MBE Advance Access originally published online on September 14, 2005
Molecular Biology and Evolution 2006 23(1):168-178; doi:10.1093/molbev/msj019
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Article |
Complex Histories of Genes Encoding 3-Hydroxy-3-methylglutaryl-CoenzymeA Reductase
* Genome Atlantic and Department of Biochemistry and Molecular Biology, Dalhousie University, Halifax, Nova Scotia, Canada; Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge; and Department of Biological Sciences, Macquarie University, Sydney Australia
E-mail: ugophna{at}dal.ca.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
The mevalonate pathway for the synthesis of isoprenoids can be found in organisms from all domains of life. It has been previously demonstrated that the first gene specific to that pathway, which encodes the enzyme 3-hydroxy-3-methylglutaryl-CoenzymeA reductase (HMGR), has been transferred between domains by lateral gene transfer on several occasions. Here we look within the domain Bacteria at lateral acquisition of HMGR, whether as a single gene or as part of a mevalonate pathway cluster. We observe a complex history of multiple transfer events probably reflecting the fact that HMGR could be beneficial in a variety of physiological and genetic contexts. We demonstrate that even in Vibrio species, where HMGR is not clustered with other genes to form an operon or a metabolic cluster, it is under strong purifying selection.
Key Words: Legionella pneumophila • Coxiella burnetii • Bdellovibrio bacteriovorus • Vibrio • horizontal gene transfer • mobilon • molecular evolution
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
Isoprenoid biosynthesis is important in all domains of life and is required for the production of a large variety of key compounds, such as cholesterol in eukaryotic membranes, isoprene alcohols in archaeal membranes, and quinones for electron transport in all three domains. Archaea, eukaryotes, and some bacteria synthesize the basic isoprenoid building block, isopentenyl pyrophosphate (IPP), from acetyl-CoA and aceto-acetyl-CoA using the mevalonate pathway (Boucher and Doolittle 2000
; Boucher, Kamekura, and Doolittle 2004). This pathway generally involves five steps. (1) Synthesis of 3-hydroxy-3-methylglutaryl-CoenzymeA (HMG-CoA) by 3-hydroxy-3-methylglutaryl-CoenzymeA synthase (HMGS). (2) Reduction of HMG-CoA to mevalonate by 3-hydroxy-3-methylglutaryl-CoenzymeA reductase (HMGR). (3). Phosphorylation of mevalonate to phosphomevalonate by mevalonate kinase (MVK). (4) Phosphorylation of phosphomevalonate by phosphomevalonate kinase (PMK). (5) Decarboxylation of diphosphomevalonate to IPP by diphosphomevalonate decarboxylase (PPMD). HMGR, the catalyst of the second (and first committed) step of the mevalonate pathway, is the focus of our analysis.
Although homologous, HMGR enzymes and the genes encoding them can be unequivocally divided into two classes. Class 1 enzymes are common in, and generally characteristic of, archaea and eukaryotes. Bacteria, when they have HMGR, usually have the class 2 enzyme. Exceptions to this generalization provided the first evidence that HMGR genes are subject to lateral transfer. For instance, the HMGR gene of the archaeon Archaeoglobus fulgidus belongs to class 2, while that of Vibrio cholerae, Paracoccus zeaxanthinifaciens, and a few Streptomyces species are of class 1 (Boucher and Doolittle 2000; Boucher et al. 2001; Humbelin et al. 2002). The current gene and genome databases have expanded severalfold since these initial studies. Here we assess just how often and widely this nomadic gene moves from three perspectives: (1) global prokaryotic phylogenetic reconstruction, (2) gene context comparisons, and (3) an experimental survey of Vibrio species.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
Bacterial Strains, DNA Extraction, and Amplification
Vibrio isolates used in this study are detailed in table 1. Templates for amplification were either 5–10 ng of bacterial DNA, purified as described previously (Thompson et al. 2005
) for the coastal isolates, or 1 µl of a single bacterial colony suspended in 0.5 ml deionized water and boiled for 10 min for the Australian isolates.
|
Polymerase chain reactions (PCRs) were carried out in a final volume of 25 µl containing 5–10 ng of template DNA, 10 x PCR buffer, 2.5 mM MgCl2, 0.2 mM deoxynucleotide triphosphates 10 pmol of each primer, and 1 unit of Taq DNA polymerase (Invitrogen).
Amplifications of rpoB gene fragments of Vibrio species were performed using primers rpoBF_2041 (5'-GGTGCGAACATGCAACGTCAG-3') and rpoBR_3201 (5'-ACGACCCGCCATCTTATCACC-3') with initial denaturation at 94°C for 5 min, 30 cycles with a denaturation at 94°C for 30 s, primer annealing at 55°C for 30 s, and DNA polymerization at 72°C for 1 min. PCR products of rpoB amplification were purified using ExoSAP-IT (USB) and sequenced from both ends utilizing the same primers using a LiCor 4000 L automated sequencer.
