Comparative Transcriptomics Sheds Light on Remodeling of Gene Expression during Diazotrophy in the Thermophilic Methanogen Methanothermococcus thermolithotrophicus

ABSTRACT

Some marine thermophilic methanogens are able to perform energy-consuming nitrogen fixation despite deriving only little energy from hydrogenotrophic methanogenesis. We studied this process in Methanothermococcus thermolithotrophicus DSM 2095, a methanogenic archaeon of the order Methanococcales that contributes to the nitrogen pool in some marine environments. We successfully grew this archaeon under diazotrophic conditions in both batch and fermenter cultures, reaching the highest cell density reported so far. Diazotrophic growth depended strictly on molybdenum and, in contrast to other diazotrophs, was not inhibited by tungstate or vanadium. This suggests an elaborate control of metal uptake and a specific metal recognition system for the insertion into the nitrogenase cofactor. Differential transcriptomics of M. thermolithotrophicus grown under diazotrophic conditions with ammonium-fed cultures as controls revealed upregulation of the nitrogenase machinery, including chaperones, regulators, and molybdate importers, as well as simultaneous upregulation of an ammonium transporter and a putative pathway for nitrate and nitrite utilization. The organism thus employs multiple synergistic strategies for uptake of nitrogen nutrients during the early exponential growth phase without altering transcription levels for genes involved in methanogenesis. As a counterpart, genes coding for transcription and translation processes were downregulated, highlighting the maintenance of an intricate metabolic balance to deal with energy constraints and nutrient limitations imposed by diazotrophy. This switch in the metabolic balance included unexpected processes, such as upregulation of the CRISPR-Cas system, probably caused by drastic changes in transcription levels of putative mobile and virus-like elements.

INTRODUCTION

Methanogenic archaea generate about 1 gigaton of methane per year (1). This amounts to about half of the greenhouse gas methane in our atmosphere (2, 3). Methanogenesis is a strictly anaerobic process occurring in habitats in which electron acceptors other than CO₂ are depleted (4). It is accepted that under natural conditions methanogenesis provides an extremely low energy yield, causing methanogens to thrive close to the thermodynamic limits of life (1, 5, 6). Therefore, it came as a surprise when in 1984 two studies proved that Methanothermococcus thermolithotrophicus (7) and Methanosarcina barkeri (8) can perform nitrogen fixation, a very energy-demanding metabolic process that requires the hydrolysis of at least 16 ATP per molecule of fixed N₂ (9).

The nifH gene is widely used as a marker to identify nitrogen-fixing organisms. Multiple environmental studies have detected nifH genes and transcripts from methanogens in diverse anoxic habitats, such as deep seawater and hydrothermal vent fluids (10), oligotrophic open seas (11), deep-sea methane seep sediments (12), and N₂-limited soils of salt marshes (13). Collectively these findings show that methanogens are indeed actively fixing N₂ in nature and thereby contributing considerably to global nitrogen cycling. They also imply that methanogens and anaerobic methanotrophs (14) can overcome the largest activation barrier in biology, +251 kJ mol⁻¹ (15), to break the N₂ triple bond.

The reduction of N₂ to NH₃ is catalyzed by the nitrogenase enzyme complex. The overall organization of this complex is highly conserved among Bacteria and Archaea. It is composed of a dinitrogenase reductase (iron protein, NifH) and dinitrogenase (iron-molybdenum protein, NifDK) containing one [MoFe₇S₉C-(R)-homocitrate] iron-molybdenum cofactor (FeMo-co) and an [8Fe-7S] P-cluster (15). The nitrogenase-encoding nif genes and a plethora of accessory proteins with roles in regulation, biosynthesis, and maturation of metal cofactors can be clustered within one or in several operons or regulons, depending on the organism (15–18). In addition to Nif, which represents the most widespread and well-studied system, at least two alternative nitrogenases are known: vanadium nitrogenase (VFe protein; Vnf) and iron-only nitrogenase (FeFe protein, Anf) (15). Both are considered to be evolutionarily related to the Nif system (19).

Phylogenetic analyses suggest that the ancestral nitrogenase originated in anaerobic methanogenic archaea, from where it was subsequently transferred into the bacterial domain, most probably initially to the Firmicutes (15, 19–22). Despite the shared origin and a minimal conserved set of genes required for functionality, archaeal nitrogenases are very distinct from their bacterial homologs. This is reflected by several features: (i) the organization of nitrogenase operon(s) (see Fig. S1 in the supplemental material), (ii) transcriptional regulation by the repressor NrpR and the activator NrpA, and (iii) the unique mode of posttranslational regulation through direct protein-protein interaction (23–25). In the latter case, the inhibition of nitrogenase activity is achieved by NifI_1,2, a P_II family regulator which blocks the NifH binding site on NifDK. Both NifI₁ and NifI₂ are required for the switch-off (26), which is reversed by 2-oxoglutarate magnesium, and ATP addition. These differences have been extensively studied in the mesophilic methanogen Methanococcus maripaludis (16, 17, 26–30) and different Methanosarcina strains (31–36) using well-established genetic systems for both.

FIG S1

Genomic environment of nif, vnf, and anf genes from selected diazotrophic methanogens and A. vinelandii. The genes nifX from M. thermolithotrophicus SN-1 (also referred to as strain DSM 2095 in the main text), nifX from M. maripaludis S2, and nifX from A. vinelandii 567 do not share sequence homologies. Download FIG S1, TIF file, 0.7 MB ^{(743.6KB, tif)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

In this context, it is noteworthy that M. maripaludis features a single operon consisting exclusively of nif genes (16, 28) (Fig. S1), whereas some Methanosarcinales harbor additional vnf and anf operons coding for V-only and Fe-only nitrogenases, respectively. However, these operons all share the core organization of the nif operon, in which nifH, nifD, and nifK genes code for the two structural subunits of the nitrogenase system while nifE and nifN play a putative role in the synthesis of FeMo-co. Genes nifI₁ and nifI₂ encode P_II family regulators that are conserved in archaea, while nifX has no homology with the homonymous bacterial gene (Fig. S1). NifX, absent in Methanosarcinales, has an unknown function. It is not required for nitrogen fixation in M. maripaludis, since nifX in-frame deletion mutants are still capable of diazotrophic growth (17).

