Repurposing degradation pathways for modular metabolite biosynthesis in nematodes

Abstract

Recent studies have revealed that Caenorhabditis elegans and other nematodes repurpose products from biochemical degradation pathways for the combinatorial assembly of complex modular structures that serve diverse signaling functions. Building blocks from neurotransmitter, amino acid, nucleoside and fatty acid metabolism are attached to scaffolds based on the dideoxyhexose ascarylose or glucose, resulting in hundreds of modular ascarosides and glucosides. Genome-wide association studies have identified carboxylesterases as the key enzymes mediating modular assembly, enabling rapid compound discovery via untargeted metabolomics and suggesting that modular metabolite biosynthesis originates from the ‘hijacking’ of conserved detoxification mechanisms. Modular metabolites thus represent a distinct biosynthetic strategy for generating structural and functional diversity in nematodes, complementing the primarily polyketide synthase- and nonribosomal peptide synthetase-derived universe of microbial natural products. Although many aspects of modular metabolite biosynthesis and function remain to be elucidated, their identification demonstrates how phenotype-driven compound discovery, untargeted metabolomics and genomic approaches can synergize to facilitate the annotation of metabolic dark matter.

Growing evidence suggests that only a small portion of the metabolomes of animal model systems, including Caenorhabditis elegans, Drosophila and mouse, has been characterized so far^1–4. The finding that uncharacterized metabolites may greatly outnumber known metabolites raises two major questions: (1) how this large number of metabolites is generated and (2) to what extent all of these as yet uncharacterized compounds serve discernable roles in biology.

Regarding the origin of the many unknowns in biological samples, it has been proposed that much of the unknown structure space may simply reflect ‘metabolic noise’, representing artefacts⁵ or products of unspecific side reactions of known metabolic pathways. Conversely, as-yet-uncharacterized metabolites could be the products of dedicated biosynthetic pathways that remain unknown. If such dedicated biosynthetic strategies could be identified, this would suggest that the associated metabolites probably serve specific functional roles. Known small molecules serve as signals that are involved in nearly every aspect of biology at intracellular, intra-organismal and inter-individual levels, as well as in interactions between species^6–10. Because delineating the biological context of newly identified metabolites is often not straightforward, the identification of corresponding biosynthetic pathways can be an important first step towards deciphering their functions.

In this Perspective, we aim to highlight one simple and intuitive strategy for generating structural diversity: the combinatorial assembly of widely available building blocks from primary metabolism via simple glycosidic, ester or amide linkages. Although all biogenic small molecules must ultimately derive from primary metabolic building blocks, recent metabolomic and biosynthetic studies have revealed that C. elegans and other nematodes make use of a simple assembly strategy based on ester- and amide-bond formation, often employing a glycoside as a central core scaffold. Several hundred such modular metabolites have now been identified in C. elegans, and ongoing studies have started to uncover central tenets of the underlying biosynthetic logic, as well as a vast range of biological functions.

In the following sections, we first describe how phenotypic observations drove identification of the first family of nematode-derived modular metabolites, the ascarosides (Fig. 1a). We then discuss the biochemical and bioinformatic approaches that revealed the unexpected role of carboxylesterase enzymes in modular ascaroside biosynthesis. Comparative metabolomic analyses of mutants of carboxylesterase homologs then uncovered the modular glucosides (MOGLs) as a distinct and even larger family of modular metabolites that connect neurotransmitter metabolism and many other biochemical pathways, promising to further advance our understanding of the role of small-molecule signals in the model system C. elegans. Finally, we compare the nematode strategy of generating structural diversity via modular assembly with that of microbial natural-product biosynthesis via polyketide synthases (PKS) and nonribosomal peptide synthetases (NRPS). Together, the identification of nematode-derived modular metabolites and their corresponding biosynthetic enzymes provides an instructive example of how biochemical, bioinformatic and phenotypic techniques can synergize to enable the chemical discovery and biological contextualization of novel metabolite families.

Fig. 1 |. Modularity of ascarosides in C. elegans and related species.

a, Activity-guided fractionation of the metabolites responsible for specific phenotypes in C. elegans and P. pacificus led to the discovery of modular ascarosides. b, Left: schematic overview of the modular assembly of ascarosides in C. elegans, P. pacificus and related nematode species. A core ascaroside is modified at the carboxyl-terminus, 2’- and 4’-hydroxyl group by catabolites derived from amino acid (red), fatty acid (green), carbohydrate (black), nucleoside (blue) and co-factor (brown) metabolism. Assembly of modular ascarosides requires specific carboxylesterases for attachment of, for example, succinylated octopamine (Cel-CEST-8), indole-3-carboxylic acid (Cel-CEST-3), tiglic acid (Cel-CEST-9.2), ureido isobutyric acid (Ppa-UAR-1), glucosyl uric acid (Cel-CEST-1.1) and p-aminobenzoic acid (Cel-CEST-2.2) to the ascaroside core. Right: structures of nematode-derived 3,6-dideoxyhexoses, selected modular ascarosides and the paratoside npar#1.

