Introduction

Sterols are polycyclic triterpenoid lipids required by most eukaryotes for many critical cellular functions, such as maintaining membrane integrity, phagocytosis, stress tolerance, embryogenesis, and cellular signaling1,2,3,4,5,6. In addition, sterols are quite recalcitrant and are preserved as sterane hydrocarbons in sedimentary rocks up to 1.6 billion years old7. Given their persistence in the geologic record and wide distribution amongst extant eukaryotes, steranes are broadly interpreted as biomarkers for ancient eukaryotic life8,9, although diverse bacteria are also known to produce sterols10. Sterols with particular modifications to the side-chain, such as methylations at the C-24 and C-26 positions, have not been reported in bacteria and remain unique to certain eukaryotic lineages. Side-chain alkylated steranes, such as ergostane and stigmastane, can therefore function as more specific biomarkers11.

According to the sponge biomarker hypothesis, two side-chain alkylated sterane biomarkers, 24-isopropylcholestane (24-ipc; Fig. 1a) and 26-methylstigmastane (26-mes), are indicative of marine sponges of the Demospongiae class12,13. The co-occurrence of 24-ipc and 26-mes in the late Neoproterozoic (660-540 Ma)12,13, preceding the earliest sponge macrofossils by ~100 million years14, may represent the first evidence of animals in the geologic record. However, debate over this interpretation continues for several reasons14,15,16,17,18,19,20,21,22 including the potential for alternative sources. Demosponges are the only extant organisms known to contain sterols with the 24-ipc and 26-mes side-chain structures as their major sterols13,23,24,25,26,27, but pelagophyte algae, which produce 24-n-propylcholesterol as their major sterol, also contain 24-isopropyl sterols28,29,30,31. Trace amounts of 24-ipc and 26-mes were also reported in hydrogenated lipid extracts from rhizarian protists22, although attempts to replicate these results were not successful20. Subsequent studies demonstrated 24-ipc and 26-mes form abiotically from certain C29 sterols during laboratory pyrolysis experiments under conditions that may mimic diagenesis18,21. These studies suggest that the previously observed Rhizarian 24-ipc and 26-mes were generated through laboratory pyrolysis rather than by in vivo synthesis, and that Neoproterozoic 24-ipc and 26-mes may have formed abiotically from algal sterols. Further, microbial symbionts, including bacteria from diverse phyla, can constitute up to 40% of sponge biomass32, and recent analyses have identified bacterial sterol biosynthesis gene clusters in several demosponge metagenomes33,34. Bacterial symbionts may therefore contribute to 24-isopropyl and 26-methyl sterol biosynthesis in sponges, although bacterial enzymes that methylate the sterol side-chain have not been previously identified.

Fig. 1: Sterol methyltransferases from sponges and yet-uncultured bacteria perform multiple side-chain methylations in vitro.
figure 1

a Structure of the sponge biomarker 24-isopropylcholestane. b Predominant pathway to 24-isopropyl sterols by bacterial sterol methyltransferases (SMTs). c Representative extracted ion chromatograms (m/z 456, 470, 484, 498) of total lipid extracts from in vitro reactions performed with desmosterol and E. coli lysates containing an empty plasmid, an SMT identified in a transcriptome of the demosponge Chondrilla nucula, a bacterial SMT identified in a metagenome from the demosponge Aplysina aerophoba, a bacterial SMT identified in a freshwater metagenome-assembled genome (MAG) assigned to the phylum Chlamydiae, and a myxobacterial SMT identified in a marine MAG assigned to the genus Sandaracinus. d Representative extracted ion chromatograms (m/z 456, 470, 484, 498) of total lipid extracts from in vitro reactions performed with the same lysates but with 24-methylenecholesterol as the substrate. e Side-chain structures of sterols identified in c and d. All lipids were derivatized to trimethylsilyls prior to GC-MS analysis. The raw data used to generate this figure are provided in Supplementary Data 2. Mass spectra of identified sterols are shown in Supplementary Fig. 3.

Full size image

One approach to constrain uncertainties in the sponge biomarker hypothesis is to identify and characterize the enzymes responsible for these distinct alkylations at C-24 and C-26. Sterol 24-C-methyltransferases (SMTs) alkylate the C-24 position at an isolated double bond in an S-adenosylmethionine (SAM) dependent mechanism35 and have been characterized extensively in fungi and plants. Bioinformatic analyses performed to understand how sponges could be synthesizing 24-propyl sterols suggested that the number of SMT copies in a eukaryotic genome corresponds to the number of carbons added at C-24 in all analyzed lineages except sponges, which contain one less SMT in their genome or transcriptome than expected based on their sterol profile36. This prompted the hypotheses that at least two SMT copies in a genome are required to produce 24-propyl sterols, and that sponges each encode bifunctional SMTs that can perform multiple methylations at C-24. However, no experimental evidence verifying this activity by any sponge-derived SMT currently exists, and SMTs that produce 24-propyl sterols have not been identified in any extant organism.