Only a few studies exist on the diazotrophic physiology of (hyper)thermophilic Methanococcales. The N₂ fixation capabilities of these hydrogenotrophic methanogens are particularly interesting for several reasons: (i) their nitrogenases form a separate evolutionary branch (19) (Fig. S2 and Fig. S3), (ii) they can perform N₂ fixation under very high H₂ partial pressure that usually inhibits nitrogenases (7, 8, 37), and (iii) they can fix nitrogen at the highest temperature ever described so far (i.e., up to 92°C) (37). Together, findings suggest a unique adaptation that allows N₂ fixation to operate at both high temperatures (7) and extreme energy limitations.

FIG S2

Evolutionary analysis of 35 NifH sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) are shown next to the branches (110). Evolutionary distances (95) are in the units of the number of amino acid substitutions per site. ChlL (light-independent protochlorophyllide reductase) from Chlorobium limicola was used as an outgroup. Accession numbers for sequences used in phylogenetic reconstruction can be found in Table S5. Download FIG S2, TIF file, 1.6 MB ^{(1.6MB, tif)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

FIG S3

Evolutionary analysis of 35 NifD and NifK sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) are shown next to the branches (110). Evolutionary distances (95) are in the units of the number of amino acid substitutions per site. ChlNB from Chlorobium limicola were used as an outgroup. The same color-coding and symbols are used as in Fig. S2. Accession numbers for sequences used in phylogenetic reconstruction are provided in Table S5. Download FIG S3, TIF file, 2.5 MB ^{(2.6MB, tif)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

We investigated the integration of N₂ fixation with other metabolic processes in thermophilic M. thermolithotrophicus by using a combined approach of physiological tests and differential transcriptomics. Our results provide new insights into the adaptive strategies of M. thermolithotrophicus to energy and nutrient limitation stresses inflicted by Mo-dependent diazotrophy.

RESULTS

Nitrogen acquisition by M. thermolithotrophicus.

We selected M. thermolithotrophicus DSM 2095 due to its remarkable chemolithoautotrophic capabilities and fast growth at 65°C in mineral medium (38, 39). Series of incubations with different NH₄Cl concentrations, ranging from 0.1 to 16.8 mM, showed that M. thermolithotrophicus required a minimum of 10 mM NH₄Cl for best growth (Fig. 1A) in our optimized medium (see Materials and Methods for composition) (39, 40). Higher NH₄Cl concentrations of 25 to 200 mM NH₄Cl did not result in higher cell yields, and a 500 mM excess of NH₄Cl led to a decrease (see Fig. S4 in the supplemental material). Continuous monitoring of ammonia consumption during growth on 10 mM NH₄Cl over a span of 26 h confirmed depletion proportional to the observed increase in biomass (Fig. 1B). The culture reached stationary phase after 18 h, during which most of the ammonia was consumed beyond the reliable detection limit.

FIG 1.

Growth of Methanothermococcus thermolithotrophicus on different nitrogen sources. (A) Final A₆₀₀ of M. thermolithotrophicus cultures after 28 h of incubation with different NH₄Cl concentrations. (B) Growth curve of M. thermolithotrophicus cultures grown on 10 mM NH₄Cl (squares) and NH₄Cl consumption during growth (triangles). (C) Growth curve of diazotrophic M. thermolithotrophicus cultures grown on N₂ as the sole nitrogen source (circles) and NH₃ release during the growth (triangles). (D) Growth curves of nondiazotrophic M. thermolithotrophicus on 16.8 mM NH₄Cl (squares), a diazotrophic culture grown on Na₂SO₃ (full circles), and a diazotrophic culture grown on Na₂SO₄ (empty circles) in a fermenter. Measurements for panel A were performed in duplicates; measurements for panels B and C were performed in triplicates.

FIG S4

Final A₆₀₀ of M. thermolithotrophicus after 24 h of incubation with different NH₄Cl concentrations. All measurements were done in triplicates. Download FIG S4, TIF file, 0.1 MB ^{(146.2KB, tif)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

M. thermolithotrophicus was adapted to diazotrophic conditions after three successive transfers to NH₄Cl-free medium. Under this condition, dissolved N₂ was the sole available nitrogen source. The cell density of the diazotrophic culture was 6 times higher than that reported by Belay and colleagues in their original study on the discovery of diazotrophy in methanogens in 1984 (e.g., Absorbance 600nm 1.85 achieved in 48 h, compared to an Absorbance 600nm of 0.4 achieved in ~20 h [7]). Ammonia possibly released in the medium was measured during diazotrophic growth (Fig. 1C), and traces were detected during the mid-exponential phase, reaching a maximum of 40 μM ammonia in the stationary phase (Fig. 1C). However, we cannot exclude that detected ammonia also originated from cell lysis rather than from active excretion.

In addition to batch diazotrophic cultures, M. thermolithotrophicus was successfully grown in a 10-liter fermenter continuously supplied with H₂/CO₂ and N₂. To maintain the sulfur source in the medium, we replaced Na₂S used in batch cultures (which would be flushed out as H₂S) with Na₂SO₃ or Na₂SO₄ (see Materials and Methods) (39). Diazotrophic growth was not affected by HSO₃⁻, a known inhibitor of methanogenesis (Fig. 1D) (41). While the final cell yield was similar to the one observed in batch cultures, division times were shorter (Fig. 1D). Fermenter-grown cultures supplemented with NH₄Cl had higher final yields than diazotrophic fermenter-grown cultures. However, in comparison to NH₄⁺-grown batch cultures, the difference was not very pronounced, as final yields were only slightly higher and the division times were similar.

N₂ fixation was molybdenum dependent.

