Chemical modulation of glycerolipid signaling and metabolic pathways

2.1.1. Enzyme activity and regulation

Phospholipase C (PLC) enzymes cleave phospholipids and produce diacylglycerol and the corresponding phospho- head group. Substrate specificity for either phosphatidylinositol-PLC (PI-PLC) or phosphatidylcholine-PLC (PC-PLC) defines the two main classes of PLCs. The cytosolic PI-PLC is the most well characterized class of PLC and localizes to the plasma membrane upon activation where it catalyzes the conversion of the minor membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P₂ or PIP₂) into the lipid second messengers inositol 1,4,5-triphosphate (IP₃) and diacylglycerol (DAG) (Fig. 1A). Both IP₃ and DAG serve as signaling molecules for Ca²⁺ mobilization or protein kinase activation, respectively. These two signaling molecules are exceptionally versatile and control distinct signaling pathways, making them responsible for dozens of cellular processes [1, 2]. Cells tightly regulate PIP₂ depletion due to its role in protein activation at the membrane. One important example is a minor form of phosphorylated PIP₂, phosphatidylinositol 3,4,5-trisphosphate (PIP₃), which controls crucial signaling cascades via the phosphoinositide 3 kinase (PI3K) pathway [3].

Mammals have 13 different PI-PLC enzymes subdivided into six different enzyme families: β,γ,δ,ε,ζ and η; each family is characterized by its unique mechanism of regulation and localization. Most major signaling events sit upstream of distinct PI-PLC isozymes. Growth factors, antigens and other extracellular stimuli activate PLCγ; extracellular stimuli, hormones, neurotransmitters, and chemosensory molecules activate PLCβ via heterotrimeric G-proteins [4]. Additionally, PLCε is activated downstream of Ras signaling affording this family of enzymes a unique role in cellular communication and signal transduction [5].

PI-PLC enzymes are highly conserved across phyla—bacteria, flies, and mammals all express PI-PLC isozymes. While the overall core structure of the PI-PLCs shows little variance between families, they share very little sequence homology. All family members contain pleckstrin homology domains (PH) (except PLCζ), EF hand motifs, X and Y domains, and a C2 domain [6]. Each of these core domains have important regulatory and catalytic functions for PLC [6]. PH domains mediate membrane recruitment and facilitate binding to both PI and PIP₂. EF hand motifs bind Ca²⁺ ions, required for enzyme activity. X and Y domains dimerize, forming a triosephosphate isomerase (TIM) barrel, with the catalytic residues on the X portion of the TIM barrel. Finally, C2 domains, also essential for Ca²⁺ activation and anionic lipid binding, are found in repeating units of either 2 or 4 on the PI-PLCs, depending on the isoform. Other PI-PLC isoforms may contain more specialized regions, such as a Ras-GEF in PI-PLCε and PDZ-motifs found in β and η isoforms, believed to scaffold large protein complexes following G-protein coupled receptor (GPCR) activation [6].

Each isozyme class has unique signaling roles and tissue distribution. The β isoforms rely on Ca²⁺ release downstream of GPCR signaling for activation. Certain β isozymes actually serve as GTPase-activating-proteins, or GAPs, for Gα, which in turn activates other PLC isozymes [7]. PLCβ isoforms often localize to the nucleus but are also found in the cytosol. PLCβ1^−/− animals have ocular and central nervous system (CNS) developmental deficiencies suggesting a critical role for PLCβ in the CNS [6]. Members of the PLCγ family are activated by receptor tyrosine kinases and cellular tyrosine kinases making them critical players in cell proliferation, and their expression in T and B lymphocytes suggests a role in immunity [8]. Ca²⁺ or Gα_i/o and Gα_q activate PLCδs and play roles in Alzheimer’s disease [9], fertilization [10], and balding [11]. Mammalian expression of PLCζ in sperm heads facilitates fertilization [2]. PLCη is expressed in the brain and neurons, resembles PLCδ in structure, sits downstream of several GPCRs and is activated by Ca²⁺ [12].

The largest and most broadly distributed PLC isoform, PLCε, interacts with Ras family member GTPases and functions as a guanine nucleotide exchange factor (GEF) [13]. Due to its close relationship with Ras family proteins, PLCε regulates cell proliferation, migration, signaling and disease processes such as those observed in cardiac hypertrophy [14]. Heterotrimeric G-proteins also activate PLCε following GPCR stimulation. The therapeutic potential for new compounds that modulate PLCs is considerable. For example, a recent report by Smrcka and colleagues [14] on the protective role played by PLCε in cardiac myocytes during stress-induced pathological hypertrophy has reinvigorated interest in the development of isoenzyme-selective modulators of PLCs.

PI-PLC cleaves PIP₂ into the critical second messengers IP₃ and DAG, distinguishing this protein family from other PLC isoforms. The varied tissue distribution, activation, and regulation of this enzyme suggest that it facilitates essential signal transduction cascades required for proper cellular function.

In addition to the PI-PLC enzyme family, a separate, phosphatidylcholine-specific Phospholipase C (PC-PLC), hydrolyzes PC to form phosphocholine and DAG [15]. PC-PLC was identified and cloned in bacteria but not yet identified in mammals. Evidence suggests that PC-PLC may rely on growth factors, cytokines, mitogens, and calcium for activation. These observations correlate with its implication in many cellular processes including proliferation, differentiation, metabolism, senescence, and apoptosis [16]. PC-PLC is likely responsible for elevation of ceramide production via DAG release resulting in macrophage activation [17]. In addition to its myriad of signaling roles, mounting evidence implicates PC-PLC activity in the expression of essential growth factor receptors like HER2 [18].

2.1.2. Tools for modulation of enzymatic activity

The integral role PLC isoforms play in key survival events makes them ideal targets for the development of future therapeutics. To better understand PLC as a therapeutic target, small molecule modulators of PLC catalytic activity were developed for further investigation.

