Evidence for Enzymatic Backbone Methylation of the Main Membrane Lipids in the Archaeon Methanomassiliicoccus luminyensis

Methylation of biomolecules is a common biochemical reaction typically catalyzed by methyltransferases or radical S-adenosylmethionine (radical SAM) enzymes. Whether the reaction is catalyzed by a methyltransferase or a radical mechanism depends on the chemical reactivity of the substrate (17). If an enzymatic methylation is responsible for the production of BDGTs and PDGTs, it would require activation of an inert sp³ carbon, the methylene group located at the C-3 position of the glycerol backbone of GDGTs. Hence, a radical mechanism is most plausible. In lipid biosynthesis, such a radical methylation was shown to be responsible for the production of hopanoids methylated at their C-2 or C-3 position (18, 19). Recently, two studies showed the involvement of radical SAM enzymes in the biosynthesis of GDGTs in Sulfolobus acidocaldarius (20, 21), and a radical mechanism was previously postulated for the lipid biosynthesis of Methanothermobacter thermautotrophicus (22, 23). A radical SAM-mediated methylation of GDGTs was thus suggested by Coffinet et al. (11) to be involved in the biosynthesis of BDGTs and PDGTs, further supported by the presence of several radical SAM enzymes in the M. luminyensis genome.

To test whether BDGTs and PDGTs derive from methylation(s) of GDGTs, a pure culture of M. luminyensis was amended with [methyl-¹³C]methionine, the known precursor of radical SAM enzymes, and grown under optimal conditions (see details in Materials and Methods). A pure culture of M. luminyensis was grown in parallel under identical conditions but without addition of any labeled substrate as a control batch. Core lipid distribution was determined by ultrahigh-performance liquid chromatography–high-resolution mass spectrometry (UHPLC-HRMS). Both cultures exhibited similar profiles in agreement with the lipid distribution previously published (16). ¹³C-label incorporation was detected by inspection of the isotopologue distributions of GDGTs, BDGTs, and PDGTs (Fig. 1), that is, the relative abundance of the molecular ions containing a different number of ¹³C atoms for each compound of interest recorded by HRMS analysis. Comparison of the isotopologue distribution between the labeled batch and the control batch (Fig. 1) revealed incorporation of ¹³C in BDGT and PDGT molecular ions (Fig. 1D to G) during incubation with [methyl-¹³C]methionine, while GDGTs did not show any label incorporation (Fig. 1B and C). The isotopologue distribution observed for BDGTs (Fig. 1D) was shifted by one unit, consistent with the transfer of one ¹³C-methyl carbon to one glycerol backbone of GDGTs to form BDGTs. The isotopologue distribution of PDGTs was shifted by two units, consistent with the addition of two ¹³C-methyl groups (Fig. 1F).

Tandem mass spectrometry was performed to examine the position where the ¹³C-labeled methyl group(s) was added. In the BDGT molecule (Fig. 2), incorporation of ¹³C was only observed in the isotopologue distribution of the butanetriol biphytanyl glycerol diether fragment (Fig. 2C), confirming that the ¹³C-methyl group was selectively added at the backbone. Contrary to the structure of BDGTs, the PDGT structure remains putative. Initial elucidation of the structure was published by Zhu et al. (12) based on tandem mass spectrometry analysis of a sediment sample from the Peru margin. MS² spectra of the PDGTs in the incubation with [methyl-¹³C]methionine, as well as in the unlabeled control incubation (Fig. SA1 in the supplemental material), revealed diverging patterns compared to the original MS² spectrum (12). Notably, absence of a fragment at m/z 1228 and presence of a fragment at m/z 1243 with an isotopologue distribution centered at the [M+1] ion in the incubation with [methyl-¹³C]methionine suggest an alternative structure for PDGTs where one methyl group would be added to each of the two lipid backbones instead of two methyl groups to the same glycerol moiety as previously hypothesized for PDGTs found in environmental samples (12) (Fig. SA1). To date, in environmental samples, as well as in M. luminyensis, concentration of PDGTs was too low to confirm these putative structures by NMR analysis. If this alternative structure were to be confirmed, the name of the lipid with an exact mass of 1,329.3467 Da in the M. luminyensis lipidome would have to be modified to butanetriol dibiphytanyl butanetriol tetraether (BDBT). In any case, the specific incorporation of methyl group(s) at the backbone(s) position observed in the present incubation provides direct evidence that BDGT and the putative PDGT compounds result from an enzymatic methylation of the common archaeal GDGT lipid, conceivably performed by a radical SAM enzyme.

