Genomic Islands Of Brucella Instability Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Brucellosis is one of the most important zoonosis worldwide, affecting people from developing countries and causing important economical losses [1]. The disease is produced by gram negative, intracellular facultative bacteria of the genus Brucella [2]. Classically, Brucella species are recognized based on the preferential host range and other phenotypic features [3]: B. abortus affects primarily cattle and other bovidae as well as cervidae; B. suis is isolated from swine and several forms of wild life; B. melitensis infects mainly sheep and goats. B. canis is found in dogs; B. neotomae is found in desert woodrats; and B. ovis is responsible for epididymitis in rams. In addition, three new species have been proposed recently: B. microti, isolated from common voles [4] and B. ceti and B. pinnipedialis, both isolated from marine mammals [5].

B. melitensis, B. suis and B. abortus [6-9] show a genome of 3.3 Mb. No classical virulence factors such as capsules, exotoxins, cytolysins, pili or fimbriae, etc. have been found [10]. Nevertheless, these bacteria are capable of infecting host macrophages, evading the innate immune system and establishing a replicative niche within cells [2]. In spite of the restricted ecological niche occupied by these intracellular pathogens, there is evidence of horizontal gene transfer (HGT) [11-14].

Genomic islands (GI) are among the DNA sequences transferred by HGT [15]. These sequences are large chromosomal regions that confer new phenotypic traits upon transfer-acquisition, often increasing bacterial fitness under hostile environmental conditions. Most of the reported GI sequences contain an atypical nucleotide composition in comparison to their host chromosome and often they are found delimited by short direct repeat sequences (DR) [16]. Two loci involved in lipopolysaccharide (LPS) biosynthesis with features of GI have been described in B. melitensis [12, 14] and the presence of a GI also related to LPS structure was reported recently in B. abortus. Interestingly, this region is absent in B. ovis [13], suggesting that GI might lead into spontaneous genome deletions through recombination events, causing genome instability and variability [15]. At present, there is a single recognized unstable GI in B. suis, B. canis, B. neotomae and marine isolates. This GI resides in the small chromosome and carries phage-related and tra genes, the latter related to conjugative machinery [17]. In addition, a whole genome microarray comparison using a chip based on B. melitensis 16M genome showed the presence of additional GI-like regions in different Brucella species [11]. Nevertheless, no experimental evidence has been presented as yet that relates GI loci with instability in Brucella.

In this study, we performed a search for GI-like loci in B. abortus 9-941 using a bioinformatic prediction which combines the presence of genomic signature deviation and flanking repeats. We identified eight GI-like loci ranging from 4-21 kb, all flanked by direct repeats, and found that these sequences are unstable in bacteria grown in rich medium, releasing a circular intermediate (CI) and leaving behind a chromosomal scar. Sequencing of CI-derived fragments suggested a site-specific recombination mechanism involved during GI excision. Altogether, our observations provide evidence for a genome plasticity linked to GI excision events in species of the genus Brucella.


Identification of the GI in the B. abortus genome

The whole genome scanning of the B. abortus genome identified 10 putative regions with genomic signature deviation in chromosomes I and 6 in chromosomes II (Figure 1, Additional file 1), Although most of these regions showed GC% deviation, eight lack of flanking DR sequences. These loci corresponded to clusters of ribosomal gene, flagellar genes and GBAI3, a(genomic island of Brucella abortus chromosome I (for a complete description of the genetic content of GI see the Additional File 1).