Amplification of HMGR gene fragments required the use of various primer combinations to obtain products from different isolates. Primers were designed based on a multiple sequence alignment of HMGR gene fragments from Vibrio species which had their genomes sequenced. First, a combination of the reverse primer HMG_994R (5'-GCAAGGGCGTTGGCGTAATG-3') and HMG_535F (5'-CTGGGCAGTTTGTCGGTTGGG-3'), HMG_313F (5'-GCACAGTGAAACTGCCTGTCGG-3'), or HMG_300F (5'-GAGCATTTCATTGGCACAGT-3') forward primer was used for amplification. Reaction conditions were initial denaturation at 94°C for 5 min, 30 cycles with a denaturation at 94°C for 30 s, primer annealing at 48°C for 30 s, and DNA polymerization at 72°C for 45 s. Subsequently, on the basis of an alignment of the sequences obtained using the above primers, a degenerate primer pair was designed and gene fragments were amplified using HMG_degF (5'-GGICARAAYATGGTIACIATIGCIAC-3') and HMG_degR (5'-GTISTCATYTKISYRAAYTKIAYCAT-3'). Reaction conditions were initial denaturation at 94°C for 5 min, 30 cycles with a denaturation at 94°C for 30 s, primer annealing at 48°C for 30 s, and DNA polymerization at 72°C for 30 s. In a few cases, the degenerate primers were used in combination with the nondegenerate primers to obtain a larger product using the combinations HMG_degF with HMG_994R and HMG_300F with HMG_degR, applying the same reaction conditions used above for degenerate primer PCR. HMGR amplicons were gel purified with the MinElute kit (Qiagen), when more than one product was observed in the agarose gel or cloned directly from PCR into pCR2.1, a TopoTA vector (Invitrogen). White colonies were PCR screened using M13 forward (5'-GTAAAACGACGGCCAGTG-3') and reverse (5'-GGAAACAGCTATGACCATG-3') primers, and positive clones' PCR products were purified by ExoSAP-IT and sequenced using a LiCor 4000 L automated sequencer.
Sequences of rpoB and HMGR gene fragments have been submitted to the European Molecular Biology Laboratory (accession numbers AJ973103–AJ973119 and AJ973229–AJ973245, respectively).
Protein Alignments
All homologous sequences were retrieved from the National Center for Biotechnology Information (NCBI) by performing a BlastP search using characterized proteins as a seed and retaining all full-length homologs with E values of 1 x 10–5 or smaller. After elimination of duplicates and partial sequences, the homologous proteins were aligned with ClustalX (Thompson et al. 1997), leading to a data set of 132, 136, and 109 for HMGR, MVK/PMK, and PPMD, respectively. The alignment was then manually refined with BioEdit (http://www.mbio.ncsu.edu/BioEdit/bioedit.html) keeping regions of unambiguous alignment. We removed almost all the gap-containing positions, except when the insertion/deletion involved only a limited number of taxa. The limits of the unambiguously aligned blocks were fixed to the first encountered constant amino acid position (or to a very conserved position displaying amino acids of the same functional category) preceding or following gap-containing parts. Alignments are available upon request.
We selected sequences for further phylogenetic analyses (by maximum likelihood) from preliminary phylogenetic analyses (by neighbor joining), keeping a representative taxonomic distribution while reducing computational time. In practice, monophyletic sequences of closely related taxa (same genus) were represented by a single sequence (i.e., Streptococcus pneumoniae would also represent Streptococcus pyogenes, Streptococcus thermophilus, Streptococcus mutans, and Streptococcus agalactiae, if they appeared grouped in a robust clade). In eukaryotic taxa, we chose a few representative taxa, which had full-length sequences and grouped together with high support. We retained 52, 67, and 32 sequences for HMGR, MVK/PMK, and PPMD, respectively.
Phylogenetic Analysis of Protein Sequences
Phylogenetic trees were based on the analysis of amino acid sequences by the maximum likelihood method, which takes into account the among-site rate variation, with the program PHYML (Guindon and Gascuel 2002) version 1.0 using the JTT model + eight gamma categories, with estimation of invariant sites. To estimate the robustness of the phylogenetic inference, we used the bootstrap method (Felsenstein 1985) with PHYML using 100 bootstrap trials.
