Reviewer's report 1
John A. Fuerst, School of Molecular and Microbial Sciences, The University of Queensland, Brisbane, Queensland 4072, Australia (nominated by Mark Ragan, The University of Queensland, Brisbane, Australia)
This study reports and analyses the complete genome sequence of an extremely acidophilic methanotrophic member of the distinctive Bacterial phylum Verrucomicrobia, "Candidatus Methylacidiphilum infernorum" isolate V4, from a New Zealand methane-emitting geothermally heated soil and growing optimally at pH 2.0–2.5 and at 60°C with 25%(v/v) methane as sole source of energy. Related thermoacidophilic organisms have also been isolated recently from other geothermal areas, and these organisms seem to represent a clade of methane-oxidizers in this phylum, a result of significance concerning our understanding of C1-compound metabolism and evolution of C1 transfer enzymes, since the only known cultivated methane utilizers are members of the phylum Proteobacteria and the only phylum other than the proteobacteria in which some C1-transfer enzymes (though not functional C1 metabolism) have been found is the phylum Planctomycetes. Significantly, phylum Planctomycetes has been proposed on the grounds of 16S rRNA, 23S RNA and ribosomal protein sequence analysis to be related to the phylum Verrucomicrobia members in a single 'PVC" superphylum also proposed to contain phyla Chlamydiae, Lentisphaerae and the candidate phyla Poribacteria and OP3, but the support for the group has been relatively weak, with stronger support for a link between Verrucomicrobia and Chlamydiae than for either of these phyla with Planctomycetes.
Possibly the most significant result of this paper is that the PVC superphylum is supported when concatenated sequences of ribosomal proteins and RNA polymerase subunits are analysed, including "Candidatus
Methylacidiphilum infernorum". This establishes firmer grounds for more detailed investigation of the relationships and links between members of the constituent phyla, at least for Planctomycetes
. It also supports the view that more comprehensive taxon sampling may assist the testing of postulated superphyla, and suggests that genomes of more genera and species representative of all sub-divisions of phylum Verrucomicrobia
should be sequenced to make this possible and confirm the present analysis. Of course, the recently explored limitations of some concatenated datasets for phylogenetic analysis of deep prokaryote nodes [73
] need to be kept in mind. Although this paper reports the first complete genome sequence for a representative of the Verrucomicrobia
, this thermoacidophilic species might not be representative of other soil, aquatic and symbiotic verrucomicrobia living in less extreme habitats, and this makes achieving completion of the genome sequencing and analysis of members of other subdivisions an urgent priority, including at the least Verrucomicrobium spinosum
, Prosthecobacter dejongeii, Chthoniobacter flavus
, Opitutus terrae
, Akkermansia muciniphila
, the soil isolate "Ellin514", and the marine verrucomicrobial strain DG1235.
This conclusion concerning a possible superphylum based on analysis of a limited set of informational – i.e. translational and transcriptional – genes is complicated and potentially contradicted by BLAST analysis of the complete set of predicted proteins, which indicated the fraction of proteins most similar to homologs from available genes of the PVC group to be only ~7%, while the largest fraction (~23% of the V4 proteins) had top hits among the phylum Proteobacteria
. Although the authors admit to certain qualifications concerning over-representation of databases by the Proteobacteria
and possible poor correlation of sequence similarity with phylogenetic affinity, they nevertheless favour a perspective or inference where horizontal gene transfer (HGT) dominates the architecture of the genome, going even further to link this to proteobacterial methylotrophy (despite the previously published result [3
] concerning pmo
genes for particulate methane monooxygenase needed for methane oxidation indicating that Methylacidiphilum
genes are completely divergent from those of methylotrophic proteobacteria). The acceptance of such a conclusion may be dependent on the validity of such BLAST analysis, the problems with which in relation to gene transfer detection have been subject to detailed analysis [74
] and such problems may also apply to this case. There appear to be no other criteria than bioinformatic BLAST or COG analysis to strengthen the conclusion of HGT in this case, and there is then a temptation to consistency with the 'global HGT' dogma without sufficient grounds for high probability that HGT is the only explanation for such results. In other words, there may be other explanations for this apparently contradictory relationship to proteobacteria which need to be considered.
