Characterization of sulphonamide-resistant Escherichia coli using comparison of sul2 gene sequences and multilocus sequence typing

The known genes encoding plasmid-borne sulphonamide resistance (Sul^R) are sul1, sul2 and sul3; each encodes a drug-resistant dihydropteroate synthase (DHPS) enzyme. Among them, the sul2 gene has previously been reported as the most prevalent one in Escherichia coli from pig and pork, poultry and poultry meat, cow and beef, mutton, human faeces and urinary tract infections (UTIs) (Blahna et al., 2006; Gow et al., 2008; Grape et al., 2003; Hammerum et al., 2006; Kerrn et al., 2002; Sunde & Norstrom, 2006; Trobos et al., 2008). Based on a few partial sequences, this gene seems highly conserved, whether it is carried on small non-conjugative or large conjugative resistance plasmids (Radstrom & Swedberg, 1988; Sorum & L'Abee-Lund, 2002). It has been suggested that horizontal gene transfer plays a much larger role than clonal expansion in the spread of trimethoprim–sulfamethoxazole resistance (Blahna et al., 2006), but there is a need to investigate this further.

Multilocus sequence typing (MLST) is used to characterize E. coli isolates, among other bacteria, on the basis of sequence variation of seven housekeeping loci. The different nucleotide sequences of each locus are assigned different allele numbers, and then each strain is defined by the alleles of the seven loci (allelic profile). Closely related organisms can be grouped in clonal complexes. Analyses of E. coli isolates by MLST help to determine the genetic relatedness of the isolates, which is important for molecular epidemiological and evolutionary studies (Feil et al., 2004). MLST is used to study the population biology of micro-organisms and provides an understanding of the population structure of important pathogens such as E. coli. Isolates that exhibit similar or identical MLST genotypes are very closely related and have descended from a recent common ancestor (Turner & Feil, 2007).

In the present study, we applied MLST, followed by PFGE typing, to characterize sul2-positive strains of E. coli from different sources, and sequenced the sul2 gene, with the following aims: (1) to see if the same clonal complexes contained resistant strains from different sources, which would be an indication that resistant strains are transferred between niches, and (2) to investigate whether horizontal gene transfer of the sul2 gene seems to have happened frequently.

Part of this study was presented at the 47th ICAAC, Chicago (USA), September 2007, and at the ASM Conference on Antimicrobial Resistance in Zoonotic Bacteria and Food-borne Pathogens, Copenhagen (Denmark), June 2008.

Bacterial isolates.
A total of 68 E. coli from different sources (pig, pork, poultry, poultry meat, mutton, healthy humans, blood stream infections and UTI), years (1969–2004) and countries (Denmark and Norway) were chosen for the study (Table 1). Except for the human blood isolates and the Norwegian animal isolates, the isolates were obtained from other studies (Hammerum et al., 2006; Kerrn et al., 2002; Sunde & Norstrom, 2006; Trobos et al., 2008). They were selected by their resistance to sulphonamides and the presence of the sul2 gene. A breakpoint of ≥512 µg ml^–1 was used for determining sulphamethoxazole resistance according to the Clinical Laboratory Standards Institute (CLSI).

Table 1. Variations in the sul2 gene of E. coli strains of animal, meat and human origin

MLST.
We used the MLST protocol standardized for E. coli maintained at the Max Planck Institute for Infection Biology website (). When two possible primers were given for the same direction, we chose the following: fumCR, gyrBR, purAF, recAF and recAR.

Sequencing of the PCR products was performed using the services of Macrogen. Sequences of the seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, recA) were analysed by BioNumerics using a special plug-in for MLST of E. coli.

PCR and sequencing of the sul2 gene.
Presence of the sul2 gene was investigated by PCR using primers Sul2-F (5'-TTT CGG CAT CGT CAA CAT AA-3') and Sul2-R (5'-GTG TGT GCG GAT GAA GTC AG-3)' (primer concentration 10 µM). A multiplex PCR kit (Qiagen) was used with the following amplification conditions: heating for 15 min at 94 °C, followed by 30 cycles at 94 °C for 30 s, 48 °C for 90 s and 72 °C for 90 s, followed by 72 °C for 10 min. E. coli NCTC 50020 was used as positive control and E. coli DH5α plasmid pUVP4401 was the negative control.

