This study describes the identification of the insertion site and partial characterization of a 43-kb region harboring the genes associated with the penta-resistant phenotype of a Canadian isolate of Salmonella enterica Typhymurium DT104 labelled 96-5227. The 43-kb fragment, here referred to as Salmonella genomic island I (SgiI), was found in the genome of S. enterica Typhymurium between the thdf and a prophage CP-4-like integrase (int2) gene and is flanked by an imperfect 18-bp direct repeat. A region downstream of sulI in the right end of SgiI contained four open reading frames which includes an IS6100 element, and a 2-kb region from the left end contained two open reading frames which showed homology to an integrase and an excisionase. Furthermore, a 1.9-kb retron sequence located between int2 and yidY was identified which may be unique to the S. enterica Typhymurium genome. The int-retron sequence is flanked by a 27-bp imperfect direct repeat.
Salmonella enterica Typhymurium DT104
The emergence of multiple antibiotic-resistant pathogens is an increasing problem worldwide. Infections caused by these organisms increase healthcare costs and limit the choice of treatment. Salmonella enterica Typhymurium phagetype DT104 (hereafter referred to as DT104) that is resistant to ampicillin, chloramphenicol, streptomycin, sulfonamides, and tetracycline (commonly abbreviated ACSSuT) has recently emerged in a number of countries. Multidrug resistant DT104 is currently the second most prevalent Salmonella phagetype of S. enterica Typhymurium in England and Wales [1,2] and is increasing in the USA [3,4] and Canada .
The genes which confer the multidrug resistance (MDR) in DT104 have been identified by a number of investigators [6–9]. In Danish isolates of DT104, two class 1 integrons have been identified encoding resistance to ampicillin (pse-1) and streptomycin (ant (3″)-Ia) . UK isolates of DT104 were found to contain two class 1 integrons that encoded ampicillin (pse-1) and streptomycin (aadA2) resistance and were localized to a 10-kb chromosomal fragment . Briggs and Fratamico have recently sequenced the MDR region in a USA strain of DT104 and the genes conferring the penta-resistance profile were located on a 13 078-bp fragment of chromosomal DNA . The region contained the two integrons described in UK isolates and were also associated with sulI (sulfonamide resistance). Tetracycline resistance encoded by tet(G) and chloramphenicol resistance encoded by pasppflo-like gene were localized between the two integrons . The identical genes and order were recently reported in a Canadian isolate of DT104 suggesting a clonal dissemination of the multidrug resistant strain .
In this report we describe the identification, size, and location of a genetic element, here referred to as Salmonella genomic island I (SgiI), harboring the antimicrobial resistance genes responsible for the MDR phenotype in S. enterica Typhymurium DT104 in the previously reported Canadian strain . Furthermore, we have identified a retron sequence which may be specific to S. enterica Typhymurium.
2 Materials and methods
2.1 Bacterial strains, bacteriophages, and media
Bacterial strains used in this study were obtained from the National Laboratory for Enteric Pathogens, LCDC, Winnipeg, Man., Canada. Escherichia coli LE392 was used for propagating phages, and λEMBL3 (Promega, Madison, WI, USA) was used in phage cloning experiments. All strains were grown at 37°C in Brain–Heart Infusion broth or on LB plates. Stock cultures were stored at −70°C in Microbank™ vials (Pro-Lab Diagnostics, Richmond Hill, Ont., Canada).