The compound most often used in the literature to manipulate the PI-PLC isozymes is U73122 (Fig. 1B and Table 1). This sterol mimetic inhibits PI-PLC-dependent processes in human platelets and neutrophils albeit with only modest potency (IC₅₀ values for collagen or thrombin induced platelet aggregation were 0.6 μM and 5 μM respectively) [19, 20]. Since its initial publication, more than 1,000 scientific papers utilize U73122 as a PI-PLC inhibitor, even showing efficacy in vivo [21]. In spite of its wide use, no evidence demonstrates U73122 as a direct modulator of enzyme activity. Several reports have detailed non-specific effects of U73122 on targets other than PLC; one study even suggests the compound is capable of stimulating certain PLC isoforms when used at micromolar concentrations [22]. Literature calls into question its isoform specificity and most reports refer to the compound as a pan-PLC inhibitor. However, some suggest it preferentially inhibits the β isoforms [6]. The indirect inhibition of PLC activity by U73122 should be carefully considered during experimental design.

Another less frequently employed PI-PLC inhibitor is the ether lysolipid mimic Edelfosine (ET-18-OCH3) [23]. Edelfosine directly inhibits the ability of PI-PLC to bind its PIP₂ substrate in a mixed micelle. At high concentrations, Edelfosine inhibited other phospholipases (PC-PLC and PLD), likely an artifact of cellular toxicity rather than specific enzyme inhibition. Ether lysolipids, like Edelfosine have well documented cellular toxicity, which should be considered prior to its use in model systems [23].

A recent study by Zheng and colleagues yielded novel small molecule inhibitors of PLC following a high-throughput screen of over 6,000 compounds. Three direct inhibitors of PI-PLC were identified (ATA, 3013, and 3017) which inhibited the three isozymes tested (PLCβ3, γ1 and δ1), suggesting they may be pan-PLC inhibitors [24]. The authors verified the cellular activity of these three novel compounds by following IP₃ accumulation upon receptor activation in intact cells. Importantly, ATA was less active than the other compounds in cellular assays due to poor cell penetration. While these compounds are direct and seem to inhibit multiple PLC isozymes, they only possess modest potency with reported IC₅₀ values around 10 μM [24].

PC-PLC inhibitors are less common in the literature. The xanthate D609 was initially investigated for its antiviral properties [25]. Later studies documented D609 preferentially inhibited PC-PLC over PI-PLC enzymes but also inhibited sphingomyelin synthase (SMS) [16]. Initial reports suggested that this compound works through a competitive mode of inhibition [26]. Both in vitro and in vivo studies demonstrated that D609 inhibits cell cycle progression and arrests proliferation either through PC-PLC or SMS inhibition. These cellular activities may be the result of an overlapping set of one or more enzymes. The exact inhibitory target of D609 remains unknown, making the mechanism unclear [27]. In spite of its nebulous activity, D609 holds promise in cancer, artherosclerosis, and cerebral infarction studies, which may be the result of activity at multiple non-specific targets[16].

2.2.1. Enzyme activity and regulation

Phospholipase D (PLD) is a phosphodiesterase responsible for the hydrolysis of the cellular membrane lipid phosphatidylcholine (PC) into the lipid second messenger phosphatidic acid (PA) and free choline (Fig. 2A). PLD activity was first described in plants in the 1940s [28, 29] but PLD superfamily members have since been found in bacteria, viruses, yeast, worms, and mammals (for review [30, 31]). Members of the PLD superfamily often contain a conserved catalytic HKD motif (HxKxxxxDx₆GSxN) [32] although there are non-HKD enzymes in this family as well. This review focuses on the conventional mammalian enzymes.

Mammalian cells contain two major PLD isozymes, PLD1 and PLD2, each of which exhibit differences in localization, activity, and regulation [33]. PLD sits downstream of G-protein coupled receptors and receptor tyrosine kinases. Thus, cellular PLD activity responds to growth factor or hormone stimulation. The most commonly studied PLD-activating receptors are those within the immune system (e.g., IL-1, IL-8, Fcγ, fMLP and TLRs)[34], and those associated with uncontrolled cell growth (specifically growth factor receptors, EGFR [35] and PDGFR [36]).

Mammalian PLD1 and PLD2 isozymes are splice variants with 50% shared sequence homology. Both isoforms have four conserved domains which include Phox homology (PX) and Pleckstrin homology (PH) domains near the N-terminus and two HKD catalytic domains [30]. The PX domain contains anionic lipid binding sites that are known to bind phosphoinositides (PI(3,4,5)P₃) which anchor PLD to the membrane [37]. This region may also facilitate protein-protein interactions. PH domains bind phosphoinositides and interact with small G-proteins such as Arf and Rho [38]. Some studies demonstrate that small G-proteins activate truncated PLD1 lacking the PH domain, suggesting an alternative recognition site [39]. Phosphoinositides, specifically PIP₂ (PI(4,5)P₂), are essential lipid cofactors required for PLD activity, which also bind the loop region of the enzyme between the two HKD catalytic motifs to further stimulate enzymatic activity [30].

PLD1 and PLD2 localize to distinct cellular compartments and possess different modes of regulation. The PLD1 isoform localizes to intracellular membranes such as the Golgi, early endosomes, or perinuclear vesicles. Upon stimulation, PLD1 translocates to late endosomes and the plasma membrane [40, 41] although this translocation event is cell-, expression-, and stimulation-dependent. Conversely, PLD2 constitutively binds to the plasma membrane [42]. Both PLD1 and PLD2 are palmitoylated and contain Ser/Thr phosphorylation sites which are not required for catalytic activity but facilitate membrane association [33].

In addition to differential subcellular localization, PLD1 and PLD2 have unique modes of regulation. PLD1 activity is robustly stimulated by Arf GTPases [43, 44], which only moderately increase PLD2 activity [43, 45]. Small G-proteins including those of the Rho family activate PLD1 catalytic activity [39, 46, 47]. Protein kinase C (PKC) is also a very well characterized activator of PLD. Phorbol esters, PMA, directly activate PKC and subsequently PLD (reviewed in [48]). PLD2 is less sensitive to protein stimulators and is constitutively active in cells.

The PLD catalytic product, PA, has a wide range of roles in many critical cellular functions including cell survival, proliferation, vesicular trafficking, and cytoskeletal reorganization. Because PLD sits downstream of several important signaling cascades, PA likely regulates the progression of specific disease states. Studies implicate PLD and PA in the progression of diabetes [49, 50], chronic inflammation [51], Alzheimer’s disease [52, 53], and several different types of cancer including gastric [54], breast [55, 56], colorectal [57], brain [58], and renal [59]. Thus, many recent reports suggested that small molecule inhibitors of PLD function are important compounds for therapeutic development [51, 60–62]. In light of the possible role of PLD as a therapeutic target, intense medicinal chemistry efforts focus on the development of modulators or inhibitors for this enzyme.