Sequence Alignments and Phylogenetic Analysis of DNA Sequences
DNA sequences obtained from sequenced genomes and cloned PCR fragments were aligned using ClustalX (Thompson et al. 1997). HMGR sequences with 353 or more positions in the HMGR sequences were selected for analysis, and rpoB fragment (880 nt) sequences of the same taxa were analyzed using the FindModel server at the HCV sequence database (http://hcv.lanl.gov/content/hcv-db/findmodel/findmodel.html), based on Modeltest. The best model identified was then used to reconstruct maximum likelihood phylogeny using PHYML (Guindon and Gascuel 2002) with bootstrapping as described above for protein sequences.
Gene Order and Gene Cluster Information
Putative operons and gene clusters were obtained from the Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg) for complete genomes and from the respective references for clusters within organisms which did not have their genomes sequenced. We assumed genes to be in an operon if they were putatively transcribed in the same direction and were less than 30 nt apart.
Approximately Unbiased Test of Tree Selection
A set of topologies including the best trees for Vibrio alignments of the rpoB gene and the HMGR gene were each used as a user-tree in PUZZLE 5.1, option–wsl, with their best respective models of evolution (see above) to estimate the likelihood of each site of a given data set and global tree likelihoods for each tree. These two sets of likelihood values were used as input for CONSEL (Shimodaira and Hasegawa 2001) to perform the approximately unbiased (AU) test (Shimodaira 2002). Trees were rejected at P < 0.05.
Ka/Ks Calculation for Determination of Selection
We calculated Ka/Ks ratios in a 351-nt-long alignment containing 20 taxa, using the Norwegian bioinformatics platform (http://www.bioinfo.no/tools/kaks). Sequence alignment was provided, and analysis was carried out using the best maximum likelihood tree obtained (see above) for the alignment, with determination of codon bias, no weighting, a moderate Li rate (Li, Wu, and Luo 1985), a discrete Grantham submatrix (Grantham 1974), and maximum likelihood tree method (Pupko et al. 2000).
|
Results and Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
Phylogeny of HMGR—Rampant Lateral Gene Transfer in Bacterial Lineages
The wealth of newly sequenced genomes allows for a better sampling of sequence space and therefore should assist in refining evolutionary inferences. We therefore performed an up-to-date analysis of HMGR homologs, to help elucidate the evolutionary history of this protein in different prokaryotic lineages (fig. 1). We performed separate tree-reconstruction analysis for class 1 (fig. 1A) and class 2 (fig. 1B) HMGR sequences to increase resolution of the outer nodes while including an all-inclusive HMGR tree in the Supplementary Material online (Supplementary Fig. S1). Generally, the HMGR trees display little agreement with established organismal phylogeny, and frequent lateral gene transfer (LGT) seems to have obliterated much of the vertical signal. In fact, even the apparently monophyletic eukaryotic and archaeal clades (fig. 1A) have been affected by LGT because members of these domains, such as the Giardia and Archaeoglobus, now possess only a class 2 HMGR (fig. 1B). The distribution of bacterial homologs is extremely patchy, and only within the class Bacilli can strictly vertical descent afford a parsimonious evolutionary scenario. Here we consider the evolutionary histories of class 1 and class 2 individually and discuss contextual genomic information that points to possible mechanisms and selective pressures driving gene transfer.
|
LGT Within Bacterial Class 2 HMGR
The class 2 HMGRs of several distantly related bacterial taxa and representatives of the euryarchaeal orders Archaeoglobales and Thermoplasmatales cluster with most of the low G + C firmicutes with good bootstrap support—including the gammaproteobacteria Legionella pneumophila and Pseudomonas mevalonii, the deltaproteobacterium Bdellovibrio bacteriovorus, the alphaproteobacterium Silicibacter pomeroyi, and the green nonsulfur bacterium Chloroflexus aurantiacus (fig. 1B, top). This clade can be divided into two groups. The top group, as represented in figure 1B (bold), includes the human pathogen L. pneumophila and B. bacteriovorus, a predator of other Gram-negative bacteria, both of which appear to possess complete mevalonate pathway gene clusters.
Legionella pneumophila and B. bacteriovorus also appear as sister taxa with very high support in the phylogenetic trees for the additional mevalonate pathway genes MVK and PPMD (figs. 2 and 3), encoding the enzymes for subsequent steps in IPP synthesis, and exhibit a great deal of similarity in gene order (fig. 4). They also share an open reading frame (ORF), designated lpg2053 in L. pneumophila and Bd1628 in B. bacteriovorus, which has no other significant homologs in the databases by BlastP search but does have an MVK/PMK domain detectable by NCBI conserved domain search. Based on the domain detected and its location within a mevalonate pathway cluster, we propose that this ORF encodes an alternative PMK in both species (which lack recognizable PMK sequences). In B. bacteriovorus, the mevalonate pathway genes are all part of a single putative operon, located downstream of two ribosomal protein genes. Because operons encoding ribosomal proteins are generally highly transcribed, it is likely that mevalonate pathway genes are highly expressed in B. bacteriovorus. Putative cotranscription of mevalonate pathway genes with ribosomal protein genes is also observed in other species such as Lactoccus lactis. On the other hand, in L. pneumophila, the mevalonate pathway genes are split between two operons, one encoding HMGR, PMK, and IPP isomerase (the enzyme which converts IPP to the isoprenoid intermediate dimethylallyl diphosphate) and the other encoding MVK and PPMD along with a putative four-gene ABC transporter system, an outer membrane protein, a peptidoglycan-associated lipoprotein, and a radical-activating enzyme.