The remaining eight loci showed a good prediction value for flanking DR, with lengths that varied from 21 to 82 bp (see Additional File 2). The presence of DR was considered as indicator of instability since DR are spots for site- specific recombination events (Figure 2, Additional file 2). The prediction results identified three putative GI previously reported by DNA microarray comparison among the six classical Brucellae species [11] Accordingly, the GI nomenclature was maintained (Table 1). The different GI are characterized by the type of sequence information that they carry. GI-1 codes mostly for hypothetical proteins (HP) and mobility genes, such as a phage-like integrase and invertase gene. GI-2 includes the LPS wboA-wboB glycosiltransferases genes, a copy of the IS711 transposase, several HP and a pseudogene putatively coding for the Omp25, an immunogenic protein, functional in B. melitensis [13]. Moreover, GI-3, one of the largest (of ~21 kb ) carry 28 ORF mostly of HP and a protein with a TIR domain [18]. Although most of the GI identified were previously reported [19, 20], the identification of flanking DR defined the boundaries of each GI. The main characteristics of the GI identified in B. abortus are summarized in Table I. GBAI1 comprises seven phage-related ORF including an active gene for integrase. GBAI2 matched with the previously reported wbk locus, involved in the biosynthesis of LPS [12]. This GI is flanked by transposases and harbors eight ORF, six of which are related to LPS biosynthesis. The GBAI4 locus encodes three ORF (a transcriptional regulator, a putative amino acid transporter, and a small HP). This loci is flanked by the insertion sequence IS2020 and has already been considered a HGT acquisition [21]. The second largest GI is GBAI5 with a length of 19.6 kb. This locus is particularly rich in information as it carries 14 ORF, a gene for DNA repair enzyme XthA-2, a putative two-component system, a transcriptional regulator of the MerR family, a cadmium transporter (CadA), several HP and two ORF annotated as pseudogenes, one coding for an adhesin and the other for an extracellular Ser-protease. One putative unstable GI was identified in chromosome II, GBAII1. This GI does not present a markedly deviated GC content and carries DR flanking sequences suggesting its instability. Except for GBAI2 and GBAI5, all the GI described are found integrated near to a tRNA genes (Table 1) Searches carried out on GenBank showed that these GI are found exclusively in brucellae.

Distribution of GI among reference strains

The genome sequences of representative strains of B. melitensis, B. suis, B. abortus and B. ovis available at the Genebank, were scanned to identify the location of the GI. Each GI was confirmed by long range PCR (Table 2). All the GI were present in B. abortus 2308 and B. melitensis 16M. However, GI-1, GI-2 and GBAI4 were absent in B. ovis 63/290, and GI-3 absent in B. suis 1330 (data not shown). , Of interest is the finding of a PCR product consistent with the presence of GBAI4 in the genome of the B. melitensis 16M strain. This GI is not included in the genome sequence deposited in GenBank for this strain [7].

GI instability assessment

The detection of flanking DR and phage-like integrases encoded in the Brucella GI suggested that these loci may be unstable. In support of this possibility, long range PCR was performed to detect both the complete GI and the chromosomal scar left after excision. The results yielded not only the expected GI amplicon, but also a product with a molecular size corresponding to the scar (data not shown). To identify the putative released GI, a PCR targeting the rejoining point of the chromosomal scar was performed (Figure 3A). Sequence analysis of PCR products derived from scars, confirmed GI deletions (Figure 3B). To confirm the formation of circularized forms of GI, we carried out the CI detection using an inverse PCR strategy with primers pairs oriented towards the right and left GI boundaries. As expected, a single product for each CI was obtained (Figure 4A). In all cases, cloning and sequencing of these PCR products showed that the CI maintain a copy of the GI-flanking DR that confirm the circularization A graphic representation of all CI detected is shown in Figure 4B.

