Genomes of the Symbiotic Nitrogen-Fixing Bacteria of Legumes

Unrooted phylogenetic tree of 16S rDNA sequences from selected rhizobial strains in the α-proteobacteria. Sequences were aligned using ClustalW and the tree was constructed by using the neighbor-joining method. GenBank/EMBL accession numbers of the sequences used for analysis are as follows: CP000133.1, R. etli CFN 42; AM236080.1, R. leguminosarum bv viciae; EF054890.1, R. tropici; D11343.1, Rhizobium galegae; BA000012.4, M. loti MAFF303099; X77123.2, Sinorhizobium fredii strain LMG 8317; X67221.1, Azorhizobium caulinodans; EF428944.1, S. meliloti; Y17047.1, Allorhizobium undicola; AY904770.1, Bradyrhizobium elkanii strain SEMIA 6169; BA000040.2, B. japonicum USDA 110; CP00494, Bradyrhizobium sp. strain BTAi1; CU234118, Bradyrhizobium sp. strain ORS278; CP000463.1, Rhodopseudomonas palustris BisA53; AB298401.1, Methylobacterium extorquens; AF220762.1, Methylobacterium nodulans. Rhizobia with fully sequenced genomes are indicated in bold. Rhizobial genomes currently being sequenced are indicated by an asterisk (*). The recent completion of the genomes of R. leguminosarum bv viciae (Young et al., 2006), Rhizobium etli (González et al., 2006), and two photosynthetic Bradyrhizobium strains (Giraud et al., 2007) has increased the number of available complete rhizobial genome sequences to seven, including sequences obtained for B. japonicum (Kaneko et al., 2002), Mesorhizobium loti (Kaneko et al., 2000), and S. meliloti (Barnett et al., 2001; Capela et al., 2001; Finan et al., 2001; Galibert, et al., 2001). The genomic architecture of these species varies considerably (Table I Architecture of rhizobial genomes Architecture of rhizobial genomes It is now apparent that multipartite genomes are typical among members of α-proteobacteria (Jumas-Bilak et al., 1998), although they are not restricted to members of this group. While multipartite genomes are often associated with species that interact with a host (Egan et al., 2005), the advantage afforded by a genome consisting of multiple replicons is not always obvious (Teyssier et al., 2004; Egan et al., 2005). However, it has been proposed that such an organization may increase the adaptive potential of a species (González et al., 2006), which, if valid, implies that the presence of multipartite genomes in many rhizobial species may reflect the complex lifestyle adopted by these plant-associated bacteria. The presence of additional replicons in the genome of the α-proteobacteria may enhance the bacterium's metabolic or symbiotic capabilities, thus ensuring stable maintenance of the replicon within the population. Genome size is influenced by environmental factors, and soil-dwelling species, such as the rhizobia, tend to have larger genomes (Bentley and Parkhill, 2004). B. japonicum has the largest chromosome size, approximately 9.2 Mb, and the difference in chromosome sizes among the rhizobia may in part be due to the presence of extrachromosomal (plasmid) DNA. While the bradyrhizobia generally lack plasmids, they are a common feature among the fast-growing microsymbionts. The complexity and heterogeneity of soil presumably necessitates a large inventory of genes to maximize survival of free-living cells (Bentley and Parkhill, 2004; Young et al., 2006), while the ability to establish a symbiotic relationship with a host plant imposes an additional genetic requirement upon rhizobia. Accordingly, it has been proposed that rhizobial genomes have evolved via expansion, primarily through lateral gene transfer and gene duplication, as a means of adjusting to the challenges imposed by this lifestyle (Batut et al., 2004; Boussau et al., 2004). Symbiotically relevant genes in rhizobia are often clustered on large plasmids (pSym), or within genomic islands (referred to as symbiosis islands [SIs]), emphasizing the accessory nature of the genes and their ability to be acquired via horizontal gene transfer. A SI present in M. loti strain ICMP3153 was found capable of transforming nonsymbiotic strains of M. loti into symbiotic counterparts (Sullivan and Ronson, 1998). While 611 and 681 kb SIs were also reported in M. loti strain MAFF303099 (Kaneko et al., 2000) and B. japonicum strain USDA110 (Kaneko et al., 2002), respectively, the transmissibility of these islands has yet to be confirmed. Intriguingly, the integration of SIs into the M. loti and B. japonicum genomes occurs within phe-tRNA and