|
|
|
It should be emphasized that the vast majority of bacteria uses the alternative 1-deoxy-D-xylulose 5-phosphate pathway for isoprenoid biosynthesis, and all prokaryotic HMGRs recovered by Blast searches have been represented in figure 1. Because, unlike most other delta- and gammaproteobacteria, L. pneumophila and B. bacteriovorus lack that pathway for isoprenoid biosynthesis, we suggest that these mevalonate pathway gene clusters are responsible for isoprenoid synthesis in these two bacteria. Another common feature of these two species is the lack of an HMGS. Because both organisms lead intracellular lifestyles, they are likely to scavenge HMG-CoA from their respective hosts (eukaryotic cells in the case of L. pneumophila and bacterial hosts for B. bacteriovorus). We suggest that both organisms acquired their mevalonate pathway from a common mobile element, whose gene arrangement might have been more similar to that observed currently in B. bacteriovorus, i.e., a single operon structure. Legionella and Bdellovibrio species have been shown to sometimes reside in the same watery environments, and Legionella may serve as prey to Bdellovibrio, making lateral transfer between the two possible (Richardson 1990
). However, the fact that the G + C content of mevalonate genes in these organisms is highly similar to the rest of their respective genomes indicates ancient transfer and subsequent sequence amelioration (Lawrence and Ochman 1997), preventing the identification of the source organism.
In contrast to L. pneumophila or B. bacteriovorus, the bottom group of the top clade (fig. 1B, italics) contains organisms where HMGR is not clustered with any other mevalonate synthesis genes (e.g., A. fulgidus) or indeed is the sole mevalonate pathway gene (as in the megaplasmid of S. pomeroyi). It seems reasonable to suggest that unlike the related L. pneumophila/B. bacteriovorus HMGR homologs the ancestral gene of this group was already separated from other mevalonate pathway genes and that the first lateral acquisition event involved no other genes of that pathway. The most likely explanation for the highly irregular taxonomical clustering in this clade is multiple transfer events of HMGR genes—first among bacteria, then from bacteria to archaea, and eventually within archaea (see also Boucher et al. 2001). Any scenario that did not invoke transfer would require many independent events of loss (to explain patchy distribution) and highly paralogous multigene families in ancestral genomes (to account for the highly nonstandard phylogenetic pattern shown by the genes that have not been lost). In the absence of other mevalonate pathway genes, transferred HMGR genes cannot contribute to IPP synthesis and must serve some other role(s) in the organisms bearing them. One such role was demonstrated in P. mevalonii, which uses mevalonate as a carbon source by converting it back to HMG-CoA with HMGR (Gill, Beach, and Rodwell 1985; Beach and Rodwell 1989). HMG-CoA can then be broken down by HMG-CoA lyase (Scher and Rodwell 1989), a highly conserved enzyme found in many organisms with or without a mevalonate pathway. HMG-CoA lyase catabolizes HMG-CoA (an intermediate in additional metabolic pathways such as the degradation of valine, leucine, and isoleucine) to acetoacetate and acetyl-CoA. HMGR and HMG-CoA lyase genes are located in the same operon in P. mevalonii which is induced in the presence of mevalonate (Gill, Beach, and Rodwell 1985; Anderson and Rodwell 1989).
Lateral Acquisition of Class 1 HMGR by Bacteria
Two separate bacterial clades containing class 1 HMGR are present in the phylogenetic tree (fig. 1A, middle). Although both these clusters are not nested within either archaeal or eukaryotic clusters, the bacterial HMGRs are probably more closely related to those of archaea because they share a four–amino acid insertion that is not found in eukaryotic homologs (Supplementary Fig. S2, Supplementary Material online, see also Boucher and Doolittle 2000). Again, a division may be drawn between the clade containing HMGR homologs that are part of a mevalonate pathway cluster (fig. 1A, bold), including those from Actinobacteria, the alphaproteobacterium P. zeaxanthinifaciens, and the first paralog from the gammaproteobacterium Coxiella burnetii, and the clade containing HMGR homologs that are not colocalized with other mevalonate genes, including those of the gammaproteobacteria Vibrio, Photorhabdus luminescens, and C. burnetii (second paralog) and the betaproteobacterium Chromobacterium violaceum (fig. 1A, italics).