Does phylogenetic analysis of individual genes hypothesized to have been transferred indicate a particular group of proteobacteria from which the transfer may have occurred recently, or is this proposed to be an ancient transfer, in which case how is it to be distinguished from transfers in the progenote? Is the polarity of transfer direction unambiguous? Of course even following such analysis, alternative explanations for phylogenetic misplacement of taxa within an alien clade may also then have to be considered.
To strengthen their conclusions, the authors do include a Correspondence Analysis, a type of gene content analysis, which isolates M. infernorum away from either members of phyla Verrucomicrobia and Chlamydiae or any other bacterial clade. They interpret this as supporting a complex history for the gene set of M. infernorum, reflecting contributions of genes from diverse groups, since nearest neighbours in genome content space are members of phylum Proteobacteria, Thermotogae, Aquificae and one Actinobacteria genus. Is there an explanation for this analysis alternative to gene transfer? The inclusion of the deep-branching Thermotogae and Aquificae is interesting, and suggests that one alternative which might be considered is that Verrucomicrobia, or at least this representative of the phylum, might harbour gene contributions which either occurred during the early radiation of the Bacteria or even earlier when phyla of the domain Bacteria were not distinguishable. More detailed phylogenetic analysis of the genes used in the genome content analysis may be needed to test such an alternative.
The authors do note the caveats which have to be applied to at least their BLAST analysis, but various methods for confirming the hypothesis of lateral gene transfer such as GC composition, codon usage, or association with possible mechanisms for transfer [76
] have not yet apparently been applied to the Methylacidiphilum
genome problem, and these might potentially reinforce the gene transfer explanation derived so far from BLAST and Correspondence Analysis, though it is also possible that ancient transfer would not be detected by such methods.
The situation of Methylacidiphilum infernorum
bears some similarity to that noted for Cenarchaeum symbiosum
by Forterre in his review of the analysis of archaeal COGs by Makarova et al. [70
] where a gene content analysis indicated possible acquisition of euryarchaeal genes via LGT, in an organism already postulated to have acquired 'lots of bacterial genes' [70
]. As in the C. symbiosum
case where the possibility remains that it represents an early branching archaeal lineage containing bacterial and archaeal homologs lost in other archaea, the alternative hypothesis should be considered that Methylacidiphilum
and perhaps also other Verrucomicrobia
(and perhaps even also other members of the PVC superphylum) represent members of an early branching lineage containing ancient homologs of genes in other Bacterial phyla which have been subject to wide (and perhaps even unparsimonious!) loss.
Further tests to estimate the relative timing of the proposed HGT [79
] might lead to insights about this possibility. Evidence regarding potential gene loss is presented supporting the interpretation that M. infernorum
may have acquired genes for several proteins belonging to the gene set conserved in archaea and eukaryotes, but again one asks whether this may alternatively be interpreted not as suggesting HGT among bacteria following initial transfer from archaea but rather as an indication of retention of an ancient signal from an organism close to the LUCA or LCA. Perhaps such an ancient signal could even stem from a lineage analogous to the uncharacterized archaeal lineage recently proposed as a root from which eucaryal archaea-like genes may have originated [80
The distribution of C1
transfer enzymes is intriguing, since the tetrahydromethanopterin-dependent enzymes found in some members of phylum Planctomycetes
do not seem to have been detected, and Methylacidiphilum
is clearly capable of methanotrophy unlike any planctomycete so far isolated. The existing controversy [81
] over the origin of the archaea-like C1
-transfer enzymes of planctomycetes relating to potential gene transfer versus ancient divergence suggests however that it may be productive to examine the phylogenetic relationships of the C1
-transfer enzymes of Methylacidiphilum
, especially since they are clearly functional.
