The plasmid pSAL-1 DNA isolated from the Escherichia coli HB101 transconjugant was partially digested with Sau3A and ligated into the BamHI site of pACYC184. Recombinant plasmid pSAL-2_ind was introduced into E. coli JM101, and the transformants had the same AmpC susceptibility pattern as the donor strain, Salmonella serovar Enteritidis, and its transconjugant E. coli HB101(pSAL-1) (5). The whole sequence (2,126 bp) of ampC and ampR genes and their intercistronic region was reported and analyzed previously (5). In this work, the recombinant plasmid pSAL-2_ind (4.2 kb) was entirely sequenced and analyzed with the BLASTN program at the National Center for Biotechnology Information (1) (Fig. 1). A segment of 1.4 kb located 319 bp downstream from ampR showed the 3′ end of ORF4 (deletion of the first 72 bp of the gene qacEΔ1), the whole gene sulI, and the 5′ end of ORF5, with 100% identity; these three open reading frames are typical of the 3′ conserved segment (3′-CS) of a class 1 integron (9). Furthermore, a segment of 0.2 kb 170 bp downstream from ampC is 100% identical to a region characteristic of integrons In6 and In7 (17, 18). Both these integrons are unusual in that they have a partial duplication of the 3′-CS, and they have a common segment of 2.1 kb, located between the two sulI genes, which includes ORF341, of unknown function. Different resistance genes (cat in In6 and dfrA10 in In7) are present downstream from ORF341 (17). E. coli JM101(pSAL-2_ind) was susceptible to sulfonamides despite the presence of a sulI gene. As on pSAL-2_ind, the sulI gene of the second copy of 3′-CS in In6 and In7 is not expressed, probably due to the absence of the promoter (in the 5′ end of the 3′-CS). The high degree of identity between regions of pSAL-2_ind and the integrons In6 and In7 suggests that the ampC gene is part of a complex integron structure.

FIG. 1 Structure of integron pSAL-1 (9.5 kb) obtained by genetic analysis of the recombinant plasmid pSAL-2ind and PCR mapping of the upstream region. (A) Sequence reported here (8,558 bp). (B) Recombinant plasmid pSAL-2ind (4.2 kb). (C) Sequence (2,126 bp) (more ...)

PCR primers corresponding to sequences of In7 were used to map the unknown region of pSAL-1 (Table 1). Genomic DNA was extracted from E. coli HB101(pSAL-1) by the method described by Grimont and Grimont (7). PCRs were carried out in 100-μl volumes containing 1× PCR buffer (ATGC Biotechnologie, Noisy-le-Grand, France); 200 mM (each) dATP, dCTP, dGTP, and dTTP (Pharmacia Biotech, Uppsala, Sweden); 0.5 mM (each) primer (Oligo-express, Paris, France) (sequences in Table 1); and 5 mM template DNA and 2 U of Taq DNA polymerase (ATGC Biotechnologie). To amplify the DNA in the thermal cycler (Perkin-Elmer Cetus, St. Quentin-en-Yvelines, France), we used a three-step profile for 40 cycles: denaturation for 30 s at 94°C, annealing for 30 s at variable temperature according to primers, and extension for 30 s at 72°C. The denaturation for the first cycle and the extension for the last cycle lasted 3 min. The temperatures used for annealing were as follows: fragment A, 66°C; fragment B, 62°C; fragment C, 58°C; fragment F, 60°C; fragment D, 58°C; fragment G, 62°C; fragment E, 58°C.

TABLE 1 Sequences and positions of the oligonucleotides used for PCR analysis of integronpSAL-1

Finally, seven overlapping PCR fragments allowed us to map the unknown region of the 5.3 kb in pSAL-1 downstream from the sequences corresponding to pSAL-2_ind. For each PCR fragment (except A), DNA sequence was determined by the procedure of Sanger et al. (16) with the PCR primers, fluorescent dye-labeled dideoxynucleotides, thermal cycling with Taq polymerase, and the ABI 373A DNA sequencer (Applied Biosystems). The BLASTN program at the National Center for Biotechnology Information was used for the alignment of the DNA sequences (1). The PCR products from pSAL-1 and the clone pSAL-2_ind overlap by 9.5 kb (Fig. 1).