Here, we characterize a variety of putative SMTs from sponges and bacteria to better understand the biochemical requirements for generating side-chain alkylated sterols, including 24-isopropyl sterols. We demonstrate that sponge SMT homologs are bona fide SMTs that can perform multiple rounds of methylation to produce sterols methylated and ethylated at the C-24 position. In addition, we show that SMTs from yet-uncultured bacteria, including demosponge symbionts, are capable of alkylating sterols at the C-24 position. We identify three symbiotic bacterial SMTs each capable of performing all three of the methylation steps required to produce the 24-isopropyl side-chain. Taken together, our results suggest yet-uncultured bacteria have the genomic capacity to synthesize side-chain alkylated sterols and demonstrate the utility of understanding the biochemistry of lipid biosynthesis to allow for more robust biomarker interpretations.

Results

Sponge SMTs are functional

Our first goal was to express putative sponge SMTs and verify methylation activity in vitro. Given the current lack of publicly available sequencing data from sponges known to contain 24-propyl sterols, the function of SMTs from these species could not be verified in this study. We therefore chose eight SMT homologs identified in publicly available genomes and transcriptomes from eight sponges of the Demospongiae, Homoscleromorpha, and Calcarea classes, each of which have been shown to contain 24-ethyl sterols at the species or genus level (Supplementary Table 1). In vitro reactions with cell lysates generated from Escherichia coli strains overexpressing each SMT were performed with two potential substrates, desmosterol and 24-methylenecholesterol, which have been previously shown to be suitable substrates for side-chain methylation in sponges37, pelagophyte algae29, and annelid worms38. Sterol products were analyzed using gas chromatography-mass spectrometry (GC-MS). All eight of the sponge SMTs methylated desmosterol to produce 24-methyl sterols including 24-methylenecholesterol and (epi)codisterol (Figs. 1c, 2). Six SMTs also methylated 24-methylenecholesterol to 24-ethyl sterols including fucosterol, isofucosterol, and (epi)clerosterol (Figs. 1d, 2). Each sponge SMT capable of producing 24-ethyl sterols from 24-methylenecholesterol could also bifunctionally methylate desmosterol. None of the sponge SMTs produced 24-propyl sterols, consistent with the sterols previously observed in these species (Supplementary Table 1).

Fig. 2: Sterols identified as products of sponge and bacterial sterol methyltransferase in vitro reactions.
figure 2

Sterols were detected using GC-MS. Products of in vitro reactions performed with both desmosterol and 24-methylenecholesterol as substrates are included. C28 sterols include (epi)codisterol (II), 24-methylenecholesterol (III), and 24-methyldesmosterol (V). C29 sterols include (epi)clerosterol (IV), fucosterol (VI), and isofucosterol (VII). C30 sterols include 24S−24-isopropylcholesta-5,25-dienol (VIII) and 24-isopropylcholesta-5,24-dienol (IX). Metagenomic SMTs identified in gene clusters with at least two other sterol biosynthesis homologs are bolded. *NMR confirmed epiclerosterol and not clerosterol as products of the “Aplysina aerophoba JGIcombinedJ30088_10000835 1 of 2” and “Chlamydiae sp., drinking water 1 of 2” sterol methyltransferases. **24-isopropylcholesta-5,23-dienol was also detected as a product of the “Chlamydiae sp., drinking water 1 of 2” sterol methyltransferase by NMR, but was not detected by GC-MS.

Full size image

Three SMTs from symbiotic bacteria produce 24-isopropyl sterols

Our analyses of sponge symbiont metagenomes revealed several SMT homologs, many of which occurred in gene clusters with homologs of other known sterol biosynthesis genes including the bacterial C-4 demethylases sdmA and sdmB39,40 (Fig. 3, Supplementary Data 1). We therefore hypothesized bacterial symbionts may contribute to side-chain alkylation in sponges. However, no bacterial enzymes, from either free-living or host-associated sources, have been experimentally shown to alkylate the sterol side-chain. We chose 10 putative bacterial SMTs identified in metagenomes from four demosponge species known to contain side-chain alkylated sterols (Supplementary Table 1) for analysis. Eight symbiont SMTs methylated desmosterol to 24-methyl sterols including (epi)codisterol, 24-methylenecholesterol, and 24-methyldesmosterol (Figs. 1c, 2). Three of these SMTs also methylated 24-methylenecholesterol to 24-ethyl sterols including fucosterol, isofucosterol, and (epi)clerosterol (Figs. 1d, 2). As with the bifunctional sponge SMTs, each bacterial symbiont SMT that produced 24-ethyl sterols from 24-methylenecholesterol also methylated desmosterol. Two SMTs identified in a metagenome from Aplysina aerophoba, a demosponge known to contain trace 24-ipc after catalytic hydropyrolysis13, also methylated 24-methylenecholesterol to two distinct C30 sterols (Figs. 1, 2). The first symbiont SMT produced 24S−24-isopropylcholest-5,25-dienol while the second produced 24-isopropylcholest-5,24-dienol. Both propylating SMTs were capable of all three methylation steps required to synthesize the 24-ipc side-chain via a single protein. One propylating SMT was identified in a sterol biosynthesis gene cluster suggesting a yet-unknown bacterial sponge symbiont has the genomic capacity to synthesize 24-isopropyl sterols de novo (Fig. 3, Supplementary Data 1).