Expression of integrase/recombinase genes

Comparative analysis of the GI loci indicates a similar gene content and genomic organization in all sequenced Brucella species According to the annotations, GI-1, GI-2, GI-3 and GBAI1 harbour a single phage-like integrase/recombinase gene (int) each, with the exception of the GI-1 of B. melitensis 16M that seems to encode an integrase of 82 aminoacid. To determine whether GI excision is linked to the activity of the putative integrases, we performed reverse transcription PCR analyses targeted the int genes. Using RNA extracted from three brucellae species grown under standard conditions, we detected integrase expression for GI-2 and GI-3 but not for GI-1 and GBAI1 (Figure 5A). This result suggests that there is variability in the expression level of int genes. This interpretation is supported by the fact that the GI-1 and GBAI-1 int products were only detected using real time RT-PCR (Figure 5B). Noteworthy, although the annotation of GI-1 int in B. melitensis 16M (the BMEI0907, a HP) predicts a protein a small protein and inversely oriented, a 400 bp amplicon was also detected. Therefore, this experimental result demonstrates an error in the annotation of BMEI0907.


Lateral transfer of GI is part of the complex mechanism that determines the pathogenicity of bacteria, depending on the ecological niche where the bacteria lives in. The classic example is the GI-encoded adherence factors (fimbriae) of Escherichia coli which in the kidney and urinary tract contribute to the pathogenicity of the bacteria; however, in the intestinal flora these factors are non-pathogenic and E. coli becomes part of the normal flora [22]. Thus, unless changes occur in the bacterium or the host that leads to pathogenicity, HGT events contribute to a successful symbiotic survival.

Our aim was to search for the GI in the genome of B. abortus to study its contribution to genome stability and pathogen physiology. We initially used a bioinformatic search based on the difference on GC content between the genomic island and the host DNA and the difference on genomic signature (based on dinucleotide frequencies values). Another criterium was the insertion of the GI in a tRNA gene and the presence of flanking DR. To avoid amelioration, that is, the lost of a GI due to its similarity in GC content and codon usage to that of the host, we used all of the above mentioned criteria for GI identification.

Our results showed 16 hypothetical GI in B. abortus of which, eight contained DR (Fig 1). Among the GI lacking DR, three corresponded to ribosomal gene clusters. These genes show a distinctive codon usage, different to the host DNA, suggestive of upregulated expression. Ribosomal genes along with other genes e.g transcriptional factors and chaperones are frequently classified as "alien" by compositional analysis techniques, but due to biological significance, it is unlikely that they are subject to HGT [23, 24]. The other five GI might have originated from ancient HGT which became stable sequences as result of mutations at GI boundaries or in genes related to DNA mobility (15). Yet, some of these regions, i. e. the virB locus and the genes for flagellar proteins, outer membrane protein, DNA repair enzymes and redox complex (27-39), are essential to Brucella virulence. Likewise, the DR-flanked GI might be considered as recent acquisitions. In addition, the sinteny displayed by these loci among brucellae suggests that HGT events must have occurred before speciation.

Brucella, like other intracellular parasites, exhibits considerable clonality. Clonality, however, does not mean genetic stability and, in this work we describe 8 regions displaying instability. Three loci, GI-1, GI-2 and GI-3, matched GI reported previously [11]. Although they do not constitute a prophage, GI-1 encodes phage related genes, and mutation analysis performed on several genes show a lack of involvement in virulence [25]. GI-2 carries two genes which participate in O-chain LPS biosynthesis: wboA and wboB. It has been shown that most genes in GI-3 encode HP and, with the exception of the integrase, are expressed [26]. Thus far, only one gene in this island has been associated with pathogenicity: BruAb1_0274 or btp1, which presents a TIR domain and is implicated in the downregulation of the NF-B signaling cascade, avoiding the maturation of host dendritic cells[18]. GBAI1, like GI-1, encodes several phage-related genes and HP. Transposon mutagenesis identified an ORF of this island in B. suis (BMEI1658) as essential for intracellular replication [27]. GBAI2 encodes six (gmd, per, wbkB, wzm, wzt, wbkC) of the ten wbk genes necessary for LPS O-chain biosynthesis and transport across the envelope. Interestingly, the detection of the corresponding CI demonstrated the instability of a tandem of the six wbk genes in GBAI2. This CI also included the 3' end insertion sequence but not the flaking wbkA gen. This gene is also essential for LPS biosynthesis and it is found close to an IS711 copy. Both, wbkA and IS711 are flanked by imperfect insertion sequences containing a putative DR pair, suggesting that the wbk locus has resulted from more than one acquisition event. The smooth LPS is a determinant virulence factor of B. abortus, B. melitensis and B. suis, and the dissociation of the smooth bacteria into rough attenuated mutants lacking the O-chain is a frequent event in vitro. Therefore, spontaneous deletions of GI related to LPS might be involved in the emergence of rough phenotype.