The metabolomic pathway analysis of gene content may have limitations for evolutionary insights in this case. The phylogenetic analysis of the pmo
genes in this organism in another publication [3
] suggested a divergent evolutionary history from methylotrophic proteobacteria, one which might even be consistent with a relatively deep branching core identity for this species, and this line of investigation might be usefully pursued with other genes involved in Methylacidiphilum
M. infernorum appears to display gene loss in the cell division system, and what is interpreted as accelerated evolution is claimed to have occurred in the ftsZ gene. Considering the non-functional nature of the ftsZ homolog in the PVC member Lentisphaera, it would be highly relevant for the data and analysis of the data relating to this claim to be described in the paper – what reasons are there for interpretation of presumably divergent sequence as accelerated evolution? Why isn't a deep branching of this and other verrucomicrobial ftsZ relative to homologs in other phyla an equally plausible explanation for its divergent sequence? Have long branch attraction artefacts been definitively demonstrated as a most probable alternative explanation for such a deep branch? What is the explanation of accelerated evolution in the apparently retained ftsZ but not in ftsK and ftsA, presumed components of the same ftsZ-dependent divisome?
The view that previous speculation interpreted as relating cell division in the PVC superphylum to evolution of the eukaryotic nucleus and cytoskeleton has not so far been supported may be a warranted view at this point. However, such conclusions rejecting PVC superphylum relevance to eukaryote evolution are not only based on limited studies but also on studies based on a very limited sampling of taxa and limited types of bioinformatic analysis. That this may be important is indicated by the occurrence of structures reactive with anti-tubulin antibodies in the 'epixenosome' verrucomicrobial symbionts of ciliate protozoa [83
] as well as the demonstrated tubulin homologs in Prosthecobacter
], indicating a potentially wider occurrence of this eukaryote cytoskeletal protein among verrucomicrobia than has been detected and one not as easily explained by lateral gene transfer as by an evolutionary retention of a deep signal. Concerning detection of potentially eukaryote-homologous features within PVC group members, analysis at the level of secondary structure may be needed to reveal unsuspected relationships to eukaryote signature proteins such as nuclear pore complex proteins [85
]. Detection of homologs of PVC proteins among eukaryote proteins may be expected to be difficult, since even within eukaryotes detection of important eukaryote signature proteins such as homologs to nuclear pore complexes may not be trivial, since exceptional heterogeneity occurs between species, and BLAST and even PSI-BLAST approaches may fail due to variation in evolutionary rate alone [86
]. This question does not appear to be as resolved as the authors have suggested, and relevance of any PVC member to eukaryotes is certainly not refuted by the analysis presented. One might in a lighter mood suggest that it is not over until the sterol-synthesizing nucleated Gemmata
This paper is important for suggesting and perhaps stimulating a number of lines of investigation for future genomic analysis of the phylum Verrucomicrobia and the PVC superphylum, but this future analysis should not be constrained by assumptions concerning easy interpretation of the paradoxes posed by Methylacidiphilum and verrucomicrobia, which still seem unsatisfactorily resolved at this point.
We appreciate this detailed, constructive review and completely agree that further phylogenetic analysis, with particular emphasis on detection of HGT, and perhaps, most importantly, sequencing and comparative analysis of additional, diverse representatives of phylum Verrucomicrobia and the PVC superphylum are required before we understand the natural history of this remarkable group of bacteria. Where we tend to be more skeptical than the reviewer, is the possibility of direct relevance of comparative genomics of the PVC superphylum for the origin of the eukaryotic nucleus. The results with anti-tubulin antibodies reported in references 83 and 84 should be treated with utmost caution. Furthermore, we strongly believe that HGT from eukaryotes is the only viable explanation for the presence of tubulins in Prosthecobacteria
]. The difficulty of detecting homologs of nuclear pore complex subunits should not be exaggerated; at least, finding orthologs throughout the eukaryotic diversity was a straightforward (if not, exactly, trivial) task
]. Although one cannot rule out surprises from new genomes, the chances that any representatives of the PVC superphylum actually possess structures that are homologous to the eukaryotic nucleus (and, in particular, the nuclear pore) are exceedingly small. It is another matter that some members of the PVC superphylum can provide extremely interesting and valuable instances of independent, convergent evolution of intracellular compartmentalization.
