Pancreatic intraepithelial neoplasms (PanINs) are the most common precursors of PDAC [32,33]. KRAS gene alterations occur in 91% of PDACs, followed by TP53 (61%), CDKN2A (44%) and SMAD4 (40%) [34]. Other genes which may be mutated in PDACs, although their mutation frequency occurs in only a small fraction (2-17%), include GATA6,ARID1A,RNF43,ATM,TGFBR2,MAP2K4,MLL3,PIK3CA,RBM10,ROBO2,SMARCA4,PBRM1,SLIT2,KDM6A,BRAF,BRCA2, among others [35]. However, the role of these and other tumor promoting genes involved in the pathogenesis of PC remains to be elucidated. For instance, little is known about genomic and proteomic changes affecting myelocytomatosis (MYC) in PC. However, deregulation of c-MYC is common in PC [36]. Recent studies support the possibility that inactivation of MYC may be an effective therapeutic strategy for KRAS mutant tumors [37].

Currently, it is unknown why PDAC is associated exclusively with KRAS mutations. Genetic mutations indicate that PDAC cells are selected based on their competitive advantages when they encounter limitations in their hypovascular, fibrotic, hypoxic and nutrient deprived TME. The fibrotic layer around the tumor, which accounts for 90% of the tumor volume, creates a barrier to the supply and systemic penetration of drugs (poor drug delivery) and affects the vascularization of pancreatic tumor tissue. Thus, PDAC cells have adapted to survive and proliferate in a harsh TME being under attack by immune system cells, deprivation of nutrients and oxygen. With limited access to blood vessels, PDAC cells must rely on their ability to reprogram metabolic pathways to survive and proliferate [38,39]. Although reprogrammed metabolism is a common feature of neoplasms, metabolic addictions vary among cancers and are determined mainly by their specific genetic mutations, tissue of origin or the TME [40]. PDAC cells show complex and heterogeneous reprogramming of glucose, amino acid and lipid metabolism. These features play an important role in disease evolution by inducing resistance to therapy [41]. In addition to changes in metabolism, PDAC cell survival and progression relies on enhancing nutrient acquisition through macropinocytosis and autophagy [42], and conducting metabolic crosstalk with other components within the TME [43].

As aforementioned, the oncogenic activation of KRAS occurs in the majority of PDAC cells. In fact, KRAS mutation is the initiating genetic event for PDAC [44]. The activation of KRAS originates most commonly from the mutation at the Gly12 residue, which prevents the interaction with GTPase activating proteins (GAPs) and keep KRAS constitutively bound to GTP, i.e. in its active form [45]. The aberrant downstream signaling pathways produce increased tumor cell proliferation, decreased apoptosis, and an invasive phenotype [46]. KRAS-GTP binds preferentially to at least 11 different downstream effector families with distinct catalytic functions. Therefore, it is not trivial to determine which effector pathways are the best to target [47]. Currently, there are no approved drugs that directly target mutated KRAS proteins. There are drugs that target KRAS indirectly by blocking proteins that interact with it, but they were ineffective against PDAC in clinical studies.

One of the relevant metabolic changes occurs in the glutamine (Gln) pathway [48]. Gln is the most abundant amino acid in the plasma and can be synthesized endogenously but becomes essential in physiological or pathological conditions of high cell proliferation such as cancer. Gln is the most highly metabolized amino acid in PDAC tumors. Oncogenic KRAS has reprogrammed Gln uptake and metabolism to serve anabolic processes. Gln enters cells through the amino acid transporters SLC6A14 (ATB^0,+), SLC6A19 (B⁰AT1) and SLC1A5 (ASCT2; AlaSerCys Transporter 2). From the three transporters, elevated expression of SLC6A14 [49] or SLC1A5 [50] were observed in cancers from diverse origins, including PDAC, and is correlated with lower patient survival. Thus, both amino acid transporters play an important role in tumor cell growth, and represent promising pathological prognosis biomarkers for PDAC outcome. SLC6A14 can transport 18 of the 20 proteinogenic amino acids excluding the acidic amino acids glutamate and aspartate. In contrast, SLC1A5 exhibits functional asymmetry with an antiport mode of transport; some amino acids are transported only inwardly, whereas others are bidirectionally transported, allowing for regulation of amino acid balance in cells. In Gln “addicted” cells, SLC1A5 exchanges mainly Gln (the preferred natural substrate) with the release of Ser. Both transporters use Na⁺ transmembrane gradients as the energy source to drive amino acid co-transport (Fig. 1).

