|Home | About | Journals | Submit | Contact Us | Français|
The Roseobacter clade is a broadly distributed, abundant, and biogeochemically relevant group of marine bacteria. Representatives are often associated with organic surfaces in disparate marine environments, suggesting that a sessile lifestyle is central to the ecology of lineage members. The importance of surface association and colonization has been demonstrated recently for select strains, and it has been hypothesized that production of antimicrobial agents, cell density-dependent regulatory mechanisms, and morphological features contribute to the colonization success of roseobacters. Drawing on these studies, insight into a broad representation of strains is facilitated by the availability of a substantial collection of genome sequences that provides a holistic view of these features among clade members. These genome data often corroborate phenotypic data but also reveal significant variation in terms of gene content and synteny among group members, even among closely related strains (congeners and conspecifics). Thus, while detailed studies of representative strains are serving as models for how roseobacters transition between planktonic and sessile lifestyles, it is becoming clear that additional studies are needed if we are to have a more comprehensive view of how these transitions occur in different lineage members. This is important if we are to understand how associations with surfaces influence metabolic activities contributing to the cycling of carbon and nutrients in the world's oceans.
The Roseobacter lineage is an abundant marine bacterial group whose members mediate key biogeochemical processes and have been the subject of several recent reviews (e.g., references 7, 11, and 91). While roseobacters are broadly distributed across diverse marine environments, roseobacter abundance is often highest near phytoplankton blooms or macroalgae, or in association with organic particles, suggesting that cell-surface interactions are a defining feature of lineage members. This is supported by culture-independent studies identifying roseobacters as ubiquitous and rapid colonizers of a variety of inorganic and organic marine surfaces, including marine algae and dinoflagellates (19, 20, 51). Though little is known of the exact mechanism(s) that roseobacters employ to physically associate with eukaryotic cell surfaces or particles, several cultivated strains have been shown to be capable of surface colonization (8, 71). Furthermore, laboratory-based studies confirm expression of traits expected to be important in colonization success, including possession of holdfast structures, motility, and chemotaxis, as well as production of quorum sensing (QS) molecules and antimicrobial metabolites in select strains. Despite the recognized importance of roseobacters in natural systems and the numerous (~40) genome sequences that are, or are to soon be, publicly available (7), laboratory investigations of phenotypes expected to contribute to the ecological success of roseobacters are limited. Thus, a comparative genome analysis of functions likely to contribute to, or even define, different colonizing abilities can provide a more holistic view of how widely distributed and conserved these features are among diverse clade members and may facilitate hypothesis-driven research in some areas.
In order to assess the prevalence and conservation of functions that can contribute to colonization success, a survey of 28 Roseobacter genomes that represent a range of phylogenetic subgroups within the clade was undertaken. The genomes were examined with respect to content and synteny for genes involved in motility, attachment, chemotaxis, QS, intracellular signaling, and secondary metabolite production. These functions were selected because they have been shown in marine bacteria, including some roseobacters, to contribute to successful associations with biotic or abiotic surfaces (see, e.g., references 8, 11, 35, 55, 56, and 71). The available complete genome sequences allow for comparisons between organisms with 16S rRNA gene identities ranging from 86 to >99% and include multiple strains of the same genus (e.g., Silicibacter, Roseobacter, and Phaeobacter) as well as strains demonstrating species-level 16S rRNA gene identities (e.g., Sulfitobacter sp. strains EE-36 and NAS-14.1 and the two Phaeobacter gallaeciensis strains share >99% identity of the 16S rRNA gene) (Fig. (Fig.1).1). Thus, this collection of genomes represents broad evolutionary distances within the clade and spans strains isolated from diverse habitats. Where available, phenotypic data (collected by us and others) are presented to provide an integrated perspective on the biological relevance of these traits.
