Emergence and Evolution of Multiply Antibiotic-Resistant Salmonella enterica Serovar Paratyphi B d-Tartrate-Utilizing Strains Containing SGI1

Abstract

The first Australian isolate of Salmonella enterica serovar Paratyphi B d-tartrate-utilizing (dT⁺) that is resistant to ampicillin, chloramphenicol, florfenicol, streptomycin, spectinomycin, sulfonamides, and tetracycline (ApCmFlSmSpSuTc) and contains SGI1 was isolated from a patient with gastroenteritis in early 1995. This is the earliest reported isolation globally. The incidence of infections caused by these SGI1-containing multiply antibiotic-resistant S. enterica serovar Paratyphi B dT⁺ strains increased during the next few years and occurred sporadically in all states of Australia. Several molecular criteria were used to show that the early isolates are very closely related to one another and to strains isolated during the following few years and in 2000 and 2003 from home aquariums and their owners. Early isolates from travelers returning from Indonesia shared the same features. Thus, they appear to represent a true clone arising from a single cell that acquired SGI1. Some minor differences in the resistance profiles and molecular profiles also were observed, indicating the ongoing evolution of the clone, and phage type differences were common, indicating that this is not a useful epidemiological marker over time. Three isolates from 1995, 1998, and 1999 contained a complete sul1 gene but were susceptible to sulfamethoxazole due to a point mutation that creates a premature termination codon. This SGI1 type was designated SGI1-R. The loss of resistance genes also was examined. When strains were grown for many generations in the absence of antibiotic selection, the loss of SGI1 was not detected. However, variants SGI1-C (resistance profile SmSpSu) and SGI1-B (resistant to ApSu), which had lost part of the integron, arose spontaneously, presumably via homologous recombination between duplications in the In104 complex integron.

SGI1 was first identified in Salmonella enterica serovar Typhimurium DT104 strains with the ampicillin, chloramphenicol, florfenicol, streptomycin, spectinomycin, sulfonamides, and tetracycline (ApCmFlSmSpSuTc) resistance phenotype (4), and SGI1, or variants of it with different resistance phenotypes and resistance genes, have since been found in many Salmonella serotypes (10, 18, 19, 24, 33) and also in Proteus mirabilis (2, 5). SGI1 is an integrating element that can move into new hosts by transduction (31) or via mobilization by an IncA/C plasmid (12), and it establishes itself by integrating into the host chromosome at the end of the thdF gene. However, these events likely are quite rare, and it is possible that all of the SGI1-containing strains of any particular serovar arose from a single cell that acquired the SGI1 genomic island and then expanded in an antibiotic-selective environment and subsequently spread around the globe. A single-cell origin for the widespread multiply antibiotic-resistant S. enterica serovar Typhimurium DT104 clone that contains SGI1 is supported by the finding that the earliest DT104 strains of this type isolated in the United States are identical to those isolated 5 and 10 years later (28) and to ones isolated in other countries. More recently, additional support for clonality comes from the finding that such strains carry a specific deletion in the genes required for allantoin utilization (21) and carry three phages that are not commonly found in allantoin-utilizing strains (13, 14). However, it has been found that the phage type of members of this clone can change to DT12 and DT120 (15).

Strains of S. enterica serovar Paratyphi B that utilize d-tartrate as a carbon source (Salmonella serovar Paratyphi B dT⁺) cause gastroenteritis (8). Multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ strains have been isolated from human infections in different parts of the world since the late 1990s. One type, with the resistance phenotype ApCmFlSmSpSuTc (Ap, ampicillin; Cm, chloramphenicol; Fl, florfenicol; Sm, streptomycin; Sp, spectinomycin; Su, sulfonamides; and Tc, tetracycline), have been recovered in Canada (23), England, Wales, Scotland and the Channel Islands (32), France (34), Australia (17, 18), and Japan (1). In most of these isolates, the resistance genes blaP1, floR (or cmlA3), aadA2, sul1, and tetA(G) are found in a complex class 1 integron (Fig. 1), designated In104 (6, 18), that is located within Salmonella genomic island 1 (SGI1) (4).

FIG. 1.

