Plasmid profiling, PCR replicon typing and plasmid size
The plasmid profiles of the 68 ST213 isolates showed that these strains carried a large number of plasmids (one to seven) in a wide range of sizes (3 to 160 kb) (Figure 1).
PCR replicon typing was performed for incompatibility groups that had been reported to be associated with either pSTV or blaCMY-2, such as IncFII, IncFIB, IncA/C, IncHI2 and IncI1 [14, 15, 21, 22]. All 36 isolates that carried blaCMY-2 were positive for the IncA/C group and negative for the other Inc groups. Unexpectedly, among the 32 ST213 isolates lacking blaCMY-2, 23 were positive for the IncA/C group. Additionally, the IncHI2 and IncI1 groups were detected in three and two isolates, respectively. Thirteen blaCMY-2-negative and IncA/C-positive isolates were selected to represent different sources, states and years of isolation for further analysis, and compared them with the blaCMY-2-positive isolates (hereafter referred to as CMY- and CMY+, respectively).
Alkaline lysis profiles and PFGE S1-digestion gels of plasmids from strains in our collection were hybridized with blaCMY-2 and repA/C probes; all of the CMY+ isolates yielded signals in the same plasmids, confirming that blaCMY-2 is carried in large IncA/C plasmids (100 to 160 kb). In contrast, only the repA/C probe hybridized in the CMY- isolates, again targeting large plasmids (100 to 160 kb) (Figure 2). Consistent with their low copy number [9, 12, 15], the IncA/C plasmids yielded faint bands in the ethidium bromide-stained gels, especially those larger than 100 kb (Figure 1), but they were unambiguously detected in the hybridization experiments.
Characterization of IncA/C plasmids based on the antibiotic resistance phenotype
To isolate and characterize the IncA/C plasmids present in the Mexican ST213 genotype, E. coli TOP10 or DH5α transformants were obtained using plasmid DNA isolated from 32 CMY+ and 13 CMY- strains. Ceftriaxone was used to select CMY+ plasmids, and chloramphenicol was used to select CMY- plasmids because this resistance has been found to be part of the IncA/C plasmid backbone [5, 6, 8].
All the transformants carrying the IncA/C plasmids also displayed resistance to ampicillin, chloramphenicol, sulphonamides, streptomycin and tetracycline. Resistance to gentamicin was conferred by most of the CMY+ plasmids, and trimethoprim-sulfamethoxazol resistance was mostly detected in the plasmids containing the IP-1 integron (dfrA12, orfF and aadA2; see below). Resistance to neither kanamycin nor nalidixic acid was transferred (Figure 2). These results indicate that the MDR phenotypes of ST213 strains can be explained largely by the presence of IncA/C plasmids.
PstI restriction fingerprints
The plasmid profiles showed that all of the E. coli transformants carried one large plasmid of between 100 and 160 kb. These transformants were analyzed by PstI restriction fingerprinting [12, 23]. Cluster analysis of the PstI fingerprints showed two main plasmid types (similarity <50%), which we named type I and type II (Figure 2). All of the CMY+ plasmids were contained in type I and were distributed into three clusters (a, c and d). The CMY- plasmids were found in two distinct groups: one in type II and the other in cluster b within type I, suggesting that the CMY- plasmids originated from two divergent IncA/C plasmid types. To put our plasmids into context, IncA/C CMY+ reference plasmids from E. coli AR060302 [6] and Newport SN11 [22] were included. The restriction profiles of these plasmids were related to our ST213 type II plasmids, which in contrast were all CMY-.
We compared the sampling information (see Methods) and our previously generated genomic DNA XbaI macrorestriction patterns [16] with the plasmid PstI restriction patterns. The observed distribution of the plasmids among genomic backgrounds was consistent with a pattern of clonal spread. The most evident association was between XbaI cluster Ib and PstI cluster e; these isolates came from Sonora and were sampled in 2004-2005 (Figure 2).