With the exception of Citricella sp. strain SE45, Silicibacter lacuscaerulensis and Phaeobacter sp. strain Y4I, Roseobacter genome sequences were accessed and analyzed with the Integrated Microbial Genomes (IMG) system (http://img.jgi.doe.gov). The three Roseobacter genomes not available through IMG were obtained from the J. Craig Venter Institute where open reading frames (ORFs) were predicted and annotated with an in-house annotation pipeline. Scaffolds were downloaded and subjected to homology searches and additional analyses locally using Geneious 3.0.6 (Biomatters Ltd., Auckland, New Zealand). Of these genomes, five have been closed (Jannaschia sp. strain CCS1, Roseobacter denitrificans OCh114, Silicibacter pomeroyi DSS-3, Silicibacter sp. strain TM1040, and Dinoroseobacter shibae DFL-12) with the remaining available as draft assemblages. Assignments for clusters of orthologous groups (COGs) are made in IMG using RPS-BLAST (reverse-position-specific BLAST) and the NCBI Conserved Domain Database (CDD) (48), using an E-value cutoff of 10−2. A single top hit is reported for each gene (49). An analysis of the presence and abundance of functionally related COGs in each of the genomes was conducted (see Table S1 in the supplemental material). For each gene associated with a COG of interest, the analysis was further extended to neighboring genes; attention was paid to functional prediction (COG), gene order, and putative operon structure in order to provide additional supportive evidence for the presence of a given function. It bears mentioning that there is a continuing debate as to whether the Silicibacter and Ruegeria genera should be folded into a single genus, designated Ruegeria (discussed in reference 7). As the issue has yet to be resolved, we retain the use of Silicibacter here.
Flagellar synthesis and/or motility has been found to be critical to surface recognition, attachment, and biofilm development in many Proteobacteria (see, e.g., references 3, 28, 52, 62, 63, 68, and 94). Thus, it was considered here as a trait that would likely contribute to colonization abilities of roseobacters. Flagellar motility has been experimentally demonstrated for many characterized Roseobacter strains, and motile strains typically display between one and five polar or subpolar flagella (see, e.g., references 5, 31, 56, 74, and 78). Polar complex flagella reminiscent of those found in rhizobial species have been documented in most detail for S. pomeroyi DSS-3 (31), while simple filaments have been characterized for Silicibacter sp. strain TM1040 flagella (56). Among the 28 Roseobacter strains considered here, there is a correlation between motility observed under defined conditions and colonization of, or attachment to, artificial and natural substrates in laboratory studies or in the marine environment (Tables (Tables11 and and2).2). For instance, Bruhn et al. (8) demonstrated that motile roseobacters rapidly attach to glass coverslips immersed in statically grown liquid cultures of these strains whereas nonmotile strains do not (8, 31). Most observations support the glass attachment assay as an adequate tool to rapidly screen for surface colonization abilities (Tables (Tables11 and and2).2). For example, the nonmotile Roseovarius nubinhibens, which was isolated from surface waters and may have a predominantly planktonic lifestyle, was unable to colonize plastic surfaces when allowed to grow for many generations in their presence. However, an exception is noted: the motile S. pomeroyi DSS-3 produced a negative result in the glass attachment assay but did demonstrate surface colonization of various plastics, suggesting that additional environmentally responsive physiological properties contribute to the surface colonization abilities of this strain. Rao et al. have shown that a motile Phaeobacter gallaeciensis strain, originally isolated from the surface of the marine alga Ulva australis, is capable of forming prolific biofilms on artificial as well as natural surfaces (i.e., U. australis leaf disks) (71). With the single exception of Silicibacter sp. strain TM1040 (described below), the specific role of flagella and/or motility in roseobacter-surface interactions has yet to be resolved. However, given the role of flagella in the ability of other motile bacteria to associate with biotic and abiotic surfaces, it is anticipated that these appendages serve analogous functions in roseobacters.
Consistent with a recent analysis suggesting that upwards of 70% of all Proteobacteria possess complete flagellar genetic systems (82), these genes appear to be a common feature of most roseobacter genomes. Bacterial flagella are complex structures: approximately 50 genes, typically organized as operons, are required for their synthesis, regulation, and function (44, 47). The presence of a gene cluster encoding components necessary for production of a functional bacterial flagellum in 24 of the 28 genomes is one of the more conserved features examined here (Tables (Tables11 and and2).2). The genomes of three of the four putatively nonflagellated roseobacters possess few (one to four) genes with homology to flagellar motor, motor switch, and/or capping proteins and are likely incapable of synthesizing a functional flagellum. Roseobacter sp. strain MED193 possesses genes falling into only 11 of the 30 COG flagellar families analyzed. However, the genome sequence shows that the flagellar genetic locus is interrupted by a transposase gene (ZP_01058050) and is found at the terminus of a contig. Thus, it is difficult to conclusively establish whether a complete set of flagellar genes is housed by this genome. However, we note that the genome of this strain does contain a single complete set of chemotaxis genes, organized as an operon, that exhibit a protein domain architecture highly suggestive of flagellar motility control (96). Of the four strains whose genomes lack sufficient evidence for flagellar assembly, phenotypic data are available only for R. nubinhibens, a nonmotile strain isolated from pelagic waters that is unable to colonize artificial surfaces in the laboratory (Tables (Tables11 and and2).2). Additional studies of the three remaining strains, also isolated from surface waters, would provide insight as to whether flagellum-mediated motility may be required for, or diagnostic of, surface colonization in these microbes.
