Comparative genome analysis of the α-proteobacteria: Relationships between plant and animal pathogens and host specificity

The phylogenetic group of α-proteobacteria contains bacterial species with a wide variety of lifestyles, including obligate intracellular (Rickettsia), facultative intracellular (Bartonella, Brucella), and extracellular pathogens (Agrobacterium), as well as symbionts of both animals and plants (Wolbachia, Sinorhizobium). With the recent completion of genome sequences of several of these organisms, it is now possible to explore the genetic basis for these biological differences. The newest addition to this small group of complete α-proteobacterial genome sequences is that of Brucella suis, described by Paulsen et al. (1) in this issue of PNAS. B. suis is an intracellular pathogen of swine, where it causes late-term abortions. This infection can be readily transmitted from swine to abattoir workers and farmers via aerosol, thus causing a febrile illness. The efficiency of its transmission via aerosol prompted the military development of B. suis as a biological weapon (2). Although the Brucellae are well known as animal pathogens and are considered to be potential agents of bioterrorism, until recently, most of the research on these organisms had been focused on eradication of brucellosis in cattle and swine and was limited to epidemiology and development of vaccines. As a result, our understanding of the basic biology of this organism is still in its infancy. The completion of the second Brucella genome sequence, therefore, provides a powerful tool for the Brucella research community to address questions on the biology, ecology, and pathogenesis of this group of organisms.

Each of these species is adapted to different hosts, yet the genomes of the two Brucella species differ by only 74 genes.

Although B. suis has been isolated from cattle, it has not been found to cause abortion in this host as it does in swine (3–6). Brucella melitensis, on the other hand, has been reported to cause abortion in cattle as well as in sheep and goats (7). A question that has interested researchers in the infectious disease field for years is, what is the genetic basis for host specificity? Some studies have shown that the initial interaction of bacteria with host tissues is an important determinant. Bacterial surface proteins, such as K88 pili of enteropathogenic Escherichia coli or internalin of Listeria monocytogenes, have been shown to bind specifically to tissues of their preferred hosts (8, 9). Accordingly, closely related bacterial species that lack these binding proteins are unable to cause disease (10). These findings suggest that bacterial host range factors can be identified by comparing the genomes of closely related pathogens that differ in their host range to find the genes unique to each pathogen. Comparison of the genomes of B. suis and B. melitensis yielded a surprising finding: whereas each of these species is adapted to different hosts, the genomes of the two Brucella species differ by only 74 genes (1, 11). B. suis contains 42 unique genes located in 22 chromosomal regions (designated genetic islands), whereas B. melitensis contains 32 unique genes on 11 islands (1). This number is quite small when compared with the differences found between pairs of host-adapted Xanthomonas or Salmonella species (Fig. 1; refs. 12 and 13). For reference, the Salmonella enterica serotype Typhi genome contains 601 genes (on 82 genetic islands) that are absent from S. enterica serotype Typhimurium, whereas Salmonella typhimurium contains 479 genes (on 80 genetic islands) that are unique relative to Salmonella typhi (13). In fact, the proportion of genes unique to the B. melitensis or B. suis genomes is much closer to that of closely related obligate intracellular bacteria, such as different Buchnera or different Chlamydia species (Fig. 1; refs. 14 and 15). It has been suggested that the low rate of genetic exchange among obligate intracellular bacteria can be attributed to their isolated location within cells of their hosts (16). The paucity of genetic exchange observed in Brucella suggests that despite their metabolic capability to grow outside the host, these pathogens spend most of their time in an intracellular location and persist only transiently in the environment between hosts. The success of programs in the U.S. and other countries for eradication of bovine brucellosis also argues against prolonged environmental persistence of Brucella.

Figure 1.

Results of comparative genomic analyses between closely related bacterial species that differ in pathogenicity or host preference. Each bar represents the number of unique genes relative to its companion genome (top vs. bottom bars) as a percentage of all ORFs in the genome. The number above each bar indicates the total number of unique genes in each genome (1, 10, 12, 13, 15, 33). *, excludes genes located on one prophage in Listeria monocytogenes and on five prophages in Listeria innocua.