With the first group (HMGR linked to other mevalonate pathway genes), several suggestive patterns are observed. Although the HMGR homologs from the marine zeaxanthin-producing P. zeaxanthinifaciens and C. burnetii do not form a distinct clade in the HMGR tree, their mevalonate clusters exhibit a very similar gene order (fig. 5) and their MVK and PPMD homologs group together with high support (figs. 2 and 3), leading to the conclusion that they are probably derived from a common ancestor. In contrast, the gene order of the actinobacterial mevalonate clusters, known to be required for the production of secondary metabolites, is similar to that of low G + C firmicutes with the exception of the HMGR gene (fig. 5, see also Humbelin et al. 2002), and their MVK and PPMD homologs are more related to those of that phylum. Likely, their present HMGR was recruited into a preexisting mevalonate cluster, which lacked an HMGR copy, or it displaced an ancient HMGR in that cluster. Because C. burnetii does not seem to possess a 1-deoxy-D-xylulose 5-phosphate pathway for isoprenoid biosynthesis, the mevalonate operon may be required for primary metabolism in this bacterium.
|
The clade of mevalonate pathway–independent homologs presents the HMGR-encoding genes in a variety of interesting genetic contexts, suggesting alternate cellular functions. The putative HMGR of the entomopathogen P. luminescens is longer than other bacterial homologs and contains an additional phosphotransferase domain. It is the second gene in a seven-gene operon which also contains a putative prenyl transferase, a phosphoenol pyruvate synthase, an AMP-binding acyl-CoA synthetase/nonribosomal peptide synthethase, and phosphatydilserine synthase homologs, as well as a putative transferase, indicating that it may be involved in the biosynthesis of an as yet uncharacterized secondary metabolite. In C. violaceum, a bacterium common in tropic soils and an occasional human pathogen, the HMGR homolog is the third gene in a six-gene operon encoding an acetyl-CoA carboxylase alpha subunit, a probable oxydoreductase protein, a probable biotin carboxylase protein, a protein containing an MmgE/PrpD family protein with homologs involved in propanoate catabolism (methylcitrate dehydratase), and a protein with a putative acetyl-CoA carboxylase domain. Based on its gene composition, we speculate that this operon may be involved in the complex set of reactions that compose propanoate metabolism in this bacterium. The putative HMGR of the causative agent of Q fever, C. burnetii, is encoded by CBU0030 (NP_819085 [GenBank] ) and is the second gene in a nine-gene operon, including genes encoding a putative oxidoreductase, two acyltransferase family proteins, an alkyl-dihydroxyacetonephosphate synthase, and a ribose 5-phosphate isomerase. At present, we cannot make any functional predictions regarding this operon.
Curiously, although C. burnetii is closely related to the genus Legionella, neither of its HMGR homologs is related to those of Legionella HMGR, and both belong to class 1 (fig. 1A).
HMGR Evolution at the Genus Level: Vibrio as a Case Study
HMGR is distributed patchily in bacteria as a whole, and the evidence for LGT from phylogenetic reconstructions and comparative genomics was not unexpected. In some more phylogenetically restricted groups, all or most members may possess HMGR, and yet recombination, orthologous replacement, and/or episodic loss and gain could still play an important evolutionary role. To examine this, we undertook a detailed study of HMGR evolution in species of Vibrio.
Vibrio species are important pathogens of mammals, fish, shellfish, and corals (for a recent review see Thompson, Iida, and Swings 2004). Although all four sequenced Vibrio species (V. cholerae, V. vulnificus, V. parahaemolyticus, and V. fischeri) possess HMGR homologs, none is located in a putative operon and there is no conservation in their neighboring genes or gene clusters. These Vibrio HMGR genes are always found on the minor chromosome, which typically has fewer genes responsible for primary metabolic functions. Another doubt regarding the importance of this gene to Vibrio metabolism is cast by the fact that in whole-genome microarray studies of V. cholerae it exhibited extremely low transcription levels when bacteria were grown in a rich liquid medium, a rabbit ileal loop model for human infection (Xu, Dziejman, and Mekalanos 2003), or during adhesion to a crab shell which models chitin binding and surface colonization of shellfish (Meibom et al. 2004). However, its presence in all sequenced vibrios did indicate some level of potential conservation and prompted us to examine its distribution and evolution in a broader sampling. We chose Vibrio isolates both from New England coastal water and Australian aquaculture to avoid either biogeographic or virulence-related biases.