Structure of the integron of SGI1 and deletion derivatives SGI1-B and SGI1-C. The In104 region of SGI1 serovar Typhimurium DT104 (from GenBank accession no. AF261825) (4) is drawn to scale. Different discrete segments, such as the gene cassettes, are represented by open boxes and lines of different thicknesses, and in the SGI1 line, 5′ and 3′ indicate the 5′-CS- and 3′-CS-derived regions, respectively. Vertical bars indicate the inverted repeats (IRi and IRt) of class 1 integrons bounding In104. The attI1 site is a tall open box, and gene cassettes (aadA2 and blaP1) are open boxes with a black bar at one end, indicating the attC sites (59-be). IS6100 and the central, non-integron-derived region are open boxes. Arrows indicate the position and orientation of genes and open reading frames. The SGI1 backbone adjacent to the integron is indicated by dashed lines.

Recently, home aquariums were identified as a source of human infections with SGI1-containing Salmonella serovar Paratyphi B dT⁺ (17, 25), and we suggested that all of these strains have arisen from a single-cell progenitor (17). The global ornamental tropical fish trade, which uses antibiotics for prophylactic treatment (25), could provide a means of amplifying and then disseminating the clone globally. The first SGI1-containing Salmonella serovar Paratyphi B dT⁺ isolate with the ApCmSmSpSuTc resistance phenotype reported so far was isolated in 1997 from a tropical fish in Singapore (22). The earliest human case of gastroenteritis caused by SGI1-containing Salmonella serovar Paratyphi B dT⁺ reported so far occurred in Canada in 1998 (23), and equivalent isolates were not recovered in the United Kingdom or France until 2000 (32, 34). The relatively recent emergence of these strains makes it possible to examine the origins of this new pathogen. In this study, the emergence in Australia of SGI1-containing multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ strains that cause gastroenteritis in humans was monitored. The relationship of these strains to one another and to strains studied previously (17) that were isolated from owners of home aquariums and their aquariums was examined. The loss of antibiotic resistance genes when cells are grown without antibiotic selection also was studied.

MATERIALS AND METHODS

Bacterial strains.

All isolates of Salmonella serovar Paratyphi B dT⁺ recovered from a human source had been submitted to the Microbiological Diagnostic Unit (MDU), where the serovar and phage type were determined, and the majority were tested for resistance to antibiotics as described elsewhere (18). The antibiotic resistance phenotypes of strains used in this study were reassessed using a disc diffusion assay, namely the calibrated dichotomous sensitivity test (http://web.med.unsw.edu.au/cdstest) with antibiotic discs (Oxoid, New Hampshire, England) containing ampicillin (25 μg), chloramphenicol (30 μg), florfenicol (30 μg), gentamicin (10 μg), kanamycin (50 μg), neomycin (30 μg), streptomycin (25 μg), spectinomycin (25 μg), sulfafurazole (300 μg), tetracycline (30 μg), tobramicin (10 μg), and trimethoprim (5 μg).

PCR mapping.

Whole-cell DNA was isolated from cultures grown overnight at 37°C on MacConkey Agar (Becton Dickinson, MD) using standard methods (30) and was used as the template for PCR amplification reactions using various combinations of the primers, which are listed elsewhere (18). The following pairs were used to detect SGI1: SGI left junction, U7-L12/LJ-R1; SGI right junction, 104-RJ/104-D; In104 left junction, S026-FW/aadA2-R2 (S026-aadA2); In104 right junction, DB-T1/MDR-B (IS6100-S044). U7-L12 was used with 104-D to amplify the original chromosomal configuration (thdF adjacent to yidY). Primers L1 and R1 amplify the gene cassettes in class 1 integrons. A multiplex PCR was used to detect the five resistance genes of SGI1 (9).

Amplification was carried out in PCR buffer (Roche Molecular Biochemicals, Mannheim, Germany) containing each deoxynucleoside triphosphate at 160 μM, 20 pmol of each primer, approximately 10 to 50 ng of template DNA, and 1 U of Taq DNA polymerase (Roche). Reaction conditions, described in detail elsewhere (18), generally were 94 to 96°C for 3 to 5 min, followed by 30 to 40 cycles of denaturation (94 to 96°C for 30 s), annealing (52 to 62°C for 30 to 60 s), and extension (72°C for 30 s to 2 min), and a final incubation at 72°C for 10 to 15 min. Products were separated on agarose gels, and sizes were estimated using 100-bp and 1-kb DNA ladders (New England BioLabs) as molecular size markers and known amplification products from SGI1 as standards.

