Identification of a conserved Orf downstream of pmoB. Directly downstream of pmoB on the USCα fragment an ORF (ORF45) was identified, which exhibited high similarity to genes found in the downstream region of amoB in autotrophic nitrifiers. Homologous genes had been reported for N. europaea, Nitrosospira sp. NpAV and Nitrosococcus oceani ("orf4"), but also for M. trichosporium Ob3b ("orfD"). Besides M. trichosporium OB3b, no other type II MB had been shown to possess orf4/orfD. Also, no gene with significant similarities to ORF45 could be identified in the 101-kb fragment of M. acidiphila.
However, Blast analysis against the complete genome of M. capsulatus (Bath) revealed a copy of this gene (MCA2130). This gene is located separately from the two pmoCAB operon copies of M. capsulatus (Bath). Based on both similarity values and phylogenetic analyses, the amino acid sequences derived from ORF45 (USCα) and ORF NE2060 (N.
europaea) are most similar (57% similarity, 40% identity). Interestingly, ORF NE2060 is not located directly downstream of amoB but instead separated from amoB by the homologous gene NE2061 ("orf4"). Both upstream and downstream of ORF45, factor-independent terminators could be identified. In addition, a Shine-Dalgarno sequence was detected upstream of ORF45. However, no putative promoter region could be identified.
acidic upland soils, previous studies had reported relative low abundances of USCα in regard to whole bacterial populations. Therefore this study is one of the first successful applications of the metagenomic strategy for the concerted retrieval of large genomic fragments of a low-abundant group of assigned uncultured members. Unfortunately, the two genomic fragments identified in the course of this survey belong to a single or two highly related genotypes, as they exhibited 100 % sequence similarity on the overlapping region. Because screening was performed with two different PCR assays, that both had been previously shown to be universal for pmoA of USCα (Holmes et al., 1996, Kolb et al., 2003), it is unlikely that identification of two identical pmoA-sequences was due to a screening artifact.
Based on phylogenetic trees constructed for partial pmoA sequences, M. acidiphila can be assumed to be the closest cultured relative of USCα. Consequently, the pmoA operon and flanking genomic regions of M. acidiphila were analysed for comparison.
Assuming a genome size of 4 Mb, the content of the M. acidiphila BAC library corresponds to roughly 20 genome equivalents of M. acidiphila. Nevertheless, we were able to identify only 8 clones carrying the pmoA gene.
Analysis of the phylogenetic origin of USCα Apart from the pmo gene cluster, which is found only in MOB, 26of 28 conserved ORFs exhibited highest similarities to members of the Alphaproteobacteria, mainly to B. japonicum and R. palustris. Moreover, gene clusters of at least two genes organised in the same order as in the analysed USCα genomic fragment were found identified in genomes of Bradhyrhizobiaceae. No gene cluster was found to be conserved in any other genome if it was not identified also in the B. japonicum genome. Moreover, no other completely sequenced genome contained regions homologous to all conserved USCα gene clusters. Taken together, the high degree of structural conservation as well as the high similarity (identity) values clearly reflect a strong accordance of the analysed USCα fragment with genomes of members of the alphaproteobacterial Bradyrhizobiaceae, in particular the B. japonicum genome. It is unlikely that this is due to random gene shuffling. Because the homologous genes of B.
japonicum as well as of R. palustris are distributed over the whole genomes, it is also unlikely that the genes were obtained as cluster by horizontal gene transfer, i.e. that the analysed fragment of USCα is not characteristic for the complete genome.
ORF16 is the only predicted gene for which no high similarity values to an alphaproteobacterial gene could be determined. It exhibited significant sequence
similarities only to a gene of N. europaea. ORF16 is localised on the only genomic fragment predicted by the SOM-based analysis to belong to the Betaproteobacteria.
Together these results strongly indicate that ORF16 was aquired via horizontal gene transfer from a member of the Betaproteobacteria.