Figure 1.

Glutamine (Gln) metabolism and redox homeostasis in PDAC cells. In PDAC, Gln enters the cell via the amino acid transporter SL6A14 and is converted by GLS to Glu. KRAS inhibits GLUD1 expression and glutamate becomes substrate of GOT2 leading to GSH production.

When comparing the amino acid transporters, SLC6A14 is the only carrier that possesses all essential characteristics to promote tumor growth. The most relevant features include broad substrate selectivity (admits all essential amino acids as well as Gln), high concentrative capability due to coupling to three different energy sources (Na⁺ gradient, Cl^- gradient, and membrane potential), as well as coupling to mTOR signaling [51]. SLC6A14 is the only amino acid transporter that generates an amino acid intracellular concentration gradient of more than 1,000-fold when compared to the extracellular milieu, making the transport practically unidirectional and directed towards the cytoplasm. More importantly, this Gln pathway is not used, extensively, by healthy cells [52].

Normal pancreas cells express SLC6A14 at much lower levels and their proliferation is not affected by blocking the transporter with α-methyl-tryptophan (α-MT). In contrast, in PDAC, SLC6A14 is clearly overexpressed (13- to 167-fold) and its pharmacological blockade with α-MT reduces the growth and proliferation of PDAC cells, in primary cultures and in xenografts of PC [52]. In SLC6A14-positive tumor cells, the inhibition of amino acid uptake induces cell death via four different mechanisms: (a) it halts the uptake of essential amino acids; (b) it targets the Gln addiction of PDAC cells; (c) it inhibits mTOR; and (d) it induces oxidative stress [51].

SLC1A5 plays a supportive role from the initial stages of PDAC formation. Tumor initiating cells have a mechanism for maximizing Gln uptake, which relies on CD9, a member of the tetraspanin family of proteins. PC cells show increased CD9 expression when compared to normal pancreatic tissues. High CD9 expression can initiate and sustain PDAC growth and correlates with poorer patient survival. CD9 augments Gln uptake by increasing the cell surface expression of SLC1A5, thereby enhancing PDAC growth and proliferation [53].

Once inside the cell, Gln is transported through the inner mitochondrial membrane before glutaminolysis can take place. However, data concerning the structure and function of the transport system are scarce [54]. Notably, a variant of SLC1A5 (SLC1A5*) induces metabolic reprogramming, ATP generation, glutathione synthesis, and gemcitabine resistance in PC cells [55]. Increased SLC1A5* expression was noted in PDAC and Kaplan-Meier survival analysis indicated a correlation with poor survival outcomes. SLC1A5* is an exclusive mitochondrial Gln transporter, while SLC1A5 localizes to the plasma membrane. Moreover, SLC1A5* is essential for PC growth.

Gln is transformed into glutamate (Glu) by glutaminase [56]. Humans have two glutaminase genes, GLS and GLS2. GLS has 3 isoforms, of which, isoform 3 is highly expressed in heart and pancreas [57]. GLS2 is highly expressed in liver and is moderately expressed in brain and pancreas. Whereas, GLS2 expression is significantly reduced in hepatocellular carcinomas [58]. All these glutaminases are known to be localized in mitochondria. Expression levels and enzymatic activity of GLS and GLS2 in different types of tumors are altered [59]. Therefore, the existing GLS or GLS2 could both be targeted in order to block tumor cell growth, taking into consideration that GLS is considerably overexpressed in PDAC cells and GLS2 is preferentially expressed in hypoxic PDAC cells [60].