Reviewer's response in a second review
I welcome the authors' agreement with my comments concerning the need for more analysis of the hypothesized HGT and for further sequencing and comparative analysis of additional representatives of the verrucomicrobia and the PVC superphylum. My only comment on the doubts of even the possibility of direct relevance of comparative genomics of the PVC superphylum for the origin of the eukaryotic nucleus is that absence of evidence does not constitute evidence of absence and that beliefs, however strongly held, are not refutations (e.g. the belief that HGT from eukaryotes is the only viable explanation for Prosthecobacter tubulins). This is perhaps especially so considering that gene annotation and decision regarding HGT appears to be a work in progress where no one annotation effort, especially if automated only, can be assumed complete. Chances that are estimated qualitatively to be exceedingly small concerning surprising structures within PVC superphylum members can nevertheless be finite, and prediction of absence of such structures cannot be made with certainty. Improved analysis as well as more genome data may be needed to solve these problems.
Reviewer's report 3
Radhey S. Gupta, Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Ontario L8N 3Z5, Canada (nominated by Jonathan Eisen, University of California Davis Genome Center, Davis, California, USA)
In this manuscript Hou et al. report the genome sequence for "Candidatus Methylacidiphilum infernorum", which is the first representative from the phylum Verrucomicrobia to be completely sequenced. The organisms from this clade of bacteria are in general very poorly characterized and only few of them have been isolated in pure culture. Hence, the availability of genome sequence for a member of this group should prove very useful for a variety of studies, particularly in clarifying its phylogenetic placement relative to other bacterial phyla. The authors have carried out detailed analyses of M. infernorum genome to reconstruct its central metabolic pathway and have identified many genes/proteins responsible for its ability to utilize methane and adaptation to the acidic environment. Several differences were noted from other methylotrophs, which are mainly alpha- and gamma-proteobacteria. The work on the annotation of various genes involved in different cellular functions and the differences seen in these regards for M. infernorum has been competently carried out and I have no questions or concerns.
Another important aspect of this manuscript relates to the phylogenetic placement of Verrucomicrobia
with respect to other bacterial phyla. Recent studies by a number of authors, primarily based on 16S and 23S rRNA [1
], have indicated that species from three bacterial phyla viz.Planctomycetes, Verrucomicrobia-Lentisphaerae
(as well as Poribacteria
and OP3), group together in phylogenetic trees. This has led to the proposal that these groups or phyla should be recognized as part of a single superphylum (PVC). In this work the authors have constructed phylogenetic trees based on concatenated sequences for 51 ribosomal proteins and also the three subunits for RNAP. The trees based on these sequences also support the grouping of these species in a single clade. Other analyses reported here to determine the closest relatives of M. infernorum
(e.g. top BLAST hits, Correspondence analysis) have provided no clarification in these regards and these results have been interpreted to suggest a complex evolutionary history of the Verrucomicrobia
1. The proposal that the PVC group of species should be recognized as a superphylum is presently entirely based on some phylogenies. As noted by the authors, other phylogenetic studies have not always supported this grouping. In our recent work [18
], phylogenetic analysis was carried out based on concatenated sequences for 11 large and conserved proteins (including alpha, beta and beta' RNAP). Although these analyses strongly supported the grouping of Chlamydiae
, a reliable grouping of the Planctomycetes
with these groups was only observed in the neighbour-joining tree, but not supported by the maximum-likelihood analysis. Thus, based upon the phylogenetic analyses alone, it is difficult to infer confidently whether the PVC groups of species should be recognized as a superphylum or not. Because, the term Superphylum has taxonomic significance, this status should not be accorded readily unless different lines of evidences support the inference. In the present work, the authors have looked for but failed to find any signature protein that was unique to the PVC groups of species. The closest they have come to identifying a signature protein for this group is the Minf_1886 protein that is only found in Planctomycetes
, but not in any of the chlamydiae.