When present in a Roseobacter genome, the vast majority of flagellum-related genes (~90%) are generally restricted to a single locus spanning ca. 30 kb that is most similar in content to the fla2 locus found in Rhodobacter sphaeroides (4, 67). Aside from the conservation of a cassette of eight genes (COG1157, COG1815, COG1558, COG1677, COG1987, COG4786, COG1261, and COG2063) in 17 of the genomes, there is little synteny among these large genetic loci except among congeners (e.g., the Phaeobacter and Sulfitobacter strains). Interestingly, the three Silicibacter strains show significant variation in the order of flagellum-associated genes housed within these loci, as do the two Oceanicola strains. This locus-specific heterogeneity is reflected in the larger repertoire of genetic and phenotypic features considered here for these two genera (Table (Table11 and Fig. Fig.1).1). The genome data can contradict the results of phenotypic assays (e.g., Oceanibulbus indolifex, Oceanicola granulosus, and Oceanicola batensis) (Table (Table1)1) and may suggest niche-specific functions for the production and/or use of these appendages. To date, there is neither genetic nor experimental evidence that roseobacters possess the lateral flagella involved in surface motility (swarming) and common to some Rhodobacteraceae.
While roseobacter motility may share similarities with that of model Alphaproteobacteria, evidence is emerging that there are aspects of roseobacter locomotion unique to the clade. The identification of complex filaments in S. pomeroyi DSS-3 flagella coupled with their polar orientation and restriction of rotation to a solely clockwise manner (31) has led to the suggestion that the flagellar machinery of some roseobacters might be most similar to that found in members of the Rhizobiaceae. However, all of the Roseobacter genomes lack homologs to MotC, -D, and -E, proteins that play critical roles in modulation of flagellar speed and hook length in Sinorhizobium meliloti and related microbes (25, 26). Caulobacter crescentus possesses a rigid flagellum capable of brief rotational switches that serve to alter the spatial orientation of moving cells (30). Flagellar rotational switches in C. crescentus are mediated by the “switch proteins” FliG, -N, and -M, which directly interact with components of the chemotaxis pathway (14, 60, 99). Functional homologs of these proteins exist in most motile chemotactic bacteria (90, 97); however, only genes of the FliN type (COG1886) are found in the flagellar genetic loci of roseobacter genomes, and not all motile roseobacters are chemotactic (discussed below). A recent genetic analysis of Silicibacter sp. strain TM1040 determined that regulation of flagellar biosynthesis utilizes a mixture of molecular components from Caulobacter and Rhodobacter but also invokes unique machinery. Ongoing investigations of two novel regulatory proteins (FlaC and -D) are expected to reveal complex mechanisms mediating the “swim-or-stick” switch critical for symbiosis in this organism (4).
The ability of bacteria to sense and navigate toward growth substrates is expected to provide a competitive advantage in the heterogeneous milieu of natural environments (90). Chemotaxis may also be intimately linked with colonization when the surface serves as a supply of, or signal for, growth compounds (22, 98). The extent of chemotactic behavior among Roseobacter group members is not well understood. Chemotaxis has been well described only for Silicibacter sp. strain TM1040, a strain known to be chemotactic toward the algal osmolyte dimethylsulfoniopropionate and a variety of amino acids produced by the dinoflagellate from which it was isolated (54, 56). Furthermore, nonmotile mutants of this strain have altered localization on, or within, their dinoflagellate hosts (55).
Sixteen of the Roseobacter genomes contain genes encoding chemotaxis pathways, and there is strong conservation of a cassette of genes (cheYAWRY). CheAYW and transmembrane chemoreceptors, referred to as methyl-accepting chemotaxis proteins (MCPs), represent the minimal set of proteins required for functional chemotaxis (96). MCPs detect specific chemicals, and their frequency in bacterial genomes is known to vary significantly (1a). Among the che operon-containing roseobacters, significant variation exists in the repertoire of chemoreceptors in a given strain (see Table S1 in the supplemental material); the MCP count varies from one (O. batsensis) to 19 (Silicibacter sp. strain TM1040), with an average of eight. While few bacterial chemoreceptors have been characterized with respect to their ligands and none are presently known for roseobacters, it is likely that the number of chemoreceptors is a direct reflection of the sensing specificity and flexibility of a given strain (1a, 36, 100). Thus, very distinct strategies for identifying and obtaining growth substrates likely exist within this metabolically versatile lineage. There is no correlation between Roseobacter genome size and the number of MCPs identified. Nor is there an obvious correlation between MCP abundance and metabolic versatility, as determined by analysis of a collection of 10 genes encoding proteins of biogeochemical relevance described in the work of Moran et al. (58).