DNA sequencing and sequence analysis.

PCR products were purified using the QIAquick PCR purification kit (Qiagen Inc., Valencia, CA) and from gels using the QIAquick gel extraction kit by following protocols supplied by the manufacturer. Automated sequencing was performed at Macquarie University on an ABI PRISM 377 DNA sequencer (AME Bioscience) or a 3130 Exel genetic analyzer (Applied Biosystems) using the Big Dye system, and sequences were assembled using Sequencher version 4.8 (Gene Codes Corporation, Ann Arbor, Michigan). Sequences were compared to standard sequences, e.g., GenBank accession no. U12338 for the 3′-conserved segment (3′-CS), using BLAST (http://blast.ncbi.nlm.nih.gov). Multiple alignments were performed using ClustalW (www.ebi.ac.uk/Tools/clustalw2). Gene construction kit version 2.5 (Textco, West Lebanon, NH) was used to create figures to scale.

RAPD profiles.

RAPD (random amplification of polymorphic DNA) profiles were determined using whole-cell DNA and primers 1290 (26) and 23L (20), which gave the highest level of discrimination of the several primers tested. Amplification was carried out in PCR buffer (Roche Molecular Biochemicals, Mannheim, Germany) containing each deoxynucleoside triphosphate at 200 μM, 50 pmol of each primer, approximately 10 to 50 ng of template, and 1 U of Taq DNA polymerase (Roche). The reaction conditions for 23L (20) and 1290 (26) were described previously. To ensure the validity of the comparisons, profiles of each strain were determined at least twice, and large sets of strains were compared directly in single experiments.

IS200 profiles and PFGE.

Whole-cell DNA was digested with restriction enzyme PstI, and the fragments were resolved by electrophoresis through 0.7% agarose. The DNA was transferred to an N⁺ nylon membrane (Amersham, Buckinghamshire, United Kingdom) by capillary action and cross-linked to the membrane by baking at 80°C for 2 h. After exposure overnight at 42°C to a denatured digoxigenin (DIG)-labeled probe for IS200, prepared as described elsewhere (34), the membrane was washed twice with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate (SDS) for 15 min at room temperature and twice more stringently with 2× SSC-0.1% SDS for 15 min at 68°C. DIG-labeled molecular mass markers (molecular marker set II; Roche Diagnostics Corporation) were included on all membranes, and bands were visualized fluorescently using standard protocols supplied by the manufacturer (Roche Diagnostics Corporation). Macrorestriction analyses of XbaI-digested whole-cell DNA by pulsed-field gel electrophoresis (PFGE) was performed as described previously (17).

Nucleotide sequence accession number.

The nucleotide sequence from the boundaries of SGI1 with flanking regions (the thdF-int and S044-yidY junctions) and the 3′-CS of SGI1 from SRC181 are deposited in GenBank under accession no. FJ477835.

RESULTS

Emergence of multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ in Australia.

