KEYWORDS: secondary replicons, genome analysis, genome organization, genomics, megaplasmids, population genetics, secondary chromosome

Purpose of This Review

The first goal of this review is to build upon previous reviews (11,23–31) and to provide an unbiased assessment of the information available on the structure, function, and evolution of divided bacterial genomes. This consists of a comprehensive review of the relevant literature as well as an analysis of all complete genomes available through the National Center for Biotechnology Information (NCBI) genome database (accessed 21 March 2016) as a way of reviewing the abundant, untapped information present within these sequences. The second goal is to synthesize the data presented throughout this review into a generalized model explaining the evolution and function of the multipartite bacterial genome.

Replicon and Secondary Replicon

We use the term “replicon” as a general term in reference to any DNA molecule regardless of its specific nature, and each replicon can be further classified based on specific characteristics, as described below. The term “secondary replicon” refers to any replicon that is not the primary chromosome of the cell. We suggest that each replicon be classified into one of the following five groups, as described below and inFig. 1: chromosome, second chromosome, chromid, megaplasmid, and plasmid.

In the strictest sense of the term replicon, it should be used only in reference to DNA molecules with a single origin of replication. While this distinction is irrelevant in reference to bacterial genomes, this definition would exclude the chromosomes of some archaea that have a chromosome containing multiple origins of replication (32,33). As such, while the classification system described here should be applicable to archaea, the term replicon should be avoided when describing the chromosomes of archaea. As this review is focused only on bacterial genomes, the term replicon is used throughout.

Chromosome

“Chromosome” refers to the primary replicon. As described by Harrison et al. (11), the chromosome is always the largest replicon in the genome and contains the majority of the core/essential genes. There is nearly a 100-fold distribution in the sizes of fully sequenced and assembled bacterial chromosomes, with average and median sizes of ∼3.65 Mb and ∼3.46 Mb, respectively (Fig. 2A). The average and median bacterial genome sizes are ∼3.87 Mb and ∼3.65 Mb, respectively (Fig. 2A), illustrating that the chromosome accounts for nearly all of the genetic material of most prokaryotic organisms. However, this is not universal. For example, the chromosomes ofSinorhizobium meliloti 1021 andBurkholderia xenovorans LB400 account for only 54.6% and 50.3% of their genomes, respectively (18,22). Nevertheless, 1,017 (∼59.5%) of the 1,708 bacterial species with a complete genome available in the NCBI database contain a chromosome but no secondary replicon (chromid, megaplasmid, or plasmid), while only 192 (∼11%) have a chromid and/or megaplasmid. The major features of genomes and all replicon classes are summarized inTable 1.

Plasmid and Megaplasmid

Many of the secondary replicons in bacterial genomes carry no core genes and are nonessential and thus dispensable for cell viability in most environments. The majority of the genes on these replicons were acquired through recent horizontal gene transfer (HGT), and their genomic signatures, such as GC content and dinucleotide composition, differ significantly from the chromosome (11). These types of replicons, defined by the lack of core genes, are referred to as “plasmids” and “megaplasmids.” The distinction between plasmid and megaplasmid is currently based solely on size, although there is no established boundary between plasmid and megaplasmid in the literature. While any size limitation is essentially arbitrary, we suggest a lower cutoff of 350 kb for megaplasmid status, as this is equal to roughly 10% of the median bacterial genome size. Any nonessential replicon of <350 kb would therefore be a plasmid. When using this boundary to distinguish megaplasmids from plasmids, the average and median plasmid sizes are ∼78.9 kb and ∼46.2 kb, respectively. In contrast, the average and median megaplasmid sizes are approximately 10 times larger, at ∼772 kb and ∼558 kb, respectively, with the pSymA replicon ofS. meliloti 2011 (1.35 Mb) and the third replicon ofBurkholderia lata 383 (1.4 Mb) being the largest to have been experimentally demonstrated to be nonessential and thus megaplasmids (34,35). It is also interesting to note that the sizes of chromosomes follow a bell-shaped distribution (Fig. 2A), whereas the size distributions of plasmids and megaplasmids are instead positively skewed (Fig. 2B andC). This is perhaps suggestive of evolutionary forces acting to limit the size of these nonessential replicons.

The term megaplasmid was originally coined in reference to a largeS. meliloti plasmid (7), and since then, megaplasmid has been used simply as a way of referring to a large plasmid. Similarly, size is used as the sole feature distinguishing megaplasmids from plasmids in this review. However, there may be a less arbitrary means of separating megaplasmids from plasmids that has yet to be elucidated. For example, megaplasmids often have a copy number equal or similar to that of the chromosome, they often encode their own partitioning systems, and the replication and partitioning of megaplasmids can be integrated into the cell cycle. It will be interesting to see if future research can identify a clear, functional distinction between plasmids and megaplasmids aside from an arbitrary size distinction.

Chromid

The term “chromid” itself is a combination ofchromosome and plasmid (11) and underscores how chromid refers to a replicon that is an intermediate between a plasmid and a chromosome (11). The replication systems of chromids are similar to those of plasmids and megaplasmids (11) but may have additional regulatory controls that integrate their replication into the cell cycle (29,36,37). However, unlike plasmids and megaplasmids, chromids carry at least one gene that is essential for cell viability (i.e., a core gene whose loss would result in cell death) and generally have genomic signatures that better resemble those of the chromosome (11). Ideally, a replicon would be classified as a chromid based on experimental evidence that the replicon carries a nondispensable core gene and not just on genome annotation. In our set of putative chromids, the average (∼1.52 Mb) and median (∼1.26 Mb) sizes are around 2-fold larger than the average and median megaplasmid sizes (∼0.77 Mb and ∼0.56 Mb, respectively), despite the lower size cutoff of chromids and megaplasmids in our classification scheme being the same. Additionally, the size distribution of the putative chromids displays a weak positive skew (Fig. 2B). The larger size and weaker positive skew of the putative chromids than of megaplasmids may suggest weaker evolutionary pressure to limit replicon size.

It was suggested by Dziewit et al. that chromids be further subdivided into primary chromids and secondary chromids (38). In this classification scheme, primary chromids are absolutely essential for cell viability. In contrast, secondary chromids may be dispensable under some conditions but are expected to be required for survival in the organism's natural habitat. We accept that these subdivisions are potentially useful but would add that the replicon must be essential in the cell's native habitat to be considered a secondary chromid; for example, a secondary replicon in a soil-dwelling, opportunistic pathogen must be essential in the soil to be considered a secondary chromid. Similarly, many secondary replicons, including small plasmids, carry antibiotic or heavy metal resistance genes. These genes are required for growth in environments containing these antibiotics or heavy metals, but we consider environment-specific essentiality such as this to be insufficient for the chromid designation. For the sake of this review, primary and secondary chromids are not differentiated.

It is worth noting that the majority of chromids are considered essential solely on the basis of the genome annotation, and the expectation that the chromid carries a single-copy essential gene is largely without experimental support. These inferences are not always correct; the third-largest replicon of theBurkholderia cepacia complex species was thought to be essential based on genome annotations but was since shown to be a virulence megaplasmid (35). Similarly, theminCDE genes of the pSymB replicon ofS. meliloti were predicted to be essential (39), although follow-up experimentation revealed them to be dispensable (40). However, there are experimentally validated cases of essential genes existing on chromids, such as theengA and tRNA^arg genes on the pSymB replicon ofS. meliloti (41). It is therefore important to experimentally validate the essential nature of more putative chromids to develop a true understanding of the prevalence of this replicon type. However, an inability to remove a replicon from the genome should not be considered sufficient to confirm its essentiality, as there may be other explanations, such as the presence of plasmid addiction systems (42–44). For example, despite being a nonessential replicon (34), the pSymA megaplasmid ofS. meliloti is nearly impossible to forcefully remove (cure) from the cell due in part to the presence of at least three active toxin-antitoxin loci (45,46).

Second Chromosome

Historically, the term “second chromosome” was used in reference to a replicon that would now be described as a chromid. As nicely described by Harrison et al., chromid is a more appropriate term to describe such replicons, and the use of second chromosome in this respect should no longer be applied (11). As we describe below in Proposed Mechanisms of Chromid Formation, it is highly likely that nearly all secondary replicons carrying essential core genes evolved from plasmids. However, very rarely, a secondary replicon may form as a result of a split of an ancestral chromosome into two replicons, and we propose that the term second chromosome continue to be used to describe this rare occurrence. No documented cases of such an event are present in the literature; however, we found two examples by scanning the complete genomes available in the NCBI genome database. Assuming that these are not errors in the genome assembly, synteny analysis revealed that the ∼0.73-Mb replicon ofSalmonella enterica strain NCTC10384 and the ∼2.66-Mb replicon ofNocardia farcinica NCTC11134 represent second chromosomes (see Fig. S1 in the supplemental material). Although it may be difficult to differentiate between a second chromosome and a chromid when the second chromosome was formed through a very ancient split, in general, differentiation between these replicon classes should be possible. Second chromosomes are expected to show high synteny to the chromosomes of related species, depending on the age of the split, and the distribution of core genes between the two replicons is expected to be random, unlike that for chromids. For the analyses presented in this review, second chromosomes were not differentiated from chromids due to their low abundance, and neither of the above-mentioned strains are included in our list of representative strains.

Classification of the Replicons Present in the NCBI Genome Database

As of 21 March 2016, the NCBI genome database contained 4,541 genomes, representing 1,708 bacterial species, that were marked as a “complete genome.” As a way of reviewing the untapped information held within these genome sequences and to examine whether conclusions based on intensive research on a limited number of species were generalizable, we downloaded the RefSeq (47) version of all 4,541 complete genomes; annotated each replicon as “chromosome,” “putative chromid,” “megaplasmid,” or “plasmid”; and performed several analyses on each class of replicon. By using the RefSeq version, we could be sure that all genomes were consistently annotated using the NCBI prokaryotic genome annotation pipeline. The complete set of replicons identified in the database is provided in File S2 in the supplemental material. In many of the subsequent analyses, we did not examine all 4,541 genomes but instead chose one random representative genome for each species in order to limit bias due to certain species being overrepresented in the database. We did not perform further controls for phylogenetic structure, such as controlling for genera or families that were overrepresented. We also did not attempt to determine whether two secondary replicons in related genomes shared a common ancestry, but if one were to attempt such an analysis, we feel that common ancestry should be based on phylogenetic analysis of the replication/partitioning proteins and not simply on gene content or synteny.

The annotation of the replicons was performed as described below, which largely follows the process outlined inFig. 1, with some exceptions. First, the largest replicon in each genome was annotated as the chromosome. This step did not involve a search for essential genes due to the inability of us to do so on such a large scale; however, as the largest replicon is always the chromosome (11), chromosomes can be reliably annotated solely on the basis of size. Next, any replicon that was below 350 kb was classified as a plasmid, as these replicons were below our size threshold for megaplasmids and putative chromids. Megaplasmids and putative chromids were distinguished on the basis of GC content and dinucleotide relative abundance. A replicon with a GC content within 1% of the corresponding chromosome and with a dinucleotide relative abundance distance from the chromosome of ≤0.4 (Fig. S2) was considered a putative chromid. Otherwise, the replicon was called a megaplasmid. Each of the 1,708 species was considered to have a megaplasmid or a putative chromid as long as one was found in at least one genome available for that species. Second chromosomes were not differentiated from putative chromids due to the rarity at which second chromosomes occur and the impracticality of manually examining each putative chromid to determine if it was a second chromosome. As defined above, a replicon must have an essential core gene to be classified as a chromid; however, the massive computational requirements to perform this analysis on such a large scale prevented us from doing so. Therefore, putative chromids and megaplasmids were differentiated on the basis of genomic signatures, which are often good proxy measures for distinguishing these replicons types. The limitations of the methods used here are discussed further in the supplemental material. While the methods used here are certainly imperfect, we believe that a small number of misclassified replicons will have a limited influence on the analyses described below and will not significantly impact the conclusions that are drawn. Indeed, the ability to detect differences between the putative chromid group and megaplasmid group with respect to several nonselected characteristics, as described below, supports this notion.

The Schism Hypothesis

The schism hypothesis states that a second essential replicon is formed as a result of a split of an ancestral chromosome into two replicons, a chromosome and a chromid. The schism hypothesis is the older of the two ideas and was initially proposed to describe the chromid formation ofBrucella suis (70) andRhodobacter sphaeroides (71). If the schism hypothesis is correct, it would predict that the properties of the two resulting replicons are highly similar, with an equal distribution of core genes between both replicons. The ability to produce viableE. coli orBacillus subtilis strains that have had their single chromosome artificially split into two self-replicating chromosomes provides support to show that such a scenario is possible (72,73). However, the strong enrichment of essential genes on the chromosomes of species with multiple replicons is inconsistent with this model (11). Additionally, evidence now indicates that the chromids ofB. suis andR. sphaeroides did not result from a schism event.

In the case ofB. suis, biovars 1, 2, and 4 contain a chromid, while biovar 3 has a single chromosome with a size equal to that of the chromosome plus the chromid of the other biovars (70). It was originally proposed that the single-replicon structure was ancestral (70); however, phylogenetic analysis subsequently showed this was not true and that the single chromosome of biovar 3 resulted from a fusion of the chromosome and chromid in this lineage (23). ForR. sphaeroides, the many genomic features being highly similar between the 3.2-Mb chromosome and the 0.94-Mb chromid (71), the large number of gene duplications between these replicons (74), and the large number of genes on the chromid predicted to be essential (75,76) led to the suggestion that the chromid resulted from a split of an ancestral chromosome. However, the lower coding density of the chromid than of the chromosome (76,77), gene functional biases as determined by Cluster of Orthologous Genes (COG) analyses (76), differences in evolutionary rates (78), and variations in gene content and size (78,79) are inconsistent with this view. Overall, there is little evidence for the formation of a secondary essential replicon through the schism hypothesis occurring in nature except perhaps very rarely, and as mentioned above, we recommend that such replicons be referred to as second chromosomes.

The Plasmid Hypothesis

Whereas the schism hypothesis predicts that a secondary essential replicon evolves from a chromosome, the plasmid hypothesis states that it evolves from a megaplasmid. According to the plasmid hypothesis, the sustained coevolution of a megaplasmid with a chromosome will result in a regression of the genomic signature of the megaplasmid to that of the chromosome and the gain of essential genes potentially through transfer from the chromosome. As summarized above and as described by Harrison et al. (11), such replicons should be referred to as chromids.

In support of this model, the replication and partitioning machinery of chromids resembles that of megaplasmids (11), although in many cases, experimental evidence for the functionality of these systems is lacking (31). That said, replicons of therepABC family that carry essential genes, and are thus chromids, haverepABC replication/partitioning genes that have a codon usage more similar to that of the chromosome than dorepABC family members that do not carry essential genes (80), consistent with these replication systems being functional. Additionally, data from a phylogenetic analysis of the plasmid partitioning protein RepA were consistent with chromids evolving from preexisting megaplasmids in theAlphaproteobacteria (11).

There are two main observations that can be explained by the plasmid hypothesis but that cannot be accounted for by the schism hypothesis. Essential genes are strongly underrepresented on chromids (11), as would be expected if the chromid originated as a megaplasmid that subsequently gained a few core genes, for example, through interreplicon gene transfers. There is also a consistently observed bias in the functional annotation of genes present on chromids versus genes on chromosomes, as determined via COG analyses (see, for example, references19,21,22, and76). This is not surprising if the chromid and chromosome originated independently. In contrast, equal distributions of core genes and of functional annotations would be expected if the chromid formed as a result of a chromosomal schism. Hence,in toto, it appears as though the plasmid hypothesis is likely to explain the formation of most, if not all, chromids studied to date.