The genome data suggest that the complement of che genes found in most Roseobacter strains are analogous to those found in plant-associated Alphaproteobacteria, including S. meliloti and Agrobacterium tumefaciens, the latter for which chemotaxis has been demonstrated to be critical for biofilm formation (52, 83). This points to a similar sensory architecture and common mode of action between chemotactic roseobacters and these well-studied strains. In all recognized bacteria, chemotaxis is initiated by MCP binding of its cognate ligand. This action initiates a CheAYW-mediated signal transduction cascade that functions to relay sensory information to flagellar motors. CheR and CheB each have roles in adaptation, and CheD contributes to MCP activation in some organisms (86). With a single exception (Roseovarius sp. strain TM1035), all of the putatively chemotactic roseobacters possess cheR, -B, and -D. In the Roseobacter genomes, cheR is found clustered with cheAYW; cheB and -D are often found together either in this same locus or in a separate region of the chromosome. Roseovarius sp. strain TM1035 possesses two separate, but homologous in content, che loci; both lack a cheD homolog which is not present elsewhere on the chromosome. While not essential for signal transduction (96), the presence of cheD is conserved in the remaining 15 che-containing roseobacter genomes. Silicibacter sp. strain TM1040 also appears to house two separate che loci that differ in content and are both found on a megaplasmid.
In Silicibacter sp. strain TM1040, both motility and chemotaxis have been found to play essential roles in this bacterium's symbiotic relationship with a marine dinoflagellate (55), and all data indicate that chemotaxis serves as a mean for this strain to navigate in chemical gradients via a mechanism largely similar to those of other motile chemotactic bacteria. Interestingly, our comparative analysis reveals a loose correlation between the production of flagella and chemotaxis, as only half (57%) of the Roseobacter strains house genes essential for both traits within their genomes (Tables (Tables11 and and2).2). Therefore, in contrast to Silicibacter sp. strain TM1040, many roseobacters are incapable of chemotaxis signal transduction and thus of sensing and responding kinetically in chemical gradients. This is an intriguing observation, and to our knowledge, such a massive loss of chemotaxis capacity across clade members with maintenance of flagellar motility is unparalleled in other bacterial groups. This suggests that flagella may serve functions other than directional motility and navigation in gradients and that chemotaxis is not essential to every roseobacter lifestyle. The distribution of chemotaxis operons among roseobacter genomes is not correlated with 16S rRNA gene phylogeny (Fig. (Fig.1),1), nor is there any evidence to suggest recent, lateral transfers of these genes to select group members. Thus, we propose that chemotaxis was present in a common ancestor of these strains and subsequently lost from individuals during the evolution of the lineage. Given the extent of chemotaxis gene loss observed among the roseobacter genomes, chemotactic and nonchemotactic roseobacters are expected to use motility differently. This may reflect distinctly different life strategies and/or the niches that they are able to occupy. In certain environmental settings, chemotaxis may be required by some strains to maintain a competitive advantage over other bacterial populations and could be indicative of a more selective niche (e.g., the phycosphere). Motile roseobacters lacking chemotaxis machinery may rely on flagella for dispersion and/or attachment, as has been hypothesized for another marine alphaproteobacterium, Hyphomonas neptunii (2). Alternatively, flagella may be involved in cellular aggregation (79), which could facilitate cells settling out of the water column. A final hypothesis is that some roseobacters may possess completely novel, uncharacterized chemotactic pathways.
The genome sizes of the 28 roseobacters considered here range from 3.0 to 5.5 Mb, and an interesting observation is the pronounced prevalence of chemotaxis genes in genomes of >4.5 Mb (97%, n = 9) relative to genomes of <4.0 Mb (14%, n = 7). Two-thirds (67%, n = 12) of the genomes that fall within the 4.0- to 4.5-Mb range contain chemotaxis genes. Thus, if not crucial to the cell, chemotaxis may be a function readily lost in the streamlining of roseobacter genomes. Another curious observation is the presence of a chemotaxis operon in a strain that lacks genetic evidence for flagellar capability (Maritimibacter alkaliphilus HTCC2654; Table Table1).1). This may be evidence of a different mode of genome restructuring. The sequenced roseobacter strains reveal the full spectrum of potential genotypes: from genomes housing complete sets of genes necessary for functional flagella and chemotaxis to those possessing partial complements of required genes for one or both functions. As a consequence, these organisms represent attractive models for investigating the evolution of motility and chemotaxis and their functional roles in promoting competitive abilities in an ecologically important lineage.
