Small Molecule Inhibitors Targeting Biosynthesis of Ceramide, the Central Hub of the Sphingolipid Network

Abstract

Ceramides are composed of a sphingosine and a single fatty acid connected by an amide linkage. As one of the major classes of biologically active lipids, ceramides and their upstream and downstream metabolites have been implicated in several pathological conditions including cancer, neurodegeneration, diabetes, microbial pathogenesis, obesity, and inflammation. Consequently, tremendous efforts have been devoted to deciphering the dynamics of metabolic pathways involved in ceramide biosynthesis. Given that several distinct enzymes can produce ceramide, different enzyme targets have been pursued depending on the underlying disease mechanism. The main objective of this review is to provide a comprehensive overview of small molecule inhibitors reported to date for each of these ceramide-producing enzymes from a medicinal chemistry perspective.

Graphical Abstract

1. INTRODUCTION

Sphingolipids belong to a class of non-glycerol-based lipids built on a sphingosine backbone. Sphingosine derivatives acylated at the amino group with fatty acids of different chain length and degree of unsaturation are termed ceramides. Because of the structural variation in the fatty acid component, ceramides can occur in several structurally different forms. For instance, ceramide formed from sphingosine (d18:1) and oleic acid (18:1(9Z)) containing a total of 18 carbons and a C8–C9 Z-double bond is referred to as Cer(d18:1/18:1(9Z)).¹ Cumulative evidence suggests that different ceramides, in terms of the fatty acid composition, may play distinct and tissue-specific physiological roles under various biological contexts.²

Ceramides are known to be synthesized through three major pathways, namely, the de novo biosynthesis pathway, the salvage pathway, and sphingomyelin hydrolysis (Figure 1).^3,4 Because of their pathophysiological implications in diseases, some of the enzymes involved in these pathways have been extensively studied as therapeutic targets including ceramide synthase, dihydroceramide desaturase, and sphingomyelinase, each of which catalyzes the final step in the de novo biosynthesis pathway, the salvage pathway, and sphingomyelin hydrolysis, respectively. Other ceramide-producing enzymes include glucocerebrosidase, galactocerebrosidase, and ceramide-1-phosphate phosphatase. With the increasing interest in ceramide biosynthesis as therapeutic targets, there are a number of well-written review articles detailing the biological and physiological aspects of these ceramide-producing enzymes.^5–16 There are also review articles on small molecule inhibitors of these enzymes,^17–21 though they are relatively limited in scope and lack detailed insights from a medicinal chemistry perspective. This Perspective provides a comprehensive and in-depth overview of small molecule inhibitors reported to date for each of these ceramide-producing enzymes including the latest development in this field with greater potential for therapeutic application.

Figure 1.

Ceramide biosynthesis pathways (adapted from ref 5). Various biosynthetic routes to ceramide are depicted for Cer(d18:1/18:1(9Z)). De novo pathway, salvage pathway, and sphingomyelin hydrolysis are highlighted in purple, orange, and cyan, respectively. Other sources of ceramide include glucocerebrosidase (gray), galactocerebrosidase (green), and ceramide-1-phosphate phosphatase (beige).

2. DIHYDROCERAMIDE DESATURASE

De novo synthesis of ceramide begins with decarboxylative condensation of l-serine and palmitoyl-CoA catalyzed by serine C-palmitoyltransferase. The resulting product, 3-dehydrosphinganine, is converted into sphinganine by 3-dehydrosphinganine reductase. Ceramide synthase catalyzes the acylation of sphinganine with fatty-acyl-CoAs of varying chain length to yield dihydroceramides. In the final step of the de novo biosynthesis pathway, dihydroceramide desaturase converts dihydroceramides into ceramides. While dihydroceramide desaturase 1 (DES1) encoded by DEGS1 in human displays only desaturase activity, its closest paralog, dihydroceramide desaturase 2 (DES2) encoded by DEGS2, exhibits both desaturase and hydroxylase activities. DEGS1 deficiency has recently been reported to lead to an imbalance between saturated/unsaturated sphingolipids and cause leukodystrophy and peripheral hypomyelination,²² highlighting the clinical significance of DES1. DES1 has been also implicated as a therapeutic target in various diseases including cancer,²³ Alzheimer’s disease,²⁴ and diabetes.²⁵ Along with these findings, a number of structurally diverse DES1 inhibitors have been reported in the literature.²¹

GT11 (1) is one of the first DES1 inhibitors described in the literature (Figure 2).²⁶ It is a ceramide derivative in which the C4–C5 double bond of the sphingosine component is replaced by a cyclopropene moiety. The design of this molecule was based on early reports on the activity of sterculic acid as a potent inhibitor of the Δ9-stearoyl-CoA desaturase.²⁷ As expected from its structure, kinetics studies using rat liver microsomes revealed that GT11 (1) inhibits DES1 in a competitive manner with respect to N-octanoylsphinganine with a K_i value of 6 μM.²⁸ Interestingly, GT11 (1) was found to inhibit dihydroceramide desaturase activity in primary cultured cerebellar neurons with an approximately 250-fold greater inhibitory potency, presumably due to a local subcellular enrichment of the inhibitor in the endoplasmic reticulum, the main site of action of the enzyme.²⁹ It should be noted, however, that GT11 (1) was also reported to display greater potency when HGC27 cell lysates were used as a source of the enzyme with an IC₅₀ value of 52 nM and a K_i value of 22 nM.³⁰

Figure 2.

Structures of ceramide-based DES1 inhibitors GT11 (1), XM462 (2), and compound 3.

XM462 (2) is a mechanism-based DES1 inhibitor in which the C5 methylene group of the sphinganine component is replaced by a sulfur atom (Figure 2).³¹ It inhibits desaturase activity in rat liver microsomes with an IC₅₀ value of 8.2 μM. The rationale for incorporating a sulfur atom into the sphinganine component is based on the previous reports on the activity of some thia fatty acids as fatty acyl-CoA desaturase inhibitors.^32,33 XM462 (2) was found to be a mixed-type inhibitor with K_iapp and α values of 2 μM and 0.83, respectively. Although the precise mechanism of inhibition was not investigated, the ceramide analog is anticipated to bind to the enzyme active site, where its C4 hydrogen is removed as a radical to afford the carbon radical intermediate in equilibrium with the sulfur radical species. The inhibition may then arise from coordination of the intermediate(s) to the enzyme active site.

There are several other ceramide analogs reported as DES1 inhibitors including those derived from GT11 (1)^28,34 and XM462 (2).³⁵ Most notably, ceramide analog 3 (Figure 2) containing a C6–C7 double bond was found to act as a potent DES1 inhibitor with an IC₅₀ value of 155 nM when HGC27 cell lysates were used as a source of the enzyme.³⁰ Contrary to expectations based on its structural similarity to DES1 substrates, compound 3 showed a noncompetitive type of inhibition.

Fenretinide (4), also known as N-(4-hydroxyphenyl)-retinamide (4-HPR), is a synthetic retinoid derivative originally developed as a chemotherapeutic agent because of its ability to attenuate cancer cell growth with relatively low toxicity (Figure 3).³⁶ Fenretinide (4) was first recognized as a DES1 inhibitor when it was found to increase dihydroceramides in a dose dependent manner in SMS-KCNR cells.³⁷ Subsequently, DES1 was confirmed as the primary target for fenretinide (4) using rat liver microsomes with an IC₅₀ value of 2.32 μM.³⁸ In 20 min incubation experiments, fenretinide (4) acted as a competitive inhibitor with a K_i value of 8.28 μM. In addition to fenretinide (4), its known metabolites were evaluated for their ability to inhibit desaturase activity. Among them, 4-oxo-N-(4-hydroxyphenyl)retinamide 5 (4-oxo-4-HPR, Figure 3) showed the highest inhibitory potency with an IC₅₀ value of 1.68 μM. N-(4-Methoxyphenyl)retinamide 6 (4-MPR) and 4-oxo-N-(4-methoxyphenyl)retinamide 7 (4-oxo-4-MPR) had minimal effects on DES1 activity (Figure 3), demonstrating the essential role played by the phenolic moiety in the potent inhibitory activity.

Figure 3.

Structures of retinoid-based DES1 inhibitors 4–7.