2.2.2. Tools for modulation of enzymatic activity

2.3.1. Enzyme activity and regulation

Phospholipase A enzymes hydrolyze the ester bonds of phospholipids at either the sn-1 (PLA₁) or sn-2 position (PLA₂) to generate lysophospholipids and free fatty acids. Organisms ranging from fungi to mammals display PLA₁ enzymatic activity, which is a prominent component of venoms. However, the physiological relevance of PLA₁ enzymes remains largely uncharacterized [89]. Currently there are nine PLA₁ isozymes known in mammals, six extracellular and three intracellular, with no shared sequence homology between them [89]. Members of the PLA₁ family exhibit broad substrate specificity, hydrolyzing lipids such as triacylglycerol, galactolipids, and phospholipids. Other PLA₁ proteins have very specific substrate specificity and hydrolyze only phosphatidylserine (PS-PLA₁) or phosphatidic acid (PA-PLA₁). General inhibitors of this lipid enzyme class are not widely represented in the literature due in part to the large variation in family members. The literature is replete with reviews of PLA₁-specific inhibitors [89, 90].

Phospholipase A₂ catalyzes the hydrolysis of phospholipids at the sn-2 position, generating free fatty acids and lysophospholipids. Both of these products are important second messengers for cellular processes. The PLA₂ class is large, containing more than 30 enzymes with PLA₂ activity [91]. Thus, PLA₂ enzymes are subcategorized into four major subclasses based on mechanism, localization, and structure: the secreted sPLA₂, the cytosolic cPLA₂, the Ca²⁺-independent iPLA₂, and the lipoprotein associated Lp-PLA₂ [92].

Secreted PLA₂s were first identified in the literature in the 1890s as a component of cobra venom. All sPLA₂ enzymes share general characteristics including an active site histidine, low molecular weight (13–15 kDa), Ca²⁺-required catalysis, and six conserved disulfide bonds [93]. Mammals express ten classes of sPLA₂ that primarily function as antimicrobial agents, hydrolyzing negatively charged gram negative bacteria. In addition, these enzymes are implicated in arachidonic acid production and eicosanoid biosynthesis during inflammatory responses although this role is still debated [94]. While little is known about sn-2 fatty acyl chain specificity for sPLA₂s, the enzymes prefer anionic phospholipids such as phosphatidylglycerol (PG), phosphatidylserine (PS), and phosphatidylethanolamine (PE) [95]. However, specific members of the sPLA₂ family are more active against the cationic PC. The role of sPLA₂s in the hydrolysis of PC may be relevant in the progression of atherosclerosis, as PC is the main phospholipid found in low density lipoprotein (LDL) particles [96].

Cytosolic PLA₂s were identified in the late 1980s from human platelets and neutrophils [95]. There are six members of the cPLA₂ family (α, β, γ, δ, ε, and ζ) which are structurally larger than the sPLA₂s and do not possess the highly conserved disulfide bond network [93]. These enzymes contain a Ser/Asp catalytic dyad and are mechanistically serine acyltransferases, a large enzymatic family. While cPLA₂s have broad substrate specificity they prefer arachidonic acid (AA) containing lipids. The arachidonate preference of these enzymes makes them key mediators of inflammatory responses [97], since AA is the essential fatty acid in prostaglandin biosynthesis. Cyclooxygenase (COX) enzymes convert AA released by cPLA₂ into the core prostaglandin scaffold and are a key step in the inflammatory pathway [98].

There are six different types of Ca²⁺-independent cytosolic enzymes (iPLA₂s): β, γ, δ, ε, ζ and η. These enzymes, also known as patatin-like phospholipase domain-containing lipases (PNPLA 9,8,6,3,2 &4) [91], rely on a characteristic Ser/Asp dyad to catalyze lipid hydrolysis. Mammalian iPLA₂s are critical to lipid metabolism as they hydrolyze a broad array of lipids including triglycerides, phospholipids, and retinol esters [91].

Lipoprotein-associated PLA₂s (Lp-PLA₂s) centrally regulate lipid metabolism and inflammation. Lp-PLA₂ circulates with lipoprotein particles, localizing onto arterial walls alongside low-density lipoprotein. Lp-PLA₂ catalyzes the release of lysophosphatidylcholine and oxidized free fatty acids, promoting a local inflammatory response. Due to the importance of inflammation in atherosclerosis, Lp-PLA₂ is recognized as a key therapeutic target in the treatment of cardiovascular disease [99].

2.3.2. Tools for modulation of enzymatic activity

Small molecule therapy development programs target PLA₂ enzymes due to their central role in lipid metabolism and involvement in a host of physiological and disease processes. As such, the literature is replete with reviews on inhibitors of the PLA₂ enzymes [95, 100]. Small molecule modulators of PLA₂ range from phospholipid analogues, to natural products and peptide mimetics. Each subclass of PLA₂ enzymes, secreted, cytosolic, Ca²⁺ independent, and lipoprotein-associated, has unique classes of chemical inhibitors. Due to the breadth and structural diversity of this enzymatic class and the sheer number of available inhibitors, we refer interested readers to excellent review articles from Dennis et al. [95] and Magrioti and Kokotos [100]. For the purposes of this review, we only highlight a select number of compounds.

Secreted PLA₂s involvement in inflammatory responses established their therapeutic potential. For years, small molecules targeted the activity of this enzyme class. Lilly Research Laboratories developed the potent and selective sPLA₂ inhibitor Varespladib (LY315920) with a low nanomolar IC₅₀ value against recombinant human enzyme (9 nM against hGIIA PLA₂) [101]. This commercially available compound targets the catalytic site of the enzyme and is selective over cPLA₂s. The Lilly group later modified Varespladib to increase cell permeability through the addition of a methyl ester, which is rapidly hydrolyzed once inside the cell. Varespladib methyl was taken on to clinical trials for the treatment of severe sepsis but was terminated in phase II due to underperformance. However, Varespladib methyl remains in phase III clinical trials for the treatment of cardiovascular disease [95, 102]. Other small molecules commercially available for the inhibition of secreted PLA₂ are reviewed elsewhere [95, 100].

