrammings were orchestrated by the RAF–MEK–ERK pathway which led to MYC upregulation⁵². Indeed, the response to MEK inhibition in PDAC correlates with a decrease in MYC and ribose-5-phosphate isomerase levels⁵². Lastly, it was shown that in addition to the transcriptional changes that oncogenic KRAS induces in cancer cells, it can also directly interact with effectors from the glycolytic pathway. The KRAS 4A isoform directly interacts with hexokinase 1, leading to enhanced glycolytic flux both in vitro and in vivo⁵³. Taken together, these studies demonstrated that oncogenic KRAS induces metabolic reprogramming that drives glucose usage for anabolic reactions (Fig. 1A).

Figure 1. Metabolic alterations in pancreatic cancers.

²²(A) In pancreatic cancer cells, oncogenic KRAS^G12D upregulates the uptake of glucose and enhances glycolysis. Additionally, glycolytic intermediates are shunted into the pentose phosphate pathway (PPP) and the hexosamine biosynthesis pathway. Ultimately, this metabolic rewiring produces metabolites required for the synthesis of DNA and/or RNA and substrates for glycosylation. (B) In pancreatic cancers, the non-essential amino acid glutamine is metabolized by a non-canonical pathway, whereby glutamine-derived glutamate is converted into aspartate. Aspartate is transported into the cytosol where it undergoes a series of metabolic reactions that ultimately lead to increased NADPH production contributing to redox balance. Abbreviations: HK, Hexokinase; GFAT, Glucosamine—fructose-6-phosphate aminotransferase isomerizing; GlcNAc, Uridine diphosphate N-acetylglucosamine; Gln, Glutamine; Glu, Glutamate; Ala, Alanine; Asp, Aspartate; Cit, Citrate; Suc, Succinate; Fum, Fumarate; Mal, Malate; OAA, Oxaloacetic acid.

Glutamine.

Mitochondria are critical for providing both cellular energy and the anabolic substrates required for cellular growth and homeostasis⁵⁴. To do so, mitochondria maintain enzymes and metabolite pools distinct from other organelles within the cell. Because many of these metabolites cannot be directly transported across mitochondrial membranes, biochemical reactions are compartmentalized within the mitochondria⁵⁵. Indeed, the mitochondria play an important part in the synthesis of ATP through the activity of the TCA cycle and oxidative phosphorylation. A major input into the TCA cycle is the entry of pyruvate into the mitochondria through the mitochondrial pyruvate carrier⁵⁶. However, other substrates such as fatty acids and certain amino acids such as glutamine can feed into the TCA cycle. Multiple groups have demonstrated that mitochondrial metabolism is critical for tumour growth in multiple types of cancer including PDAC^57,58 (discussed later).

Glutamine is a non-essential amino acid that is highly abundant in human plasma and has multiple roles in cellular metabolism^59,60. The carbon skeleton of glutamine is important to the contribution of TCA cycle intermediates. In the first step of glutaminolysis, glutamine is converted into glutamate by the glutaminase enzyme. Classically, glutamine-derived glutamate is then converted into alpha-ketoglutarate through a deamination reaction catalyzed by glutamate dehydrogenase (GLUD1). Then, α-ketoglutarate enters the TCA cycle replenishing the metabolic intermediates, a process that generates ATP through the production of NADH and FADH2.

Besides the role of glutamine in the TCA cycle, glutamine has other biosynthetic roles in cell metabolism. For example, the amine portion of glutamine can serve as a nitrogen donor for the synthesis of other amino acids, hexosamines, as well as supports the de novo synthesis of pyrimidines and purines⁶¹. Glutamine also contributes to the production of the cofactor NADPH and glutathione, both of which have roles in cellular redox balance^60,62. Given its abundance in the plasma and critical roles in multiple metabolic pathways, is not surprising that many studies have established that cancer cells rely on glutamine for growth and survival.

