Methylotrophy is a metabolic capability possessed by microorganisms that allows them to build biomass and to obtain energy from organic substrates containing no carbon–carbon bonds (C1 compounds, such as methane, methanol, etc.). This phenomenon in microbial physiology has been a subject of study for over 100 years, elucidating a set of well-defined enzymatic systems and pathways enabling this capability. The knowledge gained from the early genetic and genomic approaches to understanding methylotrophy pointed towards the existence of alternative enzymes/pathways for the specific metabolic goals. Different combinations of these systems in different organisms suggested that methylotrophy must be modular in its nature. More recent insights from genomic analyses, including the genomes representing novel types of methylotrophs, seem to reinforce this notion. This review integrates the new findings with the previously developed concept of modularity of methylotrophy.
Methylotrophy is defined as the ability by some microorganisms to utilize reduced carbon substrates containing no carbon–carbon bonds as their sole sources of carbon and energy (Anthony, 1982; Lidstrom, 2006). The substrates supporting methylotrophic growth include methane and methanol, as well as methylated amines, halogenated methanes and methylated sulfur species. As such, methylotrophs play an important role in global cycling of carbon, nitrogen and sulfur (Kelly and Murrell, 1999; Naqvi et al., 2005; Schäfer et al., 2007; Trotsenko and Murrell, 2008; Singh et al., 2010; Boden et al., 2011). As a metabolic phenomenon, methylotrophy has been known since the early 20th century (Kaserer, 1906; Söhngen, 1906). As a field, methylotrophy was formed and further defined by a number of landmark studies devoted to the biochemistry of methylotrophy that involved specific enzymes and pathways, using a handful of model organisms. The knowledge that emerged from these studies has separated methylotrophs into a few well-defined categories, such as facultative (utilize both single carbon and multicarbon compounds) versus obligate methylotrophs (utilize only single carbon compounds), Type I versus Type II methanotrophs (classification based on the main C1 assimilatory pathway utilized and the shape of intracellular membrane formations; Lidstrom, 2006; Trotsenko and Murrell, 2008), and autotrophic versus heterotrophic methylotrophs (Anthony, 1982; Lidstrom, 2006). Phylogenetically, these belonged to a small number of genera within Alpha-, Beta- and Gammaproteobacteria and within Actinobacteria (Anthony, 1982; Lidstrom, 2006). More recently, this clear and simple picture has been transformed by the new data from genomic and metagenomic approaches, and from analyses of novel groups of microbes, suggesting that not only novel combinations of known enzymes and pathways, but also entirely novel methylotrophy pathways may exist (Kane et al., 2007; Hou et al., 2008; Chistoserdova et al., 2009).
This review offers an updated account of the diversity of known methylotrophy metabolic modules, or a ‘list of parts’, provides a few examples of how these parts fit together in methylotrophs of different taxonomic backgrounds, and includes some speculations on the origins and the evolution of methylotrophy. This review further promotes the concept of modularity of methylotrophy previously developed by myself and colleagues, specifically with relation to aerobic methylotrophs (Chistoserdova et al., 2003; 2005a; 2009) for which a multitude of alternative pathways important for methylotrophy have been described. However, pathways for anaerobic methylotrophy by Clostridia or by Archaea are beyond the scope of this review and only some elements of these, relevant to aerobic methylotrophy, are mentioned.
In a very simplistic way, methylotrophy can be divided into three specific metabolic goals: (i) oxidation (demethylation, dehalogenation) of a primary methylated substrate, producing formaldehyde, methyl- or methylene radical (typically transferred onto tetrahydrofolate; H4F), (ii) oxidation of formaldehyde (methyl- or methylene-H4F) to CO2, and (iii) assimilation of a C1 unit, the latter taking place at the levels of formaldehyde [the ribulose monophosphate (RuMP) cycle], CO2[the Calvin–Benson–Bassham (CBB) cycle], or a combination of methylene-H4F and CO2 (the serine cycle; Fig. 1). Described below are the major and the most well-studied pathways and enzymes that are used for these tasks by methylotrophs, or ‘the parts’ of methylotrophy.
The two well-studied types of methane-oxidizing enzymes, the soluble methane monooxygenase (sMMO) and the membrane-bound methane monooxygenase (pMMO) remain the hallmark enzymes in the aerobic methane oxidation (Hakemian and Rosenzweig, 2007; Trotsenko and Murrell, 2008). The pMMO appears to be the enzyme almost universally widespread among methanotrophs, while the sMMO is only found in some organisms, typically in addition to pMMO, a single exception so far being members of the family Beijerinckiaceae that only possess sMMO (Dedysh et al., 2000; Chen et al., 2010a; Vorobev et al., 2011). The diversity of the pMMO enzymes has been expanded recently to include orthologues that significantly diverge in sequence, and gene clusters encoding these orthologues tend to be present in the genomes of known methanotrophs, along with the ‘canonical’ pMMO gene clusters (Tavormina et al., 2010). In some cases, a function has been determined for a divergent orthologue. For example, a divergent pMMO has been shown to be responsible for high-affinity methane oxidation in Methylocystis sp. strain SC2, with the ‘canonical’ enzyme being responsible for low-affinity methane oxidation (Baani and Liesack, 2008). However, in most cases, functions of the divergent pMMO orthologues remain unknown (Tavormina et al., 2010).
The importance of pMMO originally demonstrated in methanotrophic Proteobacteria continues to hold when it comes to the newly discovered methanotroph phyla, such as aerobic methanotrophic Verrucomicrobia (Methylacidiphilum infernorum; Hou et al., 2008; Op den Camp et al., 2009) and anaerobic methanotrophs belonging to a candidate Phylum NC10 (Candidatus Methylomirabilis oxyfera; Ettwig et al., 2010). Although these recentlydiscovered organisms are unrelated to proteobacterial methanotrophs, they appear to employ typical if evolutionarily diverged pMMOs to metabolize methane. The genomes of verrucomicrobial methanotrophs contain multiple copies of pMMO gene clusters, one of which encodes a divergent enzyme, all copies being expressed (Hou et al., 2008; Op den Camp et al., 2009), while the genome of Candidatus M. oxyfera contains a single pMMO gene cluster. In the latter case, even though the organism grows anaerobically, methane oxidation by pMMO still requires oxygen, which is proposed to be generated by a novel enzyme, NO dismutase, the identity of which remains unknown (Ettwig et al., 2010). Thus, this latter process represents a ‘false’ anaerobic methane oxidation.
True anaerobic methane oxidation so far has only been demonstrated in Archaea, and, based on metagenomic analyses of the communities enriched in ANME type archaea (Knittel and Boetius, 2009), it is proposed to proceed through a reverse methanogenesis pathway (Hallam et al., 2004; Thauer and Shima, 2008; also see discussion by Chistoserdova et al., 2005b). Recent experiments demonstrating methane activation by methyl-CoM reductase from a methanogenic archaeon (Scheller et al., 2010) not only corroborate this hypothesis, but also indicate that all methanogens carry a potential in methane oxidation. While, at the moment, oxidation of methane via reverse methanogenesis is an attribute of Archaea, it is not inconceivable that a similar pathway may be present in yet undiscovered anaerobic bacteria. It has been now well documented that gene complements enabling a partial reverse methanogenesis pathway are present in diverse groups of bacteria (Chistoserdova et al., 2004a; Kalyuzhnaya et al., 2005a), including methylotrophs, in which this pathway serves for formaldehyde oxidation (Chistoserdova et al., 1998; 2005c and see below).
