De novo phytosterol synthesis in animals

Sterols are vital for nearly all eukaryotes. Their distribution differs in plants and animals, with phytosterols commonly found in plants whereas most animals are dominated by cholesterol. We show that sitosterol, a common sterol of plants, is the most abundant sterol in gutless marine annelids. Using multiomics, metabolite imaging, heterologous gene expression, and enzyme assays, we show that these animals synthesize sitosterol de novo using a noncanonical C-24 sterol methyltransferase (C24-SMT). This enzyme is essential for sitosterol synthesis in plants, but not known from most bilaterian animals. Our phylogenetic analyses revealed that C24-SMTs are present in representatives of at least five animal phyla, indicating that the synthesis of sterols common to plants is more widespread in animals than currently known. Description Editor’s summary For most animals, cholesterol is an essential membrane lipid and signaling molecule. Plants and fungi use chemically similar molecules with one or two additional carbon atoms. Michellod et al. studied two species of gutless marine annelids and found that they produced sitosterol, a common plant lipid, as the main sterol in their lipid membranes (see the Perspective by Brocks and Bobrovskiy). A noncanonical methyltransferase that is known in plants but not previously known in bilaterian animals appears to be key in generating this sterol. Additional analyses of genomic and transcriptomic data from annelids and several other animal phyla demonstrated that this biosynthesis gene is widespread and has a complex evolutionary history. —Michael A. Funk Some annelids produce sitosterol, a biomarker lipid more commonly found in plants.

S terols are lipids that play essential roles in all multicellular eukaryotes. Their distribution and synthesis differs across eukaryotic kingdoms. Fungi and plants mainly synthesize sterols with 28 to 29 carbon atoms (C 28 and C 29 ) called ergosterols and phytosterols (1,2), whereas animals predominantly produce the C-27 sterol cholesterol. These interkingdom differences reflect the complex evolutionary history of sterol synthesis. Phylogenetic analyses suggest that most enzymes for the biosynthesis of plant, fungal, and animal sterols were present in the last eukaryotic common ancestor (LECA) (3,4), with the distribution observed in most extant eukaryotes evolving through multiple events of enzyme losses and different pathways for sterol synthesis.
Cholesterol differs from C 28 and C 29 sterols by only one methyl or ethyl group at position C 24 . This alkylation is catalyzed by C-24 sterol methyltransferase (C 24 -SMT), an enzyme widely distributed in plants, microbial eukaryotes, and fungi but not known from nearly all animals (5,6). The exceptions are some marine sponges and the marine annelid Capitella teleta (7,8). Marine sponges have an unusual sterol composition enriched in highly branched alkylated sterols (9)(10)(11)(12), of which some are synthesized by C 24 -SMT homologs (13). As with most other enzymes required for sterol synthesis, an ancestral C 24 -SMT was likely present in the LECA (3) and is assumed to have been lost early in animal evolution following the diver-gence of sponges, explaining why eumetazoans are not able to produce C 28 and C 29 sterols (6,8).
Sitosterol is the main sterol in the marine gutless annelid Olavius algarvensis O. algarvensis belongs to a group of gutless marine annelids found worldwide, mainly in coral reef and seagrass sediments. These annelid worms lack a digestive system and are obligately associated with bacterial endosymbionts that provide them with nutrition (14)(15)(16)(17). As part of our ongoing research on the O. algarvensis symbiosis, we analyzed the metabolome of single worm individuals using both gas chromatography mass spectrometry and highperformance liquid chromatography mass spectrometry. These analyses revealed an unusual sterol composition, with sitosterol accounting for most of the sterols detected (60%), and the remainder consisting of cholesterol ( Fig. 1A and fig. S1). This was unexpected, as cholesterol generally dominates the sterol pool in bilaterians, often making up more than 90% of the total sterol content (18,19). Sitosterol is abundant in most plants but among bilaterians has only been reported as the most abundant sterol in a few phytoparasitic nematodes that are not capable of de novo sterol synthesis (20)(21)(22). In these plant parasites, it is unclear whether their sitosterol is only present in the nematode gut content or incorporated into their cells and tissues. The absence of a gut in O. algarvensis excludes sterol contamination from plant matter in the digestive tract.