In PDAC, glutamine has been shown to be critical for redox homeostasis through a non-canonical pathway⁶³. Mechanistically, this process occurs by the conversion of glutamate to α-ketoglutarate and aspartate in the mitochondria by the mitochondrial aspartate aminotransferase (GOT2). Then, aspartate is transported into the cytoplasm where it is subsequently metabolized by cytosolic GOT1, malate dehydrogenase (MDH1), and malic enzyme (ME1). Ultimately, these metabolic reactions produce pyruvate and increased reducing power in the form of NADPH. Genetic inhibition of key enzymes (GOT1, MDH1, and ME1) leads to a decrease in PDAC growth both in vivo and in vitro⁶³ (Fig. 1B). This study demonstrated that the metabolism of glutamine in PDAC cells depend more on the transamination reactions of GOT1 and GOT2 than GLUD1. Along these lines, Raho and colleagues showed that the mitochondrial uncoupling protein 2 (UCP2) catalyzes the efflux of glutamate-derivate aspartate in human PDAC⁶⁴. Genetic inhibition of UCP2 disrupts redox balance in vitro and reduces tumor growth in a xenograft PDAC model⁶⁴.

In 2019, it was demonstrated that PDAC requires the glutamate ammonia ligase to synthesize glutamine and enable efficient growth in glutamine-deprived environments. Indeed, the ablation of glutamate ammonia ligase reduced PDAC tumour growth in an autochthonous model model⁶⁵. Similarly, it was demonstrated in human PDAC that the SLC1A5 gene has a novel variant (SLC1A5_var) that is localized in the inner membrane of the mitochondria and acts as a glutamine transporter⁶⁶. This variant of SLC1A5 is dependent on the transcription factor HIF-2α and ectopic expression of this transporter in human PDAC lines increased mitochondrial respiration, glutathione synthesis, and promoted chemotherapy resistance in vitro⁶⁶. In addition, inhibition of SLC1A5_var resulted in a reduction in tumour growth in a PDAC xenograft mouse model. In summary, glutamine has multiple roles in PDAC metabolism, suggesting that targeting the acquisition and utilization of this amino acid could be a promising therapeutic strategy (see later). However, in contrast to its many roles in supporting tumour growth, glutamine depletion has been shown to increase the metastatic potential of PDAC in preclinical models. Indeed, it was shown that glutamine deprivation drives the epithelial-to-mesenchymal transition (EMT) signature both in vitro and in vivo through upregulation of the transcription factor Slug1 in a process dependent on the ATF4 and the MEK–ERK signaling cascade⁶⁷. Thus, it will be critical to thoroughly evaluate the effect of inhibiting distinct aspects of glutamine metabolism on PDAC growth and/or metastasis.

Lysosomal nutrient scavenging

The TME of PDAC is characterized by high stromal content, hypoxia and low nutrients compared to adjoining pancreatic tissue^19,68. To adapt to these conditions, PDAC relies on the use of lysosome nutrient scavenging pathways such as autophagy and macropinocytosis to provide molecular substrates that can support cellular homeostasis under stressful conditions^27,69,70,71. PDAC upregulates transcriptional programmes that orchestrate lysosomal biogenesis and autophagy induction through the function of the MiT–TFE family of transcription factors— MITF, TFE3 and TFEB⁷². In PDAC, these transcription factors are constitutively active in the nucleus, leading to transcriptional activation of genes in the autophagy–lysosome network. Inhibition of MiT–TFE proteins leads to a metabolic crisis due to the impairment of both autophagy flux and macropinocytosis⁷². These findings demonstrated that nutrient scavenging is a mechanism of survival and progression in PDAC. Further investigation is necessary to understand how the various aspects of the autophagy–lysosome network are coordinated upon environmental stressors such as nutrient starvation and hypoxia.

PDAC requires autophagy to support tumour growth

Macroautophagy (hereafter referred to as autophagy) is a well-conserved catabolic process that sequesters organelles and macromolecules and targets them for lysosomal degradation to maintain cellular homeostasis and survival during periods of starvation or stress^71,73. Initially, autophagy was thought to be a non-selective process, however, studies have shown that autophagy can be highly selective⁷⁴. The molecular mechanism of autophagy is tightly coordinated by more than 30 autophagy-related (ATG) proteins that are responsible for the autophagy cascade and are broken down into four major steps: initiation and nucleation of the autophagosome; autophagosome closure; autophagosome–lysosome fusion; lysosomal degradation and recycling (Fig. 2).

Figure 2. Pancreatic cancers rely on autophagy and lysosomal scavenging.