Conversion of a Megaplasmid into a Chromid

The transition of a replicon from a megaplasmid into a chromid requires two main conversions: the amelioration of the genomic signatures to that of the chromosome and the acquisition of core/essential genes. Genomic signatures such as codon usage and dinucleotide composition are shaped by a variety of factors and can have adaptive advantages (81,82). Therefore, the similarity of the genomic signatures between chromosomes and chromids in the same species is presumably driven by evolutionary forces selecting for optimized genome function and can be caused by, for example, selection for improved translational efficiency or mutational biases of the cellular machinery (81,82).

Less intuitive is how to explain the occurrence of essential genes on a chromid when the cell was fully capable of surviving without this replicon in the past. There are two possible mechanisms for this process. The primary process is through interreplicon translocations resulting in the transfer of essential genes from the chromosome to the secondary replicon. Perhaps the best-supported example of this mechanism is inS. meliloti (Fig. 3A). Two essential genes have been experimentally demonstrated to exist on theS. meliloti pSymB chromid,engA and an unique arginine tRNA,^ARGtRNA_CCG (41). Computational analysis of the surrounding region demonstrated that their presence on pSymB is the result of the translocation of a contiguous 69-kb fragment, includingengA and the tRNA, from the chromosome to the pSymB precursor in a recentS. meliloti ancestor (41,83). InV. cholerae, there are four putatively essential genes (dsdA,thrS, L20, and L35) present in two clusters on the chromid, and all of these genes are chromosomally situated in relatedVibrio species (31). Numerous other clusters of genes in theVibrio andBurkholderia genera and the orderRhizobiales are predicted to have moved from the chromosome to a secondary replicon (84). Similarly, 25 to 30% of genes on theS. meliloti chromid have orthologs on theAgrobacterium tumefaciens chromosome (85), suggesting significant amounts of interreplicon gene flow, which is supported by the results of a genealogy study that are suggestive of recombination between theS. meliloti replicons in nature (86). Moreover, a phylogenetic analysis of individual genes betweenBacillus cereus strains indicated the frequent transfer of genes between chromosomes and plasmids (87). The precise mechanism through which gene transfer from the chromosome to a secondary replicon occurs has not been studied. However, considering that the multiple replicons of a multipartite genome can naturally form cointegrants (88,89), it may be that the integration of the replicons followed by an imprecise excision event results in interreplicon translocations (90). Alternatively, a recombination event, mediated, for example, by insertion sequence (IS) elements, may result in the excision of a chromosomal gene region that is subsequently captured by the secondary replicon.

The second putative mechanism through which secondary replicons can come to carry core genes is genetic redundancy. It was experimentally shown through transposon mutagenesis that >10% of chromosomal genes inS. meliloti may have a functionally redundant copy on one of the secondary replicons (91). Based on sequence similarity, there also appear to be many gene duplications between the chromosome and secondary replicons ofR. sphaeroides (74,92),V. cholerae (21),B. cereus (87), andBurkholderia vietnamiensis (93). Genetic redundancy between core genes on the chromosome and the chromid could be a result of an interreplicon duplication of a chromosome gene or the acquisition of an orthologous gene through horizontal gene transfer. If the copy of the gene on the secondary replicon is able to fully complement the disruption of the chromosomal version, the degeneration of the chromosomal copy would be fitness neutral, and the second copy of the gene would become the sole copy, in effect transferring a core gene to a secondary replicon.

Phylogenetic Distribution of Secondary Replicons

Of the 1,708 bacterial species examined, 11.2% included strains with a multipartite genome (megaplasmid and/or chromid), and 7.4% or 6.4% included at least one strain with at least one putative chromid or one megaplasmid, respectively (Fig. 4; see also the supplemental material). Moreover, there appeared to be an affinity for putative chromids to cooccur with megaplasmids, as ∼2.5% of all the species examined had both a chromid and a megaplasmid (although not necessarily in the same strain). While some of this apparent cooccurrence of putative chromids and megaplasmids may reflect the difficulty in clearly distinguishing between these replicons, we note that the majority of species that appeared to have both putative chromids and megaplasmids were in the genusBurkholderia and the orderRhizobiales, which are known to carry both elements. The apparently higher prevalence of putative chromids than of megaplasmids in the bacterial phylogeny was surprising given that chromids appear to have evolved from megaplasmids. This may reflect a greater instability or more dynamic nature of megaplasmids than of chromids.

By using theace function of theape package in R (95), it was predicted that putative chromids arose 45 times in the bacterial phylogeny and were lost only twice (see the supplemental material for methods and the limitations of this analysis). Of the 126 species containing a putative chromid, the large majority (∼91%) of them contained only one. At most, five putative chromids were detected in a single species, and strikingly, all three species with five putative chromids belonged to the genusAzospirillum. Similar to species with a putative chromid, ∼88% of the 109 species containing a megaplasmid had only a single megaplasmid. An additional ∼11% of these species had two megaplasmids, and only the agarolytic marine bacteriumPersicobacter sp. strain JZB09 had three. In contrast, ∼51% of the 627 species with a plasmid contained more than 1 plasmid, ∼12% had at least 5 plasmids, and nearly 2% had 10 or more plasmids. At most, 21 plasmids, accounting for ∼40% of the total genome, were identified in a single genome; this was observed forBorrelia burgdorferi. In fact, all four species with >15 plasmids belonged to the genusBorrelia.

Multipartite genomes were dispersed throughout the bacterial phylogeny, but clear clusters of species with multipartite genomes are visible (Fig. 4). In particular, megaplasmids were observed to be common in genera that contain numerous soil and marine bacteria that interact with eukaryotic species in either a symbiotic or a pathogenic relationship. These genera includedBacillus,Burkholderia,Sinorhizobium,Rhizobium,Mesorhizobium,Agrobacterium, andMethylobacterium. Megaplasmids were also enriched in the generaRhodococcus andNovosphingobium, which contain soil and marine organisms capable of inhabiting polluted environments and catabolizing the pollutants. Putative chromids were similarly found to be prevalent in several genera with species that enter into symbiotic or pathogenic relationships with eukaryotic organisms. These genera includedSinorhizobium,Rhizobium,Agrobacterium,Burkholderia,Cupriavidus,Vibrio,Pseudoalteromonas,Azospirillum,Ralstonia, andPrevotella. Putative chromids were also prevalent in a few genera containing species that are able to survive in extreme environments due to resistance to several stresses such as UV irradiation, metal ions, and aromatic compounds. These genera includedRalstonia,Deinococcus, andCupriavidus. The lack of additional clusters in the phylogeny may simply reflect sequencing biases and the underrepresentation of genome sequences from certain taxa. For example, the only representative genome for each of the generaPersicobacter (agarolytic marine bacterium),Tistrella (soil and marine bacteria),Chelatococcus (marine moderate thermophiles), andChloracidobacterium (marine moderate thermophiles) contained a putative chromid, and the sequencing of additional genomes from these genera may reveal that the presence of a multipartite genome is characteristic of these genera.

It was previously noted that chromids appear to contain genus-specific genes, and the presence of a chromid may correspond to the emergence of a new genus (11). This observation remained largely true in this expanded data set, although some exceptions were detected, where the presence of a putative chromid was not a defining characteristic of the genus. For example,R. sphaeroides contains a chromid, whereasRhodobacter capsulatus does not;Xanthomonassacchari contains a putative chromid, whereas the other sevenXanthomonas species do not; and only a fewDeinococcus species have a putative chromid, but the species having a putative chromid did not form a monophyletic group in the phylogeny (Fig. 4). On the other hand, the acquisition of a chromid may also predate the emergence of a genus. For example, it was argued that the chromids of the generaSinorhizobium,Rhizobium, andAgrobacterium were acquired by the common ancestor of these genera prior to their divergence (84).

In contrast to chromids, megaplasmids are rarely conserved at the genus level, although multiple species in a genus will often contain a megaplasmid. Even in the rare cases where megaplasmids are present in all species of a genus, different species may have unique megaplasmids. For example, allSinorhizobium species have at least one strain with a megaplasmid, but analysis of the replication and partitioning proteins suggests that they do not share a common ancestry (96).

Codon Usage

The codon usage of a gene is correlated with the expression level of the gene; highly expressed genes have a codon usage that closely mimics the relative tRNA abundance, whereas lowly expressed genes often do not (97). Differences in codon usage bias between replicons have been reported for numerous species (11,98). For example, Cooper et al. (98) examined codon usage bias in 22 species, including species of theAlphaproteobacteria,Betaproteobacteria,Gammaproteobacteria, andDeinococci, and in all cases, the codon usage biases of the chromosomes were greater than those of chromids, which in turn were greater than those of additional chromids or megaplasmids.

An analysis of codon usage bias was performed on representative genomes from 1,708 bacterial species by using CodonO to calculate synonymous codon usage order (SCUO) (see the supplemental material) (99). It was found that 85.6% of putative chromids and 85.9% of megaplasmids had less codon usage bias (a lower SCUO value) than that of the corresponding chromosome, with a median SCUO difference between the chromosome and a putative chromid or megaplasmid of −0.02 or −0.06, respectively (Fig. 5A). Somewhat surprisingly, only 61.7% of plasmids had an SCUO value lower than that of the chromosome, with a median difference of −0.02 (Fig. 5A). Additionally, the difference in SCUO values between chromosomes and megaplasmids displayed an unexpected bimodal distribution, unlike the bell-shaped distribution for putative chromids (Fig. 5A). However, both of these unexpected results appear to be due to the inclusion of genomes with low codon usage bias in the analysis (Fig. S3). When the analysis was limited to genomes with a chromosomal SCUO value above the median (Fig. 5A, inset), the bimodal distribution for megaplasmids was largely eliminated, and 77.4%, 95.5%, and 82.3% of plasmids, megaplasmids, and putative chromids, respectively, displayed less codon usage bias than the chromosome. The average differences in SCUO values from the chromosome were −0.05, −0.08, and −0.02 for plasmids, megaplasmids, and chromids, respectively, and the means of all three differences were statistically different from zero (uncorrectedP value of <1e⁻¹⁵ by a one-samplet test) and from each other (α = 0.05 by one-way analysis of variance [ANOVA] with Tukey's honestly significant difference [HSD]post hoc test). Overall, these results are consistent with secondary replicons generally displaying less codon usage bias than the chromosome and with the codon usage bias of chromids being higher than that of megaplasmids and plasmids, at least in genomes with high chromosomal codon usage bias.

GC Content

GC contents vary considerably in prokaryotic organisms and can range from ∼15% (14.55% in “Candidatus Carsonella ruddii” HT) to ∼75% (74.91% inAnaeromyxobacter dehalogenans 2CP-C). Several factors can influence the GC content of an organism, among which are environmental adaptation (100,101) and possibly recombination (102). In addition to differing between species, GC contents can vary considerably within a genome and have often been used to identify genes recently acquired through horizontal gene transfer (103). The GC content of each replicon in a multipartite genome is almost always different, and the extent of the difference is reflective of replicon type; the GC contents of a chromid and a megaplasmid usually differ by <1% and >1%, respectively, from that of the chromosome (11). In an analysis of one representative genome from 1,708 bacterial species, the median absolute GC content difference between a chromosome and a putative chromid was 0.34% (standard deviation [SD], 0.29%), that between a chromosome and a megaplasmid was 1.9% (SD, 2.0%), and that between a chromosome and a plasmid was 2.8% (SD, 3.1%) (Fig. 5B), with each difference being statistically different from zero (uncorrectedP value of <1e⁻¹⁵ by a one-samplet test) and from each other (α = 0.05 by one-way ANOVA with Tukey's HSDpost hoc test) (see the supplemental material).

Not only does the extent of differences in GC contents differ between putative chromids and megaplasmids/plasmids, the direction of the difference also appeared to differ. It was previously observed that the majority of plasmids have a GC content lower than that of the chromosome (104–106). Similarly, in the analysis presented here, the majority of plasmids (78.5%) and megaplasmids (89.4%) had a GC content lower than that of the chromosome (Fig. 5B), with the mean of each distribution being statistically different from zero (uncorrectedP value of <1e⁻⁸ by a one-samplet test). However, only little bias in the direction of the GC content difference for the putative chromids relative to chromosomes was observed; 58.2% of chromids had a GC content lower than that of the chromosome, while 41.2% had a higher GC content (Fig. 5B), with rather low statistical support for the mean of the distribution differing from zero (uncorrectedP value of 0.013 by a one-samplet test). It was suggested that the lower GC content of plasmids is due to selection for reduced energy expenditure, as the maintenance of GC-rich sequences is energetically more expensive (104). Perhaps, as chromids cannot be lost from the genome, selection for reduced energy expenditure is largely absent, and the GC content is shaped almost solely by the same forces acting on the chromosome. Alternatively, evidence suggests a general mutation bias toward AT due to G/C to A/T transitions (107,108). The reduced GC content of plasmids/megaplasmids, but not chromids, may be reflective of more relaxed selection acting on these replicons (98).

Dinucleotide Relative Abundance

The profiles of dinucleotide relative abundances in a genome have been shown to be distinct for each bacterial genome and are reflective of bacterial phylogeny (105,109). Studies have also illustrated that dinucleotide relative abundances can be used to differentiate chromosomes from chromids and from plasmids (81). The dinucleotide relative abundance distance refers to the sum of the differences in the frequencies of each dinucleotide pair between two sources of DNA. In the current analysis of 1,708 representative genomes, it was seen that the median absolute difference between a chromosome and a putative chromid was 0.21 (SD, 0.06), that between a chromosome and a megaplasmid was 0.50 (SD, 0.28), and the difference between a chromosome and plasmid was 0.91 (SD, 0.50) (Fig. 5C), with all differences being statistically different from zero (uncorrectedP value of <1e⁻¹⁵ by a one-samplet test) and from each other (α = 0.05 by one-way ANOVA with Tukey's HSDpost hoc test) (see the supplemental material). Thus, as for GC content, putative chromids appeared the most like chromosomes, while plasmids appeared the least like chromosomes.

It is interesting to note that the dinucleotide relative abundance distances of the putative chromids compared to chromosomes appeared to follow a bell-shaped distribution centered away from zero, whereas the difference in the GC contents of the putative chromids compared to chromosomes appeared centered at around zero (Fig. 5B andC). This suggests that whereas the GC content of chromids is continually ameliorated toward that of the chromosome, there is a constraint on the amelioration of the dinucleotide composition. However, whether this is simply a consequence of the original dinucleotide relative abundance difference between the two replicons or whether this reflects an adaptive function is unclear.

Conjugal Transfer and Interreplicon Genomic Signature Differences

For both GC content and dinucleotide relative abundance, it was observed that the difference between chromosomes and megaplasmids was less than that between chromosomes and plasmids despite both elements being nonessential replicons (Fig. 5). A possible explanation may be that the mobility of megaplasmids is lower than that of plasmids and that successful megaplasmid transfer to phylogenetically distant organisms occurs less frequently than it does for plasmids. Although megaplasmids can retain conjugative machinery (see, for example, references110–113) and the transfer of megaplasmids between related organisms has been observed in nature (see references114–117, among others), experimental studies have noted difficulties in promoting megaplasmid transfer between phylogenetically distant species (113). Like megaplasmids, at least some chromids retain conjugative properties and can be induced to transfer to naive cells under laboratory conditions (118). Moreover, the chromid ofS. meliloti Rm41 is naturally transmissible (119). However, there is no evidence for successful horizontal transfer (transfer and maintenance) of chromids in nature (11), and transfer of theS. meliloti pSymB chromid to the related organismA. tumefaciens in the laboratory resulted in an obvious fitness decrease (120). Hence, it is likely that plasmids are highly mobile, and many of the plasmids detected by genome sequencing are relatively newly acquired. In contrast, the successful transfer (i.e., the transfer and maintenance) of megaplasmids, and more so for chromids, is less frequent (due to either poor maintenance following transfer or the inability of the replicon to conjugate), meaning that most of the detected megaplasmids/chromids were acquired less recently, providing more time for the amelioration of the genomic signatures.