Fig. 3: Gene clusters suggest de novo side-chain alkylated sterol biosynthesis in the bacterial domain.
figure 3

a Gene clusters identified in metagenomes and metagenome-assembled genomes (MAGs) containing three or more sterol biosynthesis homologs and including functional sterol methyltransferases (bold). Genes are named for their Saccharomyces cerevisiae (erg) or Methylococcus capsulatus Bath (sdmAB) homologs. Dotted sterol methyltranferases did not methylate desmosterol or 24-methylenecholesterol in vitro. Striped patterns correspond to the number of methylations performed by each sterol methyltransferase in vitro. Gradients indicate partial sequences. b Schematic of a hypothetical bacterial side-chain alkylated sterol biosynthesis pathway. Arrow colors at each step correspond to the homologs identified in a. Detailed information on the source of each metagenomic gene cluster is available in Supplementary Table 4. Homology for each gene is described in Supplementary Data 1.

Full size image

Identification of bacterial sponge symbiont SMTs capable of side-chain alkylation led us to question if bacteria from environments outside a sponge host could also produce side-chain alkylated sterols. Through additional searches, we identified SMT homologs in a variety of environmental metagenomes and metagenome-assembled genomes (MAGs). We tested 14 of these SMT homologs from marine, freshwater, thermal spring, and coral reef environments, five of which occur in sterol biosynthesis gene clusters (Fig. 3, Supplementary Data 1). Twelve of these SMTs alkylated the sterol side-chain to produce 24-methyl and/or 24-ethyl sterols (Figs. 1, 2). Each environmental SMT that methylated 24-methylenecholesterol also methylated desmosterol. Two of the functional SMTs were identified in a sterol biosynthesis gene cluster from a Sandaracinus sp. MAG assembled from a pelagic North Atlantic Ocean metagenome41 (Fig. 3). The MAG also contains homologs for demethylation at C-4 and C-14 (Supplementary Data 1), suggesting a marine myxobacterium has the genomic potential to produce side-chain alkylated regular sterols de novo. One SMT, identified in a drinking water filter MAG assigned to the bacterial phylum Chlamydiae, produced both 24S−24-isopropylcholest-5,25-dienol and 24-isopropylcholest-5,24-dienol (Fig. 1).

The identification of functional metagenomic SMTs likely derived from bacteria, including three that produce 24-isopropyl sterols, is significant given that no bacterial proteins had previously been shown to alkylate the sterol side-chain. To confirm that these SMTs are truly of bacterial origin, we performed phylogenetic analyses of the 24 metagenomic SMTs we tested. In our maximum-likelihood tree of SMT proteins the majority of the metagenomic SMTs are distinct and more closely related to each other than to the SMTs of eukaryotes known to produce side-chain alkylated sterols (Fig. 4a). Further, 23 of the 24 putative bacterial SMTs, including those from sponge metagenomes, form two clades separate from the sponge SMT clade. It is therefore unlikely that these metagenomic SMTs were acquired from sponges via horizontal gene transfer. However, the observed bootstrap values were not robust enough to draw any additional evolutionary conclusions, nor to eliminate any remaining uncertainty of the bacterial origin of these SMTs. We therefore used the recently developed classifier Whokaryote, which predicts whether a metagenomic contig is likely eukaryotic or prokaryotic based on intergenic distance, gene density, gene length, and k-mer frequencies. Whokaryote predicts all 19 of the contigs containing the metagenomic SMTs tested here to be of prokaryotic origin (Supplementary Table 2).

Fig. 4: Functional metagenomic SMTs are bacterial.
figure 4

a Maximum likelihood tree generated from a concatenated alignment of the conserved methyltransferase and C-terminal domains of sterol methyltransferase (SMT) proteins. Metagenomic SMTs are in blue, sponge SMTs are in red, and other eukaryotic SMTs are in black. Bolded labels indicate SMTs tested in this study, and the number of circles corresponds to the number of carbons the SMT added to desmosterol at the C-24 position in vitro. The numbers in parentheses indicate the number of proteins in collapsed clades. b Maximum likelihood tree of oxidosqualene cyclases (OSCs). OSC groups are assigned as in ref. 42. Bolded labels indicate OSCs in metagenome gene clusters or metagenome-assembled genomes with SMTs tested in this study.