With NciI, AluI, and Sau3A, results of restriction patterns of the PCR A product were consistent with those of the integrase gene of sulI-type integrons (data not shown). Only the first 110 bp of the int gene (11%) was sequenced (PCR B) because of the high degree of conservation of this gene in sulI-type integrons (8). Other genes and open reading frames were identified: aadA2, ORF4, sulI, ORF341, ampC, and ampR. These findings are consistent with the susceptibility pattern of E. coli HB101(pSAL-1): resistance to streptomycin, spectinomycin, and sulfonamides. A 2.1-kb region including ORF341 between the first sulI gene and the ampC gene displays 100% identity with the region common to In6 and In7 (17, 18).

The plasmid pSAL-1 includes several characteristic elements of integrons. First, an int gene encodes a site-specific recombinase (integrase) (9). Second, the aadA2 gene is suitably oriented to be a cassette and the 59-bp element is present, downstream from this gene (15). Third, ORF4 (qacEΔ1 gene) and a sulI gene constitute a partial 3′-CS of the sulI-type integron. The plasmid pSAL-1 thus has a genetic organization similar to those of pSa and pDGO100 (In6 and In7) (17): the presence of a 2.1-kb common region including ORF341 is associated with a partial duplication of the 3′-CS. These observations indicate that the ampC and ampR genes on pSAL-1 were inserted into a complex sulI-type integron. Three distinct integrons, In6, In7, and this new one on pSAL-1, have a similar genetic organization, suggesting that they are descended from a common ancestral integron. The lengths of the deletions at the 5′ end of the second 3′-CS (CS2) are 77 bp in In7, 248 bp in In6, and 180 bp in the new integron (Fig. 2).

FIG. 2 Deletions at the 5′ end of the 3′-CS in In7, In6, and pSAL-1. Base 1 is the first base of the complete 3′-CS. Asterisks indicate sequence identity.

Nevertheless, there is a difference between pSAL-1 and integrons In6 and In7: the origin of the antibiotic resistance genes downstream from ORF341. The genes dfrA10 (resistance to trimethoprim) and cat (resistance to chloramphenicol) have the same orientation as the gene cassette(s) upstream from the first 3′-CS. It is possible that they were cassettes but have lost their mobility due to the deletion of the 59-bp element and the first bases of the adjacent 3′-CS, as suggested by Stokes et al. (17).

The genes ampC and ampR in opposite orientations are not located between a 5′-CS and a 3′-CS. No 59-bp element is present adjacent to ampC or ampR. The expression of the ampC gene was necessarily different from that of cassette genes (9). To date, only three families of β-lactamases (classification of Ambler [2]) are known to be encoded by cassettes: class A (PSE or CARB) (12, 15), class B metallo-β-lactamases (bla_IMP gene) (3), and class D (OXA-type) (13). This is the first time that a gene encoding a class C β-lactamase has been found inserted in an integron. The ampC gene is transcribed under the control of its regulatory gene ampR binding to DNA in the ampC-ampR intercistronic region, which includes overlapping promoters (5, 11).

It would be interesting to compare the genetic organization of DHA-1 with those of other plasmid-encoded cephalosporinases. Few data are available. The ampC genes and, in some cases, the neighboring regions have been sequenced. The ampC genes of the plasmid-encoded β-lactamases CMY-1 and MOX-1 (6, 10) have upstream fragments of 168 and 488 bp, respectively, which are identical to the region common to In6, In7, and pSAL-1 (Fig. 3). Both the ampC gene coding for CMY-1 and that coding for MOX-1 have the same orientation, like cassette genes, but no 59-bp element has been found nearby. These enzymes are not inducible, and they are not associated with an ampR gene.

FIG. 3 Comparison of the regions between the common regions of In7, In6, pSAL-1 (DHA-1), pMVP-1 (CMY-1), and pRMOX-1 (MOX-1) and the second 3′-CS of In7, In6, and pSAL-1 (DHA-1). Asterisks indicate sequence identity. Dark gray shading, 100% identity; (more ...)

The bla_DHA-1 gene and its regulator ampR gene were probably mobilized from the M. morganii chromosome to the conjugative plasmid of Salmonella serovar Enteritidis after recombination catalyzed by genes of the integron. The phylogenic origin of MOX-1 and CMY-1 is unknown. It is possible that ampC genes coding for these β-lactamases were inserted into a sulI-type complex integron. We suggest that a preferential recombination site is present at the end of the common region and that this leads to the insertion of a resistance gene into the complex structure of the integron. Valentine and colleagues (18) suggested that ORF341 encodes a recombinase recognizing this specific recombination site.

Nucleotide sequence accession number. The sequence has been deposited in the EMBL database under accession no. AJ237702.

This study was supported in part (C.V.) by a grant from the Fondation pour la Recherche Médicale.