Over decades, much of the work in the field of methylotrophy concentrated on methanol dehydrogenase (MDH), an enzyme catalysing methanol oxidation and another hallmark of methylotrophy (Anthony, 1982; 2004; Williams et al., 2005). A MDH first characterized in Methylobacterium extorquens (Anthony, 1982; 2004) and later detected in most of the model methylotrophs was a heterotetramer consisting of subunits encoded by genes designated mxaFI, contained pyrroloquinoline quinone (PQQ) as cofactor and required a cytochrome c electron acceptor encoded by mxaG that was part of the same gene cluster (Lidstrom et al., 1994; Vuilleumier et al., 2009), along with the accessory genes encoding functions required for inserting Ca++ into the active centre of the enzyme (mxaACKL) and other yet unknown functions (mxaRS) essential for generating an active MDH. However, while the enzyme has been now studied in great detail, it is becoming clear that the classic MDH is not every methylotroph's choice. Indeed, few mxaF genes are found in metagenomic data sets representing important environmental samples. For example, while Methylophilus-like and Roseobacter-like sequences represent some of the most abundant ribotypes in global ocean samples (Rusch et al., 2007), few mxaF homologues were recorded in the same metagenomic data set (L. Chistoserdova, unpubl. obs.). Likewise, analysis of a Lake Washington metagenomic data set specifically enriched in methylotroph sequences via stable isotope probing (Kalyuzhnaya et al., 2008a) revealed the presence of few mxaF genes, suggesting that the distribution of this metabolic module in freshwater environments is also limited.
Recently, it has been demonstrated that the newly discovered methylotrophs belonging to the orders of Burkholderiales and Rhodocyclales encode a different type of MDH, named MDH2 (Kalyuzhnaya et al., 2008b). The properties of the partially purified enzymes responsible for the MDH activity in these species were somewhat different from those exhibited by typical mxaFI-encoded enzymes, and the enzyme appeared to be composed of a single type of subunit. Protein sequences translated from mdh2 genes showed very low similarity to MxaF proteins (less than 35% amino acid identity). However, they were highly similar (up to 80% amino acid identity) to a class of PQQ-linked dehydrogenases that included alcohol dehydrogenases, typically revealing low affinity for methanol (Kalyuzhnaya et al., 2008b).
A number of methylotrophs have now been characterized that are capable of methanol metabolism but contain neither mxaFI nor mdh2 (Giovannoni et al., 2008; Hou et al., 2008; Kalyuzhnaya et al., 2008b). These must possess a different type of MDH. One gene, named xoxF, encoding a polypeptide whose sequence is very similar to the sequence of MxaF (approximately 50% amino acid identity) has been proposed to encode an enzyme replacing the classic MDH (Giovannoni et al., 2008; Hou et al., 2008; Wilson et al., 2008). However, the role of XoxF in methylotrophy has not been unequivocally proven (see discussion below).
Gram-positive methylotrophic bacteria possess PQQ-independent, NAD(P)-binding alcohol dehydrogenases revealing no homology to MDH enzymes from Gram-negative bacteria. These form decameric structures of a single type of subunit and possess broad substrate specificities, similarly to MDHs from Gram-negative bacteria (Bystrykh et al., 1993; Brautaset et al., 2004).
Oxidation of methylamine can be carried out either via direct oxidation to formaldehyde, by methylamine dehydrogenase (MADH; in Gram-negative bacteria) or by methylamine oxidase (in Gram-positive bacteria), or via an indirect pathway that involves transfer of the methyl group onto glutamate, followed by the oxidation of N-methylglutamate (NMG), presumably to methylene-H4F (Chistoserdova et al., 2009; Latypova et al., 2010). Of these known systems, the MADH system is most well studied and represents another hallmark of methylotrophy (Davidson, 2004; Wilmot and Davidson, 2009). So far, all the organisms possessing the MADH enzyme system, encoded by a number of conserved genes (mauBEDAGLMN), have been demonstrated to be bona fide methylotrophs. The mau gene clusters and respective enzymes and accessory proteins are well conserved among Alpha-, Beta- and Gammaproteobacteria, with the exception of the specific electron acceptor that may be represented by an amicyanin (Chistoserdov et al., 1994), an azurin (Gak et al., 1995) or a cytochrome (Kalyuzhnaya et al., 2008a). In some organisms, the mau gene cluster was found to be flanked by IS elements, suggesting that the methylamine oxidation capability may be a subject of lateral gene transfers (Vuilleumier et al., 2009). Indeed, there are a few examples of closely related methylotroph species differing with regard to possessing the mau genes (Kalyuzhnaya et al., 2008a; Vuilleumier et al., 2009), supporting this hypothesis.
While, in terms of its environmental distribution and the role in methylamine-degrading capability, the NMG pathway may be as important if not more important than MADH, the genetic determinants for this pathway have been revealed only recently (Latypova et al., 2010), indentifying three specific enzymes involved in the pathway, γ-glutamylmethylamide synthase, N-methylglutamate synthase and N-methylglutamate dehydrogenase, often encoded by a single cluster of genes (Latypova et al., 2010; Chen et al., 2010b). In some methylotrophs, the NMG pathway is the only pathway for methylamine oxidation (Latypova et al., 2010; Chen et al., 2010b), while in others it is present along with MADH, playing a secondary role (Hendrickson et al., 2010). The NMG pathway is also present in non-methylotrophs, which employ it for utilization of methylamine as a nitrogen source (Chen et al., 2010c). So far, like MADH, this pathway has only been found in Proteobacteria and has also been proposed to be a potential subject of lateral gene transfers (Chen et al., 2010c).
Few methylamine oxidases have been characterized in Gram-positive methylotrophs so far (Hartmann and McIntire, 1997; Wilce et al., 1997). Based on sequence homology, they are closely related to a range of amine oxidases with different substrate specificities, all covalently binding cofactor topaquinone and requiring Cu2+ ions. Such amine oxidases are widespread among Gram-positive and Gram-negative bacteria and eukaryotic organisms. As a single gene is sufficient to encode such an enzyme, genome-based predictions of methylotrophy are difficult based on a single respective gene.
Other primary C1 substrate-oxidizing pathways
A variety of other methylated compounds serve as substrates for methylotrophs, and their metabolism is enabled by a number of well-characterized specific modules. Some of these modules are involved in degradation of important green house gases other than methane, highlighting the role of methylotrophs in global cycling of not only carbon but nitrogen, sulfur and halogenated compounds (Kelly and Murrell, 1999; Schäfer et al., 2007; Trotsenko and Murrell, 2008; Boden et al., 2011). Degradation of monohalogenated methanes such as chloromethane or bromomethane involves two specific methyltransferases, CmuA and CmuB, the former transferring the methyl group of a monohalogenated methane to a corrinoid protein, and the latter further transferring it to H4F (Vannelli et al., 1999; Schäfer et al., 2007). Degradation of dihalogenated methanes is carried out by a specific dehalogenase/glutathione S-transferase (DcmA; Leisinger et al., 1994). Some of the most environmentally important methylated sulfur species are dimethylsulfoniopropionate (DMSP), dimethylsulfide (DMS) and methanesulfonic acid (MSA; Kelly and Murrell, 1999; Yoch, 2002). The latter two are known to promote growth of methylotrophs, while degradation of DMSP has not been described as a bona fide case of methylotrophy so far. Microbial degradation of DMSP is known to proceed via two alternative routes, demethylation (catalysed by DMSP-dependent demethylase DmdA) and cleavage, catalysed by a few different types of DMSP lyases producing DMS (Todd et al., 2011). The demethylation pathway appears to be the predominant pathway in nature, reportedly utilized by over half of the bacterioplankton cells in ocean surface waters (Howard et al., 2011). Some of the bacteria degrading DMSP rely on it as the source of sulfur (Tripp et al., 2008), while others, such as Ruegeria (Silicibacter) pomeroyi, must also be able to utilize it as a carbon source (see below). DMS is degraded by the action of DMS monooxygenase (DmoAB; Boden et al., 2011). The specific module responsible for degradation of MSA is the MSA-specific monooxygenase (MsmABCD; Baxter et al., 2002). Environmentally important methylated nitrogen-containing compounds upstream of methylamine include trimethylamine and its product dimethylamine, the modules for their degradation being respective dehydrogenases (TMAD, DMAD) and respective monooxygenases (TMAM, DMAM; Anthony, 1982; Fig. 1). The TMA/DMA-oxidizing enzymes are not methylotrophy-specific as they are known to function in nitrogen metabolism of diverse bacteria (De Boer et al., 1989; Liffourrena et al., 2010). It is likely that additional, not yet recognized primary oxidation modules supporting methylotrophic metabolism exist, and these are still awaiting to be discovered.