We then investigated the nitrogenase type used for diazotrophy, taking into account that the M. thermolithotrophicus genome features only one nifDK. It must be noted that M. thermolithotrophicus resembles M. maripaludis in terms of physiology, and the latter has been shown to require molybdenum (Mo) for its diazotrophic growth (28). Therefore, we first tested for Mo dependency. Growth of cultures lacking either Mo, vanadium (V), or both metals in the medium were monitored simultaneously. Depletion of Mo and V from the medium was achieved by three successive culture transfers to the same medium without Mo and V. The highest optical density at 600 nm (OD₆₀₀) was reached when M. thermolithotrophicus was grown in the control medium containing both Mo and V. Growth was similar in the diazotrophic culture incubated without V, but with a lower final OD₆₀₀ (Fig. 2A). This trend was reproducible and suggested that V acts as a potential growth stimulator under such conditions; however, the potential mechanism remains unclear. No growth was observed in the absence of Mo, with or without V (Fig. 2A). The minimal required Mo concentration of 0.1 μM was determined in a separate series of incubations with Mo concentrations ranging from 0.01 to 100 μM (Fig. 2B). To check if Mo was essential for methanogenesis (42) or metabolic processes other than nitrogen fixation, we supplemented the culture grown with V in the absence of Mo with NH₄Cl, which restored growth after an overnight incubation (Fig. 2A).

FIG 2.

Influence of trace metal availability on diazotrophic growth of M. thermolithotrophicus. (A) Growth curves of diazotrophic M. thermolithotrophicus grown in the medium with both Mo and V (full circles), without V (empty circles), without Mo (gray circles), and without both Mo and V (full rhomboids) and of the negative control (empty rhomboids). The arrow indicates the supplementation of the culture containing V without Mo with 16.8 mM NH₄Cl. (B) Final A₆₀₀ of M. thermolithotrophicus as a function of MoO₄²⁺ concentration. (C) Final A₆₀₀ of nondiazotrophic (dark gray bars) and diazotrophic (light gray bars) M. thermolithotrophicus as a function of WO₄²⁺ concentration. A concentration of 10 μM Na₂MoO₄ was used for this experiment. Measurements for panel A were performed in triplicates; measurements for panels B and C were performed in duplicates.

An inhibitory effect of tungstate (WO₄²⁻) on diazotrophic growth has already been observed in Azotobacter vinelandii OP (43), Methanosarcina barkeri 227 (32), and M. maripaludis strain S2 (28). In this case, WO₄²⁻ can inhibit MoO₄²⁻ uptake (44), and it can be incorporated into the nitrogenase, thereby rendering the protein inactive (43, 45, 46). As shown by Siemann and coworkers in their work on Rhodobacter capsulatus, a WFe nitrogenase does not exhibit N₂ or C₂H₂ reducing activity but can reduce protons (46). These inhibition mechanisms are competitive and thus dependent on the MoO₄²⁻/WO₄²⁻ ratio. Surprisingly, there was no observable inhibitory effect in M. thermolithotrophicus (Fig. 2C), even at a 1:1 MoO₄²⁻/WO₄²⁻ ratio. In contrast, diazotrophic growth of M. maripaludis was inhibited at a 10:1 MoO₄²⁻/WO₄²⁻ ratio (28). This observation might be explained by different affinities of the WO₄²⁻ and MoO₄²⁻ transport systems in both species. The ModABC transporter (47, 48) is present in both organisms and should transport MoO₄²⁻ and WO₄²⁻ (49). In addition, M. thermolithotrophicus has a highly specific WO₄²⁻ transporter, TupABC (50), that is lacking in M. maripaludis. Instead, M. maripaludis has a third type of tungstate transporter, WtpABC (49, 51), which can transport WO₄²⁻ and MoO₄²⁻ but has a higher affinity for WO₄²⁻ than has been shown for Mod and Tup in Pyrococcus furiosus (51). Furthermore, it is also known that Mod transporters can have different affinities for both oxyanions in different organisms (49, 52, 53), suggesting that both transporter specificities and the specificity of the FeMo-co insertion machinery contribute to the W tolerance during diazotrophy.

Diazotrophy shifted expression of a large number of genes.

Transcriptomics and DNA microarrays have been used to investigate the complex metabolic and regulatory networks that control N₂ fixation in two model organisms: Azotobacter vinelandii (54) and Methanosarcina mazei Gö1 (36, 55, 56). These studies led to the discovery of three gene clusters (rnf1, rnf2, and fix) coding for electron transfer systems that provide reducing equivalents to the nitrogenase in A. vinelandii, as well as multiple potential transcriptional and small RNA (sRNA) regulators in Methanosarcina mazei Gö1. Here, we used comparative transcriptome profiling of N₂-fixing M. thermolithotrophicus cultures versus NH₄⁺-grown cultures to investigate the metabolic adaptations induced by nitrogen fixation. The experiment was conducted in biological triplicates at three different time points: 3 h (early exponential phase), 21 h (stationary-phase starvation due to the exhaustion of the gas phase), and 25 h postinoculation (see Materials and Methods) (Fig. 3A and Table S1). The gas phase was exchanged after sampling at the 21-h time point, since the cultures consumed H₂/CO₂. The robustness of our sample data was reflected in a corresponding principal-component analysis (PCA), in which triplicates from different conditions and time points were clustered (Fig. 3B). More importantly, it revealed that 92% of the variation could be explained by the different sample treatments.

FIG 3.

Differential gene expression between NH₄Cl-grown and diazotrophic cultures. (A) Points during growth of the control (squares) and diazotrophic (circles) cultures at which samples for transcriptomic profiling were taken (each of the triplicates is shown separately). The arrow indicates the complete gas-phase exchange of the cultures after the sample at 21 h was taken. (B) PCA plot showing the first two principal components that explain the variability in the data using the regularized log count data. Control corresponds to the NH₄Cl-grown cultures. (C) Volcano plots of differentially expressed genes in the diazotrophic culture versus control at different time points. Negative and positive log₂ fold changes of >1 with adjusted P of <0.05 are shown in red and green, respectively. Values in red and green indicate the number of down- and upregulated genes, respectively.