The exploration of cytosolic PLA₂ inhibitors as novel pharmaceutical agents for inflammatory diseases led to the generation of several potent and selective small molecules. Wyeth developed several submicromolar indole-based cPLA₂ inhibitors including Ecopladib [103], Efipladib [104], and Giripladib [105]. Giripladib advanced to clinical trials for the treatment of osteoarthritis but terminated early because it could not improve upon the standard of care [100]. A recent study using Giripladib (PLA-695) showed that the compound sensitizes non-small cell lung cancer xenograft tumors in mice to radiation treatment and inhibits angiogenesis [106].

The pool of small molecules targeting the Ca²⁺-insensitive class of PLA₂ enzymes is very limited. Lipid or peptide-like inhibitors of iPLA₂ reported to be direct, reversible and specific showed poor pharmacokinetic profiles in vivo. Thus far the best iPLA₂ inhibitors are the polyfluoro ketones, including the commercially available FKGK11, which directly inhibits iPLA₂ activity [107]. Additional polyfluoro ketones (FKGK18 and GK187) with improved potency and selectivity have been reported, but are not commercially available [108, 109].

The Lp-PLA₂-targeting Darapladib is a direct, reversible and highly potent small molecule inhibitor with sub-nanomolar potency (IC₅₀=0.25 nM against rhLp-PLA₂) [110]. Due to the involvement of Lp-PLA₂ in atherosclerotic inflammation, Darapladib is currently in phase III clinical trials for treatment of cardiovascular disease [111].

2.4.1. Enzyme activity and regulation

Autotaxin (ATX) belongs to the nucleotide pyrophosphatase/phosphodiesterase (NPP) family of enzymes responsible for the hydrolysis of phosphodiester bonds of nucleotides. Unlike other family members, however, ATX shows unique specificity for lipid substrates. ATX is a secreted lysophospholipase D that catalyzes the hydrolysis of lysophosphatidylcholine (LPC) into lysophosphatidic acid (LPA) and choline (Fig. 3A) [112]. ATX was initially identified in the 1990s as a motility factor capable of stimulating cell migration yet the catalytic activity remained unknown. Subsequent reports identified ATX as the prominent source of signaling-specific lysophosphatidic acid (LPA), a ligand for a subclass of lipid-binding GPCRs.

There are three isoforms of ATX resulting from alternative splicing: α, β, and γ. All three isoforms exhibit similar catalytic activities but differ in tissue distribution and expression, with ATXβ as the predominant and most well characterized isoform. ATX is initially expressed as a pre-proenzyme that contains an N-terminal secretion signal later removed by peptidase hydrolysis. The cleaved protein is then converted to its active form through the action of a proprotein convertase [113]. The catalytic activity of ATX relies on an interaction between a threonine residue and zinc ions in the active site. An adjacent hydrophobic pocket facilitates lipid substrate binding. Subsequent crystallization of mammalian ATX enzymes in 2011 provided clearer insight into the activation, mechanism, and potential inhibition of this enzyme [114, 115]. The large hydrophobic pocket of ATX is notoriously non-selective. While its most well-documented role is the production of LPA, ATX can also generate sphingosine 1-phosphate (S1P) from sphingosylphosphorylcholine (SPC), which is chemically distinct from its native substrate [116].

LPA, the catalytic product of ATX, is a potent agonist at six GPCRs. LPA mediates distinct physiological roles such as initiation of migration, proliferation and cell survival. The LPA receptors are divided into two subgroups: endothelial differentiation gene family (Edg) members and a less well characterized family more closely related to the purinergic receptors [117]. LPA receptor stimulation results in the activation of essential signal transduction pathways including the MAP Kinase cascade and PI3K signaling, each of which are key nodes in cancer pathogenesis [118]. In addition, LPA receptor stimulation also promotes the production of cytokines and growth factors, further contributors to survival [119]. Since the LPA receptors activate these critical survival pathways in the cell, LPA generated by ATX remains a major focus for cancer therapeutic development [120]. Several breast cancer lineages display aberrant LPA receptor and ATX expression, a sufficient mechanism to induce late onset invasive, metastatic breast cancers in transgenic mice [121]. ATX itself is differentially expressed in cancerous tissues compared to normally differentiated tissues and may function as a marker of cancer prognosis and a biomarker for cancer detection [112]. Importantly, overexpression of ATX in non-neoplastic diseases such as acute coronary syndrome and liver disease, suggests that modulators of this enzyme promote disease states other than cancer [122].

2.4.2. Tools for modulation of enzymatic activity

3.1.1. Enzyme activity and regulation

Diacylglycerol kinases (DGKs) catalyze the phosphorylation of diacylglycerol (DAG) to PA (Fig. 4A); most kingdoms of life—from bacteria and yeast to mammals—express DGK enzymes. Importantly, the specialization of DGK isoforms and their regulatory intricacies increase with organism complexity: bacteria and yeast express a single DGK enzyme while humans express ten isozymes, most possessing multiple variants [132, 133].

The product of DGK enzymatic activity, PA, is an important signaling lipid with multiple functions. Enzymes as diverse as mTOR [134], sphingosine kinase [135], and phosphatidylinositol-4-phosphate 5-kinase [136] are all activated by PA. Yet the substrate of the DGK enzymes, DAG, is an important signaling molecule in its own right. DAG prominently regulates PKC, which regulates several signaling cascades [137]. Thus, the catalytic activity of the DGKs governs the concentrations of two critical classes of signaling molecules and places the DGKs at a unique metabolic nexus. The importance of DGKs in such diverse signaling pathways makes them immensely important targets for a variety of disease states and physiological processes, ranging from hyperproliferative diseases to neurotransmitter release.