Based on whole-genome sequencing data, HMGR nucleotide sequences of Vibrio species can vary by up to 30%. We designed several different PCR primers, based on the DNA alignment of the representatives of all four sequenced Vibrio genomes. We amplified sections of the HMGR gene in various isolates using different primer sets and cloned and sequenced the PCR products. We then designed degenerate primers on the basis of the now expanded sequence alignment, which now included more Vibrio sequences. Overall, we were successful in obtaining an HMGR gene fragment with one or more primer pairs in 17 out of 19 Vibrio isolates. DNA from the remaining two isolates gave a strong positive signal when hybridized in a low-stringency DNA dot blot with a probe from the V. fischeri HMGR gene. In contrast, we were unable to amplify an HMGR gene fragment from any of the four Photobacterium isolates in our sampling, and only one gave a strong positive dot blot signal in the hybridization (data not shown). The complete sequence of Photobacterium profundum does not contain any HMGR homologs. The high prevalence observed in several (possibly all) species lineages within the genus Vibrio, including the deep-branching V. fischeri (Thompson, Iida, and Swings 2004), suggests that the HMGR gene was acquired in a single LGT event by an ancestral Vibrio.
Although the scenario inferring a single LGT event in the root of Vibrio is appealing in its simplicity, there is also the possibility that the observed prevalence is due to the fact that the gene was acquired by LGT, multiple times in separate lineages. We therefore reconstructed the phylogeny of HMGR based on amplified fragments and gene sequences from the available whole-genome sequences (see above), and compared it to that of the highly conserved rpoB gene, which can be considered as reflecting the species phylogeny. To that end, we amplified and sequenced 880-bp fragments of rpoB from each of the 19 Vibrio and 4 Photobacterium DNAs. We constructed an rpoB phylogeny and verified that all Vibrio and Photobacterium sequences clustered into their respective monophyletic clades. Sequences of the rpoB amplicons of Vibrio species that gave HMGR products of 353 nt or longer and rpoB sequences from sequenced Vibrio genomes were used for analysis. To rule out the effects of a possible mutational saturation of third codon positions, we constructed an alignment in which third codon nucleotides were eliminated. We then generated best neighbor-joining and maximum likelihood trees, with bootstrapping, from both the original alignment and an alignment composed of first and second codon positions only. The HMGR tree topologies observed were highly similar, and all highly supported clades were observed in both trees, indicating that third codon positions were not saturated in the HMGR phylogeny. The rpoB phylogeny was affected by the removal of the third nucleotide position, and one supported relationship was lost but no new significant relationships emerged. This is due to the fact that this gene is highly conserved at the protein level with very few amino acid substitutions, as reflected by the identity between a tree based on a corresponding rpoB amino acid sequence alignment and the one based on the two first nucleotides alignment. Thus, this gene does not appear to have mutational saturation in the third codon position but rather has most of its phylogenetic signal concentrated within it.
Although the trees are in general agreement and are compatible with our hypothesis of a single ancestral acquisition event, a few incongruencies could be observed (fig. 6, top). To determine whether the extent of incongruence between the rpoB tree and the HMGR tree was statistically significant, we used the AU test of tree selection (Shimodaira 2002). The trees were found to be statistically different, each rejecting the other's topology at the threshold of 5%. Detection of recombination in HMGR using RDP2 (Martin, Williamson, and Posada 2005) identified several putative recombination events, but none of these events were recognized by more than one algorithm, making results inconclusive. This is likely to be caused by the relatively short sequence length, which greatly impedes recombination detection. For rpoB, significant recombination events were detected by two or more algorithms for several Vibrio isolates (7A03, 12F11, 14A09, 12F02) and the sequences derived from the genomes of V. fischeri ES114 and V. vulnificus YJ016. Thus, it appears that subsequent to the acquisition of HMGR recombination or orthlogous replacement between different Vibrio species in either or both HMGR and rpoB genes has obscured some of the vertical phylogenetic signal. To our knowledge, this is the first evidence for interspecies recombination in vibrios other than for genes in these species' "superintegrons." It suggests that a multispecies multilocus sequence–typing study would reveal recombination involving many loci in this genus.