TABLE S1

Mapping statistics. Download Table S1, XLSX file, 0.01 MB ^{(11KB, xlsx)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

Transcriptome profiling across the three time points revealed prominent changes (Fig. 3C). Of the total 1,751 predicted genes (open reading frames [ORFs]), 1,737 genes were found to have nonzero read counts. At 3 h, 12.3% (214) were differentially expressed, with most (122/214) being upregulated under diazotrophic conditions. More than half of these genes (70/122) were of an unknown function without any homologs in the databases (their probable role is discussed further below). The difference was smaller at 21 h, with only 3.3% (57) of the total expressed genes having differential expression; however, at 25 h 17.4% (302) were differentially expressed, with most of them downregulated (257/302) under diazotrophic conditions (Table S2).

TABLE S2

Summary of up- and downregulated genes pairwise. Download Table S2, XLSX file, 0.1 MB ^{(59KB, xlsx)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

To check the overall expression of genes irrespective of differential expression, we mapped reads to predicted genes in terms of transcripts per million (TPM). Details of this analysis are summarized in the sections below.

Nitrogen and molybdate acquisition genes were mostly affected during the early exponential phase.

As anticipated, upregulated genes at 3 h included the nif operon (nifHI₁I₂DKXEN) and molybdate-acquisition genes (modABC). This illustrated a swift response of M. thermolithotrophicus to diazotrophic conditions. The highest log₂ fold change (FC) was observed for nifI₁I₂HDKX genes, with values ranging from 8.8 to 7.0 (Table S2). The proteins of the FeMo-co biosynthetic machinery, nifE and nifN, were also among the upregulated genes, with log₂FC of 5.3 and 4.6, respectively (Fig. 4A and B). An essential gene for FeMo-co biosynthesis, nifB (57, 58), was transcribed at a lower level, with a log₂FC of 1.8.

FIG 4.

Changes in the expression over time of selected genes involved in nitrogen acquisition. (A) Arrangement of the nif, mod, amtB/glnK, and nar/fsr operons in M. thermolithotrophicus. (B to D) Log₂ fold changes (light gray bars) and TPM of the controls (orange squares) and diazotrophic cultures (green circles) of nitrogen acquisition genes, compared to the mcrA gene as reference, after 3 h (B), 21 h (C), and 25 h (D). Genes that were not differentially expressed at a given time point are marked by an asterisk.

The mod operon, coding for the three subunits of a molybdate ABC transporter (modABC), was also upregulated (Fig. 4A), with modA being strongly expressed with a log₂FC of 5.5 (Table S2). This agreed with our physiological data, that showed a clear Mo dependency under diazotrophic conditions. Although three modA instances are present in the genome, only the modA gene in the vicinity of the nif operon, which is part of the complete mod operon, was highly expressed. The tupA gene coding for the tungstate transporter was also upregulated (log₂FC, 1.7), strengthening the hypothesis that both transporter specificities and adaptation of the cofactor biosynthetic machinery contribute to the W tolerance during diazotrophy.

The other genes upregulated at 3 h included the ammonium transporter amtB₂ and its P_II family regulatory protein glnK₂, with the log₂FCs of 6.2 and 5.8, respectively (Table S2). The genome of M. thermolithotrophicus harbors two different amtB genes with their associated glnK regulators (40). Interestingly, amtB₁ and glnK₁ were not differentially expressed (Table S2; Fig. 4A and B), which might point to the existence of an internal regulator or an additional promoter. In addition to the nif and amtB/glnK operons, which are known to be upregulated during diazotrophy from previous studies (54, 55, 59), the ammonium-assimilating glutamine synthetase (glnA) gene was also upregulated (Table S2). However, the transcript level of the second enzyme involved in N-assimilation, glutamate synthase, remained unchanged, suggesting that GlnA is the rate-limiting step in ammonium assimilation.

Unexpectedly, narK coding for a putative nitrate transporter and narB coding for a molybdopterin-dependent nitrate reductase (60) were also highly expressed during diazotrophy in the early exponential stage (Fig. 4A and B; Table S2). If nitrate were imported and reduced in the cell, oxidant nitrite would be generated and could damage the highly oxidation-sensitive methanogenic machinery (61). A gene coding for a F₄₂₀-sulfite reductase isoform (also belonging to Fsr group 1 but different from the one naturally expressed under sulfite conditions [39]), which cooccurs with narK and narB, would be a plausible candidate for nitrite detoxification, since Fsr has been recognized to catalyze nitrite reduction (39). M. thermolithotrophicus has been reported to be able to grow on nitrate as the only source of nitrogen (62), which might involve these genes.

In addition, three genes coding for the enzymes of the molybdopterin biosynthetic pathway (moaE, mobB, and moeB) were upregulated. They might be involved in supplying the putative nitrate reductase with molybdopterin, while the formylmethanofuran dehydrogenase enzyme used for methanogenesis harbors a tungstopterin (42). The pathway of molybdopterin biosynthesis has been extensively studied, and it is highly conserved (63). The pathway of tungstopterin biosynthesis is thought to be homologous up to the step of metal insertion (64). It has been proposed that the biosynthetic machinery is able to distinguish between the two metals and insert the correct metal into the respective enzyme by employing MoeA isoenzymes selective for either molybdate or tungstate (64, 65). All archaeal genomes sequenced so far encode two moeA isoforms sharing around 40% identity (64). This might explain how organisms are able to correctly express different proteins with molybdopterin and tungstopterin simultaneously (64). However, this hypothesis still lacks experimental validation.

As shown in Fig. 4C and D, the expression of genes partaking in N₂ fixation and molybdenum acquisition remained unchanged at 21 h and 25 h; therefore, no differential expression was observed. The overall expression of genes in terms of TPM followed a similar pattern. An established marker for metabolic activity of methanogens, the transcription level of mcrA (66) was used as a reference point for comparison (Fig. 4B to D).