Humans express a total of 10 DGK isoforms (α–κ), classified into five sub-families based on sequence homology. In most cases, these differences in sequence homology result in alternate modes of regulation. Type I DGKs include α, β, and γ and are characterized by their calcium-binding EF motif, making calcium an important regulator of these enzymes [138]. Empirical evidence also suggests that other lipids such as sphingosine may play an important regulatory role for the type I DGKs [139]. The second sub-family of DGKs, type II enzymes (δ, η, and κ), each possess a pleckstrin homology (PH) domain containing PKC phosphorylation sites for regulation. Phosphorylation of type II DGKs in the PH domain inhibits translocation to the plasma membrane thereby decreasing PA turnover [140]. DGKε is the only known type III isoform and does not contain an identifiable regulatory domain. However, empirical evidence suggests that anionic lipids negatively regulate enzymatic activity [141]. DGKζ and DGKι are classified as type IV DGKs and possess four ankyrin repeats and a myristoylated alanine-rich C-Kinase (MARCKS) phosphorylation site. MARCKS domains localize the type IV DGKs to the nucleus, a process negatively regulated by PKC phosphorylation [142]. RhoA association negatively regulates the enzymatic activity of DGKθ, the only known type V sub-family member. [143].

All tissues contain at least one member of each DGK sub-family with most tissues expressing multiple isozymes. However, the various DGK isoforms display stark contrasts in subcellular localization, suggesting that distinct pools of DAG and PA within specific cellular compartments perform discrete physiological roles despite catalyzing the same reaction. For instance, DGKβ closely associates with actin filaments and is vital for cytoskeletal processes [144], while DGKδ closely associates with endosomes [145]. DGKα localizes to membranes upon activation and regulates global DAG content in cells [146] while DGKε selectively phosphorylates arachidonoyl-containing substrates making it a critical regulator of the PI turnover cycle [147, 148]. DGKδ, regulates de novo fatty acid synthesis and lipogenesis; as a result, knockout mice perish shortly after birth[148]. In sum, the various DGK isozymes regulate a host of physiological processes, reviewed elsewhere in detail [132, 133].

3.1.2. Tools for modulation of enzymatic activity

3.2.1. Enzyme activity and regulation

Perhaps the most well characterized and promising area for lipid metabolism and cancer research focuses on phosphatidylinositol 3-kinase (PI3K) and its downstream signaling cascade. The PI3K enzymes phosphorylate the 3′-hydroxyl group of phosphatidylinositols (PIs). This phosphorylation event regulates cell growth, migration, metabolism, angiogenesis, and survival [156]; other intracellular signaling proteins rely on the products of PI3K for activation. Importantly, many cancers display hyper-activation of the PI3K pathway, a key component to their survival [157]. There are 3 classes of PI3K enzymes some with additional isoforms. Denoted as classes I, II, and III, each class exhibits different substrate specificity [158]. Class I PI3K enzymes, the most relevant class for cancer survival, sit downstream of cell surface receptors and primarily phosphorylate PI(4,5)P₂ to generate PI(3,4,5)P₃. Class II PI3Ks produce PI(3)P and PI(3,4)P₂ from PI and PI(4)P, respectively. Class III only contains one enzyme, Vps34p which converts PI into PI(3)P. In addition to PI3K, other classes of lipid kinases phosphorylate the remaining positions of PI(4′ and 5′) on the inositol ring, generating the other PIP_n lipids.

Class I PI3K enzymes are heterodimers made up of a regulatory subunit, p85 for class IA or p101 for class IB, and a catalytic subunit, p110. Various regulatory subunit isoforms on the PI3Ks lead to differential activation by cell surface receptors, specifically receptor tyrosine kinases [159]. Class IA enzymes have a highly conserved catalytic domain, expressed as one of three isoforms, p110α, β, and δ. The p110 catalytic subunit is made up of a p85 binding region, a Ras-binding domain, a C2 domain, a PIK homology domain and a C-terminal catalytic domain [159]. The PIK domain is found in other kinases including mTOR, suggesting that these proteins are phylogenically related.

PI3K signaling pathways centrally sustain cell growth under a plethora of stimulation conditions. Receptor tyrosine kinase stimulation activates class I PI3Ks. Meanwhile, insulin, insulin growth factor, or platelet derived growth factor all trigger PI3K activity directly or through adaptor protein recruitment [159]. GPCRs also stimulate Class IB PI3K proteins through binding of the Gβγ subunit to the regulatory domain of PI3K.

The canonical PI3K-signaling pathway focuses on the production of PIP₃, a critical lipid second messenger. The opposing actions of PI3K and PTEN (phosphatase and tensin homologue) balance this system; PTEN dephosphorylates PIP₃ generating PIP₂ and terminating PIP₃ signaling. Aggressive cancers often display mutations in or complete loss of PTEN, leading to hyperactive PI3K signaling. The protein kinase Akt, considered the primary downstream effector of PI3K signaling, phosphorylates multiple target proteins affecting a broad range of cellular processes [159].

Class IA PI3Ks regulate cell survival, sitting downstream of growth factor receptors; many members of this signaling pathway work together in promoting tumor pathogenesis. In certain cancer types, specifically cervical or squamous cell lung cancer, mutations or amplifications in PI3K are present in 60% of cases [160]. The PI3K-Akt signaling nexus is often considered the major effector of metabolic insulin action though the insulin receptor and insulin-sensitive kinase activation. Proper metabolic control relies on the correct functioning of these signaling pathways. Dysregulation of these pathways often leads to impaired glucose homeostasis [161]. Because of its prevalence in cancer and diabetes progression, interest in PI3K as target of drug development grew substantially in the past decade.

3.2.2. Tools for modulation of enzymatic activity

4.1.1. Enzyme activity and regulation

Monacylglycerol lipase (MGLL) facilitates the hydrolysis of monoacylglycerol (MAG) lipids to free fatty acids and glycerol, the final step in fatty acid catabolism (Fig. 5A). First cloned from mice, the MGLL gene is a relatively small enzyme with a molecular weight of 33 kDa [167]. Mammals only express one isoform of MGLL, which contains the characteristic serine hydrolase GXSXG catalytic motif in its active site. MGLL activity is detected in a variety of other tissues including brain, kidney, liver, heart, brown adipose tissue, and lung [168]. Primarily a cytosolic protein, MGLL can associate with membranes via a lid domain adjacent to the catalytic site [169]. Initially thought of as solely important to lipid metabolism, recent studies confirmed MGLL as a prominent regulator of three critical signaling nodes.