The ability to produce BDGTs and PDGTs has, so far, only been observed in M. luminyensis; Becker et al. (16) suggested these lipids may be specific to the order Methanomassiliicoccales after screening 25 archaeal strains representing several archaeal phyla. In this study, we analyzed the lipid composition of another Methanomassiliicoccales representative, “Candidatus Methanogranum gryphiswaldense,” enriched from peat soil (24, 25) (see supplementary methods for details of the enrichment procedure). Absence of BDGTs and PDGTs in this enrichment was revealed by analysis of the acid-hydrolyzed lipid extract of its biomass (Fig. SA2). Interestingly, “Ca. Methanogranum gryphiswaldense” belongs to the family Methanomethylophilaceae, while M. luminyensis belongs to the family Methanomassiliicoccaceae. We therefore suggest that the radical SAM enzyme(s) responsible must be restricted to a limited number of archaeal taxa, including the family Methanomassiliicoccaceae.

An extended bioinformatics search relative to our previous paper (11) of the M. luminyensis genome for the radical SAM motif (protein family [Pfam] identifier pfam04055 and close homologs) identified 37 putative radical SAM enzymes. To further discriminate which gene could be responsible for backbone methylation of GDGTs, we performed a BLASTP search of these 37 proteins in the genomes of “Ca. Methanogranum gryphiswaldense” (25), as well as of the 25 strains previously shown to be devoid of BDGTs and PDGTs (16) (Fig. SA3). Seven radical SAM enzymes identified in M. luminyensis were found to share little homology, based on their E values, with proteins from the other 26 strains and are likely candidates to perform the backbone methylation (Table SA1). Interestingly, a BLASTP search of these seven proteins against the National Center for Biotechnology Information (NCBI) non-redundant protein sequence database revealed close homologues in genomes of some methanogenic archaea and anaerobic methane-oxidizing archaea and with members of “Candidatus Bathyarchaeaota” phylum (Fig. SA4). This phylogenetic distribution is in agreement with the BDGT distribution and isotopic composition observed in environmental samples so far (11, 13). Notably, Meador et al. (13) suggested, on the basis of the BDGT abundance patterns in relation to bathyarchaeotal 16S genes in sediment samples, that members of “Candidatus Bathyarchaeaota” are candidates for their production. This preliminary bioinformatic analysis will have to be complemented by microbial genetics once the genetic manipulation of a BDGT-producing microorganism has been developed.

M. luminyensis biomass was harvested at two time points in the course of the incubation, during the exponential phase and during the stationary phase (Fig. 3A and B). An increase in the relative abundance of BDGTs and PDGTs in the stationary phase was observed in incubations with and without label addition (Table SA2). In the present incubations, we observed a decrease in the concentration of both the major carbon (acetate) and energy (methanol) sources (26) of M. luminyensis (Fig. SA5). This suggests that additional methylation of the membrane lipid backbone might be a response to limitation in energy or carbon substrate availability. Previous studies reported modification of the membrane lipid structures when energy or/and nutrients were scarce, notably via methylation mechanisms. For example, a methylation reaction converts cis-unsaturated fatty acids to cyclopropane derivatives in many bacteria once they reach the stationary growth phase (27). In the archaeal realm, studies on cultured species from the euryarchaeal, crenarchaeal, and thaumarchaeal phyla showed that the degree of cyclization of GDGTs was increasing when nutrient availability and/or energy supply were limited (2–4). However, secondary processes, in particular, pH modification, could also influence the lipid distribution in the membrane and cannot be ruled out at the moment. Other variables whose impact on the lipid distribution are currently not well understood, such as accumulation of metabolic waste products in batch experiments, could also warrant further investigation. Dedicated experiments using continuous cultivation in a chemostat could help to further constrain the environmental parameter(s) that are triggering backbone methylation (2, 4).