O. algarvensis sterols have an isotopic composition that is distinct from their environment We next asked whether O. algarvensis could acquire its sterols from the environment through passive diffusion. Chemical analyses of porewater collected in the vicinity of seagrass meadows-the habitat of many gutless annelids (including O. algarvensis)-revealed that sterols were present in the environment in concentrations sufficient to sustain the growth of small sterol-auxotrophic invertebrates (supplementary text and fig. S2). Therefore, we further investigated the origin of sterols in O. algarvensis by analyzing the carbon isotopic signature (d 13 C) of sterols in the worms, their environment (which includes the seagrass Posidonia oceanica) and the porewater of the sediments these worms live in. Carbon isotopic signatures are used to reveal carbon sources and their paths through the food web. As a rule, the bulk d 13 C values of animals reflect their dietary sources [0.5 per mil (‰) to 2 ‰ difference] (23,24), but sterols are typically depleted in 13 C relative to bulk biomass by as much as 5 ‰ to 8 ‰ (25,26). Results from gas chromatography isotope ratio mass spectrometry (GC-IRMS) with single metabolite resolution showed that sitosterol in the seagrass and porewater had d 13 C values ranging from −30 ‰ to −15 ‰ (Fig. 1E and supplementary text). The sterols in O. algarvensis, as well as those in another co-occurring gutless annelid species, Olavius ilvae, had much lower d 13 C values: −38 ‰ to −36 ‰ for sitosterol and −40 ‰ to −31 ‰ for cholesterol (Fig. 1E). The difference in the isotopic signature of sterols in both Olavius species and their environment excludes that these worms acquired sterols from their environment, and instead indicates an endogenous origin. O. algarvensis, as all other Olavius and Inanidrilus species, derives all its nutrition from its chemosynthetic bacterial symbionts, and this is reflected in its bulk isotopic composition with d 13 C values of −30.6 ‰ (27). The 13 C-depleted signatures of both cholesterol and sitosterol by 1 ‰ to 10 ‰ compared with bulk biomass in O. algarvensis and O. ilvae led us to hypothesize that these animals synthesize both sterols de novo, using carbon derived from their chemosynthetic symbionts.
Bacterial symbionts are not the source of sitosterol in O. algarvensis Having shown that sitosterol in O. algarvensis and O. ilvae did not originate from the uptake of plant sterols from their environment, we next investigated whether their bacterial symbionts were the source of this C 29 sterol. The bacterial symbionts of these worms form a thick layer between the cuticle and the epidermis of the animal (Fig. 1B). To localize the distribution of sterols in O. algarvensis, we used two high spatial resolution metabolite imaging techniques. Time-of-flight secondary ion mass spectrometry (TOF-SIMS) data revealed that, at a spatial resolution of 0.4 mm, both sitosterol and cholesterol were uniformly distributed throughout the animals' tissues ( Fig. 1, C and D). We found no evidence for a tissue-specific distribution of these two sterolsthat is, there was no correlation between symbiont location and sitosterol distribution. These findings are supported by a second mass spectrometry imaging method, matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-2-MSI), of cross and longitudinal sections at a spatial resolution of 5 mm. The MALDI imaging data of longitudinal worm sections confirmed a uniform distribution of sitosterol and cholesterol throughout the animal (figs. S3 and S4) and the identity of these sterols (table S1). This homogeneous sterol distribution suggests that the bacterial symbionts are not the source for sitosterol in O. algarvensis.
As a second approach to investigate whether the bacterial symbionts are the source of sitosterol, we sequenced and assembled the genomes of the O. algarvensis symbionts and screened them for enzymes involved in de novo sterol synthesis. These analyses revealed that the symbionts, as in most bacteria, do not encode enzymes involved in de novo sterol synthesis (for more details see materials and methods).