In addition to gene- or protein-based similarity studies, we applied a method focusing on the analysis of (taxon specific) genomic signatures, i.e. di- tri- and tetra-nucleotide distribution patterns. This technique has been widely discussed during the last years (Deschavanne et al., 1999; Edwards et al., 2002; Karlin et al., 1994; Karlin et al., 1992;
Karlin et al., 1997; Knight et al., 2001; Oliver et al., 1993; Sandberg et al., 2001, Sandberg et al., 2003; Teeling et al., 2004). The signature analysis was performed with the program XanaMap, developed at the Xanagene Inc (http://www.xanagen.com/). XanaMap bases on the unsupervised neuronal network algorithm self-organizing map (SOM), which was intensively tested and demonstrated to recognize taxon- and even species-specific characteristics (key combinations of oligonucleotide frequencies). It therefore allows the taxon- or species-specific classification of genomic sequences without any regard on coding regions or sequence similarities (Abe et al., 2002; Abe et al., 2003). The taxonomic SOM-based affiliation was attained by comparison of 5 kb genomic subfragments of USCα against a map computed out of 5 kb genomic fragments of the public available complete sequenced genomes. The SOM-based analysis clearly confirmed the gene- and genome-based phylogenetic assignment of USCα to the Alphaproteobacteria. Interestingly, the analysed fragments of M. acidiphila genome as well as of type II MOB were placed in similar regions of the "alphaproteobacterial map area", indicating at least that this sequences are closely related within the alphaproteobacteria
The results of both comparative sequence analysis of single genes or gene clusters and SOM-based analysis provide final evidence that USCα has an alphaproteobacterial origin. The two methodologically independent approaches group USCα with members of the Bradyrhizobiaceae, in particular with B. japonicum. However, this is also true for M.
acidiphila, despite the fact that, based on 16S rDNA data as well as NifHD phylogenies (Dedysh et al., 2002), M. acidiphila clearly bid more closely related to Beijerinckia spp.
than to the Bradyrhizobiaceae. Therefore, the implications of the genera-related placement of USCα as well as M. acidiphila should not be overestimated. Due to the lack of available genomic information from other Alphaproteobacteria necessary for a more detailed phylogenetic placement of USCα (in particular of the genus Beijerinckia, as well as
sufficient data of type II MB) our results do not provide evidence that USCα is less related to Beijerinckia or type II MB than to the Bradyrhizobiaceae.
To gain a deeper insight into the relationship between USCα, M. acidiphila and representatives of the Bradyrhizobiaceae, we analysed the phylogenies of five genes identified on the genomic fragmentsof both USCα and M. acidiphila, as well as in the genomes of various other bacterial strains. The resulting phylogenies will not be discussed in detail, as the derived trees are not always in complete agreement with 16S rDNA-based phylogenies. This may be due to ancient events of horizontal gene transfer, but it is also possible that the degree of conservation of the respective amino acid sequences is not high enough to produce a meaningful phylogenetic relationship for highly divergent lineages.
However, these phylogenies clearly reflect the common history of the organisms under study, including USCα, M. acidiphila, and members of the Bradyrhizobiaceae and therefore are in deep agreement with the genomic analyses described above. They also show consistently the closer relationship of the two analysed MB on one hand and on the other hand of the two Bradyrhizobiacea. Again, due to the lack of sequence information neither type II MB nor members of the genus Beijerinckia could be included in the analysis.
The inclusion of this organisms would have been of specific interest because the direct comparison of type II MB and of USCα still is only possible based on pmoCAB / PmoCAB data. Nonetheless, we have been able to show that the close relationship of M. acidiphila and USCα is supported not only by pMMO but also by the phylogenies of various other genes.
Comparative Analysis of pmoCAB gene clusters. MB belonging to the so fare uncultured group USCα are of specific interest not only because they are highly abundant in various acidic upland soils but also because it has been assumed for a long time, that they might be able to oxidate methane at atmospheric concentrations. Consequently, a major aim of this study was a comparative analysis of the genes encoding pMMO. The presumed capability of USCα to oxidize methane at atmospheric concentrations might be reflected at the molecular/enzymatic level. It should therefore be investigated whether the derived polypeptide sequences possess any unusual properties or characteristics. The pMMO sequences of USCα and M. acidiphila were compared with those of five pmo operons of type II and type X MOB and with three homologous amo operons of autotrophic nitrifiers.