SKI II (8) is an orally available dual inhibitor of sphingosine kinases 1 (SK1) and 2 (SK2) with K_i values of 16 μM and 7.9 μM, respectively (Figure 4).^39,40 It was later discovered that SKI II (8) also inhibits DES1.⁴¹ Unlike fenretinide (4), SKI II (8) was found to act as a noncompetitive inhibitor of DES1 in HGC-27 cell lysates with a K_i value of 0.3 μM despite that they share the 4-aminophenol moiety. On the basis of these findings, it is speculated that SKI II (8) targets the upstream enzyme, cytochrome b5 reductase (Cb5R), which plays an essential role in the desaturation process by regenerating ferrocytochrome b5. Compounds 9 and 10 (Figure 4), analogs of SKI II (8), were also reported as inhibitors of desaturase activity in Jurkat cells with varying degrees of inhibitory potency and selectivity for SK1 and SK2,⁴² which should, together with other SKI II (8) analogs devoid of desaturase inhibitory activity, serve as useful probes to understand the biology of SK1, SK2, and DES1.

Figure 4.

Structures of SKI II (8) and its derivatives 9 and 10.

More recently, a series of DES1 inhibitors containing a 4-acylaminophenol or 6-acylamino-3-pyridinol scaffold were disclosed in a patent application (Figure 5).⁴³ Compounds 11 and 12 were reported to potently inhibit DES1 activity in Jurkat cells with an IC₅₀ value of 1 nM, while fenretinide (4), used as a control, exhibited an IC₅₀ value of 100–250 nM under the same assay conditions. The impact of converting a phenol to a 3-pyridinol moiety was most profound for compound 13, which exhibited >20-fold improvement in potency over the corresponding phenol derivative 14 possessing the same acyl group. At present, only limited pharmacological data are available for these highly potent DES1 inhibitors. It remains to be seen whether these compounds have the potential to be developed into viable therapeutic agents.

Figure 5.

Structures of DES1 inhibitors 11–14 containing a 6-acylamino-3-pyridinol or 4-acylaminophenol scaffold.

There are other non-ceramide analogs reported as DES1 inhibitors including resveratrol, celecoxib, curcumin, and Δ9-tetrahydrocannabinol (THC).²¹ Their utility as DES1 inhibitors, however, remains questionable given their rather weak inhibitory activity against DES1 coupled with the fact that they act on other targets more potently and/or display polypharmacological behaviors.

3. CERAMIDE SYNTHASE

Ceramide synthase (CerS) catalyzes the acylation of sphingoid bases with fatty-acyl-CoAs of varying chain length.⁴⁴ As mentioned earlier, CerS can recognize sphinganine as a substrate and produce dihydroceramides, penultimate intermediates of the de novo biosynthesis pathway. CerS can also directly produce ceramides when sphingosine is used as a substrate via the salvage pathway, reutilizing sphingosine formed as a result of degradation of higher order sphingolipids. CerS is encoded by six distinct genes (CerS1–6), each of which shows preference for a particular range of Acyl-CoA substrates (Figure 6).^10,44,45 For instance, CerS1, which is primarily expressed in the brain, has a high substrate preference for stearoyl-CoA. Indeed, C18-ceramide was reported to be the most abundant ceramide in the rat brain, representing nearly half of the total ceramide amount.⁴⁶ Given the distinct substrate specificity and tissue distribution shown by each of the six isoforms, it is not surprising that the clinical and pathological significance of CerS varies among them.^9,10,47 Therefore, the therapeutic potential of CerS inhibition highly depends on the ability to target a specific isoform of CerS. Until recently, though, there had been no isoform-selective CerS inhibitors, hindering efforts to develop CerS-targeting therapeutic agents.

Figure 6.

Acyl-CoA substrate preference of CerS isoforms.¹⁰

Fumonisin B1 (15) (Figure 7), a mycotoxin produced by Fusarium moniliforme, is one of the first reported CerS inhibitors.⁴⁸ In rat liver microsomes, fumonisin B1 (15) was found to inhibit conversion of [³H]sphingosine to [³H]ceramide with an IC₅₀ value of 0.1 μM. It is postulated that the two tricarballylic acid side chains of fumonisin B1 (15) play a crucial role in interacting with the acyl-CoA binding site while its amino group mimics that of sphinganine/sphingosine.⁴⁹ Inhibition kinetics studies using mouse brain microsomes indicated that fumonisin B1 (15) is a competitive inhibitor with respect to sphinganine while it displayed a mixed-type inhibition with respect to stearoyl-CoA.⁵⁰ Although speculative, it is conceivable that a fraction of fumonisin B1 (15) was hydrolyzed in brain microsomes to form aminopentol (16) (Figure 7), which is known to serve as a CerS substrate by making the fatty acyl-CoA binding site accessible to an acyl donor.⁵¹ This may explain the observed mixed-type inhibition with respect to stearoyl-CoA by fumonisin B1 (15). It should also be noted that the resulting N-acyl derivatives of aminopentol likely serve as CerS inhibitors as seen with N-palmitoyl and N-nervonoyl derivatives of aminopentol.^51,52

Figure 7.

Structures of fumonisin B1 (15) and aminopentol (16).

Fumonisin B1 (15) has been widely used to study effects of CerS inhibition on the sphingolipid metabolism and cell viability.^9,10 Fumonisin B1 (15) is cytotoxic to various mammalian cell lines, at least in part due to its lack of isoform selectivity.⁵³ This has hindered efforts to assess the therapeutic potential of selectively targeting each isoform.

FTY720 (17), also known as fingolimod, is a sphingosine-like molecule approved by FDA for the treatment of the relapsing form of multiple sclerosis (Figure 8). After enantiospecific monophosphorylation, the resulting metabolite 18 acts as a functional antagonist of sphingosine-1-phosphate (S1P) receptors in vivo. Interestingly, FTY720 (17) itself was found to inhibit N-acylation of dihydrosphingosine with docosanoyl-CoA with an IC₅₀ value of 6.4 μM in the human pulmonary artery endothelial cell (HPAEC) lysate.⁵⁴ It appears to act as a competitive inhibitor with respect to dihydrosphingosine with a K_i value of 2.15 μM for CerS2 when C22-CoA was used as an acyl donor. A different research group, however, reported more complex inhibition kinetics of FTY720 (17) using homogenates prepared from HEK cells overexpressing CerS4, namely, noncompetitive and uncompetitive inhibition toward C18-CoA and sphinganine, respectively.⁵⁵ Yet another group showed that FTY720 (17) is competitive with respect to fluorescent NBD-sphinganine and noncompetitive with respect to C16:0-CoA.⁵⁶

Figure 8.

Structures of FTY720 (17) and its derivatives 18–24.

The (R)- and (S)-enantiomers of 2-amino-4-(4-heptylox-yphenyl)-2-methylbutanol (AAL) are derivatives of FTY720 (17) in which the benzylic carbon was replaced by an ether oxygen and one of the two prochiral hydroxymethyl groups is replaced by a methyl group (Figure 8).⁵⁷ It was reported that (R)-AAL (19) but not (S)-AAL (20) is phosphorylated by recombinant mouse sphingosine kinase 1a.⁵⁷ This finding is consistent with the enantiospecificity of sphingosine kinases displayed toward FTY720 (17). Interestingly, the nonphosphorylatable enantiomer (S)-AAL (20) was subsequently found to inhibit C16:0 CerS activity of HEK293 lysates (∼50% inhibition at 10 μM) in a competitive manner with respect to NBD-sphinganine.⁵⁶ None of the three additional analogs 21–23 (Figure 8) showed substantial CerS inhibitory activity at 10 μM, suggesting that both the amino group and the long carbon chain of (S)-AAL (20) are essential for the potent inhibitory activity.⁵⁶

Investigation on the isoform selectivity of (S)-AAL (20) revealed that it inhibits CerS1 potently (>90% inhibition at 10 μM) and, to a lesser extent, CerS5 and CerS6 (nearly 50% inhibition at 10 μM).⁵⁸ Although benzyloxy analog 24 (Figure 8) was found to be a somewhat weaker CerS1 inhibitor (76% inhibition at 10 μM), it showed no inhibitory activity toward CerS2 and CerS4–6, representing one of the first isoform-selective CerS inhibitors. It should be noted that (S)-AAL (20) and compound 24 (at 10 μM) have CerS1 inhibitory activity–cytotoxicity ratios of 0.2 and 3.6, respectively. The results suggest the possibility of differentiating CerS1 inhibitory activity from cytotoxicity, an important step forward for the future therapeutic development of CerS inhibitors.