O. algarvensis encodes and expresses enzymes involved in sitosterol synthesis that overlap with those of cholesterol synthesis Having ruled out a symbiotic origin and an environmental source of sitosterol in O. algarvensis, we next investigated whether the animals themselves can synthesize this typical plant sterol. To identify and characterize the biosynthetic pathways involved in sterol production, we sequenced and assembled the genome of O. algarvensis and analyzed metatranscriptomic and metaproteomic data to search for enzymes involved in de novo sterol synthesis. The host possessed the full enzymatic toolbox required for cholesterol and sitosterol synthesis. For cholesterol synthesis, the genome of O. algarvensis encodes homologs of 11 enzymes known to be involved in the synthesis of this sterol ( fig. S5). The intron-exon structure of these genes confirms their eukaryotic origin and excludes bacterial contamination ( Fig. 2A and table S2). The cholesterol biosynthesis pathway, starting with squalene, is a series of 10 connected enzymatic reactions encoded by 11 genes ( fig. S5 and table S3). Homologs of all enzymes were transcribed (11 out of 11 enzymes) and 5 out of 11 proteins were detected in the proteome of O. algarvensis ( fig. S5 and tables S4 and S5), indicating active expression of the genes involved in cholesterol synthesis. Phylogenetic analysis allowed us to assign each homolog to an ortholog group and thus to a potential function (figs. S6 to S15). Collectively, these data show that O. algarvensis has all the enzymes required for de novo cholesterol synthesis, which in combination with the isotopic signature of their cholesterol suggests that these annelids are able to synthesize cholesterol.
Notably, our analyses also identified a homolog of C 24 -SMT in the genome of O. algarvensis, an enzyme essential to sitosterol synthesis in plants (Fig. 2). As described above, most bilaterians are assumed to lack C 24 -SMT. C 24 -SMT catalyzes the transfer of a methyl group from S-adenosyl-L-methionine (SAM) to the sterol side chain and is essential for the biosynthesis of sitosterol and other C 28 -C 29 sterols commonly found in plants and fungi. As with the cholesterol synthesis genes, the intron-exon structure of the putative C 24 -SMT gene confirmed its eukaryotic origin and excluded bacterial contamination ( Fig. 2 and table S2). The putative C 24 -SMT gene is a 1071-base pair (bp) open reading frame (ORF) encoding a 356-amino acid polypeptide and contains all the conserved residues characteristic of C 24 -SMT as well as the 4 conserved signature motifs responsible for substrate binding (fig. S16) (28)(29)(30)(31). We identified the C 24 -SMT gene in O. algarvensis transcriptomes and proteomes, confirming that these animals express this enzyme ( Fig. 2 and tables S4 and S5). Our findings suggest that the O. algarvensis C 24 -SMT gene encodes a functional enzyme involved in sitosterol metabolism.
The O. algarvensis C 24 -SMT homolog is bifunctional and consecutively transfers methyl groups to sterol intermediates Two methylation reactions are required for the final steps of sitosterol synthesis, one that adds a methyl group at C-24 and one at C-28.    (table S6). However, the enzyme was able to methylate zymosterol and desmosterol, two intermediates of the cholesterol biosynthetic pathway. When incubated with either of these sterol substrates and SAM, the O. algarvensis C 24 -SMT produced a methylated sterol product (C 28 ) ( Fig. 2 and figs. S17 to S20). Zymosterol was methylated to fecosterol and desmosterol to 24-methylene-cholesterol. The shift in retention times and changes in mass spectra of the products indicated that a methyl group was added to their sterol side chain, likely at the C 24 -position (Fig. 2, and figs. S19 and S20). These results suggest that the cholesterol and sitosterol synthesis pathways overlap in O. algarvensis, as the two C 24 -SMT substrates, zymosterol and desmosterol, are intermediates produced in the second half of the animal cholesterol synthesis pathway (Fig. 2).