MGLL activity negatively regulates cellular concentrations of the endocannabinoid 2-arachidonoylglycerol (2-AG). Along with anandamide, 2-AG is known as a retrograde synaptic suppressor, binding the endocannabinoid (EC) receptors, the same GPCRs responsible for binding, Δ⁹-tetrahydrocannabinol, the psychoactive component of marijuana [170]. When MGLL cDNA was transfected and overexpressed in neurons, 2-AG signaling decreased at the EC receptors, supporting the hypothesis that MGLL negatively regulates 2-AG concentrations [171]. These findings were later corroborated in vivo with a small molecule inhibitor of MGLL that demonstrated behavioral effects consistent with sustained cannabinoid receptor agonism [172].

A second critical signaling node regulated by MGLL activity centers on the arachidonic acid (AA) product of 2-AG hydrolysis by MGLL. AA is the essential fatty acid utilized for prostaglandin biosynthesis; oxidation of AA by the cyclooxygenase (COX) enzymes generates the core prostaglandin scaffold. COX enzymes are recognized as key regulators of inflammatory responses and are the targets of non-steroidal anti-inflammatory drugs [98]. As discussed previously, cPLA₂ activity generates the critical pool of AA needed for COX-mediated prostaglandin biosynthesis in most tissues [173]. Indeed, cPLA₂-knockout mice display altered levels of AA in nearly all tissues except, notably, the CNS [174]. A recent study demonstrated that MGLL activity on 2-AG regulated prostaglandin-specific AA production in the brain. Blockade of MGLL enzymatic activity with a small molecule inhibitor blunted LPS-induced inflammatory responses mediated by prostaglandins in neural tissues [175].

Aside from its roles in the nervous system regulating 2-AG and prostaglandin biosynthesis, MGLL also contributes to cancer malignancy. A recent study found that MGLL activity was upregulated in aggressive cancer types and that transfection of the MGLL gene into more benign tumor lines increased their aggressiveness and invasiveness [176]. This initial report found that MGLL regulated free fatty acids (FFAs) in aggressive cancer types and contributed to lipogenic phenotypes. A follow up study found that in addition to FFAs, MGLL regulation of 2-AG in prostate cancer regulated metastasis [177], identifying MGLL as a vital component to pro-survival signaling pathways in aggressive cancer types.

4.1.2. Tools for modulation of enzymatic activity

Early classes of MGLL inhibitors suffered from either insufficient potency or selectivity. The first reported MGLL-selective inhibitor, URB602 (Fig. 5B), was based on the structure of a previously identified carbamate-based inhibitor of fatty acid amide hydrolase (FAAH) [178], the primary enzyme responsible for the biosynthesis of the second endocannabinoid anandamide [179]. However, subsequent studies found URB602 lacked selectivity for MGLL over FAAH [180] as well as the potency necessary for in vivo characterization [181].

While initial MGLL inhibitors were not suitable chemical tools, the use of carbamates to disrupt the serine hydrolase activity of MGLL remained a fruitful strategy. These privileged scaffolds function by entering the enzyme’s active site and introducing an irreversibly-bound carbamate moiety (Fig. 6). Use of this strategy led to the identification of the MGLL-selective inhibitor JZL-184 (Fig. 5B and Table 1) and marked a turning point in MGLL studies. Identified through a screen of N-piperidine and N-piperidine-linked carbamate scaffolds, JZL-184 displayed an IC₅₀ value of 10 nM at MGLL and nearly 5 μM at FAAH. Thus JZL-184 achieved 500-fold selectivity for regulating 2-AG-specific signaling pathways in the CNS [182]. Consistent with the proposed covalent binding mechanism, LC-MS studies detected the carbamate product of JZL-184 hydrolysis from tryptic residues of the MGLL active site [168]. JZL-184 is currently the standard chemical tool for the analysis of the physiological roles of MGLL both in vitro and in vivo and is currently commercially.

In spite of the successes of JZL-184, the compound does display some off-target activity against other serine hydrolase targets. Studies showed that prolonged doses of JZL-184 in vivo result in low levels of FAAH inhibition [168]. A new structural class of compounds identified in the Cravatt laboratory utilizes a hexafluoroisopropyl moiety as part of the carbamate scaffold (KML29). These compounds more closely resemble the structure and reactivity of 2-AG and thus display greater selectivity for MGLL [183]. A second, more recent class of MGLL inhibitors, based on a piperidine triazole urea scaffold, utilizes the same general strategy for irreversible serine hydrolase inhibition (Fig. 6). These series of compounds identified the most potent inhibitor of the enzyme yet reported: JJKK048 inhibits MGLL with an IC₅₀ value of 0.2 nM (Fig. 5B) [184].

4.2.1. Enzyme activity and regulation

Diacylglycerol lipase (DAGL) preferentially hydrolyzes DAG species at the sn-1 position yielding sn-2-substituted MAG and FFAs (Fig. 5A) [185]. DAGL exists as two isoforms, designated as DAGLα and DAGLβ [186]. The two isoforms of DAGL regulate tissue-specific pools of DAG metabolism, with DAGLα primarily localized to the nervous system and DAGLβ localized to peripheral tissues such as the liver [187, 188]. Although the two isozymes vary greatly in size, with DAGLα and DAGLβ having a 1042 and 672 amino acid sequences, respectively, they display extensive sequence homology and similar subcellular distribution [185].

The substrate specificity of the DAGL isoforms grants them two important signaling roles. The specificity of the DAGL enzymes for sn-1 hydrolysis of DAG species makes them tissue-specific producers of 2-AG, counteracting the activity of MGLL [186]. Due to its high level of expression in the nervous system, DAGLα is the primary regulator of 2-AG levels in the brain. Indeed, studies on DAGLα^−/− mice revealed these animals lack retrograde synaptic suppression [188]. A higher level of DAGLβ expression outside the nervous system makes this isoform the critical regulator of 2-AG in peripheral tissues [187].

As with MGLL, a second, subtler role of DAGL lies in its ability to regulate tissue-specific pools of AA. Studies on DAGLβ^−/− mice identified this isoform as the critical regulator of AA in liver [187]. Subsequent studies on DAGLβ in vivo revealed that decreases in AA pools in peripheral tissues led to decreased prostaglandin biosynthesis and suppressed local inflammatory responses [189].