Phenotypic screens reveal a library of modular ascarosides

The discovery of nematode-derived modular metabolites began with the identification of a family of glycolipids, the ascarosides (Box 1 and Fig. 1), which, over the past 15 years, have been shown to play a central role in nematode biology by regulating development, behavior and many other life history traits^8,11–16. One such life history trait is the ability of C. elegans larvae to interrupt normal reproductive development and form highly stress-resistant ‘dauer’ (German for ‘enduring’) larvae in response to adverse environmental conditions. Dauer larvae have been a major focus of C. elegans research, and work starting in the 1980s had shown that C. elegans produces a small molecule that triggers dauer larva formation^17–19. Dauer-promoting constituents of the C. elegans metabolome were finally identified using multi-stage activity-guided fractionation. The most active (that is, dauer-inducing) fractions were pooled and characterized using NMR spectroscopy and mass spectrometry, which resulted in the identification of several dauer-inducing ascarosides (for example, ascr#3 (1); Fig. 1b), consisting of the 3,6-dideoxyhexose, l-ascarylose, with a fatty acid-derived side chain of variable length attached to the anomeric position^20,21. Following identification of the first dauer-inducing ascarosides, additional activity-guided fractionation efforts in C. elegans and related nematodes (for example, Pristionchus pacificus) led to the identification of a large number of structurally more elaborate ascaroside derivatives with diverse biological functions (Fig. 1b). These ‘modular ascarosides’ incorporate additional building blocks from amino acid, neurotransmitter, nucleoside and other primary metabolic pathways (for example, the indole carboxyl ascaroside icas#9 (2), which integrates an indole carboxyl moiety, or ascr#8 (3), featuring a p-aminobenzoic acid moiety). Figure 1b provides an overview of the structural diversity and biological activities of ascarosides identified in C. elegans and the satellite model organism P. pacificus.

Box 1. A brief history of ascarosides—110 years of discovery.

‘Ascarosides’ were first described by Flury in 1912 as an unusual type of lipid (‘unsaponifiable matter’) that accounted for nearly 25% of the total lipid content in parasitic nematodes of the genus Ascaris. It was not until 1957 that these ascaroside lipids were chemically characterized as glycosides of a dideoxysugar named ascarylose linked to very long-chain aliphatic side chains. It took another 50 years before the role of shorter side-chained ascarosides as signaling molecules was uncovered. A series of studies in 2005–2008 reported that short-chain ascarosides are major components of the population density signal that triggers developmental arrest at the dauer stage, an important model for aging research. Subsequently, it became apparent that ascaroside signaling is involved in virtually every aspect of the life history of C. elegans and many other nematodes, including parasitic species, regulating not only development but also many behaviors, as well as lifespan and reproduction. Most recently, ascarosides have been found to modulate immune responses in animals and plants, suggesting that other phyla have evolved to recognize ascarosides as an evolutionarily conserved signature of nematodes.

Importantly, the identification of new structural classes and subtypes went hand-in-hand with the identification of new biological functions. For example, indole ascarosides such as icas#9 (2) were shown to function as aggregation signals, whereas osas#9 (4), incorporating the neurotransmitter octopamine as a building block, acts as a dispersal signal, and ascr#8 (3), featuring a likely folate-derived p-aminobenzoic acid moiety, synergizes with other ascarosides to strongly induce larval diapause (dauer)^22–24. Investigations of other nematode species have revealed incorporation of additional modules and use of 3,6-dideoxysugars other than ascarylose as scaffolds. For example, one of the major components of the dauer pheromone in P. pacificus is a compound named npar#1 (5), which is based on L-paratose and further incorporates an unusual xylose-based nucleoside^25–27. An additional 3,6-dideoxyhexose core, l-caenorhabdose, was identified in Caenorhabditis nigoni, a nematode closely related to C. elegans (Fig. 1b)²⁸. Several other ascarosides incorporating nucleoside-derived moieties have also been reported from C. elegans^3,27, for example uglas#11 (6) and its isomer uglas#15 (7)²⁹, which is strongly enriched in male animals³⁰.