After confirming the first methylation step at C-24, we next searched for potential substrates for the second methylation step at C-28. This second methylation is essential as sitosterol is a C 29    presence of an ethyl group on its C-24 position. To test our hypothesis that both of these methylations are catalyzed by the O. algarvensis C 24 -SMT, we selected the product of the first methylation, 24-methylene-cholestrol, as well as campesterol, as potential substrates for the second methylation ( fig. S21). Only 24-methylenecholesterol, but not campesterol, was methylated by the O. algarvensis C 24 -SMT, producing a C 29 sterol compound, most likely (epi)clerosterol ( Fig.  2 and fig. S22). 24-methylene-cholesterol is the product of the methylation of desmosterol, providing evidence to support our hypothesis that in O. algarvensis, the C-24 and C-28 methylations are catalyzed by the same enzyme and occur consecutively. That is, the O. algarvensis C 24 -SMT first methylates desmosterol at C-24 to produce the C 28 sterol 24-methylene-cholesterol, and then adds a second methyl group to 24methylene-cholesterol at C-28, to produce the C 29 sterol (epi)clerosterol. (Epi)clerosterol differs from sitosterol by the presence of a double bond at position C-25 (26 or 27). This double bond is most likely removed by delta(24)-sterol reductase (DHCR24), which was expressed based on its presence in O. algarvensis transcriptomes and proteomes (figs. S5 and S15). Our results provide evidence for an animal C 24 -SMT that catalyzes the two methylation steps needed to synthesize sitosterol from a cholesterol intermediate, revealing a previously unknown pathway for C 28 -C 29 sterol synthesis in animals (Fig. 2).

C 24 -SMT homologs are widespread in annelids
Having demonstrated the activity of an animal C 24 -SMT homolog that enables O. algarvensis to synthesize sitosterol de novo, we asked whether other gutless annelids also encode functional C 24 -SMTs. To answer this question, we analyzed the sterol contents of six additional gutless annelid species. All six species had lipid profiles similar to that of O. algarvensis, with sitosterol as their major sterol (table S7 and supplementary text).
In addition to lipid profiling, we screened the transcriptomes of nine Olavius and Inanidrilus species and found that all expressed a C 24 -SMT homolog, including O. ilvae (table S8). Moreover, heterologous gene expression analyses confirmed the C-24 methylation ability of a C 24 -SMT homolog from another gutless annelid, O. clavatus. The C 24 -SMT from O. clavatus is also a bifunctional sterol methyltransferase, capable of methylating zymosterol, desmosterol, and 24-methylene-cholesterol (figs. S17, S18, and S22).
We next asked whether C 24 -SMT homologs are present in other annelids. We screened published transcriptomes and genomes and identified C 24 -SMT homologs in 3 deep-sea gutless tubeworm species and in 17 gut-bearing annelid species from marine, limnic, and terrestrial environments (table S9 and fig. S23). Despite the presence of C 24 -SMT homologs in these annelids, the published sterol profiles of annelids-including four gut-bearing species analyzed in this study-were dominated by cholesterol (table S9 and Fig. 1A). However, sitosterol as well as other C 28 and C 29 sterols accounted for a considerable proportion of total sterols (15 to 30%) in some of these species, including the vent and seep tubeworms Riftia pachyptila and Paraescarpia echinospica (table S9). These deep-sea siboglinid annelids are only distantly related to Olavius and Inanidrilus, but also lack a gut and gain all of their nutrition from their chemosynthetic symbionts (35). The sterol contents of these tubeworms are dominated by cholesterol and desmosterol, but the C 28 sterol campesterol as well as other C 28 and C 29 sterols make up as much as nearly one-third of their sterol contents (36)(37)(38). Our discovery of C 24 -SMT homologs in these deep-sea annelids suggest that these tubeworms are able to synthesize C 28 and C 29 sterols as well. Further support that annelid C 24 -SMT homologs methylate sterols was recently shown for the enzyme encoded by the nonsymbiotic annelid C. teleta (39). Taken together, these results indicate that functional C 24 -SMT homologs are widespread in the annelid phylum and that the presence of this gene is not restricted to hosts that harbor symbiotic bacteria. To assess the broader distribution of C 24 -SMT homologs in animals, we performed protein searches against public databases (see materials and methods for details). Hits were found in six additional animal phyla: sponges, cnidarians, rotifers, mollusks, nematodes, and chordata (supplementary text and table S10). For the latter two, we concluded they are not animal C 24 -SMTs for the following reasons. Among the chordata, we found a single homolog in a metatranscriptome from a fruit bat. It had 99.1% identity to sequences from a plant and fell in a clade with these, suggesting that this C 24 -SMT is a contamination (supplementary text). The nematode sequences belonged to a group of SMTs called C-4 sterol methyltransferases (C 4 -SMTs) that are specific to nematodes (40,41). Our phylogenetic analyses showed that these nematode C 4 -SMTs are not closely related to the monophyletic clade of C 24 -SMTs from plants, fungi, microbial eukaryotes, sponges, cnidarians, rotifers, annelids, and mollusks (Fig. 3B).