In healthy cells, Glu enters the cycle of tricarboxylic acids (TCA), but oncogenic activation of KRAS in PDAC cells repurpose Glu through a distinct pathway in which mitochondrial glutamic-oxaloacetic transaminase 2 (GOT2) transforms Glu into aspartate (Asp), which is transported into the cytoplasm. Afterwards, successive reactions catalyzed by glutamic-oxaloacetic transaminase 1 (GOT1), malate dehydrogenase 1 (MDH1), and malic enzyme 1 (ME1), convert Asp to pyruvate and produce NADPH, maintaining the redox balance and ensuring cell proliferation (Fig. 1). Thus, KRAS holds an important role in Gln metabolic reprogramming in PDAC through the transcriptional upregulation of GOT1 and the inhibition of glutamate dehydrogenase 1 (GLUD1) expression [61]. Relative to other cancer types, GLUD1 is not upregulated in PDAC [62]. Cancer cells depend on GLUD1 for the conversion of glutamate into α-ketoglutarate. However, PDAC cells rely on GOT1 and GOT2 to transform aspartate into pyruvate, which supports PDAC cell growth by maintaining the redox balance. Therefore, there must be a great deal of complex crosstalk among the metabolic processes of different energy sources that cooperatively regulate the malignant behavior of PDAC. The relevant druggable enzymes of this non-canonical Gln pathway, reprogrammed by the oncogenic KRAS, are GOT1, GOT2, MDH1 and ME1 [41, 63].

GOT is a pyridoxal phosphate-dependent enzyme which exists in both cytoplasmic and inner-membrane mitochondrial forms, namely GOT1 and GOT2, respectively. In healthy cells, GOT plays a role in amino acid metabolism and the urea and tricarboxylic acid cycles [64]. In PDAC cells, GOT1 [65] and GOT2 [66] were found to be overexpressed. GOT1 is critical for connecting the mitochondria and cytosolic compartments in Gln anaplerosis, and hence, allows this metabolic process to complete [61]. The status of GOT1 in tumor tissue serves as an independent prognostic biomarker in PDAC [65]. Reduced NAD-dependent protein deacetylase sirtuin-3 (SIRT3) expression leads to an increase in GOT2 acetylation in PDAC cells. GOT2 acetylation at three lysine residues (K159, K185, and K404) augments the protein interaction between GOT2 and malate dehydrogenase 2 (MDH2), thereby stimulating the malate-aspartate shuttle stimulating and net transfer of cytosolic NADH into mitochondria to support ATP production [67].

Cytosolic MDH1 and mitochondrial MDH2 enzymes are overexpressed in PDAC patients. However, only high expression of MDH1 is associated with poor prognosis of the disease [66]. MDH1 is a cytoplasmic enzyme that exists as a mixture of monomers and dimers, where the homodimeric state is the catalytically active form. PDAC cells require MDH1 to maintain their cellular redox state by reprogramming Gln metabolism, and MDH1 knockdown inhibits the viability of PDAC cells. Arginine 248 (R248) methylation of MDH1 by protein arginine methyltransferase 4 (PRMT4/CARM1) inhibited the enzyme through disrupting its homodimerization [68]. In clinical PDAC samples, MDH1 is overexpressed and hypomethylated. KRAS suppresses MDH1 methylation, contributing to Gln metabolism in PC [66].

In human cells, three isoforms of malic enzyme (ME) are known. ME1 localizes in the cytoplasm and it is important for NADPH production as well as keeping the redox balance in PDAC cells. In PC3 cells, ME1 depletion induced cellular senescence and suppressed tumor cell growth [69]. Thus, ME1 and GOT1 represent potential prognostic or sensitivity markers of radiotherapy [70]. Malic enzyme 2 (ME2) and malic enzyme 3 (ME3) are two redundant enzymes that reside in the mitochondria, where they help keep reactive oxygen species (ROS) levels under control. In PDAC, the homozygous deletion of SMAD4 is often accompanied with homozygous deletion of ME2 as well. A compensatory increase in ME3 expression in ME2-null cell lines occurs [71]. In the absence of ME2, ME3 maintains indispensable NADPH synthesis in mitochondria. ME3 expression was higher in PC tissues of patients that had significantly shorter survival [72].