None of the methanogenesis pathway genes showed any difference in transcription levels at the early exponential phase. However, after 25 h nearly all genes involved in methanogenesis were downregulated. The expression of F₄₂₀-reducing hydrogenase was maintained at the same level at all time points to supply reduced F₄₂₀ from H₂ oxidation. Hydrogenotrophic Methanococcales can alternatively use formate as electron source (67, 68), and in this case a putative formate transporter (fdhC) and formate dehydrogenase (fdhF) were found to be downregulated at 3 h. In addition to methanogenesis, numerous anabolic processes were shut down at 25 h, such as carbon assimilation (e.g., pyruvate:ferredoxin oxidoreductase), amino acid metabolism (e.g., ketol-acid reductoisomerase), lipid biosynthesis (e.g., hydroxymethylglutaryl-coenzyme A synthase), ATP synthesis (i.e., ATP synthase), and vitamin and coenzyme biosynthesis (e.g., hemE). Such a decrease in catabolic and anabolic processes combined with the downregulation of genes involved in S-layer formation and cellular division (e.g., ftsZ) suggests a fine-tuned metabolic mode of energy saving to prioritize nitrogen fixation.

Transcriptional and translational machineries were considerably downregulated.

Both the transcriptional and translational machineries responded negatively to diazotrophic conditions after 3 h. This included downregulation of RNA polymerase subunits (rpoA₂H), the sigma factor 70 (rpoD, controlling the transcription of housekeeping genes), and 35 ribosomal proteins (Table S2). Our rRNA expression data have the limitation that rRNA removal treatment was performed before sequencing. Still, we could detect downregulation of rRNA expression, which was corroborated by downregulation of ribosomal proteins. Some genes with a putative function in tRNA and ribosome biogenesis and biosynthesis of nucleotide and amino acid precursors were also negatively affected, corroborating a deep impact on the overall translation process under N₂-fixing conditions. For example, two key enzymes (transketolase and transaldolase) of the nonoxidative branch of the pentose phosphate pathway (69) for synthesis of nucleotides and histidine (from ribose 5-phosphate), as well as aromatic amino acids (from erythrose 4-phosphate) precursors were downregulated (Table S2). This was another notable difference from Methanosarcina, in which the upregulation of genes involved in the synthesis of aromatic amino acids has been observed (55). Notably, tRNA^Thr and tRNA^Ser were upregulated, while tRNA^Ala and tRNA^Pro were downregulated at 3 h. Taken together, changes in the expression of all the mentioned genes contributed to the restriction of the entire translation process. The DNA replication system seemed not to be impacted, since we could not observe any changes in the transcript levels of DNA polymerase-encoding genes. This was in line with the results of Belay and coworkers, who showed that during diazotrophic growth the cellular protein content of M. thermolithotrophicus was significantly reduced (7).

High expression of CRISPR-Cas genes and putative viral genes.

Components of the CRISPR-Cas virus defense system were also upregulated under diazotrophy at 3 h, including Csm1 to -5 of the Csm effector complex and CRISPR-associated proteins (Cas proteins) Cas4 to -6 (Table S2). This might be explained by presence of a prophage that is expressed when the cells are energy depleted or otherwise stressed. The Phaster server, a tool for finding prophages in bacterial genomes (70), was unsuccessful in detecting any complete prophages in the M. thermolithotrophicus genome. However, we identified a locus of 28 cooccurring ORFs that were highly expressed at all time points (Fig. 5). Twenty of these genes had no characterized homologs, while eight encoded putative virus-like, replication, and mobile genetic elements. We used models generated by Alphafold2 (71) and a membrane prediction tool to gain further information on these sequences and examined 18 confident models (Fig. 5). Five of these proteins were predicted to be secreted or embedded in the membrane, but structural homologs were scarce based on the predicted models (Table S3). The protein encoded by the third gene of the locus (Fig. 5) has already been structurally characterized in Thermococcales (72) and is believed to be a virus-like element. While Methanococcales share some of these genes, Methanosarcina mazei, a model organism for the Methanosarcinales, contains only two of them in its genome (Table S3). These 28 cooccurring genes might be derived from a small plasmid transferred by conjugation or represent a yet-unknown prophage, although we could not identify a protein that could be responsible for independent insertion or replication of this element. This region could therefore be considered part of the dark matter in archaeal genomes (73). However, active transcription of all the genes of this locus might be a plausible explanation for the observed upregulation of the CRISPR-Cas system.

FIG 5.

Log₂ fold changes over time for 28 cooccurring genes of unknown function that are of putative viral origin. AlphaFold 2 (71) models are represented by cartoons and color-coded as follows: black, intracellular segments; cyan, predicted extracellular segments; ocher, transmembrane segments; red, predicted signal sequences. The topology prediction was made via the DeepTMHMM server (109). Genes with numbered red labels highlight models with an overall per-residue confidence score (pLDDT) of <75. These models were not presented due to their low confidence scores.

TABLE S3

Upregulated genes coding for putative virus-like elements in region 1 and their conservation among selected methanogens. Download Table S3, XLSX file, 0.01 MB ^{(14.5KB, xlsx)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

Another locus of 26 cooccurring ORFs was downregulated under diazotrophic conditions at all time points (Fig. S5). Again, most of these genes have unknown functions, with the exception of a putative transcriptional regulator, a mini-chromosome maintenance protein, and a recombinase (Table S4). Four of these genes were predicted to encode secreted proteins, and eight contained transmembrane segments. Based on these observations, we assume that these 26 cooccurring open reading frames are also derived from mobile elements or prophage-associated genes.

TABLE S4

Downregulated genes coding for putative virus-like elements in region 2 and their conservation among selected methanogens. Download Table S4, XLSX file, 0.01 MB ^{(14KB, xlsx)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

FIG S5

Log₂ fold changes over time for the 26 cooccurring genes of unknown function. The downregulation of this genomic region in coordination with the upregulation of the other detected putative region of viral origin (Fig. 5) might imply the competition of the two for the availability of transcriptional and translational machineries, or it is the result of a general cellular stress. The topology prediction was made via the DeepTMHMM server and is color-coded as follows: black, intracellular proteins; cyan, secreted proteins; green, proteins containing transmembrane segments. WP numbers and predicted functions can be found in Table S4. A gene that was not differentially expressed at 3 h is marked by an asterisk. Download FIG S5, TIF file, 1.1 MB ^{(1.1MB, tif)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

DISCUSSION

Diazotrophy allows microbes to survive when nitrogen becomes limiting in their natural habitats. This is also the case for (hyper)thermophilic methanogens, such as M. thermolithotrophicus, which has been shown to rely on diazotrophy in different environments (10–13). A notable exception among Methanococcales is Methanocaldococcus jannaschii, a nondiazotrophic methanogen isolated from a white smoking crater on the East Pacific Rise. While ammonium concentrations at the M. jannaschii isolation site were not measured (74), some parts of the East Pacific Rise, such as Guaymas Basin, are known to feature notable ammonium concentrations (e.g., 15.3 mM [75]). Availability of this inorganic nitrogen source might have resulted in a complete loss of nitrogen-fixing abilities in M. jannaschii.

