Global Emergence of Trimethoprim/Sulfamethoxazole Resistance in Stenotrophomonas maltophilia Mediated by Acquisition of sul Genes

The first sul2 genes have been found in S. maltophilia from several different countries.

N osocomial Stenotrophomonas maltophilia are intrinsically resistant to a plethora of antimicrobial agents that severely limit commonly used empiric standard antimicrobial therapies. S. maltophilia is resistant to many β-lactams, β-lactamase inhibitors, and aminoglycosides (1,2). A recent survey of SENTRY (www.jmilabs.com) Antimicrobial Surveillance Program isolates indicated that the newer fluoroquinolones demonstrated good efficacy; the most active were levofloxacin (6.5% resistance) and gatifloxacin (14.1%) (3). Furthermore, the resistance to the polymyxins (20%-32%) is higher than observed in Pseudomonas aeruginosa (3,4). Because of low resistance levels (≈5%), trimethoprim/sulfamethoxazole (TMP/ SMX) remains the therapy of choice worldwide. A recent study encompassing data from Europe, Latin America, and North America indicates that the level of resistance to TMP/SMX is 3.8%; however, previous studies indicate that the level is higher in Latin America than North America (5,6). Although surveillance studies are few, resistance to TMP/SMX appears to be emerging, and recent in vitro modeling studies have shown that combination therapies of TMP/SMX plus ciprofloxacin and TMP/SMX plus tobramycin exhibit a greater killing capacity then TMP/SMX alone (7,8).
S. maltophilia exhibits an array of mechanisms that singularly or collectively contribute to its multidrug resistance status. Intrinsic resistance includes inducible efflux pumps (2) and multiple β-lactamase expression (1) but not mutations in the quinolone resistance-determining region (9). In addition, S. maltophilia can acquire resistance through integrons, transposons, and plasmids (10). Recently, class 1 integrons have been characterized from S. maltophilia strains isolated in Argentina and Taiwan, which indicates that they contribute to TMP/SMX resistance through the sul1 gene carried as part of the 3′ end of the class 1 integron (10).
In addition to class 1 integrons, other mobile elements are associated with sul genes. For example, Vibrio cholerae serogroup 0139 is resistant to several antimicrobial agents, including SMX, and it has been recently shown that the sul2 gene was part of a cluster located on a novel genetic element of the integrative conjugative element group named SXT. The resistance genes harbored by SXT are embedded in a composite transposon-like structure and were probably acquired recently (11). Within this antimicrobial drug resistance region, an insertion element common region (ISCR) sequence, ISCR2, is adjacent to a sul2 gene that moves by 1-ended transposition. Thus, the possibility exists that sul2 genes can transfer intra-and intergenerically, including into S. maltophilia. Herein, we describe the molecular characterization of an international collection of S. maltophilia isolates and determine their mechanism of resistance to TMP/SMX, including the first report of sul2 genes and the first description of insertion element common region (ISCR) elements carried in S. maltophilia.

Bacterial Strains
During 1998-2003, a total of 1,744 S. maltophilia isolates collected worldwide were forwarded to the SENTRY Program (Europe, USA, and Australia) and tested for antimicrobial drug susceptibility. A TMP/SMX resistance phenotype was demonstrated for 71. From these isolates, 25 nonclonal strains from patients in North America, Latin America, and Europe were analyzed by using molecular methods together with 30 representative isolates that were TMP/SMX-susceptible. Isolates were identified by using the Vitek System and confirmed by using API20NE (bioMérieux, Hazelwood, MO, USA).

Susceptibility Methods
Isolates were tested for susceptibility to TMP/SMX according to procedures of the Clinical and Laboratory Standards Institute (CLSI, formerly the National Committee for Clinical Laboratory Standards [NCCLS]) (12,13) by using broth microdilution methods (TREK Diagnostics, Cleveland, OH, USA). MIC results were confirmed with TMP/SMX. Etests were performed according to the manufacturer's directions (AB Biodisk, Solna, Sweden).

Molecular Materials
PCR primers were purchased from Sigma-Genosys Ltd. (Pampisford, UK) and are listed in the Table. General reagents for DNA manipulation were obtained from Invitrogen (Groningen, the Netherlands). All other reagents were obtained from Sigma Chemical Co. or BDH (both of Poole, England, UK).

Strain Typing
Clonality among the S. maltophilia isolates was assessed by pulsed-field gel electrophoresis (PFGE) fol-lowed by XbaI digestion of genomic DNA. This assessment was conducted according to the standard 1-day protocol (16).