Oxidation of formaldehyde is an important step in methylotrophy, not only in terms of energy generation but also in terms of keeping the intracellular levels of formaldehyde at non-toxic levels. Methylotrophs have a variety of formaldehyde-oxidizing systems from which to chose. These range from a single gene/single enzyme systems [NAD-linked formaldehyde dehydrogenases (FaDH); mycothiol-linked FaDH] to multigene/multienzyme cofactor-linked C1 transfer pathways (Vorholt, 2002; Chistoserdova et al., 2009). One clever way to oxidize formaldehyde is by employing the reactions of the assimilatory RuMP cycle with the addition of one single enzyme, 6-phosphogluconate dehydrogenase (Anthony, 1982). Most well-studied methylotrophs possess more than one means for oxidizing formaldehyde, while in novel and less well-studied methylotrophs, formaldehyde-oxidizing systems are yet to be experimentally defined (Giovannoni et al., 2008; Hou et al., 2008).
It is intriguing to observe that some of the simple and very efficient single enzyme systems such as NAD-linked, glutathione-independent FaDH (exemplified by the well-studied enzyme from Pseudomonas putida; Tanaka et al., 2003) are not found in most methylotrophs. However, they appear to be the major formaldehyde detoxification systems in organisms such as Pseudomonas and non-methylotrophic Burkholderia species that, like methylotrophs, tend to possess multiple formaldehyde oxidation systems (Marx et al., 2004; Roca et al., 2009). Conversely, one of the most elaborate pathways, the one employing tetrahydromethanopterin (H4MPT) as a cofactor and encoded by at least 20 genes, is the most widespread pathway for formaldehyde oxidation in methylotrophs (Chistoserdova et al., 2009). This pathway was first discovered in M. extorquens AM1 (Chistoserdova et al., 1998) and demonstrated to be an indispensable pathway for both formaldehyde oxidation with generation of NAD(P)H and formaldehyde detoxification is this organism (Hagemeier et al., 2000; Marx et al., 2003a). Mutants of M. extorquens with lesions in the genes involved in this pathway are all methylotrophy-negative, and some demonstrate a remarkable threshold for methanol sensitivity (as low as 90 µM while wild-type organism grows at 100 mM; Chistoserdova et al., 2005c). However, in the organisms possessing dissimilatory RuMP cycle, the H4MPT-linked pathway appears to serve an auxiliary function as the cyclic pathway is the major pathway in oxidizing and detoxifying formaldehyde, and mutants with lesions in this pathway can be easily generated (Chistoserdova et al., 2000; Kalyuzhnaya et al., 2005a). Some autotrophic methylotrophs, such as Paracoccus and the closely related Rhodobacter species (the latter use methanol photosynthetically), rely on the glutathione (GSH)-linked formaldehyde oxidation pathway as the major means for both energy generation and formaldehyde detoxification (Ras et al., 1995; Barber and Donohue, 1998), and these species do not contain genes or activities for the H4MPT-linked pathway. With this respect it is interesting to point out that heterologously expressed genes for this pathway were able to complement mutants of M. extorquens AM1 deficient in genes of the H4MPT-linked pathway (Marx et al., 2003a).
Gram-positive methylotrophs, beside the cyclic route, employ mycothiol-dependent FaDHses (Vorholt, 2002). In addition, dye-linked enzymes have been described for some methylotrophs (Klein et al., 1994) but their role in methylotrophy remains poorly understood.
A C1 transfer pathway linked to tetrahydrofolate (H4F), analogous to the H4MPT-linked pathway, exists in many methylotrophs (and across all life). Typically, this pathway involves FolD, a bifunctional enzyme possessing methylene-H4F dehydrogenase and methenyl-H4F cyclohydrolase activities. Most methylotrophs for which genome sequences are available possess this gene (see Table 1), and in methylotrophs that grow on compounds whose primary degradation involves demethylation reactions, such as chloromethane-degrading M. extorquens (formerly chloromethanicum) CM4, FolD appears to be specifically involved in dissimilation of these compounds, along with respective methyltransferases (CmuAB), methyl-H4F reductase (MetF) and formyl-H4F hydrolase (PurU; Vannelli et al., 1999).
Table 1. Examples of occurrence of methylotrophy metabolic modules in major functional groups.
However, some methylotrophs, for example M. extorquens AM1 and Methylococcus capsulatus, do not posses FolD. Instead, they posses two alternative enzymes, MtdA and Fch, responsible for methylene-H4F dehydrogenase and methenyl-H4F cyclohydrolase activities respectively. These enzymes have been characterized in M. extorquens AM1, and, based on enzyme kinetics (Vorholt et al., 1998; Pomper et al., 1999), mutant phenotypes (Chistoserdova and Lidstrom, 1994; Marx and Lidstrom, 2004) and metabolite flux analysis (Marx et al., 2005; Crowther et al., 2008), they appear to be involved in a pathway that acts in a reductive rather than oxidative direction in this organism, i.e. this pathway enables transfer of C1 units into the serine cycle using H4F as an adduct (Crowther et al., 2008). The latter study thus challenged the long-standing dogma of spontaneous reaction between formaldehyde and H4F being of physiological significance. The difference in physiological roles of MtdA and FolD is highlighted by mutant complementation experiments: while MtdA could act as a functional substitute for FolD (Studer et al., 2002), FolD could not complement a lesion in MtdA (Marx and Lidstrom, 2004).
To complicate the matters further, a novel class of Mtd enzymes, named MtdC, have been recently described. These are so far found in representatives of the division Planctomycetes and in a novel deeply divergent phylum of uncultivated bacteria from Lake Washington (Vorholt et al., 2005), which I will term here Phylum LW for convenience (Supporting information, Fig. S1). These enzymes had high affinities for methylene-H4MPT and NADP but low affinities for methylene H4F or NAD, distinguishing them from MtdA and MtdB enzymes (Supporting information, Table S1). Accordingly, it has been demonstrated that MtdC cannot functionally substitute for either MtdA or MtdB (Vorholt et al., 2005). Thus, while phylogenetically more related to MtdA, MtdC must fulfil a function more similar to the function of MtdB, as part of the H4MPT-linked pathway for formaldehyde oxidation/detoxification (Vorholt et al., 2005).
