The IncP Island in the Genome of Brucella suis 1330 Was Acquired by Site-Specific Integration

Abstract

An 18,228-bp region containing open reading frames predicted to be derived from the IncP plasmid or phage ancestors is present in the genomes of Brucella suis biovars 1 to 4, B. canis, B. neotomae, and strains isolated from marine mammals, but not in B. melitensis, B. abortus, B. ovis, and B. suis biovar 5. The presence of circular excision intermediates and the results of an analysis of sequenced bacterial genomes suggest that the region downstream of the guaA gene is a hotspot for site-specific integration of foreign DNA mediated by a CP4-like integrase.

Bacteria of the genus Brucella are responsible for zoonotic infections in a wide range of mammals, from rodents to domestic farm animals to cetaceans (1). Comparison of the genomes of Brucella melitensis and B. suis showed that more than 90% of the genes exhibit between 98% and 100% sequence identity (2, 10). Thirty-three regions longer than 100 bp that are unique to one of these two species (22 regions in B. suis and 11 regions in B. melitensis) were identified; only 6 of these regions are longer than 1 kb, showing how closely related the two species are. The longest insertion in the B. suis genome is 18,228 bp and is of interest since it has hallmarks of a sequence acquired by horizontal gene transfer (genes of the IncP plasmid or phage or of unknown origin, a G+C content of 55.3% [compared to 57.3% for the Brucella genome], and the presence of a CP4-like site-specific integrase), a somewhat rare event in the stable genomes of the genus. In this paper we report the distribution of this IncP island in the genus Brucella and provide evidence for the presence of circular excision intermediates which suggest that the integrase is still active. Analysis of previously published bacterial genomes showed that the region downstream of the guaA gene is a site of insertion of horizontally acquired DNA also found in other related and unrelated bacteria.

18.3-kb B. suis-specific region contains plasmid-related genes.

The 18,228-bp IncP island is located on B. suis 1330 chromosome II, downstream of the guaA gene encoding a GMP synthase (BRA0361/BMEII0887) and upstream of BRA0380. BRA0380 encodes a small hypothetical protein found only inBrucella spp. (but only annotated in B. suis) that overlaps BMEII0886 in the B. melitensis genome. We disregarded BMEII0886 as it is a poorly annotated open reading frame (ORF) (starting at a TTG codon) that has not been annotated in the B. suis 1330 genome. Following BRA0380 we found a gene encoding a 19-kDa protein that has been annotated in both B. suis and B. melitensis (BRA0381/BMEII0885).

The region contains 18 open reading frames (BRA0362 to BRA079) (Fig. 1), 11 of which encode proteins of phage or plasmid origin. The protein encoded by BRA0362 is a member of the Int family of site-specific recombinases. CP4-like integrases are found in temperate bacteriophages, integrative plasmids, pathogenicity and symbiosis islands, and other mobile genetic elements. We designated this protein BciA, for Brucella CP4-like integrase A (see below). BRA0363 encodes a small putative DNA binding protein with homology to the phage-encoded AlpA family of regulators. BRA0364 encodes a protein similar to the replication protein (RepA) of Pseudomonas fluorescens plasmid pVS1. Several genes in the region encode proteins with homologues in IncP incompatibility group plasmids; traI, trbL, and trbJ are homologous to genes from the RP4 and R751 plasmids of Escherichia coli, while traJ and traC have homologs on pNGR234a, a plasmid of Rhizobium sp. strain NGR234 (5), and R18 of E. coli. TraI is a relaxase that, in conjunction with several auxiliary proteins, including TraJ and TraH (not found in the Brucella genome), forms the relaxation complex or relaxosome which is required for the first step in conjugative DNA transfer. The protein encoded by BRA0379 has a RelB domain and exhibits high homology to DinJ an “antitoxin” belonging to the toxin-antitoxin regulatory families that play a role in maintaining replicons (6). Usually the “antitoxin” gene is found in association with a gene encoding the “toxin” protein that it counteracts; interestingly, no toxin gene was found to be associated with this locus or with another antitoxin, PemK, which is encoded elsewhere on the genome by BRA1113.

FIG. 1.

Organization of the IncP island of B. suis biovar 1 (A) and comparison with the guaA-19-kDa region of B. melitensis (B). The BRA ORF numbers are those in the accession no. AE014292.2 sequence. Genes of phage or plasmid origin are indicated by gray arrows, and genes encoding proteins of unknown function or origin are indicated by cross-hatched arrows. EcoRI restriction sites are indicated by vertical arrows. (C) Southern blot of EcoRI-digested genomic DNA hybridized with a probe encompassing the guaA-int junction, showing that the IncP island is present in the genomes of B. suis biovar 1 (lane S1) and B. neotomae (lane N) but absent in B. suis biovar 5 (lane S5). The arrowheads indicate the 52-bp imperfect direct repeats at the boundaries of the IncP island.