On the other hand, GBAI4 encodes phage related genes, and both GBAI4 and GBAII1 have putative transporters. These genes may be involved in metabolic fitness inside the host. Finally, Wattam et col [20] have reported the existence of a putative island named SAR 1-17, which corresponds to the GBAI5 identified in this work. This island harbors genes with similarity to those encoding virulence factors such as a putative adhesin and extracellular protease. However, these genes are inactivated in B. abortus. One possibility is that these pseudogenes in GI loci of B. abortus but not in B. melitensis relate to the differences in the pathobiology of these two species. In this context, those inactivations affecting hypothetical virulence genes may represent advantageous genetic changes that positively impact on a better adaptation to a specific host. Further studies with a large number of strains would be necessary to assess if these genes are actually pseudogenes derived from selective inactivations or events related to a genetic decay of GI.

Concerning the instability, the experimental evidence suggests the excision of the four GI encoding int genes is likely to be mediated by the cognate integrases. Even though expression of GI-3 int in B. abortus and B. melitensis grown on trypticase soy broth has not been detected before [26], we demonstrated the expression of the int genes of GI-2 and GI-3 by RT-PCR using RNA from bacteria grown under standard conditions. On the other hand, the expression of GI-1 and GBAI1 integrases could only be detected by real time RT-PCR. These different levels of expression suggest that some integrases are expressed constitutively whereas others may be downregulated. Another interesting result was the identification of the misannotation for BMEI0907, the GI-1 integrase in B. melitensis. Automatic ORF prediction by Glimmer using the sequence of B. melitensis 16M as template showed an ORF located on the +3 reading frame encompassing 1305 bp (Additional File 3). The gene would be similar in both size an orientation to its B. abortus orthologue, as shown by the detection of a PCR product derived from cDNA with identical molecular size to the amplicons obtained for B. abortus and B. suis. The expression of int genes in other brucellae suggests that GI circularization (and deletion) may not be restricted to B. abortus genome.

The instability of the four GI not encoding int genes should be explained by alternative mechanisms. Recombination between DR is a conserved mechanism that contributes to phenotypic diversity in prokaryotes [28]. It has been demonstrated that RecA catalyzes the homologous recombination between long DNA segments displaying sequence similarity [29]. GBAI4 is flanked by DR which are part of IS2020. Thus, GBAI2 is flanked by several copies of degenerated IS. Nevertheless, sequencing of PCR products that include putative attP sites derived from the corresponding CI indicates site-specific recombination events. One possibility is that there is a similarity between flanking DR of unrelated islands such that offers recombination sites recognized by unrelated integrases, as reported in other bacteria [29]. However, with the exception of GBAI5, sequence analyses of all DR of Brucella GI resulted negative for the presence of a similarity that could account for the necessary short canonical inverted repeats sequences. In fact, the DR of GBAI5 island contains internal palindromic repeats of 7 bp separate by 11 bp which resembles a point of recombination. Interestingly, sequence analysis indicates that such DR belongs to Bru-RS repeat family reported previously as a hot-spot insertion for IS711 element [30]. Further mutational analysis of int genes and quantitative studies using recA mutants will be necessary to determine the specificity of Brucella integrases and the possible role of RecA in GI instability.