In another study, N-alkyl and N-acyl derivatives 25–28 (Figure 9) were profiled for their effects on ceramide synthase activity of cell lysates and live cells by measuring the levels of C14–C24 ceramides or dihydroceramides.⁵⁹ For instance, compounds 25 inhibited C16-, C18-, and C24-dihyroceramide synthesis with IC₅₀ values of 25.7, 50.7, and 60.6 μM, respectively, in a microsomal fraction of HCT-116 cells. Subsequent experiments using a cell lysate of CerS2, CerS4, or CerS6 overexpressing HT-116 cells indicated that compound 25 preferentially inhibits CerS2 and CerS4.

Figure 9.

Structures of CerS inhibitors 25–29 derived from FTY720 (17).

More recently, P053 (29), a dichloro analog of 24, was reported as a potent and selective inhibitor of CerS1 (Figure 9).⁶⁰ P053 (29) inhibited human and murine CerS1 (hCerS1 and mCerS1) with IC₅₀ values of 0.54 and 0.46 μM, respectively, while it showed substantially lower inhibitory potency against hCerS2, mCerS2, hCerS4, mCerS5, and hCerS6 (IC₅₀ = 7–30 μM). The inhibition of CerS1 by P053 (29) appears to be noncompetitive with respect to either sphinganine or C18 fatty acyl-CoA. Consistent with its in vitro isoform selectivity, P053 (29) selectively reduced C18 ceramide and other C18 sphingolipids in HEK293 cell culture. P053 (29) is orally available in mice and reduced C18 ceramide levels in skeletal muscles of mice on a high-fat diet, leading to the enhancement of fatty acid oxidation and reduction in overall adiposity. These findings demonstrate the therapeutic potential of CerS1 inhibition for the treatment of obesity.

4. SPHINGOMYELINASE

Sphingomyelin, the most abundant form of sphingolipids, is a non-glycerol-based phospholipid in which the primary alcohol of ceramide is linked to choline through a phosphodiester bond.^61,62 Sphingomyelin is synthesized from ceramide by sphingomyelin synthase in the Golgi apparatus. Because of varying chain length (C14 to C26) of the fatty acid moiety of the ceramide molecule, sphingomyelin occurs in a number of structurally different forms. Upon synthesis, sphingomyelin is exported to other membrane systems. Indeed, sphingomyelin constitutes a significant fraction of the total lipid composition of the plasma and lysosomal membranes.⁶³ Membrane sphingomyelin can be hydrolyzed back to ceramide via hydrolysis of the phosphodiester catalyzed by sphingomyelinase, thereby serving as a reservoir for ceramide.⁶⁴ Sphingomyelinase can be divided into three evolutionarily unrelated classes based on their pH optima: acid sphingomyelinase (aSMase), neutral sphingomyelinase (nSMase), and alkaline sphingomyelinase. No potent inhibitors have been reported for alkaline sphingomyelinase to date, and thus this topic will not be covered in this review. On the other hand, tremendous efforts have been made to identify potent inhibitors of aSMase and nSMase, in part driven by the therapeutic potential of targeting these enzymes for the treatment of various diseases, including cancer, neurodegenerative diseases, major depression disorder, pulmonary inflammation, atherosclerosis, and viral infections.

4.1. Acid Sphingomyelinase.

Two forms of aSMase, both encoded by the SPMD1 gene, have been known to exist as a result of two distinct post-translational trafficking pathways within the ER-Golgi network.¹³ Lysosomal aSMase is transported to the endolysosomal compartment and anchored to the inner lysosomal membrane. Secretory aSMase is transported to the outer leaflet of the plasma membrane and secreted into the extracellular space under certain conditions.

Clinical significance of aSMase has been well recognized as some forms of Niemann–Pick disease, a lysosomal storage disorder, were found to be caused by loss-of-function mutations in the SMPD1 gene.⁶⁵ Upregulation of aSMase has also been implicated in a number of diseases including several neurodegenerative disorders where excess ceramide is believed to play a pathogenic role.¹² As described later, these findings have spurred efforts to develop aSMase inhibitors as therapeutic agents.

Several crystal structures have been reported for mammalian aSMase.^66–68 Two distinct conformations were observed in the crystal structures of murine aSMase depending on the folding and positioning of the membrane-interacting N-terminal saposin domain.⁶⁶ A closed globular conformation (PDB code 5FI9) is believed to be predominant in the absence of membranes and render the enzyme inactive (Figure 10A), while an open conformation (PDB code 5FIB) is expected to establish an interface with the catalytic domain essential for activity (Figure 10B). The active site carries a binuclear zinc complex, playing a central role in the catalytic process at the core of the active site. A hydrophobic track that extends from the edge of the active site to the saposin domain is predicted to accommodate the fatty acid chains of sphingomyelin. aSMase is positively charged at acidic lysosomal pH and is tightly associated with lysosomal membrane surfaces, presumably through the positively charged residues in the saposin domain. This formed the mechanistic basis for indirect inhibition of aSMase by cationic amphiphilic molecules, which constitute one of the two major classes of aSMase inhibitors, the other being those directly interacting with the enzyme.

Figure 10.

Crystal structures of (A) the closed form (PDB code 5FI9) and (B) the open form (PDB code 5FIB) of murine aSMase.⁶⁶ The saposin domain is shown in blue, and the active site zinc ions are shown as gray spheres.

AY-9944 (30) (Figure 11), originally reported as an inhibitor of cholesterol biosynthesis,⁶⁹ was later found to recapitulate the pathological features similar to those found in Niemann–Pick disease in rats.⁷⁰ Furthermore, it was found that aSMase activity was significantly reduced in the tissues of rats treated with AY-9944 (30), although it failed to inhibit aSMase activity in a cell-free system.⁷¹ Tricyclic antidepressants, imipramine (31) and desipramine (32) (Figure 11), were also reported to decrease aSMase activity in murine neuroblastoma and human fibroblast cell cultures.⁷² Interestingly, iminodibenzyl (33) (Figure 11) containing the same tricyclic core but devoid of the side chain showed no effect on aSMase activity, indicating the essential role of the cationic amphiphilic characteristics of these molecules. Additional cationic amphiphilic drugs were subsequently found to reduce aSMase activity in cell cultures.^73–76 It is postulated that these molecules accumulate in lysosomes as positively charged species and compete for the aSMase binding site at the inner lysosomal membrane, resulting in a detachment and proteolytic degradation of the enzyme.^76–78 These indirect aSMase inhibitors are now collectively referred to as functional inhibitors of acid sphingomyelinase (FIASMAs).⁷⁸ It should be noted that many FIASMAs are FDA approved for clinical use. For instance, the clinical benefit of amitriptyline (34) (Figure 11) was investigated in patients with cystic fibrosis, in which ceramide is known to accumulate in bronchial epithelial cells.⁷⁹ While FIASMAs offer a unique mechanism to indirectly reduce aSMase activity, their safety profile remains to be seen given that many of these molecules act on various CNS targets.

Figure 11.

Structures of representative functional inhibitors of aSMase (FIASMAs).

Screening of tropical plant extracts and microbial secondary metabolites using a crude preparation of aSMase from the bovine brain led to the discovery of some xanthone-based molecules as aSMase inhibitors (Figure 12).⁸⁰ α-Mangostin (35), cowanol (36), and cowanin (37) inhibited aSMase with IC₅₀ values of 14.1, 10.9, and 19.2 μM, respectively. These inhibitors differ structurally from FIASMAs in that they lack a basic nitrogen atom necessary for amphiphilic properties. Indeed, all of the three compounds inhibited aSMase in a competitive manner with respect to sphingomyelin, suggesting direct interaction with the enzyme. α-Mangostin analog 38 containing saturated side chains retained the inhibitory activity (IC₅₀ = 10.9 μM), while diacetyl analog 39 showed no inhibitory activity, highlighting the important role played by the two phenolic OH groups of α-mangostin (35) and its derivatives.