4.2.2. Tools for modulation of enzymatic activity

Given the important role of DAGL in 2-AG biosynthesis and retrograde synaptic suppression, inhibitors of DAGL traditionally focused on the inhibition of the DAGLα isoform. The first functionally-identified DAGL inhibitor, RHC80267, utilizes a cyclohexanone O-acetyl oxime moiety for irreversible enzyme inhibition, outlined in Fig. 6. RHC80267 decreased 2-AG levels in cortical neurons but was not evaluated against recombinant DAGL [190]. Later cloning of both DAGL isoforms confirmed that RHC80267 inhibited DAGL at 100 μM concentrations but could not distinguish between either of the DAGL isoforms [186]. High IC₅₀ values and poor specificity make RHC80267 an unsuitable tool compound for interrogating DAGL enzymatic activity in native biological systems.

Fluorophosphonate (FP) lipid analogues are well known irreversible inhibitors of serine hydrolase activity. Studies on FP lipid analogues identified two additional DAGL inhibitors. Initial studies on the DAG-FP analogue O-3841 (Fig. 5C) reported an IC₅₀ of 160 nM on the DAGLα isoform [191]. Following suit, reports surfaced of a second arachidonic acid fluorophosphonate inhibitor, known as MAFP. Not surprisingly, this compound also displayed substantial off-target activity towards MAGL [191]. FP lipid analogues are highly-reactive and bind non-selectively to a majority of serine hydrolases [192], calling into question their reliability as scaffolds for selective inhibitors. Indeed, activity-based protein profiling (ABPP) regularly takes advantage of this broad reactivity of lipid FP analogues to survey global cellular serine hydrolase activity [193, 194].

A third class of early DAGL inhibitors, based on tetrahydrolipstatin (THL, Orlistat), utilizes a beta-lactone functionality to irreversibly bind the DAGL active site [195]. Figure 6 describes the unique inhibitory mechanism of these analogues, reminiscent of β-lactam antibiotics. Later studies on THL raised suspicion of off-target effects at the CB receptors. Medicinal chemistry efforts have since focused on eliminating this activity from THL analogues. These studies identified three compounds having improved potency at DAGL over THL but with only modest gains in selectivity [196]. When evaluated by ABPP, this class of inhibitors proved more selective than the lipid-FP analogues but still displayed substantial off-target effects, making them unsuitable for in vivo use [193].

A recent study identified 1,2,3-triazoleureas as a dynamic scaffold for the inhibition of serine hydrolases. Like the carbamate moieties used in the development of MGLL inhibitors, 1,2,3-triazoleureas irreversibly bind to the active site of serine hydrolases (Fig. 6). Importantly, their streamlined, modular synthesis makes them amenable to rapid library expansion [197]. Using this strategy as a starting point, Cravatt and coworkers devised a fourth class of DAGL inhibitors. A recent report identified two 1,2,3-triazoleurea DAGLβ-selective inhibitors (KT109 and KT172), which displayed mid-nanomolar potencies with good selectivity over the other EC biosynthetic targets FAAH and MGLL [189]. However, KT109 and KT172 have prominent off-target activity against ABHD6. The negative control published alongside these inhibitors, KT195 displayed no activity toward either DAGLα or DAGLβ but retains the off-target effects on ABHD6 and is best used alongside KT109 and KT172 to study DAGL in vivo and in vitro.

4.3.1. Hormone-sensitive lipase

4.3.2. Adipose triglyceride lipase (ATGL)

4.3.3. CGI-58 (ABHD5)

4.3.4. PNPLA3

5.1.1. Enzyme activity and regulation

In the de novo biosynthesis of TAG, lipin catalyzes the penultimate step dephosphorylating PA to produce DAG (Fig. 8A) [277]. Kennedy first detected phosphatidic acid phosphatase activity from mitochondrial preparations in the 1950s [278], but the lipin enzyme remained uncharacterized until 2001 [279]. Mammals express three lipin isoforms (lipin1-3) each with differing tissue expression profiles. A broad set of tissues express lipin1 but the enzyme is predominantly found in the muscle and adipose tissue of humans. Lipin2 and 3 have much lower levels of expression with lipin2 localized to brain, liver, placenta, and adipose tissue while lipin3 is broadly expressed albeit at lower levels [280].

Lipin1 is the most biochemically and pathophysiologically characterized member of the lipin family. Genetic studies on the fatty liver dystrophic (fld) mouse identified the lipin gene. fld mice exhibit extreme hypertriglyceridemia, developmental retardation, early mortality, and abnormal adipose tissue development [281]. Interestingly, these animals maintained normal TAG levels in the liver. This characteristic fld phenotype was subsequently linked to a loss-of-function mutation in the lipin1 protein [279]. Later studies found that lipin2 compensates for the defective lipin1 in hepatocytes [282].

Given the importance of lipin to anabolic metabolism of fats, researchers postulated that nutrient-sensing mechanisms regulated the activity of this enzyme. Less than a year after the initial characterization of the lipin gene from fld mice, Lawrence and coworkers identified several mTORC1-specific phosphorylation sites on the lipin protein, placing it under the umbrella of insulin signaling [283]. In spite of these findings, regulation of lipin activity is not conducted strictly via post-translational modifications. Rather, sub-cellular translocation of lipin results in phosphorylation and increases enzymatic activity [284]. Carefully controlled expression mechanisms also account for regulations in lipin activity. For example peroxisome proliferator-activated receptor γ (PPARγ) regulates lipin1 expression to increase fatty acid oxidation and oxidative phosphorylation in the mitochondria [285].

Loss-of-function lipin1 mutations in humans do not exhibit the extreme lipodystrophy observed in fld mice. However, recent studies demonstrated that perturbations in lipin1 expression in humans result in insulin-insensitivity and poor glucose metabolism [286–288]. Thus small molecule activators of lipin1 activity may prove an important class of small molecule chemical tools to study the role of lipin in normal metabolic flux.

5.1.2. Tools for modulation of enzymatic activity

Metabolic disorders such as lipodystrophy and type 2 diabetes mellitus are mounting health crises facing the developed world. Identification of loss-of-function lipin1 mutants in the fld mouse demonstrates the enzyme’s critical role in hepatic TAG production and circulating lipid levels. Hormonal and transcriptional regulators critical to energy balance, PPARγ and insulin, are known modulators of the expression and activity of lipin1. Taken together, these studies suggest that lipin1 is central to lipid metabolism and energy balance in vivo.

