Introduction

Methylobacterium (Methylobacteriaceae, Hyphomicrobiales, Alphaproteobacteria; type species: Methylobacterium organophilum [1]) is a genus of Gram-negative, rod-shaped bacteria able to metabolize single-carbon compounds like methanol. Because these bacteria are often pink or reddish due to carotenoid pigments, they are referred to as pink-pigmented facultative methylotrophs. Methylobacterium species are primarily free-living and commonly found in various environments, including soil, water, and particularly in association with plants. Some Methylobacterium, like Methylobacterium nodulans, can form root nodules, promoting nitrogen fixation and uptake by plants [2], whereas others across the genus are found on the leaf surface (phyllosphere), establishing beneficial interactions with their host [3]. In the phyllosphere, Methylobacterium uses methanol as a primary carbon source, which is produced by plants as a byproduct of pectin metabolism and emitted through stomata [4]. They benefit the host by preventing pathogens, promoting growth through the production of phytohormones, and aiding in nutrient uptake [5][6][7]. The ability of Methylobacterium to utilize methanol and its benefits to plants makes it a model for biotechnology, already used commercially for protein production, bioremediation, and as a biostimulant in agriculture [3,8,9].

In 2018, Green and Ardley [10] proposed a significant taxonomic revision within the genus Methylobacterium, based on 16S rRNA gene sequences, multi-locus sequence analysis, variation in methylotrophic capabilities, and reddish pigmentation. The authors suggested that 11 species of Methylobacterium form a genetically and phenotypically homogeneous clade, and classified them into a new genus called Methylorubrum. Notable reclassified species included Methylobacterium extorquens, the best-studied methylotroph and the species in the genus most used in biotechnology, which was subsequently proposed as the type species of Methylorubrum.

Several recent phylogenomic-based studies have raised doubts about the genetic distinctions used to separate Methylorubrum from Methylobacterium. Hördt et al. found that although Methylorubrum is monophyletic, it causes Methylobacterium to be paraphyletic, violating key taxonomic guidelines [11]. Alessa et al. also argued against classifying Methylorubrum as a separate genus [12] and more recently, we supported this perspective, concluding that the genetic differences between the two groups are minor and should be viewed as variations within a single genus rather than justifying a taxonomic split [13]. Despite those controversies, both Methylobacterium and Methylorubrum are officially adopted as genera in major taxonomic databases, such as the List of Figure 1. Comparison of recent Methylobacterium classifications showing that the 16S rRNA gene is a reliable taxonomic marker to distinguish most species but is inadequate to infer evolutionary relationships among them. (A) Clades based on the 16S rRNA gene tree [10]. Phylogeny was performed de novo from the complete nucleotide sequence of the 16S rRNA gene (neighbor-joining tree; 1000 replicates; sequences extracted from genomes previously used [13], but for Psychroglaciecola). Scale represents substitutions per site. (B) Clades based on DNA-DNA hybridization from complete genomes of 59 species (DDH clades; [12]). (C) Groups based on the consensus phylogeny from core genome (right; 384 core gene phylogenies combined with ASTRAL-III, simplified from previously [13]; only topology is showed; see Fig. 2A) and refined in well-supported lineages (this perspective). Arrows indicate type species M. organophilum and Methylorubrum extorquens. Clades, groups, lineages, and nodes in the consensus tree that are also monophyletic in the 16S rRNA gene tree are indicated by asterisks. Trees were arranged to minimize crossing branches (dotted lines in the consensus tree). In the 16S rRNA gene tree, species that could not be distinguished were collapsed together. Lineage C2 is missing (no complete 16S rRNA gene sequence available for M. Crusticola).

Prokaryotic names with Standing in Nomenclature, the National Center for Biotechnology Information, and the most recent release of SILVA (v. 138) database for 16S rRNA gene microbial sequences [14], creating great inconsistencies among microbial studies.