Plasmid Isolation
Bacterial plasmids were isolated by the alkaline lysis method described by Grinsted and Bennett (17). Essentially, an overnight 10-mL culture was centrifuged (12,000× g) and suspended in water (250 µL) before 200 µL of lysis solution (0.2 mol/L NaOH, 1% sodium dodecyl sulfate [SDS]) was added. After lysis, 125 µL of neutralizing solution (0.3 mol/L potassium acetate, 1 mmol/L EDTA) was added. After precipitation, the suspension was centrifuged (12,000× g) and washed twice with 500 µL of a 50/50 (v/v) phenol/chloroform solution. The DNA was precipitated from the solution with the addition of 0.7 volumes of iso-amyl alcohol. The DNA/RNA pellet was washed twice in 1 mL 70% ethanol before being dried. The DNA was dissolved in 30 µL with 0.1 U RNase.

Southern Hybridization
ISCR and sul2 PCR amplicons generated with primers CRF/CRFF-r were labeled with P 32 -CTP by random primer extension by using a commercially available kit (Stratagene, Amsterdam, the Netherlands) according to the manufacturer's instructions. Unincorporated nucleotides were removed by passing the labeled DNA through a Sephadex column (Nick column, Pharmacia Bio-tech, Uppsala, Sweden).

PCR Analysis
The presence of class 1 integrons in each strain was assessed by using class 1 specific primers. Gene cassettes embedded within the class 1 integrons were determined by using primers listed in the Table. Isolates were also screened for sul1, sul2, and sul3 by using sul1-F and -R, sul2-F and -R, and sul3-F and -R, respectively. Seven positive class 1 integron PCR products were chosen randomly, extracted from agarose gels after size separation, and sequenced with IntF, IntR, and custom-made oligonucleotide primers (Table).
The presence of ISCR elements in each strain was also determined by using primers CRF/CRFF-r designed to amplify the same 700-bp fragment internal to the open reading frames (ORFS) of ISCR1-5 (Table). Full-length ISCR2 elements were amplified with primers designed to target the ends of ISCR2. Primers used to amplify genes often associated with ISCR2 or ISCR3 are also given (Table). Because dhfr genes are associated with ISCR elements, we also performed molecular analysis of them.
PCRs were conducted in a final volume of 20 µL by using 10 µL ABgene Expand Hi-fidelity Master Mix (ABgene House, Surrey, UK). Primers were used at final concentrations of 10 µmol/L, and 1 µL of an overnight bacterial culture (optical density 1.0 at 600 nm) was added as source of DNA template. The cycling parameters were as follows: 95°C for 5 min, followed by 30 cycles of 95°C for 1 min, 55°C for 1 min, and 68°C for 1-4 min, depending on the sequence to be amplified, and ending with a 5min incubation at 68°C.

DNA Sequencing and Analysis
Sequencing was conducted on both strands by the dideoxyl-chain termination method with a Perkin-Elmer Biosystems 377 DNA sequencer (Perkin-Elmer, Waltham, MA, USA). Sequence analysis was performed with the Lasergene DNASTAR software package (SelectScience Ltd., Bath, UK). Sequence alignments were conducted with the ClustalW program (www.ebi.ac.uk/clustalw) and the PAM 250 matrix.
The sequence of ISCR2, together with the adjacent sul2 region and the novel ISCR9 and ISCR10, has been deposited in GenBank. The genetic locus ISCR2-glmM/sul2 from isolates 5232, 4647, 3800, and 2107 has been attributed the accession nos. AM182031, 182030, 182029, and 181666, respectively. ISCR9 and ISCR10 have been given the numbers AM182033 and AM182032, respectively.

Detection and Determination of Class 1 Integrons
Of the 25 TMP/SMX-resistant S. maltophilia isolates that we analyzed, 17 possessed the sul1 gene as part of the 3′ end of a class 1 integron. None of the TMP/SMX-susceptible S. maltophilia isolates yielded positive sul1 PCR products. PFGE analysis (data not shown) showed that only 2 isolates (9189 and 12221 from Chile) are clonally related (online Appendix Table). To our knowledge, this is the first report of sul1-positive S. maltophilia isolates from North America and Europe. The sul1-positive isolates are widespread, being from Europe, North America, and South America. Most (5) were isolated from Brazil. The integrons associated with the sul1 gene vary in size; however, when 2 strains were isolated from the same country (e.g., 3438 and 3444, 9189 and 12221, and 98 and 14469), they possessed integrons of the same size, despite not being clonally related (Online Appendix Table). Seven of these integrons were randomly selected to examine their gene cassettes. The genetic context of the class 1 integrons and procured gene cassettes are shown in Figure 1. Strains 1893 (Germany) and 9431 (Brazil) possessed only the int and sul/qac genes. The class 1 integrons from strains 4891 (USA), 9189 (Chile), and 12221 (Chile) contained an embedded aacA4 gene cassette. The 2 Mexican strains (3438 and 3444) contained 2 aminoglycoside-modifying genes (aacA7 and aadA5) and an unknown ORF (Figure 1) yet were clonally unrelated, as judged by PFGE profiling. None of the integrons were the same as those characterized from strains isolated from Argentina (10).