Full size image

The genomic context of the SMTs we tested provides additional evidence of their bacterial origin. Of the 24 metagenomic SMTs, 13 were in gene clusters or MAGs with a homolog of oxidosqualene cyclase (OSC), a key enzyme in sterol biosynthesis responsible for cyclizing oxidosqualene to lanosterol10 (Fig. 3, Supplementary Data 1). OSCs cluster into robust phylogenetic clades with high bootstrap support42. Thus, phylogenetic analyses of OSCs associated with the SMTs we tested could confirm their bacterial origin. In our maximum-likelihood tree of OSC proteins (Fig. 4b), we see that seven of the OSCs associated with the SMTs analyzed here, including four from demosponge metagenomes, fall into the Group 1 clade which contains only bacterial OSCs. The remaining 2 OSCs, from soil and hot spring sediment metagenomes, fall into bacterial Group 2. The position of these OSCs provides additional evidence that the metagenomic SMTs we tested originate from a bacterial source.

Side-chain propylation mechanism

It was surprising to observe that one SMT was capable of propylating the sterol side-chain on its own as it has been hypothesized that at least two SMTs would be required for this reaction36. To better understand this unexpected biosynthetic pathway, we performed additional in vitro reactions with two of the bacterial propylating SMTs using 13C-labeled sterol substrates and analyzed the resulting products by nuclear magnetic resonance (NMR; Supplementary Table 3, Supplementary Figs. 2, 4–7). In experiments with [28-13C] 24-methylenecholesterol as the substrate of the A. aerophoba sponge symbiont SMT, the major C29 products were confirmed as isofucosterol and fucosterol present in a 7:1 ratio. Trace epiclerosterol, but not clerosterol, was also detected. The product of triple side-chain methylation was confirmed as 24S−24-isopropylcholesta-5,25-dienol. The label was found exclusively at C-25, and the 24 R isomer was absent (Supplementary Fig. 1a). Fucosterol was not a substrate for the A. aerophoba symbiont SMT, but [29-13C] isofucosterol yielded 24S−24-isopropylcholesta-5,25-dienol with the label distributed in a 5:1 label ratio between C-27 and C-26 (Supplementary Fig. 1b). This outcome shows interconversion of the C-28 and C-25 cations by hydride-shifts is not involved in the biosynthesis of this compound, and deprotonation of the initially formed cation is unspecific.

The same set of reactions with the propylating Chlamydiae MAG SMT gave similar results. Fucosterol was not a substrate and epiclerosterol, but not clerosterol, was detected. However, minor differences were measured between these two bacterial SMTs. In experiments with [29-13C] isofucosterol and the Chlamydiae MAG SMT, the ratio of isofucosterol and fucosterol was 30:1, and the ratio of label at C-27/C-26 of 24S−24-isopropylcholesta-5,25-dienol was 6:1. In addition, the C30 Δ23 and Δ24 isomers, 24-isopropylcholesta-5,23-dienol and 24-isopropylcholesta-5,24-dienol, were also detected. The label was detected in the Z-vinylic methyl of 24-isopropylcholesta-5,24-dienol and in one of the two methyls of the Z-isopropyl group of 24-isopropylcholesta-5,23-dienol. These results provide evidence for one hydride-shift to generate the C-24 cation. The ratio of 24S−24-isopropylcholesta-5,25-dienol and its Δ24 and Δ23 isomers was estimated to be 20:4:1, respectively. Though minor mechanistic differences between these two propylating SMTs exist, the predominant pathway to 24-isopropyl sterols observed here proceeds from desmosterol to 24-methylenecholesterol to isofucosterol to 24S−24-isopropylcholest-5,25-dienol (Fig. 1b).

Discussion

The results presented here experimentally verify that sponges do indeed encode functional sterol methyltransferases capable of sequentially methylating desmosterol to 24-methylenecholesterol to 24-ethyl sterols, similarly to the bifunctional SMTs recently identified in annelid worms38. However, as in Annelida43, this activity was inconsistent among sponges. All five of the demosponge SMTs we tested sequentially methylated sterols. However, this activity was not observed by the SMTs from Oscarella carmela (Homoscleromorpha) and Sycon ciliatum (Calcarea), contrary to a previous hypothesis that all sponges encode bifunctional SMTs36. Sterol profiles from O. carmela and S. ciliatum have not been reported, but 24-ethyl sterols have been identified in other members of Oscarella and Sycon44,45. Sponge-derived bifunctional SMTs that produce 24-ethyl sterols may therefore be restricted to certain species of Homoscleromorpha and Calcarea, although we recognize these SMTs may require a different substrate than we provided in our assays in order to sequentially methylate sterols. Substrate specificity could also explain why no sponge SMT produced 24-isopropyl sterols in our experiments, although this result was unsurprising given that we did not test SMTs from sponges known to contain propylated sterols due to the lack of available genomic data. Future sequencing of genomes or transcriptomes from demosponges known to contain sterols with the 24-ipc carbon skeleton may reveal sponge-derived SMTs sufficient for the biosynthesis of 24-propyl sterols.