Over the past 15 years, a large library of more than 200 different ascarosides have been identified in nematodes, some of which feature additional modifications such as phosphorylation of the ascarylose or additional glucose moieties^31,32 (see Supplementary Table 1 for a list of structures, source species and phenotypic associations). The sheer number of modular metabolites begged the question as to whether they are the products of dedicated biosynthetic pathways, or rather represent just biosynthetic ‘noise’. For example, one might reason that modular ascarosides simply represent products of non-enzymatic reactions between abundant reactive intermediates, for example, activated acyl CoA esters, and diverse biogenic nucleophiles. However, the assembly of different moieties in modular ascarosides is highly specific; for example, p-aminobenzoic acid is combined almost exclusively with 7-carbon side chain ascarosides (ascr#8 (3)), but not the more abundant 9-carbon side chain ascarosides, whereas indole carboxyl and octopamine-derived moieties (for example, in icas#9 (2) and osas#9 (4)) are preferentially attached to the 4′-position of 5- and 9-carbon side ascarosides. Moreover, multiple studies have shown that modular ascaroside profiles are species-specific^25,33,34. For example, ubas#1 (8) has only been detected in P. pacificus, whereas uglas#11 (6) has only been detected in C. elegans^3,29. Taken together, species specificity and their peculiarly selective assembly strongly suggest that the ascaroside library must be derived from directed biosynthetic pathways.

Genome-wide association studies uncover modular ascaroside biosynthesis

When we and others began our search for enzymes involved in the assembly of the modular ascaroside library, it was unclear what enzyme families could be responsible. It was known that the ascaroside fatty acid-like side chains are derived from peroxisomal β-oxidation of very long-chain precursors, and that many of the other building blocks are derived from conserved primary metabolism pathways, such as amino acid catabolism^23,32. However, the origin of the crucial ester and amide linkages remained enigmatic. Genetic screens targeting putative enzymes that could plausibly generate ester or amide bonds, such as O-acyl and N-acyl transferases (OACs and NATs), were unsuccessful; production of the identified modular ascarosides was not abolished in any of the screened mutants³⁵. Ultimately, the use of genome-wide association studies (GWASs) led to success (Fig. 2a,b). This a priori unbiased approach takes advantage of natural genomic variation among nematode strains isolated from the wild to discern genomic loci where sequence variation is associated with significant changes in a phenotypic trait of interest, in this case, the abundance of a specific modular ascaroside.

Fig. 2 |. GWAS uncovers the enzymes required for modular ascaroside biosynthesis.

a, The discovery of enzymes required for modular ascaroside biosynthesis allows for the characterization of homologous enzymes in related species. b, GWAS-based strategy for linking genomic and metabolomic natural variation to identify gene candidates involved in modular ascaroside biosynthesis. The biosynthetic role of the candidate gene, Ppa-uar-1, was validated by means of metabolomics of knockout mutants generated via CRIPSR/Cas9 and heterologous expression of Ppa-uar-1 in C. elegans. Knockout of Ppa-uar-1 homologs in C. elegans and Caenorhabditis briggsae revealed the broad scope of cest-family carboxylesterases; that is, Cel-CEST-3 is required for the production of carboxyl-indole ascarosides (icas#9 (2)), and Cel-CEST-2.2 is required for the production of the p-aminobenzoic acid ascaroside ascr#8 (3). c, Dendrogram relating Ppa-UAR-1 (red) to homologous predicted proteins in C. elegans (black and blue) and C. briggsae (green). Proteins in red, blue and green have been shown to be required for assembly of modular ascarosides or modular glucosides. Asterisks in b denote proposed structures.