Representative roseobacter isolates demonstrate cellular polarity and/or display of peritrichous polysaccharide fibrils (likely pili) and holdfast structures (see, e.g., references 31 and 32), all of which may be important in the colonization of surfaces. Rosette formation, a type of cellular aggregation in which cells clump by their poles, is common among roseobacters (11) and more prevalent in cultures grown under static conditions (8, 10). While the mechanism by which these rosettes develop is not yet clear, the ability to form rosettes has been correlated with an ability to readily attach to glass in some roseobacters (8). Furthermore, Sagittula stellata E-37 selectively attaches to lignocellulose particles via a polar, holdfast structure (32). Holdfast structures are present in other aquatic Alphaproteobacteria (see, e.g., reference 73), including C. crescentus. In this microbe, holdfasts are located on the tips of cellular stalks and are a key component of the strain's complex developmental cycle but also function in the formation of rosettes (61).
The genome sequences were queried for the presence of genes encoding components of appendages shown to be involved in attachment in other Proteobacteria. All genomes possess a genomic island encoding the machinery required for assembly of adhesive Flp (fimbrial low-molecular-weight protein) pili important in surface colonization, biofilm formation, and pathogenesis in diverse bacteria (87) (Fig. (Fig.1).1). There is strong synteny of seven genes across all roseobacter genomes (cpaBC-ompA-cpaEF-tadBC) that is nearly identical to the cpa (Caulobacter pilus assembly) locus of C. crescentus (80) and appears to be well maintained across most Alphaproteobacteria (e.g., Caulobacter spp., Rhizobium spp., and Parvularcula spp.). The exception is that most other Alphaproteobacteria contain a cpaD (NP_421738) homolog in place of the ompA homolog found in roseobacters. The exact role of CpaD is not known, but it is required for pilus biogenesis in C. crescentus (80). M. alkaliphilus HTCC2654 is the only Roseobacter genome to contain two complete copies of the Flp-related suite of genes; only one shares the previously mentioned gene order. The presence of multiple flp loci has been documented in other bacteria, where the function of the redundancy is unknown but suggested to be linked to distinct physiologies (87). While relatively little is known of the ecology and phenotypic capabilities of M. alkaliphilus HTCC2654 (42), it is clear that this strain is distinct from other roseobacters with respect to this redundancy, and it can be hypothesized that it is reflective of unique cell-surface interactions in this strain. Only two strains possess genes encoding components of a type II secretion system, a mechanism used by some pathogens to translocate virulence factors that may contribute to biofilm formation (43, 46, 64). These genes are housed at discrete, but dissimilar, loci, and both strains possess a gene encoding the pseudopilin PulG found in type IV pili (89). The roles of type IV pili among Proteobacteria are varied, but their expression has been linked to colonization and maturation of biofilms on different surfaces (see, e.g., reference 63). Half of the genomes contain a genetic locus encoding components of type IV secretion systems (T4SS), and a quarter possess multiple (two to three) copies (see Table S1 in the supplemental material). A core organization of genes is highly conserved (virB2B3B4-COG0741-virB8B9B10) among the strains and is often maintained across homologous operons within a single genome. The prevalence of T4SS among roseobacters has been previously documented, and it has been hypothesized that they may play important roles in facilitating interactions between these bacteria and phytoplankton (58, 65), but this has yet to be demonstrated. In the three fully sequenced genomes possessing these genes, the cassettes are located on plasmids, suggesting that the T4SS of roseobacters are analogous to that of the A. tumefaciens VirB/D4 type (17). Holdfast-related genes were not readily apparent; homologs of gene clusters encoding holdfast attachment proteins in C. crescentus, including hfaABD and hfsDAB (39, 81), are absent from the Roseobacter genomes. Thus, rosette formation in roseobacters likely proceeds via a manner distinct from that described in C. crescentus. Identification of the appendages or surface structures involved in forming Roseobacter rosettes will provide insight into how different cell surface features result in behaviorally similar cellular aggregates in phylogenetically and morphologically distinct microbes.