From an evolutionary point of view, the genomic instability displayed by the GI explains several deletions reported among the genomes of the six different brucellae [11]. GI-1, GI-2 and GBAI4 are absent in B. ovis and GI-3 is not present in B. suis and B. canis. Our work also draws attention on the existence of significant variations in the genomes of collection strains. We found that GBAI4 was present in the B. melitensis 16M strain conserved at SAG and also detected the scar resulting from its excision. This instability can account for the absence of the corresponding sequence in the 16M genome sequence available in GenBank, an observation that stresses the variability of Brucella in vitro. Despite the lifestyle that keeps the bacteria restricted to an ecological niche genetically isolated from foreign DNA acquisitions, Brucella has the potential to circularize several GI as shown by the CI detection. GI circularization has been reported in free living microorganisms and it is considered the first step in the potential horizontal propagation, thus contributing to genome plasticity and bacterial evolution [15, 31]. An intriguing question arises from the results; what is the biological significance of GI circularization in an intracellular parasite like Brucella, if any? Since the brucellae are not naturally susceptible to transformation and neither transduction nor conjugation have been reported [32], it is likely the circularization is not associated to HGT of GI. Probably, the selective pressure to maintain GI is lessened under rich growth conditions, favoring the formation of CI. Quantitative experiments done on biological samples derived from infected animals will help to determine whether island excision contributes to virulence attenuation in vivo.


The results show that the intracellular pathogen Brucella carries several unstable GI. The instability displayed by the GI strongly suggests that HGT has been a process by which Brucella has gained pathogenic traits. More interestingly, genome plasticity and putative selective inactivations within the GI found in B. abortus are in agreement with the idea that these regions can play a role in host specificity and adaptation to an intracellular lifestyle. Future studies on hypothetical genes with unknown functions carried on the GI will allow elucidating their role in the pathogenicity of Brucella.


Bacterial strains and culture conditions

All strains used in this study are listed in Table 2. Bacterial cultures were obtained by transfering a loopfull of frozen glycerol stock (conserved at 70°C) to Brucella agar plates and incubation for 48 hr at 37°C. Liquid cultures from a single colony were grown on 20 ml of Brucella broth, maintained for 24 hr at 37°C and 100 rpm until the stationary phase was reached. For E. coli JM109 growth, Luria-Bertani broth (LB), supplemented with ampicillin (100 µg/ml) was used.

Bioinformatic strategy

Complete genome sequences in GenBank format for Brucella abortus 9-941 (accession numbers AE017223 and AE017224) were retrieved from NCBI and used to perform genomic island prediction analysis using the Artemis software [33] at Briefly, we used Karlin signature difference ( ≥ 0.05) and GC% content plots to detect deviation peaks, adjusting to 10 kb the sliding window through entire sequences. Boundaries of these GI were defined by DR searches using the REputer web server ( with default values and FASTA sequence segments retrieved from the genome sequence [34]. Predicted flanking repeats pairs with E values < 10-10 were considered likely putative GI boundaries. Oligonucleotides were designed considering 500-1000 bp upstream and downstream of theoretical limits to amplify the entire regions by long range PCR. Long amplification products were analyzed first by computational simulation of restriction digestion, and then confirmed experimentally. PCR products from chromosomal junction (which include the attB site) of GI were cloned and sequenced. Because GI show inherent instability mediated by a site-specific recombination mechanism and are often found transiently excised from the chromosome as CI, putative CI were simulated using Vector NTI software [35] (available at Invitrogen website, taking account predicted flanking DR as cutting and rejoining points (attL and attR sites). These circular sequences were used as templates for primer design to amplify a specific segment generated by excision mechanism (the attP site).