In a major tour de force, the metabolomes of 300 fully sequenced natural strains of P. pacificus were analyzed via HPLC-MS to measure variation of modular ascaroside production across these 300 strains³⁶. It was found that abundances of many modular ascarosides varied greatly between strains. Variation was particularly large for modular ascarosides containing a 4′-attached ureido isobutyric acid group, referred to as ubas#1-3 (8–10). GWAS revealed that variation of ubas#1-3 (8–10) production correlated with genomic variation in a region of chromosome I that contained several genes coding for putative α/β-hydrolase-fold family enzymes with homology to mammalian liver carboxylesterases (for example, CES2). Using CRISPR-Cas9, mutant strains for the best candidate gene, later named Ppa-uar-1 (ubas-ascaroside-required), were generated. Metabolomic analysis demonstrated that Ppa-uar-1 mutants are deficient specifically in the production of the three ubas-ascarosides, whereas production of other ascarosides, including the putative precursor ascr#9 (11), was not affected. Moreover, levels of ureido isobutyric acid remained unchanged in the Ppa-uar-1 mutants, suggesting that Ppa-UAR-1 is specifically required for attachment of the ureido isobutyric acid moiety in position 4′ of ascr#9 (11), To test this idea, Ppa-uar-1 was heterologously expressed in C. elegans, which does not normally produce ubas-ascarosides. Metabolomic analysis of C. elegans expressing Ppa-uar-1 revealed production of a modular ascaroside not present in either the wild-type C. elegans or P. pacificus, featuring a phenylalanyl-moiety attached to the 4′-position of ascr#3 (1) (phascr#3 (12); Fig. 2b). A corresponding ureido isobutyryl derivative was not detected; nonetheless, these results further supported direct involvement of Ppa-UAR-1 in the attachment of acyl moieties at the 4′ position of ascarosides³⁶. This result was surprising, given that α/β-hydrolase-fold enzymes act primarily as esterases, peptidases and lyases in animals, plants and bacteria, often as part of detoxification pathways^37–40. However, there are a few examples for bacterial α/β-hydrolase-fold enzymes catalyzing acyl-transfer reactions resulting in the formation of ester and amide bonds. In these enzymes, residues near the active site appear to favor binding an acceptor alcohol instead of water, thus promoting acyl transfer rather than hydrolysis^39,40.

The discovery of the biosynthetic role of Ppa-uar-1 via GWA in P. pacificus enabled rapid progress toward understanding modular ascaroside biosynthesis in other nematode species. The C. elegans genome includes ~40 carboxylesterase (cest) homologs of Ppa-uar-1 (Fig. 2c), and ongoing loss-of-function genetic screens have revealed that CEST enzymes mediate formation of diverse ester as well as amide bonds in modular ascarosides, resulting in attachment of specific building blocks via the ascarylose hydroxyl groups or the carboxyl terminus of the fatty acid side chain^41,42. For example, Cel-cest-3 and Cel-cest-8 are required for 4′-attachment of the indole carboxyl and octopamine succinate moieties to ascr#9 (11) in precursors of icas#9 (2; Fig. 2b) and osas#9 (4), respectively, whereas Cel-cest-2.2 is required for amide-bond formation in ascr#8 (3), from the ascr#7 (13) precursor (Fig. 2b)^41,42.

Although diverse ester and amide bonds are formed, cest-mediated acyl transfer reactions are often highly substrate-specific; for example, Cel-cest-1.1 is required for the attachment of ascr#1 at the carboxyl terminus to the 2′-position of uric acid glucoside, forming uglas#11 (6), whereas attachment at the 6′-position (uglas#15) (7) is Cel-cest-1.1-independent (Fig. 1b). The discovery of the roles of CEST enzymes in modular ascaroside biosynthesis provides new tools to study how these particular modular metabolites affect C. elegans behavior and other life history traits and, furthermore, enable additional phenotypic screens. For example, Cel-cest-3 mutant worms, which lack the ability to form the aggregation and anti-foraging signal icas#9 (2), have an increased tendency to migrate away from their bacterial food lawn⁴¹. In another example, it was shown that Cel-cest-2.2 overexpression increases C. elegans lifespan. However, the role of Cel-cest-2.2-dependent ascarosides, for example, ascr#8 (3), in C. elegans lifespan extension remains to be determined⁴³. Transcriptional regulation of CEST enzymes can reveal unexpected connections to other pathways; for example, Cel-cest-1.1 expression and production of the Cel-cest-1.1-dependent uglas#11 (6) were found to be upregulated in insulin signaling mutants²⁹. Much work remains to be done to characterize cest-dependent phenotypes. Additionally, it is unclear whether cest-family enzymes can explain the entire diversity of more than 100 modular ascarosides, as other enzymes or mechanisms may contribute^35,44.