val-tRNA genes, respectively. The association of SIs in both species with a phage-related integrase implies that SIs may have originated from the ancient integration of a bacteriophage. In contrast, in S. meliloti, R. etli, R. leguminosarum, and Sinorhizobium sp. strain NGR234, the majority of genes involved in symbiosis (nod, nif, and fix) are plasmid borne (Freiberg et al., 1997; Barnett et al., 2001; Galibert et al., 2001; González et al., 2003, 2006; Young et al., 2006) and several pSyms have the potential or documented ability to be transferred among bacteria via conjugation (Rao et al., 1994; Freiberg et al., 1997; Barnett et al., 2001; González et al., 2003; Brom et al., 2004; Pérez-Mendoza et al., 2004). The organization of symbiotic genes within mobile islands or plasmids permits the conversion of nonsymbiotic saprophytes into nitrogen-fixing plant endosymbionts (and vice versa) in a single step. Rhizobial genomes appear to be highly dynamic entities and this is particularly reflected by the presence of many insertion sequence (IS) elements, transposases, and related genes, within regions encoding symbiotic functions. The SIs of M. loti and B. japonicum encode the majority of transposase genes and IS-related sequences in these genomes (Kaneko et al., 2000; Göttfert et al., 2001; Kaneko et al., 2002; Sullivan et al., 2002). In B. japonicum, an incredible 60% of all transposase genes (100 of 167) were located within the SI, which comprises a mere 7.5% of the entire genome. The islands of M. loti strains ICMP3153 and MAFF303099 differ in size by >110 kb; much of the interstrain variation is attributable to DNA of external origin (Sullivan et al., 2002). This is consistent with the dynamic nature of the islands (Finan, 2002). Similar trends for the accumulation of IS elements have also been reported for plasmid pSymA of S. meliloti (Galibert et al., 2001), pNGR234a from Sinorhizobium sp. strain NGR234 (Viprey et al., 2000), and plasmid p42d from R. etli (González et al., 2006), which is likely due to the nonessential nature of many of the genes within these genomic regions (Viprey et al., 2000). Reiterated sequences offer potential sites for recombination (Mavingui et al., 2002) and are likely involved in the movement of symbiotic genes within and among the genomes of the root- and stem-nodule bacteria. For example, in R. leguminosarum, almost identical symbiotic regions are present on the plasmids pRL10 and pRL1 (Young et al., 2006). Similarly, genes clustered within the SI of M. loti strain R7A are located on plasmid pMLa in strain MAFF303099 (Sullivan et al., 2002), and an approximately 42-kb region of pSymA was likely transferred to an accessory plasmid (pSmeSM11a) in S. meliloti (Stiens et al., 2006). Palacios and colleagues have performed elegant studies of genome plasticity in rhizobia, demonstrating that extensive recombination occurs within the symbiotic plasmid of Sinorhizobium sp. NGR234 (Flores et al., 2000), as well as cointegration of this plasmid with the chromosome and/or megaplasmid of this species (Mavingui et al., 2002). They have also demonstrated cointegration of the S. meliloti megaplasmids with the chromosome (Guo et al., 2003). The plasticity and instability of rhizobial genomes are likely due to the presence of repeated DNA sequences (Flores et al., 1987), as well as IS elements and multiple replicons (Romero and Palacios, 1997). It has been postulated that large-scale recombinational events such as these offer a means of reshuffling genes among replicons over successive generations. It should be noted, however, that this potential source of genetic variation may be limited should excision events preferentially occur within cointegration sites, as has been demonstrated in S. meliloti (Guo et al., 2003). The multiphasic lifestyle adopted by rhizobia has influenced the size, complexity, and also the content of their genomes (Bentley and Parkhill, 2004; Boussau et al., 2004). This appears to be a common trend among other members of the α-proteobacteria that have a varied lifestyle, including Rickettsia (an obligate intracellular animal pathogen), Bartonella and Brucella (facultative intracellular animal pathogens), Agrobacterium (an extracellular pathogen of plants), and the animal symbiont Wolbachia (Tsolis, 2002). Comparative genomic analyses of members of the α-proteobacteria have shown that these phylogenetically diverse bacteria share many similarities