The sul2 PCR products (727 bp) were purified by using the NucleoFast 96 PCR plates (Macherey-Nagel) and sequenced at Macrogen. The sequences were analysed with KODON software (Applied-Maths) and compared by multiple alignment using CLUSTAL W and MEGA4 software (Tamura et al., 2007; Thompson et al., 1997).

PFGE.
PFGE was used to type E. coli from poultry, poultry meat and Danish blood infections. The one-day PFGE Standardized Laboratory Protocol for Molecular Subtyping of E. coli O157 : H7 from PulseNet () was used with the following variation: the restriction enzyme used was SpeI (20 U µl^–1) (5 µl 10x buffer 2, 42 µl sterile water, 1 µl BSA and 2 µl SpeI). Digestions were performed in 50 µl volumes.

Computer-assisted analysis of SpeI profiles was performed using the BioNumerics software (Applied Maths). The dendrogram was constructed using Dice similarity coefficients of SpeI PFGE patterns and the UPGMA clustering method (see Supplementary Fig. S1, available with the online version of this paper).

Diversity of sul2-positive E. coli STs among animals, food, humans and infections
A minimal spanning tree (MS_TREE) shows the distribution of the multilocus sequence types (STs) and their relation to the sul2 gene variants (see below) of the 68 E. coli isolates from all the sources (see Fig. 1).

(26K): Fig. 1. Distribution of MLST STs and their relation to the sul2 gene variants within a minimal spanning tree (MSTREE) of 68 Escherichia coli from various sources. Each ST is represented by a circle. Ellipses containing several STs indicate clonal complexes (CC). The lines connecting the STs within a clonal complex are shaded in colours. Black lines connecting pairs of STs indicate that they differ in one allele (thick lines), two alleles (thin), or three to seven alleles (dotted). Green circles represent isolates that contain the sul2(a) gene variant, red circles represent the sul2(b) variant and blue circles represent both variants [except ST10, which contains the three variants: sul2(a), sul2(b) and sul2(c)]. S, swine; PK, pork; P, poultry; PM, poultry meat; HH, healthy human; BI, blood infection; UTI, urinary tract infection.

Among the 68 E. coli isolates studied, 45 different STs were identified (Supplementary Table S1). The largest group of isolates (24 %) was part of the clonal complex ST10 (CC10). They originated from pigs (n=2), pork (n=2), poultry (n=3), poultry meat (n=2), healthy humans (n=6) and blood infection (n=1). This was not surprising, since CC10 is the biggest clonal complex of the species, and E. coli isolates from very different hosts are assigned to it (e.g. healthy human, UTI, septicaemia, diarrhoea, pig, dog, chicken and soil) (). CC10 comprised mainly the non-pathogenic E. coli isolates of our study. One-third of the pig (3 out of 9) and pork (4 out of 11) isolates belonged to the clonal complex CC23. A UTI isolate was also associated with this group.

Seven isolates were part of CC168; they were isolated from poultry (n=1), pig (n=2), pork (n=1), blood infections (n=2) and UTI (n=1). CC350 contained E. coli from poultry meat (n=2) and blood infection (n=1). Three isolates from poultry meat, a healthy human and a blood infection belonged to CC69. This clonal complex is mainly represented by human UTI and bacteraemia E. coli isolates, although it has also been found in cow and sheep isolates (Tartof et al., 2005).

Fig. 2 shows an eBURST diagram of the current E. coli MLST global population structure. The spatial distribution of Sul^R E. coli in the population seemed random; however, we have not tested this statistically. The Sul^R strains typed in this study were scattered among the rest of the Sul^R and Sul^S E. coli from all over the world. This indicates that there is not a cluster of this specific resistant population, and that all E. coli could potentially become Sul^R.

(21K): Fig. 2. Population snapshot of E. coli clusters of related and unrelated STs within the entire E. coli MLST database (performed using eBURST v3). This diagram compares the STs identified in this study (SulR E. coli, with pink haloes) with all the E. coli isolates in the MLST database (SulS and SulR). It does not show the genetic distance between unrelated STs and CCs. In each CC the primary founder is at a central position. A node between two STs represents a single-locus variation. [Performed on 11 September 2008.]

Comparison of sul2 gene sequences
In total six point mutations were recognized (Fig. 3). We decided to name these variants of the sul2 genes in the following manner: if a mutation resulted in an amino acid change in the DHPS protein we gave it a new letter. Extra non-synonymous mutations are simply named as the sul2 variant they derive from.