Cest genes are abundantly expressed in the intestine, one of the primary sites of ascaroside biosynthesis, although RNA-seq data suggest that some cest genes are also expressed in other tissues; for example, Cel-cest-1.2 and Cel-cest-2.1 may also be expressed in the AWA sensory and RID motor neurons^29,42,45. Clues about the subcellular localization of CEST enzymes emerged from the earlier finding that production of most modular ascarosides is dependent on Cel-glo-1, a Rab GTPase required for the formation of fluorescent lysosome-related organelles (LROs), known as gut granules^35,42. An mCherry-tagged Cel-cest-2.2 construct was found to co-localize with intestinal gut granules, consistent with a model in which modular ascaroside assembly via CESTs occurs at or within the gut granules⁴². All nematode cest homologs that have been biochemically characterized so far feature a predicted C-terminal transmembrane domain, suggesting that they are anchored to LROs and potentially other membranes^41,42. Efforts toward heterologous expression of C. elegans cest-family carboxylesterases in bacteria or cell culture have not yet been successful^41,42, perhaps in part due to difficulties arising from the presence of the transmembrane domain. Moreover, the highly acidic environment of the LROs may be required for proper CEST function, assuming that CESTs, like other type I transmembrane proteins featuring a C-terminal transmembrane domain, have the majority of the enzyme facing the inside of the LRO^46,47. However, CESTs may not exclusively localize to the acidic LROs in the intestine⁴², and their intracellular localization in other tissues, for example, Cel-cest-1.2 and Cel-cest-2.1 in neurons, remains unknown.

Cest mutants reveal modular glucosides

The initial targeted metabolomic analyses of cest mutants revealed their roles in modular ascaroside biosynthesis. However, many of the tested mutants did not exhibit any significant changes in ascaroside production. To clarify whether cest homologs perhaps serve additional biosynthetic functions, untargeted metabolomics was used to investigate the metabolomes of a CRISPR/Cas9-generated library of cest loss-of-function mutants by comparison with wild-type C. elegans (Fig. 3a). Untargeted metabolomics is a discovery-focused strategy that provides a global and comprehensive inventory of all detectable metabolites present in a sample, independent of any prior knowledge of metabolite structures. Comparative analysis of datasets from untargeted metabolomics of wild-type and mutant strains can thus enable discovery of entirely unexpected or unprecedented metabolites associated with a specific genetic background. The use of precise CRISPR–Cas9-edited mutants is particularly advantageous for untargeted comparative metabolomic studies, which essentially aim to find the ‘needle in the haystack’, that is, to identify one or a few metabolites dependent on a specific gene against a vast background of other, mostly unannotated metabolites. Untargeted metabolomics of several cest mutants not involved in modular ascaroside biosynthesis ultimately uncovered a distinct, even larger library of modular compounds, the MOGLs. In contrast to ascarosides, MOGLs are based on glucosides of a range of mostly aromatic compounds derived from neurotransmitter, amino acid and nucleoside metabolism. These glucoside scaffolds are then further elaborated with a wide range of acyl moieties, in a manner similar to the modular ascarosides (Fig. 3b).

Fig. 3 |. Comparative metabolomics of cest mutants reveals a previously uncharacterized family of modular glucosides.

a, Metabolomic analysis of mutants of homologs of the carboxylesterase Ppa-uar-1 led to the discovery of a novel family of modular metabolites. b, Schematic overview of MOGL assembly in C. elegans. MOGL assembly relies on carboxylesterases for the attachment of anthranilic acid to N-acetylserotonin and indole glucosides (orange, Cel-CEST-4), ascr#1 to uric acid glucoside (green, Cel-CEST-1.1), and a wide range of acyl moieties to tyramine and indole glucosides (blue, Cel-CEST-1.2 and Cbr-CEST-2). c, Structures for selected modular glucosides from C. elegans. d, Nematode phylogenetic tree (derived from small subunit-ribosomal RNA gene sequencing)⁵². Clades for which BLAST analysis revealed an expansion of Cel-CEST-1.2 homologs including a predicted C-terminal transmembrane domain are marked with a red circle. e, Structures of N-acetylserotonin glucoside (sngl#1), its vertebrate analog N-acetylserotonin glucuronide, and buprestin B from jewel beetles. Asterisks denote proposed structures.