The proteins encoded by six ORFs in the region have no assigned function; the BRA0375 protein exhibits high homology with hypothetical proteins in several bacteria, including Xanthomonas axonopodis, Nitrosomonas europaea, Legionella pneumophila, and Pseudomonas species. BRA0377 homologues are conserved in many organisms, ranging from bacteria to higher eukaryotes, including plants, mammals, and birds. The protein may be an ATPase as a conserved domain search revealed similarities to both the Lon protease and the Sbc protein involved in DNA repair. Interestingly, genes encoding homologues of BRA0377 and BRA0378 are linked in both Thermobifida fusca and Vibrio vulciticus (both phylogenetically distant from the Brucella genus). The other ORFs have no GenBank homologues.

Distribution of the region in the genus Brucella.

The genus Brucella is classically divided into six species (Table 1), each of which infects a particular, although nonexclusive, mammalian host. B. melitensis infects sheep and goats; B. abortus infects cattle; B. suis infects swine and rodents; B. canis infects dogs; B. ovis infects sheep; and B. neotomae infects the desert wood rat (1). The species B. melitensis, B. abortus, and B. suis are further subdivided into biovars on the basis of antigenic and metabolic phenotypes. Strains recently isolated from marine mammals form a new group of species (3). We were interested in determining whether the IncP island is restricted to B. suis 1330 or is also found in other members of the genus.

TABLE 1.

Bacterial strains

Species	Strain	Biovar or host	IncP	Source
B. melitensis	16MT	Biovar 1	−	ATCC 2356
B. melitensis	63/9	Biovar 2	−	ATCC 2357
B. melitensis	Ether	Biovar 3	−	ATCC 2358
B. suis	1330T	Biovar 1	+	ATCC 23444
B. suis	Thomsen	Biovar 2	+	ATCC 23445
B. suis	686	Biovar 3	+	ATCC 23446
B. suis	40	Biovar 4	+	ATCC 23447
B. suis	513	Biovar 5	−	A. Macmillan
B. canis	RM/666		+	ATCC 23365
B. abortus	544T	Biovar 1	−	ATCC 23448
B. abortus	B3169	Biovar 1	−	ATCC 23452
B. neotomae	5K33		+	ATCC 23459
B. ovis	63/290		−	ATCC 25840
Brucella sp.	5	Dolphin	+	A. Macmillan
Brucella sp.	34	Porpoise	+	A. Macmillan
Brucella sp.	2	Seal	+	A. Macmillan

Oligonucleotide primers guaAF and 19kDaR (Table 2) annealing in the 5′ region of guaA and downstream of the 19-kDa gene were designed based on the B. suis and B. melitensis genome sequences to amplify a 2,345-bp fragment in B. melitensis or a 19,547-bp fragment in B. suis (Fig. 1). A fragment of the predicted size was amplified when B. melitensis biovar 1 to 3 genomic DNA were used as the templates for PCR (not shown) and was also amplified from B. abortus biovars 1 and 5 and B. ovis DNA. No PCR product was observed when genomic DNA from B. suis biovars 1 to 4, B. canis, B. neotomae, and the three marine isolates were used as templates, probably because the 18-kb fragment is too large to be amplified under the conditions used. To confirm the presence of the region, we used PCR to amplify internal fragments of the int, trbL, BRA0375, and traI genes from the B. suis genome and to generate probes (Table 2). Genomic DNA from the panel of strains was digested with EcoRI, separated on a 0.7% agarose gel, transferred to a nylon membrane, and hybridized with the four probes as described previously (4). Southern blots confirmed that the 18-kb region was absent from the genomes of B. melitensis, B. abortus, B. ovis, and B. suis biovar 5 and present in B. suis biovars 1 to 4, B. neotomae, and the strains isolated from marine mammals (Fig. 1 and data not shown).

TABLE 2.