(13K): Fig. 3. Mutation sites in the sequence of the sul2 gene: representation of the three sul2 gene variants [sul2(a), sul2(b) and sul2(c)]. Mutations are at positions 159, 197 and 427. The extra mutations in three isolates derive from the sul2(a) type. These isolates contain an extra point mutation at position 288 (A instead of G), position 489 (A instead of T) and position 672 (T instead of G), respectively.

The first mutation, at position 159, was a silent thymine (T) or cytosine (C) third-codon substitution. The second, at position 197, was a C or adenine (A) second-codon non-silent substitution leading to alanine or glutamic acid, respectively. Almost all of the E. coli strains tested fell into one of these two variants: sul2(a) (TC variant) or sul2(b) (CA variant) (GenBank accession numbers FJ200242 and FJ200243, respectively) (Table 1). Two isolates, within the TC variant, contained an extra non-synonymous mutation at position 427 (A instead of G); this mutation was named sul2(c) (GenBank accession number EU653286).

The sul2(a) variant was the most frequently found among the strains tested. Three strains with the sul2(a) variant contained an extra synonymous mutation at position 288, 489 or 672 (GenBank accession numbers EU653287, EU653288 and EU653289, respectively) (see Fig. 3).

This study confirms, as previously reported on a smaller sample (Radstrom & Swedberg, 1988) that the sul2 gene is highly conserved, with few point mutations, though not identical as was previously concluded (Sorum & L'Abee-Lund, 2002). Presence of only two main types of the sul2 gene in both human and animal isolates, irrespective of time and geography, suggests horizontal transfer of the gene. It has been concluded by others that this had probably happened within a short time, since signs of more point mutations, indicative of genetic drift, were not found (Sørum & L'Abee-Lund, 2002). However, there may also be a second possible explanation, namely in order to function efficiently as a DHPS enzyme, the protein cannot tolerate many changes to its amino acid sequence. Therefore, it could also have been that an ancient transfer occurred which resulted in a wide dissemination.

The two main sul2 STs were present in the same clonal lineage. This indicates that horizontal gene transfer has happened several times, into the same clonal lineage, i.e. spread of sul2 happens independently of the evolution of the clonal lineage.

Link between poultry, poultry meat and human blood E. coli isolates and their sul2 gene ST
Interestingly, all poultry, poultry meat, and Danish human blood isolates had the same sul2(a) variant and some of these strains clustered under the same multilocus STs (ST10, ST57 and ST69) (Supplementary Tables S1 and S2). These isolates were PFGE typed in order to dismiss clonality (data not shown). We found that many PFGE profiles clustered (≥80 % genomic similarity) (Supplementary Fig. S1). Most clustered according to their source, but there were exceptions to this (Supplementary Fig. S1). One cluster associated two poultry isolates, two isolates from poultry meat and one from a blood infection. Another linked one poultry strain with five human blood isolates. This is an indication that poultry isolates of E. coli may be a source of extra-intestinal E. coli infection in humans; however, investigations with more strains are needed to confirm this.

Transfer mechanism of sul2
Resistance genes can spread via two phenomena, horizontal gene transfer and clonal expansion. In this study, we found the same MLST STs, clonal complexes and sul2 sequence among E. coli isolates from very different ecological niches, regions and years, just as we found the same sul2 gene variant among different MLST types. In the first case, this suggests that strains were of clonal origin with a stable presence of sul2, while the other case indicates a widespread horizontal transfer of sul2 between different hosts. It may be that clonal lineages differ in their ability to act as recipients for horizontally transferred genes, in particular plasmid-encoded genes such as sul2.

To conclude, this study shows that the same type of E. coli carrying an identical sul2 gene can colonize both animals and humans, and that strains which can be found among animals may be implicated in human infections such as UTI and septicaemia. Whether the strains or the sul2 gene transfer from animals to humans, or vice versa, cannot be assessed from the current study, due to the lack of a specific sul2 type colonizing each of the hosts.

Bacterial strains were provided by DANMAP (the Danish Integrated Antimicrobial Resistance and Research Program) and NORM/NORM-VET (Usage of Antimicrobial Agents and Occurrence of Antimicrobial Resistance in Norway). This study is part of the multi-disciplinary Early Stage Training (EST) site: TRAINAU Training Risk-Assessment In Non-human Antibiotic Usage. It is funded by a European Marie Curie Host Fellowship for EST (MEST-CT-2004-007819). The study was furthermore supported by a grant from the Scandinavian Society for Antimicrobial Chemotherapy.

Edited by: S. D. Bentley