During the period 1990 to 2005, between 70 and 150 cases of gastroenteritis caused by S. enterica serovar Paratyphi B dT⁺ were reported in Australia, with isolates recovered from all states. The number of S. enterica serovar Paratyphi B dT⁺ isolates rose slightly in the mid- to late 1990s, and at the same time the number of multiply antibiotic-resistant isolates also increased (Table 1). To identify the earliest SGI1-containing strains, all recoverable isolates with the resistance phenotype ApCmSmSpSuTc isolated in Australia prior to 2000 from cases of human disease (Table 2) were examined for the presence of SGI1, and the integron In104 that lies within it, using PCRs as described previously (17, 18). Strains with the closely related resistance phenotype ApCmSmSuTc (Sp^s) or ApCmSmSpTc (Su^s) and isolates that could be derived from an SGI1-containing strain (SGI1-C, resistance phenotype SmSpSu) also were included in the study. Three strains from 2000 and two from 2001 that were described previously (17, 18) were included for comparison. Only a few isolates were associated with travel to Indonesia, with the remainder being locally acquired infections (Tables 1 and 2). The isolates listed in Table 2 contained the left and right junctions of SGI1 with the chromosome and the left and right junctions of In104 with the SGI1 backbone. The ApCmSmSuTc (Sp^s) PT Worksop strain from 1992 (Table 1) did not contain SGI1 or a class 1 integron; the ApCmSmSuTc (Sp^s) strain from 1994 (Table 1) could not be recovered. Thus, isolates SRC181 and SRC182 from January and May 1995, respectively, represent the first multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ strains containing SGI1 known to appear in Australia. These strains predate by 3 years the earliest isolation in other countries (23). However, one of these strains, SRC181, does not display resistance to sulfonamides.

TABLE 1.

Antibiotic resistance in S. enterica serovar Paratyphi B dT⁺ from humans in Australia

Yr	No. of isolates	No. tested	No. resistant	No. resistant to ApCmSmSpSuTca	Phage typeb (overseas origin)
1990	43	41	2
1991	68	65	8
1992	61	59	3	1 (Sps)	Worksop
1993	89	89	1
1994	69	68	6	1 (Sps)	1var3 (India)
1995	71	71	7	1	RDNC
				1 (Sus)	1var3
1996	89	87	1
1997	106	105	11	3	1var3
				1	3bvar3
				1	RDNC (Bali)
				1 (Sps)	RDNC
1998	160	160	18	7	1var3
				1 (Tpr)	1var3
				1 (Sus)	3bvar
				1	RDNC (Indonesia)
1999	112	112	14	1	1var
				2	1var3
				1 (Sus)	1var
2000	100	100	23	1	3bvar
				2	3bvar3
				7	RDNC Aus2
				1	RDNC
				1	RDNC (Indonesia)
2001	103	103	8	1	1var
				1	1var3
				1	RDNC Aus3
				1	RDNC
				1	RDNC Aus3 (India)
2002	85	85	21	1	var1
				1	3bvar
				8	RDNC Aus3
				1 (Tpr)	RDNC Aus3
				1 (Sus)	RDNC Aus3
				2	RDNC Aus4
				1 (Tpr)	3bvar3
				1	RDNC
				2	RDNC Aus1
2003	78	76	26	20	RDNC Aus3
				1	1var15
				1	3bvar
				1	3bvar3
2004	111	110	25	12	RDNC Aus3
				2	3bvar3
				1	RDNC
				1	RDNC Aus1 (Indonesia)
				1	3bvar3 (Indonesia)
2005	139	139	12	6	RDNC Aus3
				2 (Tpr)	RDNC Aus3
				1	1var3
				1	3bvar
				1 (Sus)	3bvar9

^aMaterial in parentheses indicates the difference from the ApCmSmSpSuTc resistance phenotype.

^bPhage types were determined using standard procedures and designations (http://www.geocities.com/avinash_abhyankar/typing.htm). RDNC, reaction does not conform. RDNC Aus1, RDNC Aus2, etc., are identifiable and reproducible phage typing patterns awaiting formal assignment by the WHO-designated International Reference Laboratory at Colindale, United Kingdom. RDNC Aus2 is a 1var variant, and RDNC Aus3 and Aus4 are 3bvar variants.

TABLE 2.