In contrast to other tumor types, PDAC cells do not rely extensively on glucose metabolism for energy demand and macromolecular biosynthesis. However, glycolysis is significantly higher than in normal cells. Furthermore, a high glycolysis phenotype in PDAC correlates with cancer metastasis [73]. KRAS reprogramming enhances glucose uptake and upregulates the primary glucose transporter SLC2A1 (also known as GLUT1), which correlated with worse prognosis for PDAC [74]. Furthermore, SLC2A1 is indispensable for the preservation of PC stem cells [75]. Several rate-limiting glycolytic enzymes are also overexpressed in PDAC, including hexokinase 1 (HK1), HK2, phosphofructokinase 1 (PFK1) and lactate dehydrogenase A (LDHA). Additionally, the overexpression of NAD-dependent protein deacetylase sirtuin-2 (SIRT2) keeps LDHA deacetylated at K5 and retaining its enzymatic activity [76]. Clinical studies revealed that patients with a strong pyruvate kinase M2 (PKM2) and LDHA expression had significantly worse survival [77]. The dependence on glycolysis presents additional demands on mobilization and excretion of lactate to avert its intracellular accumulation and decreased cytosolic pH [78]. In PDAC cells, the transporter proteins SLC16A1 (also known as MCT1) and SLC16A3 (also known as MCT4) are overexpressed, with SLC16A3 playing a predominant role in this detoxification process and in the progression to metastasis [79].

In PDAC treatment, patients are treated with chemotherapeutic agents irrespective of tumor subtypes. However, diverse studies have shown the relationship between biomarkers and PDAC prognosis (Table 2). These proteins have great functional and prognostic importance for PDAC patients. The establishment of predictive biomarkers is essential for therapeutic decision-making and for treatment with targeted therapies. Routine cancer markers (like carbohydrate antigen 19-9 known as CA19-9) do not seem to be reliable in prediction and detection of early stage PDAC [80]. In practice, PDAC biomarkers are not established for diagnosis purposes. However, there is hope that emerging biomarkers may significantly have increased specificity and sensitivity in early PDAC detection. Liquid biopsy [81], proteomics [82], metabolomics [83], genomics [84], and miRNAs [85] appear most promising and might provide valuable biomarkers to improve selection of patients for optimal treatment regimens.

The evaluation of tryptophan derivatives led to the identification of α-MT, which is not a transportable substrate, but is a weak inhibitor that blocks the transport function of SLC6A14 [87]. At present, no additional SLC6A14 inhibitors (iSLC6A14) are known. Recently, α- and γ-glutamyl tryptophan dipeptides [88], and naphthol-derived Betti bases [89] were proposed as iSLC6A14. Further studies are necessary in order to confirm these compounds as inhibitors of SLC6A14-mediated transport.

In addition to a plausible therapeutic target, SLC6A14 represents a strong candidate for the selective delivery of amino acid-based prodrugs to tumors [90-92].

In contrast to SLC6A14, diverse small molecule compounds were discovered as pharmacological inhibitors of SLC1A5 (iSLSC1A5). Initial efforts consisted of the development of compounds derived from amino acids, the preferred natural substrates of the transporter. Thus, l-γ-glutamyl-p-nitroanilide (L-GPNA) was reported as one of the first synthetic iSLC1A5. Unfortunately, this compound showed very weak potency toward the transporter [93]. Subsequent research focused on obtaining new iSLC1A5 using amino acids (either l or d) as the main backbone, such as 2-substituted glutamylanilides (CHEMBL3576929) [94], phenylglycine derivatives (L-3,4diFPG and L-3OH,4FPG) [95], serine esters (CHEMBL3576945) [96],O-benzyl-serine (BnSer) and S-benzyl-cysteine (BnCys) [97], γ-(2-flurobenzyl)-proline (γ-FBP) [98], 4-aryl-prolines (CHEMBL4116473) [99], sulfonamides based on the 3-amino-alanine scaffold (12b) and sulfonic acid esters based on hydroxyproline (16b) [100], and 2,4-diaminobutanoic acid derivatives (CHEMBL3754498 and V-9302) [101,102]. The most potent iSLC1A5 among all of these amino acid derivatives were CHEMBL3754498 and V-9302, which showed IC₅₀ values in the low micromolar range (Table 3). All of these compounds blocked, with different potencies, SLC1A5-mediated amino acid uptake in live cells. In particular, V-9302 reduced cancer cell growth and proliferation, augmented cell death, and increased oxidative stress, both in vitro and in vivo. However, the study also showed that V-9302 efficacy is unrelated to SLC1A5 inhibition [103].