We have found no obvious correlations between possession of specific appendage-related genes (such as those described above) and surface colonization, or even rosette formation, as determined in laboratory studies. This suggests that different cellular structures are used for surface colonization and rosette formation and/or that different environmental conditions are required for expression of necessary genes and their associated phenotypes across this collection of strains. Regardless of this discrepancy, this genome analysis does highlight the universal distribution of genes encoding adhesive Flp (fimbrial low-molecular-weight protein) pili, suggesting that these appendages may be a core feature of roseobacter biology and that their function in mediating surface formation or rosette formation can be hypothesized.
QS is known to control a number of physiological responses in diverse bacteria (53); chiefly among these are mediation of surface colonization and production of secondary metabolites (21, 24, 40). Experimental evidence for QS systems has been demonstrated in Roseobacter lineage members, and genomic evidence for this form of cell-to-cell communication has been recognized previously (8, 33, 57-59, 93). In fact, recent surveys suggest that roseobacters may be the dominant producers of the common bacterial signaling molecules collectively referred to as acyl homoserine lactones (HSLs) in the marine environment (18) and that they produce a myriad of HSL congeners (93). Of particular interest is a recent report showing that S. pomeroyi DSS-3 produces a novel class of HSLs derived from aromatic compounds (likely plant derived) rather than fatty acids (77). S. pomeroyi DSS-3 was isolated from waters adjacent to expansive salt marsh estuaries that are likely abundant sources of the precursor of this novel HSL. Evidence that QS systems facilitate colonization in roseobacters is not yet definitive, but correlative data are available. For instance, Rao et al. have shown that P. gallaeciensis strain 2.10 produces acyl-HSLs and suggest that these QS molecules are instrumental in colonization success (71). In the HSL-producing Phaeobacter sp. strain 27-4, colonization has been linked to the ability of these organisms to grow in a multicellular rosette form, a condition that is affected by growth conditions (static versus stationary) and correlated with production of the antibacterial compound tropodithietic acid (TDA) (8, 9). Finally, preliminary work by Cicirelli et al. suggests that QS may mediate motility in a Roseobacter sponge symbiont (18).
Twenty-two of the genomes house canonical HSL signaling systems of the AI-I type prevalent in many Proteobacteria and comprised of an HSL synthase gene (COG3916) and a LuxR-family transcriptional regulator (IPR000792) (see Table S1 in the supplemental material). Half (50%) of the genomes contain multiple copies (two to three) of the synthase genes. Of the 43 putative HSL synthases identified, 30 display a common luxRI-type QS organization; they are found immediately adjacent to a gene encoding a LuxR-family transcriptional regulator (53). In all but one case (Roseovarius sp. strain 217) the luxR gene is found on the same strand as is the putative synthase gene. In seven of the remaining instances, the HSL synthase gene is found immediately adjacent to, and divergently transcribed from, a signal transduction histidine kinase (COG0642); in three cases no ORF having homology to known regulatory elements is found adjacent to the HSL synthase; and in three cases non-LuxR-like transcriptional regulators (COG3706, COG2197, and COG0583) are immediately adjacent to the synthase gene. Five of the strains appear to possess two distinct copies of the luxRI-type QS systems: D. shibae, Rhodobacterales strain HTCC2150, Roseobacter litoralis, S. pomeroyi, and Phaeobacter sp. strain Y4I. Here again the number of QS systems harbored by a strain is not consistent across members of the same genus (i.e., Roseobacter, Silicibacter, and Phaeobacter), suggesting that the cell-to-cell signaling mechanisms may not be phylogenetically conserved. The production of non-HSL LuxS-type furanosyl borate diesters is not evident in roseobacter genomes. Nor is there genetic evidence for production of other well-characterized quorum molecules common to gram-positive bacteria (e.g., peptides). Finally, the roseobacter genomes contain a number of “orphan” luxR genes that are not physically linked with a cognate luxI and may indicate that roseobacters are equipped to sense a myriad of chemical cues from other marine bacteria.
As mentioned above, little experimental evidence exists as to the genetic elements that might be regulated by HSL QS systems in roseobacters, and the genome sequences provide few obvious clues. Furthermore, a correlation between possession of a QS system and biofilm formation is not perfect. For example, Silicibacter sp. strain TM1040 is a prolific surface colonizer in laboratory studies but does not harbor a recognized QS system (8). However, as Silicibacter sp. strain TM1040 displays phenotypic traits consistent with density-dependent behavior, it is tempting to speculate that this strain employs a yet-to-be discovered QS system. Both S. lacuscaerulensis and S. pomeroyi possess multiple HSL QS systems but differ in their abilities to readily colonize surfaces (Table (Table1).1). Thus, distinct surface colonization strategies likely exist within this genus alone.