SGI1-containing S. enterica serovar Paratyphi B dT⁺ isolates^a

Isolate	Phage type	Source	Yr	Antibiotic resistance profile	Sizes of L1-R1 products	IS200 profile	RAPD profile		PFGE profile, XbaI
							23L	1290
SRC182	RDNC	NSW	1995	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC181b	1var3	NSW	1995	ApCmFlSmSp Tc	1, 1.2	IP1-1	A	A	A
SRC177	3bvar3	NSW	1997	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC184	1var3	Tas	1997	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC185	1var3	NSW	1997	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC178	1var3	Vic	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A-2	A	A
SRC189	1var3	WA	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC190	1var3	WA	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC192	1var3	ACT	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-2	A	A	A
SRC193	1var3	NSW	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC194	1var3	ACT	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC313	1var3	WA	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC191c	1var3	NSW	1998	SmSpSu	1	IP1-1	A	A-1	ND
SRC195b	3bvar	NSW	1998	ApCmFlSmSp Tc	1, 1.2	IP1-3	A	A	A
SRC199	1var3	Qld	1999	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC200	3bvar	NSW	1999	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC198b	1var	Vic	1999	ApCmFlSmSp Tc	1, 1.2	IP1-1	A	A	A
SRC289c	1var3	Qld	2000	SmSpSu	1	IP1-1	A	A	ND
SRC290	3bvar	Qld	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	ND
SRC315	AUS2	Qld	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC316	AUS2	NSW	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC229d	AUS2	ACT	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC230d	AUS2	ACT	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC232d	AUS2	Vic	2000	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC49e	1var	Qld	2001	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A-3	A-2	A
SRC50e	RDNC	Vic	2001	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A-3	A-2	A
SRC176	RDNC	Vic/Bali	1997	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A-1	A	A
SRC196	RDNC	ACT/Indonesia	1998	ApCmFlSmSpSuTc	1, 1.2	IP1-1	A	A	A
SRC295c	1var3	Indonesia	2001	SmSpSu	1	ND	A	A	ND

^aNSW, New South Wales; Tas, Tasmania; Vic, Victoria; WA, Western Australia; ACT, Australian Capital Territory; Qld, Queensland; RDNC, reaction does not conform; ND, not determined.

^bThese strains are susceptible to sulfamethoxazole.

^cThese strains are not individually listed in Table 1.

^dSome of the information on these strains, which are associated with fish tank ownership, is from reference 17.

^eSome of the information on these strains is from reference 18.

Salmonella serovar Paratyphi B dT⁺ strains usually contain SGI1, which carries the aadA2 (resistance to streptomycin and spectinomycin) and blaP1 (resistance to ampicillin) gene cassettes in the attI sites of the integron (4, 17, 18, 22, 23, 32, 34). For all strains that were ApCmSmSpSuTc or ApCmSmSpTc resistant, two fragments containing gene cassettes, with sizes of 1.0 and 1.2 kb, were amplified using standard primers in the 5′- and 3′-CS of class 1 integrons. The digestion of these amplicons with RsaI generated a profile (data not shown) that was indistinguishable from the pattern for the two amplicons containing the aadA2 and blaP1 cassettes found in In104 and Salmonella serovar Paratyphi B dT⁺ isolates SRC49 and SRC50 from 2001 (18). The multiplex PCR that detects the five resistance genes of In104 (9) revealed the presence of the aadA2, blaP1, floR, sul1, and tetA(G) genes. The remaining strains that were resistant only to Sm, Sp, and Su yielded only the 1.0-kb amplicon and only the aadA2 and sul1 genes in the multiplex, confirming that they contained SGI1-C (Fig. 1). A set of overlapping PCRs that cover the whole of the SGI1 backbone (16) were used to demonstrate that the backbone was complete in the two first isolates, as it is in SRC49 (16).

SGI1-containing strains are closely related.

The isolates listed in Tables 1 and 2 were assigned to phage types using standard procedures, and several different phage types, mostly belonging to the phage type 1var and 3bvar groups, are represented. However, previous molecular studies of SGI1-containing Salmonella serovar Paratyphi B dT⁺ isolates using PFGE or IS200 analysis indicated that they form a relatively homogeneous group, in contrast to the considerable genetic heterogeneity found among Salmonella serovar Paratyphi B dT⁺ strains that do not carry SGI1 (23, 27, 32, 34). We therefore examined our early isolates using a number of molecular methods with a set of six multiply resistant non-SGI1-containing Salmonella serovar Paratyphi B dT⁺ isolates as outliers. Strains were examined by PFGE (Table 2) and had the same profile as that reported previously (17), which appears to differ from the most common profile among French isolates, X1.1 (34), by only one band.