in their metabolism, physiology, overall genome architecture, and processes required for interaction with their cognate hosts (Paulsen et al., 2002; Tsolis, 2002). Interestingly, the obligate intracellular α-proteobacteria have genome structures indicative of prolonged gene loss, whereas the large size and complexity of the genomes of free-living species indicates expansion by both gene duplication and horizontal gene transfer events (Batut et al., 2004; Boussau et al., 2004; Giraud et al., 2007). Rhizobial genomes are richly endowed with transport, regulatory, and stress-related systems, all of which are essential for interacting with the external environment and were acquired subsequent to the divergence of plant-associated Rhizobiales from related α-proteobacteria (Boussau et al., 2004). Some nodule bacteria, like the recently sequenced Bradyrhizobium sp. strains BTAi1 and ORS278 (Giraud et al., 2007) are amazingly metabolically and ecologically diverse and have very complex genomes. These strains can grow as heterotrophs, autotrophs, phototrophs, and symbiotically with members of the plant genus Aeschynomene. Analysis of the R. leguminosarum genome sequence indicates that this species shares more genes in common with two other plant symbionts, S. meliloti and M. loti, than with the more closely related plant pathogen Agrobacterium tumefaciens, consistent with the observation that genome content (as well as size) is directly influenced by a species' lifestyle (Young et al., 2006). In addition to the rhizobia discussed above, genome sequencing projects have also been initiated or completed for a number of plant-associated α-proteobacteria, including Agrobacterium vitis S4, and Agrobacterium radiobacter K84. Genome sequencing of Rhizobium tropici and Sinorhizobium medicae, and the nitrogen-fixing, plant-growth-promoting bacteria Azospirillum brasilense and Gluconacetobacter diazotrophicus, is also nearing completion. Since nodulation and nitrogen fixation genes in many rhizobia are plasmid borne, the sequences of several large rhizobial plasmids were determined prior to the completion of the whole genome scale projects. The first such initiative involved the complete sequencing of the symbiotic plasmid pNGR234a (Freiberg et al., 1997). This work was among the first to report a high frequency of IS and related sequences within rhizobial genomes and led the authors to suggest that lateral gene transfer played a prominent role in the disparate evolution of a plant pathogen and endosymbiont from a common ancestor. The sequence of the symbiotic plasmid (p42d) of R. etli strain CFN42 (González et al., 2003) also reflects the mosaic nature of symbiotic regions in rhizobia. Intriguingly, IS sequences located on either side of a 125 kb symbiotic region and a putative integrase gene raised the possibility that this plasmid-borne symbiotic region may itself be motile. To date, there have been no published reports of plasmids in the slow-growing bradyrhizobia, which is likely due to the fact that no symbiotic genes have been localized to Bradyrhizobium plasmids. Many rhizobial strains appear to carry one or more nonsymbiotic, accessory plasmids that may confer benefits with regard to overall fitness. One such plasmid, the 144 kb pSmeSM11a from S. meliloti, has been sequenced (Stiens et al., 2006). The majority of genes in pSmeSM11a were predicted to encode for proteins involved in DNA replication, recombination, and repair, as well as genes encoding various metabolic enzymes and transport systems. A gene encoding an 1-aminocyclopropane-1-carboxylic acid deaminase (acdS) was also identified and it was postulated that AcdS indirectly promotes plant growth by decreasing ethylene levels (Glick, 2005). Consistent with this suggestion is the finding that the introduction of acdS and its regulatory gene into an S. meliloti strain lacking these genes resulted in a strain with considerably superior nodulating ability (Ma et al., 2004). This study demonstrates that genes localized to accessory plasmids may have a role in microbial fitness, and underscores the importance of undertaking similar projects in the future. The transcriptome for S. meliloti has been examined under a variety of conditions, including in planta (Ampe et al., 2003; Bèrges et al., 2003; Rüberg et al., 2003; Barnett et al., 2004; Becker et al., 2004; Krol and Becker, 2004; Uchiumi et al., 2004; Capela et al., 2005; Barnett