Figure 12.

Structures of xanthone-based aSMase inhibitors 35–39.

Using the γ-pyrone fragment of the xanthone-based inhibitors discussed above as the branching point of origin, a collection of 500 γ-pyrones spanning three hierarchy levels in the branch were assembled for screening.⁸¹ Two potent compounds 40 and 41 containing a benzopyran scaffold (Figure 13) were identified as potent inhibitors of aSMase prepared from rat brain homogenates with IC₅₀ values of 3.1 and 9.5 μM, respectively. These compounds displayed no inhibitory activity toward nSMase at concentrations up to 50 μM.

Figure 13.

Structures of γ-pyrone-based aSMase inhibitors 40 and 41.

Epicatechin-3-O-gallate (42) (ECg, Figure 14) was reported to inhibit secretory SMase, a secreted form of aSMase.⁸² Further investigation identified a synthetic catechin, 3′-O-methylepigallocatechin-3-O-gallate (43) (EGCg-3′-O-Me, Figure 14), as a low micromolar inhibitor of secretory SMase from rat plasma with an IC₅₀ value of 1.7 μM. Secretory SMase was found to be activated by oxidation and inactivated by reduction. Thus, these compounds are believed to act as reducing agents toward the enzyme to achieve its inhibition. Indeed, EGCg-3′-O-Me displayed noncompetitive inhibition when C6 NBD sphingomyelin was used as a substrate, indicating the presence of an allosteric site targeted by the epicatechin-based compounds.

Figure 14.

Structures of catechin-based secretory SMase inhibitors 42 and 43.

Weak aSMase inhibitory activity observed for phosphatidylinositol-4,5-bisphosphate (PtdIns4,5P2) against aSMase⁸³ prompted exploration of other phosphatidylinositol derivatives including PtdIns3,5P2 (44) (Figure 15).⁸⁴ PtdIns3,5P2 (44) potently inhibited a recombinant human aSMase (>90% inhibition at 5 μM) in a micellar assay system. It displayed a noncompetitive mode of inhibition with a K_i value of 0.53 μM and showed no inhibitory activity against rat brain microsomal nSMase at concentrations of up to 50 μM. Subsequently, racemic sulfonate derivatives 45–49 with alkyl groups of varying length (Figure 15) were synthesized in an attempt to remove the polar phosphate moiety of PtdIns3,5P2 (44).⁸⁵ Compound 49 containing a dodecyl group was found to be most potent against aSMase prepared from rat brain homogenates with an IC₅₀ value of 0.90 μM, comparable to that of PtdIns3,5P2 (44) (IC₅₀ = 0.93 μM). Unlike PtdIns3,5P2 (44), however, compound 49 acted as a competitive inhibitor with a K_i value of 0.42 μM, suggesting a mode of binding distinct from that of PtdIns3,5P2 (44) despite common structural features shared by the two molecules.

Figure 15.

Structures of inositol-4,5-bisphosphate-based aSMase inhibitors 44–49.

α-Aminobisphosphonate 50 (Figure 16) was identified as a potent aSMase inhibitor (IC₅₀ = 0.04 μM) through screening of a bisphosphonate compound library using rat brain homogenates as a source of the enzyme.⁸⁶ Additional compounds were synthesized in order to gain further insight into the SAR of bisphosphonate derivatives. Among them, compounds 51–53 (Figure 16) exhibited IC₅₀ values below 100 nM. None of these compounds inhibited nSMase of rat brain homogenates.

Figure 16.

Structures of bisphosphonate-based aSMase inhibitors 50–53.

Cocrystal structure of murine aSMase with compound 51 (PDB code 5FI9) shows that one of the two phosphonate groups in the inhibitor interacts with both zinc atoms in the active site (Figure 17).⁶⁶ Compound 51 appears to act as a transition state analog inhibitor by mimicking the intermediate formed by a nucleophilic attack of a water molecule to the phosphodiester of sphingomyelin. Surprisingly, the nine-carbon chain of compound 51 does not seem to bind to the hydrophobic track of aSMase, a putative binding site for the alkyl chains of sphingomyelin.

Figure 17.

Crystal structure of the murine aSMase active site in complex with compound 51 (PDB code 5FI9).⁶⁶ Key residues are shown as a green stick model, zinc ions are shown as gray spheres, and compound 51 is shown as a cyan stick model.

α-Substituted serinol derivatives 54–57 were reported as aSMase inhibitors in a patent application (Figure 18).⁸⁷ The most potent compound 57 containing a 1-nonyl-1H-1,2,3-triazol-4-yl group inhibited aSMase with an IC₅₀ value of 1.14 μM in a biochemical assay. It is conceivable that the two primary hydroxyl groups of these compounds serve as alternatives to the highly polar phosphonate moieties of 50–53. Indeed, compound 57 was found to be orally available in mice and distributed to the brain. Furthermore, compound 57 was found to normalize the enhanced brain aSMase activity in APP/PS1 mice to that of wild type.

Figure 18.

Structures of serinol-based aSMase inhibitors 54–57.

By use of a pharmacophore model developed from the previously reported xanthone-based aSMase inhibitors, hydroxamate-containing compound 58 (Figure 19) was identified as a submicromolar competitive inhibitor of human aSMase from the supernatants of Huh7 cell lysates with an IC₅₀ value of 0.48 μM.⁸⁸ Compound 58 was also found to dose-dependently reduce generation of ceramide and LPS-induced inflammation in human NIH3T3 cells. Subsequent efforts guided by molecular docking led to the discovery of compound 59 devoid of a flexible alkyl chain (Figure 19).⁸⁹ Compound 59 inhibited aSMase from the supernatants of Huh7 cell lysates with an IC₅₀ value of 0.32 μM. In Sprague-Dawley rats, compound 59 showed good CNS permeability and reduced aSMase activity in cerebral cortex and hippocampus following intraperitoneal injection. Neither 58 nor 59 inhibited nSMase at concentrations up to 100 μM.

Figure 19.

Structures of hydroxamate-based aSMase inhibitors 58 and 59.

4.2. Neutral Sphingomyelinase.

As indicated by its name, the catalytic phosphodiesterase activity of nSMase is optimal at neutral pH. In mammals, four distinct genes, SMPD2–5, have been identified to date, encoding nSMase1–3^90–92 and mitochondria-associated nSMase,⁹³ respectively. nSMase1–3 show little sequence homology while substantial sequence homology (50.6% similarity) exists between nSMase2 and mitochondria-associated nSMase.^93,94 These isoforms differ in tissue distribution and subcellular localization, suggesting distinct physiological and pathogenic roles.⁹⁵

Similar to aSMase, the therapeutic utility of nSMase inhibition has gained increasing attention as another strategy to regulate ceramide levels in various disease conditions. Most efforts have focused on nSMase2, which is the most extensively studied isoform in terms of its potential as a therapeutic target in a number of disease areas.

nSMase2 contains a hydrophobic N-terminal domain (NTD) and a C-terminal catalytic domain (CAT). The NTD is a lipid-binding domain that interacts with phosphatidylserine (PS), which is required for its full enzymatic activity. The CAT is interrupted by a large insertion (175–339), and the crystal structure of the human nSMase2 CAT devoid of the insertion (residues 117–651 Δ175–339) was recently determined at 1.85 Å resolution (PDB code 5UVG).⁹⁶ The structure reveals a deoxyribonuclease I (DNase I)-like fold as expected from the structural and catalytic similarity between the two enzymes (Figure 20A). Its active site contains conserved residues within DNase I-like superfamily involved in Mg²⁺ binding and catalysis. The hydrophobic track stretching from the active site is believed to be the binding site for the alkyl chains of sphingomyelin. In this structure, however, an evolutionarily conserved motif termed “DK switch” is positioned such that it obstructs the active site entrance (Figure 20B) and places the conserved Asp430 away from the active site. This observation suggests that “DK switch” adopts different conformations to regulate the enzymatic activity. It was subsequently postulated that PS binding triggers conformational change of “DK switch”, resulting in the formation of the active enzyme.⁹⁷ These structural insights should serve as a guide for facilitating the rational design of nSMase inhibitors.

Figure 20.