While we were unable to demonstrate the production of propylated sterols through the expression of sponge SMTs, we did show that bacterial symbionts have the genomic capacity to synthesize these lipids and may therefore contribute to the 24-isopropyl sterol pool in demosponges. Previous studies have provided evidence of de novo sterol biosynthesis in demosponges37,46,47, but the experiments were performed prior to the development of 16S community sequencing using cell-free sponge extracts or whole sponge tissue from which the presence of bacterial symbiont-derived enzymes cannot be excluded. Without confirmation that the bacterial symbionts in these samples did not contain functional SMTs, the possibility remains that bacterial symbionts contribute to sterol biosynthesis in demosponges. Further, holo-transcriptomic sequencing of the demosponge Vaceletia sp. found that some sterol biosynthesis transcripts were sponge-derived, while others, including those for squalene synthase, squalene monooxygenase, and oxidosqualene cyclase, were derived from bacteria48. Both sponge and bacterial symbiont proteins may therefore be required to produce sterols in demosponges, similarly to what has been observed in certain deep-sea mussels49,50. However, the metagenomic gene clusters containing functional SMTs from A. aerophoba symbionts identified here, along with steroid gene clusters containing SMT homologs previously identified in bacterial MAGs associated with demosponges of the Ircinia genus33, suggest bacterial symbionts may produce 24-isopropyl sterols de novo without demosponge input. Aplysterol, a 26-methyl sterol, is the major sterol of A. aerophoba, and trace 24-n-propyl sterols have also been reported in this demosponge51. While trace 24-ipc was reported after catalytic hydropyrolysis of A. aerophoba biomass, 24-isopropyl sterols have not been reported as components of A. aerophoba’s native sterols. It is therefore unlikely that bacterial symbionts produce significant amounts of 24-isopropyl sterols in this sponge. However, bacterial sterol biosynthesis gene clusters are enriched in demosponges compared to pelagic marine environments33, and it remains to be seen if bacteria produce 24-isopropyl sterols in other demosponges. Our work highlights the need for robust sequencing of both the sponge host and its microbiome coupled to sterol analysis in demosponge species that contain 24-isopropyls as their major sterols. These coupled genomic and lipidomic analyses from relevant sponges would help confirm whether sponges, their bacterial symbionts, or some combination of both are potential sources of 24-ipc biomarkers.

The production of 24-isopropyl sterols by an SMT identified in a freshwater Chlamydiae MAG was surprising given that this bacterium is unlikely to produce sterols de novo due to the absence of other sterol biosynthesis homologs in the MAG. All characterized members of Chlamydiae are obligate endosymbionts of eukaryotes, and none have been shown to produce sterols. However, some Chlamydiae are known to recruit sterols from their eukaryotic host into their membranes52. Further, homologs of sterol reductases have been identified in members of this phylum, including in demosponge-associated MAGs53. Chlamydiae are found in both freshwater54,55 and marine sponges56,57,58, and a sterol identified in the demosponge Callyspongia implexa was shown to have antichlamydial activity59. The potential roles of chlamydial sterol modification during pathogenesis and symbiosis therefore warrant further investigation, and it remains to be seen if Chlamydiae methylate host-acquired sterols to give the 24-isopropyl side-chain in sponges and other eukaryotic hosts.

All three propylating SMTs we identified originated from bacterial symbionts associated with eukaryotic hosts, and SMTs from free-living bacteria capable of this reaction have yet to be discovered. However, we did identify SMTs capable of producing 24-methyl and 24-ethyl sterols that likely originated in free-living bacteria. The two functional SMTs present in a sterol biosynthesis gene cluster from a marine Sandaracinus sp. MAG are particularly significant given that myxobacteria are rare in host-associated environments60, and that several myxobacteria, including a Sandaracinus species, are known to produce sterols10. Further, a recent study demonstrated de novo cholesterol biosynthesis in the marine myxobacterium Enhygromyxa salina and identified bacterial sdmB homologs capable of complete demethylation at C-440. Given that cultured marine myxobacteria are known to produce complex sterols fully demethylated at C-4 and C-14, and that a yet-uncultured marine Sandaracinus species encodes functional SMTs, we hypothesize that marine myxobacteria are the most promising candidates for de novo side-chain alkylated sterol biosynthesis in the bacterial domain. However, this activity remains to be confirmed in a myxobacterial isolate.

The identification of three bacterial SMTs each capable of producing the 24-isopropyl side-chain via a single protein is also significant. Molecular clock analyses suggest demosponges were the likely source of Cryogenian 24-ipc biomarkers given that demosponges acquired a second SMT copy prior to their deposition, while pelagophyte algae did not acquire a third SMT copy until the Phanerozoic36. In addition, researchers have argued against rhizarian protists as a source of 24-isopropyl sterols given that Bigelowiella natans encodes only two SMT copies and is thus unlikely to synthesize 24-propyl sterols20. These interpretations rely on the hypothesis that multiple SMT copies are required to produce 24-isopropyl sterols. However, our results demonstrating one SMT is sufficient for side-chain propylation make SMT copy number an unreliable predictor of an organism’s ability to produce 24-propyl sterols. Now that propylating SMTs have been identified, site-directed mutagenesis as performed with yeast61,62 and plant63 SMTs may reveal specific amino acid changes that allow an SMT protein to perform a third methylation. If such key residues are identified, they could act as a more reliable predictor of 24-isopropyl sterol biosynthesis directly from sequencing data than SMT copy number alone.