We have shown previously that the IS200 profiles of several Australian SGI1-containing isolates from 2000 and 2003 (17), which were determined by the hybridization of an IS200 probe with PstI-digested whole-cell DNA, differ from the profile IP1, which was described for the equivalent French isolates (34), by the absence of one band. Almost all of the SGI1-containing strains in the current collection yielded the same variant IP1 profile, designated IP1-1 in Table 2. The remaining two strains, SRC192 and SRC196, had one additional copy of IS200 (IP1-2 and IP1-3), while the six unrelated antibiotic-resistant strains that lacked SGI1 each showed quite different patterns and numbers of bands (data not shown).

As a further measure of relatedness, we determined RAPD profiles by using two different primers (23L and 1290). Four antibiotic-resistant strains that were isolated in the same time period but lacked SGI1 gave distinct profiles with these primers (Fig. 2). In contrast, the SGI1-containing strains all had the same profile or a minor variation of it (Fig. 2 and Table 2), except for SRC49 and SRC50 from 2001, which showed more extensive differences from this profile. However, isolates from later years returned to the A profiles for both primers (data not shown).

FIG. 2.

RAPD profiles of various serovar Paratyphi B dT⁺ strains. Whole-cell DNA was amplified with primer 23L (upper panel) or primer 1290 (lower panel), and the products were separated on a 1% agarose gel and stained with ethidium bromide. SRC numbers are above the gels, and designated types are shown below. Strains contain SGI1, except for SRC183, SRC187, SRC197, and SRC201, which were used as controls.

Sequence typing.

Two types of S. enterica serovar Derby containing SGI1, representing independent acquisition events, have been identified using differences in the sequences of the chromosomal region on either side of SGI1 that is amplified by the primer pairs used to identify the left and right junctions (3). Such sequence types have the potential to provide a powerful measure of clonality, as it is very unlikely that precisely the same mutations would arise twice independently after SGI1 had been incorporated. The sequence of the SGI left junction (U7-L12/LJ-R1) and SGI right junction (104-RJ/104-D) product was determined for six strains, SRC181 and SRC182 from 1995, SRC177 from 1997, SRC178 from 1998, SRC199 from 1999, SRC289 from 2000, and SRC49 from 2001. These sequences all were identical to one another, and the chromosomally derived regions were identical to a continuous segment of the sequenced serovar Paratyphi B strain SPB7 (GenBank accession no. CP000886), which does not contain SGI1 (Fig. 3). However, this chromosomal sequence type does not appear to be common and was not found in the genomes of any other Salmonella serovars for which sequence is available, in the SGI1-containing strains of Salmonella serovar Derby (3), or in other Salmonella serovars for which sequences have been determined (16, 19) (Fig. 3), all of which differ from it in at least three positions. The sequence type also was not detected among five further SGI1-free Salmonella serovar Paratyphi B dT⁺ isolates from the same time period that were examined here. The corresponding chromosomal region (amplified with U7-L12/104-D) from four Salmonella serovar Paratyphi B dT⁺ strains that did not contain SGI1 differed from the SGI1-containing strain at three positions on the left of the SGI1 insertion site (Fig. 3). The sequence in one further strain that did not contain SGI1 differed at two positions on the left and five on the right (Fig. 3). Hence, the serovar shows significant diversity within this region, and this further supports the notion that the SGI1-containing isolates all have a common ancestor that is a close relative of the SPB7 strain.

FIG. 3.