The pmo genes of both USCα and M. acidiphila are organised in a single gene
highly conserved among all MOB analysed so far. However, we detected on the USCα genomic fragment directly downstream of pmoB an ORF (ORF45) for which no homologous gene could be identified on the genomic fragment of M. acidiphila. In contrast, a homologon was present in the corresponding genomic regions of M. trichosporium OB3b and the betaproteobacterial nitrifiers ("OrfD"). Interestingly, despite the fact that pmoCAB of USCα is based on both sequence similarity values and phylogenies clearly more closely related to other pmo operons than to amo operons, the homologous gene most similar to ORF45 was identified in N. europaea (NE2060). The similarity value between the amino acid sequences derived from ORF45 and NE2060 is in the same range than similarity values between USCα PmoCAB and N. europaea AmoCAB. Both pMMO and AMO are key enzymes of the respective metabolic pathways and therefore subject to strong purifying evolutionary selection. The conserved gene organisation clearly suggests that ORF45 and NE2060 are separated for the same evolutionary period than the adjacent monoxygenase operons are. It therefore can be concluded that the purifying selection pressures acting on either ORF45/ORFD or pMMO/AMO are similar strong.
As reported previously (Norton et al., 2002) orf4 seems not to be part of the amoCAB operon as it is separated from amoB by a terminator region. Similarly, ORF45 seems not to be part of the pmoCAB operon in USCα because a factor-independent terminator is located between pmoB and ORF45. However, the high degree of conservation between ORF45 and NE2060 suggests functional importance. Indeed, a Shine-Dalgarno sequence could be identified upstream of ORF45 and a terminator is located downstream of it, whereas no putative promoter could be identified in the 100-bp region between pmoB and ORF45. If ORF45 is expressed, as indicated by the purifying selection pressure acting on it, there are only two possibilities for transcription: a) ORF45 is a facultative part of pmoCAB and regulated e.g. via antitermination or b) its promoter is localised in the gene pmoB.
The secondary structures predicted for the derived pMMO`s of both USCα and M.
acidiphila are highly similar. They agree well with the secondary structures reported previously for pMMO of other MOB, including gammaproteobacterial M. capsulatus Bath as well as alphaproteobacterial M. trichosporium OB3b and Methylocystis sp. strain M (Gilbert et al., 2000; Ricke et al., 2004; Stolyar et al., 1999). In summary, our findings provide further evidence that both the structural organization of pmoCAB and the secondary structure of derived proteins are highly conserved among all MOB, regardless of their phylogenetic positions.
Phylogenies of complete pMMO sequences. With the sequence information obtained in this study for M. acidiphila and USCα, and the recently analysed pmoCAB1 and pmoCAB2 of Methylocystis strain SC2, the data set of fully sequenced pMMO /AMO operons became large enough for a phylogenetic analysis of concatenated gene sequences. Compared to treeing analyses based only on partial pmoA, our approach sixtouples the underlying phylogenetic information. Nevertheless, regardless the evolutionary models and algorithms used, the phylogenies always corresponded well to pmoA-based trees with the only exception that the USCα /M. acidiphila group branched separated while the recently reported pMMO2 of Methylocystis strain SC2 fromed a common branch with PmoCAB1 of type II MOB. Because pMMO2 had been shown to exhibit unusual properties, this finding again demonstrates the unique position of USCα / M. acidiphila among the known MOB.
The possibility that the separate branching of USCα / M. acidiphila is only due to the small data set could be excluded by testing a small PmoA data set consisting only of the PmoA sequences used for concatenated PmoCAB analysis. In addition, trees based only on either PmoB or PmoC were computed. It could be shown that the separate branching of USCα / M. acidiphila PmoCAB is mainly due to the inclusion of PmoB (Fig.4).