C 24 -SMTs are widespread across the tree of life
We reconstructed the evolutionary relationships of animal C 24 -SMTs from sponges, cnidarians, rotifers, mollusks, and annelids and compared these with previously described C 24 -SMTs from bacteria, microbial eukaryotes, fungi, and plants (Fig. 3A). Our phylogenetic analyses revealed that C 24 -SMTs are widespread across the tree of life and fall into nine well-supported clades (Fig. 3B, A to I). These clades were not congruent with the phylogeny of bacteria and eukaryotes, indicating a complex evolutionary history for C 24 -SMTs, which we discuss below. Many eukaryotes had more than one C 24 -SMT homolog, with members of some groups such as the choanoflagellates and cnidarians encoding as many as four homologs that belonged to different, phylogenetically distant clades (Fig. 3C). Most animal C 24 -SMTs clustered in seven clades (Fig. 3A and figs. S24 to S27), with each clade consisting of animal sequences only. The exceptions were animal C 24 -SMTs that fell on single, long branches that formed sister lineages to other eukaryotic clades (Fig. 3A, supplementary text, and  figs. S25 and S26).
For the animal C 24 -SMTs, two of the seven clades (2.F and 3.F) contain homologs from sponges and annelids whose C-24 methylation function has been experimentally verified [this study and (13,39)]. The five remaining clades consist of predicted C 24 -SMTs, but their high sequence homology to verified C 24 -SMTs together with conserved protein domains suggest that these homologs also methylate sterol intermediates at C-24.

The complex evolutionary history of C 24 -SMTs
In two animal phyla, Annelida and Rotifera, we observed strong congruence between their C 24 -SMT homologs and their evolutionary history ( Fig. 3D and figs. S27 and S28). For annelids, in addition to the gutless Olavius and Inanidrilus sequences from our study, we found C 24 -SMT homologs in nine annelid orders from limnic, terrestrial, and marine environments. Annelid C 24 -SMTs fell within two sister clades, one from Errantia and the other from Sedentaria, corresponding to two major subgroups of Annelida (Fig. 3D). The only exception was Megasyllis neponica, which belongs to the Errantia but has a C 24 -SMT that falls in the Sedentaria (Fig. 3D). Congruence was also visible within the two annelid clades (Fig. 3D and fig. S28). These results suggest that C 24 -SMTs from extant annelids evolved through direct inheritance from a common ancestor. Similarly, the rotifer C 24 -SMT clade consisted largely of sequences from the Bdelloidea, a phylogenetically well-defined class within the phylum Rotifera (supplementary text). Within this clade, the C 24 -SMT topology corresponded well to the phylogeny of bdelloid rotifers, consistent with direct inheritance driving the evolution of this gene within this animal phylum as well ( fig. S27).
At longer evolutionary time scales, there was no congruence between C 24 -SMT homologs and the evolutionary history of the lineages they came from, not even between closely related phyla. For example, Annelida and Mollusca are more closely related to each other than to Porifera, yet their C 24 -SMTs often clustered with those from Porifera. The lack of congruence at deeper nodes of the C 24 -SMT tree with the evolutionary history of eukaryotes is also visible in the interspersed phylogeny of C 24 -SMT homologs from fungi, plants, and animals throughout the tree (Fig. 3A). The consistent incongruencies throughout most of the C 24 -SMT tree are hard to reconcile with direct inheritance from a common eukaryotic ancestor. Repeated, independent events of lateral gene transfer (LGT) may provide a more likely explanation for the C 24 -SMT tree topology.