The first MOGLs were identified in comparative metabolomic analyses of Cel-cest-4 mutants. Four compounds consistently abolished in Cel-cest-4 mutant metabolomes (iglu#3, iglu#4, sngl#3 and sngl#4, 14–17; Fig. 3c) were identified as derivatives of indole glucoside and N-acetylserotonin-O-glucoside bearing an anthranilic acid moiety in the 6′ position of the glucose^42,48. Subsequent analysis of Cel-cest-1.2 mutants then revealed that MOGLs constitute a large library of metabolites. Cel-cest-1.2 mutants are defective in the biosynthesis of more than 150 MOGLs, primarily based on glucosides of indole and the neurotransmitter tyramine and featuring a wide range of different acyl moieties attached to the 2′-O and/or 6′-O-positions of the glucose (Fig. 3c)⁴⁹. Although it may seem bewildering at first that a single enzyme could be required for the production of such a large number of highly diverse metabolites, closer inspection of the identified structures showed that Cel-CEST-1.2 is specifically required for the attachment of acyl moieties to the 2′-O-position in the parent glucoside. For example, production of iglu#121 (18), featuring a 2′-O-benzoyl moiety, is abolished in Cel-cest-1.2 mutants, whereas abundance of the corresponding 6′-O-acylated isomer, iglu#12 (19), is slightly increased. Nonetheless, unlike the other so far characterized cest homologs, which are selective with regard to the acyl moieties and ascaroside or glucoside scaffolds they combine, Cel-cest-1.2 appears to have a much wider substrate scope, including a range of different glycoside scaffolds and diverse acyl moieties for attachment. Examples include the tyramine-derived tyglu#2 (20) and octopamine-derived oglu#401 (21). The broad substrate scope and selectivity inferred for 2′-O-attachment of Cel-cest-1.2 is conserved for Cbr-cest-2, the Cel-cest-1.2 ortholog in the closely related species Caenorhabditis briggsae. Cbr-cest-2 is required for the biosynthesis of at least 117 2′-O-acylated MOGLs in C. briggsae, of which 97 can also be detected in C. elegans and are dependent on Cel-cest-1.2⁴⁹ (Supplementary Table 1). Notably, substrate promiscuity is common among other, non-cest-members of the α/β hydrolase-fold enzyme family^50,51.

It is possible that, like the modular ascarosides, MOGLs are produced by a wide range of nematode species. The large number of cest family enzymes in C. elegans, C. briggsae and P. pacificus (20, 18 and 24 CEST homologs containing a C-terminal transmembrane domain, respectively), may originate from recent lineage-specific gene duplication events. BLAST analysis of Cel-CEST-1.2 against genomes of species from other branches of the nematode phylogenetic tree revealed species in several clades whose genomes feature expansion of α/β-hydrolase-fold enzymes with a predicted C-terminal transmembrane domain, suggesting that cest family genes have undergone multiple independent expansions (Fig. 3d)⁵². Even closely related species can differ greatly with regard to the number of CEST homologs and their sequence characteristics; for example, Caenorhabditis japonica lacks predicted CEST homologs featuring a C-terminal transmembrane domain. Similar lineage-specific differences are found within the Caenorhabditis genus for G-protein coupled receptors (GPCRs), which have undergone dramatic expansion in nematodes⁵³; for example, the number of predicted GPCRs is much reduced in Caenorhabditis inopinata compared to C. elegans⁵⁴. These species-specific differences are consistent with the idea that capabilities for both metabolite biosynthesis and perception can evolve rapidly. Notably, all CESTs with demonstrated involvement in modular ascaroside and MOGL biosynthesis in C. elegans, C. briggsae and P. pacificus group together in the dendrogram (Fig. 2c), which may provide clues for identifying additional enzymes participating in modular metabolite assembly.

Much remains to be learned about the biological roles of MOGLs. In C. elegans, the more abundant phosphorylated MOGLs are specifically retained within the worm body and not secreted into the environment, suggesting that they may play a role in intra-organismal signaling. Interestingly, MOGL production greatly increases during starvation in C. elegans and is required for fitness; Cel-cest-1.2 mutants, which lack most 2′-O-acylated MOGLs, have significantly decreased starvation survival compared to wild-type worms⁴⁹. These preliminary findings suggest that MOGLs play a role in starvation survival or are involved in stress response pathways. Although many of the recently identified MOGLs are based on indole glucoside scaffolds, it should be noted that indole glucoside production in C. elegans is dependent on indole production by the Escherichia coli bacteria used as food for C. elegans in the laboratory. In the absence of bacterial indole, Cel-CEST-4 and Cel-CEST-1.2 show a strong preference for acylating O-glucosides of N-acetylserotonin or tyramine, respectively, indicating an interplay between the bacterial diet and neurotransmitter metabolism. In fact, a recent study demonstrated that most serotonin in C. elegans is converted into Cel-cest-4-dependent MOGLs, and that the abundance of these serotonin-incorporating metabolites is modulated by the presence of bacterial indole⁴⁸. Given the specificity and integration of microbial and host metabolism, it is tempting to speculate that MOGLs bind to dedicated receptors, for example, GPCRs, similar to modular ascarosides⁵⁵, or act as nuclear hormone receptor ligands. Additionally, MOGLs could ostensibly function as allosteric modulators of other receptors or enzymes.