(A)⁷⁰The molecular mechanism of autophagy can be categorized by four main stages: initiation and/or elongation, closure, fusion with the lysosome, and degradation. Nutrient starvation is a major trigger for autophagy activation through mTOR and AMPK. A series of protein–lipid interactions convert pro-LC3B into LC3B-I by ATG4B. LC3B is subsequently conjugated into phosphatidylethanolamine (PE), producing LC3B-II, via ubiquitin conjugation-like E1-E2-E3 series enzymes ATG7, ATG3, and ATG12-ATG5-ATG16L1 respectively. LC3B-II is associated with the autophagosome and acts as a docking site for autophagy cargo receptors. Cargo, such as damaged proteins or organelles are sequestered in a double membrane vesicle called an autophagosome. Ultimately, the autophagosome fuses with the lysosome, and the content is degraded. Degraded products are shuttled back into the cytoplasm where they can be used in various metabolic pathways. Autophagy and lysosomal degradation can be inhibited by compounds such as hydroxychloroquine, Lys05, and bafilomycin (BAFA1). Other compound tools targeting other modulators of autophagy such as ATG4B inhibitors are currently being tested. (B) Macropinocytosis, a non-selective endocytic pathway, allows extracellular content, such as albumin, to be degraded by the lysosome and can provide amino acids and other metabolites to the cell. Macropinocytosis can be inhibited by EIPA (5-(N-ethyl-N-isopropyl)-Amiloride).

A decade ago, our lab demonstrated that PDAC has elevated levels of basal autophagy and inhibition of autophagy either by RNA interference or pharmacologically using chloroquine (an inhibitor of lysosomal acidification), impaired mitochondrial metabolism, decreased proliferation in vitro, and reduced tumour growth in vivo⁷⁵. The importance of autophagy in PDAC tumorigenesis was further confirmed by crossing the autochthonous PDAC mouse model with a conditional knockout of the essential autophagy gene, Atg5 (Atg5^L/L). This study, along with others, supported the critical role of autophagy in the progression of PDAC^76,77,78.

To further investigate how autophagy supports the progression of PDAC in vivo, a new GEMM PDAC model was developed that enables the acute and reversible inhibition of autophagy under temporal and spatial control using a combination of the Cre-Lox and Dox-inducible systems⁷⁹. This model utilized a dominant negative mutant of ATG4B (ATG4B^C74A), a cysteine protease essential for the conjugation of LC3 with phosphatidylethanolamine, which potently inhibits autophagy. The ATG4B mutant was crossed into the well-established pancreatic cancer ‘KPC’ (LSLKras^G12D/+; Trp53^L/+;Ptf1^Cre) mouse model. Here, it was shown that autophagy inhibition could decrease the growth of established tumours in an autochthonous mouse model. An additional key finding was the fact that autophagy promotes PDAC growth through both cell-autonomous and non-autonomous mechanisms. Inhibition of autophagy in the host (while intact in the tumour cells) impaired tumour growth, likely in part through disruption of a metabolic crosstalk (see later)⁷⁹. Furthermore, these studies also demonstrated an important role in the antitumour response to autophagy inhibition

Although progress has been made in the study of the role of autophagy in PDAC, little is known about the molecular substrates that autophagy provides under environmental stress. While autophagy inhibition decreases oxidative phosphorylation in PDAC, the metabolic causes of this phenotype have not been fully identified⁷⁵. For example, it has been shown that in a Kras-driven lung cancer mouse model, that autophagy-deficient cells (Atg7^−/−) were more susceptible to nutrient starvation than wild-type cells. It was further demonstrated that, in this context, autophagy is critical to maintaining nucleotide pools under starvation conditions, to support cellular survival⁸⁰. In another study using an LKB1-deficient lung cancer model in which the essential autophagy gene Atg7 was conditionally deleted, it was demonstrated that autophagy-deficient cells increase de novo fatty acid synthesis to maintain lipid levels and support energy production through fatty acid oxidation⁸¹. These findings suggest that autophagy is an important mechanism to ensure metabolic homeostasis in other types of cancers during periods of stress.