Here, we examine recent phylogenomic and phenotypic studies that challenged the classification of Methylorubrum as a distinct genus and advocate for its abandonment and the reinstatement of the original and widely accepted Methylobacterium taxonomy (Patt et al. 1976). We argue that Methylobacterium sensu lato (Patt et al. 1976) represents a cohesive, monophyletic genus. We acknowledge that the significant phenotypic and genetic diversity within Methylobacterium sensu lato is however larger than expected for a genus and indicates complex evolutionary and adaptive processes, warranting further genomic, phenotypic, and ecological research on this group.

Phylogenomic evidence against Methylorubrum

The original motivation for the Methylorubrum description by Green and Ardley [10] was to solve the large genetic diversity observed at the 16S rRNA gene within Methylobacterium-larger than what was expected for a single genus-as well as the fact that M. organophilum, originally chosen as the type species of the genus by Patt et al. [1], was much less studied than its famous counterpart, M. extorquens. Based on a phylogeny of the complete 16S rRNA gene sequence, the authors noted that Methylobacterium consisted of three monophyletic clades: A, B, and C. For the purpose of this perspective, we performed de novo a phylogenetic analysis using complete 16S rRNA gene nucleotide sequences found across 213 Methylobacteriaceae genomes from our previous study [13] and recovered the three clades described by Green and Ardley (Fig. 1A). Clade A included the type species M. organophilum (A) Consensus phylogeny from core genome (right; 384 core gene phylogenies combined with ASTRAL-III) collapsed for well-supported lineages and rooted on microvirga and Enterovirga. Scale represents coalescent units. (B) Lineage names, number of genomes and species, and type species per lineage. (C) Genome size and GC content (mean and standard deviation per lineage). (D) Pairwise dissimilarity in gene content (above diagonal) and pairwise similarity in genome organization (below diagonal). (E) Average gene occurrence per lineage for four methylotrophic metabolic pathways: NMG (8 genes encoding the N-methylglutamate pathway for methylamine utilization); MADH: (11 genes encoding for methylamine dehydrogenase); CaMDH: (14 genes encoding for the calcium-dependent methanol dehydrogenase) and LaMDH (6 genes encoding for the lanthanide-dependent methanol dehydrogenase). and, accordingly, was retained in Methylobacterium by Green and Ardley (Methylobacterium sensu stricto). Clade B was renamed as Methylorubrum and included the model species M. extorquens, which was naturally chosen by Green and Ardley as the type species. Finally, the basal clade C was retained in Methylobacterium, pending further taxonomic revisions.

At this point, it is worth noting several issues raised by phylogenetic analyses used by the authors to support Methylobacterium (Patt et al., 1976) emendation in Methylobacterium (clades A + C) and Methylorubrum (clade B). First, it is surprising that, given its clear genetic distinction and basal position, Green and Ardley did not propose an emendation for group C, as they did for group B, probably because it did not include enough genetic data or a model species like M. extorquens, or represented too much phenotypic complexity in comparison to its apparent genomic homogeneity. Alternative gene phylogenies explored by Green and Ardley (housekeeping, ribosomal, methylotrophy, and serine cycle genes), although confirming that clade B (Methylorubrum) was consistently monophyletic, also highlighted C as the most basal clade, making A + C paraphyletic. Second, Clade A, in addition to being poorly supported in the 16S rRNA gene phylogeny (<70%), was also poorly supported-or even paraphyletic-in alternative gene phylogenies presented by Green and Ardley. Finally, in the two alternative gene phylogenies in which M. organophilum was included by the authors (gyrB and mxaF), the Methylobacterium type species did not branch with clade A but formed a well-supported monophyletic group with clade B (Methylorubrum). Despite these profound discrepancies between the 16S rRNA gene phylogeny and alternative gene phylogenies, these were not addressed by the authors. Eventually, Methylorubrum was progressively adopted as a valid genus name in taxonomic databases.