Analysis of the recently available genomic sequences suggests that the diversity of Mtd enzymes must extend beyond the Mtd enzymes described above. While the divergent Mtd protein in the newly described genome of NC10 phylum methylotroph Candidatus M. oxyfera is likely a divergent MtdB (Supporting information, Fig. S1), the divergent Mtd proteins in the genomes of Methylophaga thiooxidans, Nitrosococcus halophilus and Anaerobaculum hydrogeniformans may represent new classes and may express a different set of substrate specificities, based on their phylogenetic positions (Supporting information, Fig. S1). It is noteworthy that so far genes encoding Fch enzymes have only been found in organisms possessing MtdA but not MtdC, suggesting that the novel Mtd enzymes are likely employed in the dissimilatory rather than assimilatory pathways.
Understanding the specific functions of different variants of H4F-utilizing C1 transfer pathways is further complicated by the use of two alternative enzymes for transferring C1 units between the levels of formyl-H4F and formate: formyl-H4F ligase (FtfL) being a reversible enzyme (Marx et al., 2003b) and formyl-H4F hydrolase (PurU) being a non-reversible enzyme (Nagy et al., 1995). Thus, in the organisms possessing FtfL the pathway can in principle operate in both directions while in the organisms possessing only PurU the pathway can only operate in the oxidative direction. It is remarkable that the organisms utilizing the MtdA/Fch variant of the pathway always employ FtfL, while organisms utilizing the FolD variant can employ either FtfL or PurU or both (Table 1). Few methylotrophs, exemplified by Methylibium petroleiphilum and M. extorquens CM4 encode both variants of the pathway (Vannelli et al., 1999; Studer et al., 2002; Kane et al., 2007). In the latter, PurU was demonstrated to specifically operate in degradation of chloromethane, and FtfL cannot functionally substitute for PurU (Vannelli, 1999; Studer et al., 2002). However, from the experimental data available, it is not clear whether FtfL is simply not present at sufficient levels under these growth conditions or whether it can only operate in the reductive direction in this organism.
Methylotrophs, like other organisms, tend to encode multiple formate dehydrogenases (FDH). For example, M. extorquens species possess four different FDH enzymes, which include a tungsten-containing FDH (FDH1; Laukel et al., 2003), a predicted molybdenum-containing FDH (FDH2; Chistoserdova et al., 2004b), a predicted cytochrome-linked FDH that is likely periplasmic (FDH3; Chistoserdova et al., 2004b), and a novel type of FDH (FDH4; Chistoserdova et al., 2007a), and all of these appear to be functional. However, when each of the enzymes was disrupted in a wild-type background, mutants in only one of them, FDH4, revealed a phenotype: these mutants accumulated formate in the late stationary phase of growth on methanol (Chistoserdova et al., 2007a). If an intact FDH4 was present, the three other enzymes appeared redundant during growth on methanol. However, the situation was different when the organism grew on formate. FDH4 alone could not sustain the growth, and at least one of the other FDH enzymes was needed (Chistoserdova et al., 2004b). FDH4 in this organism also appeared to be somehow involved in acid stress response, but the mechanism of this involvement remains unknown (Chistoserdova et al., 2007a).
In organisms oxidizing formaldehyde via the RuMP cycle, the formate oxidation step is expected to be less critical for methylotrophy, and, accordingly, activities of FDH enzymes are typically measured at very low levels (Anthony, 1982). Relative contributions of alternative FDHses to methylotrophy were recently investigated in a RuMP cycle methylotroph, Methylobacillus flagellatus that possesses FDH enzymes homologous to FDH2 and FDH4 of M. extorquens (Hendrickson et al., 2010). It was concluded from this study that the formate oxidation step is essential even when the dissimilatory RuMP cycle is functionally present, as no mutants could be generated simultaneously lacking both FDHses, and FDH4 appeared to play a more important function in fitness of this organism compared with FDH2 (Hendrickson et al., 2010).
Overall, based on the available genomic data, FDHses are not conserved among methylotroph groups, but rather these are randomly distributed. For example, close homologues of the FHD2 enzymes described above are widespread across methylotrophic and non-methylotrophic Proteobacteria, while non-proteobacterial methylotrophs encode FDH enzymes homologous to the ones of other representatives of respective (or related) phyla (Hou et al., 2008; Ettwig et al., 2010).
In methylotrophic bacteria, C1 assimilation takes place either at the level of formaldehyde, via the RuMP cycle, at the level of methylene-H4F and CO2, via the serine cycle, or at the level of CO2, via the CBB cycle. The two former pathways are considered specific to methylotrophy while the latter pathway is shared with non-methylotrophic autotrophs (as well as plants).
The RuMP cycle
The RuMP cycle functions in adding a molecule of formaldehyde onto a C5 sugar (RuMP), thus producing a familiar currency of universally widespread cellular metabolism, glucose 6-phosphate, part of which is then converted into C3 molecules and used for cell biosyntheses, another part is oxidized generating NAD(P)H, and the remaining part is used to regenerate the acceptor molecule (Anthony, 1982). Only two specific enzymes, hexulosephosphate synthase (HPS) and hexulosephosphate isomerase (HPI), are needed to enable the RuMP cycle, in addition to the enzymes of glycolysis, pentosephosphate cycle and Entner–Doudoroff pathways. High activities of HPS and HPI are typically signatures of methylotrophy via the RuMP cycle (Anthony, 1982; Lidstrom, 2006). However, the presence of the hps/hpi gene pair is not always a sign of methylotrophy capability as the same cycle is utilized by some non-methylotrophs, serving as an additional means for formaldehyde detoxification (Yasueda et al., 1999). Indeed, homologues of hps/hpi are found in the genomes of a variety of non-methylotrophic species, including Bacillus, Escherichia, etc. and also in some Archaea (Kato et al., 2006).
The serine cycle. In the serine cycle, methylene-H4F and CO2 are assimilated through a series of reactions involving amino acids and organis acids, to result in C3 and C4 metabolites for building biomass and glyoxylate that serves as the acceptor for methylene-H4F (Anthony, 1982; Chistoserdova et al., 2009; Peyraud et al., 2009). The cycle, while specific to methylotrophy, borrows a number of genes/enzymes from other common metabolic pathways and some of these (such as serine hydroxymethyltransferase, enolase and malate dehydrogenase) have dual functions in methylotrophs, being involved in pathways with different metabolic goals. Key enzymes that are typically measured to indicate the operation of the serine cycle are hydroxypyruvate reductase and serine glyoxylate aminotransferase. Serine cycle genes are typically found in clusters, and are subjects of coordinated regulation (Kalyuzhnaya and Lidstrom, 2003; 2005). Such clusters have been analysed in a number of methylotrophs and potential methylotrophs (Vuilleumier et al., 2009), and their structure and content suggest a complex history. For example, the structure of the serine gene cluster first characterized in M. extorquens AM1 (Chistoserdova et al., 2003) is highly conserved in some alphaproteobacterial methylotrophs, and these clusters are remarkably syntenic with the cluster in a betaproteobacterial methylotroph, M. petroleiphilum (Vuilleumier et al., 2009). However, the genome of R. pomeroyi, also an Alphaproteobacterium, contains a cluster that shows no synteny to the M. extorquens cluster (Vuilleumier et al., 2009). In addition, some of the genes in the cluster reveal very low levels of homology to their counterparts in other Alphaproteobacteria, suggesting independent evolution of the serine cycle in different bacterial lineages. Gene clustering is also different in Hyphomicrobium denitrificans, and some of the genes (such as that encoding phosphoenolpyruvate carboxylase, PPC, HPR) show little homology with other PPC and HPR genes.
Genomes of gammaproteobacterial methanotrophs also harbour most of the serine cycle genes, but few of them are clustered (Ward et al., 2004). However, the gene for one key enzyme of the serine cycle, the PPC, is missing from the genomes. Thus the cycle might be incomplete in these bacteria, or a modification of the cycle using an alternative enzyme is operational. The proteomic analysis of M. capsulatus revealed that most of the serine cycle genes are expressed during growth on methane (Kao et al., 2004).