PCR screening and nucleotide sequence analysis of the plasmids
The E. coli transformants were subjected to PCR screening using primer pairs that detect seven regions (repA/C, floR, CMY region, R-7, R-8, mer and IP-1; Figure 3 and Additional file 1, Table S1) distributed throughout the reported IncA/C plasmids [5–8, 10]. All the plasmids were positive for the repA/C, floR and mer regions (Figure 2); only one plasmid did not contain the mer region (strain YUHS 05-78). The R-7 segment was detected in all the CMY+ plasmids but in none of the CMY- plasmids. We analyzed the CMY region assuming that the right junction would consist of an insertion of dsbC upstream of traC and that the left junction would consist of an insertion of tnpA downstream of traA (PCRs G and A, respectively; Figure 4). However, during the nucleotide sequence analysis, we realized that dsbC and the hypothetical protein 0093 gene are part of the plasmid core of other closely related IncA/C plasmids lacking the CMY island (see below). Thus, PCR D was also used to detect the insertion of the CMY island at the right junction, demonstrating the insertion of blc, sugE and ΔentR upstream of the 0093 gene (Figure 4). To determine if the flanking region of traA is similar in the CMY+ and CMY- plasmids, the left junction was assessed by PCR B (Figure 4). As expected, the CMY- plasmids did not amplify the CMY junctions, whereas most of the CMY+ plasmids amplified the right and left junctions (Figure 2), indicating that with only one exception (strain MIPOLS 03-75), the CMY island is inserted in the same position in these plasmids. The most variable regions of the IncA/C plasmids were the R-8 segment and the IP-1 integron (dfrA12, orfF and aadA2). R-8 was present only in a small fraction of the CMY+ plasmids, including all the plasmids that belong to cluster d. Most (25 out of 35) of the Salmonella strains that were positive for the IP-1 integron transferred this region along with their IncA/C plasmids. The exceptions were six CMY+ plasmids and four CMY- plasmids (Figure 2). The presence of integrons has been reported for other IncA/C plasmids [6, 7, 9].
Ten strains representing different geographic locations, years and sources were chosen and their regions analyzed in the PCR screening were sequenced (Additional file 2, Table S2). The sequences were identical for all the plasmids (both CMY+ and CMY-); only the mer region showed a single nucleotide substitution (Additional file 2, Table S2). It was surprising that even intergenic regions and third codon positions were invariable. BLAST searches showed that our sequences are identical (100% identity) to the IncA/C plasmids pAR060302 (E. coli), peH4H (E. coli), pAM04528 (Newport) and pSN254 (Newport); are closely related (99-98%) to the IncA/C plasmids pIP1202 (Yersinia pestis), pYR1 (Yersinia ruckeri), pP91278 (Photobacterium damselae), pP99-018 (P. damselae) and pMRV150 (Vibrio cholerae); and are related (88-89% identity) to pRA1 (Aeromonas hydrophila) [5–10]. The repA gene displays the repA/C
2
allele described for other IncA/C CMY+ plasmids [19]. Call et al. [6] reported that one of the most variable parts of the IncA/C plasmids is the floR gene; the floR allele found in the ST213 isolates is identical to that of pSN254, which differs from that of pAR060302 by three non-synonymous substitutions. The traC-dsbC junction (PCR G) of the CMY island (Figure 4) was found in all the plasmids mentioned above and in the recently described integrating conjugative element ICEPmiJpn1 of Proteus mirabilis [GenBank:AB525688]. The finding that traC-dsbC is present in pIP1202, pYR1, pP91278, pP99-018, pMRV150 and pRA1, which lack the CMY island, revealed that this gene cluster is part of a conserved core region of these closely related IncA/C plasmids. However, this region did not match with any other plasmids in the database, and it was not amplified in the CMY- plasmids of ST213 (Figure 2). Therefore, to assess the insertion of the right CMY junction, a second marker was used: PCR D spanning from sugE to the hypothetical protein 0093 (Figure 4). The complete traVA-tnpA right junction (PCR A and B) of the CMY island was identical to that of the E. coli and Newport plasmids, but only traVA (PCR B) was present in the other CMY- IncA/C reference plasmids. This result indicates that this marker is the left CMY island junction. Interestingly, the ST213 CMY- plasmids did not amplify the traVA region, indicating that the region around the CMY island is not present in these plasmids. R-7 and R-8 were found to be present in all the IncA/C reference plasmids, with the only exception being peH4H, which lacks R-7. The mer region was detected only in the E. coli pAR060302 and Newport plasmids; however, it was found to be related to other mer operons present in several plasmids such as pRMH760 (Klebsiella pneumoniae).