Crystal structure of human nSMase2 CAT (PDB code 5UVG).⁹⁶ (A) Ca²⁺ shown as a gray sphere occupies the position of the primary Mg²⁺ ion required for the Mg²⁺-dependent activity of nSMase2. DK switch is shown in blue, and the conserved Asp430 is shown as a purple stick model. (B) Close-up view of the substrate binding site. DK switch shown in blue obstructs the entrance to the hydrophobic track stretching from the active site (painted in yellow).

Scyphostatin (60), isolated from Trichopeziza mollissima, represents the first low-micromolar inhibitor of nSMase.⁹⁸ As shown in Figure 21, it bears some structural resemblance to ceramide in that they both contain the N-acylamino alcohol scaffold.⁹⁹ Scyphostatin (60) is a far more potent nSMase inhibitor than other naturally occurring compounds previously reported^100,101 with an IC₅₀ value of 1.0 μM using rat brain microsomes as an enzyme source.¹⁰² It showed 50-fold selectivity over aSMase and no inhibitory activity against Staphylococcus aureus and Bacillus cereus SMases.¹⁰² In preliminary kinetics studies, scyphostatin (60) displayed mixed-type inhibition with respect to sphingomyelin. The presence of the epoxide moiety suggests the possibility of covalent bond formation between the inhibitor and the enzyme though the precise mechanism of inhibition has not yet been elucidated.

Figure 21.

Structures of scyphostatin 60 and its derivatives 61–66.

Subsequent to the discovery of scyphostatin (60), several analogs have been evaluated as nSMase inhibitors (Figure 21). Compound 61 containing a spiroexpoxide moiety exhibited 80% inhibition of nSMase activity in rat brain microsomes following 90 min preincubation at 200 μM.^103,104 It was found that inhibition by compound 61 was time dependent, suggestive of a covalent modification of the enzyme.¹⁰³ Compounds 62 and 63 displayed substantially lower inhibitory activity compared to 61, demonstrating the crucial role played by the primary hydroxyl group in the enzyme inhibition.¹⁰⁴ Because of its structural similarity to scyphostatin (60) and compound 61, manumycin A (64) was evaluated for its ability to inhibit nSMase using a partially purified enzyme from rat brain microsomes.¹⁰⁵ Manumycin A (64) was, indeed, found to inhibit nSMase in a time-dependent manner (34% inhibition without preincubation and 79% inhibition following 60 min preincubation at 100 μM). Its truncated analog 65 displayed enhanced inhibitory activity (90% inhibition without preincubation and 98% inhibition following 60 min preincubation at 100 μM), while analog 66 (racemic) containing a saturated acyl group showed weaker activity (10% inhibition without preincubation and 88% inhibition following 60 min preincubation at 100 μM). Although these analogs were not directly (head-to-head) compared to scyphostatin, it appears that none of them exceeded the inhibitory potency achieved by scyphostatin (60).

Alutenusin (67) (Figure 22), originally isolated from Alternaria sp., was identified as an nSMase inhibitor through microbial screening.¹⁰⁶ Alutenusin (67) containing a biphenyl scaffold is one of the first nSMase inhibitors structurally distinct from sphingolipids and was found to inhibit nSMase from rat brain microsomes with a K_i value of 20 μM. As expected from its non-sphingolipid-like structure, alutenusin (67) displayed a noncompetitive inhibition with respect to sphingomyelin. Furthermore, it showed no inhibitory activity against aSMase from rat brain microsomes at concentrations up to 950 μM.

Figure 22.

Structure of alutenusin (67).

Hydroquinones 68–70 (Figure 23), isolated from Acre-monium murorum, were reported to inhibit nSMase of rat brain microsomes with IC₅₀ values of 7.2 μg/mL (34 μM), 3.6 μg/mL (17 μM), and 3.2 μg/mL (16 μM), respectively.¹⁰⁷ Compound 71 (Figure 23), the corresponding quinolone derivative of 69, was also found to potently inhibit nSMase of rat brain microsomes with a IC₅₀ value of 0.8 μg/mL (4.2 μM).¹⁰⁸ Another hydroquinone-based compound, ubiquinol (72) (CoQ₁₀H₂, Figure 23), was reported to inhibit Mg²⁺-dependent nSMase of pig liver plasma membranes (>90% inhibition at 100 μM following 15 min preincubation) in a noncompetitive manner with respect to sphingomyelin.¹⁰⁹ Subsequent studies releveled that inhibition by ubiquinol (72) is time-dependent, displaying a greater degree of potency when preincubated for 60 min.¹¹⁰ Furthermore, CoQ₆H₂ (73) (Figure 23) was found to have the optimal isoprenoid side chain length for nSMase inhibition among several homologs.¹¹⁰

Figure 23.

Structures of benzoquinol- and benzoquinone-based nSMase inhibitors 68–73.

In an attempt to mimic the phosphodiester moiety of sphingomyelin, some phosphorus-based compounds have also been examined as nSMase inhibitors. After some early efforts resulting in identification of weak nSMase inhibitors,^111,112 compound 74 (Figure 24) was found to inhibit nSMase of bovine brain microsomes with an IC₅₀ value of 3.3 μM.¹¹³ It should be noted that the stereochemistry of 74 is opposite that of sphingomyelin at both of the two chiral centers. Surprisingly, compound 75 possessing the same stereochemistry as sphingomyelin (Figure 24) showed much weaker inhibitory potency (IC₅₀ = 181 μM). These findings indicate that compound 74 does not act as a substrate-based inhibitor despite its structural similarity to sphingomyelin. Indeed, compound 74 displayed noncompetitive inhibition with respect to sphingomyelin with a K_i value of 1.6 μM. Consistent with its mode of inhibition, compound 74 showed no inhibitory effect on SMase from B. cereus even though it shares common structural and mechanistic features in the catalytic domain with mammalian nSMase.

Figure 24.

Structures of phosphonate-based nSMase inhibitors 74 and 75.

Some sphingomyelin analogs possessing a carbamate moiety as a replacement for the phosphodiester group were identified as low-micromolar nSMase inhibitors (Figure 25).¹¹⁴ By use of rat brain microsomes as the enzyme source, carbamates 76 and 77 were found to inhibit nSMase with IC₅₀ values of 2.8 and 1.8 μM, respectively. SAR studies indicate that only a narrow range of acyl groups are tolerated at the amino group. Both N-acetyl and N-stearoyl derivatives 78 and 79 showed no inhibitory activity at concentrations up to 100 μM.

Figure 25.

Structures of carbamate-based nSMase inhibitors 76–79.

Ceramide derivatives 80 and 81 containing a thiourea moiety (Figure 26) were reported to inhibit nSMase semipurified from bovine brain microsomes with low micromolar potency in a time-dependent manner (∼90% inhibition after 60 min preincubation).¹¹⁵ Compounds 80 and 81 displayed competitive inhibition with respect to sphingomyelin with K_i values of 1.7 and 2.5 μM, respectively, suggesting direct binding of these ceramide derivatives to the active site of the enzyme. Assuming that the primary hydroxyl group of these inhibitors corresponds to that of ceramide, the stereochemistry of the amino-attached (R)-chiral carbon appears opposite to that of ceramide. The corresponding (S)-derivatives of 80 and 81, however, were not explored in this report.

Figure 26.

Structures of thiourea-based nSMase inhibitors 80 and 81.

Ceramide derivatives containing a lactone moiety in their sphingosine component were found to inhibit nSMase from rat brain microsomes (Figure 27).¹¹⁶ Compounds 82 and 83, the so-called sphingolactones, achieved nearly 50% and 90% inhibition following 15 min of preincubation at 350 μM. It is speculated that the time-dependent inhibition resulted from irreversible covalent binding of the lactone carbonyl moiety to the enzyme.

Figure 27.

Structures of γ-lactone-based nSMase inhibitors 82 and 83.