2.2 Recombinant DNA methodology and PCR
A genomic DNA library was constructed in λEMBL3 using 15–20 kb Sau3A genomic DNA fragments from S. enterica Typhymurium 96-5227 using standard methods . The library was screened by plaque lift using Hybond N+membrane and hybridization carried out with appropriate DNA probes as recommended by the manufacturer (Amersham Pharmacia Biotech, Arlington Heights, IL, USA). DNA probes were labelled and hybrids were detected using an ECL system (Amersham Pharmacia Biotech, Arlington Heights, IL, USA). Synthesis of oligonucleotides and DNA sequencing was carried out in the DNA Core Facility (Laboratory Centre for Disease Control, Health Canada, Winnipeg, Canada). Standard PCR reactions were carried out using 2.5 U AmpliTaq Gold DNA polymerase in PCR Buffer II (PE Applied Biosystems, Foster City, CA, USA), containing 3 mM MgCl2, 0.2 mM dNTPs, 0.5 μM primers, and 50 ng DNA. PCR cycling conditions were 95°C for 10 min followed by 30 cycles of 94°C for 30 s, 54°C for 30 s, 72°C for 1 min, followed by a final extension at 72°C for 5 min. Other annealing temperatures may have been used depending on the Tm of the primers in the reaction. Long PCR was carried out using the Expand™ Long Template PCR System (Roche Diagnostics, Laval, Que., Canada) as recommended by the manufacturer. The primers used in a nested-PCR assay for detection of SgiI excision were: U7-L11, 5′-TGAAGCGGCCATTGTCACCG; C9-L1, 5′-TCCCCTGGAAATTGCATAG; U7-L12, 5′-ACACCTTGAGCAGGGCAAAG; and C9-L2, 5′-AGCAAGTGTGCGTAATTTGG. The following primers were used to test for the presence/absence of SgiI and/or retronphage-like sequences between the thdf and yidY genes; primers: U7-L9, 5′-TGAAGATGTGCTGGAGCTAC; 104-D, 5′-ACCAGGCAAAACTACACAG; LJ-R1, 5′-AGTTCTAAAGGTTCGTAGTCG; 104-RJ, 5′-CTGACGAGCTGAAGCGAATTG; C9-L2, 5′-AGCAAGTGTGCGTAATTTGG; and DB-T7, 5′-ACCAGTGTTTTGTTGATTATGC.
2.3 Cloning of the MDR region
The 96-5227 genomic library was first screened using a labelled amplicon of tet(G)  and λSt2 was isolated. Overlapping clones were isolated by obtaining approximately 500 bp of sequence from each end of the insert in λSt2 and then using PCR to generate new probes for a further round of screening of the library. In this way we obtained a series of overlapping clones covering a region of approximately 60 kb that contained within it the MDR region. Identification of genes related to homologs in the genome of E. coli K12 , S. typhi (these sequence data were produced by the Salmonella typhi Sequencing Group at the Sanger Centre and can be obtained from ftp://ftp.sanger.ac.uk/pub/pathogens/st/, Sanger Centre, Wellcome Trust Genome Campus, UK), and S. enterica Typhymurium LT2 (Genome Sequencing Center, Washington University, St. Louis, MO, USA. http://genome.wustl.edu/gsc/projects/bacterial/salmonella.shtml, personal communication) allowed us to ascertain whether a clone harbored a potential junction site between a putative genomic island and chromosomal DNA. The complete nucleotide sequence of the genes surrounding the left and right junctions of the genomic island has been deposited in the GenBank database under accession numbers AF261825 and AF261826, respectively.
2.4 Computer-aided analysis
The sequence obtained was used to query the GenBank database in order to identify putative genes using the BLAST suite of programs  via the world wide web interface of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih/BLAST).
2.5 Antimicrobial susceptibility testing
Antimicrobial susceptibilities were performed following the NCCLS agar dilution guidelines and the interpretation was based on the criteria in NCCLS M7-A4 .
3.1 Characterization of clones upstream of MDR region
A 60-kb region containing the MDR cluster, which was partially characterized in this study, is shown in Fig. 1. An initial clone, λSt2, was identified from the genomic library of 96-5227 using a tet(G) probe which has been shown previously to be associated with the MDR region of DT104 [6,7]. Genomic walking was used to identify a clone upstream of λSt2, and labelled λSt3-1. Partial sequence analysis of the left end of λSt3-1 did not reveal any significant homology to known E. coli, S. typhi or S. enterica Typhymurium sequences which indicated this clone did not harbor native genomic DNA. PCR was used to amplify a probe from the left end of λSt3-1 which was used in a further round of library screening to isolate λU7. Partial sequence analysis of the left end of λU7 revealed an open reading frame (ORF) which showed homology to the recF gene in E. coli. Using sequences from the S. enterica Typhymurium LT2 genome sequencing project, PCR mapping was used to identify and order the recF-dnaN-rpmH-rnpA-yidC-thdf genes in λU7 (data not shown). The gene int2 found directly downstream of thdf in the S. enterica Typhymurium LT2 genome was not identified in λU7 suggesting SgiI harboring the MDR was inserted between these two genes. We sequenced a region of 2047 bp from λU7 (accession no. AF261825) and identified a partial ORF whose product has 93% identity to the Thdf protein from E. coli, a putative integrase gene (int1), and a putative excisionase gene (xis) which overlaps the int1 gene by 4 bp and is transcribed in the opposite direction (Fig. 1B). The coordinates, percent homology to other proteins, and G+C content of the ORFs are described in Table 1. By the identity and arrangement of the genes, we surmised that this region may contain one insertion point for a transposon or prophage-like sequence associated with the MDR region.