2. The question thus arises whether the placement of PVC groups of species into a single superphylum is supported by any other line of evidence besides some of the phylogenetic trees. Interestingly, contrary to the authors' observation that they did not find any protein that was unique to the PVC groups of species, we have identified a protein CT421.2 that is uniquely present in all available sequences from the Planctomycetes, Verrucomicrobia-Lentisphaerae
phyla. The sequence alignment of this protein is presented in Supplementary Figure 8 [see Additional file 2
]. A large number of positions in this small protein are highly/completely conserved in all of the species. Besides, the PVC group of species, no other BLAST hit for this protein was observed. The identification of this signature protein that is unique for the PVC groups of species provides additional independent evidence for their grouping into a Superphylum. Sequence information for M. infernorum
was not available to us, but we expect that this protein should be present there as well. It is also of interest to note that within this protein, a 2 aa indel is present in various Chlamydiae
species, but not in any of the Planctomycetes
species. This indel is indicated to be an insert in the Chlamydiae
species rather than a deletion in the other groups (see next comment). It is unclear why the authors in their analyses did not identify this protein. However, since these results are of importance for the inferences drawn here (and in earlier studies) concerning the PVC superphylum and the authors should discuss their significance in the main manuscript.
Indeed, this protein (Minf_0061) is encoded in M. infernorum genome and, as predicted by the reviewer, lacks the 2-aa indel, as do other members of Planctomycetes or Verrucomicrobia-Lentisphaerae species. This sequence has been missed by our automated analysis because of an error of the automated gene calling procedure, which resulted in a truncated 37-aa protein (YP_001938720.1) that missed 13 N-terminal amino acid residues and did not produce sufficiently significant BLAST hits to be recognized in our analysis. We greatly appreciate the reviewer's comment thanks to which the sequence of Minf_0061 has been corrected (YP_001938720.2) and included in the alignment.
Since this manuscript concerns with the phylogeny of M. infernorum
, it will also be of interest for the author to indicate whether the sequences from this species contain the two conserved inserts in the LysRS and RpoB proteins that were previously reported to be specific for various Chlamydiae
and Ver. spinosum
]. My updating of the RpoB sequences indicates that the 3 aa insert in this protein is uniquely found in all of the Chlamydiae, Verrucomicrobia
(V. spinosum DSM 4136, Opitutus terrae PB90-1, Opitutaceae bacterium TAV2, Akkermansia muciniphila ATCC BAA-835 and Bacterium Ellin514
) and Lentisphaerae
(Victivallis vadensis ATCC BAA-548
and Lentisphaera araneosa
) species, but not in any other RpoB homologs from different groups of bacteria including the Planctomycetes
(>400 sequences available in the NCBI database), see in the attached file (Supplementary Figure 9 [see Additional file 2
]). The insert in the RpoB protein is thus specific for the Chlamydiae
phyla and it provides strong and direct evidence that species from these groups shared a common ancestor exclusive of the Planctomycetes
. This inference is also strongly supported by various phylogenetic trees. Since this work reports the first genome of a Verrucomicrobia
species, the close relationship that this group exhibits to the Lentisphaerae
species (exclusive of the Planctomycetes
) deserves to be emphasized, apart from the fact that all of these groups are also part of a higher clade (i.e. the PVC clade).
Authors' response Indeed, RpoB of M. infernorum (Minf_0804) has the same 3-aa insert, and indeed, this indel supports the existence of a separate Chlamydia/Verrucomicrobia clade, as do our results shown in Figure and Supplementary Figure 1 [see Additional file 1]. However, the Chlamydia/Verrucomicrobia group appears to be widely accepted in the scientific community genome, which is why we did not feel it was necessary to stress this point.
1. The authors should provide some further details regarding the phylogenetic analyses. They should indicate whether the aligned sequences were edited (if so, how) and how many aligned characters were present in the final two datasets that they have employed. The sequence alignments for these dataset could also be included as supplemental data. It will also be useful if it could be mentioned in the Figure legends that the trees shown are maximum-likelihood trees.
Authors' response The necessary details on the alignment filtering and the size of the dataset have been added to the text. In the legend to Figure , it is now indicated that Maximum Likelihood trees are shown. Sequence alignments in the aligned FASTA text format have been made available on the FTP site
2. Page 7, first line, I am not sure whether Lentisphaerae is now regarded as a distinct phylum. In the Bergey's manual (2005) Victivallaceae is indicated to be a family within the phylum Verrucomicrobia.
Authors' response The suggestion by Cho and colleagues
] that Lentisphaerae forms a distinct phylum has been subsequently validated by Euzéby
]. There seems to be a general agreement that Lentisphaerae should be regarded as a separate phylum