Oligonucleotide primers

Use	Oligonucleotidea	Sequence
Amplify region between guaA and 19-kDa gene	guaAF	GTTCGGCCGTGCTTTTCTCG
	19kDaR	TAACCCCGTCCTTGAACTCA
	Demonstrate excision	excA	GGCGGATTTCTACCACTACG
		excB	GGGTGCTTGGCGATGCGAAA
	excC	TAAGCCCATTCCCGATAACA
	excD	CAAGAAAAAGCCCGCAGTTA
Amplify internal fragments for probes and mutagenesis	int F	GCG AGA TCT AAA GCC GAA GT
	int R	AGT CAG AGT CAT AGC GCG AT
	BRA0375 F	GCA CAG GCC GCA GAA CAG AT
	BRA0375 R	AGT CGT CGG CAA GCT CAA CA
	traI F	AGA AAC GCT TAC GAC CTC AG
	traI R	TAC ATC AGG CGT TTG GTC GG
	trbL F	ATG GGT TTC GGT GGC TCT CA
	trbL R	CCT GTT TGG CGG CTT GCT TC

^aF, forward; R, reverse.

Region does not contain essential virulence factors.

Many bacterial pathogens have acquired virulence factors by horizontal transfer of DNA. To assess the role of this region in virulence, the internal fragments of the int and trbL genes and the gene encoding a hypothetical protein (BRA0375) (Fig. 1) were amplified by PCR and cloned into pGEM-T (Promega). The resulting plasmids were introduced into B. suis by electroporation to inactivate the genes with insertional mutagenesis by homologous recombination. The three mutants were verified by PCR and Southern blotting (not shown), and their virulence was tested in three cell lines, differentiated THP-1 human macrophages, J774 murine macrophages, and HeLa cells, as described previously (9). The three mutants exhibited wild-type virulence in the three cell lines, showing that these genes were not virulence factors (data not shown). These results were not unexpected; both B. melitensis and B. abortus, which lack the genes, are fully virulent in cell infection models. It is, however, possible that genes in this region are determinants required for virulence in an animal model or for full virulence in a specific host.

Is the guaA locus a hotspot for insertion?

Members of the CP4 integrase family mediate integrative and excisive site-specific recombination between two sequences at sites called attachment sites, which are located on the phage genome (attP) and the bacterial chromosome (attB). The integrase gene is generally found inserted in close proximity to the insertion site. The majority of tyrosine integrases target mRNA or tmRNA genes (13); however, there is no such sequence in this region of the Brucella genome. Analysis using The Institute for Genome Research Comprehensive Microbial Resource (http://www.tigr.org/tigr-scripts/CMR2/CMRHomePage.spl) showed that a gene encoding a CP4-like integrase (often followed by other phage, transposon, or plasmid-related genes) is found downstream of guaA in several gram-negative and gram-positive bacteria. In the Shigella flexneri 2a genome (accession no. AE016986.1), the guaA gene is followed by an 18-kb region starting with a phage integrase and containing phage, transposon, and plasmid genes (including TrbL and TrbJ), as well as genes encoding arsenate resistance (12). Similar situations are seen in Mesorhizobium loti (accession no. BA000012), Rhodopseudomonas palustris (BX571963), Xanthomonas axonopodis (AE011866.1), and Listeria monocytogenes (AL591977), while in Fusobacterium nucleatum (AE009951) and Streptococcus pneumoniae (AE007317) a gene encoding a degenerate transposase is found. Site-specific integration of DNA carrying an attP site into an attB site in the bacterial chromosome generates two junctions (attL and attR). In B. suis 1330 a 52-bp almost perfect direct repeat can be recognized on either side of the insertion (Fig. 1 and 2A). When this 52-nucleotide sequence was used to BLAST search bacterial genomes, the last 27 bp of the sequence was found to be present in the guaA-CP4 int intergenic region in the genomes of several bacteria, including A. anonopodis, Pseudomonas syringae, Azoarcus, and S. flexneri, and downstream of guaA in Methylococcus (which has no integrase gene) (Fig. 3). This 27-nucleotide sequence has the characteristics of a tyrosine recombinase recognition site, with a conserved 8-bp perfect inverted repeat interrupted by a 6-bp asymmetric core binding site (Fig. 2A), suggesting that the sequence may represent a minimal att site. The SaPlbov2 pathogenicity island in Staphylococcus aureus carrying the Sip integrase has also been found to integrate at the 3′ end of guaA; however, the attB site is different from that found inBrucella (11). BLAST searches with this short S. aureus sequence showed that it is found at the 3′ end of guaA in manybacteria, including Leptospira interrogans (accession no. AE016823), Shewanella oneidensis (AE014299), Vibrio parahaemolyticus (BA000031), and Clostridium perfringens (BA000016). These observations strongly suggest that the region downstream of guaA may be the target site for phage integration in many bacterial species.

FIG. 2.