DNA extraction and PCR analysis

Genomic DNA was extracted from heat inactivated (30 min at 90°C) pure cultures using lysozime-SDS-proteinase K lysis, followed by DNA-CTAB extraction and isopropanol precipitation [36]. To determine the presence of GI, long range PCR assays were performed in a final volume of 25 ul containing 25-50 ng of gDNA, 0.4 mM dNTPs, 12.5 pmol primer and 1 U of Long Range Enzyme Mix (Fermentas). The PCR conditions were: initial denaturation at 95°C for 1 min; 30 cycles of denaturation at 95°C for 20 sec; annealing at 60°C for 30 sec; extension at 68°C for 1 min/kb product; final extension step of 10 min at 68°C. PCR were followed by restriction analysis to confirm the product identity. GI instability assessment was carried out by detection of chromosomal junctions and CI, using conventional PCR assay in a final volume of 25 ul containing 0.5-1.0 µg of gDNA, 10 pmol of each primer, 0.2 mM of dNTPs, 2 mM of MgCl2 and 1U of Taq Platinum (Invitrogen). For this PCR, the conditions were: pre-incubation at 95°C; 30 cycles at 95°C for 20 sec: 60°C for 30 sec; and 72°C for 1 min with a final extension at 72°C for 5 min. To confirm identity, PCR products from chromosomal junctions and CI-derived amplicons were purified and cloned in pGEMT-easy vector (Promega) and sequenced (Macrogen). All PCR rounds were amplified in a GeneAmp PCR System 9700 (Applied Biosystems). Amplicons and restriction fragments were resolved by electrophoresis on 1.0-2.0 % TAE agarose gels. All primers used are listed in Additional File 4.

RNA extraction and RT-PCR analysis of int genes

Total RNA was prepared with bacteria harvested from liquid cultures using RNA purification system (Qiagen). Following RNA isolation, samples were treated with DNAse free RNAse Turbo (Ambion) and the yield was quantified spectrophotometrically. After that, 100 ng of total RNA were reverse transcribed to produce cDNA using Transcriptor RT (Roche) with random hexamers at 55°C for 30 min. PCR amplification rounds were performed on 2 µl of cDNA template using 12.5 pmol of forward and reverse primers of each int gen in separated experiments with 0.4 mM dNTPs, 2 mM MgCl2, Buffer 1x and 1 U of Taq Platinum in a final volume of 25 µl. Samples were incubated at 95°C for 2 min followed by 30 cycles of 95°C for 20 sec, 60°C for 30 sec and 72°C for 30 sec with a final extension for 5 min at 72°C. RT-PCR rounds were also performed in a GeneAmp 9700 thermocycler (Applied Biosystems).

For int genes that were not detected by the conventional approach, real time PCR rounds were performed on a LightCycler 1.5 PCR machine (Roche) using the same amount of RNA samples, 0.3 M of forward and reverse primers, MgCl2 4 mM, 1 l of LightCycler FastStart DNA Master SYBR Green I mix and PCR grade water in a final volume of 20 l. After enzyme activation for 10 min at 95°C, 45 cycles (denaturation at 95°C for 8 sec, annealing at 65°C at 8 sec and extension at 72°C for 20 sec) were performed to amplify a specific fragment derived from int cDNA. Then the melting curve of PCR products was analyzed with LightCycler3 software to verify PCR specificity. Data was acquired by heating the amplicons to 95°C for 1 sec, 15 sec at 70°C and then slowly heating at 0.1°C/sec to 99°C under continuous fluorescence monitoring. Negative control without DNA was included for each round.

Authors' contributions

MM conceived the study, participated in its design, accomplished computational analysis, carried out PCR assays, performed RT-PCR of integrases and drafted the manuscript. AMZ participated in the design, coordination and financial support of the study and helped to draft the manuscript.


We are grateful to Minie Villarroel from Servicio Agrícola y Ganadero, SAG, for supplying bacterial strains. We thank also to Dr. Ignacio Moriyón from Universidad de Navarra for his valuable help to draft the manuscript. This work was supported by doctoral thesis project D-2005-17 from Dirección de Investigación y Desarrollo, Universidad Austral de Chile and FONDEF D02I1111 grant. MM was the recipient of a doctoral fellowship from CONICYT-Chile.