Glyoxylate regeneration. The assimilation of C1 units via the serine cycle requires regeneration of glyoxylate from acetyl-CoA. It has been recognized early on that, while this task can be carried out by the classic glyoxylate shunt, many serine cycle methylotrophs employ instead an alternative glyoxylate regeneration pathway (Anthony, 1982) that remained a puzzle for over three decades. This pathway has been recently resolved by a combination of efforts from different laboratories, and it is now known as ethylmalonyl-CoA pathway (EMCP; Erb et al., 2007; 2008; Chistoserdova et al., 2009; Peyraud et al., 2009). The pathway shares reactions and enzymes with the serine cycle (malate thiokinase, malyl-CoA lyase), the tricarboxylic acid cycle (succinate dehydrogenase, fumarase), the polyhydroxybutyrate cycle (beta-ketothiolase, acetoacetyl-CoA reductase) and with other metabolic pathways (ethylmalonyl-CoA mutase, propionyl-CoA carboxylase), in addition to the specific reactions such as ethylmalonyl-CoA mutase (Erb et al., 2008) and crotonyl-CoA reductase/carboxylase (Erb et al., 2007). Respectively, genes for the latter enzymes serve as markers for probing genomes for the presence of EMCP. The EMCP is not specific to methylotrophy. It is also utilized for metabolism of C2 compounds by both facultative methylotrophs, such as M. extorquens (Okubo et al., 2010), and non-methylotrophic bacteria such as Rhodobacter and Streptomyces (Alber, 2010). In the latter case, genes for the EMCP are present in the genomes but not the genes for the serine cycle. It is interesting to point out that the EMCP functions as a cycle during growth on C1 compounds (regenerating a molecule of glyoxylate per each molecule of methylene-H4F and each molecule of CO2 assimilated; Chistoserdova et al., 2009), while during growth on C2 compounds, it functions as a liner pathway (Erb et al., 2010; Okubo et al., 2010).
Some methylotrophs, however, opt to use the glyoxylate shunt instead of the EMCP. For example, the genomes of M. petroleiphilum and Methylocella silvestris, while lacking genes for the key enzymes of the EMCP, contain the genes for the glyoxylate shunt (Kane et al., 2007; Chen et al., 2010a) and the genome of H. denitrificans encodes both pathways. In M. capsulatus, neither the EMCP nor the glyoxylate shunt appears to be encoded, consistent with the hypothesis of this pathway fulfilling an auxiliary metabolic task.
The CBB cycle. A number of methylotrophs possess genes and enzymes of the CBB cycle, some as the only means for C1 assimilation and others in addition to other C1 assimilatory cycles. In the latter case, the contribution of the CBB cycle to methylotrophy remains poorly understood (Chistoserdova et al., 2005b). The three major groups that rely on the CBB for methylotrophy are (i) the alphaproteobacterial autotrophs (such as Paracoccus denitrificans and Xanthobacter autotrophicus; Baker et al., 1998), (ii) methanotrophic Verrucomicropbia (M. infernorum; Op den Camp et al., 2009) and (iii) methanotrophs belonging to the NC10 phylum (Ettwig et al., 2010). Based on sequence comparisons, the key CBB genes (cbbLS) may be subjects of lateral transfers. For example, two cbb gene clusters are found in the genome of M. petroleiphilum, one more related to the clusters in other Betaproteobacteria, the other more related to the clusters in Alphaproteobacteria (Kane et al., 2007). While analysis of the genome of the NC10 representative Candidatus M. oxyfera revealed that most of the C1 metabolism genes significantly diverge from the genes found in other groups of methylotrophs, the cbb operon showed high homology with the operons from Proteobacteria (Ettwig et al., 2010). The cbb genes of M. infernorum are most closely related to the genes in Chloroflexi and Actinobacteria.
How ‘the parts’ fit together (module integration)
It has been previously postulated that, to enable methylotrophy, at least one of each of the key modules needs to be present: (i) a primary oxidation (demethylation, or dehalogenation) module, (ii) a formaldehyde oxidation/detoxification [or methyl (methylene)-H4F oxidation] module, and (iii) an assimilation module (Chistoserdova et al., 2003; 2009). This concept not only suggested the existence of organisms possessing a minimal set of methylotrophy metabolic modules, but it also encompassed metabolic scenarios, i.e. module combinations, not yet known to exist. These postulates can now be supported by the analysis of the genomes of some ‘non-canonical’ methylotrophs. Below, a few examples are presented of such ‘non-canonical’ module combinations, along with a few examples of module variations, to support the idea that alternative enzymes/enzymatic systems (modules) are capable of performing a single specific metabolic task in methylotrophy, the modules appearing to be subjects of deletion, duplication, exchange or replacement.
The classic combinations
These have been historically defined by studies employing a finite number of model methylotroph strains, such as Type I methanotrophs (e.g. M. capsulatus), Type II methanotrophs (e.g. Methylosinus trichosporium OB3b), alphaproteobacterial (true) methylotrophs (e.g. M. extorquens), alphaproteobacterial autotrophic methylotrophs (e.g. P. denitrificans) and beta- and gammaproteobacterial (Methylophilus, Methylophaga) and Gram-positive (Amycolatopsis, Bacillus) methylotrophs. From these studies, the serine cycle was attributed to alphaproteobacterial (Type II) methanotrophs and non-methane-utilizing methylotrophs not possessing the CBB cycle. Typically the former would possess a pMMO and a MDH (always MxaFI) but not possess a methylamine oxidation system (Fig. 2B), while the latter would typically possess a MADH (Fig. 2E). The RuMP cycle would be attributed to Type I methanotrophs (Fig. 2A) and beta- and gammaproteobacterial (and Gram-positive) non-methane utilizers (Fig. 2I). The Gram-negative organisms would always possess a MDH (MxaFI) and often possess a MADH while the Gram-positive organisms would possess NAD-MDH and a methylamine oxidase. Autotrophic methylotrophs would employ the classic MDH and MADH enzymes but use the CBB cycle for assimilation (Fig. 2F). Based on these studies, pMMO, MDH and MADH appeared to be the most persistent and most important methylotrophy modules, and these became the hallmark enzymes of methylotrophy, demanding much attention and experimental effort in the field (Hakemian and Rosenzweig, 2007; Zhang et al., 2007; Balasubramanian et al., 2010; Jensen et al., 2010; Meschi et al., 2010). However, emergence of novel types of methylotrophs and the genomic knowledge, combined in some cases with expression and biochemical data, now demonstrate that other combinations of methylotrophy modules are not only possible but may prevail in environmentally significant methylotrophs.
The genome of strain HTCC2181 is one of (if not) the smallest genomes of a free-living organism, at 1.3 Mbp (Giovannoni et al., 2008). With such a small genome, the organism is a true minimalist, as it appears to encode only the essential functions, with few gene (function) duplications. Based on comparisons with other, much larger, Methylophilaceae genomes, it is clear that the genome (and metabolism) of strain HTCC2181 is a result of a massive gene loss (Giovannoni et al., 2008). For this reason, this genome is convenient in defining a minimal set of genetic determinants enabling methylotrophy. Interestingly, the genome encodes neither MDH no MADH, and it also lacks genes for the H4MPT-linked pathway that otherwise serves as a signature of methylotrophy. However, a gene encoding XoxF is present, along with accessory genes (xoxGJ) as well as the genes for PQQ biosynthesis and MDH accessory functions (MxaRSACKL). All the genes involved in the reactions of the RuMP cycle are present. The H4F-linked pathway (involving FolD but not MtdA) is also encoded, along with a single FDH (Fig. 2H).