Genetic Variability

The levels of sequence and gene conservation between related bacterial strains and species are different for chromosomes, chromids, and megaplasmids. This can be visualized inFig. 3, which illustrates how theS. meliloti chromosome is highly conserved within theSinorhizobium genus, theS. meliloti chromid shows less conservation, and theS. meliloti megaplasmid is poorly conserved. Studies have shown thatS. meliloti chromosomal genes show the highest level of conservation, followed by chromid genes and finally by megaplasmid genes, both when comparing different strains ofS. meliloti and in an interspecies comparison withSinorhizobium medicae (122,123). Similarly, in both theVibrio andBurkholderia genera, higher percentages of chromosomal genes are conserved between species than are chromid genes, and where applicable, genes on the megaplasmid or the smaller chromid were the least conserved (98,124,125). InR. sphaeroides, greater synteny between the chromosomes of related strains than between the chromids of the same strains was observed (78). Chromosome- and megaplasmid-specific pangenome analyses of 11Bacillus thuringiensis strains with a megaplasmid by using Roary (126) revealed that whereas 5,153 chromosomal genes were present in at least 5 genomes (1,984 in all 11 strains), only 163 megaplasmid genes were present in at least 5 of the strains (none in all 11 strains) (see methods and Fig. S4 in the supplemental material). Interestingly, comparisons between the Gram-negativeCupriavidus species showed greater ortholog conservation between chromosomes than between chromids, whereas comparison between strains belonging to the sameCupriavidus species showed only slightly greater ortholog conservation on the chromosome than on the chromid (27,127,128). Similarly, genes on all replicons inB. cenocepacia showed strong conservation between strains, but whereas the level of conservation of the chromosome remained high compared to those of otherBurkholderia species, a low level of conservation of the smaller chromid/megaplasmid was seen, while the larger chromid is highly conserved only in closely related species (125).

The same general observations are detected when nucleotide conservation, instead of gene conservation, is examined. For 7 of the 9 examined species belonging to the generaBrucella,Rhodobacter,Burkholderia, andVibrio, the level of nucleotide identity between the chromosomes of strains belonging to the same species was greater than the level of nucleotide identity between the chromids (129). The same pattern was observed for 9 of 10 intergenus comparisons of related species (129). In a population genomics study, Epstein et al. observed that similar percentages (∼95%) of nucleotides of the chromosome and chromid of the referenceS. meliloti andS. medicae genomes were conserved across 32 and 12 strains, respectively, while the level of conservation of the megaplasmid was much lower (<80%) (123).

All considered, these data suggest that chromosomes are the most genetically stable replicons, followed by chromids and finally by megaplasmids. Chromosomes display high synteny at both the species and genus levels. Chromids may be conserved nearly as strongly as chromosomes at the species level; however, the level of conservation drops off at the genus level. In contrast, megaplasmids can display high variability even between strains of the same species.

Evolutionary Rates

Several studies have observed different rates of evolution on each replicon in a multipartite genome. The substitution rate of the chromid ofVibrio species is higher than that of the chromosome, whereas purifying selection is weaker on the chromid (98). Similarly, the substitution rate of the chromid inBurkholderia multivorans is higher than that of the chromosome but lower than that of the megaplasmid, while purifying selection is greatest on the chromosome, then the chromid, and finally the megaplasmid (98). InS. meliloti, the rate of positive selection is highest for the chromid (121). A comparison of orthologous gene products ofBurkholderia xenovorans to those ofBurkholderia cepacia indicated that the percent amino acid identity was highest for the chromosome, intermediate for the chromid, and lowest for the small chromid/megaplasmid (22). In contrast, genes involved in rhizobium-legume symbiosis carried by the megaplasmid ofSinorhizobium species showed less divergence than those carried by the chromosome or chromid (130). While at first glance, this result conflicts with theBurkholderia observations, it is perhaps not surprising, as the megaplasmid is the primary replicon with respect to the symbiosis.

Mutation accumulation studies withB. cenocepacia (131,132) indicated that the rates of the different types of substitution mutations differed across replicons. Additionally, the overall rate of substitution mutations, but not indels, was highest on the chromosome and lowest on the chromid (131). Given that the evolutionary rate of the chromosome is lower than those of the other replicons, it was suggested that the above-mentioned results are consistent with much stronger purifying selection on the chromosome (131). Somewhat different results were observed forVibrio species. InVibrio fischeri, the substitution mutation rate was higher on the chromid than on the chromosome, while no difference was detected inV. cholerae (133). The rates of the particular substitutions also varied between replicons (133).

Comparison of conserved sequences betweenR. sphaeroides strains suggested that the chromid is experiencing more rapid evolution than the chromosome (78). However, when only duplicated genes were considered, there was only a non-statistically significant difference in selective constraint for genes where one duplicate was on the chromosome and the other was on the chromid compared to when both duplicates were on the same replicon (134). This may suggest that the higher rate of divergence of secondary replicons is not an intrinsic property of secondary replicons but instead reflects differences in the types of genes. However, a separate study determined that the elevated evolutionary rates of genes on secondary replicons were due to both an intrinsic property of secondary replicons as well as differences in the types of genes (98). This was done by comparing rates of evolution of genes in the multipartiteBurkholderia andVibrio genomes to conserved orthologs of related genomes (Bordetella andXanthomonas, respectively) that lack secondary replicons (98). Overall, these data suggest that each replicon in a multipartite genome may experience different rates of evolution and unique types of evolutionary pressures and that these differences are at least partially independent of the differences in the gene content of each replicon.

Global Replicon Functional Biases

As functional analyses of multipartite genomes have focused on individual species, it is unclear whether the types of functions showing a biased distribution will vary between phylogenetically distant taxa. Therefore, a global COG analysis was performed (see the supplemental material). All genes for each replicon class from a single representative genome of 1,708 species (1,708 chromosomes, 139 chromids, 99 megaplasmids, and 1,114 plasmids) were pooled, regardless of whether the genome was multipartite, and COG analyses were performed. Indeed, several global biases were evident, as summarized inTable 2, consistent with each replicon class having specific functions enriched regardless of phylogeny.

Not surprisingly, core functions were enriched on chromosomes, such as COG classes J, A, and Z. Chromids also generally appeared to be enriched in some core functions compared to megaplasmids, although only the differences in COG class J (translation) and class M (cell wall/membrane/envelope biogenesis) were statistically significant. There was a large overlap in the functional groups enriched in chromids and megaplasmids, although the extents of enrichment often differed (Table 2). COG classes I, Q, and W were similarly enriched on chromids and megaplasmids, while a small but statistically significant difference in COG class K (transcription) was observed. The enrichment in COG class K likely represents a larger number of transcription factors present on these replicons allowing gene regulation in response to numerous environmental signals (e.g., carbon availability). The transport and metabolism of a few types of compounds (COG classes E, G, and P) as well as motility and signal transduction (COG classes N and T), both of which may be related to movement in response to external stimuli, were primarily enriched on chromids and less so, if at all, on megaplasmids. No classes were enriched specifically on megaplasmids. Plasmids were enriched in replication-related functions (COG classes D and L) and COG class U, which may be related to the replication and conjugal transfer of the plasmid and to resistance to toxic compounds.

In toto, the global functional analysis revealed that many functional categories of genes are universally enriched on secondary replicons, consistent with secondary replicons playing a conserved role in the biology of these organisms. It was also notable that functional biases could be detected between each of chromids, megaplasmids, and plasmids, supporting that these classifications are biologically relevant.

Distribution of Transposable Elements

As a proxy to examine the prevalence of transposable elements on each type of replicon, the RefSeq protein fasta files for each of the 1,708 bacterial genomes were searched for the term “transposase” (see the supplemental material). Approximately 3.1% of the chromosomal genes were annotated as a transposase, compared to 4.3% of putative chromid genes, 6.3% of megaplasmid genes, and 21.6% of plasmid genes, and all differences between replicon classes were statistically significant (Table 2). Thus, the pattern of prevalence of transposable elements on each replicon appeared to mimic that of the genomic signatures of these replicons (Fig. 5); i.e., putative chromids appeared most like chromosomes, followed by megaplasmids, with plasmids being very different from the others. Given that the gain of insertion elements is generally deleterious (138–140), perhaps the biases in transposase prevalence reflect differences in the expendabilities of genes on each type of replicon and differences in purifying selection (98).

TABLE 3.

Increased Genome Size

It has been suggested that multipartite genomes allow for further genome expansion once the chromosome has reached its maximal size (84). In support of this, it was noted that as of 2010, the mean total size of genomes lacking a chromid was 3.38 Mb (SD, 1.81 Mb), whereas the mean size of genomes with a chromid was 5.73 Mb (SD, 1.66 Mb) (11). In contrast, it was pointed out that some small genomes, like that ofBrucella melitensis, are multipartite, whereas some large genomes have a single chromosome, such as the 9-Mb chromosome ofMyxococcus xanthus (31). When 1,708 representative genomes from the NCBI genome database were examined, the mean and median total genome sizes of species containing a putative chromid/megaplasmid were ∼3.67 and ∼3.41 Mb, respectively, whereas the mean and median genome sizes for species with a megaplasmid and/or a putative chromid were ∼5.72 and ∼5.56 Mb, respectively (Fig. 6A). The difference in genome sizes between these two groups can be associated primarily with the secondary replicon, as there was little difference in the mean and median chromosome sizes (Fig. 6B). Hence, it is clear that multipartite genomes are, on average, larger than genomes lacking megaplasmids and chromids. That said, multipartite genomes are not a prerequisite for a large genome, and in fact, fewer than one-third of the genomes with a size of >6 Mb are multipartite (Fig. 6A), while none of the 26 largest genomes, and only 3 of the top 50, are multipartite. Thus, it seems unlikely that the multipartite genome organization evolved simply to allow increased gene accumulation, as the majority of large genomes are not multipartite. Additionally, causality has not been demonstrated; i.e., it has not been established whether genomes are multipartite to allow increased size or whether the increased size is a consequence of having chromids/megaplasmids.

Increased Rate of Bacterial Growth

A second consideration is that the multipartite genome organization may allow faster bacterial division by decreasing the time required to replicate the genome, as each replicon can replicate concurrently (55,65). Indeed, some of the fastest-replicating species, such as those in the generaVibrio, have a multipartite genome (31), and the “fast-growing” rhizobia contain chromids, whereas the “slow-growing” rhizobia do not (161). However, multipartite genomes can be found in many species with relatively long generation times (162), such asR. sphaeroides (31), and multipartite genomes are certainly not a requirement for fast growth. For example, nonmultipartiteClostridium perfringens strains can have generation times as low as 7 min (163). Moreover, an analysis of 214 genomes failed to detect a correlation between genome size and minimal generation time (164). If the emergence and maintenance of a multipartite genome are driven by selective pressure to reduce the time required to replicate the genome, coresident chromosomes and chromids should be equally sized. However, this is not observed (162). Chromids are always smaller than the chromosome (11), and although there is a large range of disparity in the sizes of chromids versus chromosomes, chromids are on average less than half the size of the coresident chromosome (Fig. 7). Overall, while the multipartite genome may impart an ability to replicate the genome more rapidly, the data on the whole do not support this as a driving force for the evolution of the multipartite genome. It may, however, help promote the maintenance of this genome organization in fast-replicating species once it has formed.

Coordinated Gene Regulation

A third hypothesis states that the division of genes between multiple replicons facilitates their coordinated regulation. This could be accomplished through the modulation of gene dosage. Initiation of the replication of the chromosome ofVibrio species occurs prior to that of the chromid, resulting in a higher average gene dosage, and, thus, transcription, of chromosomal genes (16). However, this effect is expected to be limited to fast-growing bacteria (162), and indeed, evidence suggests that there is no gene dosage bias among the three replicons of theS. meliloti genome (165). Recently, a related suggestion was put forth, stating that the localization of genes to different replicons may facilitate their correct subcellular positioning (137). Although an exciting possibility, experimental support for this hypothesis is currently lacking.

Alternatively, the grouping of genes together on individual replicons may facilitate their coordinated regulation by transcription factors. This hypothesis has merit, considering that the transcription machinery is not equally distributed throughout the cell (166). Supporting this, anin silico regulon analysis of 41S. meliloti transcription factors indicated a bias for transcription factors to regulate genes on the same replicon (147). Furthermore, multiple transcriptomic studies have observed that particular replicons are enriched in differentially regulated genes during niche adaptation. Comparison of theV. cholerae transcriptome under laboratory growth conditions to that under intestinal growth conditions illustrated that many more chromid genes are expressed in the intestine than under laboratory conditions (167). In a comparative transcriptomics study ofB. cenocepacia J2315, the number of expressed genes during soil colonization was biased toward the larger chromid, whereas the number of genes expressed underin vitro cystic fibrosis conditions was biased toward the chromosome (168). Additionally, transcriptional differences betweenB. cenocepacia strain J2315 and a related strain were biased toward the smaller chromid of J2315 (168). The pRL8 replicon ofR. leguminosarum is enriched for genes upregulated during growth in the pea rhizosphere (169), and similarly, secondary replicons ofRhizobium phaseoli are enriched in genes upregulated during rhizosphere colonization (170). During symbiosis with legumes, genes downregulated inS. meliloti are overrepresented on the chromosome, whereas upregulated genes are overrepresented on the megaplasmid (151,171). These studies clearly demonstrate that replicon-specific patterns of gene expression can be observed. However, causation has not been demonstrated, and thus, it is unclear if the multipartite genome evolved to facilitate coordinated gene regulation or whether these transcriptional patterns are a by-product of functionally related genes being colocalized to different replicons.

Adaptation to Novel Niches

To account for many of the observations related to multipartite genomes, including replicon-specific regulation patterns, functional biases, and distinct evolutionary patterns, it was suggested previously that individual replicons within multipartite genomes contribute to adaptation to unique environments (22,121,152). It is our opinion that the primary advantage of the multipartite genome architecture is to mediate adaptation to novel niches. In this section, we present a generalized evolutionary model based on niche adaptation that attempts to explain the observations reported throughout this review. This model is summarized inFig. 8 and is an extension of ideas proposed previously (22,121,152).

The main tenant of the proposed model is that secondary replicons act as specialized entities for adaptation to unique environments. This implicitly states that the primary chromosome is sufficient for growth and survival in nonspecialized soil or aquatic environments. Computational analyses suggested that the ancestor of theAlphaproteobacteria harbored ∼3,300 genes, with lower and upper bounds of 3,000 and 5,000 proteins, respectively (172). The median bacterial chromosome size of 3.46 Mb is therefore likely similar to that of the ancestral alphaproteobacterial genome. Additionally, anS. meliloti strain with a genome reduced by 45% through the removal of the chromid and megaplasmid was capable of growing quite well in bulk soil mesocosms (152). Together, these observations are consistent with the chromosomal gene repertoire being sufficient for high fitness in nonspecialized environments.

Accepting that the plasmid hypothesis explains the formation of chromids, then all multipartite genomes must originate from the acquisition of a megaplasmid through horizontal gene transfer (HGT). Secondary replicons often carry the key determinants required for initial colonization of new environments, such as symbiotic and virulence factors, although they generally account for only a small portion of large secondary replicons (161,173). The high level of gene variations of megaplasmids, as discussed above (27,78,98,122–125,127,128), suggests that they are undergoing rapid gene loss and gene gain through HGT, as was observed forS. meliloti (121). Comparative genomics and metabolic modeling studies illustrated that most genes acquired through HGT are involved primarily in adaptation to different environments (174–177), with different genomic regions being responsible for different ecologies (178).

