INTRODUCTION

Top Abstract Introduction Materials & Methods Results & Discussion References

Penicilloyl serine transferases, routinely called -lactamases, cleave the cyclic amide bond of -lactam antibiotics via the formation of a serine ester-linked penicilloyl enzyme giving a product devoid of antibacterial activity (46). A close inspection of databases indicated that in the last 3 years, a collection of at least 150 DNA sequences from plasmid-mediated and chromosomal bla genes has been acquired. Analysis of deduced peptides confirmed that most have conserved motifs typical of serine active-site enzymes that are divided into three major classes (classes A, C, and D) on the basis of a level of amino acid sequence identity of more than 20% between members in each class (10).

In 1969, a -lactamase was found in Pseudomonas aeruginosa Dalgleish, it was noticed to be "markedly active against carbenicillin," and the enzyme was named PSE-4 (13, 32). As other -lactamase enzymes were found, it was noticed that some -lactamases have better activities than others against carbenicillin. All -lactamases except class C enzymes hydrolyze carbenicillin at very different levels; class C enzymes hydrolyse it poorly. Genes encoding enzymes similar to PSE-4 were subsequently discovered in other bacterial species and are now known to be part of multidrug resistance transposons (24). In addition to the four original -lactamases called PSE-1, PSE-2, PSE-3, and PSE-4, a plethora of plasmid-mediated enzymes capable of hydrolyzing carbenicillin at a high rate, such as LCR-1 (10), AER-1 (16), CARB-3 (22), NPS-1 (26), CARB-5 (35), and CARB-4 (36), were identified; but these enzymes have subtle differences in their biochemical properties and in their substrate profiles (7). The amino acid sequences of PSE-1 (17), PSE-2 (18), PSE-3 (8), PSE-4 (4), and CARB-3 (23) have been compared to those of other class A and class D enzymes, and it has been confirmed that PSE-2 (OXA-10) is a class D enzyme (10).

The relationship of CARB-4 to other plasmid-mediated -lactamases has been tested by determining the neutralization of benzylpenicillin-hydrolyzing activity with antisera prepared against purified the TEM-1, OXA-4, and CARB-3 -lactamases (36). Antisera prepared with CARB-3 antigen inactivated the CARB-4 -lactamase as well as the PSE-1, PSE-4, and CARB-3 enzymes (22, 36). The bla_CARB-3 and bla_CARB-4 genes are localized within transposons Tn1413 (7 kb) and Tn1408 (25 kb), respectively; these mobile elements were from plasmids isolated from bacterial strains of distinct origins (24, 27, 47).

Unusual -lactamases such as a metalloenzyme have been reported in Aeromonas hydrophila, a water-borne, gram-negative rod known to be highly resistant to -lactam antibiotics, including carbenicillin (42). A carbenicillin-hydrolyzing -lactamase has been discovered in an isolate of A. hydrophila from blood (16). The substrate profile of the AER-1 enzyme resembled those of plasmid-mediated carbenicillin-hydrolyzing enzymes, but it had a different isoelectric point (pI 5.9) and molecular mass (29 kDa); these values are reminiscent of those for BRO-1 (pI 5.45), PSE-1 (pI 5.7), CARB-3 (pI 5.75), and CARB-5 (pI 5.35). The gene coding for AER-1 is part of the 7711 unit which is IncP mobilizable but RecA dependent and which inserts only between purC and guaB at a specific site in the Escherichia coli chromosome (16). The 7711 unit cotransfers resistance to the antibiotics chloramphenicol, streptomycin, and sulfonamide; the transfer of multidrug resistance and insertion at a unique site are properties analogous to those of Tn7.

In the study described in this report, we have focused on a carbenicillin-hydrolyzing enzyme identified from a clinical isolate, P. aeruginosa p83372, containing the pUD12 plasmid and producing CARB-4. This enzyme has an acidic isoelectric point (pI 4.3) and hydrolyzes carbenicillin very efficiently (36). We also present the nucleotide sequences of bla_AER-1 and bla_CARB-4, including flanking sequences containing integrons that explain their distribution and presence in different mobile genetic elements (15). We compared the deduced AER-1 and CARB-4 polypeptides with those of other group 2c enzymes (7) via a multiple alignment.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results & Discussion References