Related metabolites and pathways in other animals

Caenorhabditis elegans extensively glucosylates a wide range of molecular moieties derived from primary metabolism, including neurotransmitters and other amino acid degradation products, as well as nucleoside breakdown products such as uric acid (for example, gluric#1, 22) and methylguanine^42,49,56,57. Similarly, C. elegans can glucosylate many exogenous compounds, including synthetic compounds such as the antihelminthic albendazole (23; Fig. 4a)^58–60. As a bacterivore living in soil and leaf litter, C. elegans encounters a wide range of potentially toxic bacterial and fungal metabolites, so glucosylation may represent a detoxification mechanism. As mentioned above, C. elegans efficiently converts E. coli-derived indole (24), which at high concentrations becomes toxic, into glucosides iglu#1 and iglu#2 (25)^59,60. Analogously, hydroxyphenazine (26), a toxin produced by pathogenic strains of Pseudomonas spp., is rapidly converted into the corresponding glucoside (27), which is less toxic⁶⁰. Glucosylation in C. elegans thus parallels glucuronidation and glycosylation of exogenous toxins and endogenous catabolites as part of phase II detoxification pathways in vertebrates^61–63. For example, N-acetylserotonin (28) and the soy isoflavone genistein (29) are glucosylated (producing for example, sngl#1 (30)) in C. elegans, but glucuronidated in mammals (for example, N-acetylserotonin glucuronide (31);Fig. 3e)^64–66. Glycosylation in mammals relies primarily on uridine 5′-diphospho-glucuronosyltransferases (UGTs) and β-glucosidases. The C. elegans genome contains ~60 UGT homologs, but little is known about their functions. Given that UGTs often have wide and overlapping substrate scopes^61,63,67,68, elucidating their potential roles in MOGL biosynthesis could be challenging.

Fig. 4 |. Strategies utilized by C. elegans to create structural diversity.

a, Combinatorial strategy for the assembly of modular metabolites in C. elegans and other nematodes. Primary metabolic building blocks, often derived from degradation pathways, as well as xenobiotics are glycosylated. CEST family carboxylesterases then attach additional moieties to the resulting ascaroside and glucoside scaffolds via ester and amide bonds, producing modular ascarosides and MOGLs. b, Biosynthesis of nemamide A (35) in C. elegans as an example for assembly-line style biosynthesis of PKS/NRPS-derived metabolites^87,88. Shown are the domain architectures of Cel-PKS-1 and Cel-NRPS-1 and key steps of nemamide biosynthesis. Asterisks denote proposed structures.

In contrast to vertebrates, in which glycosylation usually facilitates excretion, the C. elegans glucosides serve as scaffolds for the attachment of other moieties via cest-dependent pathways and are mostly retained in the worm body in the form of phosphorylated derivatives (for example, iglu#401 (32), tyglu#6 (33) and sngl#4 (17); Fig. 4a). Essentially, Cel-CEST-4 and Cel-CEST-1.2 appear to repurpose intermediates from detoxification or degradation pathways for the biosynthesis of structurally diverse MOGLs. Similarly, modular ascarosides are derived from combining products from β-oxidation of nematode-specific ascaroside lipids with catabolites from other pathways. However, in contrast to most MOGLs, the modular ascarosides are primarily excreted to mediate inter-organismal communication.

To what extent modular metabolites similar to those identified from nematodes are produced by other animals is unclear. Production of glucuronates and glucosides in diverse vertebrate species has been extensively documented^61,63, but, in most cases, it is unknown whether they have additional metabolic roles beyond facilitating transport and excretion. An interesting example is N-acetylserotonin glucuronide (28), the mammalian analog of sngl#1 (30), the production of which was recently shown to be regulated by gut microbiota⁶⁶. Finally, there is the intriguing case of jewel beetles, for example the emerald ash borer, which produce a family of glucosides, the buprestins, with striking similarity to the nematode-derived MOGLs. The buprestins are β-glucosides of pyrrolic acid that bear additional acyl moieties–preferentially additional pyrrolic acid moieties–at the 2′-O- and 6′-O-positions, mimicking the substitution patterns of the nematode-derived MOGLs. An example, buprestin B (34), is shown in Fig. 3e. Buprestin production by jewel beetles appears to confer protection from predators, but their biological functions have not been studied in great detail^69–72. Nothing is known about buprestin biosynthesis. However, it is intriguing that many beetle species, including jewel beetles, are intimately associated with nematodes^73,74.