Other computational work illustrated that genome expansion within lineages of theAlphaproteobacteria and the orderRhizobiales was linked to an association with plants and the evolution of symbiosis, respectively (172,179). Those genome expansions involved mainly the acquisition of transcriptional, transport, and metabolic functions (172,179), which are the same functions commonly enriched on secondary replicons (Table 2). Hence, it is reasonable to suggest that the gain of secondary replicons first allows an association with a novel niche, after which the new replicon gains genes associated with environmental adaptation. In particular, we hypothesize that these environments often represent new niches formed due to the emergence of eukaryotic organisms, such as the rhizosphere and the animal gut. However, the association with eukaryotic organisms may reflect biases in the organisms chosen for whole-genome sequencing, and sequencing of more environmental isolates may reveal that secondary replicons may be generally associated with diverse niche colonization independent of eukaryotes.

As chromids are on average twice as large as megaplasmids (Fig. 2), the evolution of a megaplasmid must therefore involve significant gene accumulation. However, it has been argued that most HGT events, including those that are eventually fixed in a population, are initially deleterious (180,181). As a result, most genes acquired through HGT are lost from the genome, as the costs outweigh the benefits (175,180–183). The high variability of megaplasmids is consistent with most genes acquired by these replicons eventually being lost either because they provide no benefit or because the costs are too high. Only genes whose benefits outweigh the costs to the cell are eventually fixed in the population, and as described above, these genes are expected to be associated with adaptation to the new environment. This conclusion is supported by the transcriptional studies summarized above (151,167–171), by anin silico metabolic modeling study that suggested that the metabolic abilities associated with theS. meliloti chromid are rhizosphere specialized (184), and by experimental work showing that the loss of the pATC58 megaplasmid decreases fitness in the plant rhizosphere (153).

As chromids evolve from megaplasmids, chromids could be considered a subclass of megaplasmids, and it might therefore be expected that megaplasmids would be more prevalent than chromids. However, chromids are found in approximately twice as many species as megaplasmids (Fig. 4). This may reflect two divergent fates of megaplasmids. The potentially high costs of megaplasmids (83,113,120,152,155,159), as discussed above, may result in the loss of the megaplasmid in particular environments and may limit the ability of megaplasmids to be successfully maintained following horizontal transfer of the entire replicon. On the other hand, if the megaplasmid provides enough of a benefit to remain in the cell, rapid conversion into a chromid is likely to occur. The observation that megaplasmids are limited to more narrow taxonomic groups than are chromids (Fig. 4) supports that secondary replicons remain an evolutionarily stable component of the genome only if a conversion to a chromid occurs.

Genes recently acquired through HGT undergo rapid evolution (181). As described above, megaplasmids experience high evolutionary rates (22,78,98,121,130), which is possibly reflective of the rapid amelioration of the replicon to reduce the associated costs. This can include modification of genes and regulatory elements, such as through the amelioration of codon usage (185), and promoter modifications to better integrate the genes into existing transcriptional networks (186). At the same time, the gain of essential genes from the chromosome results in the formation of a chromid (41,84). This interreplicon gene flow also contributes to the increased stability of the secondary replicon, in terms of reducing both the rate of gene loss and the loss of the entire replicon (11,51,84,152), and contributes to the integration of the replicon into the cell's core metabolism (187). In this way, environmental adaptation can drive the emergence of a multipartite genome.

The primary observation opposing this hypothesis is that many species lack multipartite genomes, yet they can still show high genetic variability and colonization of multiple niches. Species of the generaSinorhizobium andBradyrhizobium are legume N₂-fixing endosymbionts that also colonize bulk soil and the plant rhizosphere. Both genera also contain large pangenomes, and thus, both genera have high genetic variability (188). However, only species of the genusSinorhizobium have a multipartite genome (18,189). Moreover, the organisms of the genusProchlorococcus have a genome size of just 2,000 genes but have an estimated pangenome size of >58,000 genes (190). Thus, even if the evolution of the multipartite genome is driven by niche adaptation, a multipartite genome is by no means a prerequisite for niche adaptation or genetic variability.

Maintenance of the Multipartite Genome

Even if the evolution of the multipartite genome is driven by environmental adaptation, it remains unclear why the secondary replicon remains an independent unit and why it does not eventually become integrated into the chromosome, as appears to have occurred in a few cases (84). One possibility is that, at least in some species, the potential benefits of divided genomes, as summarized above, may help maintain the multipartite genome structure once it has formed. However, we hypothesize that the multipartite architecture is often an evolutionary relic limited by what came before. Megaplasmids are transferable (110–117). Hence, a gene on a megaplasmid may have higher fitness than a gene on a chromosome due to the increased frequency of HGT mediated through megaplasmid conjugation. We note that chromosomes can also carry mobile elements as well, but as the gain of a chromosomal mobile element can result in the disruption of important chromosomal genes (191,192), plasmid-mediated HGT may be more efficient. Selection for increased HGT may be lost in larger megaplasmids and chromids due to their increased costs; however, integration into the chromosome may still be unfavorable due to the large size of such replicons. The chromosomal origin of replication and terminus region normally, but not always, split chromosomes into two roughly equal halves referred to as replichores (193–195). Genome rearrangements that significantly perturb this equal distribution appear to have a negative impact on fitness and can be selected against (13,196–200), meaning that the integration of a 1.5-Mb chromid into a bacterial chromosome is likely to be unfavorable. Integration may be further selected against if the gene strand bias of the chromosome is not maintained (13,17,194,201–203). Indeed,S. meliloti strains with all three replicons fused together display a fitness decrease (88), and it has been proposed thatV. cholerae strains with both replicons fused together are less fit (89). Hence, the maintenance of the multipartite genome architecture may reflect selection for increased HGT early in its development and selective pressures against chromosome disruptions later during its maintenance.

Enrichment of Environmental Adaptation Genes on Secondary Replicons

If the niche adaptation model outlined above is correct, secondary replicon enlargement occurs as a result of the acquisition of niche-specific genes. However, it is unclear why these genes would be preferentially acquired by the secondary replicon and not equally acquired by the chromosome. It may be that secondary replicons more readily acquire new DNA (84) or because insertions into the chromosome are more likely to disrupt growth-promoting genes than are insertions into a megaplasmid, leading to greater selection against chromosomal insertions (191,192). This is supported by the reduced purifying selection observed on secondary replicons (98) and the higher prevalence of transposable elements (Table 2). Additionally, and perhaps more importantly, megaplasmids can move through conjugation, and thus, genes that integrate into these replicons may have higher fitness, as they can more readily spread horizontally throughout the population. Moreover, the colocalization of related/interacting pathways on the same replicon results in their genomic linkage. If this replicon is transferable, then both pathways will move together, and if this is beneficial to the recipient, evolution may select for the linkage of the pathways.

Fixation of Essential Gene Transfer Events

The key characteristic of a chromid is that it contains core biological functions. One way in which this occurs is through the transfer of genes from the chromosome to the secondary replicon (31,41,83–87). However, it is unclear why such a translocation event would become fixed in the population. Many secondary replicons within theAlphaproteobacteria belong to therepABC family and encode a partitioning system that helps ensure high stability and segregation of the replicon to both daughter cells (204,205). However, segregation is sufficiently imperfect so that the replicon could be lost from the population within a few thousand generations (205). Thus, it may be that the transfer of essential genes results in the stabilization of the replicon that, combined with the loss of replicons without essential genes from the population through genetic drift, results in the fixation of essential genes on the chromid. This could also explain why chromids generally contain only a few essential genes; the first transfer of essential genes would stabilize the replicon, while additional transfers of essential genes would provide little additional advantage (51).

Multipartite Genome Topology

An exciting research direction that has so far remained largely unexplored is the study of the genome topology and DNA physical interactions in species with multipartite genomes through techniques such as chromosome conformation capture methods like Hi-C (206,207) and multicolor fluorescencein situ hybridization (FISH) (208). The sole study examining three-dimensional (3D) genome topology in a multipartite genome was performed withV. cholerae (60). It would be interesting to use these techniques and various model systems to address general questions related to multipartite genome evolution. Potential research topics include examining whether there are interreplicon chromatin interactions and if such interactions are correlated with regions of increased interreplicon transcriptional interactions or gene flow. It would also be fascinating to examine whether the replicons are intermingled or if each replicon occupies a distinct and stable location in the cell, similar to how each chromosome occupies an unique nuclear territory in eukaryotic organisms (209), and if the removal of one or more replicons impacts the localization of the remaining replicons. In the case ofV. cholerae, the data were consistent with the chromosome and chromid having very different organizations, with each replicon occupying distinct locations in the cell, and with there being direct physical interactions between the two replicons (60). However, as the data were analyzed with respect to one specific question, additional work is required to address general questions related to multipartite genome topology and the generalizability of the observations.

It is also worth noting that many secondary replicons encode nucleoid-associated proteins (NAPs) that can influence the topology of themselves as well as the chromosome, influence chromosomal transcription, and impact host fitness. This topic was recently reviewed by Shintani et al. (210), and we refer readers to that article for an in-depth discussion of this topic. Interestingly, they found that only a low percentage of plasmids carried NAPs, while over one-third of megaplasmids/chromids encoded NAPs (210); however, as chromids and megaplasmids were not differentiated, the relative frequencies of these two classes of replicons cannot be compared.

Loss of Conjugal Properties

It is also unclear why megaplasmids, but not chromids, appear to transfer via conjugation in nature (110–117) despite at least some replicons of both classes retaining conjugal properties in laboratory settings (11,118,119). This may simply reflect a high cost of acquisition. It has been argued that most genes acquired through HGT are maintained due to their low costs as opposed to their benefits (180), and thus, the costs associated with HGT (180,211,212) may mean that chromids are rapidly lost if they are transferred to a new cell. Additionally, core “information” genes are less likely to be successfully transferred via HGT (213,214), and such genes are found on chromids but not megaplasmids. Thus, the large size of chromids and the types of genes that they carry may lead to their rapid loss in the event that they are horizontally transferred.

George C. diCenzo began work on this topic in 2010 as a Molecular Biology and Genetics undergraduate student. He obtained his B.Sc. in 2012 from McMaster University and defended his Ph.D. at the same institution in late 2016 under the supervision of Turlough Finan in the Department of Biology. He is currently a postdoctoral fellow at the University of Florence, Italy, under the supervision of Alessio Mengoni. During his Ph.D. work, he combined genome reduction approaches with molecular genetic, genomics, system-level, andin silico analyses to study the contributions of each replicon ofSinorhizobium meliloti to the biology of the free-living and symbiotic forms of the bacterium. His current work is focused on usingin silico genome-scale metabolic network reconstruction and flux balance analysis, together with genetic studies, to characterize the genetics and metabolism ofS. meliloti during symbiosis with alfalfa in order to develop strategies for manipulating this interaction.

Turlough M. Finan is a professor of Biology at McMaster University, Hamilton, Ontario, Canada. His interest in secondary replicons originated during his B.Sc. and M.Sc. in Microbiology under the supervision of Kieran Dunican, National University, Galway, Ireland. He obtained his Ph.D. in 1981 under the supervision of Carl Jordan, Microbiology Department, University of Guelph, Canada. Following a year at the Connaught Research Institute studying the neutralizing antigens of poliovirus, he then performed postdoctoral studies onSinorhizobium under the supervision of Ethan Signer in the Department of Biology, Massachusetts Institute of Technology. He teaches undergraduate and graduate courses on Microbiology, Molecular Genetics, and Environmental Microbiology. For over 30 years, he has examined genomic and metabolic aspects of the interaction betweenSinorhizobium meliloti and alfalfa. Initial studies focused on the detection and analysis of symbiotic loci and expanded to the biology of the 1.7-Mb pSymB replicon, including carbon and phosphate metabolism.