The small number of genetic islands in the Brucella species makes them an attractive model for examining the contribution of their unique genes to host specificity. Which host–pathogen interactions might be determinants of host range? The pathogenic Brucella species are known to cause systemic infection in a variety of animal hosts, including humans, where the bacteria traverse epithelial barriers and localize to the reticuloendothelial system to cause infection. Similarly, in vitro, Brucella spp. are able to survive in macrophages from different species, including mice and humans (34–37). In contrast, the ability of Brucella species to cause abortion is restricted to one or a few hosts, suggesting that factors that contribute to tropism and growth in the placenta may determine whether a particular Brucella species can cause abortion in its pregnant host. For example, interspecies differences in morphology and function of the placenta, including nutrient transport functions and maternal recognition of pregnancy, have been documented (17–19). Thus, bacterial genes involved in host restricted disease phenotypes may encode factors that specifically enhance localization to or rapid growth within the placental cells of a particular host. Several of the genes unique to B. suis encode putative transport systems or outer membrane proteins that could contribute to these functions (1).

In addition to unique genes, point mutations may contribute to the differences in host specificity between B. suis and B. melitensis. These point mutations may lead to truncations or frameshifts in genes that may play a role in host–pathogen interactions. Finally, the role of point mutations that change the amino acid sequence of a protein should also be considered, as it has been shown that a single amino acid change in the E. coli FimH adhesin leads to loss of collagen-binding activity (20). Some of the variable surface proteins identified in the comparative genome analysis of B. suis and B. melitensis, therefore, may be additional candidates for host-specific virulence factors. With only 7,301 single nucleotide polymorphisms identified between the B. suis and the B. melitensis genomes, it should also be possible, by using the genome sequence data, to design experimental approaches aimed at assessing the contribution of point mutations in coding genes to host specificity of the Brucellae (1).

Comparative genome analysis between B. suis and members of the family Rhizobiaceae, which includes Agrobacterium tumefaciens, Sinorhizobium meliloti, and Mesorhizobium loti, revealed similarities in metabolic capability and genome structure (1). This result extends and confirms results from several groups that have identified virulence factors in Brucella, whose counterparts in S. meliloti or A. tumefaciens are required for endosymbiosis or pathogenesis in plants. For example, a two-component regulatory system, BvrR/BvrS, has been shown to be required for intracellular replication of Brucella abortus (21). Its homolog in Agrobacterium, ChvG/ChvI, is required for tumor formation in plants, and the S. meliloti homolog ExoS/ChvI regulates production of succinoglycan, which is crucial for the development of endosymbiosis (22, 23). Cyclic β-1–2-glucan, a polysaccharide produced by S. meliloti, A. tumefaciens, and B. abortus also contributes in these three species to host–pathogen interactions, as mutants are impaired in their persistence in the host (24–26). Similarly, a cytoplasmic membrane transporter, BacA, was shown to be required for both symbiosis of S. meliloti with alfalfa and survival of B. abortus within the macrophage (27). More recently, a type IV secretion with homology to the A. tumefaciens virB genes was identified in B. suis and B. abortus (28, 29). The virB genes are essential for tumor formation on plants by A. tumefaciens, and similarly, the B. abortus virB genes are required for infection in animals (28, 30). Although the precise functions of all of these shared proteins in virulence remain to be elucidated, these findings suggest that animal pathogens, plant symbionts, and animal pathogens all use common mechanisms for interactions with their hosts, and that these may have been adapted to respond to signals in the plant and animal environments.

Unlike A. tumefaciens, M. loti, and S. meliloti, B. suis does not contain any extrachromosomal DNA. However, B. suis chromosome 2 contains numerous homologs of genes located on the linear chromosome of A. tumefaciens or on plasmids of S. meliloti. These include the virB genes, encoding a type IV secretion system and conjugation genes trbL-traI (1, 29). Together, these findings suggest that parts of chromosome 2 may have been originally acquired from a plasmid. The recent identification of the closest sequenced homologs to date of the Brucella virB genes on conjugative plasmids from microbes of the rhizosphere of wheat and alfalfa lends further support to this idea (31, 32).

In analyzing a vast amount of comparative data, Paulsen et al. (1) have managed to give us an overview of the complex relationships between the genomes of the α-proteobacterial species and how we can use the genomic data to gain insight into the ecology and biology of a poorly characterized organism. The completion of genome sequences of a third Brucella species, B. abortus, and its close phylogenetic relative Ochrobactrum anthropi, an opportunistic pathogen, will allow researchers to extend the comparative genomic analysis presented here to better understand the contribution of genomic differences to the diverse interactions of the different α-proteobacterial species with their hosts.