|
HMGR Is Under Purifying Selection in Vibrio
The function of HMGR in Vibrio species is yet unknown, but because they have no other mevalonate pathway genes, it is probably not related to isoprenoid synthesis. Only two of the four sequenced Vibrio species have an HMG-CoA lyase (see above), required along with HMGR for the degradation of mevalonate to acetoacetate and acetyl-CoA (Anderson and Rodwell 1989
). It therefore seems unlikely that the role of HMGR in this genus is in the utilization of exogenous mevalonate as a carbon source. Because HMGR exhibits extremely low transcription (see above) in V. cholerae under various conditions, one may speculate that it could be a pseudogene. We performed analysis of the ratios of asynonymous/synonymous substitutions in the HMGR DNA alignment (Ka/Ks). We observed Ka/Ks ratios of between 0.04 and 0.18 along the tree (Supplementary Fig. S3, Supplementary Material online), indicating the existence of strong purifying selection on the HMGR gene. It is therefore reasonable to conclude that despite the fact that the induction conditions have not yet been determined for HMGR, this gene is physiologically important, at least in some Vibrio species. |
Conclusion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
An evolutionary history so rich in transfer events indicates that a complete mevalonate pathway confers a broadly applicable biochemical property, the ability to synthesize IPP for the production of isoprenoids, either for primary or secondary metabolism. Lateral acquisition of such a cluster could be beneficial for bacteria that retain a 1-deoxy-D-xylulose 5-phosphate pathway because it allows regulation of isoprenoid synthesis for secondary metabolite to be separated from primary metabolic functions (as in several Actinobacteria). Alternatively, auxotrophic bacteria which had come to generally rely on various host metabolites may find it metabolically advantageous to utilize host HMG-CoA for isoprenoid synthesis and subsequently lose their ancestral pathway altogether (e.g., as in Legionella). Attainment of an HMGR without any other mevalonate pathway genes, on the other hand, is more difficult to explain, as it is likely to involve recruitment of this gene for other physiological functions, and could involve alteration in substrate specificity. Characterizing the novel roles of HMGR when integrated into new operons, biochemically and genetically, will be of great interest and may lead to the discovery of new metabolites. Elucidation of the physiological role of the isolated HMGR of Vibrio will prove particularly challenging because although it is under strong purifying selection, induction conditions for its expression have not been identified thus far.
The traditional way to reliably infer an interdomain LGT event has been to find a single taxon or a few closely related taxa which nest within a robust clade of highly distributed taxa. As our coverage of sequence space improves and hundreds of microbial genomes accumulate, recovery of such simple clear-cut cases of LGT becomes impossible. This is due to the fact that genes and especially operons which can contribute to a variety of microbial lifestyles and ecological settings will tend to be transferred multiple times. Clearly, some genes, like HMGR, are prone to be transferred multiple times, often over large evolutionary distances, and merit a unique classification, reflecting their irregular evolutionary nature. Bapteste and colleagues have recently suggested such LGT-prone genes should be named "mobilons" (Bapteste, Macleod and Doolittle, unpublished data). Mobilons will be hard to detect using traditional approaches (such as best Blast hit outside a certain taxonomical division and not inside it) and hard to support based on phylogeny alone, forcing us to rely heavily on patchiness of distribution as well as phylogenetic signal.
It is remarkable that two close sister taxa (L. pneumophila and C. burnetii) have converged on the same metabolic solution, namely, replacing the 1-deoxy-D-xylulose 5-phosphate pathway for isoprenoid biosynthesis, which is broadly distributed and probably ancestral in Proteobacteria with a laterally acquired mevalonate pathway from different sources. Both species have no HMGS gene presumably because they can obtain HMG-CoA from the host—another convergent feature in their evolution. Thus, two separate LGT events followed by convergent evolution can sometimes create a false impression of an ancestral state that can only be rejected by phylogenetic reconstruction.
|
Supplementary Material |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
Supplementary Figures S1–S3 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).
SUPPLEMENTARY FIG. S1. Phylogeny of HMGR protein sequences. A maximum likelihood tree for class HMGR protein sequences was generated using a JTT + gamma model with eight categories and estimation of invariant sites. Bootstrap values were obtained using PHYML bootstrapping. Bootstrap support values exceeding 50% are displayed. Size bar represents substitutions per site.
SUPPLEMENTARY FIG. S2. Conserved insertion/deletion in archaeal type class 1 HMGR.
SUPPLEMENTARY FIG. S3. Values of Ka/Ks along the phylogenetic tree based on HMGR gene fragment multiple alignment.
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
U.G. is supported by a Killam postdoctoral fellowship, W.F.D. is supported by the Canada Research Chair Program, and this research is supported by Genome Atlantic and the Canadian Institutes for Health Research. The authors wish to thank David Byers for his help with various marine Vibrio species, David A. Walsh and Olga Zhaxybayeva for their help with phylogentic analysis, and Ellen R. Boudreau for critical reading of the manuscript.
|
Footnotes |
|---|
Takashi Gojobori, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResults and DiscussionConclusionSupplementary MaterialAcknowledgementsReferences |
|---|
Anderson, D. H., and V. W. Rodwell. 1989. Nucleotide sequence and expression in Escherichia coli of the 3-hydroxy-3-methylglutaryl coenzyme A lyase gene of Pseudomonas mevalonii. J. Bacteriol. 171:6468–6472.
Beach, M. J., and V. W. Rodwell. 1989. Cloning, sequencing, and overexpression of mvaA, which encodes Pseudomonas mevalonii 3-hydroxy-3-methylglutaryl coenzyme A reductase. J. Bacteriol. 171:2994–3001.