Methylotrophy in Verrucomicrobia
Methylacidiphilum infernorum possesses the second smallest genome for a characterized methylotroph (2.3 Mbp), and it also encodes a very minimal set of metabolic modules to enable growth on methane. While three pmoCAB gene clusters are present, the only candidate for a methanol-oxidizing system is XoxF. The only recognizable assimilatory pathway, at least from genome analysis, is the CBB cycle. This is a novel and so far very unusual metabolic make up for a methanotroph as the only module in common with the well-characterized proteobacterial methanotrophs is the pMMO (Fig. 2D). While the CBB cycle is encoded in the genomes of some proteobacterial methanotrophs (and in some cases is known to be expressed; Kao et al., 2004), in each such case, either the RuMP (Type I methanotrophs) or the serine cycle (Type II methanotrophs) are present, and these latter are considered as the main assimilatory pathways in these organisms. One problem M. infernorum has in common with strain 2181 is the lack of recognizable formaldehyde oxidation systems, while both encode the H4F-linked pathway (involving in both cases FolD in combination with FtfL). This suggests that either novel formaldehyde-oxidizing systems exist in these microbes, or there must be a way of generating methyl- or methylene-H4F from formaldehyde. This problem also exists for organisms utilizing the serine cycle but not possessing MtdA/Fch, and it is discussed in more detail below.
Methylotrophy in Phylum NC10
As already mentioned, the novel methylotrophs of the yet unnamed Phylum NC10, represented by Candidatus M. oxyfera, are disguised as anaerobic methanotrophs, while still employing the classic oxygen-requiring pMMO enzyme. In this case, the denitrification pathway appears to be directly connected to methylotrophy being a source of oxygen, which is proposed to be generated as a result of a NO dismutase reaction (Ettwig et al., 2010). The details of this novel process require further research. Downstream of the methane oxidation step, methylotrophy takes a more predictable course, employing an MxaFI-type methanol dehydrogenase and the H4MPT-linked formaldehyde oxidation pathway for the dissimilatory metabolism and the CBB cycle for the assimilatory metabolism (Fig. 2C). Thus, Candidatus M. oxyfera is a second example of an obligately autotrophic methanotroph as judged from its genome annotation (Ettwig et al., 2010).
Methylotrophy in Burkholderiales and Rhodocyclales
Until recently, all known betaproteobacterial methylotrophs belonged to the family Methylophilaceae, and all followed the same metabolic scheme: oxidizing methanol and methylamine via, respectively, MDH and MADH, and oxidizing and assimilating formaldehyde via the oxidative and assimilatory branches of the RuMP cycle respectively. The first non-canonical betaproteobacterial methylotroph discovered was M. petroleiphilum of the family Comamonadaceae, order Burkholderiales. With the exception of the H4MPT-linked formaldehyde oxidation pathway, M. petroleiphilum utilized a completely different scheme of methylotrophy, employing a novel MDH type (MDH2) for primary oxidation and the serine cycle for formaldehyde assimilation (Fig. 2G). The latter appears to be borrowed from alphaproteobacterial methylotrophs, based on gene synteny and sequence conservation (Kane et al., 2007; Vuilleumier et al., 2009). Methylotrophs have also been recently described belonging to the families of Rhodocyclaceae and Burkholderiaceae, whose methylotrophy schemes are metabolically and genetically similar to the one of M. petroleiphilum: MDH2 or the NMG pathways being their choices for primary oxidation and the serine cycle their choice for the assimilatory needs (Kalyuzhnaya et al., 2006a; 2008b; Latypova et al., 2010). Thus, methylotrophy in betaproteobacteria is of polyphyletic origin, highlighting the flexibility of this metabolic feature.
Other types of methylotrophy
Clearly, a variety of compounds remain unidentified (such as the components of the dissolved organic matter; McCarren et al., 2010) whose primary oxidation (demethylation) will result in C1 units that could be metabolized or co-metabolized using the typical methylotrophy modules. Some organisms not characterized in the laboratory experiments as bona fide methylotrophs may be reliant on methylotrophy in their natural niches. For example, a number of marine bacteria such as Fulvimarina pelagi, R. pomeroyi, Aurantimonas sp. possess one or more of the specific methylotrophy metabolic modules (the serine cycle, the EMCP, the H4MPT-linked pathway, etc.; Table 1) while missing the well-characterized primary C1 oxidation modules. Likely, these methylotrophy modules are used to metabolize C1 units originating from methylated compounds, such as DMSP (Fig. 2J), or other yet unknown methylated compounds.
Most persistent modules
From the theoretical predictions and from the genomic and experimental data available at the moment, methylotrophy appears to be more versatile than previously thought. Even based on the current knowledge on discrete methylotrophy modules that likely does not represent a complete picture, an almost infinite number of module combinations are possible, and indeed, many different combinations have been already recorded (Table 1). Some modules appear to be more persistent than others thus potentially indicating the metabolic tasks both most critical to methylotrophy and most conserved in terms of the evolution of methylotrophy. Remarkably, these do not include pMMO, MDH or MADH. Based on the current genomic data, one of the most persistent modules is XoxF(JG), so far present in all characterized methylotrophs. Two other most persistent modules are the H4F-linked C1 transfer module involving FolD and FtfL and the module enabling H4MPT-linked C1 transfers. It is important to point out that neither of these modules is strictly methylotrophic. For example, XoxF(JG) are encoded in many bacterial phyla not known to be methylotrophs, being especially widespread among non-methylotrophic Rhizobiales and Burkholderiales. Interestingly, Planctomycetes, most of which encode the H4MPT-linked C1 transfer reactions, do not encode XoxF. FolD is universally widespread in Bacteria and Eukaryotes and is present in some Archaea. The H4MPT-linked C1 transfer pathway (specifically, the bacterial variant involving NAD(P)-specific Mtd enzymes) appears to be more specific to methylotrophy. However, it is also present in most Planctomycetes, in some Synergistetes (A. hydrogeniformans) and in some Proteobacteria without demonstrated methylotrophy capability (Burkholderiales, Nitrosococcales)
While genomic knowledge aids significantly in understanding the biochemical mechanisms of methylotrophy in specific organisms, not only some questions remain unresolved, but new questions arise from this knowledge. Some of the most burning unresolved problems are discussed below.