and Fisher, 2006; Capela et al., 2006; Domínguez-Ferreras et al., 2006). To simultaneously and systematically measure both plant and bacterial gene expression, Long and colleagues developed a dual-genome chip that carried a probe set specific for the entire S. meliloti genome, and an additional approximately 10,000 gene probe set corresponding to the host plant Medicago truncatula (Barnett et al., 2004). One of the more interesting results obtained from this study was the observation that the majority of tentative plant consensus sequences that were up-regulated in nitrogen-fixing nodules produced by wild-type S. meliloti were similarly expressed in Fix− nodules (i.e. unable to fix nitrogen) induced by a fixJ − mutant strain. This result further supports the hypothesis that the nitrogen-fixing ability of colonizing rhizobia is likely not important with respect to plant root and nodule gene expression (Barnett et al., 2004; Barnett and Fisher, 2006). While there have been reports of host sanctions imposed upon rhizobia in nonfixing nodules in soybeans (Glycine max), and that host plants are sensitive to the nitrogen-fixing status of individual nodules (Kiers et al., 2003), this likely takes place long after the establishment of nodules. The functional genomics of rhizobia is ultimately dependent on the ability to rapidly develop mutants in interesting loci identified using transcriptional analyses. In S. meliloti, this issue was addressed by the development of an ORFeome functional genomics platform (Schroeder et al., 2005). In addition, in vivo expression technology has been used for the identification of genes with nodule-inducible expression (Oke and Long, 1999) and to define S. meliloti promoters that were actively expressed during the earliest stages of infection (Zhang and Cheng, 2006). One of the difficulties inherent in working with plant hosts with indeterminate root nodules (such as M. truncatula and alfalfa [Medicago sativa]) is that colonizing bacteria (and plant cells) may be present at various developmental stages. Such a heterogeneous population makes the identification of genes relevant to a particular stage of symbiosis challenging, and may result in apparent discrepancies between different studies (Barnett et al., 2004). The use of plant and bacterial developmental mutants offers a means of analyzing gene expression at specific stages of infection, albeit with the caveat that the expression patterns of such mutants may not always be representative of that exhibited by the wild-type bacterium. In S. meliloti, a bacA mutant, which fails to survive upon entry into plant cells, permitted expression analysis of cells contained within infection threads (Ampe et al., 2003; Capela et al., 2006). Similar studies are required to tease apart gene expression patterns at each stage of nodulation. Transcriptomics and functional analyses in other rhizobial species are less developed, although micro- and macroarray studies have been performed in B. japonicum and M. loti, respectively. In B. japonicum, an array was used to identify potential members of the NifA and RegR regulons using probes sets specific for >700 candidate genes (Hauser et al., 2006). Microarray analysis in this species has also been used to examine cell response to iron-limiting conditions, and to identify iron-regulated genes (Rudolph et al., 2006; Yang et al., 2006a, 2006b). Recently, full genome oligo-based microarrays for B. japonicum have been developed (see http://www.bradygenome.org/) and likely other functional genomic studies in this bacterium. macroarray performed upon from M. loti strain MAFF303099 resulted in the identification of expression islands the symbiotic et al., 2004). In contrast, gene expression in of the was during a similar to that by Barnett et for S. The symbiotic islands of M. loti strains MAFF303099 and R7A differ by and several expression islands were located within MAFF303099 Intriguingly, in analyses indicated that many genes expression was up-regulated in are likely not by a required for the of Uchiumi et have that may be used by these bacteria as a means of expression of genes within the symbiotic While the of genome sequence a in the of a bacterial species, gene to is essential to within the One of this issue is through the of an as has been reported in S. et the and of all predicted within the S. meliloti genome into an entry which has been to the transfer via recombination of