Finally, although we identified SMTs that produce 24-isopropyl sterols, proteins that methylate sterols to give other geologically important side-chain structures remain undiscovered. Identification of enzymes that methylate sterols at the C-26 position to give the 26-mes side-chain structure remains especially important when considering the biomarker evidence for Precambrian sponges. SMTs from sponges and bacteria produced (epi)codisterol suggesting both groups have the genomic capacity to produce substrates for hypothetical 26-SMTs37,64. However, it remains to be seen if sponges and bacteria, sponge-associated and otherwise, encode SMTs capable of methylation at C-26. If bacterial 26-SMTs are identified, verified side-chain alkylated sterol biosynthesis in a bacterial isolate would remain necessary in order to determine if extant bacteria produce 24-ipc and 26-mes sterol precursors in ratios corresponding to the C30 sterane distributions observed in the Cryogenian. The identification of any unique isotopic signatures resulting from the reactions of 24- and 26-SMTs could also confirm bacteria as a contributing source of these and other side-chain alkylated biomarkers. Until then, eukaryotes remain the most likely source of side-chain alkylated steranes, and the co-occurrence of 24-ipc and 26-mes in high relative abundance continues to provide compelling evidence for derived sponges in the Cryogenian, consistent with molecular clock analyses65,66,67. However, the potential for sterol side-chain methylation by symbiotic and free-living bacteria should no longer be ignored both when analyzing the sterols of extant eukaryotes and when interpreting the presence of ergostane, stigmastane, and 24-isopropylcholestane in the geologic record.

Methods

Bioinformatics analysis

SMT homologs were identified in the Joint Genome Institute Integrated Microbial Genomes & Microbiomes (JGI IMG; https://img.jgi.doe.gov), GenBank (https://www.ncbi.nlm.nih.gov/genbank), and Compagen (previously at http://compagen.org, now at http://compagen.unit.oist.jp) databases using the Saccharomyces cerevisiae S299C SMT amino acid sequence as the BLASTP search query (locus tag: YML008C, maximum e-value: 1e−50, minimum percent identity: 30%). The BLAST search of the JGI IMG metagenomic database resulted in 7966 SMT homologs as of May 2022. A subset of SMT homologs that contained the conserved methyltransferase and C-terminal domains and covered multiple environments were chosen for further analysis, with particular focus on SMT homologs from sponge microbiomes and those present in sterol biosynthesis gene clusters. Additional demosponge SMT sequences from C. nucula, S. fasciculatas, and P. ficiformis transcriptomes were provided by Gold36. Identifying sequence information is available in Supplementary Table 4. Conserved sterol methyltransferase C-terminal and methyltransferase 11 domains were identified using HmmerWeb.

Oxidosqualene cyclase homologs were identified through BLASTP searches of the metagenome scaffolds and MAGs containing the SMTs analyzed in this study and all bacterial and eukaryotic genomes available in JGI IMG as of October 2019 using the Methylococcus capsulatus Bath OSC protein sequence (locus tag: MCA2873, maximum e-value: 1e−50, minimum percent identity: 20%, minimum sequence length: 350 aa). Redundancy was decreased through manual removal of OSC homologs with 100% similarity at the amino acid level. The N-termini of fused SMO/OSC homologs were trimmed out of the dataset to begin the sequences at the squalene-hopene cyclase N-terminal domain as identified using HmmerWeb.

All protein sequence datasets were aligned via MUSCLE. Conserved SMT domains were aligned separately, then concatenated using Geneious. Maximum-likelihood trees were generated using IQ-Tree on XSEDE with 1000 ultra-fast bootstrap replicates using the best amino acid substitution models, gamma shape parameters, and invariable sites proportions under the Bayesian Information Criterion and Akaike Information Criterion using ModelTest-NG on XSEDE (Supplementary Table 5). Resulting phylogenetic trees were visualized in iTOL68.

All contigs containing metagenomic SMTs were retrieved from JGI IMG or GenBank and analyzed with Whokaryote69 using the default inputs with the minimum contig size decreased to 2000 base pairs. Calculated features and predictions are shown in Supplementary Table 2.

Gene synthesis and molecular cloning

SMT DNA sequences were codon-optimized for expression in E. coli and artificially synthesized by GeneArt (ThermoFisher Scientific) or through the Department of Energy Joint Genome Institute (DOE JGI) DNA Synthesis Science Program. SMT sequences from DOE JGI were obtained in the IPTG-inducible plasmid pSRKGm-lacUV5-rbs570 in E. coli TOP10. SMT sequences from GeneArt were subcloned into pSRKGm-lacUV5-rbs5 by sequence and ligase independent cloning (SLIC)71 and transformed into electrocompetent E. coli DH10B (Invitrogen) using a MicroPulser Electroporator (BioRad). Oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). PCR was performed according the to the manufacturer’s protocol using Phusion DNA Polymerase (New England Biolabs). DNA fragments were purified using the GeneJET Gel Extraction Kit (ThermoFisher Scientific). Plasmid DNA was isolated using the GeneJET Plasmid Miniprep Kit (ThermoFisher Scientific). Plasmid DNA was sequenced to confirm promoter and SMT sequences by ELIM Biopharm (Hayward, CA) using the following primers: 5′-AATGCAGCTGGCACGACAGG-3′ (forward) and 5′-CCAGGGTTTTCCCAGTCAC-3′ (reverse).