In addition to cell-to-cell signaling, intracellular signaling mediated by the universal bacterial secondary messenger bis-(3′-5′)-cyclic dimeric guanosine monophosphate (c-di-GMP) has also been demonstrated to play important roles in biofilm formation, motility, and multicellular behavior in diverse bacteria (38, 95). The genomes were searched for ORFs containing domains known to be involved in c-di-GMP regulation (i.e., EAL, GGDEF, and HD-GYP domains) (72). The abundance of these domains in a given genome ranged from 3 to 11 with an average of seven (see Table S1 in the supplemental material). Strains for which interactions with eukaryotic hosts have been unequivocally shown, Silicibacter sp. strain TM1040 and the two P. gallaeciensis strains, had the greatest number (11) of domains. Two strains isolated from surface waters of the oligotrophic open ocean, Rhodobacterales strain HTCC2150 and Sulfitobacter strain NAS-14.1, were among the strains with the fewest (four) domains. The Roseobacter average is less than the number found in nutritionally versatile freshwater alphaproteobacterial genomes (e.g., C. crescentus CB15  and Rhodobacter sphaeroides ) but more than the number found in the streamlined genomes characteristic of marine lineages geographically limited to the oligotrophic open ocean (e.g., Prochlorococcus spp. [none] and Pelagibacter ubique [none]). The variation in the number of putative intracellular signaling domain proteins among roseobacters does not appear correlated with genome size. In fact, the two Sulfitobacter strains each possess four domains and differ in their genome content by 0.5 Mb. As of yet, no obvious correlations between domain abundance and surface colonization abilities, motility, or rosette formation are evident either. It has been suggested that the relative abundance of EAL, GGDEF, and HD-GYP domain proteins is a reflection of an organism's environmental responsiveness, as specific domain proteins appear dedicated to specific environmental conditions (76). If this is indeed the case, the variation in domain abundance across Roseobacter lineage members may be reflective of, or linked with, environmental niche and life strategy, as hinted at above. Along those same lines, the strains with the greatest number of these intracellular signaling domains are also chemotactic, while the four strains with the fewest lack chemotaxis genes. While the exact biological significance of this correlation remains unclear, it is obvious that studies directed at addressing the role of both cell-to-cell and intracellular signaling in roseobacter activities are necessary if we are to understand how environmental cues influence these organisms in the natural environment.
Competition and cooperation among microbes play an important role in the development of assemblages on surfaces in natural environments (85). Previous studies have demonstrated the ability of representative roseobacters to produce secondary metabolites with antimicrobial activities against a variety of marine bacteria (see, e.g., references 6 and 92), pointing to a competitive advantage for these organisms. It has been reported that this competitive advantage may have practical application. Studies have promoted the use of roseobacters as probiotics in aquaculture facilities (37) or as antifouling agents in marine systems (24, 69). The production of inhibitory compounds, or antagonistic behavior, appears correlated with growth on surfaces. Particle-associated representatives of the Roseobacter clade were demonstrated to be >10 times more likely to produce antimicrobial compounds than were free-living members (45). P. gallaeciensis strain 2.10, referred to as an “aggressive” colonizer, has been demonstrated to invade preestablished biofilms of marine bacteria on the alga Ulva australis (71). Silicibacter sp. strain TM1040 and some Phaeobacter strains are known to produce TDA, a sulfur-containing seven-member aromatic tropolone ring compound with antibiotic properties presumed to be the active agent of well-documented antagonistic behavior in these strains. All of these strains were isolated from the surfaces of eukaryotic organisms (Table (Table1),1), and TDA production is elevated when the strains are grown under static conditions and form biofilms (6, 10). However, possession of the genes encoding TDA production does not appear to be a universal trait of antagonistic roseobacters (29), including the two other Silicibacter strains for which genome sequences are available and Phaeobacter strain Y4I. As these latter strains were all isolated from surface waters and not obviously associated with any colonizable surface, this may indicate a link between TDA production and interactions with eukaryotic organisms. These findings point to the production of alternate and varied antimicrobial molecules by group members and underscore the possibility that members of the roseobacters represent a source of novel antimicrobial compounds. This is supported, in part, by the recent identification of a bipyridyl pigment specific to Phaeobacter sp. strain Y4I whose production is correlated with antagonistic behavior (J. P. Mooney and A. Buchan, unpublished data).