Aminoguanidine derivatives C11AG (84) and 85 (Figure 28) were reported to inhibit nSMase activity of rat brain microsomes with IC₅₀ values of 8.2 and 5 μM, respectively.¹¹⁷ Although C11AG (84) is less potent than compound 85, it displayed superior antiviral potency against HSV-1 in Rita cells with a favorable therapeutic index. C11AG (84) showed 30-fold selectivity over aSMase and no inhibitory activity toward other phospholipases including phospholipase A2, phospholipase D, and phosphatidylcholine-specific phospholipase C. In subsequent studies, C11AG (84) was found to suppress LPS-stimulated sphingomyelin degradation and ceramide synthesis in RAW cells, resulting in suppression of NF-κB liberation.¹¹⁸ Another aminoguanidine derivative, ES048 (86) (Figure 28), was reported to inhibit nSMase activity in mouse splenocytes though it exhibited a U-shaped dose–response curve with the maximal effect (∼80% inhibition) at 1 μM.¹¹⁹ It showed no inhibitory effects on either aSMase from mouse splenocytes or recombinant human aSMase.

Figure 28.

Structures of aminoguanidine-based nSMase inhibitors 84–86.

Some N-dialkylated amino acids were reported as low micromolar nSMase inhibitors (Figure 29).¹²⁰ Among the compounds tested, serine and alanine derivatives 87 and 88 were most potent with IC₅₀ values of 1.8 and 2.8 μM, respectively, using U937 cell lysate as a source of the enzyme. Preliminary SAR studies indicated that the carboxyl group and the propenyl (−C=CH–CH₂−) linkers are essential for the potent inhibitory activity.

Figure 29.

Structures of α-amino acid-based nSMase inhibitors 87 and 88.

GW4869 (89) (Figure 30) was identified by a high throughput assay using a partially purified and delipidated rat brain nSMase.¹²¹ GW4869 (89) is a noncompetitive inhibitor with respect to sphingomyelin with an IC₅₀ value of 1 μM. GW4869 (89) showed no inhibitory activity against human aSMase. Interestingly, PS, a known endogenous activator of nSMase,¹²² was found to reduce the inhibitory potency of GW4869 (89). The two substances, however, were unlikely to compete for the same binding site since a PS-independent B. cereus SMase was also inhibited by GW4869 (89) with similar potency. GW4869 (89) significantly inhibited TNF-induced sphingomyelin hydrolysis in a dose-dependent manner in MCF7 cells, while it showed no effects on the de novo ceramide biosynthetic pathway.¹²² Since its discovery, GW4869 (89) has been by far the most widely used prototype nSMase2 inhibitor for studying the physiological and pathological roles of nSMase. Indeed, as an in vivo probe molecule, GW4869 (89) has demonstrated the therapeutic utility of nSMase2 inhibition in a variety of animal disease models, including Alzheimer’s disease,^123,124 traumatic brain injury,¹²⁵ lung injury,¹²⁶ metabolic disease,¹²⁷ heart failure,¹²⁸ and cancer.^129,130 Despite its proven track record as a tool compound, systematic structural optimization efforts on GW4869 (89) has not yet been reported, presumably due to its high lipophilic core scaffold contributing to poor aqueous solubility.

Figure 30.

Structures of GW4869 (89).

Various 4H-1,2,4-triazole-3(2H)-thione derivatives were reported as low micromolar nSMase inhibitors (Figure 31). Compounds 90 and 91 inhibited mouse nSMase with IC₅₀ values of 2.5 and 2.4 μM, respectively.¹³¹ The same group also reported a series of 3,4-dihydropyrimidine-2(1H)-thione derivatives including 92 and 93 (Figure 31) as nSMase inhibitors with IC₅₀ values of 0.9 and 0.86 μM, respectively. More recently, cambinol (94) (Figure 31), a 2-thioxo-2,3-dihydropyrimidin-4(1H)-one derivative originally reported as a SIRT1/2 inhibitor,¹³² was identified as a nSMase2 inhibitor from screening assays using the recombinant human enzyme.¹³³ Cambinol (94) displayed uncompetitive inhibition with respect to sphingomyelin with a K_i value of 7 μM. It also inhibited nSMase activity of B. cereus and rat brain homogenate with IC₅₀ values of 5 and 6 μM, respectively. Furthermore, cambinol was found to block TNF-α induced increase in ceramide levels in the rat primary neurons. It should be noted that inhibitors 90–94 possess the cyclic thiourea moiety as the common structural feature. In particular, it is conceivable that the pyrimidine-based core rings of compounds 92–94 play a similar role in binding to the enzyme. Cambinol (94) was independently identified as a tau propagation inhibitor from cell-based functional assays.¹³⁴ Molecular docking and molecular dynamic simulation studies using recently published crystal structure of nSMase2 CAT⁹⁶ indicate that it binds to nSMase2 at the DK switch and inhibits the enzyme by directing Asp430 away from the active site.¹³⁴

Figure 31.

Structures of nSMase inhibitors 90–94 containing a 1,2,4-triazole-3-thione or pyrimidine-2-thione ring.

2,6-Dimethoxy-4-(5-phenyl-4-thiophen-2-yl-1H-imidazol-2-yl)phenol (95) (DPTIP, Figure 32) was identified as a potent inhibitor of human nSMase2 through a high throughput screening of >365 000 compounds from the Molecular Libraries Small Molecule Repository (MLSMR) and the NCGC Pharmaceutical Collection (NPC).¹³⁵ DPTIP (95) inhibited a human recombinant nSMase2 with an IC₅₀ value of 30 nM in a noncompetitive manner with respect to sphingomyelin. DPTIP (95) inhibited release of extracellular vesicles from the rat primary astrocyte cultures in a dose dependent manner. Furthermore, DPTIP (95) attenuated IL-1β-induced astrocyte-derived extracellular vesicle (ADEV) release in GFAP-GFP mice following intraperitoneal administration. Subsequent medicinal chemistry efforts identified 4-(1H-imidazol-2-yl)-2,6-dimethoxyphenol as a key pharmacophore essential for the potent nSMase2 inhibition, and several derivatives containing this pharmacophore exhibited inhibitory potency comparable to that of DPTIP, including compounds 96–99 (Figure 32).¹³⁶

Figure 32.

Structures of nSMase inhibitors 95–99 based on the 4-(1H-imidazol-2-yl)-2,6-dialkoxyphenol scaffold.

Compound 100 containing an imidazo[1,2-b]pyridazine ring is another inhibitor identified from screening efforts using a human recombinant nSMase2.^137,138 Preliminary SAR studies revealed that the imidazo[1,2-b]pyridazin-8-amine scaffold is essential for the potent inhibition as compounds possessing other 6,5-fused ring scaffolds failed to show any inhibitory activity. On the other hand, modifications of the pyrrolidine ring of 100 led to the discovery of phenyl (R)-(1-(3-(3,4-dimethoxyphenyl)-2,6-dimethylimidazo[1,2-b]pyridazin-8-yl)-pyrrolidin-3-yl)carbamate (101) (PDDC) with improved potency (IC₅₀ = 0.3 μM).¹³⁷ It displayed a noncompetitive inhibition with respect to sphingomyelin with a K_i value of 0.3 μM. Consistent with its noncompetitive mode of inhibition, PDDC (101) showed no inhibitory activity against aSMase and two other phosphodiesterases (PDE3A and PDE4D2). It was found to be metabolically stable and orally available in mice with excellent CNS permeability. As seen with DPTIP, but by oral administration, PDDC (101) inhibited ADEV release in a dose dependent manner in GFAP-GFP mice.¹³⁷ It was also reported to reverse cognitive impairment in 5XFAD mice following intraperitoneal administration.¹³⁸ Subsequently, extensive structure–activity relationship (SAR) studies were conducted using PDDC as a molecular template, providing additional potent nSMase2 inhibitors (Figure 33) within this structural series including compounds 102 (IC₅₀ = 0.1 μM) and 103 (IC₅₀ = 0.07 μM).¹³⁸

Figure 33.

Structures of nSMase inhibitors 100–103 containing an imidazo[1,2-b]pyridazine ring.