Due to its active N₂ fixation, M. thermolithotrophicus represents one of the contributors to the available nitrogen pool in specific environments (10–13). In our laboratory cultures, M. thermolithotrophicus released only minute amounts of up to 40 μM ammonia to the medium, which might have simply resulted from passive diffusion or cell lysis. It therefore seems that N₂ fixation in M. thermolithotrophicus is precisely controlled to avoid any losses, likely by the NifI_1,2 regulation system (Fig. 6). The excessive energy cost inflicted by nitrogenase activity was noticeable in our physiology experiments (Fig. 1), as it reduced final yields by a factor of 2 in both batch- and fermenter-grown cultures.

FIG 6.

Influence of nitrogen fixation on the metabolism of M. thermolithotrophicus. In the scheme, differentially expressed genes involved in methanogenesis, nitrogen fixation, ammonia assimilation, nitrate reduction, molybdate import, molybdopterin biosynthesis, immunity, mobile elements, and transcription and translation, as well the tRNAs, are shown. Differentially upregulated and downregulated transcripts at 3 h postinoculation are highlighted with a green and red glow, respectively. An orange glow highlights differentially downregulated gene transcripts at 25 h postinoculation. Full names of enzymes can be found in Table S2. This picture was created with Biorender.

Transcriptome profiling under diazotrophic conditions revealed that M. thermolithotrophicus relies on multiple synergistic strategies that ensure both sufficient N₂ fixation and energy preservation to support cellular growth (Fig. 6). In addition, M. thermolithotrophicus enhances nitrogen acquisition by increasing ammonia uptake via its amtB₂ transporter and glutamine synthetase overall activity, a strategy that has been described before based on proteomics of nitrogen-starved M. maripaludis cultures (59). While upregulation of the nif and amtB/glnK operons under nitrogen limitation has been previously reported in other diazotrophs (54, 55, 59), the upregulation of a putative nitrate transporter (narK), a nitrate reductase (narB), and a new isoform of F₄₂₀-dependent sulfite reductase (fsr) reported in this study is so far unique to M. thermolithotrophicus.

Like M. thermolithotrophicus, the hyperthermophile Methanocaldococcus infernus has also been reported to grow on nitrate as sole nitrogen source (76). M. infernus was isolated from a smoking crater of the Logatchev hydrothermal vent field (76, 77), from which other nitrate-reducing organisms have been isolated as well (78). Although hydrothermal fluids have been reported to be depleted in nitrate and nitrite (79), bottom seawater can contain nitrate for use as electron acceptor and nitrogen source. The enzyme which can possibly be used for intracellular nitrate reduction by these methanogens is the nitrate reductase NarB. NarB is expected to harbor a tungstopterin or molybdopterin cofactor, which might explain upregulation of molybdopterin biosynthesis genes (moaE, mobB, moeB) under diazotrophic conditions. In this context, the molybdate transporter Mod would play a dual role, supplying Mo for both the nitrogenase and NarB metallo-cofactor.

Strict dependence on Mo for diazotrophy has also been described for Methanococcus maripaludis (28). Likewise, our results support that the single nitrogenase operon of M. thermolithotrophicus encodes a molybdenum nitrogenase. This is corroborated by the M. thermolithotrophicus phylogenetic position (19), as well as high transcription levels of the mod operon (encoding a molybdate ABC transporter) adjacent to the nitrogenase operon under diazotrophy. In addition to modA within the mod operon, M. thermolithotrophicus features two additional modA instances. These are not colocalized with modBC genes and were not differentially expressed in our experiments. Although multiple modA genes are present in different diazotrophic organisms, including A. vinelandii and the methanogens M. maripaludis and M. mazei, the benefit of multiple modA genes is not yet understood. Interestingly, in M. mazei (55), no upregulation of Mod transporters was observed. In Methanosarcinales, expression of modABC is under the control of the transcriptional regulator ModE. This regulator also affects transcriptional regulation of different nitrogenases types, if these are present (80, 81). Methanococcales are devoid of modE homologs, which might explain the observed upregulation of the mod operon under diazotrophic growth in M. thermolithotrophicus.

The nif operon exhibited the highest log₂FC values between both conditions. However, changes in transcript levels of the nifE and nifN FeMo-co biosynthesis genes were less prominent than those of the structural nif genes. The former are possibly required in lower amounts, as was the case for the FeMo-co biosynthesis gene nifB. Levels of nifX transcription were similar to those of the nifDK nitrogenase genes, suggesting potential association with the nitrogenase complex. The nifX gene is not essential for nitrogen fixation in M. maripaludis and has not been detected in the nitrogenase complex of this archaeon. Instead, it has been proposed to take on the role of the FeMo-co biosynthesis genes (17). However, since M. maripaludis nifX is not homologous to M. thermolithotrophicus nifX, future experimental evidence is required to clarify its putative function and association to NifDK in (hyper)thermophilic methanogens.

The enormous ATP investment required for diazotrophy did not affect the transcription of methanogenesis genes at the onset of diazotrophic growth. This observation corroborates previous studies (55, 59) that reported that the proportion of the overall energy required for maintenance increased when switching to diazotrophy (82). The high energy demand for nitrogen fixation is counterbalanced by drastic downregulation of the transcription and translation machineries. The question is whether this is an a priori and thus targeted metabolic adaptation, or a mere general consequence of energy starvation. It has been described that transcription and translation are reduced upon nutrient limitations. After all, 40 to 70% of the cellular ATP pool in growing bacteria is attributed to protein synthesis (83). It is also known that bacteria can balance tRNA abundances when under stress to selectively regulate the translation of stress-induced proteins. These proteins contribute to the response and adaptation to different types of stress, including nutrient limitation (84). It has recently been shown that M. maripaludis adopts a resource relocation strategy upon energy depletion (85): instead of reducing ribosome numbers, the cells rather redistribute available energy and decrease catabolic and ribosomal activities, a strategy also described in bacteria (86). Therefore, our results suggest a targeted response associated with switching to a diazotrophic lifestyle. In comparison, no such changes in the transcript levels of transcriptional and translational genes were detected in either Azotobacter vinelandii (54) or M. mazei (55) when switching to diazotrophic growth, even though in the case of the former 30% of the genes were differentially expressed (as observed by differential transcriptomics), and in the case of the latter, 5% were differentially expressed (as observed by microarray analysis).