• 1.Cairns J.1963. The bacterial chromosome and its manner of replication as seen by autoradiography. J Mol Biol6:208–213. doi: 10.1016/S0022-2836(63)80070-4. [DOI] [PubMed] [Google Scholar]
• 2.Wake RG.1973. Circularity of theBacillus subtilis chromosome and further studies on its bidirectional replication. J Mol Biol77:569–575. doi: 10.1016/0022-2836(73)90223-4. [DOI] [PubMed] [Google Scholar]
• 3.Bode HR, Morowitz HJ. 1967. Size and structure of theMycoplasma hominis H39 chromosome. J Mol Biol23:191–199. doi: 10.1016/S0022-2836(67)80026-3. [DOI] [PubMed] [Google Scholar]
• 4.Hayakawa T, Tanaka T, Sakaguchi K, Ōtake N, Yonehara H. 1979. A linear plasmid-like DNA inStreptomyces sp. producing lankacidin group antibiotics. J Gen Appl Microbiol25:255–260. doi: 10.2323/jgam.25.255. [DOI] [Google Scholar]
• 5.Baril C, Richaud C, Baranton G, Saint Girons I. 1989. Linear chromosome ofBorrelia burgdorferi. Res Microbiol140:507–516. doi: 10.1016/0923-2508(89)90083-1. [DOI] [PubMed] [Google Scholar]
• 6.Ferdows MS, Barbour AG. 1989. Megabase-sized linear DNA in the bacteriumBorrelia burgdorferi, the Lyme disease agent. Proc Natl Acad Sci U S A86:5969–5973. doi: 10.1073/pnas.86.15.5969. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 7.Rosenberg C, Boistard P, Dénarié J, Casse-Delbart F. 1981. Genes controlling early and late functions in symbiosis are located on a megaplasmid inRhizobium meliloti. Mol Gen Genet184:326–333. [DOI] [PubMed] [Google Scholar]
• 8.Bonhoeffer F, Messer W. 1969. Replication of the bacterial chromosome. Annu Rev Genet3:233–246. doi: 10.1146/annurev.ge.03.120169.001313. [DOI] [Google Scholar]
• 9.Suwanto A, Kaplan S. 1989. Physical and genetic mapping of theRhodobacter sphaeroides 2.4.1 genome: presence of two unique circular chromosomes. J Bacteriol171:5850–5859. doi: 10.1128/jb.171.11.5850-5859.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 10.Anda M, Ohtsubo Y, Okubo T, Sugawara M, Nagata Y, Tsuda M, Minamisawa K, Mitsui H. 2015. Bacterial clade with the ribosomal RNA operon on a small plasmid rather than the chromosome. Proc Natl Acad Sci U S A112:14343–14347. doi: 10.1073/pnas.1514326112. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 11.Harrison PW, Lower RPJ, Kim NKD, Young JPW. 2010. Introducing the bacterial “chromid”: not a chromosome, not a plasmid. Trends Microbiol18:141–148. doi: 10.1016/j.tim.2009.12.010. [DOI] [PubMed] [Google Scholar]
• 12.Junier I.2014. Conserved patterns in bacterial genomes: a conundrum physically tailored by evolutionary tinkering. Comp Biol Chem53:125–133. doi: 10.1016/j.compbiolchem.2014.08.017. [DOI] [PubMed] [Google Scholar]
• 13.Rocha EPC.2008. The organization of the bacterial genome. Annu Rev Genet42:211–233. doi: 10.1146/annurev.genet.42.110807.091653. [DOI] [PubMed] [Google Scholar]
• 14.Lawrence JG.2003. Genome organization: selection, selfishness, and serendipity. Annu Rev Microbiol57:419–440. doi: 10.1146/annurev.micro.57.030502.090816. [DOI] [PubMed] [Google Scholar]
• 15.Bryant JA, Sellars LE, Busby SJW, Lee DJ. 2015. Chromosome position effects on gene expression inEscherichia coli K-12. Nucleic Acids Res42:11383–11392. doi: 10.1093/nar/gku828. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 16.Dryselius R, Izutsu K, Honda T, Iida T. 2008. Differential replication dynamics for large and smallVibrio chromosomes affect gene dosage, expression and location. BMC Genomics9:559. doi: 10.1186/1471-2164-9-559. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 17.Rocha EPC.2004. The replication-related organization of bacterial genomes. Microbiology150:1609–1627. doi: 10.1099/mic.0.26974-0. [DOI] [PubMed] [Google Scholar]
• 18.Galibert F, Finan TM, Long SR, Pühler A, Abola AP, Ampe F, Barloy-Hubler F, Barnett MJ, Becker A, Boistard P, Bothe G, Boutry M, Bowser L, Buhrmester J, Cadieu E, Capela D, Chain P, Cowie A, Davis RW, Dréano S, Federspiel NA, Fisher RF, Gloux S, Godrie T, Goffeau A, Golding B, Gouzy J, Gurjal M, Hernández-Lucas I, Hong A, Huizar L, Hyman RW, Jones T, Kahn D, Kahn ML, Kalman S, Keating DH, Kiss E, Komp C, Lelaure V, Masuy D, Palm C, Peck MC, Pohl T, Portetelle D, Purnelle B, Ramsperger U, Surzycki R, Thébault P, Vandenbol M, et al. 2001. The composite genome of the legume symbiontSinorhizobium meliloti. Science293:668–672. doi: 10.1126/science.1060966. [DOI] [PubMed] [Google Scholar]
• 19.Goodner B, Hinkle G, Gattung S, Miller N, Blanchard M, Qurollo B, Goldman BS, Cao Y, Askenazi M, Halling C, Mullin L, Houmiel K, Gordon J, Vaudin M, Iartchouk O, Epp A, Liu F, Wollam C, Allinger M, Doughty D, Scott C, Lappas C, Markelz B, Flanagan C, Crowell C, Gurson J, Lomo C, Sear C, Strub G, Cielo C, Slater S. 2001. Genome sequence of the plant pathogen and biotechnology agentAgrobacterium tumefaciens C58. Science294:2323–2328. doi: 10.1126/science.1066803. [DOI] [PubMed] [Google Scholar]
• 20.DelVecchio VG, Kapatral V, Redkar RJ, Patra G, Mujer C, Los T, Ivanova N, Anderson I, Bhattacharyya A, Lykidis A, Reznik G, Jablonski L, Larsen N, D'Souza M, Bernal A, Mazur M, Goltsman E, Selkov E, Elzer PH, Hagius S, O'Callaghan D, Letesson J-J, Haselkorn R, Kyrpides N, Overbeek R. 2002. The genome sequence of the facultative intracellular pathogenBrucella melitensis. Proc Natl Acad Sci U S A99:443–448. doi: 10.1073/pnas.221575398. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 21.Heidelberg JF, Eisen JA, Nelson WC, Clayton RA, Gwinn ML, Dodson RJ, Haft DH, Hickey EK, Peterson JD, Umayam L, Gill SR, Nelson KE, Read TD, Tettelin H, Richardson D, Ermolaeva MD, Vamathevan J, Bass S, Qin H, Dragoi I, Sellers P, McDonald L, Utterback T, Fleishmann RD, Nierman WC, White O, Salzberg SL, Smith HO, Colwell RR, Mekalanos JJ, Venter JC, Fraser CM. 2000. DNA sequence of both chromosomes of the cholera pathogenVibrio cholerae. Nature406:477–483. doi: 10.1038/35020000. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Chain PSG, Denef VJ, Konstantinidis KT, Vergez LM, Agulló L, Reyes VL, Hauser L, Córdova M, Gómez L, González M, Land M, Lao V, Larimer F, LiPuma JJ, Mahenthiralingam E, Malfatti SA, Marx CJ, Parnell JJ, Ramette A, Richardson P, Seeger M, Smith D, Spilker T, Sul WJ, Tsoi TV, Ulrich LE, Zhulin IB, Tiedje JM. 2006.Burkholderia xenovorans LB400 harbors a multi-replicon, 9.73-Mbp genome shaped for versatility. Proc Natl Acad Sci U S A103:15280–15287. doi: 10.1073/pnas.0606924103. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 23.Moreno E.1998. Genome evolution within the alphaProteobacteria: why do some bacteria not possess plasmids and others exhibit more than one different chromosome?FEMS Microbiol Rev22:255–275. doi: 10.1111/j.1574-6976.1998.tb00370.x. [DOI] [PubMed] [Google Scholar]
• 24.Prozorov AA.2008. Additional chromosomes in bacteria: properties and origin. Mikrobiologiia77:437–447. [PubMed] [Google Scholar]
• 25.Schwartz E. (ed). 2009. Microbial megaplasmids. Springer, Berlin, Germany. [Google Scholar]
• 26.Choudhary M, Cho H, Bavishi A, Trahan C, Myagmarjav B-E. 2012. Evolution of multipartite genomes in prokaryotes, p 301–323.InPontarotti P. (ed), Evolutionary biology: mechanisms and trends. Springer, Berlin, Germany. [Google Scholar]
• 27.Van Houdt R, Mergeay M. 2012. Plasmids as secondary chromosomes, p 1–4.InWells RD, Bond JS, Klinman J, Masters BSS, Bell E (ed), Molecular life sciences: an encyclopedic reference. Springer, Berlin, Germany. [Google Scholar]
• 28.Val M-E, Soler-Bistué A, Bland MJ, Mazel D. 2014. Management of multipartite genomes: theVibrio cholerae model. Curr Opin Microbiol22:120–126. doi: 10.1016/j.mib.2014.10.003. [DOI] [PubMed] [Google Scholar]
• 29.Ramachandran R, Jha J, Paulsson J, Chattoraj D. 2017. Random versus cell cycle-regulated replication initiation in bacteria: insights from studyingVibrio cholerae chromosome 2. Microbiol Mol Biol Rev81:e00033-16. doi: 10.1128/MMBR.00033-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 30.Ramachandran R, Jha J, Chattoraj DK. 2015. Chromosome segregation inVibrio cholerae. J Mol Microbiol Biotechnol24:360–370. doi: 10.1159/000368853. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 31.Egan ES, Fogel MA, Waldor MK. 2005. Divided genomes: negotiating the cell cycle in prokaryotes with multiple chromosomes. Mol Microbiol56:1129–1138. doi: 10.1111/j.1365-2958.2005.04622.x. [DOI] [PubMed] [Google Scholar]
• 32.Barry ER, Bell SD. 2006. DNA replication in the archaea. Microbiol Mol Biol Rev70:876–887. doi: 10.1128/MMBR.00029-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 33.Wu Z, Liu J, Yang H, Xiang H. 2014. DNA replication origins in archaea. Front Microbiol5:179. doi: 10.3389/fmicb.2014.00179. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 34.Oresnik IJ, Liu SL, Yost CK, Hynes MF. 2000. Megaplasmid pRme2011a ofSinorhizobium meliloti is not required for viability. J Bacteriol182:3582–3586. doi: 10.1128/JB.182.12.3582-3586.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 35.Agnoli K, Schwager S, Uehlinger S, Vergunst A, Viteri DF, Nguyen DT, Sokol PA, Carlier A, Eberl L. 2012. Exposing the third chromosome ofBurkholderia cepacia complex strains as a virulence plasmid. Mol Microbiol83:362–378. doi: 10.1111/j.1365-2958.2011.07937.x. [DOI] [PubMed] [Google Scholar]
• 36.Orlova N, Gerding M, Ivashkiv O, Olinares PDB, Chait BT, Waldor MK, Jeruzalmi D. 2017. The replication initiator of the cholera pathogen's second chromosome shows structural similarity to plasmid initiators. Nucleic Acids Res45:3724–3737. doi: 10.1093/nar/gkw1288. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 37.Venkova-Canova T, Chattoraj DK. 2011. Transition from a plasmid to a chromosomal mode of replication entails additional regulators. Proc Natl Acad Sci U S A108:6199–6204. doi: 10.1073/pnas.1013244108. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Dziewit L, Czarnecki J, Wibberg D, Radlinska M, Mrozek P, Szymczak M, Schlüter A, Pühler A, Bartosik D. 2014. Architecture and functions of a multipartite genome of the methylotrophic bacteriumParacoccus aminophilus JCM 7686, containing primary and secondary chromids. BMC Genomics15:124. doi: 10.1186/1471-2164-15-124. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 39.Finan TM, Weidner S, Wong K, Buhrmester J, Chain P, Vorhölter FJ, Hernández-Lucas I, Becker A, Cowie A, Gouzy J, Golding B, Puhler A. 2001. The complete sequence of the 1,683-kb pSymB megaplasmid from the N₂-fixing endosymbiontSinorhizobium meliloti. Proc Natl Acad Sci U S A98:9889–9894. doi: 10.1073/pnas.161294698. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 40.Cheng J, Sibley CD, Zaheer R, Finan TM. 2007. ASinorhizobium meliloti minE mutant has an altered morphology and exhibits defects in legume symbiosis. Microbiology153:375–387. doi: 10.1099/mic.0.2006/001362-0. [DOI] [PubMed] [Google Scholar]
• 41.diCenzo G, Milunovic B, Cheng J, Finan TM. 2013. The tRNA^arg gene andengA are essential genes on the 1.7-Mb pSymB megaplasmid ofSinorhizobium meliloti and were translocated together from the chromosome in an ancestral strain. J Bacteriol195:202–212. doi: 10.1128/JB.01758-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 42.Hayes F.2003. Toxins-antitoxins: plasmid maintenance, programmed cell death, and cell cycle arrest. Science301:1496–1499. doi: 10.1126/science.1088157. [DOI] [PubMed] [Google Scholar]
• 43.Van Melderen L, Saavedra De Bast M. 2009. Bacterial toxin-antitoxin systems: more than selfish entities?PLoS Genet5:e1000437. doi: 10.1371/journal.pgen.1000437. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 44.Gerdes K, Christensen SK, Løbner-Olesen A. 2005. Prokaryotic toxin-antitoxin stress response loci. Nat Rev Microbiol3:371–382. doi: 10.1038/nrmicro1147. [DOI] [PubMed] [Google Scholar]
• 45.MacLellan SR, Smallbone LA, Sibley CD, Finan TM. 2005. The expression of a novel antisense gene mediates incompatibility within the largerepABC family of α-proteobacterial plasmids. Mol Microbiol55:611–623. doi: 10.1111/j.1365-2958.2004.04412.x. [DOI] [PubMed] [Google Scholar]
• 46.Milunovic B, diCenzo GC, Morton RA, Finan TM. 2014. Cell growth inhibition upon deletion of four toxin-antitoxin loci from the megaplasmids ofSinorhizobium meliloti. J Bacteriol196:811–824. doi: 10.1128/JB.01104-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 47.O'Leary NA, Wright MW, Brister JR, Ciufo S, Haddad D, McVeigh R, Rajput B, Robbertse B, Smith-White B, Ako-Adjei D, Astashyn A, Badretdin A, Bao Y, Blinkova O, Brover V, Chetvernin V, Choi J, Cox E, Ermolaeva O, Farrell CM, Goldfarb T, Gupta T, Haft D, Hatcher E, Hlavina W, Joardar VS, Kodali VK, Li W, Maglott D, Masterson P, McGarvey KM, Murphy MR, O'Neill K, Pujar S, Rangwala SH, Rausch D, Riddick LD, Schoch C, Shkeda A, Storz SS, Sun H, Thibaud-Nissen F, Tolstoy I, Tully RE, Vatsan AR, Wallin C, Webb D, Wu W, Landrum MJ, Kimchi A, et al. 2016. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res44:D733–D745. doi: 10.1093/nar/gkv1189. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 48.Suzuki K, Iwata K, Yoshida K. 2001. Genome analysis ofAgrobacterium tumefaciens: construction of physical maps for linear and circular chromosomal DNAs, determination of copy number ratio and mapping of chromosomal virulence genes. DNA Res8:141–152. doi: 10.1093/dnares/8.4.141. [DOI] [PubMed] [Google Scholar]
• 49.Sibley CD, MacLellan SR, Finan T. 2006. TheSinorhizobium meliloti chromosomal origin of replication. Microbiology152:443–455. doi: 10.1099/mic.0.28455-0. [DOI] [PubMed] [Google Scholar]
• 50.Döhlemann J, Wagner M, Happel C, Carrillo M, Sobetzko P, Erb TJ, Thanbichler M, Becker A. 2017. A family of single copyrepABC-type shuttle vectors stably maintained in the alpha-proteobacteriumSinorhizobium meliloti. ACS Synth Biol6:968–984. doi: 10.1021/acssynbio.6b00320. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 51.Du W-L, Dubarry N, Passot FM, Kamgoué A, Murray H, Lane D, Pasta F. 2016. Orderly replication and segregation of the four replicons ofBurkholderia cenocepacia J2315. PLoS Genet12:e1006172. doi: 10.1371/journal.pgen.1006172. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 52.Li H, Angelov A, Pham VTT, Leis B, Liebl W. 2015. Characterization of chromosomal and megaplasmid partitioning loci inThermus thermophilus HB27. BMC Genomics16:317. doi: 10.1186/s12864-015-1523-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 53.Hartmann EM, Badalamenti JP, Krajmalnik-Brown R, Halden RU. 2012. Quantitative PCR for tracking the megaplasmid-borne biodegradation potential of a model sphingomonad. Appl Environ Microbiol78:4493–4496. doi: 10.1128/AEM.00715-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 54.Srivastava P, Chattoraj DK. 2007. Selective chromosome amplification inVibrio cholerae. Mol Microbiol66:1016–1028. doi: 10.1111/j.1365-2958.2007.05973.x. [DOI] [PubMed] [Google Scholar]