Non-metabolic role of autophagy–lysosome in PDAC

Pancreatic cancers are characterized by a lack of response to immunotherapies such as anti PD-1 and anti-CTL4A and multiple reasons have been proposed to explain the immunosuppressive PDAC TME^82,83. Human and murine PDAC tumours have been shown to have decreased expression of major histocompatibility complex (MHC) class I molecule on the plasma membrane, which is further reduced in metastatic lesions in the liver^84,85. In 2020, it was demonstrated in human and murine PDAC that MHC-I is targeted for degradation through an autophagy-dependent mechanism using the autophagy cargo receptor NBR1 to induce immune evasion⁸⁶. Genetic inhibition of autophagy machinery or pharmacological inhibition (chloroquine) restored the surface levels of MHC-I. Autophagy inhibition led to an increase in anti-tumour T cell activation, infiltration and cytotoxicity in a syngeneic orthotopic transplant PDAC model. Lastly, autophagy inhibition synergizes with dual immune checkpoint blockade therapy in a syngeneic PDAC model⁸⁶. Taken together, these findings demonstrate a molecular mechanism by which autophagy enhances immune evasion and provides further rationale for targeting the autophagy–lysosomal system in PDAC.

Macropinocytosis

Macropinocytosis, a nutrient-scavenging pathway that has been identified as an essential mechanism to support pancreatic cancer growth⁶⁹n, is a non-selective endocytic programme that can uptake content from extracellular fluid through large vesicles known as macropinosomes. PDAC cells can uptake extracellular protein (such as serum albumin) from the extracellular fluid, and this content is then degraded by the lysosome, providing amino acids that eventually fuel central carbon metabolism⁸⁷. This mechanism can support survival and growth of PDAC and other Ras-driven tumours under nutrient-limiting conditions⁶⁸.

It has been known for many years that oncogenic RAS induces macropinocytosis⁸⁷. Moreover, the activity of other effectors such as the PI3K signalling pathway and small GTPases, Rac1 and Cdc42 are required⁸⁸. Using a short-interfering RNA screen to comprehensively identify regulators of micropinocytosis, it was demonstrated that the vacuolar ATPase (V-ATPase) is a critical effector of Ras-dependent macropinocytosis⁸⁹. This study supports a model in which oncogenic KRAS promotes the translocation of the v-ATPase to the plasma membrane through the activity of protein kinase A and Rac1. In another study, it was demonstrated that human PDAC xenografts have decreased levels of amino acids, such as glutamine, at their core⁹⁰, which correlates with the amount of macropinocytosis. This work provides a model whereby epidermal growth factor receptor signaling regulates glutamine deprivation-induced macropinocytosis through the activation of the p21 activated kinase. Using an unbiased functional proteomic screen in the doxycycline-inducible KRAS mouse model of PDAC, Yao and colleagues identified syndecan 1 as a novel effector of macropinocytosis in PDAC⁹¹. Furthermore, Hobbs and colleagues demonstrated that PDAC cell lines harbouring KRAS^G12R activating mutations are unable to induce macropinocytosis as this mutant fails to interact with p110α, an essential effector of KRAS signalling. It was further shown that PDAC cells harboring KRAS^G12R have distinct drug sensitivity profiles compared to those harbouring KRAS^G12D or KRAS^{G12V 92}. Understanding the metabolic and signalling differences between distinct mutant alleles of KRAS could help stratify patients with PDAC for appropriate therapies.

Taken together, these studies suggest that macropinocytosis could be a potential target for PDAC. As autophagy and macropinocytosis both converge at the lysosome, it will be important to determine whether there are any compensatory scavenging mechanisms under stress when either of these pathways are inhibited. Understanding how macropinocytosis and autophagy coordinate in response to nutrient deprivation, hypoxia, or other TME stressors will be critical to understanding the metabolic scavenging response in PDAC.