Studies in M. mazei and M. maripaludis confirmed the prominent regulation of nitrogen metabolism at the transcriptional level, particularly the importance of the global nitrogen regulatory repressor NrpR of nitrogen assimilation genes (24, 25, 35, 87). NrpR represses transcription by binding to the nif and glnA promoter regions in a 2-oxoglutarate-dependent manner. NrpR from M. thermolithotrophicus shares 38.2% identity with the one from M. mazei. Interestingly, NrpR expression levels are not regulated by nitrogen supply (24, 25). Thus, although a homolog of NrpR does exist in M. thermolithotrophicus, its transcription level remained unchanged during diazotrophic growth. In contrast to NrpR, NrpA, the nif promoter-specific activator, sharing 32% identity with the one from M. mazei, is known to be upregulated upon nitrogen limitation (88). We did not observe this in our experiments, suggesting that NrpA could be constitutively expressed in M. thermolithotrophicus.

One major difference between M. mazei and M. thermolithotrophicus, however, is the role of regulatory sRNAs. Methanosarcina mazei Gö1 expresses multiple sRNA genes in response to nitrogen limitation (56), with sRNA₁₅₄ being the best characterized (36). In M. mazei, sRNA₁₅₄ stabilizes the nitrogenase, glnA₁, and nrpA mRNAs (36). While sRNA₁₅₄ is highly conserved within Methanosarcinales, there are no sRNA₁₅₄ homologs in Methanococcales (89). Consequently, we did not detect such a pattern in M. thermolithotrophicus.

Transcriptome analysis can provide valuable information on proteins involved in nitrogen fixation. For instance, transcriptome analysis allowed the discovery that in A. vinelandii electron bifurcation is coupled to a flavodoxin that fuels the nitrogenase (54). However, we could not detect any putative novel candidates that could shuttle electrons to the nitrogenase, and we propose reduced ferredoxin as a candidate for electron delivery, a known electron carrier for anabolic reactions. M. thermolithotrophicus would use H₂ oxidation for ferredoxin reduction, an endergonic process that requires the coupling of the influx of sodium ions by the Ech complex. The reduced ferredoxin would then drive the reduction of NifH for N₂ fixation (Fig. 6).

Here, we have presented novel insights into the metabolic rebalancing that methanogens employ to accommodate Mo-dependent diazotrophy. Our findings include additional cellular responses to this stress, such as the unexpected CRISPR-Cas upregulation. This upregulation is most probably caused by the expression of putative virus-like elements (Fig. 6). Further studies at the protein level are required to decipher the mechanism of this complex adaptation in greater detail. In particular, molecular investigations of the regulatory functions of NifI_1,2, the role of NifX, and the intrinsic properties of the thermostable NifHDK complex will unveil the secrets of the astonishing diazotrophic capabilities of M. thermolithotrophicus.

MATERIALS AND METHODS

Growth conditions.

Methanothermococcus thermolithotrophicus DSM 2095 (Leibniz Institute DSMZ, Braunschweig, Germany) was grown under anoxic conditions in minimal mineral medium with 1 ×10⁵ Pa overpressure of either H₂:CO₂ (80%:20%) for nondiazotrophic cultures or 1.2 × 10⁵ Pa overpressure H₂:CO₂:N₂ (58.2%:14.5%:27.3%) for diazotrophic cultures. Cultures were grown in 250-mL serum flasks (Glasgerätebau Ochs, Bovenden, Germany) sealed with rubber stoppers and aluminum crimps in final volumes of 10 mL with a 1:10 inoculum. Serum flasks and media were made anoxic prior to inoculation by sparging with N₂ and two final gas exchanges with H₂:CO₂ (80%:20%). Incubation was done at 65°C, in the dark, without shaking.

Medium composition.

The minimal mineral medium was prepared as described by Müller et al. (40), but with replacement at an equal final concentration of Fe(NH₄)₂(SO₄)₂·12H₂O by FeCl₂·4H₂O and of Na₂SeO₃·5H₂O by Na₂SeO₄. The used trace metal solution when 100-fold concentrated contained 7.1 mM nitrilotriacetic acid, 0.45 mM MnCl₂·2H₂O, 0.68 mM FeCl₃·6H₂O, 0.41 mM CaCl₂, 0.76 mM CoCl₂, 0.66 mM ZnSO₄·6H₂O, 0.28 mM CuSO₄, 0.19 mM Na₂MoO₄·2H₂O, 0.38 mM NiCl₂·6H₂O and 0.19 mM VCl₃. The final pH was adjusted to 6.0 by addition of NaOH pellets. The final media were subsequently made anoxic by several degassing and N₂ addition cycles (minimum, 25 cycles). The same medium, but without NH₄Cl, was used for diazotrophic cultures. Na₂S was used as both a reductant and sulfur source at a final concentration of 1.5 mM in all cases.

Adaptation to diazotrophic conditions.

M. thermolithotrophicus was adapted to growth under diazotrophic conditions after NH₄Cl depletion from the media by three successive transfers to the same media without NH₄Cl, with a headspace containing 1.2 × 10⁵ Pa H₂:CO₂:N₂ (58.2%:14.5%:27.3%) as described above. Cultures grown in media without NH₄Cl with a gas phase of 1 ×10⁵ Pa H₂:CO₂ (80%:20%) were used as negative controls.

Influence of trace metal availability on diazotrophic growth and tungstate inhibition.