Bacterial culture and heterologous expression

E. coli expression strains were cultured in 50 mL terrific broth (TB) supplemented with gentamycin (15 μg/mL) at 37 °C while shaking at 225 rpm. Cultures were induced with 500 μM isopropyl β-D-1-thiogalactopyranoside (IPTG) at an OD600 of ~0.6, then incubated an additional 4 h at 30 °C while shaking at 225 rpm. Cells were harvested by centrifugation at 4500 × g for 10 min at 4 °C. Cell pellets were stored at −80 °C until sonication.

Sterol methyltransferase assay

Cells pellets were resuspended in 5 mL buffer containing the following: 50 mM Tris-HCl, 2 mM MgCl2, 20% glycerol (v/v), and 0.1% β-mercaptoethanol (v/v), pH 7.5. Cells were then lysed on ice via a Qsonica Sonicator Q500 equipped with a 3.2 mm probe at 30% amplitude pulsing at 5 s on, 15 s off for 8 min of total on time. Lysates were then partially clarified by centrifugation at 4500 × g for 20 min at 4 °C. Total protein concentration in the resulting supernatant was quantified using a Coomassie (Bradford) Protein Assay Kit (Thermo Scientific) according to the manufacturer’s Standard Microplate Protocol and a BioTek Synergy HT Microplate Reader. SMT assays were then immediately prepared using enough supernatant to give 4500 μg total protein. Other reaction components included 100 μM desmosterol (Sigma-Aldrich), 24-methylenecholesterol (Avanti Polar Lipids, Inc.), [28-13C] 24-methylenecholesterol (Avanti Polar Lipids, Inc.), or synthesized [29-13C] fucosterol or isofucosterol, 100 μM S-(5′-sdenosyl)-L-methionine chloride dihydrochloride (Sigma-Aldrich), and 0.1% Tween-80 (v/v) to a total reaction volume of 400 μL. Reactions were held at 30 °C for 20 h and then stored at −20 °C until lipid extraction.

Lipid extraction

Lipids were extracted using a modified Bligh-Dyer method72,73 with the completed in vitro reactions as the water phase. Reactions were sonicated in 10:5:4 (vol:vol:vol) methanol:dichloromethane:water for 1 h. The organic phase was then separated with twice the volume of 1:1 (vol:vol) dichloromethane:water followed by storage at −20 °C for >1 h. Following centrifugation at 2800 × g for 10 min at 4 °C, the organic phase was transferred and evaporated under N2 to give total lipid extracts (TLEs). TLEs were derivatized to trimethylsilyl ethers in 1:1 (vol:vol) pyridine:Bis(trimethylsilyl)trifluoroacetamide for 1 h at 70 °C prior to GC-MS analysis.

GC-MS analysis

Lipids were separated with an Agilent 7890B Series GC equipped with two Agilent DB-17HT columns (30 m × 0.25 mm i.d. × 0.15 μm film thickness) in tandem with helium as the carrier gas at a constant flow of 1.1 ml/min and programmed as follows: 100 °C for 2 min, then 12 °C/min to 250 °C and held for 10 min, then 10 °C/min to 330 °C and held for 17.5 min. 2 μL of each sample was injected in splitless mode at 250 °C. The GC was coupled to an Agilent 5977 A Series MSD with the ion source at 230 °C and operated at 70 eV in EI mode scanning from 50 to 850 Da in 0.5 s. Data were analzyed using Agilent MassHunter Qualitative Analysis. Sterols were identified based on retention time and comparison to previously published spectra and laboratory standards. In vitro reactions for each SMT with both sterol substrates were performed at least twice, and the same sterol products were observed by GC-MS each time.

NMR analysis

In addition to GC-MS analysis, structures of sterol products formed by the metagenomic sterol methyltransferases referred to as “Aplysina aerophoba JGIcombinedJ30088_10000835 1 of 2” and “Chlamydiae sp., drinking water 1 of 2” were confirmed by NMR. 13C-labeling experiments were used to assist the identification of the biosynthetic products by 2D-NMR (heteronuclear single quantum coherence (HSQC) and heteronuclear multiple bond correlation (HMBC)) by increasing the amount of 13C at a specific position by two orders of magnitude compared to natural abundance. The two terminal isopropyl groups of 24-isopropyl sterols come from biosynthetically disparate origins, the isoprenoid pathway and SAM methylation. Because of the possibilities of hydride-shifts and different positions of deprotonation, each of the four terminal methyl groups can have multiple different origins within the product27. In addition to enhanced sensitivity, 13C-labeling provided mechanistic details of a pathway which, because of the intrinsic symmetry of 24-isopropyl sterols, would be otherwise hidden.