The rat but not the human SLC1A5 isoform contains two cysteine residues (Cys207 and Cys210) that form a CXXC metal binding motif. Mercurial compounds react with this site and inactivate the protein. This result was the rationale to design and synthesize iSLC1A5 containing a 1,2,3-dithiazole ring as the common core group. It was anticipated that the dithiazole group could covalently interact with the thiol group of C207/C210. The results provided the first inhibitors lacking an amino acid, which displayed potent activity at the low micromolar range (CHEMBL3753379) [104].

One of the earliest GLS inhibitors (iGLS) is 6-diazo-5-oxo-l-norleucine (DON), who failed in clinical trials due to its low therapeutic index and no substantial activity in cancer patients [105]. However, the preclinical results of DON led to the search for new inhibitors. Allosteric iGLS 968 was shown to inhibit the growth of cancer cells, highlighting the potential of this enzyme as a druggable target [106]. Bis-2[5-phenylacetamido-1,2,4-thiadiazol-2-yl]ethylsulfide (BPTES) is another allosteric, time-dependent, and specific iGLS that also blocks tumor growth. Despite its remarkable selectivity, BPTES has poor solubility, which has limited its clinical development [107]. A recent study showed that nanoparticle encapsulation of BPTES (BPTES-NP) with dense PEG surface coatings provides an effective modality to deliver the inhibitor to pancreatic tumors while minimizing untoward toxicity [108]. However, hypoxic PDAC cells, which preferentially express GLS2, survived BPTES-NP monotherapy. The inability to target hypoxic PDAC cells with BPTES-NPs was overcome by treating the tumors with metformin. The promising results enhanced the search for new iGLS, most of which were based on the structure of BPTES [109-111]. As a result, the BPTES derivative telaglenastat (CB-839), a potent, selective, and orally bioavailable iGLS, has advanced to clinical trials [112]. Currently, 20 phase I/II clinical trials include CB-839 alone or in combination with other drugs [107]. However, none of these clinical trials includes PC patients. CHEMBL4080388 is a thiazolidine-2,4-dione that was optimized after a preliminary, high throughput screening against GLS of a library of 40,000 small molecule compounds [113]. In a virtual screen for more iGLS-like compounds conducted in vivo on 1,280 active drugs, ebselen, chelerythrine and (R)-apomorphine exhibited 10- to 1500-fold greater affinities than DON and BPTES [114]. Ebselen behaves as a mixed non-competitive inhibitor, while chelerythrine and (R)-apomorphine are competitive inhibitors.

Overall, the exact disease context where GLS inhibition will be most effective remains an area of active investigation. This opens a debate emphasizing the importance of defining the patient subpopulation likely to benefit from GLS inhibition. However, due to the widespread expression of GLS throughout the body, long-term and high-dose administration of an iGLS would not circumvent its likely toxicity. The results of the completed clinical trials conducted with CB-839 clearly suggest that combination therapy is a plausible option for developing future iGLS [107].

While diverse studies report on inhibitors of GOT1 (iGOT1), the corresponding studies on GOT2 inhibition were limited to knockdown of GOT2 [66]. The iGOT1 aminooxyacetic acid (AOA), while cytotoxic to triple negative breast cancer cells, showed acceptable toxicity profiles in small clinical trials of patients with tinnitus and Huntington’s disease [115]. The prodrug approach demonstrated an effective strategy to improve the anti-proliferative potency of AOA in vitro and in vivo while reducing its untoward toxicity in vivo [116]. iGOT1-01 was discovered as iGOT1 during the screening of a large library of 800,000 small molecules [117]. Medicinal chemistry-based optimization of iGOT1-01 caused the identification of several analogs with an improvement in potency of at least 10-fold (e.g. CHEMBL4238792), along with the discovery of a tryptamine-based series (e.g. CHEMBL4239817) of iGOT1 [118]. PF-04859989, a known kynurenine aminotransferase II (KAT II) inhibitor, was developed to be applied in the treatment of several psychiatric and neurological disorders. Notably, it inhibited GOT1 in a time- and pyridoxal-5′-phosphate-dependent manner and showed selective growth inhibition of PDAC cell lines [119]. In the same study, PF-04859989 displayed lower inhibitory activity against GOT2. Aspulvinone O was identified from an in-house natural compound library as a new iGOT1 that significantly reduced proliferation of PDAC in vitro and in vivo [120].