Two studies have since questioned whether the genetic differences used to justify the separation of Methylorubrum from Methylobacterium are significant enough to warrant a distinct genus. By analyzing the genome of 62 type strains from Methylobacterium and Methylorubrum, Alessa et al. redefined clades on the basis of whole-genome-based DNA-DNA Hybridization (DDH) [12]. Authors showed that Methylorubrum was likely nested within Green and Ardley's clade A, and that M. organophilum, the type species of Methylobacterium, and other relatives (e.g. M. brachythecii; clade B2 in Alessa et al. study; see Fig. 1B), shared more genomic similarity with Methylorubrum than with other Methylobacterium species from clade A. Accordingly, authors amended Methylobacterium back to the Patt et al. (1976) description, but the emendation was not fully adopted by the community [12]. Similarly, we recently reconstructed the consensus evolutionary tree of Methylobacteriaceae using 384 core genes and demonstrated that Methylobacterium-Methylorubrum consisted of four wellsupported monophyletic groups (A, B, D, C; Figs 1C and 2A) [13], broadly consistent with DDH-based clades (Fig. 1B) [12].

We summarized the characteristics of 213 Methylobacteriaceae genomes we previously examined: genome size, GC content, gene content, and core genome architecture (synteny) (Fig. 2A-D) [13]. For this study, we also calculated Average Nucleotide Identity (ANI) among those genomes (Fig. S1). Although a 74% threshold is traditionally used to delineate bacteria genera with this statistic [15], Methylobacterium sensu lato formed a cohesive group for ANI = 80% and could not be distinguished from its sister genera for ANI = 79%, stressing the previously noted need to modify the criterion for genus delineation in Hyphomicrobiales [16]. All of the genome characteristics examined show both that (1) Group C can be clearly distinguished from the rest of Methylobacterium and that (2) Methylorubrum (lineages B2 and B1) does not represent a natural group (Fig. 2A-D; S1). Group C corresponded to Green and Ardley's clade C (Methylobacterium aquaticum, M. nodulans, and relatives); its basal position remained unchanged compared to the 16S rRNA gene phylogeny, suggesting that it could be described a fortiori as a distinct genus, provided that we consider a highly conservative ANI threshold (82%; Fig. S1). Group B contained Methylorubrum, as well as some species from Green and Ardley's clade A, including the Methylobacterium type species M. organophilum. The co-occurrence of two type species, M. organophilum and Methylorubrum extorquens, in this same group is in violation of taxonomic rules and invalidates de facto the S1 for references). (B) Summary of environmental source for each group. Left: Environmental sources of isolated bacteria. For each group, the information of environmental source of isolation comes from isolates from which genomes were previously analyzed [13]. Proportions were weighted by the number of isolates per species to avoid overrepresentation. Right: Environmental sources of metagenomes [19]. previous emendation of Methylobacterium by Green and Ardley. The rest of the species from Green and Ardley's clade A could be divided in two distinct groups: A (Methylobacterium brachiatum and relatives) and D (Methylobacterium gossipiicola and relatives).

Although the basal group C is distinct by all methods, the genetic boundaries of the later-diverging groups A, B, and D are more difficult to delineate. For instance, deep-branching species from group A, like Methylobacterium jeotgali, Methylobacterium soli, and relatives, share more ancestry in gene content with groups B and D than with other species from group A (Fig. 2D). Similarly, M. soli, M. jeotgali, M. trifolii, Methylobacterium cerastii, and relatives share more synteny with some species from group D than with other species from group A (Fig. 2D). Because of these deep-branching species, no clear threshold in ANI could be defined to distinguish groups A, B, and D (Fig. S1). Many of these deep-branching species also showed discrepancies between different individual gene phylogenies explored by Green and Ardley. Together with the inconsistencies of these species between the consensus phylogeny, synteny, gene content and ANI, these observations suggest that groups A, B, and D may have experienced too many gene exchanges since their divergence to be considered as distinct genera [13].