Nevertheless, the substantial increase of methylated lipids observed in the stationary phase indicates a metabolic response to a change in the growth status of M. luminyensis that possibly could also be observed in sediments. We thus introduced the backbone methylation index (BMI) to express the change in methylation degree with change of growth status; the BMI corresponds to the relative proportion of PDGTs as fraction of the sum of BDGTs and PDGTs [PDGTs/(BDGTs + PDGTs)] (Fig. 3C). GDGTs were not included in the calculation because of their highly diverse sources in marine sediments (28), which would reduce the sensitivity to detecting responses of BDGT- and PDGT-producing archaea. BMI was then assessed in a set of marine sediment samples collected in the Mediterranean Sea that were previously investigated for their BDGT and PDGT content (11). An obvious variable that could affect BMI is sediment age because, with increasing age, the rate of microbial decomposition of organic matter decreases (29), and the proportion of microbial cells in stationary phase or dormancy increases (30). Indeed, BMI increases with the estimated age of the sediment (Fig. 4). The observed higher ratios of backbone methylation in the oldest, most energy-deprived sediment samples are therefore generally consistent with the hypothesis derived from our laboratory observations. Based on these preliminary observations in pure cultures and sediment samples, we propose that the herein introduced backbone methylation index could be used in natural settings to explore energetic status and growth activity of microbial communities along environmental gradients. Backbone methylation could serve to reduce membrane permeability to limit energy loss or could enhance the intact polar lipid stability, thus reducing maintenance energy requirements. To fully understand the environmental triggers and cellular processes responsible for backbone methylations will require further research, but this work opens new research avenues, notably for studies of the deep sedimentary biosphere.

This study demonstrates that biosynthesis of BDGTs and PDGTs results from the enzymatic methylation of typical archaeal GDGT lipids in M. luminyensis. This methylation is likely catalyzed by radical SAM enzyme(s) specific to a small range of taxa, including some uncultured methanogens and Bathyarchaeotal strains commonly observed in subsurface sediments. We further observed an increase of the degree of backbone methylation during the stationary phase of the batch pure culture incubation as well as in relatively old marine subsurface sediments. Based on these preliminary observations, we propose a novel index, the BMI, which could be used to probe energetic status and activity of microbial communities in natural environments.

Pure cultures of M. luminyensis strain B10 (purchased at DSMZ) were grown in 100-mL batch cultures under an H₂/CO₂ (80:20, 100 kPa) atmosphere at 37°C using an anaerobic medium described by the DSMZ (protocol 1637) and modified after Kröninger et al. (26). Half of the medium was amended with 15% [methyl-¹³C]methionine and distributed in six 250-mL serum vials, while the nonamended medium was dispatched in six additional 250-mL serum vials. Each vial was inoculated with 10% (vol/vol) of a M. luminyensis culture in its exponential growth phase. Cell growth was monitored every 2 to 4 days by headspace CH₄ measurements (supplementary methods in the supplemental material). Concentrations of acetate, methanol, and CH₄, as well as stable carbon isotopic composition of CH₄ and methanol, were regularly monitored in the course of the incubation (supplementary methods). Cell biomass was harvested in triplicate during exponential growth (16 days) and after reaching the stationary growth phase (28 days) by multiple 1-h-long centrifugation steps at 2,700 × g followed by 20 min at 14,000 × g in Eppendorf tubes.