Characterization of the junction regions of SgiI. A: Restriction map of the 60-kb region harboring the MDR from S. enterica Typhymurium 96-5227. Restriction enzymes used were as follows: E, EcoRI; A, AvrII; X, XbaI; S, SalI. Lines above the restriction map denote the inserts from the overlapping lambda clones. B: Predicted ORFs from the sequences obtained from the SgiI junction regions. Arrows above ORFs indicate direction of transcription.
The ORFs in and around the SgiI DNA and the identities of their deduced products to extant protein sequences
Description of gene product (accession no.)
Left junction region: accession no. AF261825
THDF protein (AE000447)
Right junction region: accession no. AF261826
dihydropteroate synthase (AP000342)
putative acetyltransferase (AP000342)
hypothetical protein (U38230)
hypothetical protein (M64113)
reverse transcriptase (AJ249205)
Asterisk indicates that this ORF contains an in-frame stop codon (see text) however the complete ORF amino acid product with the stop codon represented by an X was used as the query sequence in the BLAST search.
aORFs or genes in parentheses are located outside of the genomic island.
bNucleotide position in the sequence deposited under accession no. AF261825 or accession no. AF261826.
cExpect value. Sometimes returned as zero for matches with 100% identity.
dn.s.h – no significant homology. E value >1.
3.2 Characterization of clones downstream of the MDR region
Using the genomic walking technique described in Section 3.1, two clones were isolated downstream of λSt2 labelled λC9 and λC12, respectively. We sequenced 8640 bp from lC12 (accession no. AF261826) and identified nine ORFs (Fig. 1B). The coordinates, percent homology to other proteins, and G+C content of the ORFs are listed in Table 1. The first two ORFs, the 3′-end of the previously identified sulI gene and orf5, constitute the 3′-CS of the class I integron that contains the pse-1 β-lactamase cassette previously identified in the MDR region [6,7]. Downstream of orf5 is orf6 which is associated with some but not all 3′-CS of class I integrons . IS6100, a mobile genetic element first identified in Mycobacterium fortuitum and also found in Tn1696 of Pseudomonas aeruginosa and pACM1 of Klebsiella oxytoca, was identified downstream of orf6. Following IS6100 is an unique ORF, labelled SgiI1 (Salmonella genomic island I orf1), which exhibits no significant homology to any entry in GenBank. A prophage-like integrase pseudogene, int2, that has an in-frame stop codon (TGA) beginning 492 bp downstream of the start codon is located adjacent to SgiI1. Downstream of int2 are two ORFs, urt (upstream of reverse transcriptase) and rt, which overlap by 1 bp and appear to constitute a bacterial retron element. The arrangement most closely resembles that of retron Ec-73 which is part of retronphage φ-R73 in E. coli. In addition we have detected a large inverted repeat structure upstream of urt which may be part of the msd of the retron similar to Ec-73 and other retrons (Fig. 1B). We have identified a 27-bp direct repeat (five mismatches) labelled retDR-L and retDR-R flanking the int2 and rt genes (Fig. 1B) which may be an insertion point for the int2–urt–rt sequences. Beginning 802 bp downstream of the end of rt is a gene whose product exhibits 87% identity to the YidY protein of E. coli which itself has homology to the bicyclomycin resistance protein (bcr) of the E. coli genome.
3.3 Identification of the putative SgiI insertion site
Sequence analysis of the region directly downstream of thdf and upstream of int2 genes did not reveal any significant homology to sequences from the S. enterica Typhymurium LT2 genome sequencing project suggesting the insertion of SgiI is between these two genes. Analysis of these putative junction regions has revealed the last 18 bp of the thdf gene are imperfectly directly repeated (16/18 matches) 110 bp upstream of int2. We have named these direct repeats DR-L and DR-R (Fig. 1B). Thus, the MDR region of 96-5227 appears to be contained within a 43-kb genomic island (SgiI).