Conclusions and outlook

The identification of modular metabolites in nematodes provides an instructive example of how phenotypic analyses, metabolomics and bioinformatic approaches can synergize to enable discovery (Fig. 3a). Identification of the first ascarosides was driven by behavioral and developmental phenotypes. The chemical structures of the identified ascarosides then motivated a search for the enzymes and pathways required for their production, which, using GWAS, revealed the roles of cest homologs for modular assembly. In turn, comparative metabolomics of mutants of cest homologs revealed the MOGLs as a large family of previously unrecognized metabolites.

Ultimately, identification of the biosynthetic roles of cest homologs has led back to phenotype discovery. Although the large diversity of identified MOGLs may pose great challenges in terms of their functional characterization, mutant phenotypes, such as reduced starvation survival of Cel-cest-1.2 mutants, have provided a starting point for elucidating their biological roles. The specific incorporation of neurotransmitter or amino acid catabolism-derived building blocks into the MOGLs may provide additional functional clues; for example, combining building blocks from different pathways may integrate information about the metabolic state of the animal.

Although considerable progress has been made, our understanding of modular metabolite biosynthesis in C. elegans remains incomplete. For example, none of the cest mutants screened so far are responsible for attachment of the wide range of acyl moieties to the 6′-position of glucosides, except for Cel-CEST-4, which is involved exclusively in attachment of anthranilic acid to a small number of scaffolds. Moreover, it is unclear whether attachment of acyl moieties is reversible. For example, it is conceivable that CEST homologs participate in both ester formation and hydrolysis, depending on cellular conditions. In addition, a better understanding of the roles of metabolite localization and transport between different cell compartments is needed. For example, it is unclear whether the peroxisomal β-oxidation-derived ascaroside scaffolds are actively transported from the peroxisomes to the LROs, or perhaps peroxisomal degradation via the autophagy–lysosome pathway is involved⁷⁵. Inter-organelle transport of metabolites and its role for small-molecule biosynthesis and signaling also remain poorly understood in other eukaryotes and present an interesting challenge^76–78.

Finally, the potential roles of UGTs or other glycosylation enzymes in the biosynthesis of the scaffold glucosides and ascarosides, as well as the origin of the curious 3′-O-phosphorylation found in most MOGLs, remain to be determined. Given the large number of UGTs in nematodes as well as the potential involvement of other, unexpected enzyme families, GWAS using the extensive collection of available natural variants of C. elegans may provide a viable alternative to exhaustive screening of CRISPR-Cas9-derived mutant libraries^79–81. More generally, GWAS using natural variants or recombinant inbred lines⁸² may enable linking yet entirely uncharacterized metabolites with biosynthetic or otherwise related pathways at scale, a tantalizing prospect that could dramatically accelerate both structural and functional annotation of unknown metabolites.

Taken together, biosynthesis of the modular ascaroside and glucoside libraries shows how nematodes repurpose enzymes and metabolites generally associated with catabolic and detoxification pathways to generate structural diversity (Fig. 4a). This strategy contrasts with the biosynthesis of most bacterial and fungal secondary metabolites via assembly-line-style PKSs⁸³ and NRPSs⁸⁴. The chemical diversity of NRPS- and PKS-derived natural products is generated via sequential condensation of amino acids and amino acid analogs or a limited set of fatty acyl building blocks (for example, acetate and propionate)⁸⁵. Further structural diversification arises from reduction, oxidation and other modifications of the growing chain by other enzymes⁸⁶. In contrast, nematode-derived modular metabolites integrate a wider range of building blocks from all major metabolic pathways, including nucleoside derivatives, carbohydrates, as well as diverse products from amino acid, neurotransmitter and fatty acid metabolism. However, there is no obvious equivalent to the extensive tailoring found in many PKS- or NRPS-based pathways. Interestingly, the C. elegans genome includes two genes with close homology to microbial PKS and NRPS, which participate in the biosynthesis of the macrocyclic nemamides, such as 35 (Fig. 4b)^87,88. The nemamides′ structures and biosyntheses provide an example of the complementary characteristics of nematode-derived modular metabolites and PKS/NRPS pathways.

In contrast to PKSs and NRPSs, carboxylesterases and other α/β-hydrolase-fold family enzymes are abundant across animal phyla. Like many other enzyme families, their endogenous biochemical roles have not yet been comprehensively studied using untargeted metabolomics. The role of CEST homologs in the biosynthesis of modular metabolites in nematodes suggests the possibility that α/β-hydrolase-fold family enzymes in other phyla also contribute to the biosynthesis of yet uncharacterized secondary metabolite families.