Tumor–stroma crosstalk

Cancer-associated fibroblasts (CAFs) are a predominant cellular population within the TME of PDAC^93,94. Several years ago, it was demonstrated that pancreatic CAFs can secrete metabolites to support cellular metabolism in PDAC cells in vitro and in vivo⁹⁵. Through a combination of biochemical and metabolic assays, it was shown that CAFs provide alanine to pancreatic cancer cells to support tumour growth in PDAC. Further characterization utilizing stable-isotope tracing and flux analysis demonstrated that alanine has pleiotropic roles in the metabolism of PDAC cells by contributing to the TCA cycle, synthesis of proteinogenic amino acids, de novo synthesis of fatty acids and transamination reactions. Interestingly, pancreatic cancer cells selectively express the mitochondrial glutamic-pyruvic transaminase 2, creating a unique metabolic situation whereby alanine is used to supply mitochondrial pyruvate to fuel the TCA cycle while allowing cytosolic pyruvate to be utilized to maintain cytosolic NAD+ regeneration⁹⁶. This metabolic crosstalk between PDAC cells and CAFs was characterized in a panel of PDAC cell lines and xenograft models demonstrating that this interaction is enabled by specific transporters⁹⁶. Further analyses demonstrated that PDAC requires the neutral amino acid transporter SLC38A2 to uptake alanine⁹⁶. On the other hand, CAFs utilize SLC1A4 to secrete alanine. Loss of SLC38A2 causes an unrecoverable metabolic crisis and impairs PDAC growth. Disruption of SLC38A2 in non-transformed cells was dispensable, suggesting that targeting this transporter could be promising in PDAC therapy. This work highlights the importance of dissecting the metabolic niches that occur between tumour cells and their environment.

Another example of how CAFs can interact with PDAC cells demonstrated that CAFs can secrete specific lipids to support PDAC progression⁹⁷. Auciello and colleagues showed that CAFs can secrete lipids such as lysophosphatidylcholines, which support biomass production in PDAC cells through a mechanism dependent on autotaxin–lysophosphatidic acid signalling. Inhibition of autotaxin signalling, both genetically and pharmacologically, leads to a decrease in PDAC tumour progression in vivo⁹⁷. Another study demonstrated that CAFs not only can support the metabolism and progression of PDAC, they can also contribute to chemotherapy resistance. CAFs secrete the pyrimidine, deoxycytidine, which acts as an inhibitor of the effects of the antimetabolite gemcitabine in human PDAC cells in vitro^98,99 A similar mechanism was shown by the Lyssiotis group, who demonstrated that tumour-associated macrophages in the TME of PDAC secreted deoxycytidine, which inhibits the uptake and incorporation of gemcitabine both in vitro and in vivo¹⁰⁰. The pharmacological or genetic ablation of tumour-associated macrophages led to an increase in gemcitabine response in vivo using a syngeneic PDAC model¹⁰⁰. In addition to metabolism, CAFs can also secrete cytokines and chemokines such as CXCL12, IL-6, and LIF that support PDAC progression^101,102,103. In another report, Francescone and colleagues identified Netrin G1 (NetG1) as an important player in PDAC progression¹⁰⁴. They showed that NetG1+CAFs support PDAC growth under nutrient deprivation through secretion of glutamine. Interestingly, genetic inhibition ablation or neutralization of NetG1 in CAFs increase anti-tumor function of natural killer (NK) and decrease tumor burden¹⁰⁴.

It has been shown that human PDAC tumors have high neuronal innervation, which correlates with poor prognosis^105,106. In this regard, multiple studies have demonstrated the protumorigenic effects of neurotrophic factors or neurotransmitters secreted by neurons in PDAC^{107,108,109,110}. However, these studies have not investigated the metabolic contribution of neurons to PDAC. Banh and colleagues demonstrated a novel metabolic crosstalk where peripheral axons release serine to support PDAC growth and survival¹¹¹. Indeed, the exogenous serine (Ser) was critical to support the growth of ~40% of human PDAC lines that do not express PHGDH, an important enzyme in the Ser biosynthesis pathway. Interestingly, Ser appeared dispensable for the canonical metabolic roles that have been described in cancer. Depletion of Ser selectively reduced mRNA translation rates on specific Ser codons. This allowed the translation and secretion of particular proteins with a favorable codon mix, such as nerve growth factor (NGF), thereby increasing neuronal innervation in the setting of low Ser levels. To evaluate the role of Ser depletion in tumor innervation, the authors injected human PDAC lines into the pancreata of mice and exposed them to a nutrient-rich diet or a Ser-Gly free diet. Strikingly, the Ser-Gly free diet induced increased innervation in PDAC tumors. Pharmacological inhibition of neuronal innervation by LOXO-101, an FDA-approved inhibitor of the NGF receptor, decreased tumor burden in Ser-Gly deprivation in Ser-dependent PDAC¹¹¹. In summary, this study demonstrated a novel metabolic crosstalk between neuronal axons and PDAC that could lead to novel therapeutic opportunities¹¹¹.