3.4 Test for excision of SgiI
To test whether SgiI behaves like a mobile genetic element, PCR experiments were performed to examine if spontaneous excision occurs at the DR-L and DR-R regions. Primers U7-L11 with C9-L2, and U7-L12 with C9-L1, were used for first and second rounds of PCR, respectively. Primers U7-L11 and U7-L12 were located in the thdf gene upstream of DR-L, and primers C9-L2 and C9-L1 were located downstream of DR-R (Fig. 2). If precise excision has occurred via DR-L and DR-R, 1440- and 393-bp products would be expected to be amplified. Using 96-5227 DNA as a template no product was detectable using the U7-L11 and C9-L2 primer pair. To detect whether an extremely low frequency of excision had occurred, an aliquot of the first PCR was used in a second round of PCR with the U7-L12 and C9-L1 primer pair, and again no product was visibly detectable (data not shown). Using S. enterica Typhymurium DT104 strain 97-05548 DNA as a template (DT104, no SgiI, see below), 1.4- and 0.4-kb products were obtained. Thus, SgiI in 96-5227 does not appear to excise from this location in the chromosome, or the frequency of excision is too low to be detected by this assay.
Schematic representation of the regions between thdf and yidY in the organisms listed. Open boxes represent the retDR-R and retDR-L and solid boxes represent DR-R and DR-L. Name and location of primers referred to in the text are shown. The drawing is not to scale.
3.5 The presence of SgiI and the retron in other Salmonella
We used PCR to screen several other Salmonella strains for the presence of SgiI and/or the retronphage-like sequence between the thdf and int2 genes (Table 2). Using a primer pair derived from thdf and yidY (U7-L9 and 104-D, respectively; Fig. 2), a 5.3-kb product was expected from strains harboring only the retronphage sequence, whereas a 1.3-kb product was expected if this element was absent. If SgiI was present, no PCR product would be expected due to the large size of the predicted amplicon (∼44 kb). Two DT104 ACSSuT and two ACSSuTK strains gave no product, three DT104 strains with antibiograms other than ACSSuT (including strain 97-05548) and two susceptible strains gave the 5.3-kb product, and eight non-Typhimurium serovars gave the 1.3-kb product (Table 2). The strains which gave no product were tested to confirm the presence of SgiI with a thdf primer and a primer downstream of DR-L (U7-L12 and LJ-R1, respectively; Fig. 2), and a primer from near the start of int2 and a primer downstream of DR-R (104-RJ and C9-L2, respectively; Fig. 2). The ACSSuT and ACSSuTK strains gave the products expected (both 0.5 kb) if they harbored SgiI, while a susceptible S. enterica Typhymurium and a S. heidelberg failed to produce products indicating the absence of SgiI (Table 2). The presence of the right end of the retroelement was confirmed in the S. enterica Typhymurium strains by using a primer immediately downstream of rt and the yidY primer (DB-T7 and 104-D, respectively; Fig. 2). All S. enterica Typhymurium strains tested gave the product expected (0.8 kb) while the S. heidelberg gave no product (Table 2).
The presence of SgiI and/or retron-like sequences between the thdf and yidY genes in various Salmonella strains as tested for by PCR
No. of isolates
PCR primer set, size of product (kb)
S. enterica Typhymurium
S. enterica Typhymurium
S. enterica Typhymurium
S. enterica Typhymurium
S. enterica Typhymurium
S. enterica Typhymurium
aResistance to ampicillin (A), chloramphenicol (C), streptomycin (S), sulfonamides (Su), tetracycline (T), kanamycin (K), and nitrofurantoin (F).
bNo PCR product expected under the conditions used due to the distance between primers.
cND – not done.
dOne strain tested.
The increasing prevalence of multidrug resistant DT104 has led to studies into the genetics of resistance responsible for the ACSSuT phenotype [6–9]. In this report we have determined the overall size of the of the element harboring the MDR genes of S. enterica Typhymurium DT104 strain 96-5227 to be 43 kb and located between the thdf and int2 genes located at 3.89 Mb in the Salmonella genome. Due to the large size of inserted DNA, the antimicrobial resistance genes harbored within the element, and the characteristics of the element, we have referred to the sequence as SgiI.