LGT is widespread and well-studied in bacteria, but less is known about LGT within extant eukaryotic lineages. Recent evidence from high quality genome sequencing suggests that LGT may play a more important role in Michellod    eukaryotes than previously assumed (42)(43)(44)(45).
LGT within eukaryotes is assumed to be most commonly mediated by viruses and bacteria, particularly in species that are intimately associated with beneficial or pathogenic microorganisms (46). However, we ruled out LGT mediated by associated bacteria in the gutless Olavius and Inanidrilus, as we exhaustively searched their symbiont metagenomes for C  LGT from bacteria to eukaryotes of C 24 -SMTs is also unlikely to have played a role in the more recent evolution of other animals, as bacterial C 24 -SMTs were not closely related to those from animals, with the exception of sponges (Fig. 3A). Alternatively, LGT mediated by viruses could explain how genes are transferred between eukaryotes (47). Independent of the precise route for LGT across eukaryotes, our study provides evidence for rampant LGT of gene homologs within the animal kingdom and suggests that the acquisition of C 24 -SMTs provides animals with a strong selective advantage.

Conclusions
Although cholesterol is commonly the most abundant sterol in animals, through studies of non-model organisms and technological advances, evidence is growing that C 28 and C 29 sterols are also often present, sometimes in considerable amounts (48)(49)(50)(51). Our findings highlight the value of reconsidering the source of C 28 and C 29 sterols in both extant animals and the fossil record and determining whether these originated from the animal's diet as is often assumed, or if they were synthesized by the animals themselves. It is also timely to reconsider the widespread use of the term phytosterols for C 28 and C 29 sterols, as we now know that microbial eukaryotes, plants, and animals can synthesize these lipids.
How can we explain the unusually high abundance of sitosterol in gutless Olavius and Inanidrilus? Studies have shown that C 28 and C 29 sterols can be incorporated into animal membranes (52,53) and provide beneficial effects when added to animal diets or cell lines. For example, they act as cholesterol-lowering agents, have anti-tumor, anti-inflammatory, antibacterial, and antifungal properties (54,55), and modulate interactions between bacterial pathogens and eukaryotic hosts (56). Therefore, the anti-inflammatory and antibacterial properties of sitosterol, as well as its ability to protect animal cells against toxins that target cholesterol (57), might play a role in the symbiosis between Olavius and Inanidrilus and their chemoautotrophic symbionts, by preventing the symbionts from entering the host cytoplasm. Changes in sterol composition also affect the fluidity and permeability of membranes, and these physical changes in turn affect many cellular processes. For example, sitosterol has been shown to enhance mitochondrial energy metabolism in a mouse cell line (58) and might enable Olavius and Inanidrilus to gain more energy under low oxygen concentrations in their environment. Our findings highlight the need for studies that elucidate the physiological and ecological roles of sitosterol in animals. Olavius and Inanidrilus are valuable model systems for studying the impact of C 28 and C 29 sterols on animal membrane properties in vivo and furthering our understanding of the roles sterols play in eukaryotic cells. support to N.D. from the Gottfried Wilhelm Leibniz Prize from the German Research Foundation (DFG) and a Gordon and Betty Moore Foundation Marine Microbiology Initiative Investigator Award (grant GBMF3811), and to M.L. from sea4society, a CDRmare campaign in the German Marine Research Alliance under Award number 03F0896D. M.K. received financial support from the National Institute of General Medical Sciences of the National Institutes of Health under award number R35GM138362 and the US National Science Foundation (grant IOS 2003107). Competing interests: The authors declare no competing interests. Data and materials availability: The metaproteomic mass spectrometry data have been deposited at the ProteomeXchange Consortium via the PRIDE partner repository (59) with the following dataset identifier: PXD014881. The sequencing data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession numbers PRJEB52460 (PolyA libraries) and PRJEB52678 (PacBio reads). The C24-SMT protein alignment and the original tree file can be found on Figshare (60). The metabolomics and other MS-based data are on Figshare. The sterol profiles (61), the sterol GC-IRMS data (62), the metabolite imaging data (63), and the enzyme assay results (64)