In summary, the role of stroma in the progression of PDAC is an area of extensive research as these cells can have both tumorigenic and anti-tumour effects. The implementation of single-cell analyses is beginning to provide insights into the heterogeneous role of CAFs and other cell types in the TME of multiple types of cancers, including PDAC⁹⁴. Although not in pancreatic cancer models, additional types of nutrient sharing and/or competitions have been identified that affect nutrient viability^112,113. With the increased understanding of the microbiome in cancer^114,115 as well as the multitude of cell types in the complex TME of PDAC^{116,117,118,119}, new insights into tumor-stroma crosstalk are anticipated.

Targeting metabolic liabilities

lysosomal scavenging in pancreatic cancers

Having demonstrated the importance of autophagy in the progression of pancreatic cancer, multiple clinical trials have begun to evaluate the use of the lysosomal inhibitor, hydroxychloroquine for PDAC treatment. Although hydroxychloroquine treatment demonstrated only minimal activity as a monotherapy, the combination of hydroxycholoroquine with standard chemotherapy (gemcitabine and nab-paclitaxel (GA)) markedly improved response rates in randomized trials of localized and metastatic PDAC^{120,121,122,123}. In the past few years, through various approaches, several groups found that inhibition of either MEK or ERK in pancreatic cancers increases the level of autophagy in preclinical experimental models^124,125,126. Treatment with hydroxychloroquine and MEK and/or ERK inhibitors showed synergistic responses in various models of pancreatic cancer including patient-derived xenografts. Moreover, the treatment of hydroxychloroquine and a MEK inhibitor (Trametinib) reduced the tumour burden in a single patient, leading to the development of multiple clinical trials to test this approach in patients with PDAC (NCT04145297; NCT03825289; NCT04132505)^127,128,129. Given the preclinical results of combining autophagy inhibition with immune checkpoint inhibitors⁸⁶, clinical trials are also in development to assess these combinations.

Targeting glutamine metabolism in pancreatic cancers

The reliance of PDAC on glutamine and the availability of clinical-grade inhibitors to glutaminase, a key enzyme that converts glutamine to glutamate, provided a strong rationale to study this approach. Although short-term in vitro assays showed suppression of growth, efficacy was lost with prolonged treatment through compensatory mechanisms¹³⁰. Moreover, the translation of these findings to in vivo PDAC mouse models (including autochthonous and allograft models) failed to reduce tumour growth. Proteomic and metabolomic analyses showed that PDAC cells globally rewire their metabolism in response to this metabolic perturbation¹³⁰. By studying these compensatory metabolic shifts, co-targeting approaches were determined. Indeed, combining the glutaminase inhibitor with buthionine sulfoximine to induce redox stress, produced a significant decrease in tumour growth in a nude mouse model. This study demonstrated the rapid rewiring that PDAC cells undergo upon metabolic perturbation and that understanding these adaptations can lead to effective therapeutic combinations¹³⁰.

Previously, there have been attempts to target glutamine utilization in human cancers by using the glutamine antagonist, 6-diazo-5-oxo-l-norleucine (DON), which broadly inhibits all of the metabolic reactions that require glutamine¹³¹. However, these clinical studies failed, in a large part, due to gastrointestinal toxicities and poor pharmacological properties of the drug¹³¹. A study by Sharma et al. demonstrated that DON treatment decreased tumor growth of pancreatic cancer cells in an athymic nude mouse model¹³². Interestingly, DON treatment enhanced the response to anti-PD1 treatment in a syngeneic PDAC mouse model. Another study from Leone et al. developed a compound called JHU083, a prodrug version of DON, with better pharmacological properties and less toxicity than DON¹³³. In this study, there were robust effects seen in multiple syngeneic tumour models (colon cancer, lymphoma, and melanoma), demonstrating that glutamine blockade resulted in tumour growth decreases and a better immune response when combined with immune checkpoint blockade^133,134. The role of this new drug has not been tested in PDAC, however it might provide an opportunity to overcome the rapid metabolic compensatory response observed upon glutaminase inhibition, given its broad mechanism of action. Conversely, concerns remain regarding the potential toxicities of inhibiting all glutamine-dependent reactions. Further studies are required to validate this approach.