Alignment of the sequences at the boundaries of SGI elements with sequences from the chromosome of Salmonella serovar Paratyphi strains lacking SGI1. The sequences of the chromosome to the left and the right of SGI1 are separated by a short segment from SGI1, which is shown in italics. This region is dotted when SGI1 is not present. The remainder of the sequence internal to the SGI, which is identical in all sequences, is not shown. PtB+SGI1 represents sequences from SRC181 and SRC182 from 1995, SRC177 from 1997, SRC178 from 1998, SRC199 from 1999, SRC289 from 2000, and SRC49 from 2001; PtB represents sequences from four multidrug-resistant Salmonella serovar Paratyphi B dT⁺ strains lacking SGI1, SRC183, SRC187, SRC197, and SRC201, and PtB 272 is the sequence from a further multidrug-resistant Salmonella serovar Paratyphi B dT⁺ strain lacking SGI1, SRC272. Sequences from other SGI1-containing serovars were included for comparison. Kentucky represents the boundaries of SGI1-K from serovar Kentucky (GenBank accession no. AY463797), and Emek represents the boundaries of SGI2 from serovar Emek (GenBank accession no. AY963803). Derby W4 and W7 sequences, representing boundaries with SGI1-A and SGI1-C, respectively, are from reference 3. DT104 represents the equivalent regions of SGI1 from serovar Typhimurium DT104 (GenBank accession no. AF261825), and dotted lines indicate bases that are not present; the gap labeled retron phage delineates a segment that is missing from serovar Typhimurium. PtB SPB6 is from the genome sequence of strain SPB7 (GenBank accession no. CP000886). The initiation codon for the yidY gene and primers located in the chromosomal region and used to generate the PCR products sequenced are underlined. The recombination crossover region (shown in boldface), which includes the termination codon of the thdF gene (underlined), marks the boundaries between the SGI and the chromosome. Differences from the SGI1-K sequence are white on black.

A new SGI1 variant.

Three strains, isolated in 1995, 1998, and 1999, were susceptible to sulfonamides, and further isolates of this type were recovered in 2002 and 2005 (Table 1). SGI1 contains one complete and one partial copy of the sul1 gene in the two 3′-CS regions (Fig. 1), and the multiplex primers used above can detect both of these regions. To determine if a complete copy of the sul1 gene was present, a primer preceding the beginning of sul1 was combined with a primer in IS6100 and a PCR product of the predicted size was obtained, indicating that SRC181, SRC195, and SRC198 all include a complete copy of the sul1 gene. The amplified fragment from these three strains was sequenced, and all contained the same G-to-T single-base change at base 442 in the sul1 gene, which creates a premature termination codon at codon 148. This variant was named SGI1-R. Though this base (base 890 of the 3′-CS) is present in the left partial copy of the 3′-CS (3′-CS 1), which extends from positions 1 to 978, the left copy does not appear to contain the same mutation. A derivative equivalent to SGI1-C that had lost the central region of SGI1 but retained the aadA2 gene was recovered from the frozen stock of the sulfonamide-susceptible strain SRC195 and was found to be resistant to sulfamethoxazole. This indicates that it arose via recombination between the duplicated portions of the 3′-CS on the left and right, with the crossover located between the mutant base (bp 890) and the end of the partial copy on the left (bp 978).

Loss of antibiotic resistance.

Theoretically, SGI1-containing strains could revert to an antibiotic-susceptible phenotype if the complete island were excised from the chromosome and lost from a daughter cell on division, leading to SGI1-free subclones. Such loss appeared to occur frequently in one study of SGI1 in Salmonella serovar Typhimurium DT104 (7) but was very rare in another (12). Two variants of SGI1, SGI1-B and SGI1-C, that have lost some but not all of the resistance genes can arise by homologous recombination between the duplicated parts of the 5′-CS or 3′-CS found in SGI1 (Fig. 1). To determine if such events occur readily, strains SRC49 and SRC50 were grown for more than 460 generations in the absence of antibiotic selection, and a total of 1,140 single colonies were isolated and screened for the resistance markers. The loss of all resistance phenotypes, indicating the loss of SGI1, was not detected. More stringent tests, using PCR on DNA isolated from the culture to detect the ancestral chromosomal configuration of thdF adjacent to yidY that would reform in cells that have lost the SGI, did not detect any SGI1 deletions. However, small numbers of cells that were resistant to either ApSu (one colony) or SmSpSu (three colonies) were recovered. These were shown to retain SGI1 but to have lost the central region containing the floR and tetA(G) genes together with either the left or right cassette to give rise to SGI1-C and SGI1-B, respectively (Fig. 1). SRC191, SRC289, and SRC295 presumably arose in this way.