Despite the growing importance of lipin1 to metabolic homeostasis, no small molecules capable of inhibiting the enzymatic activity of lipin currently exist. Interestingly, these enzymes regulate the same biosynthetic equilibrium as the DGKs, cellular concentrations of PA and DAG, which are critical lipid second messengers in cells. While genetic perturbation studies identified critical roles of lipin in metabolic development, the ability to acutely inhibit the activity of this enzyme may reveal new and important roles for lipin in physiological processes.

6.2.1. Enzyme activity and regulation

First described by Kennedy [290], GPAT commences de novo phospholipid and TAG biosynthesis by esterifying G3P with fatty acyl CoA at the sn-1 position (Fig. 8A). Initial studies identified two GPAT isoforms that differed in cellular and tissue localization as well as substrate preference [291, 292]. Further study identified two additional GPAT isoforms, each as a unique gene product [289, 293]. GPAT1 is an intrinsic membrane protein primarily located in the mitochondria of hepatic cells [294] and preferentially acetylates G3P with saturated fatty acyl-CoAs [293], particularly 16:0. The enzyme regulates hepatic TAG levels by controlling TAG biosynthesis [295]. GPAT1^−/− mice, despite being obese, display 60% less TAG content in the liver compared to their wild-type littermates [296]. Gene expression, primarily facilitated by SREBP-1c, regulates GPAT1 expression in response to nutritional and hormonal cues. During periods of fasting and decreased fat storage, GPAT1 activity also decreases in the liver [297].

GPAT2, the second mitochondrial isoform, displays no obvious substrate preference. The enzyme has 72% sequence similarity to GPAT1 and an estimated mass of 89 KDa. GPAT2 is much more highly expressed in testis than GPAT1 and is likely to have a specialized function there [297]. Experimental evidence suggests that GPAT2 expression in the liver occurs independently of nutritional fluxes [297].

The recently identified GPAT3 and 4 associate closely with the endoplasmic reticulum [298–300]. Both exhibit similar levels of expression in adipose tissues, liver, and heart. However, 3T3-L1 differentiating adipocytes specifically induce high levels of GPAT3 [301]. GPAT3 and 4 lack the extended C-terminal region found in GPAT1 and 2, suggesting important differences in protein regulation.

6.1.2. Tools for modulation of enzymatic activity

6.2.1. Enzyme activity and regulation

A number of mechanisms account for total PA production in the cell. De novo biosynthesis relies on the sn-2 acylation of LPA by LPAAT (Fig. 8A). Kennedy published the first reports of LPAAT activity in the early 1960s [306, 307], but the six human LPAAT isoforms (α-ζ) remained uncharacterized until the late 1990s [308]. The α and β isoforms are the most well studied LPAAT isozymes [309] and possess relatively low molecular weights (31.7 and 30.9 KDa, respectively [308]). Humans express LPAATα in most tissues but the enzyme is more abundant in brain, heart, spleen, thymus, kidney, and skeletal muscle [308]. Expression of the β isoform is limited to the liver, pancreas, adipose tissue and heart. Interestingly, select tumor types overexpress LPAATβ, specifically lung, colon, cervix, brain, breast, ovary, and prostate [308, 310, 311]. The α isomer preferentially adds 14:0, 16:0, and 18:2 fatty acyl CoAs to LPA while the β isomer preferentially adds 18:1, 18:2, and 18:3 fatty acyl CoAs [309]. Specific mechanisms for LPAAT regulation remain uncharacterized.

6.2.2. Tools for modulation of enzymatic activity

6.3.1. Enzyme activity and regulation

Whether TAG is synthesized through the canonical “Kennedy pathway” [318] or the MAG pathway [319], TAG biosynthesis requires DGAT activity (Fig. 8A) [289, 291, 320]. Two DGAT isoforms exist in eukaryotes, both localized to the endoplasmic reticulum. Tissues specializing in TAG storage and biosynthesis display elevated DGAT1 and 2 mRNA levels expression, which correlates to increased TAG production [289]. There are significant differences between the two isoforms: DGAT1 belongs to the acyl-CoA:cholesterol acyltransferase (ACAT) enzyme family, while DGAT2 is part of a 7 member family more closely related to the MGAT enzymes.

DGAT1 is a 55 kDa protein abundantly expressed in adipose tissue, small intestine, liver, and mammary glands [321]. This versatile enzyme catalyzes DAG, retinol and wax esterification due to its broad acyltransferase activity [320]. The DGAT2 isozyme was not identified until 2001 [322] and is smaller than DGAT1 with a molecular mass of 40–44 kDa but contains the conserved amino acid sequences essential for neutral lipid binding. Unlike DGAT1, DGAT2 has greater substrate specificity and does not catalyze the synthesis of wax or retinyl esters. There is no obvious substrate preference for either isoform with respect to chain length and unsaturation of acyl CoAs. Both isoforms exhibit similar expression patterns, mostly in tissues and organs that generate large amounts of TAG and are regulated by transcriptional regulators of adipogenesis. Nutritional and hormonal cues regulate DGAT1 and 2 activities at the transcriptional level [323, 324].

Dysregulation of either DGAT enzyme contributes to metabolic disorders such as diabetes, liver steatosis, obesity, and insulin resistance. DGAT’s role in energy metabolism, leptin sensitivity, glucose metabolism [325], lipotoxicity [326], and inflammation [327] makes it an important therapeutic target. While a DGAT1 deficiency results in reduced TAG production and resistance to high-fat diet-induced obesity [328], loss of DGAT2 is lethal in neonatal mice [329]. These observations suggest that DGAT1 and 2 have non-redundant, non-compensatory functions in spite of overlapping catalytic activity and tissue expression. Wurie and colleagues published a report suggesting that DGAT2 acts upstream of DGAT1 and serves as the rate-limiting step for de novo TAG synthesis. Meanwhile, DGAT1 is largely responsible for the production of TAG via DAG re-esterification [330].

6.3.2. Tools for modulation of enzymatic activity