• 55.Rasmussen T, Jensen RB, Skovgaard O. 2007. The two chromosomes ofVibrio cholerae are initiated at different time points in the cell cycle. EMBO J26:3124–3131. doi: 10.1038/sj.emboj.7601747. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 56.De Nisco NJ, Abo RP, Wu CM, Penterman J, Walker GC. 2014. Global analysis of cell cycle gene expression of the legume symbiontSinorhizobium meliloti. Proc Natl Acad Sci U S A111:3217–3224. doi: 10.1073/pnas.1400421111. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 57.Duigou S, Knudsen KG, Skovgaard O, Egan ES, Lobner-Olesen A, Waldor MK. 2006. Independent control of replication initiation of the twoVibrio cholerae chromosomes by DnaA and RctB. J Bacteriol188:6419–6424. doi: 10.1128/JB.00565-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 58.Egan ES, Waldor MK. 2003. Distinct replication requirements for the twoVibrio cholerae chromosomes. Cell114:521–530. doi: 10.1016/S0092-8674(03)00611-1. [DOI] [PubMed] [Google Scholar]
• 59.Baek JH, Chattoraj DK. 2014. Chromosome I controls chromosome II replication inVibrio cholerae. PLoS Genet10:e1004184. doi: 10.1371/journal.pgen.1004184. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 60.Val M-E, Marbouty M, de Lemos Martins F, Kennedy SP, Kemble H, Bland MJ, Possoz C, Koszul R, Skovgaard O, Mazel D. 2016. A checkpoint control orchestrates the replication of the two chromosomes ofVibrio cholerae. Sci Adv2:e1501914. doi: 10.1126/sciadv.1501914. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 61.Srivastava P, Fekete RA, Chattoraj DK. 2006. Segregation of the replication terminus of the twoVibrio cholerae chromosomes. J Bacteriol188:1060–1070. doi: 10.1128/JB.188.3.1060-1070.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 62.Demarre G, Galli E, Muresan L, Paly E, David A, Possoz C, Barre F-X. 2014. Differential management of the replication terminus regions of the twoVibrio cholerae chromosomes during cell division. PLoS Genet10:e1004557. doi: 10.1371/journal.pgen.1004557. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 63.Galli E, Poidevin M, Le Bars R, Desfontaines J-M, Muresan L, Paly E, Yamaichi Y, Barre F-X. 2016. Cell division licensing in the multi-chromosomalVibrio cholerae bacterium. Nat Microbiol1:16094. doi: 10.1038/nmicrobiol.2016.94. [DOI] [PubMed] [Google Scholar]
• 64.Kadoya R, Chattoraj DK. 2012. Insensitivity of chromosome I and the cell cycle to blockage of replication and segregation ofVibrio cholerae chromosome II. mBio3:e00067-12. doi: 10.1128/mBio.00067-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 65.Frage B, Döhlemann J, Robledo M, Lucena D, Sobetzko P, Graumann PL, Becker A. 2016. Spatiotemporal choreography of chromosome and megaplasmids in theSinorhizobium meliloti cell cycle. Mol Microbiol100:808–823. doi: 10.1111/mmi.13351. [DOI] [PubMed] [Google Scholar]
• 66.Dubarry N, Pasta F, Lane D. 2006. ParABS systems of the four replicons ofBurkholderia cenocepacia: new chromosome centromeres confer partition specificity. J Bacteriol188:1489–1496. doi: 10.1128/JB.188.4.1489-1496.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 67.Passot FM, Calderon V, Fichant G, Lane D, Pasta F. 2012. Centromere binding and evolution of chromosomal partition systems in theBurkholderiales. J Bacteriol194:3426–3436. doi: 10.1128/JB.00041-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 68.Żebracki K, Koper P, Marczak M, Skorupska A, Mazur A. 2015. Plasmid-encoded RepA proteins specifically autorepress individualrepABC operons in the multipartiteRhizobium leguminosarum bv.trifolii genome. PLoS One10:e0131907. doi: 10.1371/journal.pone.0131907. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 69.Fricke WF, Kusian B, Bowien B. 2009. The genome organization ofRalstonia eutropha strain H16 and related species of theBurkholderiaceae. J Mol Microbiol Biotechnol16:124–135. doi: 10.1159/000142899. [DOI] [PubMed] [Google Scholar]
• 70.Jumas-Bilak E, Michaux-Charachon S, Bourg G, O'Callaghan D, Ramuz M. 1998. Differences in chromosome number and genome rearrangements in the genusBrucella. Mol Microbiol27:99–106. doi: 10.1046/j.1365-2958.1998.00661.x. [DOI] [PubMed] [Google Scholar]
• 71.Choudhary M, Mackenzie C, Nereng K, Sodergren E, Weinstock GM, Kaplan S. 1997. Low-resolution sequencing ofRhodobacter sphaeroides 2.A.1T: chromosome II is a true chromosome. Microbiology143:3085–3099. doi: 10.1099/00221287-143-10-3085. [DOI] [PubMed] [Google Scholar]
• 72.Liang X, Baek C-H, Katzen F. 2013.Escherichia coli with two linear chromosomes. ACS Synth Biol2:734–740. doi: 10.1021/sb400079u. [DOI] [PubMed] [Google Scholar]
• 73.Itaya M, Tanaka T. 1997. Experimental surgery to create subgenomes ofBacillus subtilis 168. Proc Natl Acad Sci U S A94:5378–5382. doi: 10.1073/pnas.94.10.5378. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 74.Choudhary M, Fu Y-X, Mackenzie C, Kaplan S. 2004. DNA sequence duplication inRhodobacter sphaeroides 2.4.1: evidence of an ancient partnership between chromosomes I and II. J Bacteriol186:2019–2027. doi: 10.1128/JB.186.7.2019-2027.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 75.Choudhary M, Mackenzie C, Nereng KS, Sodergren E, Weinstock GM, Kaplan S. 1994. Multiple chromosomes in bacteria: structure and function of chromosome II ofRhodobacter sphaeroides 2.4.1T. J Bacteriol176:7694–7702. doi: 10.1128/jb.176.24.7694-7702.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 76.Mackenzie C, Choudhary M, Larimer FW, Predki PF, Stilwagen S, Armitage JP, Barber RD, Donohue TJ, Hosler JP, Newman JE, Shapleigh JP, Sockett RE, Zeilstra-Ryalls J, Kaplan S. 2001. The home stretch, a first analysis of the nearly completed genome ofRhodobacter sphaeroides 2.4.1. Photosynth Res70:19–41. doi: 10.1023/A:1013831823701. [DOI] [PubMed] [Google Scholar]
• 77.Kontur WS, Schackwitz WS, Ivanova N, Martin J, Labutti K, Deshpande S, Tice HN, Pennacchio C, Sodergren E, Weinstock GM, Noguera DR, Donohue TJ. 2012. Revised sequence and annotation of the Rhodobacter sphaeroides 2.4.1 genome. J Bacteriol194:7016–7017. doi: 10.1128/JB.01214-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 78.Choudhary M, Zanhua X, Fu YX, Kaplan S. 2007. Genome analyses of three strains ofRhodobacter sphaeroides: evidence of rapid evolution of chromosome II. J Bacteriol189:1914–1921. doi: 10.1128/JB.01498-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 79.Nereng KS, Kaplan S. 1999. Genomic complexity among strains of the facultative photoheterotrophic bacteriumRhodobacter sphaeroides. J Bacteriol181:1684–1688. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 80.Castillo-Ramírez S, Vázquez-Castellanos JF, González V, Cevallos MA. 2009. Horizontal gene transfer and diverse functional constrains within a common replication-partitioning system inAlphaproteobacteria: therepABC operon. BMC Genomics10:536. doi: 10.1186/1471-2164-10-536. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 81.Wong K, Finan T, Golding B. 2002. Dinucleotide compositional analysis ofSinorhizobium meliloti using the genome signature: distinguishing chromosomes and plasmids. Funct Integr Genomics2:274–281. doi: 10.1007/s10142-002-0068-0. [DOI] [PubMed] [Google Scholar]
• 82.Carbone A, Zinovyev A, Képès F. 2003. Codon adaptation index as a measure of dominating codon bias. Bioinformatics19:2005–2015. doi: 10.1093/bioinformatics/btg272. [DOI] [PubMed] [Google Scholar]
• 83.diCenzo GC, Zamani M, Milunovic B, Finan TM. 2016. Genomic resources for identification of the minimal N₂-fixing symbiotic genome. Environ Microbiol18:2534–2547. doi: 10.1111/1462-2920.13221. [DOI] [PubMed] [Google Scholar]
• 84.Slater SC, Goldman BS, Goodner B, Setubal JC, Farrand SK, Nester EW, Burr TJ, Banta L, Dickerman AW, Paulsen I, Otten L, Suen G, Welch R, Almeida NF, Arnold F, Burton OT, Du Z, Ewing A, Godsy E, Heisel S, Houmiel KL, Jhaveri J, Lu J, Miller NM, Norton S, Chen Q, Phoolcharoen W, Ohlin V, Ondrusek D, Pride N, Stricklin SL, Sun J, Wheeler C, Wilson L, Zhu H, Wood DW. 2009. Genome sequences of threeAgrobacterium biovars help elucidate the evolution of multichromosome genomes in bacteria. J Bacteriol191:2501–2511. doi: 10.1128/JB.01779-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 85.Wong K, Golding GB. 2003. A phylogenetic analysis of the pSymB replicon from theSinorhizobium meliloti genome reveals a complex evolutionary history. Can J Microbiol49:269–280. doi: 10.1139/w03-037. [DOI] [PubMed] [Google Scholar]
• 86.Sun S, Guo H, Xu J. 2006. Multiple gene genealogical analyses reveal both common and distinct population genetic patterns among replicons in the nitrogen-fixing bacteriumSinorhizobium meliloti. Microbiology152:3245–3259. doi: 10.1099/mic.0.29170-0. [DOI] [PubMed] [Google Scholar]
• 87.Zheng J, Guan Z, Cao S, Peng D, Ruan L, Jiang D, Sun M. 2015. Plasmids are vectors for redundant chromosomal genes in theBacillus cereus group. BMC Genomics16:6. doi: 10.1186/s12864-014-1206-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 88.Guo X, Flores M, Mavingui P, Fuentes SI, Hernández G, Dávila G, Palacios R. 2003. Natural genomic design inSinorhizobium meliloti: novel genomic architectures. Genome Res13:1810–1817. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 89.Val M-E, Kennedy SP, Soler-Bistué AJ, Barbe V, Bouchier C, Ducos-Galand M, Skovgaard O, Mazel D. 2014. Fuse or die: how to survive the loss of Dam inVibrio cholerae. Mol Microbiol91:665–678. doi: 10.1111/mmi.12483. [DOI] [PubMed] [Google Scholar]
• 90.Ng WV, Ciufo SA, Smith TM, Bumgarner RE, Baskin D, Faust J, Hall B, Loretz C, Seto J, Slagel J, Hood L, DasSarma S. 1998. Snapshot of a large dynamic replicon in a halophilic archaeon: megaplasmid or minichromosome?Genome Res8:1131–1141. doi: 10.1101/gr.8.11.1131. [DOI] [PubMed] [Google Scholar]
• 91.diCenzo GC, Finan TM. 2015. Genetic redundancy is prevalent within the 6.7 Mb Sinorhizobium meliloti genome. Mol Genet Genomics290:1345–1356. doi: 10.1007/s00438-015-0998-6. [DOI] [PubMed] [Google Scholar]
• 92.Bavishi A, Lin L, Schroeder K, Peters A, Cho H, Choudhary M. 2010. The prevalence of gene duplications and their ancient origin inRhodobacter sphaeroides 2.4.1. BMC Microbiol10:331. doi: 10.1186/1471-2180-10-331. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 93.Maida I, Fondi M, Orlandini V, Emiliani G, Papaleo MC, Perrin E, Fani R. 2014. Origin, duplication and reshuffling of plasmid genes: insights fromBurkholderia vietnamiensis G4 genome. Genomics103:229–238. doi: 10.1016/j.ygeno.2014.02.004. [DOI] [PubMed] [Google Scholar]
• 94.Landeta C, Dávalos A, Cevallos MÁ Geiger O, Brom S, Romero D. 2011. Plasmids with a chromosome-like role in rhizobia. J Bacteriol193:1317–1326. doi: 10.1128/JB.01184-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 95.Popescu A-A, Huber KT, Paradis E. 2012. ape 3.0: new tools for distance-based phylogenetics and evolutionary analysis in R. Bioinformatics28:1536–1537. doi: 10.1093/bioinformatics/bts184. [DOI] [PubMed] [Google Scholar]
• 96.Österman J, Marsh J, Laine PK, Zeng Z, Alatalo E, Sullivan JT, Young JPW, Thomas-Oates J, Paulin L, Lindström K. 2014. Genome sequencing of twoNeorhizobium galegae strains reveals anoeT gene responsible for the unusual acetylation of the nodulation factors. BMC Genomics15:500. doi: 10.1186/1471-2164-15-500. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 97.Quax TEF, Claassens NJ, Söll D, van der Oost J. 2015. Codon bias as a means to fine-tune gene expression. Mol Cell59:149–161. doi: 10.1016/j.molcel.2015.05.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 98.Cooper VS, Vohr SH, Wrocklage SC, Hatcher PJ. 2010. Why genes evolve faster on secondary chromosomes in bacteria. PLoS Comput Biol6:e1000732. doi: 10.1371/journal.pcbi.1000732. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 99.Wan XF, Zhou J, Xu D. 2007. CodonO: a new informatics method for measuring synonymous codon usage bias within and across genomes. Int J Gen Syst35:109–125. doi: 10.1080/03081070500502967. [DOI] [Google Scholar]
• 100.Foerstner KU, von Mering C, Hooper SD, Bork P. 2005. Environments shape the nucleotide composition of genomes. EMBO Rep6:1208–1213. doi: 10.1038/sj.embor.7400538. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 101.Mann S, Chen Y-PP. 2010. Bacterial genomic G+C composition-eliciting environmental adaptation. Genomics95:7–15. doi: 10.1016/j.ygeno.2009.09.002. [DOI] [PubMed] [Google Scholar]
• 102.Lassalle F, Périan S, Bataillon T, Nesme X, Duret L, Daubin V. 2015. GC-content evolution in bacterial genomes: the biased gene conversion hypothesis expands. PLoS Genet11:e1004941. doi: 10.1371/journal.pgen.1004941. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 103.Ravenhall M, Škunca N, Lassalle F, Dessimoz C. 2015. Inferring horizontal gene transfer. PLoS Comput Biol11:e1004095. doi: 10.1371/journal.pcbi.1004095. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 104.Rocha EPC, Danchin A. 2002. Base composition bias might result from competition for metabolic resources. Trends Genet18:291–294. doi: 10.1016/S0168-9525(02)02690-2. [DOI] [PubMed] [Google Scholar]
• 105.van Passel MWJ, Bart A, Luyf ACM, van Kampen AHC, van der Ende A. 2006. Compositional discordance between prokaryotic plasmids and host chromosomes. BMC Genomics7:26. doi: 10.1186/1471-2164-7-26. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 106.Nishida H.2012. Comparative analyses of base compositions, DNA sizes, and dinucleotide frequency profiles in archaeal and bacterial chromosomes and plasmids. Int J Evol Biol2012:342482. doi: 10.1155/2012/342482. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 107.Hershberg R, Petrov DA. 2010. Evidence that mutation is universally biased towards AT in bacteria. PLoS Genet6:e1001115. doi: 10.1371/journal.pgen.1001115. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 108.Hildebrand F, Meyer A, Eyre-Walker A. 2010. Evidence of selection upon genomic GC-content in bacteria. PLoS Genet6:e1001107. doi: 10.1371/journal.pgen.1001107. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 109.Karlin S, Burge C. 1995. Dinucleotide relative abundance extremes: a genomic signature. Trends Genet11:283–290. doi: 10.1016/S0168-9525(00)89076-9. [DOI] [PubMed] [Google Scholar]