Targeting mitochondrial metabolism in PDAC

Mitochondria are critical organelles that have key roles in a myriad of metabolic processes¹³⁵. In PDAC, it has been demonstrated that upon the ablation of KRAS using the iKRAS model (described earlier) a small population of pancreatic cancer cells survive²⁵. Through a series of metabolic and transcriptomic analyses, it was shown that cells that survive genetic Kras extinction were less dependent on glycolysis, but are highly sensitive to inhibition of oxidative phosphorylation (OXPHOS)¹³⁶. However, it will be critical to carefully assess any potential toxicity with this approach given the importance of OXPHOS in normal cells⁵⁷.

Other aspects of the mitochondria biology have been explored in pancreatic cancers. For example, Yu and colleagues demonstrated that genetic or pharmacological ablation of dynamin-related protein-1 by the small-molecule Mdivi-1 or overexpression of mitofusin-2 stabilized mitochondrial fusion, and increased survival in a PDAC autochthonous mouse model¹³⁸. Additionally, MacVicar, Ohba and colleagues demonstrated that YME1L, a mitochondrial inner membrane protease, is responsible for rewiring the mitochondrial proteome under hypoxia, a well-documented feature of pancreatic cancers¹³⁹. Genetic inhibition of YME1L reduced tumour progression in a PDAC xenograft model, suggesting YME1L as a potential target. Collectively, these findings suggest that mitochondria function and metabolism might provide therapeutic targets in PDAC. Indeed, the concept of targeting different aspects of mitochondrial biology has been studied in the clinic¹³⁴. For example, devimistat (CPI-613), a lipoate analogue that inhibits the activity of pyruvate dehydrogenase and α-ketoglutarate dehydrogenase has shown some anti-tumour activity in the clinic in early phase trials in pancreatic cancer and acute myeloid leukemia (AML)^{140,141,142,143}. Early findings have demonstrated that CPI-613 treatment in combination with a standard chemotherapy regimen for PDAC (FOLFIRINOX), demonstrated a good therapeutic index, and might have activity in this disease¹⁴⁴. Currently, CPI-613 is in a phase III trial for the clinical assessment of metastatic pancreatic cancers (NCT03504423)¹⁴⁴ and acute myeloid leukemia (NCT03504410).¹⁴⁵

The role of cysteine in pancreatic cancers

Cysteine is an essential amino acid for pancreatic cancer progression¹⁴⁶. The uptake of cystine (the oxidized dimer of cysteine) occurs through the system xCT (Slc7a11), which mediates the exchange of intracellular glutamate for extracellular cystine, serving as a precursor for glutathione synthesis. Boutaina and colleagues deleted SLC7A11 in a panel of human-derived PDAC lines¹⁴⁷. They demonstrated that SLC7A11 was important to retain the intracellular levels of glutathione¹⁴⁷. Depletion of cysteine induces ferroptosis, a form of cell death resulting from excessive accumulation of lipid ROS in human PDAC lines in vitro^147,148. Further investigation using the KPC model and a dual recombinase system showed that systemically targeting Slc7a11 increased ferroptosis in the tumours and extended survival of mice compared to control¹⁴⁹. Treatment with cyst(e)inase¹⁵⁰, an enzyme that depletes cysteine, also reduced the tumour progression and induced ferroptosis in vivo¹⁴⁹. These studies provide a strong rationale for the use of specific inhibitors of the SLC7A11 system in PDAC.

Functional genomic screens as a tool to uncover metabolic dependencies in PDAC.