Schematic representation and analysis of site-specific integration of the IncP island in Brucella. (A) Region surrounding the proposed guaA hotspot for site-specific integration in B. suis 1330 and B. melitensis 16M. The sequences of attB, the bacterial integration site, attR, attL, and right and left recombination sites are shown. Bm, B. melitensis; Bs, B suis. Single-nucleotide polymorphisms between attB or attR and attL are enclosed in boxes (SNP-A and SNP-B). The proposed minimal recombination site consists of two short inverted (8-bp) repeats (open arrows) and a 6-bp asymmetric core region (arrow). (B) PCR and sequence analysis of excision events in B. suis. Lanes 1, 2, 3, and 4 contained the PCR amplification reaction mixtures with the corresponding primer combinations, as indicated on the right. Sizes are indicated in base pairs. Lane M contained Smartladder (Eurogentec).

FIG. 3.

Alignement of att sites from Brucella and from the guaA-int intergenic region in unrelated bacteria. The inverted repeats are indicated by open arrows, and the sequence differences are underlined. The genome sequence accession numbers are as follows: B. suis, AE014292.2; B. abortus, AE017224.1; B. melitensis, AE009722.1; X. axonopodis, AE011866.1; Salmonella enterica serovar Choleraesuis (S. choleraesuis), AE017220.1; S. flexneri, AE016986.1; Azoarcus,, CR555306.1; Methylococcus capsulatus, AE017282.2; and Erwinia carotovora, BX950851.1.

Brucella CP4-like integrase gene, bciA, encodes a functional integrase.

To investigate whether the integrase could mediate recombination in B. suis between the attL and attR sites, we performed a PCR analysis. We designed primers (Table 2) that specifically amplified the att region of circular recombinant molecules, reminiscent of excision (Fig. 2B shows the results obtained with primer set 4 [excB/excC]), as well as the chromosomal junction that would remain after excision of the IncP island (primer set 1 [excA/excB]). Control primer sets 2 (excA/excB) and 3 (excC/excD), which were indicative of the attL and attR junctions present in the B. suis genome, amplified fragments of the expected size. Primer set 4 reproducibly resulted in amplification of a 543-bp fragment using genomic DNA from B. suis biovars 1 to 4, B. canis, B. neotomae, and marine isolate 5 as templates, suggesting that excision did occur (Fig. 2B and not shown). Although in our preliminary experiments there was no evidence of a junction left by excision of the IncP island, after optimizing the PCR conditions, we were able to amplify a 1,307-bp PCR product using primer set 1, supporting the conclusion that excision had occurred, the first observation of such an event in Brucella. We therefore concluded that the integrase encodes a functional protein, which we designated BciA for Brucella CP4-like integrase A.

We sequenced the amplified region of the recombinant circle and found that it was the result of a precise recombination event between the B. suis attL and attR sites (Fig. 2B and not shown), which provided further evidence for the functionality of BciA. The attL and attR sites contain two single-nucleotide polymorphisms, called SNP-A and SNP-B for simplicity (Fig. 2A). In site-specific recombination, the integration products attL and attR are chimeras of the attB and attP site; we detected SNP-A_attR and SNP-B_attL in the attP site of the recombinant circle (Fig. 2B), suggesting that the cleavage and strand-rejoining reactions occurred between these nucleotides. Cleavage and rejoining generally occur in the asymmetric spacer region, suggesting that the cleavage site must be located in the sequence GTAA (Fig. 2A). Based on the sequences of the attL and attR sites of B. suis and the attP site of the excised circle, the attB site is expected to contain SNP-A_attL. The attB site of B. melitensis does not contain SNP-A_attL. Interestingly the attB site of B. abortus is identical to the B. melitensis attB site, whereas the attL sites of B. neotomae and marine strain 34, both containing the IncP island, are identical to the B. suis 1330 attL site (not shown).

Two intriguing questions are how and when the region was acquired. Our current hypothesis for the evolution of the genus Brucella is that an ancestor with a single circular chromosome gave rise to two separate lineages with two chromosomes following recombination at rrn loci (7). B. suis biovar 1 is predicted to be the ancestor of B. canis, B. abortus B. melitensis, and B. neotomae and the marine strains. The presence of the region at the same genomic site and containing the same nucleotide at SNP-A in strains of Brucella, which would never normally encounter each other, argues in favor of the hypothesis that the insertion was an ancient event that was followed by speciation. The region was then lost, and there was an accompanying mutation that gave SNP-A_attB in the ancestor of B. abortus and B. melitensis. The Brucella genome is extremely stable (8), and the fact that we have not isolated B. suis 1330 that has lost the IncP island, despite the fact that it does excise, suggests that there is selective pressure for it to be maintained. Ongoing experiments will hopefully shed more light on the function of this region.