• 110.Pérez-Mendoza D, Sepúlveda E, Pando V, Muñoz S, Nogales J, Olivares J, Soto MJ, Herrera-Cervera JA, Romero D, Brom S, Sanjuán J. 2005. Identification of therctA gene, which is required for repression of conjugative transfer of rhizobial symbiotic megaplasmids. J Bacteriol187:7341–7350. doi: 10.1128/JB.187.21.7341-7350.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 111.He X, Chang W, Pierce DL, Seib LO, Wagner J, Fuqua C. 2003. Quorum sensing inRhizobium sp. strain NGR234 regulates conjugal transfer (tra) gene expression and influences growth rate. J Bacteriol185:809–822. doi: 10.1128/JB.185.3.809-822.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 112.Yang JC, Lessard PA, Sengupta N, Windsor SD, O'Brien XM, Bramucci M, Tomb J-F, Nagarajan V, Sinskey AJ. 2007. TraA is required for megaplasmid conjugation inRhodococcus erythropolis AN12. Plasmid57:55–70. doi: 10.1016/j.plasmid.2006.08.002. [DOI] [PubMed] [Google Scholar]
• 113.Romanchuk A, Jones CD, Karkare K, Moore A, Smith BA, Jones C, Dougherty K, Baltrus DA. 2014. Bigger is not always better: transmission and fitness burden of ∼1MBPseudomonas syringae megaplasmid pMPPla107. Plasmid73:16–25. doi: 10.1016/j.plasmid.2014.04.002. [DOI] [PubMed] [Google Scholar]
• 114.Brom S, García-de los Santos A, Cervantes L, Palacios R, Romero D. 2000. InRhizobium etli symbiotic plasmid transfer, nodulation competitivity and cellular growth require interaction among different replicons. Plasmid44:34–43. doi: 10.1006/plas.2000.1469. [DOI] [PubMed] [Google Scholar]
• 115.Herrera-Cervera JA, Caballero-Mellado J, Laguerre G, Tichy H-V, Requena N, Amarger N, Martínez-Romero E, Olivares J, Sanjuán J. 1999. At least five rhizobial species nodulatePhaseolus vulgaris in a Spanish soil. FEMS Microbiol Ecol30:87–97. doi: 10.1111/j.1574-6941.1999.tb00638.x. [DOI] [Google Scholar]
• 116.Brom S, Girard L, García-de los Santos A, Sanjuan-Pinilla JM, Olivares J, Sanjuán J. 2002. Conservation of plasmid-encoded traits among bean-nodulatingRhizobium species. Appl Environ Microbiol68:2555–2561. doi: 10.1128/AEM.68.5.2555-2561.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 117.Young JPW, Wexler M. 1988. Sym plasmid and chromosomal genotypes are correlated in field populations ofRhizobium leguminosarum. Microbiology134:2731–2739. doi: 10.1099/00221287-134-10-2731. [DOI] [Google Scholar]
• 118.Blanca-Ordóñez H, Oliva-García JJ, Pérez-Mendoza D, Soto MJ, Olivares J, Sanjuán J, Nogales J. 2010. pSymA-dependent mobilization of theSinorhizobium meliloti pSymB megaplasmid. J Bacteriol192:6309–6312. doi: 10.1128/JB.00549-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 119.Banfalvi Z, Kondorosi E, Kondorosi A. 1985.Rhizobium meliloti carries two megaplasmids. Plasmid13:129–138. doi: 10.1016/0147-619X(85)90065-4. [DOI] [PubMed] [Google Scholar]
• 120.Finan TM, Kunkel B, De Vos GF, Signer ER. 1986. Second symbiotic megaplasmid inRhizobium meliloti carrying exopolysaccharide and thiamine synthesis genes. J Bacteriol167:66–72. doi: 10.1128/jb.167.1.66-72.1986. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 121.Galardini M, Pini F, Bazzicalupo M, Biondi EG, Mengoni A. 2013. Replicon-dependent bacterial genome evolution: the case ofSinorhizobium meliloti. Genome Biol Evol5:542–558. doi: 10.1093/gbe/evt027. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 122.Guo H, Sun S, Eardly B, Finan T, Xu J. 2009. Genome variation in the symbiotic nitrogen-fixing bacteriumSinorhizobium meliloti. Genome52:862–875. doi: 10.1139/G09-060. [DOI] [PubMed] [Google Scholar]
• 123.Epstein B, Branca A, Mudge J, Bharti AK, Briskine R, Farmer AD, Sugawara M, Young ND, Sadowsky MJ, Tiffin P. 2012. Population genomics of the facultatively mutualistic bacteriaSinorhizobium meliloti andS. medicae. PLoS Genet8:e1002868. doi: 10.1371/journal.pgen.1002868. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 124.Holden MTG, Seth-Smith HMB, Crossman LC, Sebaihia M, Bentley SD, Cerdeño-Tárraga AM, Thomson NR, Bason N, Quail MA, Sharp S, Cherevach I, Churcher C, Goodhead I, Hauser H, Holroyd N, Mungall K, Scott P, Walker D, White B, Rose H, Iversen P, Mil-Homens D, Rocha EPC, Fialho AM, Baldwin A, Dowson C, Barrell BG, Govan JR, Vandamme P, Hart CA, Mahenthiralingam E, Parkhill J. 2009. The genome ofBurkholderia cenocepacia J2315, an epidemic pathogen of cystic fibrosis patients. J Bacteriol191:261–277. doi: 10.1128/JB.01230-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 125.Holden MTG, Titball RW, Peacock SJ, Cerdeño-Tárraga AM, Atkins T, Crossman LC, Pitt T, Churcher C, Mungall K, Bentley SD, Sebaihia M, Thomson NR, Bason N, Beacham IR, Brooks K, Brown KA, Brown NF, Challis GL, Cherevach I, Chillingworth T, Cronin A, Crossett B, Davis P, DeShazer D, Feltwell T, Fraser A, Hance Z, Hauser H, Holroyd S, Jagels K, Keith KE, Maddison M, Moule S, Price C, Quail MA, Rabbinowitsch E, Rutherford K, Sanders M, Simmonds M, Songsivilai S, Stevens K, Tumapa S, Vesaratchavest M, Whitehead S, Yeats C, Barrell BG, Oyston PCF, Parkhill J. 2004. Genomic plasticity of the causative agent of melioidosis,Burkholderia pseudomallei. Proc Natl Acad Sci U S A101:14240–14245. doi: 10.1073/pnas.0403302101. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 126.Page AJ, Cummins CA, Hunt M, Wong VK, Reuter S, Holden MTG, Fookes M, Falush D, Keane JA, Parkhill J. 2015. Roary: rapid large-scale prokaryote pan genome analysis. Bioinformatics31:3691–3693. doi: 10.1093/bioinformatics/btv421. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 127.Janssen PJ, Van Houdt R, Moors H, Monsieurs P, Morin N, Michaux A, Benotmane MA, Leys N, Vallaeys T, Lapidus A, Monchy S, Médigue C, Taghavi S, McCorkle S, Dunn J, van der Lelie D, Mergeay M. 2010. The complete genome sequence ofCupriavidus metallidurans strain CH34, a master survivalist in harsh and anthropogenic environments. PLoS One5:e10433. doi: 10.1371/journal.pone.0010433. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 128.Van Houdt R, Monsieurs P, Mijnendonckx K, Provoost A, Janssen A, Mergeay M, Leys N. 2012. Variation in genomic islands contribute to genome plasticity inCupriavidus metallidurans. BMC Genomics13:111. doi: 10.1186/1471-2164-13-111. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 129.Bavishi A, Abhishek A, Lin L, Choudhary M. 2010. Complex prokaryotic genome structure: rapid evolution of chromosome II. Genome53:675–687. doi: 10.1139/G10-046. [DOI] [PubMed] [Google Scholar]
• 130.Guo HJ, Wang ET, Zhang XX, Li QQ, Zhang YM, Tian CF, Chen WX. 2014. Replicon-dependent differentiation of symbiosis-related genes inSinorhizobium strains nodulatingGlycine max. Appl Environ Microbiol80:1245–1255. doi: 10.1128/AEM.03037-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 131.Dillon MM, Sung W, Lynch M, Cooper VS. 2015. The rate and molecular spectrum of spontaneous mutations in the GC-rich multichromosome genome ofBurkholderia cenocepacia. Genetics200:935–946. doi: 10.1534/genetics.115.176834. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 132.Dillon MM, Cooper VS. 2016. The fitness effects of spontaneous mutations nearly unseen by selection in a bacterium with multiple chromosomes. Genetics204:1225–1238. doi: 10.1534/genetics.116.193060. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 133.Dillon MM, Sung W, Sebra R, Lynch M, Cooper VS. 2017. Genome-wide biases in the rate and molecular spectrum of spontaneous mutations inVibrio cholerae andVibrio fischeri. Mol Biol Evol34:93–109. doi: 10.1093/molbev/msw224. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 134.Peters AE, Bavishi A, Cho H, Choudhary M. 2012. Evolutionary constraints and expression analysis of gene duplications inRhodobacter sphaeroides 2.4.1. BMC Res Notes5:192. doi: 10.1186/1756-0500-5-192. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 135.Tatusov RL, Fedorova ND, Jackson JD, Jacobs AR, Kiryutin B, Koonin EV, Krylov DM, Mazumder R, Mekhedov SL, Nikolskaya AN, Rao BS, Smirnov S, Sverdlov AV, Vasudevan S, Wolf YI, Yin JJ, Natale DA. 2003. The COG database: an updated version includes eukaryotes. BMC Bioinformatics4:41. doi: 10.1186/1471-2105-4-41. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 136.Paulsen IT, Seshadri R, Nelson KE, Eisen JA, Heidelberg JF, Read TD, Dodson RJ, Umayam L, Brinkac LM, Beanan MJ, Daugherty SC, Deboy RT, Durkin AS, Kolonay JF, Madupu R, Nelson WC, Ayodeji B, Kraul M, Shetty J, Malek J, Van Aken SE, Riedmuller S, Tettelin H, Gill SR, White O, Salzberg SL, Hoover DL, Lindler LE, Halling SM, Boyle SM, Fraser CM. 2002. TheBrucella suis genome reveals fundamental similarities between animal and plant pathogens and symbionts. Proc Natl Acad Sci U S A99:13148–13153. doi: 10.1073/pnas.192319099. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 137.Frank O, Göker M, Pradella S, Petersen J. 2015. Ocean's twelve: flagellar and biofilm chromids in the multipartite genome ofMarinovum algicola DG898 exemplify functional compartmentalization. Environ Microbiol17:4019–4034. doi: 10.1111/1462-2920.12947. [DOI] [PubMed] [Google Scholar]
• 138.Mahillon J, Chandler M. 1998. Insertion sequences. Microbiol Mol Biol Rev62:725–774. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 139.Elena SF, Ekunwe L, Hajela N, Oden SA, Lenski RE. 1998. Distribution of fitness effects caused by random insertion mutations inEscherichia coli. Genetica102–103:349–358. [PubMed] [Google Scholar]
• 140.Wu Y, Aandahl RZ, Tanaka MM. 2015. Dynamics of bacterial insertion sequences: can transposition bursts help the elements persist?BMC Evol Biol15:288. doi: 10.1186/s12862-015-0560-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 141.Villaseñor T, Brom S, Dávalos A, Lozano L, Romero D, García-de los Santos A. 2011. Housekeeping genes essential for pantothenate biosynthesis are plasmid-encoded inRhizobium etli andRhizobium leguminosarum. BMC Microbiol11:66. doi: 10.1186/1471-2180-11-66. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 142.García-de los Santos A, Brom S. 1997. Characterization of two plasmid-bornelpsβ loci ofRhizobium etli required for lipopolysaccharide synthesis and for optimal interaction with plants. Mol Plant Microbe Interact10:891–902. doi: 10.1094/MPMI.1997.10.7.891. [DOI] [PubMed] [Google Scholar]
• 143.Brom S, García-de los Santos A, Stepkowsky T, Flores M, Davila G, Romero D, Palacios R. 1992. Different plasmids ofRhizobium leguminosarum bv.phaseoli are required for optimal symbiotic performance. J Bacteriol174:5183–5189. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 144.Hynes MF, Simon R, Müller P, Niehaus K, Labes M, Pühler A. 1986. The two megaplasmids ofRhizobium meliloti are involved in the effective nodulation of alfalfa. Mol Gen Genet202:356–362. doi: 10.1007/BF00333262. [DOI] [Google Scholar]
• 145.Hynes MF, McGregor NF. 1990. Two plasmids other than the nodulation plasmid are necessary for formation of nitrogen-fixing nodules byRhizobium leguminosarum. Mol Microbiol4:567–574. doi: 10.1111/j.1365-2958.1990.tb00625.x. [DOI] [PubMed] [Google Scholar]
• 146.González V, Santamaría RI, Bustos P, Hernández-González I, Medrano-Soto A, Moreno-Hagelsieb G, Janga SC, Ramírez MA, Jiménez-Jacinto V, Collado-Vides J, Dávila G. 2006. The partitionedRhizobium etli genome: genetic and metabolic redundancy in seven interacting replicons. Proc Natl Acad Sci U S A103:3834–3839. doi: 10.1073/pnas.0508502103. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 147.Galardini M, Brilli M, Spini G, Rossi M, Roncaglia B, Bani A, Chiancianesi M, Moretto M, Engelen K, Bacci G, Pini F, Biondi EG, Bazzicalupo M, Mengoni A. 2015. Evolution of intra-specific regulatory networks in a multipartite bacterial genome. PLoS Comput Biol11:e1004478. doi: 10.1371/journal.pcbi.1004478. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 148.Pini F, De Nisco NJ, Ferri L, Penterman J, Fioravanti A, Brilli M, Mengoni A, Bazzicalupo M, Viollier PH, Walker GC, Biondi EG. 2015. Cell cycle control by the master regulator CtrA inSinorhizobium meliloti. PLoS Genet11:e1005232. doi: 10.1371/journal.pgen.1005232. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 149.Bobik C, Meilhoc E, Batut J. 2006. FixJ: a major regulator of the oxygen limitation response and late symbiotic functions ofSinorhizobium meliloti. J Bacteriol188:4890–4902. doi: 10.1128/JB.00251-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 150.Ronson CW, Nixon BT, Albright LM, Ausubel FM. 1987.Rhizobium meliloti ntrA (rpoN) gene is required for diverse metabolic functions. J Bacteriol169:2424–2431. doi: 10.1128/jb.169.6.2424-2431.1987. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 151.Barnett MJ, Toman CJ, Fisher RF, Long SR. 2004. A dual-genome symbiosis chip for coordinate study of signal exchange and development in a prokaryote-host interaction. Proc Natl Acad Sci U S A101:16636–16641. doi: 10.1073/pnas.0407269101. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 152.diCenzo GC, MacLean AM, Milunovic B, Golding GB, Finan TM. 2014. Examination of prokaryotic multipartite genome evolution through experimental genome reduction. PLoS Genet10:e1004742. doi: 10.1371/journal.pgen.1004742. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 153.Morton ER, Platt TG, Fuqua C, Bever JD. 2014. Non-additive costs and interactions alter the competitive dynamics of co-occurring ecologically distinct plasmids. Proc Biol Sci281:20132173. doi: 10.1098/rspb.2013.2173. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 154.Dougherty K, Smith BA, Moore AF, Maitland S, Fanger C, Murillo R, Baltrus DA. 2014. Multiple phenotypic changes associated with large-scale horizontal gene transfer. PLoS One9:e102170. doi: 10.1371/journal.pone.0102170. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 155.Baltrus DA, Nishimura MT, Romanchuk A, Chang JH, Mukhtar MS, Cherkis K, Roach J, Grant SR, Jones CD. 2011. Dynamic evolution of pathogenicity revealed by sequencing and comparative genomics of 19Pseudomonas syringae isolates. PLoS Pathog7:e1002132. doi: 10.1371/journal.ppat.1002132. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 156.Lee M-C, Marx CJ. 2012. Repeated, selection-driven genome reduction of accessory genes in experimental populations. PLoS Genet8:e1002651. doi: 10.1371/journal.pgen.1002651. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 157.Hessen DO, Jeyasingh PD, Neiman M, Weider LJ. 2010. Genome streamlining and the elemental costs of growth. Trends Ecol Evol25:75–80. doi: 10.1016/j.tree.2009.08.004. [DOI] [PubMed] [Google Scholar]
• 158.Vieira-Silva S, Touchon M, Rocha EPC. 2010. No evidence for elemental-based streamlining of prokaryotic genomes. Trends Ecol Evol25:319–320. doi: 10.1016/j.tree.2010.03.001. [DOI] [PubMed] [Google Scholar]