Enrichment cultures of “Ca. Methanogranum gryphiswaldense” were grown in 20-mL batch cultures under a H₂/CO₂ (80:20, 200 kPa) atmosphere at 37°C (24, 25). Modified MpT1 medium (31) containing 50 μM methanol as electron acceptor was inoculated with 200 μL enrichment culture for 14 days until exponential growth phase (see details in the supplementary methods). Cell mass was harvested by centrifugation in Eppendorf tubes for 20 min at 13,000 × g.

Marine sediments were sampled in the Mediterranean Sea during the DARCSEAS cruises I and II onboard RV Meteor, cruise M84/1 (32), and RV Poseidon, cruise POS450 (33), respectively. Total lipid extracts were obtained following a modified Bligh and Dyer extraction (34) and analyzed on a Dionex Ultimate 3000 RS UHPLC system coupled to a Bruker maXis ultrahigh-resolution quadrupole time of flight tandem mass spectrometer (Q-TOF MS) equipped with an electrospray ion source and a C₁₈ reversed-phase column (ACE3, 3 μm, 2.1 by 150 mm; Advanced Chromatography Technologies Ltd.), as described previously (35). Details of the sampling and analytical procedures can be found in Coffinet et al. (11).

Freeze-dried cell pellets of M. luminyensis and “Ca. Methanogranum gryphiswaldense” were amended with combusted sand, and core lipids (CLs) were extracted by subjecting the mixture to a mild acid treatment as described before (16). Following acid hydrolysis, CLs were extracted three times for 20 min by sonication using a solvent mixture of dichloromethane (DCM)/MeOH (5:1 [vol/vol]). CL analysis of acid hydrolyzed biomass was performed on a Dionex Ultimate 3000 RS UHPLC system coupled to a Bruker maXis Q-TOF MS equipped with an APCI-II ion source (36). Lipids were separated using normal-phase chromatography with two coupled Acquity BEH amine columns (1.7 μm; 2.1 by 150 mm; Waters) maintained at 50°C. Detection of lipid isotopologues was performed in positive ionization mode scanning a mass range of 150 to 2,000 Da. All spectra acquired at a resolution of 27,000 were calibrated both externally with a loop-injected tuning mix at the end of each analysis and internally with a lock mass (m/z 922.0098), leading to typical mass errors <3 ppm. Lipids were identified by retention time and exact mass. All extracted mass chromatograms were calculated at ±0.01 Da. Diagnostic product ions for the investigation of intramolecular isotope distributions were obtained by systematic fragmentation in MS/MS mode using collision energies of 40 to 45 eV. Specific precursor ions were selected (10-Da isolation width) and sequentially fragmented in three time segments, 11.5 to 13.7 min for C-PDGT-0, 13.7 to 16.8 min for C-BDGT-0, and 16.8 to 18.0 min for C-GDGT-0.

Putative radical SAM proteins in M. luminyensis were identified based on homology to known radical SAM enzymes (pfam 04055) using Joint Genome Institute Integrated Microbial Genomes and Microbiomes (JGI IMG/M; https://img.jgi.doe.gov/). Successive Basic Local Alignment Search Tools (BLAST) homology searches of the candidate radical SAM enzymes identified in M. luminyensis were carried out against the complete “Ca. Methanogranum gryphiswaldense” genome (25) (metagenomics rapid annotation [MG-RAST] project ID 6666666.490459) and genomes of the 25 strains previously shown not to produce BDGTs (16).

The complete genome of “Ca. Methanogranum gryphiswaldense” can be accessed in NCBI (GenBank accession number CP076745) (https://www.ncbi.nlm.nih.gov/nuccore/2067799480). Lipid data of the samples collected in the Mediterranean Sea were deposited in the PANGAEA repository (https://doi.pangaea.de/10.1594/PANGAEA.907964).