MtdA versus FolD. Most methylotrophs possess two different pathways for transferring C1 units between different oxidation levels: one linked to H4MPT and one linked to H4F. The latter can be enabled either by the MtdA enzyme (in combination with Fch) that so far remains specific to methylotrophs, and more specifically to methylotrophs possessing the serine cycle, or by FolD that has much more universal distribution. All so far known RuMP methylotrophs only possess FolD, and few organisms (such as M. petroleiphilum) possess both (Table 1). In the case of M. extorquens species, distinct roles for two variants of the pathway have been demonstrated: the MtdA-enabled variant was shown to serve for the synthesis of methylene-H4F used for assimilation, with energy consumption, and the FolD-enabled variant was shown to be essential for oxidation of methyl-H4F to CO2. Based on these discoveries, a new concept of the lack of non-enzymatic condensation between formaldehyde and H4F, replaced the old concept of a spontaneous reaction playing a physiological role (Crowther et al., 2008). However, this new concept now creates a problem. A number of newly sequenced genomes, such as the ones of H. denitrificans and M. silvestris, encode FolD but do not encode MtdA while relying on the serine cycle for assimilation. Thus in these cases either FolD functionally replaces MtdA, or methylene-H4F is generated using a different enzymatic system, or a non-enzymatic spontaneous condensation takes place. Concerning the latter two scenarios, no enzyme analogous to formaldehyde activating enzyme (Fae, Vorholt et al., 2000) or glutathione-dependent formaldehyde activating enzyme (Gfa, Goenrich et al., 2002) that would facilitate condensation between free formaldehyde and H4F have yet been discovered. At the same time, it would be hard to imagine that spontaneous condensation is sufficient for delivering C1 units into the serine cycle in some organisms but not others (such as Methylobacterium). Thus, it is more likely that FolD can carry out a function in feeding methylene-H4F into the serine cycle (i.e. it must function in the reductive direction). This hypothesis is supported by an observation that organisms utilizing the serine cycle always posses a gene for FtfL (the reversible enzyme) while organisms not depending on the serine cycle may encode either PurU (the irreversible enzyme) or FtfL or both (Table 1). However, for the lack of experimental evidence, the question remains open. FolD enzymes in methylotrophs are quite divergent, suggesting a complex history (not shown), and it is possible that some FolD enzymes, unlike the FolD of M. extorquens (Studer et al., 2002), can functionally substitute for MtdA. It is still an interesting fact that most organisms chosing MtdA for H4F-mediated C1 transfers do not possess FolD, likely due to a specific selection against the latter. Thus, the kinetic properties of MtdA must have a certain advantage over the properties of FolD in terms of methylotrophy, while both can serve a function in C1 transfer for general metabolic purposes.
XoxF. The exact function of XoxF remains a mystery. Originally discovered in bona fide methylotrophs (Harms et al., 1996; Chistoserdova and Lidstrom, 1997), xoxF-like genes are now found in many genomes, most frequently in representatives of the phyla containing methylotrophs (such as Rhizobiales, Burkholderiales), but also in some deeply branching phyla (Aquificales, Acidobacteria). In many methylotroph genomes, multiple, sometimes divergent copies are encoded. Disruption of xoxF when present in the mxaF or mdh2 backgrounds was demonstrated to have little effect on methanol-oxidizing capacity of cells (Harms et al., 1996; Chistoserdova and Lidstrom, 1997; Kalyuzhnaya et al., 2008b). However, a phenotype could be demonstrated for a xoxF mutant in Rhodobacter sphaeroides, an organism not containing mxaFI or mdh2, indicating that XoxF may be involved in C1 metabolism and possibly in methanol oxidation (Wilson et al., 2008). Moreover, a low MDH activity was demonstrated for XoxF purified from M. extorquens AM1 (Schmidt et al., 2010).
Different expression patterns have been observed for XoxF in different organisms, via transcriptomic and/orproteomic studies. In organisms possessing true MDH enzymes, XoxF proteins are typically not highly expressed, at least in laboratory conditions. For example, while MxaFI are some of the most abundant proteins in both M. extorquens AM1 and M. flagellatus grown on methanol or methylamine, the multiple XoxF (two in the former and four in the latter) are expressed at low levels (Bosch et al., 2008; Hendrickson et al., 2010). Interestingly, in mutants defective in regulation of MDH in M. extorquens AM1, the xoxF gene was highly upregulated (Chistoserdova and Lidstrom, 1997). The XoxF protein was also found as one of the most abundant proteins of M. extorquens in planta (Delmotte et al., 2009).
The expression patterns are different in methylotrophs not possessing true MDH enzymes (and lacking measurable MDH activity), such as Methylotenera mobilis (Kalyuzhnaya et al., 2006b). In the latter, both XoxF proteins are found at high levels of abundance (Bosch et al., 2009; Kalyuzhnaya et al., 2009; E.L. Hendrickson, T. Wang, M.G. Kalyuzhnaya, M. Hackett and L. Chistoserdova, unpublished). In non-methylotrophic Burkholderia xenovorans LB400 expression of the only xoxF gene was found to be induced by starvation conditions, along with other C1 genes (Denef et al., 2005). XoxF is also one of the most abundant proteins expressed by the communities of the nutrient-limited costal ocean waters (Sowell et al., 2010). It is tempting to speculate that XoxF is a low activity MDH (Giovannoni et al., 2008; Wilson et al., 2008; Schmidt et al., 2010). However, further evidence such as null mutants of organisms not possessing bona fide MDH enzymes (MxaFI, Mdh2) is needed.
It is important to point out that the sequences that are named here XoxF and that typically share over 40% identity with MxaF show a lot of divergence among themselves, separating into at least five different phylogenetic groups (Supporting information, Fig. S2). Therefore it is possible that different XoxF proteins possess different substrate specificities and potentially have different functions.
Fae homologues. Fae (formaldehyde activating enzyme) is a key enzyme in the operation of the H4MPT-linked C1 transfer pathway in bacteria, catalysing condensation of free formaldehyde with H4MPT (Vorholt et al., 2000). In organisms that rely on the H4MPT-linked pathway as a major pathway for formaldehyde oxidation, such as M. extorquens, loss of Fae function leads to severe consequences: not only mutants lacking Fae are methylotrophy-negative, but they are also extremely sensitive to methanol/formaldehyde (Vorholt et al., 2000). Interestingly, in many methylotroph genomes, including the ones of M. extorquens, one or more homologues of Fae are encoded (Chistoserdova et al., 2007b; Vuilleumier et al., 2009) and in some cases these are parts of C1 gene clusters (Kane et al., 2007), suggesting a potential involvement in methylotrophy. In cases when expression studies have been conducted, Fae homologues were found to be expressed in methylotrophic growth conditions (Bosch et al., 2009; Hendrickson et al., 2010). So far, no specific function has been proposed for these. In M. extorquens AM1, both Fae homologues, named Fae2 and Fae3, have been mutated, without a visible phenotype (Kalyuzhnaya et al., 2005a). Meantime, additional homologues of Fae and Fae-like enzymes are being detected in prokaryotic genomes (Supporting information, Fig. S3). Interestingly, genes for Fae2 and Fae3 are often found in methylotroph genomes, in addition to the gene for true Fae, and different combinations may be present. From the data available to date, Fae2 appears to be more frequently present in Proteobacteria compared with Fae3, but both may be absent. However, Fae3 is persistently present in all the Planctomycete genomes, in addition to true Fae. In the latter case, these genes appear to be of proteobacterial origin, based on sequence similarity (Supporting information, Fig. S3). Two other divergent types of Fae-like proteins (Fae4 and Fae5) are found, respectively, in Proteobacteria and in Gram-positive bacteria, neither of these possessing any other genes for the H4MPT-linked pathway. It is likely that the divergent homologues of Fae carry out functions different from the one of Fae, and these remain to be determined.