Targeting both nSMase2 and acetylcholinesterase (AChE) may lead to a new therapeutic approach to Alzheimer’s disease since it could not only suppress exosome biogenesis but also enhance cholinergic synaptic plasticity. To this end, a ∼70-compound library largely comprising phenserine analogs and other known AChE inhibitors was screened for nSMase2 inhibitory activity.¹³⁹ One hit from the screening, compound 104 (Figure 34), was validated as a weak inhibitor of nSMase2 (60% inhibition at 50 μM). Subsequent hit-to-lead optimization led to the discovery of compounds 105 and 106 (Figure 34) displaying nearly 100-fold more potent inhibitory activity. As anticipated from their phensvenine-like structures,¹⁴⁰ compounds 105 and 106 were also found to inhibit AChE with IC₅₀ values of 7 and 1.7 μM, respectively. Both compounds displayed an uncompetitive mechanism of inhibition with respect to sphingomyelin. Consistent with these findings, molecular docking analysis using a crystal structure of the nSMase2 CAT (PDB code 5UVG) indicates that both compounds bind to nSMase2 at the DK-switch and stabilize its inactive conformation¹³⁹ in a manner similar to what was proposed for cambinol (94).¹³⁴ Both compounds were reported to suppress the release of tau-bearing extracellular vesicles (EVs) in cell-based assays, and compound 106 significantly reduced IL1β-induced release of tau-carrying exosomes in tau P301S (PS19 line) mice.¹³⁹

Figure 34.

Structures of phensvenine-based nSMase inhibitors 104–106.

4.3. Bacterial Sphingomyelinase.

There are two different types of bacterial SMase depending on the phosphodiester cleaving site.¹⁶ Like human SMase, sphingomyelinase C (SMaseC) catalyzes the hydrolysis of the phosphoester linkage between ceramide and phosphocholine. Sphingomyelinase D, on the other hand, hydrolyzes the other phosphoester bond, resulting in the release of ceramide-1-phosohate and choline. Given the primary focus of this review article on ceramide-producing enzymes, only SMaseC inhibitors will be discussed herein. SMaseC can be found in many species of bacteria, including S. aureus, B. cereus, Clostridium perfringens, and Leptospira interrogans.¹⁶ It appears that SMaseC is evolutionally related to mammalian neutral sphingomyelinases as they share a number of conserved residues in the catalytic region and adopt the same fold as DNase I, another metal-dependent phosphodiesterase. Indeed, mammalian nSMase1 and nSMase2 were cloned based on their sequence homology to bacterial SMaseC.^90,91 SMaseC, however, differs from the mammalian homologs in that it is a secretory protein that can target the host cell plasma membrane. For instance, increase in ceramide caused by SMaseC is known to alter the physical properties of the host cell plasma membrane and, possibly, contribute to the pathogenesis of some infectious diseases.¹⁶ In fact, antibodies against SMaseC, but not those against phosphatidylcholine and phosphatidylinositol-specific phospholipases C, were found to protect mice from a lethal dose of B. cereus,¹⁴¹ underscoring the potential therapeutic utility of SMaseC inhibitors.

Prior to the elucidation of human nSMase2 crystal structure,⁹⁶ a number of crystal structures have been solved for SMaseC.^142–145 The first crystal structure of SMaseC was determined for Listeria ivanovii SMaseC (SmcL) at 1.9 Å resolution (PDB code 1ZWX).¹⁴² Like human nSMase2, SmcL adopts a DNase I-like protein fold and possesses a number of key residues in the active site conserved among bacterial SMases and eukaryotic nSMases (Figure 35A), suggesting a common catalytic mechanism for binding and hydrolysis of sphingomyelin. The structure also revealed a large hydrophobic β-hairpin and hydrophobic loop surrounding the active site. In the absence of identifiable membrane-spanning regions, these segments unique to bacterial SMases are thought to play an important role in mediating protein–host membrane interactions to guide the phosphodiester moiety of sphingomyelin into the active site (Figure 35B). It should be noted that, unlike the human nSMase2 CAT structure (PDB code 5UVG) discussed earlier, the DK switch in this structure forms a short α-helix that directs the conserved Asp160 residue into the active site, conceivably representing the active conformation of the enzyme.

Figure 35.

(A) Overall structure of SmcL (PDB code 1ZWX).¹⁴² The hydrophobic β-hairpin and the hydrophobic loop are shown in green and red, respectively. Phosphate ion shown as an orange/red stick model is likely located in the position occupied by the phosphate moiety of sphingomyelin in the enzyme–substrate complex. DK switch shown in blue forms a short α-helix and directs the conserved Asp160 residue shown as a purple stick model into the active site. (B) Schematic diagram of SmcL-catalyzed hydrolysis of sphingomyelin into ceramide on the outer leaflet of the plasma membrane (adapted from ref 142).

As could be expected from the close evolutionary relationship between the two classes of enzymes, some mammalian nSMase inhibitors, including GW4869 (89) and cambinol (94), were also found to inhibit B. cereus SMaseC as described earlier.^122,133 Conversely, it is conceivable that some of the SMaseC inhibitors described below act as mammalian nSMase inhibitors.

Phosphonate-based sphingomyelin analogs 107 and 108 (Figure 36) were explored as transition-state analog inhibitors of sphingomyelinase.^146,147 Compounds 107 and 108, however, showed rather weak inhibitory potency against B. cereus SMaseC with IC₅₀ values of 120 μM and 78 μM, respectively.¹⁴⁷ Phosphoramidate 109¹⁴⁸ and difluoromethylene phosphonate 110¹⁴⁹ (Figure 36) also exhibited low inhibitory potency against B. cereus SMaseC with IC₅₀ values of 53 and 57 μM, respectively. The weak inhibitory activity of these compounds can be at least partially attributed to the lack of the double bond corresponding to the C4–C5 double bond of sphingomyelin. Despite the presence of the C4–C5 double bond, however, the inhibitory activity of thiophosphate derivative 111 (Figure 36) was reported to be even weaker than other phosphorus-based compounds.¹⁵⁰ It was found, however, that compound 111 can be recognized as a substrate by B. cereus SMaseC, releasing 1-thiosphingosine. Since the free thiol group can be captured by various labeling reagents as generated, compound 111 may serve as a useful substrate to continuously monitor SMaseC activity.

Figure 36.

Structures of phosphorus-based SMaseC inhibitors 107–111.

On the basis of the inspection of the B. cereus SMaseC crystal structures,¹⁴³ sphingomyelin analog 112 containing a bipyridyl moiety (Figure 37) was designed to interact with the active site Mg²⁺ ion, predicted to play a critical role in the catalytic process.¹⁵¹ Compound 112 inhibited B. cereus SMaseC with an IC₅₀ value of 1.2 μM. It displayed a competitive inhibition with respect to sphingomyelin with a K_i value of 5.2 μM, indicating that the inhibitor competes with the substrate at the active site of the enzyme as designed. Compound 113 (Figure 37), a regioisomer of 112 with a bipyridyl moiety linked to the C3 secondary hydroxyl group, showed only weak inhibitory activity (20% inhibition at 100 μM). Docking simulation analyses suggest that the bipyridyl moiety of compound 112 is optimally positioned to participate in the coordination with the active site Mg²⁺ ion while that of compound 113 is oriented away from the active site.¹⁵¹ Subsequent structural optimization efforts led to the discovery of two submicromolar inhibitors 114 and 115 (Figure 37).¹⁵² Compound 114, devoid of an acyl group, inhibited B. cereus SMaseC with an IC₅₀ value of 0.9 μM. Compound 115 has a phenyl group replacing the alkenyl chain extending from the C3 position and inhibited B. cereus SMaseC with an IC₅₀ value of 0.8 μM. Like compound 112, the two inhibitors displayed a competitive inhibition with respect to sphingomyelin with K_i values of 2.8 and 1.3 μM, respectively. Intraperitoneal injection of compound 115 in mice decreased the lethality of the B. cereus treatment in a dose-dependent manner, demonstrating for the first time in vivo efficacy of a small molecule SMaseC inhibitor.¹⁵²

Figure 37.

Structures of SMaseC inhibitors 112–115 containing a bipyridyl moiety.