DISCUSSION

The availability of early isolates of multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ (ApCmSmSpSuTc phenotype) containing SGI1 provided an opportunity to examine the global emergence of this new pathogen. Broadly, there are two possible ways that multiply antibiotic-resistant Salmonella strains carrying SGI1 could have emerged in several different countries in the last decade: either an SGI1-containing variant arose once, i.e., there was a single SGI1 acquisition event, and this cell was subsequently amplified and spread globally, or an SGI1-containing strain arose independently in each country. The two isolates recovered in 1995 in Australia represent the earliest ones known, and since 1997 they have occurred every year. However, only small numbers, up to 23 isolates, are reported in any one year, and they occur in all states of Australia (Table 2), indicating that infection occurs sporadically. Here, we have used multiple measures to assess the relatedness of the SGI1-containing strains isolated in Australia, and though some variations were seen, all molecular measures indicate that they have a common origin. The sequences of small stretches of the chromosome on either side of SGI1, which were identical in all of the SGI1-containing strains tested, may provide the most definitive measure of this relationship. While it is not possible to eliminate the possibility that SGI1 has been acquired more than once by very closely related cells, the considerable diversity of strains belonging to Salmonella serovar Paratyphi B dT⁺ argues against it.

It is possible that this clone is globally distributed. However, it seems unlikely that this strain arose in Australia, because Salmonella serovar Typhimurium DT104 strains are not endemic and other SGI1-contining serovars are rare (18). In the earliest time period, closely related strains also were recovered occasionally from travelers returning from Indonesia, and an SGI1-containing Salmonella serovar Paratyphi B dT⁺ strain was found in a tropical fish in Singapore in 1997 (22). Over the next few years, Salmonella serovar Paratyphi B dT⁺ isolates with the ApCmSmSpSuTc resistance phenotype and containing SGI1 were detected in other countries. Where comparison was possible, namely for PFGE and IS200 profiles, the properties of the isolates described in this study are close to those reported for strains isolated in France (34). In both this study and the French study, several non-SGI1-containing strains included as references each gave a different IS200 profile, which strengthens the conclusion of identity both within and between the data sets. This suggests that the isolates from other countries also have the same origin, though comparable data for isolates from the United Kingdom and Canada (23, 32) are not available for comparison. Though further data, such as the chromosomal flanking sequences, that are readily comparable between laboratories would strengthen this conclusion, it appears possible that a single cell originally acquired SGI1 and its progeny were disseminated globally.

Travel and the global trading of food are well-established routes for the dissemination of bacterial strains, including pathogens. However, there is clear evidence that at least some of the Salmonella serovar Paratyphi B infections were derived from tropical aquariums in the patient's homes (17, 25). Ornamental tropical fish for home aquariums are produced mainly in Southeast Asia but are traded globally (25), raising the possibility that this is the route by which the original strain was spread. Under the selective pressure of antibiotic use—tropical fish-rearing industries use antibiotics such as chloromycetin, florfenicol, tetracycline, and sulfadiazine (25)—the first cell would have gained a selective advantage, leading to its amplification. The export of tropical fish to Australia and other countries around the globe and then to pet shops in each country would provide a simple means for the clone to be disseminated.

The isolates that contain SGI1 display a wide variety of phage types, and a few other variant types, such as a mutation in the sul1 gene, were detected. It is to be expected that some variations will arise during the expansion of a clone, and lysogenization with a new phage can explain the observed differences in phage types, which also have been seen in other studies of multiply antibiotic-resistant serovar Paratyphi B dT⁺ strains (32, 34). Recently it has been reported that the acquisition of a new phage is the most frequent source of variation among S. Typhimurium strains (29). Hence, it appears that phage typing is useful as an epidemiological marker only for the short term. A number of related but slightly different XbaI PFGE patterns also were observed in previous studies (32, 34), and a recent study with serovar Typhimurium DT104 strains has identified changes in the phage and plasmid content as the source of these differences (11). Occasional differences in other features, for example, the additional band seen in the IS200 profiles of strains isolated in France, would arise if IS200 moved to an additional location in the chromosome. Hence, the observations reported here support the conclusion that all multiply antibiotic-resistant Salmonella serovar Paratyphi B dT⁺ found in Australia, Indonesia, and France and potentially globally have a single point of origin but that variations continue to arise over time. As the transfer of SGI1 cannot occur without assistance, it is likely to be infrequent in the wild, and hence a similar scenario is predicted for other SGI-containing Salmonella serovars.