New insights into the evolution of methylotrophy
The H4MPT-linked pathway
It has been previously speculated that of all known methylotrophy metabolic modules, the H4MPT-linked formaldehyde oxidation module must be very ancient, based on gene/peptide divergence and based on its occurrence in Archaea (Chistoserdova et al., 2004a; 2005a). This is also one of the most persistently present modules in methylotrophs, and only a few functional groups are lacking it, such as the minimalist methylotroph Methylophilales strain HTCC2181, the autotrophic P. denitrificans and the methanotrophic Verrucomicrobia. However, until recently, alternative scenarios for the evolution of this pathway remained unresolved (Bauer et al., 2004; Chistoserdova et al., 2004a), due to the limited distribution of this pathway. Based on the analysis of the recently sequenced genomes, the pathway appears to be much more widespread in both bacterial and archaeal phyla than previously thought. Complete or partial sets of genes have now been identified, besides methanogenic and sulfate-reducing archaea, Proteobacteria and Planctomycetes, in the genomes representing Crenarchaeota (such as Ignisphaera), the yet unclassified Phyla NC10 and LW (the latter representing uncultivated bacteria of an unknown lifestyle), phylum Synergistetes and Gram-positive bacteria (Supporting information, Figs S1 and S3). This broad distribution of the pathway and further expansion of the phylogenetic diversity of the respective genes/enzymes clearly point to the likelihood of the presence of this pathway in the last common universal ancestor. Remarkably, some of the new members of the bacterial domain possessing this pathway are obligate anaerobes (Candidatus M. oxyfera, A. hydrogeniformans), corroborating the hypothesis of the early emergence of the pathway (Chistoserdova et al., 2004a), possibly prior to the emergence of oxygenic photosynthesis. Likely, this pathway, in its formaldehyde-oxidizing capacity, has evolved before any of the primary oxidation modules, such as pMMO, MDH or MADH. Indeed, genes responsible for the latter tend to be more conserved even among deeply diverging phyla.
While the existence of the deeply diverging genes in major microbial lineages must reflect the long history of this pathway, the recent evolution must have involved multiple lateral transfers. For example, Planctomycetes and some Proteobacteria share highly similar fae-like genes (encoding Fae3; Chistoserdova et al., 2004a; Supporting information, Fig. S3). At least in Betaproteobacteria, genes encoding the H4MPT-linked pathway appear to be of polyphyletic origin, with the sequences of Burkholderiaceae separating from the sequences of Methylophilaceae (Kalyuzhnaya et al., 2005b; Chistoserdova et al., 2007b). Overall, the pathway appears to be a currency easily gained and easily lost. The pathway loss (along with many other non-essential genes) appears to be the case with the minimalist genome of strain HTCC2181 (Giovannoni et al., 2008). Experiments with laboratory-evolved strains of M. extorquens AM1 demonstrate that such losses may happen very quickly, as a matter of 1500 generations without selection for the methylotrophy ability (Lee et al., 2009). Indeed, in the case of the Nitrosococcus species, one can observe results of the very recent deletion events: while the gene synteny and high gene identity are maintained between the gene clusters found in N. halophilus and Nitrosococcus oceani, in the latter, key genes are missing from the cluster, thus suggesting that the pathway is no longer operational in this species. In the Burkholderiales, both types of events may be observed: of the 68 sequenced genomes belonging to Burkholderia, only eight encode the pathway, suggesting frequent losses. However, one of the genomes, of Burkholderia phymatum STM815 contains two non-homologous clusters, both on a megaplasmid, along with other methylotrophy genes, suggesting a mechanism for lateral transfer and for the acquisition of methylotrophy capability (Chistoserdova et al., 2009).
From the sequence data now available, XoxF also appears quite ancient, even though likely it emerged in Bacteria after they separated from Archaea. As (divergent) homologues of XoxF are encoded in the genomes of anaerobic bacteria, it is possible that early in its evolution XoxF primarily acted anaerobically. If XoxF is indeed a primordial, low activity MDH, then MxaF(I) would be an evolved, high efficiency type allowing for faster growth on methane and methanol and providing competitive advantage to methylotrophs possessing this enzyme, at least in some environments. Phylogenetic comparisons of the available MxaF peptides shows that they are rather closely related, including the sequence of the MxaF from Phylum NC10 (Supporting information, Fig. S2), therefore its emergence must be relatively recent in the bacterial evolution. Mdh2 must have evolved in Betaproteobacteria, before the separation of Burkholderiales and Rhodocyclales, via convergent evolution (Kalyuzhnaya et al., 2008b).
So far, genes encoding the MtdC enzymes are only found in Planctomycetes and Phylum LW. The former are considered some of the earliest bacteria on this planet (Brochier and Philippe, 2002; Di Giulio, 2003), while the latter so far lack a position on the tree of life as no data on 16S rRNA for this phylum exist. It has been previously proposed that MtdC enzymes must function in formaldehyde detoxification rather than energy generation or assimilatory pathways (Chistoserdova et al., 2004a; Vorholt et al., 2005). While possessing different substrate specificities, the MtdA enzymes are evolutionarily more closely related to MtdC than to MtdB, thus they must pre-date the MtdB enzymes. In current life, the distribution of MtdA seems to be limited to a small number of specific functional groups, likely due to losses from other bacterial groups, possibly as a result of MtdA specializing in the assimilatory function, in combination with Fch and the serine cycle. MtdA enzymes found in Alpha, Beta- and Gammaproteobacteria are closely related, suggesting a possibility of more recent distribution via lateral transfers. Phylogenetically distinct Mtd enzymes are encoded in the genomes of M. thiooxidans, N. halophilus, and A. hydrogeniformans, and these, while forming two distinct clusters, are more related to MtdA and MtdC enzymes than to MtdB enzymes (Supporting information, Fig. S1). The specificities and the functions of these enzymes remain unknown. The MtdB enzymes, while appearing the youngest evolutionary branch of the group, appear to pre-date the separation of Phylum NC10 and Proteobacteria. So far, MtdA enzymes are only found in bona fide methylotrophs, MtdB and MtdD enzymes are found in both true methylotrophs and organisms with no demonstrated methylotrophy capability, while MtdC and MtdE enzymes are found in species so far lacking connection to methylotrophy.
RuBisCo enzymes are considered some of the most ancient enzymes (Ashida et al., 2003), and it is likely that the CBB cycle served as the assimilatory pathway in early methylotrophs. In this sense the streamlined metabolism of verrucomicrobial methanotrophs may be reminiscent of the metabolism of primordial methanotrophs. The enabling enzymes of the RuMP cycle (HPS/HPI) are also widespread in the microbial world, including Archaea, and in non-methylotrophs they play a role in formaldehyde detoxification (Kato et al., 2006). This may have been the original function of the RuMP cycle before it was adopted by some methylotrophs for the assimilatory needs. The serine cycle appears to have evolved more recently as a methylotrophy module, and it appears to have evolved multiple times, likely by modifying enzyme specificities and evolving mechanisms of specific regulation.
Conclusions and future perspectives
The recent breakthroughs enabling analysis of genomes, transcriptomes and proteomes of a variety of methylotrophs, including both laboratory models and environmentally relevant organisms, as well as the modifications of these approaches known as meta-omics that assess genomic information for organisms not available in pure cultures, open new horizons in understanding methylotrophy as a physiological phenomenon. The recent data highlight the previously unsuspected diversity and flexibility in the use of a variety of known methylotrophy enzymes and pathways. These new data further reinforce the concept of methylotrophy being modular in its nature, i.e. requiring the presence of discrete metabolic pathways or enzymes for discrete metabolic goals, with the modules being in some cases interchangeable and/or replaceable and in some cases redundant. Clearly, the current knowledge is far from being complete, and more methylotrophs employing novel modes of metabolism and potentially completely novel modules (novel enzymes or entire novel pathways) may be discovered in the future. On the other hand, the definition of methylotrophy may need to be revisited as many organisms encoding (and expressing) methylotrophy pathways may use them for co-metabolism of C1 compounds, rather than for exclusively C1-based metabolism. Overall, the occurrence of methylotrophy modules in a variety of environmentally important and sometimes environmentally dominant organisms point to the importance of this type of metabolism, beyond the well-understood role of aerobic methanotrophs, in the global carbon, nitrogen and sulfur cycles.
The author acknowledges support from the National Science Foundation (Grants MCB-0604269 and MCB-0950183).