the into a (Schroeder et al., 2005). Such plasmids may be used in the of a gene of to expression of the or to mutants in which gene expression has been The in which genes are expressed may offer into the of the For that et constructed a gene of S. meliloti that genes at to of the genes in the S. meliloti genome (see These have been for the identification of genes expression is by environmental et al., 2006) and in studies to the identification of by the present in different were and specific and were for over transport et al., 2006). gene offers means of gene via of et reported the of all putative in the S. meliloti genome, and identified two and as being essential for A consisting of S. meliloti mutant strains has also been constructed et al., 2006). Such in the identification of genes essential for symbiosis and that are involved in in the of nodulation. analyses have now additional into the symbiosis between S. meliloti and B. japonicum and their host et al., 2000; et al., 2004; and 2006). One trend that has during studies of B. japonicum and S. meliloti is the expression of a of in nodule bacteria 2004; et al., 2004; and 2006). cells of both species a large number of and expression of these is in planta 2004; and 2006). A different set of transport appears to be specific to with predicted including S. and et that acid between plant cells and R. leguminosarum occurs via and and is required for the of nitrogen by plants et al., 2003). While it is that similar are in place in the B. japonicum and S. meliloti expression of the majority of these transport was not found to be similarly up-regulated in analyses (Barnett et al., 2004). Since several studies have shown that there is often between and et al., 2004; Barnett and Fisher, 2006; Capela et al., 2006), both of functional studies to be to a of the role of specific genes and proteins in the symbiotic The and of differ considerably from of free-living cells et al., et al., 2000; et al., 2003; Barnett et al., 2004; Becker et al., 2004; 2004; et al., 2004). gene expression is in (Barnett et al., 2004; Becker et al., 2004; Capela et al., 2006), as a result of growth and a et al., 2006). a considerably larger number of proteins in are than consistent with a in cell et al., 2000). the the plant rhizobia with a source in the of which are via the acid In both S. meliloti and B. japonicum, enzymes involved in have been reported in 2004; and 2006), that this may be in these the involved in the first of was in B. japonicum and 2005), it was not found in the of S. meliloti and expression of was as in studies (Barnett et al., 2004; Becker et al., 2004). was not in alfalfa et al., may differ in these two In for nodule bacteria the host plant with a source of In S. meliloti of as a source of nitrogen is This was in both gene expression (Barnett et al., 2004; Becker et al., and analyses 2004). This result is consistent with the hypothesis that produced via nitrogen fixation is not by for their nitrogen The nitrogen-fixing root nodules on and the genetic of the symbiotic between strains and host plants is Accordingly, the recent of genomic sequences of sp. strains is an development et al., 2007). In this work has into the of the sequenced and has in genome content and size with the of the host plants associated with each strain. This work the first of both genome in the host strain and expansion in the host strain within closely related strains of bacteria. This further the that genome content is a of a species' lifestyle and that the genome is influenced by the and of their studies rhizobia have upon a of genes involved in root nitrogen and other symbiotic However, the recent completion of genome sequencing projects in several rhizobial species permits a more analysis of gene and One trend that has is the highly dynamic and complex nature of rhizobial genomes. it is that genome architecture and content is influenced by the multiphasic lifestyle adopted by and analyses have into however, within these of This is in studies indeterminate root the issue of analyzing expression patterns at various stages of infection has been addressed through the use of bacterial and plant developmental The that has been documented between and the importance of additional functional genomic such as studies upon the of and bacteria et al., 2004; et al., 2004). Such are to of the that between and their microsymbionts. for with the of the phylogenetic tree of 16S rDNA