Boucher, Y., and W. F. Doolittle. 2000. The role of lateral gene transfer in the evolution of isoprenoid biosynthesis pathways. Mol. Microbiol. 37:703–716.[CrossRef][ISI][Medline]
Boucher, Y., H. Huber, S. L'Haridon, K. O. Stetter, and W. F. Doolittle. 2001. Bacterial origin for the isoprenoid biosynthesis enzyme HMG-CoA reductase of the archaeal orders Thermoplasmatales and Archaeoglobales. Mol. Biol. Evol. 18:1378–1388.
Boucher, Y., M. Kamekura, and W. F. Doolittle. 2004. Origins and evolution of isoprenoid lipid biosynthesis in archaea. Mol. Microbiol. 52:515–527.[CrossRef][Medline]
Felsenstein, J. 1985. Confidence intervals on phylogenies: an approach using the bootstrap. Evolution :783–791.
Gill, J. F. Jr., M. J. Beach, and V. W. Rodwell. 1985. Mevalonate utilization in Pseudomonas sp. M. Purification and characterization of an inducible 3-hydroxy-3-methylglutaryl coenzyme A reductase. J. Biol. Chem. 260:9393–9398.
Grantham, R. 1974. Amino acid difference formula to help explain protein evolution. Science 185:862–864.
Guindon, S., and O. Gascuel. 2002. Efficient biased estimation of evolutionary distances when substitution rates vary across sites. Mol. Biol. Evol. 19:534–543.
Humbelin, M., A. Thomas, J. Lin, J. Li, J. Jore, and A. Berry. 2002. Genetics of isoprenoid biosynthesis in Paracoccus zeaxanthinifaciens. Gene 297:129–139.[Medline]
Lawrence, J. G., and H. Ochman. 1997. Amelioration of bacterial genomes: rates of change and exchange. J. Mol. Evol. 44:383–397.[CrossRef][ISI][Medline]
Li, W. H., C. I. Wu, and C. C. Luo. 1985. A new method for estimating synonymous and nonsynonymous rates of nucleotide substitution considering the relative likelihood of nucleotide and codon changes. Mol. Biol. Evol. 2:150–174.[Abstract]
Martin, D. P., C. Williamson, and D. Posada. 2005. RDP2: recombination detection and analysis from sequence alignments. Bioinformatics 21:260–262.
Meibom, K. L., X. B. Li, A. T. Nielsen, C. Y. Wu, S. Roseman, and G. K. Schoolnik. 2004. The Vibrio cholerae chitin utilization program. Proc. Natl. Acad. Sci. USA 101:2524–2529.
Pupko, T., I. Pe'er, R. Shamir, and D. Graur. 2000. A fast algorithm for joint reconstruction of ancestral amino acid sequences. Mol. Biol. Evol. 17:890–896.
Richardson, I. R. 1990. The incidence of Bdellovibrio spp. in man-made water systems: coexistence with legionellas. J. Appl. Bacteriol. 69:134–140.[Medline]
Scher, D. S., and V. W. Rodwell. 1989. 3-Hydroxy-3-methylglutaryl coenzyme A lyase from Pseudomonas mevalonii. Biochim. Biophys. Acta 1003:321–326.[Medline]
Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51:492–508.[CrossRef][ISI][Medline]
Shimodaira, H., and M. Hasegawa. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17:1246–1247.
Thompson, F. L., T. Iida, and J. Swings. 2004. Biodiversity of vibrios. Microbiol. Mol. Biol. Rev. 68:403–431.
Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins. 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876–4882.
Thompson, J. R., S. Pacocha, C. Pharino, V. Klepac-Ceraj, D. E. Hunt, J. Benoit, R. Sarma-Rupavtarm, D. L. Distel, and M. F. Polz. 2005. Genotypic diversity within a natural coastal bacterioplankton population. Science 307:1311–1313.
Xu, Q., M. Dziejman, and J. J. Mekalanos. 2003. Determination of the transcriptome of Vibrio cholerae during intraintestinal growth and midexponential phase in vitro. Proc. Natl. Acad. Sci. USA 100:1286–1291.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. A. S. Snyder, S. McGowan, M. Rogers, E. Duro, E. O'Farrell, and N. J. Saunders The Repertoire of Minimal Mobile Elements in the Neisseria Species and Evidence That These Are Involved in Horizontal Gene Transfer in Other Bacteria Mol. Biol. Evol., December 1, 2007; 24(12): 2802 - 2815. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Juttner and S. B. Watson Biochemical and Ecological Control of Geosmin and 2-Methylisoborneol in Source Waters Appl. Envir. Microbiol., July 15, 2007; 73(14): 4395 - 4406. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||