Characterization of the CMY region
When we started this study, the only completely sequenced plasmid carrying blaCMY-2 was that of the Newport strain [GenBank:NC_009140] [8]. PCR mapping experiments were performed to compare the CMY region of our Typhimurium isolates with that of Newport pSN254 (Figure 4 and Additional file 1, Table S1). To determine if the blaCMY-2 gene is present as an inverted repeat element as in pSN245, we performed PCR H and I, which we expected to produce bands of around 3.2 and 2.3 kb, respectively, based on the in silico prediction. The Newport strain SN11 was used as a positive control for these amplifications. No bands were obtained with our Typhimurium isolates, consistent with the notion that our isolates possess only a single blaCMY-2 gene. We designed a set of primer pairs to amplify overlapping fragments covering the complete CMY region and to obtain the nucleotide sequence of one of our isolates, YUHS 07-18, which is the most recent strain in our collection and which displays XbaI and PstI fingerprints prevalent in the ST213 population. The 12,621-bp sequence of YUHS07-18 [GenBank:HQ203988] spanning from traV to traC was found to be identical to the corresponding region of pSN254 [GenBank:NC_009140] [8], and it has the same CMY region configuration as the E. coli plasmid pAR060302 [GenBank:FJ621588] [6].
Southern blot hybridization of PstI plasmid restriction fingerprints
Representative examples of Southern hybridizations of the PstI fingerprints are shown in Figure 5. Hybridization with the blacmy-2 probe demonstrated that all CMY+ plasmids were of Giles type A [20], displaying two hybridization bands of about 12 and 0.6 kb. This type has been associated with plasmids that carry one copy of the CMY island, such as pAR060302 [6]. The repA/C probe hybridized with the larger band in all the strains, which should be about 55 kb according to an in silico PstI restriction of the complete sequence of pAR060302. This band also hybridized with the mer probe for most of the plasmids, in agreement with the in silico prediction. However, some polymorphisms were detected using the mer probe (Figure 5). The floR probe produced a single band of 8 kb, with one exception (Figure 5; MIPOLS 03-75, 7 kb). Finally, hybridizations were performed using the first two genes of IP-1 (dfr12 and orfF); the aadA region was not included in the probe because this gene has been associated with other integrons often present in IncA/C plasmids, such as that of transposon Tn21 [7–9]. Most of the strains produced a hybridization band of 6 kb, but there were polymorphisms (Figure 5).
Conjugative transfer of IncA/C plasmids
Ten CMY+ and seven CMY- ST213 isolates were evaluated for conjugative transfer of their A/C plasmids to E. coli DH5α. Transconjugants were only obtained for the CMY+ strain YUHS 05-78 and at a very low frequency (10-7 to 10-9), but they were positive for all nine PCR markers of the donor plasmid, which lacked the mer region (Figure 2). However, no transconjugants were observed when an E. coli strain carrying the YUHS 05-78 CMY+ plasmid was used as the donor. The highest efficiencies were obtained with a donor:recipient ratio of 1:10 and an incubation for 18 hr on a solid medium (see Methods). In our hands, conjugation efficiencies for AR060302 and SN11 strains were in the order of 10-5 and 10-6, respectively. Nevertheless, these frequencies were lower than those reported for these plasmids (i.e. 10-3) [6, 22].