The role of MDH in cancer metabolism is not relevant at present. Inhibitors of MDH (iMDH) are scarce, although there is evidence of cancer-associated functions for MDH1 and MDH2. These findings motivated the search for iMDH. Thus, hypoxia-inducible factor 1 (HIF-1) inhibitor LW6 is also a dual iMDH1/2 [121]. In PC cells, LW6 inhibited migration, proliferation and cell viability. These effects were enhanced synergistically when cells were treated with LW6 in combination with metformin [122]. LW6 served as basis for structure-activity relationship studies on a series of (aryloxyacetylamino)benzoic acids that led to the identification of novel iMDH [123]. In that study, the lead compound (CHEMBL4068781) competitively inhibited MDH1 and MDH2, and demonstrated significant in vivo antitumor efficacy in xenograft models using HCT116 cells. Affinity investigations revealed that paullones bind and inhibit MDH from various tissues. Subsequent studies showed that alsterpaullone, gwennpaullone and kenpaullone inhibited MDH1 and MDH2 in the low micromolar concentration range [124,125]. Moreover, alsterpaullone induced apoptosis and inhibited proliferation via the p38MAPK signaling pathway [126]. The virtual screening of the compound library of Ambinter revealed 16 candidate molecules for further in vitro testing against MDH2. From this set, only 5 compounds were identified as iMDH2, with IC₅₀ values in the range of 3.9-18.2 µM [127].

Among the three MEs, ME1 and ME2 were predominantly studied. However, the number of small molecules reported as inhibitors of ME (iME) is limited. A fragment-based virtual library design and virtual screening allowed synthesizing several compounds that were tested against ME1. The derivatives from this library combining the piperazine and 2,5-dioxopyrrolidine fragments have shown sub-micromolar inhibitory activity against ME1 (e.g. CHEMBL372408) [128]. The natural compound embonic acid (EA) inhibited ME2 and induced anti-proliferative effects in the non-small cell lung cancer H1299 cells [129]. A set of 12,683 natural products from the Chinese National Compound Library were tested against ME2 revealing 15 ME2 inhibitors with different structures [130]. From this set, compound NPD387 was the most potent inhibitor. Through structural modification, an even more potent iME2 was generated, NPD389, which is a fast-binding iME2 and acts as an uncompetitive inhibitor. The study of the effects of fumarate analogs on ME1 and ME2 led to the conclusion that diethyl oxaloacetate behaves as a weak allosteric iME2 [131]. The work paves the way to rational design of allosteric iME2.

PROteolysis TArgeting Chimeras (PROTACs) are small molecules that selectively degrade target proteins by exploiting the intracellular ubiquitin-proteasome system (UPS) [132]. PROTACs have three connected chemical components: a ligand binding to a target protein, a ligand binding to E3 ubiquitin ligase, and a linker bridging these two ligands. Once the PROTAC-mediated target protein-E3 complex is formed, an E2 ubiquitin-conjugating enzyme transfers ubiquitin to lysine residues on the surface of the target protein. The recognition of polyubiquitination signal by UPS facilitates the degradation of the target protein [133]. In contrast to the stoichiometric occupancy-driven process of traditional inhibitors, PROTACs induce target protein degradation, in multiple rounds, at sub-stoichiometric levels. PROTACs allow degradation of previously “undruggable” proteins [133]. Even target proteins with low affinities with PROTACs can be effectively degraded if PROTACs can induce extensive protein-protein interactions between target proteins and E3 ligases. PROTACs that trigger KRAS degradation would effectively shut down the alternative Gln pathway overexpressed in PDAC cells.

In the WO2019/19560A2 patent highlight report, there are six examples of PROTAC molecules that target KRAS [134]. In cells treated with 1 µM of the compound, two of these molecules triggered degradation of more than 50% of KRAS. These PROTACs were found to recruit either VHL or CRBN E3 ligases.

Recently, over 100 PROTACs targeting oncogenic KRAS^G12C were described [135]. The lead PROTAC successfully recruited the E3 ligase CRBN in cells, bound to KRAS^G12C in vitro, promoted CRBN/KRAS^G12C complex formation, and degraded GFP-KRAS^G12C in reporter cells in a CRBN-dependent manner. However, it failed to degrade endogenous KRAS^G12C in pancreatic and lung cancer cells. Although unsuccessful, this effort indicates the shortcomings that must be surpassed to achieve KRAS degradation in cancer cells.