5. OTHER CERAMIDE-PRODUCING ENZYMES

In addition to the enzymes involved in the three major ceramide biosynthesis pathways described above, glucocerebrosidase, galactocerebrosidase, and ceramide-1-phosphate phosphatase are known to produce ceramide from distinct substrates (Figure 1). Ceramide-1-phoshate can be hydrolyzed by phosphatidate phosphohydrolase,¹⁵³ presumably owing to its overall structural similarity to phosphatidate. Although definitive evidence for the existence of monophosphatase specific to ceramide-1-phosphate remains to be seen, it has been implicated that ceramide-1-phosphate phosphatase activities detected in brain synaptosomes¹⁵⁴ and liver plasma membrane fractions¹⁵⁵ are distinct from that of phosphatidate phosphatase. Such enzyme, if identified, could serve as another therapeutic target aimed at regulating ceramide biosynthesis. Glucocerebrosidase and galactocerebrosidase are lysosomal glycoside hydrolases that cleave glucosylceramide and galactosylceramide, respectively, resulting in the generation of ceramide (Figure 1). The physiological and pathological significance of ceramide produced by these enzymes has not yet been extensively studied. However, accumulation of glucosylceramide due to inherited deficiency of glucocerebrosidase is linked to a rare genetic lysosomal storage disorder, Gaucher disease.¹⁵⁶ On the other hand, Krabbe disease, another genetic lysosomal storage disorder, is known to be caused by a deficiency in galactocerebrosidase, which leads to the accumulation of galactosylceramide.¹⁵⁷ Mutations in these enzymes commonly result in protein misfolding and subsequent premature degradation in the ER. To this end, interests in developing inhibitors of these enzymes have been predominantly driven by the goal of identifying molecular chaperones capable of stabilizing the native conformation of the mutant enzymes and increasing translocation of functional enzymes to the lysosomes. For instance, ambroxol (116) (Figure 38) was reported to act as a mixed-type inhibitor of human recombinant glucocerebrosidase at the neutral pH found in the ER and to display no inhibitory activity at the acidic pH of lysosomes.¹⁵⁸ This pH dependent inhibitory action makes ambroxol (116) an ideal pharmacological chaperone for mutant glucocerebrosidases. Indeed, treatment with ambroxol (116) resulted in the increase of glucocerebrosidase activity and protein levels in N370S/N370S GD-1 fibroblast cell line.¹⁵⁸ It should be noted that there have been active research efforts to develop pharmacological chaperones for lysosomal enzymes as new therapeutic options for various types of genetic lysosomal storage disorders including Gaucher and Krabbe diseases. Excellent review articles focused on the medicinal chemistry aspects of these efforts have been published in recent years.^159,160

Figure 38.

Structure of a pH-dependent glucocerebrosidase inhibitor, ambroxol 116.

6. CONCLUSION

The intricate metabolic network surrounding ceramide reflects the dynamic role played by ceramide at the core of sphingolipid metabolism pathways in both physiological and pathological conditions. The fact that several distinct enzymes can each produce ceramide has provided potentially important therapeutic opportunities by inhibiting specific ceramide-producing enzymes depending on the disease of interest. As described in this review article, tremendous efforts have been devoted to identifying small molecule inhibitors of these enzymes. Upon review of the structural evolution of these inhibitors, one notable trend appears to be a departure from lipid-like scaffolds originating from the structures of substrates. This movement has been most likely driven by a strong desire to identify more drug-like inhibitors that can serve not only as biological probe molecules but also as lead molecules for future therapeutic development. In particular, DES1 inhibitor SKI II (8), CerS1 inhibitor P053 (29),⁶⁰ aSMase inhibitor 59,⁸⁹ and nSMase2 inhibitors DPTIP (95),¹³⁵ PDDC (101),¹³⁷ and 106¹³⁹ represent promising leads for further optimization as these compounds displayed not only submicromolar inhibitory potency but also desirable in vivo pharmacokinetics and promising in vivo profile as summarized in Table 1.

Table 1.

Representative Inhibitors of Ceramide Biosynthesis

compd	target, *additional known target	pharmacokinetics	preliminary in vivo findings
SKI II (8)39,41	DES1 (Ki = 0.3 μM in HGC-27 cell lysates), *SK1/SK2	Orally available in mice	Showed a significant inhibition of tumor growth in mice bearing JC tumors following oral admin.
P053 (29)60	CerS1 (IC50 = 0.54 μM against human CerS1)	Orally available in mice	Increased fatty acid oxidation in skeletal muscle and reduced overall adiposity in mice fed a high-fat diet following oral admin.
5989	aSMase (IC50 = 0.32 μM in Huh7 cell lysates)	Distributed to the brain in rats (ip admin)	Improved depression-like behaviors of rats, reduced the cortical and hippocampal aSMase activity, and restored neurogenesis in the brain following ip admin.
DPTIP (95)135	nSMase2 (IC50 = 0.03 μM against human nSMase2)	Distributed to the brain in rats (ip admin)	Attenuated IL-1β-induced ADEV release in in GFAP-GFP mice following ip admin.
PDDC (101)137,138	nSMase2 (IC50 = 0.3 μM against human nSMase2)	Distributed to the brain in rats (oral admin)	Attenuated IL-1β-induced ADEV release in GFAP-GFP mice following oral admin and reversed cognitive impairment in 5XFAD mice following ip admin.
106139	nSMase2 (IC50 = 0.5 μM against human nSMase2), *AChE	Distributed to the brain (sc admin)	Diminished IL1β-induced brain EV release in tau P301S (line PS19) mice following sc admin.

Despite the growing interest in ceramide-producing enzymes as therapeutic targets, exploration of structure-based drug design has been limited to aSMase⁸⁹ and B. cereus SMaseC.¹⁵¹ Further progress in structural biology is crucial for successful development of structure-based drug design strategies for inhibitors of other ceramide-producing enzymes. While the recent elucidation of the crystal structure of human nSMase2 catalytic domain⁹⁶ is an important breakthrough, the full-length structure of nSMase2 would be highly desirable given that many nSMase inhibitors do not appear to bind to the active site.

Finally, recent advancement in the field of sphingolipidomics^161,162 is highly encouraging and relevant to inhibitors of ceramide-producing enzymes given that the use of reliable biomarkers is becoming increasingly important in successful translation of experimental drugs into clinical development.¹⁶³ Application of the most advanced sphingolipidomics analysis techniques in clinical studies should provide in-depth information on the metabolic consequence(s) of inhibiting specific ceramide producing enzymes, enabling informed decision making at various stages of clinical development. It is certainly a very exciting time to explore therapeutics for many disease areas by targeting ceramide biosynthesis.

ABBREVIATIONS USED

AChE

acetylcholinesterase

CerS

ceramide synthase

DES1

dihydroceramide desaturase 1

FIASMAs

functional inhibitors of acid sphingomyelinase

PDB

Protein Data Bank

PS

phosphatidylserine

SMase

sphingomyelinase

Biographies

Jan Skácel received his Ph.D. in Medicinal Chemistry from the University of Chemistry and Technology, Prague. As a Ph.D. student, he worked at the Institute of Organic Chemistry and Biochemistry of the Czech Academy of Sciences under the direction of Dr. Zlatko Janeba. His doctoral research was mainly focused on the design and synthesis of novel PNP and HG(X)PRT inhibitors. Currently, Dr. Skácel is a postdoctoral fellow at Johns Hopkins University, where he explores the design and synthesis of novel nSMase2 inhibitors under the direction of Prof. Takashi Tsukamoto.

Barbara S. Slusher is Professor of Neurology, Pharmacology, Neuroscience, Psychiatry, Medicine, and Oncology and the Director of Johns Hopkins Drug Discovery at Johns Hopkins School of Medicine. Prior to joining Johns Hopkins in September 2009, she spent 18 years in the pharmaceutical industry. She has extensive experience in drug discovery spanning through phase I/IIa clinical development. She is an inventor of over 80 issued patents and applications and an author of over 250 scientific articles. She serves on the board or as a scientific consultant to multiple biotechnology companies and cofounded the first International Academic Drug Discovery Consortium.

Takashi Tsukamoto is Associate Professor of Neurology at Johns Hopkins University and the Director of Medicinal Chemistry at Johns Hopkins Drug Discovery. He received his Ph.D. in Chemistry from Tokyo Institute of Technology and pursued his postdoctoral studies at the University of Michigan. Prior to joining Johns Hopkins in 2009, Dr. Tsukamoto has held a variety of research positions in the pharmaceutical industry. During the course of his career, he has served as a lead medicinal chemist in a number of drug discovery projects, exploring new therapeutics for neurological disorders and cancer. He is the lead inventor of the cytidine deaminase inhibitor cedazuridine and its combination with decitabine, which received FDA approval as Inqovi in 2020.