Polyketide synthases (PKS) and nonribosomal peptide synthases (NRPS) are prevalent among diverse bacteria, where they have been demonstrated to be essential in the production of secondary metabolites, some antimicrobial in nature (84). A previous report identified PKS and NRPS genes in Roseobacter strains by using a PCR screen and showed a correlation between their presence, antagonistic behavior, and production of HSLs in Phaeobacter isolates (50). Over half (61%) of the Roseobacter genomes analyzed here contain homologs that likely encode PKS and/or NRPS proteins (Table (Table1).1). Most (82%) of these genomes contain homologs for both PKS and NRPS, and six contain multiple copies of these genes (see Table S1 in the supplemental material), suggesting that these strains are capable of producing a myriad of structurally diverse compounds. In fact, O. indolifex HEL-45 is known to produce upwards of eight bioactive metabolites (92). Generally speaking, the cadre of natural products produced by roseobacters have been largely unexplored and likely underestimated due to the known dependence of secondary metabolite production on appropriate cultivation conditions, including the presence of other bacteria (23, 34, 75). Yet, with even a limited knowledge of the spectrum of Roseobacter products, it is evident that this lineage is as an untapped resource for novel compounds with practical application. The extent to which the production of structurally diverse inhibitory compounds dictates competition and cooperation among roseobacters in mixed-species biofilms remains to be seen but will undoubtedly provide a fascinating view of allegiances among this phylogenetically cohesive but metabolically heterogenous lineage.
The ability of roseobacters to form associations with surfaces (either living or nonliving) is arguably a defining trait of lineage members. Collectively, Roseobacter genomes harbor genes encoding functions expected to be pivotal in surface attachment and colonization. Genomic data often corroborate phenotypic data but also reveal significant variation in terms of gene content and synteny among group members, even among closely related strains (congeners and conspecifics). Perhaps the most striking is illustrated by the Sulfitobacter strains that share nearly identical 16S rRNA genes but show surprising variation in genomic content with regard to the functions analyzed here (Fig. (Fig.1).1). This may be explained, in part, by the substantial difference in genome size between Sulfitobacter sp. strain NAS-14.1 and Sulfitobacter sp. strain EE-36, 4.0 Mb and 3.5 Mb, respectively, or possibly the different environments in which they are found (i.e., coastal versus open ocean). Similarly, genomic variations are also evident between the two P. gallaeciensis strains and among species of the Silicibacter and Oceanicola genera that are not accompanied by major differences in genome size but often do coincide with isolation from different environments of various geographic and environmental parameters (Table (Table1).1). These variations may reflect the versatility necessary for clade members to predominate in disparate and varied environments but also provide a unique opportunity for detailed studies of the evolution of specific functions in one of the most dominant marine lineages. Future studies of representative strains will likely reveal the varied strategies employed by this diverse group and help elucidate the molecular underpinnings and environmental cues that dictate interactions between roseobacter cells and surfaces in marine environments. This knowledge is critical if we are to understand how transitions between planktonic and sessile lifestyles influence Roseobacter metabolic activities that contribute to biogeochemical cycles and make them valid candidates as beneficial bioagents.
We are grateful to G. M. Alexandre, M. A. Moran, and G. LeCleir for providing valuable comments on drafts of the manuscript. We thank the Gordon and Betty Moore Foundation for providing genome sequence for most of the strains.
The work in our lab is supported by grants from the National Science Foundation (MCB-0534203 and OCE-0550485).
Rachael N. Slightom is a microbiologist working as a research associate in the Insect Control Discovery group at Monsanto Company in St. Louis, MO. She received her bachelor's degree in biology from the University of Evansville in Evansville, IN. It was there that she was first introduced to the captivating field of microbiology, particularly microbial ecology and how it impacts the environment. She is interested in learning as much as possible about the roles of microbes in nature. She obtained her master's degree in microbiology from the University of Tennessee.
Alison Buchan is an assistant professor of microbiology at the University of Tennessee, where her research group studies various aspects of Roseobacter biology. She was first introduced to this fascinating group of marine heterotrophic bacteria as a graduate student under the direction of Mary Ann Moran at the University of Georgia. She later received an NSF Postdoctoral Fellowship in Microbial Biology and was fortunate to work on catabolism in a soil-borne bacterium, Acinetobacter, with L. Nicholas Ornston at Yale University.
Published ahead of print on 7 August 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.