Bacterial strains, plasmids, and phages. The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were grown on tryptic soy agar plates (Difco Laboratories, Detroit, Mich.) containing appropriate antibiotics (ampicillin, 100 µg/ml; kanamycin, 50 µg/ml; chloramphenicol, 30 µg/ml; tetracycline, 10 µg/ml). The cloning vector used initially was pACYC184 (chloramphenicol resistant [Cm^r] and tetracycline resistant [Tet^r]) (9, 37). E. coli HB101 (5) transformants were selected on chloramphenicol-containing plates and were susceptible to tetracycline. E. coli HB101 was the recipient of pMON1028, pMON510, and recombinant plasmid derivatives coding for CARB-4 and AER-1 -lactamases. The selected transformants were confirmed to produce the prototype enzymes by isoelectric focusing (23). For single-stranded DNA production and sequencing, the bla_CARB-4 and bla_AER-1 genes were subcloned into phages M13mp18 and M13mp19 (29). E. coli JM101 was used as a recipient and was kept on minimal medium without proline (44). Bacteriophages M13mp18 and M13mp19, the replicative forms of phagemids pBGS18⁺ and pBGS19⁺, and single-stranded DNA were prepared by standard procedures (29, 30, 39, 40, 43).

Enzymes and chemicals. All chemicals were of the highest grade commercially available. Restriction enzymes were used with the manufacturer's recommendations and were from Pharmacia LKB, Baie d'Urfé, Montréal, Québec, Canada; New England Biolabs, Mississaugua, Ontario, Canada; and Gibco BRL, Mississaugua, Ontario, Canada. The radioisotope [³⁵S]dATP was from Amersham.

Preparation of DNA and related techniques. Large plasmid DNA preparations were prepared by the cleared lysate method, with modifications for cell lysis (1 mg of lysozyme per ml, 7.5 mM disodium EDTA, 1% sodium dodecyl sulfate [pH 8.0]), and were purified by cesium chloride-ethidium bromide gradient ultracentrifugation (14, 39). Plasmid minipreparations for double-stranded DNA sequencing were prepared with Qiagen plasmid mini and midi kits (Chatsworth, Calif.) according to the manufacturer's suggestions. Restriction enzymes were digested as recommended by the manufacturer; ligation, transformation, and selection of recombinant DNA molecules were done by standard procedures (39, 43).

Physical mapping and subcloning. We previously reported the cloning of a 4.3-kb BamHI DNA fragment containing bla_CARB-4 (CARB-4) and a 1.5-kb Sau3AI fragment coding for bla_AER-1 (AER-1) isolated from the multidrug resistance transposon Tn1413 (Ap^r, Gm^r, Km^r, Su^r, Tm^r) and from a genomic library of E. coli J53-1 7711, respectively (25). E. coli HB101 transformants were selected for ampicillin resistance containing recombinant plasmids pMON1025 and pMON511 and were confirmed to produce CARB-4 (pI 4.3) and AER-1 (pI 5.9) -lactamases by isoelectric focusing (data not shown). From pMON1025, we constructed plasmids pMON1026, pMON1027, pMON1028, and pMON1035 (Table 1). Phenotypic analysis of E. coli HB101 transformants with ampicillin, kanamycin, and sulfonamide and correlation with the plasmids' physical DNA maps constructed with selected restriction endonucleases indiated that bla_CARB-4 is within the 1.9-kb BamHI plus HindIII fragment. The physical map of pMON511 is shown in Fig. 1C. The bla_AER-1 structural gene was centrally mapped by deletion of an AvaI fragment and was associated with ampicillin susceptibility in the recipient E. coli strain.

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Physical maps of recombinant plasmids pMON1035 (A) and pMON511 (C). Only the DNA inserts are depicted as open-boxed lines that include the restriction sites used for subcloning into M13mp18 and M13mp19 sequencing vectors. The sequencing strategies used for blaCARB-4 and flanking DNA (B) and blaAER-1 (D) are indicated below each map. The extent and direction of each sequencing reaction are indicated by arrows, and sequencing primers, synthesized as 21-mers from the last nucleotides read, were used.