Detection and Location of sul2 Genes
All 55 isolates (both TMP/SMX resistant and sensitive) were screened for sul2 genes with the primers listed in the Online Appendix Table. Nine of the isolates gave PCR products for sul2. None of the TMP/SMX-susceptible S. maltophilia isolates displayed positive sul2 PCR products. Sequence analysis showed 100% identity with previous sul2 sequences.
Given that sul2 is normally located on medium-tolarge sized plasmids, plasmids were isolated and characterized for sul2 carriage. Plasmid DNA was prepared from each isolate and used as a template for PCRs by using the sul2 primer detection set. In every case, a product of the size expected of sul2 sequence amplification was obtained. The purity of each plasmid preparation was evaluated by attempted PCR amplification of the host cell chromosomal gyrA gene. In no case was an amplification product obtained when plasmid DNA was used as template; in contrast, a gyrA amplification product of the correct size was obtained from genomic DNA. These data were later confirmed by Southern hybridization that used the labeled sul2 gene as a probe (data not shown). Unsurprisingly, in most cases sul2 was found on a large plasmid of ≈120 kb; however, in 2 of 9 sul2-positive isolates, sul2 gene was chromosomally encoded.

Detection of ISCR Elements in TMP/SMX-sensitive and -resistant Strains
The sul2 gene and dhfr genes are often found on plasmids and in close association with class 1 integrons or ISCR mobile genetic elements (10,15,18,19). Accordingly, we investigated the 55 S. maltophilia isolates for ISCR elements. Seven of the 25 TMP/SMX-resistant isolates yielded PCR products of the expected size (≈700 bp) when the ISCR specific primers CRF/CRFF-r were used, and 6 of 23 TMP/SMX-sensitive S. maltophilia isolates also yielded the correct-sized amplification products.
To determine whether the locations of the ISCR sequences in the S. maltophilia isolates are chromosomal or plasmid mediated, plasmid DNA was prepared from each isolate and used as a template for ISCR-PCR and Southern hybridization analysis in a similar manner as described for sul2. In every case, a product of the size expected of ISCR sequence amplification was obtained. Hence, in those isolates that possess an ISCR element, the element is located on a plasmid (data not shown). The PCR ISCR amplification products were recovered, purified, and ligated into the cloning vector, PCR-Topo-2.1 (Invitrogen) and recombinant plasmids were recovered by transformation of Escherichia coli DH5α. One clone from each transformation was chosen for further study. Sequence analysis showed that 5/7 amplicons obtained from TMP/SMX-resistant S. maltophilia isolates were identical to the equivalent sequence of ISCR2; the other 2 amplicons were identical to that of ISCR3 (Online Appendix Table). ISCR2 sequences were identified in isolates originating from North and South America, as well as from Europe. In contrast, the ISCR3 sequence was identified only in isolates that originated from Spain.
The ISCR-like elements carried by the sensitive isolates, while clearly related to ISCR1-5, differed markedly from known ISCR sequences (15). Two variants were found, which we have designated ISCR9 and ISCR10. The putative amino acid sequences of ISCR9 and ISCR10 are ≈95% identical to each other and display 30%, 48%, and 74% identity to ISCR2, ISCR,3 and ISCR5, respectively (Online Appendix Figure, available from http://www.cdc. gov/EID/13/4/content/559-appG.htm). These novel ISCRs are harbored in isolates from several different regions, including South American countries, the United States, and Turkey (Table).