• 159.Morton ER, Merritt PM, Bever JD, Fuqua C. 2013. Large deletions in the pAtC58 megaplasmid ofAgrobacterium tumefaciens can confer reduced carriage cost and increased expression of virulence genes. Genome Biol Evol5:1353–1364. doi: 10.1093/gbe/evt095. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 160.Mauchline TH, Fowler JE, East AK, Sartor AL, Zaheer R, Hosie AHF, Poole PS, Finan TM. 2006. Mapping theSinorhizobium meliloti 1021 solute-binding protein-dependent transportome. Proc Natl Acad Sci U S A103:17933–17938. doi: 10.1073/pnas.0606673103. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 161.MacLean AM, Finan TM, Sadowsky MJ. 2007. Genomes of the symbiotic nitrogen-fixing bacteria of legumes. Plant Physiol144:615–622. doi: 10.1104/pp.107.101634. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 162.Couturier E, Rocha EPC. 2006. Replication-associated gene dosage effects shape the genomes of fast-growing bacteria but only for transcription and translation genes. Mol Microbiol59:1506–1518. doi: 10.1111/j.1365-2958.2006.05046.x. [DOI] [PubMed] [Google Scholar]
• 163.Labbe RG, Huang TH. 1995. Generation times and modeling of enterotoxin-positive and enterotoxin-negative strains ofClostridium perfringens in laboratory media and ground beef. J Food Prot58:1303–1306. doi: 10.4315/0362-028X-58.12.1303. [DOI] [PubMed] [Google Scholar]
• 164.Vieira-Silva S, Rocha EPC. 2010. The systemic imprint of growth and its uses in ecological (meta)genomics. PLoS Genet6:e1000808. doi: 10.1371/journal.pgen.1000808. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 165.Cowie A, Cheng J, Sibley CD, Fong Y, Zaheer R, Patten CL, Morton RM, Golding GB, Finan TM. 2006. An integrated approach to functional genomics: construction of a novel reporter gene fusion library forSinorhizobium meliloti. Appl Environ Microbiol72:7156–7167. doi: 10.1128/AEM.01397-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 166.Weng X, Xiao J. 2014. Spatial organization of transcription in bacterial cells. Trends Genet30:287–297. doi: 10.1016/j.tig.2014.04.008. [DOI] [PubMed] [Google Scholar]
• 167.Xu Q, Dziejman M, Mekalanos JJ. 2003. Determination of the transcriptome ofVibrio cholerae during intraintestinal growth and midexponential phasein vitro. Proc Natl Acad Sci U S A100:1286–1291. doi: 10.1073/pnas.0337479100. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 168.Yoder-Himes DR, Konstantinidis KT, Tiedje JM. 2010. Identification of potential therapeutic targets forBurkholderia cenocepacia by comparative transcriptomics. PLoS One5:e8724. doi: 10.1371/journal.pone.0008724. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 169.Ramachandran VK, East AK, Karunakaran R, Downie JA, Poole PS. 2011. Adaptation ofRhizobium leguminosarum to pea, alfalfa and sugar beet rhizospheres investigated by comparative transcriptomics. Genome Biol12:R106. doi: 10.1186/gb-2011-12-10-r106. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 170.López-Guerrero MG, Ormeño-Orrillo E, Acosta JL, Mendoza-Vargas A, Rogel MA, Ramírez MA, Rosenblueth M, Martínez-Romero J, Martínez-Romero E. 2012. Rhizobial extrachromosomal replicon variability, stability and expression in natural niches. Plasmid68:149–158. doi: 10.1016/j.plasmid.2012.07.002. [DOI] [PubMed] [Google Scholar]
• 171.Becker A, Bergès H, Krol E, Bruand C, Rüberg S, Capela D, Lauber E, Meilhoc E, Ampe F, de Bruijn FJ, Fourment J, Francez-Charlot A, Kahn D, Küster H, Liebe C, Pühler A, Weidner S, Batut J. 2004. Global changes in gene expression inSinorhizobium meliloti 1021 under microoxic and symbiotic conditions. Mol Plant Microbe Interact17:292–303. doi: 10.1094/MPMI.2004.17.3.292. [DOI] [PubMed] [Google Scholar]
• 172.Boussau B, Karlberg EO, Frank AC, Legault B-A, Andersson SGE. 2004. Computational inference of scenarios for α-proteobacterial genome evolution. Proc Natl Acad Sci U S A101:9722–9727. doi: 10.1073/pnas.0400975101. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 173.Johnson TJ, Nolan LK. 2009. Pathogenomics of the virulence plasmids ofEscherichia coli. Microbiol Mol Biol Rev73:750–774. doi: 10.1128/MMBR.00015-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 174.Pál C, Papp B, Lercher MJ. 2005. Adaptive evolution of bacterial metabolic networks by horizontal gene transfer. Nat Genet37:1372–1375. doi: 10.1038/ng1686. [DOI] [PubMed] [Google Scholar]
• 175.Lawrence JG, Ochman H. 1998. Molecular archaeology of theEscherichia coli genome. Proc Natl Acad Sci U S A95:9413–9417. doi: 10.1073/pnas.95.16.9413. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 176.Ochman H, Lawrence JG, Groisman EA. 2000. Lateral gene transfer and the nature of bacterial innovation. Nature405:299–304. doi: 10.1038/35012500. [DOI] [PubMed] [Google Scholar]
• 177.Wiedenbeck J, Cohan FM. 2011. Origins of bacterial diversity through horizontal genetic transfer and adaptation to new ecological niches. FEMS Microbiol Rev35:957–976. doi: 10.1111/j.1574-6976.2011.00292.x. [DOI] [PubMed] [Google Scholar]
• 178.Niehus R, Mitri S, Fletcher AG, Foster KR. 2015. Migration and horizontal gene transfer divide microbial genomes into multiple niches. Nat Commun6:8924. doi: 10.1038/ncomms9924. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 179.Pini F, Galardini M, Bazzicalupo M, Mengoni A. 2011. Plant-bacteria association and symbiosis: are there common genomic traits inAlphaproteobacteria?Genes2:1017–1032. doi: 10.3390/genes2041017. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 180.Park C, Zhang J. 2012. High expression hampers horizontal gene transfer. Genome Biol Evol4:523–532. doi: 10.1093/gbe/evs030. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 181.Hao W, Golding GB. 2006. The fate of laterally transferred genes: life in the fast lane to adaptation or death. Genome Res16:636–643. doi: 10.1101/gr.4746406. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 182.Kurland CG, Canback B, Berg OG. 2003. Horizontal gene transfer: a critical view. Proc Natl Acad Sci U S A100:9658–9662. doi: 10.1073/pnas.1632870100. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 183.Hao W, Golding GB. 2010. Inferring bacterial genome flux while considering truncated genes. Genetics186:411–426. doi: 10.1534/genetics.110.118448. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 184.diCenzo GC, Checcucci A, Bazzicalupo M, Mengoni A, Viti C, Dziewit L, Finan TM, Galardini M, Fondi M. 2016. Metabolic modelling reveals the specialization of secondary replicons for niche adaptation inSinorhizobium meliloti. Nat Commun7:12219. doi: 10.1038/ncomms12219. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 185.Lawrence JG, Ochman H. 1997. Amelioration of bacterial genomes: rates of change and exchange. J Mol Evol44:383–397. doi: 10.1007/PL00006158. [DOI] [PubMed] [Google Scholar]
• 186.Lercher MJ, Pál C. 2008. Integration of horizontally transferred genes into regulatory interaction networks takes many million years. Mol Biol Evol25:559–567. doi: 10.1093/molbev/msm283. [DOI] [PubMed] [Google Scholar]
• 187.Fei F, diCenzo GC, Bowdish DME, McCarry BE, Finan TM. 2016. Effects of synthetic large-scale genome reduction on metabolism and metabolic preferences in a nutritionally complex environment. Metabolomics12:23. doi: 10.1007/s11306-015-0928-y. [DOI] [Google Scholar]
• 188.Tian CF, Zhou YJ, Zhang YM, Li QQ, Zhang YZ, Li DF, Wang S, Wang J, Gilbert LB, Li YR, Chen WX. 2012. Comparative genomics of rhizobia nodulating soybean suggests extensive recruitment of lineage-specific genes in adaptations. Proc Natl Acad Sci U S A109:8629–8634. doi: 10.1073/pnas.1120436109. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 189.Kaneko T, Nakamura Y, Sato S, Minamisawa K, Uchiumi T, Sasamoto S, Watanabe A, Idesawa K, Iriguchi M, Kawashima K, Kohara M, Matsumoto M, Shimpo S, Tsuruoka H, Wada T, Yamada M, Tabata S. 2002. Complete genomic sequence of nitrogen-fixing symbiotic bacteriumBradyrhizobium japonicum USDA110. DNA Res9:189–197. doi: 10.1093/dnares/9.6.189. [DOI] [PubMed] [Google Scholar]
• 190.Baumdicker F, Hess WR, Pfaffelhuber P. 2012. The infinitely many genes model for the distributed genome of bacteria. Genome Biol Evol4:443–456. doi: 10.1093/gbe/evs016. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 191.Rankin DJ, Rocha EPC, Brown SP. 2011. What traits are carried on mobile genetic elements, and why?Heredity (Edinb)106:1–10. doi: 10.1038/hdy.2010.24. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 192.Lerat E, Ochman H. 2004. Psi-Phi: exploring the outer limits of bacterial pseudogenes. Genome Res14:2273–2278. doi: 10.1101/gr.2925604. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 193.Song J, Ware A, Liu S-L. 2003. Wavelet to predict bacterialori andter: a tendency towards a physical balance. BMC Genomics4:17. doi: 10.1186/1471-2164-4-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 194.Hendrickson H, Lawrence JG. 2006. Selection for chromosome architecture in bacteria. J Mol Evol62:615–629. doi: 10.1007/s00239-005-0192-2. [DOI] [PubMed] [Google Scholar]
• 195.Morton RA, Morton BR. 2007. Separating the effects of mutation and selection in producing DNA skew in bacterial chromosomes. BMC Genomics8:369. doi: 10.1186/1471-2164-8-369. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 196.Darling AE, Miklós I, Ragan MA. 2008. Dynamics of genome rearrangement in bacterial populations. PLoS Genet4:e1000128. doi: 10.1371/journal.pgen.1000128. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 197.Lesterlin C, Pages C, Dubarry N, Dasgupta S, Cornet F. 2008. Asymmetry of chromosome replichores renders the DNA translocase activity of FtsK essential for cell division and cell shape maintenance inEscherichia coli. PLoS Genet4:e1000288. doi: 10.1371/journal.pgen.1000288. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 198.Hill CW, Gray JA. 1988. Effects of chromosomal inversion on cell fitness inEscherichia coli K-12. Genetics119:771–778. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 199.Liu G-R, Liu W-Q, Johnston RN, Sanderson KE, Li S-X, Liu S-L. 2006. Genome plasticity andori-ter rebalancing inSalmonella typhi. Mol Biol Evol23:365–371. doi: 10.1093/molbev/msj042. [DOI] [PubMed] [Google Scholar]
• 200.Esnault E, Valens M, Espéli O, Boccard F. 2007. Chromosome structuring limits genome plasticity inEscherichia coli. PLoS Genet3:e226. doi: 10.1371/journal.pgen.0030226. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 201.Mao X, Zhang H, Yin Y, Xu Y. 2012. The percentage of bacterial genes on leading versus lagging strands is influenced by multiple balancing forces. Nucleic Acids Res40:8210–8218. doi: 10.1093/nar/gks605. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 202.Koonin EV.2009. Evolution of genome architecture. Int J Biochem Cell Biol41:298–306. doi: 10.1016/j.biocel.2008.09.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 203.Srivatsan A, Tehranchi A, MacAlpine DM, Wang JD. 2010. Co-orientation of replication and transcription preserves genome integrity. PLoS Genet6:e1000810. doi: 10.1371/journal.pgen.1000810. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 204.Cevallos MA, Cervantes-Rivera R, Gutiérrez-Ríos RM. 2008. TherepABC plasmid family. Plasmid60:19–37. doi: 10.1016/j.plasmid.2008.03.001. [DOI] [PubMed] [Google Scholar]
• 205.MacLellan SR, Zaheer R, Sartor AL, MacLean AM, Finan TM. 2006. Identification of a megaplasmid centromere reveals genetic structural diversity within therepABC family of basic replicons. Mol Microbiol59:1559–1575. doi: 10.1111/j.1365-2958.2006.05040.x. [DOI] [PubMed] [Google Scholar]
• 206.Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, Sandstrom R, Bernstein B, Bender MA, Groudine M, Gnirke A, Stamatoyannopoulos J, Mirny LA, Lander ES, Dekker J. 2009. Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science326:289–293. doi: 10.1126/science.1181369. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 207.Le TBK, Imakaev MV, Mirny LA, Laub MT. 2013. High-resolution mapping of the spatial organization of a bacterial chromosome. Science342:731–734. doi: 10.1126/science.1242059. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 208.Ried T, Schröck E, Ning Y, Wienberg J. 1998. Chromosome painting: a useful art. Hum Mol Genet7:1619–1626. doi: 10.1093/hmg/7.10.1619. [DOI] [PubMed] [Google Scholar]
• 209.Cremer T, Cremer M. 2010. Chromosome territories. Cold Spring Harb Perspect Biol2:a003889. doi: 10.1101/cshperspect.a003889. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 210.Shintani M, Suzuki-Minakuchi C, Nojiri H. 2015. Nucleoid-associated proteins encoded on plasmids: occurrence and mode of function. Plasmid80:32–44. doi: 10.1016/j.plasmid.2015.04.008. [DOI] [PubMed] [Google Scholar]
• 211.San Millan A, Toll-Riera M, Qi Q, MacLean RC. 2015. Interactions between horizontally acquired genes create a fitness cost inPseudomonas aeruginosa. Nat Commun6:6845. doi: 10.1038/ncomms7845. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 212.Baltrus DA.2013. Exploring the costs of horizontal gene transfer. Trends Ecol Evol28:489–495. doi: 10.1016/j.tree.2013.04.002. [DOI] [PubMed] [Google Scholar]
• 213.Rivera MC, Jain R, Moore JE, Lake JA. 1998. Genomic evidence for two functionally distinct gene classes. Proc Natl Acad Sci U S A95:6239–6244. doi: 10.1073/pnas.95.11.6239. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 214.Jain R, Rivera MC, Lake JA. 1999. Horizontal gene transfer among genomes: the complexity hypothesis. Proc Natl Acad Sci U S A96:3801–3806. doi: 10.1073/pnas.96.7.3801. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 215.Wu M, Scott AJ. 2012. Phylogenomic analysis of bacterial and archaeal sequences with AMPHORA2. Bioinformatics28:1033–1034. doi: 10.1093/bioinformatics/bts079. [DOI] [PubMed] [Google Scholar]
• 216.Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W, Lopez R, McWilliam H, Remmert M, Söding J, Thompson JD, Higgins DG. 2011. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol7:539. doi: 10.1038/msb.2011.75. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 217.Capella-Gutiérrez S, Silla-Martínez JM, Gabaldón T. 2009. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics25:1972–1973. doi: 10.1093/bioinformatics/btp348. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 218.Stamatakis A.2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics30:1312–1313. doi: 10.1093/bioinformatics/btu033. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 219.Miller MA, Pfeiffer W, Schwartz T. 2010. Creating the CIPRES Science Gateway for inference of large phylogenetic trees, p 1–8.InProceedings of the Gateway Computing Environments Workshop, New Orleans, LAACM, New York, NY. [Google Scholar]
• 220.Wu S, Zhu Z, Fu L, Niu B, Li W. 2011. WebMGA: a customizable Web server for fast metagenomic sequence analysis. BMC Genomics12:444. doi: 10.1186/1471-2164-12-444. [DOI] [PMC free article] [PubMed] [Google Scholar]

Supplementary Materials