The genomic island harboring the resistance genes is flanked by a near perfect 18-bp direct repeat which appears to be a duplication of the 3′ end of the thdf gene whose product is involved in thiophene and furan oxidation in E. coli (Fig. 1). The duplication, in the form of a direct repeat, indicates a site-specific recombination event that results from the insertion of plasmids or prophages . This duplication has been observed in pathogenicity islands, however unlike the SgiI described in this report, the majority of pathogenicity islands described to date are found in tRNA genes . Our analysis has identified several genes related to phages or mobile genetic elements such as transposons or insertion elements. Immediately downstream of DR-L are the int1 and xis genes whose products exhibit between 29 and 31% identity to similar proteins from the Bacteroides fragilis transposon Tn4555, but whose arrangement is more similar to that of the int and xis genes found in the LT2 prophage Gifsy-1 . Upstream of DR-R we have identified the sulI and orf5 genes typically comprising the 3′-CS of the class I integrons and adjacent to orf5 is orf6 which has been associated with integron In5 . Downstream of orf6 is IS6100 which was originally identified in M. fortuitum and more recently described in pACM1 from K. oxytoca. Interestingly, the sulI–orf5–orf6–IS6100 sequence is identical in pACM1 and SgiI . As in pACM1, the IS6100 element is flanked by the last 123 bp of In2 located in Tn21, with each copy inverted relative to the other (see Fig. 1B). Downstream of this 123-bp sequence is part of the last 152 bp of In2. Finally, we have identified one more ORF, SgiI1, before the DR-R of the genomic island. This sequence had no significant homology to any entries deposited in GenBank. It is interesting to note that the two genes at the left junction region, int1 and xis, and SgiI1 at the right junction have similar G+C contents of 43, 46.1 and 47.3%, respectively, which may suggest a similar origin.
We have also identified a retronphage-like sequence located between the SgiI insertion site (DR-R) and the yidY gene (Figs. 1 and 2). This 4.3-kb fragment contains three genes, int2, urt, and rt which display homology at the amino acid sequence level (Table 1) to a previously described retronphage sequence in E. coli named φ-R73 . The φ-R73 retronphage contains an int at the 3′-end followed by 11 genes related to phage P4 followed by the retron sequence containing orf316 and rt. The gene order observed in strain 96-5227 suggests a similar phage may have inserted into the Salmonella genome. A nonsense mutation has been identified in the int2 gene (Fig. 1B) which may have rendered the phage inactive. Over time, this nonfunctional retronphage may have lost the majority of the phage-related genes leaving the int2 and retron sequences in the genome. The three genes (int2–urt–rt) are flanked by direct repeats consisting of 27 bp with five mismatches, labelled retDR-L and retDR-R (Fig. 1B), which may indicate a possible site of insertion for the cryptic phage sequences. The bacterial retron sequences synthesize satellite DNA called multicopy single-stranded DNA (msDNA) which has an extensive secondary structure. Located 91 bp upstream of urt is a 62-bp inverted repeat which has the potential to form a stem loop structure with a ΔG0 of −217 kJ/mol (Fig. 1B), and may be involved in the formation of msDNA. Further studies are needed to confirm this observation. We have not detected the retron sequence in four other Salmonella serotypes but have identified it in all S. enterica Typhymurium strains tested to date (Table 2). Using this limited data and the sequence data from S. enterica Typhymurium LT2 and S. typhi genome sequencing projects, the structure of the genome in this region in various Salmonella serotypes hints at a possible evolution of events involving several genetic elements. We hypothesize that a phage containing retron sequences inserted into the genome of S. enterica Typhymurium after it had diverged from S. typhi and other Salmonella serotypes. The SgiI appears to have inserted adjacent to the phage-retron in the S. enterica Typhymurium DT104 genome after it diverged from other S. enterica Typhymurium phage types (Fig. 2). Further studies with a wider range of Salmonella serotypes and phage types are underway to determine if this hypothesis is correct.
It has been suggested that the chromosomal integration of resistance genes may result in a stable structure which may persist even in the absence of selective pressure . Attempts to transfer the MDR phenotype of S. enterica Typhymurium DT104 to strains of E. coli have been unsuccessful , which provides support for this hypothesis. To further support the stability of SgiI in the genome, we could not detect the loss of the element using a PCR-based assay. This supports the hypothesis that even if antibiotics are withdrawn as growth promoters in the agricultural industry, there may be no significant effects on the current epidemic of DT104 ACSSuT in animals as has been suggested previously .
Studies are underway to elucidate the remaining sequence associated with the genomic island to better understand the origin and function of this region.
The authors wish to thank the Genome Sequencing Centre, Washington University, St. Louis and the Salmonella typhi Sequencing Group at the Sanger Centre for communication of DNA sequence data prior to publication. Also, we would like to thank Rae Bose and Romeo Hizon for performing the susceptibility testing, the National Laboratory for Enteric Pathogens for the identification of Salmonella and phagetyping, and Mr. Shaun Tyler and the staff of the DNA Core Facility, Bureau of Microbiology, LCDC for DNA sequencing and synthesis of oligonucleotides.
National Committee for Clinical Laboratory Standard (1997) Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically, 4th edn., approved standard. M7-A4 vol. 17, no. 2, NCCLS, Wayne, PA.