Studying the in vivo metabolism of PDAC have been challenging due to the complex TME and the heterogenous cellular populations within it. The application of isotope tracing experiments in multiple cancers (e.g., non-small-cell lung cancers) have been informative to delineate differences in fuel source utilization between in vivo versus in vitro systems^151,152,153. However, despite the information obtained from these studies, the metabolic pathways that are required for the progression of PDAC in vivo or what metabolic pathways are required in different hosts (e.g. immune competent or immune compromised animals) have not been fully elucidated. To address this gap, two groups, independently performed the first in vivo metabolism-focused CRISPR screens in PDAC to determine the pathways that are essential for PDAC tumor growth^154,155. Both studies demonstrated a striking similarity between in vivo and in vitro metabolic dependencies, indicating that standard 2D culture is a reliable system to study tumor metabolism. Interestingly, the heme biosynthesis pathway was identified as a major metabolic dependency in vivo^154,155. Biancur and colleagues demonstrated that although pooled in vitro screens are an excellent method for globally studying metabolic dependencies, they have certain limitations such as dependencies being masked by intercellular nutrient sharing. Additionally, they demonstrated the benefits of using other in vitro systems such as 3-dimensional (3D) growth in the case of metabolic-cell signaling cross-talk¹⁵⁴. Squalene synthase (FDFT1) was a significant metabolic dependency in vivo and in 3D culture through its regulation of PI3K/AKT signaling¹⁵⁴. There was a more potent dependency in syngeneic models, suggesting a potential role of the immune system. Indeed, genetic or pharmacological inhibition of Fdft1 decreased tumor growth and increased infiltration of CD8+ T cells, suggesting Fdft1 as a potential target in PDAC¹⁵⁵. In a similar fashion, Zhu and colleagues showed the essential autophagy gene Atg7 was an important mediator of immune evasion in PDAC¹⁵⁵. Together, both studies provide an important resource for the field. We anticipate that these types of approaches will be more widely employed in the future and will allow characterization of metabolic dependencies in distinct genetic drivers or in other relevant situations, such as metastasis.

Conclusions

Pancreatic cancer is considered one of the most lethal malignancies. Despite the advances achieved during the past decades in terms of genetics and tumour biology, many gaps in the pathobiology of this disease remain. The field of tumour metabolism has provided key insights into the understanding of the mechanism by which pancreatic cancers survive and thrive in the TME, and how they become resistant to conventional therapy. It is important to highlight that other metabolites such as fatty acids¹⁵⁷, cholesterol^158,159 and other amino acids such as proline¹⁶⁰ have been identified to support PDAC growth and progression. Future studies will certainly elucidate additional fuel sources for PDACs.

A current challenge in the metabolism field broadly is the lack of proper tools to deconvolute the metabolic contributions of the multiple cellular populations within the TME. Although some studies have implemented the use of stable isotope labeling in the in vivo settings^151,152,153, these cannot decipher how tumour cells interact with other cellular populations. Specifically, in PDAC the tumour bulk is often comprised of non-PDAC cells (fibroblast, immune cells, neurons and others)^13,161. Thus, understanding the metabolic contribution of each cell population in the complex PDAC ecosystem will provide important insights. A potential approach to overcome this limitation is to investigate the incorporation of labeled fuel sources into biomass or proteins, which are more stable molecules than metabolites¹⁶². Similarly, future studies will need to accurately assess the metabolic compartmentalization that occurs in different cellular organelles such as the mitochondria¹⁵⁸ and the lysosome. Traditionally, methodologies to study metabolism focus on measuring changes in whole-cell metabolite pools, which fail to consider the compartmentalization of mitochondrial metabolic processes and metabolites. Rapid immunoprecipitation tools for metabolic profiling of both mitochondria^163,164,165 and lysosomes,^166,167 have been developed. These new technologies could elucidate key findings in tumour metabolism, not only in PDAC but in other types of cancers as well as in normal physiology. We anticipate that the use of in vivo functional genetic screens, improved in vitro culture systems (e.g.,organoids, co-cultures)¹⁶⁸, and metabolomic imaging will provide a better understanding of the metabolic plasticity of tumours temporally and spatially. Such studies hold promise to improve therapeutic and diagnostic approaches.

Key point:

Pancreatic ductal adenocarcinoma is characterized by hypoxia, low nutrients, high interstitial pressure and desmoplasia.

To survive and thrive in this hostile environment, PDAC cells reprogramme their metabolism.

PDAC cells also utilize lysosomal scavenging pathways (e.g., autophagy and macropinocytosis) to support and maintain metabolic homeostasis.

Other cellular populations present at the tumor microenvironment of PDAC such as cancer-associated fibroblasts, neurons and immune cells can support PDAC growth in nutrient-limiting conditions.

Targeting the metabolic vulnerabilities of PDAC could provide new therapeutic interventions.