To determine influences of Mo and V on diazotrophic growth, we depleted media of Mo and V by three successive transfers of diazotrophic M. thermolithotrophicus cultures to media prepared as described above, but without Na₂MoO₄·2H₂O or VCl₃ or without both. The minimal concentration of Mo for diazotrophic growth was determined in a series of incubations, in which an already Mo-depleted diazotrophic culture was supplemented with Na₂MoO₄·2H₂O concentrations ranging from 0.01 to 100 μM. W inhibition was tested by supplementing the already W-depleted diazotrophic culture with 0.001 to 10 μM Na₂WO₄·2H₂O; a supplementation with the concentration of 100 μM gave unreproducible results (data not shown). W depletion was done as described above, with transfers to the media prepared without Na₂WO₄·2H₂O.

Cultivation in a fermenter.

M. thermolithotrophicus was continuously grown in a 10-liter fermenter (BIOSTAT B plus; Sartorius, Göttingen, Germany) under diazotrophic conditions with either 10 mM Na₂SO₃ or Na₂SO₄ as sulfur source instead of Na₂S. The final culture volumes were 7 liters for the culture grown with Na₂SO₃ and 6 liters for the culture grown with Na₂SO₄. The culture was continuously sparged with H₂:CO₂ (80%:20%) and N₂ in the ratio of 1:1 and stirred with the speed of 500 rpm, at 65°C. As an inoculum, cultures cultivated in the same media were used in a ratio of 1:10.

Ammonia measurement.

Ammonia concentrations were measured in the culture supernatant. Aliquots of 0.5 mL of culture were subsampled aerobically at each time point, cells were pelleted by centrifugation for 5 min at 15,700 × g using 5415R microcentrifuge (Eppendorf, Hamburg, Germany), and the supernatant was frozen at −20° C until further use. Ammonia was measured by the salicylate-nitroprussidine method (90) in ROTILABO F-profile microtitration plates (Carl Roth GmbH, Karlsruhe, Germany). Standards in the range from 0 to 600 μM NH₄Cl were prepared in the same medium used for cultivation. An 80 μL volume of salicylate reagent (424.7 mM sodium salicylate, 193.8 mM trisodium citrate dihydrate, 193.8 mM disodium tartrate dihydrate, and 0.95 mM sodium nitroprusside dihydrate) and 80 μL volume hypochlorite reagent (10% sodium hypochlorite and 1.5 M NaOH mixed in the ratio of 1:36) were added to 40 μL of each sample. The plate was additionally mixed for 5 min on a shaker and incubated in the dark at the room temperature for 45 min. The absorbance was measured at 650 nm using an Infinite 200 PRO plate reader (Tecan, Männedorf, Switzerland) at room temperature.

Evolutionary analyses.

NifHDK, VnfHDK, and AnfHDK sequences of 35 selected species were aligned using MUSCLE (91) (default parameters) in MEGA11 (92, 93). Afterwards, ambiguous positions were removed for each possible pairing (pairwise deletion option). The final alignment contained a total of 485 NifH positions, 729 NifD positions, and 563 NifK positions. The sequence’s evolutionary history was inferred using the neighbor-joining method (94) with the JTT matrix model for multiple substitutions (95). The analysis was conducted in MEGA11 with ChlLNB from Chlorobium limicola as outgroup.

Transcriptomics setup.

The cultures for transcriptomic profiling were grown in triplicates in batches as described above, but scaled up to culture volumes of 60 mL in 1-liter pressure-resistant Duran bottles. Inocula used to start the cultures were adapted to the respective conditions prior to inoculation, as described above. Subsamples were taken anaerobically on ice after 3 h, 21 h, and 25 h and transferred to an anaerobic chamber, where they were pelleted and subsequently frozen in liquid nitrogen immediately after being taken out of the anaerobic chamber. Samples were stored at −80°C until further use. Samples were shipped on dry ice to the Max Planck Genome Centre in Cologne, where they were processed and sequenced. RNA extraction and quality control, including rRNA removal, were also done at the Max Planck Genome Centre in Cologne.

Transcriptome sequencing and analysis.

Transcriptome sequencing was performed on an Illumina HiSeq 3000 system (San Diego, CA, USA). Information on the raw reads is summarized in Table S1. Raw RNA reads were quality trimmed and repaired using the bbduk and repair.sh scripts of the BBMap v35.14 suite (https://sourceforge.net/projects/bbmap/). Reads with a minimum length of 70 bp and a quality score of 20 were filtered for rRNAs using SortMeRNA v3.0.3 (96). Remaining mRNA reads were mapped against the M. thermolithotrophicus reference genome using Bowtie2 (97) as part of the SqueezeMeta v1.3.1 pipeline (98). The DESeq2 R package (99) was subsequently used to calculate log₂ fold changes, standard errors, test statistics, and adjusted P values. Changes in expression levels with adjusted P values (P_adj) of <0.05 and a minimum 2-fold change ratio (log₂ fold change of ≥1) were considered significant (54).

The high-quality genome of M. thermolithotrophicus was used as a reference for mapping RNA reads. The genomic DNA of M. thermolithotrophicus was isolated using the protocol from Platero et al. (100) and sequenced on PacBio Sequel II platform using a single SMRT cell at the Max Planck Genome Center in Cologne. The genome was assembled using Flye v2.7 (101) and annotated as part of the SqueezeMeta pipeline. Briefly, the ORFs were predicted using Prodigal (102), and similarity searches for GenBank (103), eggNOG (104), and KEGG (105), were done using Diamond (106). HMM homology searches were done with HMMER3 (107) for the Pfam database (108).

Structural modeling and bioinformatic analysis.

Alphafold2 was run with default parameters for all generated models. Predictions of membrane regions and overall topology were run on the DeepTMHMM server (https://dtu.biolib.com/DeepTMHMM/) (109) as of 20 June 2022.

Data availability.

The M. thermolithotrophicus genome sequence and the transcriptome raw reads are available under ENA project number PRJEB53446.

TABLE S5

Accession numbers for sequences used in the phylogenetic reconstruction. Download Table S5, XLSX file, 0.01 MB ^{(12.1KB, xlsx)} .

Copyright © 2022 Maslać et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.