The unfortunate renaming of Methylobacterium based on the phylogenetics of the 16S rRNA gene should serve as a wider warning for others in microbial genetics. The 16S rRNA gene is widely used as a marker in prokaryote taxonomy but does not always accurately ref lect evolution below the genus level [17] for several reasons: 1) insufficient polymorphism for robust phylogenetic reconstructions; 2) large variation in 16S rRNA gene copy number among groups A (4-6), B (4-5), D (3)(4), and C (6-13), sometimes resulting in within-species sequence variation above the inter-species variation; 3) the 16S rRNA gene has its own evolutionary history, which does not ref lect the consensus phylogenetic tree accounting for ancestry. This applies to any gene potentially subject to selection or horizontal gene transfers, and with the availability of genomics, one should prioritize inferring taxonomy upon consensus phylogenetic reconstructions from several dozen core genes or from whole-genome information, rather than from a single marker gene.

Phenotypic evidence against Methylorubrum

To further support the split into Methylorubrum, Green and Ardley proposed the use of phenotypic data such as the utilization of methylotrophic substrates, growth on multi-carbon compounds, and pigmentation differences. The ability to use methanol as a sole carbon and energy source has been historically used to define Methylobacterium sensu lato [18]. This criterion can be retained as all species carry pathways that allow methanol assimilation (Table S1, Fig. 2E), and, so far, there have been no strains unable to use methanol as a sole carbon source, particularly with added lanthanides (Table S1, Fig. 3A). It was suggested that growth on methylamine distinguishes group B (Methylorubrum) from the other groups, but this does not serve as a good distinguishing feature, as methylamine use is found throughout groups A, C, and D (Table S1, Figs 2E and 3A). The other substrates used by Green and Ardley to distinguish subgroups of Methylobacterium were betaine and citrate. Betaine is an N-trimethylated glycine that is used by plants as an osmolyte which helps protect against abiotic stress and thus would be an important carbon source for plant-associated bacteria [20]. Citrate is a well-known metal ion chelator in the soil and can be used by a wide variety of bacteria as a carbon source [21]. Although there may be some trends in the available data, such as the lack of citrate utilization in group B strains that have been tested, otherwise both substrates are used by some members of each group throughout the genus (Fig. 3B).

Distribution of Methylobacterium in the environment

Whereas the phenotyping of strains on selected substrates is liable to be arbitrary in terms of which are chosen and have an uncertain connection to their ecology, it is perhaps preferable to use the occurrence in the environment in distinguishing them. By looking at both the isolation source and metagenome location, we can determine a clear differentiation between group C and groups A, B, and D [19]. The main environmental sources of species from group C were soil and rhizosphere, similar to the outgroup Microvirga (Fig. 3B). In contrast, the later-diverging groups (A, B, and D) are mostly associated with the phyllosphere. These data suggest that, although all groups can be found across varied environments, there is no evidence that the strains that had been assigned to Methylorubrum have a distinct ecology from other Methylobacterium.

Conclusion and future directions

Phylogenomic, phenotypic, and ecological data all provide evidence supporting removing the name Methylorubrum [10] and reinstating Methylobacterium sensu lato [1]. Accordingly, we call for the definitive adoption of Methylobacterium emendation as previously proposed [12]. Although the genus sensu lato still contains a large amount of diversity, the naming of Methylorubrum did not address this problem. We believe the clearest way to address the large diversity with Methylobacterium would be to rename the C group as it follows the consistent pattern of being phylogenetically distinct, phenotypically different, and found in unique locations. Work on this renaming is outside the scope of this perspective and will be handled in proper discourse.

These challenges experienced with Methylobacterium highlight the broader issues in bacterial taxonomy, where the resolution of single gene data, especially from 16S rRNA gene sequences, might not provide a clear-cut distinction on the genus level. We hope that this perspective can show how renaming a genus based on single gene phylogeny in the absence of genomic information can fail to capture true relationships among organisms. With the onset of modern genomic techniques, the standard should be to combine information from core genome phylogeny, genome architecture, gene content, and ANI to distinguish genera and to prioritize metagenomic occurrence over arbitrary phenotypes to evaluate whether there is a realized difference in phenotype between taxa.