Total lipid extracts from pooled in vitro reactions using [28-13C] 24-methylenecholesterol or [29-13C] isofucosterol as the substrate for both SMTs were fractionated using small silica pipet columns (1 mL) eluting with 5 mL each of hexanes:ethyl acetate 9:1, 2:1, and 1:1 (vol:vol). Steroidal constituents were found in the 2:1 solvent fraction and concentrated with a stream of N2 prior to further purification by preparative thin-layer chromatography (TLC; 4:1 hexanes:ethyl acetate). Sterols were separated by high-performance liquid chromatography (HPLC) and the fractions were analyzed by NMR39. Sterols were identified by comparison of the HSQC-DEPT and HMBC spectra with reference compounds.

In experiments with [28-13C] 24-methylenecholesterol, the presence of isofucosterol was confirmed by a 1H-13C cross peak corresponding to C-28 at 5.108 and 116.48 ppm. The presence of fucosterol was shown by an HSQC cross peak at 5.183 and 115.58 ppm. Ratios of isofucosterol to fucosterol were determined by comparison of the integral ratios of the cross peaks39 for these two isomers. The presence of epiclerosterol was confirmed by cross peaks for C-28 at 1.290/1.380 and 26.01 ppm. Structures of the C29 sterol products were confirmed by long-range HMBC correlations from the 29-methyl group. Isofucosterol showed an HMBC cross peak for H-29 and C-28 at 1.59 and 115.58 ppm; fucosterol and epiclerosterol showed HMBC cross peaks at 1.57 and 116.48 ppm and 0.799 and 26.01 ppm, respectively. The presence of 24S−24-isopropylcholesta-5,25-dienol labeled at the C-25 position was confirmed by HMBC from the 27-methyl group (cross peak between 1.570 and 147.38 ppm). The presence of 24-isopropylcholesta-5,24-dienol was confirmed by HMBC showing two adjacent cross peaks correlating the C-26 and C-27 methyl protons (1.633/1.651 ppm) with C-25 (123.1 ppm).

In experiments with [29-13C] isofucosterol, the label distribution between the C-27 and C-26 positions of 24S−24-isopropylcholesta-5,25-dienol was shown by integration of the HSQC signals for C-27 (1.570 and 18.95 ppm) and C-30 (4.602/4.740 and 111.85). The label on the Z-vinylic methyl of 24-isopropylcholesta-5,24-dienol was detected by HSQC (1.651 and 19.73 ppm). 24-Isopropylcholesta-5,23-dienol was detected by label at 20.96 ppm exclusively, which is one of the two methyls of the Z-isopropyl group. The ratio of 24S−24-isopropylcholesta-5,25-dienol, 24-isopropylcholesta-5,24-dienol, and 24-isopropylcholesta-5,23-dienol was estimated by integration of the HSQC.

Preparation of [29-13C] fucosterol and isofucosterol

Specifically labeled C29 sterol precursors were prepared by modification of the procedure reported in ref. 74. The precursor 24-formylcholesterol i-methyl ether was obtained either through hydroboration of i-Me 24-methylenecholesterol followed by Dess-Martin periodinane oxidation or ozonolysis of i-Me 24-vinylcholesterol. The latter reaction was carried out in 90:9:1 (vol:vol:vol) dichloromethane/methanol/pyridine and yielded the desired 28-formylcholesterol i-methyl ether and its methyl hemiacetal in a 1:5 ratio. The hemiacetal was isolated as a mixture of four isomers. 1H-NMR (600 MHz): 8.06-8.03 (1H, 28-OH), 4.68-4.63 (1H, 28-H), 3.56-3.52 (3H, 28-Ome) 3.332 (s, 3H, 6-Ome), 2.767 (s, 1H, 6-H), 1.108 (s, 3H, 19-H), 0.710 (s, 3H, 18-H). This mixture was reacted with 13C-methylmagnesium iodide to give four isomers of [29-13C] i-Me 28-hydroxysitosterol: 24R,28R (H-28: 3.938 ppm; C-29: 21.97 ppm), 24S,28S (H-28: 3.938 ppm; C-29: 22.20 ppm), 24R,28 S (H-28: 3.820 ppm; C-29: 21.05 ppm), and 24S,28R (H-28: 3.803 ppm; C-29: 21.01 ppm). Preparative TLC (4:1 hexanes:ethyl acetate) gave two bands corresponding to a mixture of the RR and SS isomers (higher Rf) and a mixture of the RS and SR isomers (lower Rf) in ca. 2:1 ratio. These two mixtures were separately dehydrated with phosphorus oxychloride to obtain after deprotection [29-13C] fucosterol (C-29: 13.16 ppm) and [29-13C] isofucosterol (C-29: 12.76 ppm), respectively.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.