Identification of Resistance Genes and Sequences Adjacent to ISCR Elements
ISCR2 is often associated with various antimicrobial resistance genes, not least, genes mediating TMP/SMX resistance ( Figure 2) (15). These and other genes normally associated with ISCR2 were therefore analyzed; these included dhfrA10, dhfrA9, dhfrA20, floR, tetR, strA, sul2, and glmM encoding a truncated phosphoglucosamine mutase. Pairs of oligonucleotides were used (Table) to genetically characterize all those S. maltophilia isolates that possessed an ISCR element.
The floR gene was detected in isolates 2139 and 2170 (which also contains ISCR2) from Turkey and the United States, respectively, and in isolates 12044 and 12049 (which also contains ISCR3) from Spain. A truncated glmM allele (∆glmM) was detected in all ISCR2-containing isolates, and sul2 was found in all ISCR2-and ISCR3containing isolates. The dhfr, tetR, and strA genes were not detected.
Linkage of the ISCR element to ∆glmM, sul2, or floR was then investigated by PCR analysis, i.e., the oligonucleotide pair CRFF/sul2F is expected to generate a product if the ISCR sequence is close to sul2 and downstream of it ( Figure 2). Using this strategy, we found that ISCR2 was linked to ∆glmM and sul2 in all isolates that possess ISCR2. The floR gene was also found to be linked to ISCR2, on the opposite side from ∆glmM and sul2, in isolates 2139 and 2170 ( Figure 2). Linkage of ISCR3 to either sul2 or floR was not demonstrated.

Discussion
We report sul2 genes being present in S. maltophilia and contributing to TMP/SMX resistance. In most cases, sul2 was carried on large plasmids (≈120 kb), but as judged by Southern hybridization data, a few appear to be chromosomally encoded. This study also supports the findings of Barbolla et al. that sul1 present in S. maltophilia is associated with class 1 integrons (10). Herein, we have characterized S. maltophilia sul1 genes from North America, South America, Spain, Turkey, Italy, and Germany, and observed that all of them were associated with class 1 integrons.
Most studies of the location and dissemination of sul2 genes have concentrated on Enterobacteriaceae, such as E.  (23). However, none of the studies demonstrated the genetic origin of the sul2.
In addition to sul genes associated with plasmids and class 1 integrons, we investigated whether the S. maltophilia isolates possessed ISCR elements and whether these could be linked to dhfr or sul genes, as has been shown (18). Of the 25 TMP/SMX-resistant isolates, 6 harbored sul2 linked to ISCR2. However, we could not detect any sul3 genes. In the isolates with ISCR2, the element was directly linked to a deleted version of a phosphoglucosamine mutase gene, ∆glmM, as has been reported on other occasions (Figure 2). This arrangement is identical to those of 5 other sequences in the EMBL database, in E. coli isolated from cattle in France and Germany (24), in the plasmid pRVS1 isolated from a strain of Vibrio salmonicida from Norway, in a plasmid from a strain of S. enterica isolated in Japan, and on the chromosome of Shigella flexneri isolated in the United States (18,24). In all cases, ∆glmM and sul2 are linked to the end of ISCR2 that accommodates the IS91 oriIS equivalent (Figure 2). The dual arrangement of ∆glmM and sul2 is also found in plasmids of marine psychotrophic bacteria isolated in Norway (GenBank accession no. AJ306553/4), but in these cases the ISCR2 element appears not to be present.
Two of the isolates harbored a copy of the floR gene immediately upstream of a copy of ISCR2 (Figure 2), an arrangement identical to that reported on plasmids found in isolates of E. coli from cattle in France and Germany (24). The S. maltophilia isolates investigated in this study came from Turkey and the United States. Two isolates from Spain also carry the floR gene but not ISCR2. Instead, the isolates possess copies of ISCR3, which do not appear to be linked to floR. The finding of florfenicol-resistant traits on plasmids in different bacterial species from different countries highlights the wide geographic spread of this resistance mechanism. The location of floR next to ISCR2 is such that it is possible, if not probable, that the resistance gene can be cotransposed with the ISCR element.
The findings within this study are important for several reasons. First, this is, to our knowledge, the first report of ISCR elements being found in S. maltophilia isolates. In 6 cases, these were linked to sul2 genes responsible for the TMP/SMX-resistant phenotype. Moreover, these isolates were unrelated strains found in different countries. Second, since TMP/SMX is the mainstay therapy for S. maltophilia infections, the mobilization of sul genes by means of class 1 integrons and ISCR elements is likely to increase with TMP/SMX consumption. Third, most sul2 genes in this study have been found on plasmids, and sul2-containing plasmids can potentially confer an increase in bacterial "fitness" (25). As yet, such phenomena have only been explored in Enterobacteriaceae, and it has yet to be established whether sul2-carrying plasmids have such an additive effect in S. maltophilia or for that matter, other nonfermenting gram-negative bacilli.
These data suggest that microbiology laboratories need to carefully monitor S. maltophilia TMP/SMX resistance, which has the potential to increase by means of mobile elements. We also advocate the continued international surveillance of antimicrobial drug resistance that may act as early warning systems for this kind of resistance. Furthermore, yearly monitoring with molecular probes is advisable.