DNA sequencing. The 650-bp XbaI-XbaI DNA fragment (in which the first XbaI is bla_CARB-4 and the second site is in the pACYC184 moiety at position 1424 [37]), the 548-bp HindIII-HindIII fragment, and the 1.1-kb BglII-XhoI fragments isolated from pMON1025 and pMON1039 (Table 1) were cloned in both orientations into the sequencing phage vectors M13mp18 and M13mp19 and into the phagemids pBGS18⁺ and pBGS19⁺ (30, 44). For bla_AER-1, a 2.2-kb HindIII-SalI fragment from pMON511 (Table 1) was cloned into M13mp8 and M13mp9 (29). The 2.2-kb SalI-HindIII fragment from pMON511 was subcloned into phagemid pBGS18⁺ for sequencing (Fig. 1C and Table 1). Complementary DNA strands were completely sequenced by the primer walking sequencing strategy outlined in Fig. 1B and D. Nucleotide sequencing of single-stranded DNA was done by the dideoxynucleotide T7 polymerase chain termination method (40) with the Pharmacia LKB sequencing kit and [-³⁵S]dCTP (Amersham).

Informatics and computer software analysis. DNA sequence analyses were done with software from ABI (Factura, Gene Navigator, and AutoAssembler) and the Genetics Computer Group (version 9.0) of the University of Wisconsin (11). Comparisons of the sequences with the sequences in the GenBank, European Molecular Biology Laboratory, and National Biomedical Research Foundation databases were done with the Genetics Computer Group software package adapted to a UNIX-based system by using the FASTA, TFASTA, and BLAST programs. Identification of signal peptides and prediction of protein localization sites were done by two methods (31, 33, 34). Molecular masses and pI values were predicted from the amino acid sequences as described previously (3). The Terminator and Stemloop programs were used to identify terminator sequences by using a primary structure threshold cutoff P value of 3.5 (6). Multiple alignments were done with CLUSTALW in the MOLPHY (version 2.2) package of software. Phylogenies were obtained with the PHYLIP software package (version 3.57c), obtained from J. Felsenstein, University of Washington (12).

Nucleotide sequence accession numbers. The sequences reported here have been assigned GenBank accession nos. U14748 and U14749.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials & Methods Results & Discussion References

Sequence of bla_CARB-4 encoding CARB-4. Comparisons of the physical maps of pMON1025 and Tn1413 published previously (24, 28) and the map of pMON1035 described in Fig. 1A showed minor differences in the positions of restriction endonuclease sites (as a BamHI site) that were attributed to the limits of DNA mapping done previously (24, 28). The nucleotide sequences of the physical DNA maps for BglII, ClaI, HindIII, PvuII, SstI, and XbaI done for pMON1035 shown in Fig. 1A were also identified (Fig. 2). We refer to the gene as bla_CARB-4 encoding CARB-4, and complementary DNA strands were completely sequenced by using subclones and the primer walking sequencing strategy illustrated in Fig. 1B and D. The nucleotide sequence of bla_CARB-4 (Fig. 2A) is 1,878 nucleotides (nts) long. Searches in databases identified an integron that comprised open reading frame (ORF) ORF4 (from nts 1 to 211) and that was in frame with part of aadB and a 59-bp element (from nts 212 to 271) (15, 41). We identified the bla_CARB-4 gene cassette encoding CARB-4 (from nts 572 to 1679) and part of ORF5 encoding sul-1 (from nts 1680 to 1878). The bla_CARB-4 gene cassette had significant DNA homology with bla_PSE-1 (86.5%) and bla_N-29 (86.3%), as summarized in Table 2. DNA analysis indicated G+C contents of 69.2% for ORF4 and 58.8% for aadB. Curiously, the bla_CARB-4 gene cassette encoding CARB-4 had a G+C content of 39.1% and ORF5 had a G+C content of 38.7%. Similar G+C contents for part of the integron and bla_CARB-4 would indicate a role of the integron in the assembly of the 1.87-kb locus. In addition, the typical G+C content of genes from P. aeruginosa is 67%.

View larger version (58K): [in this window] [in a new window]   View larger version (59K): [in this window] [in a new window]   FIG. 2.   Nucleotide sequences of blaCARB-4 (A) and blaAER-1 (B) genes and their flanking regions. The deduced amino acid sequence is indicated by the one-letter code beginning under the first nucleotide of each codon. The secretion peptides in CARB-4 and AER-1 as well as conserved amino acids found in PRPs are underlined (1, 19, 20). The stop codon in each sequence is indicated by an asterisk. Ribosome-binding sites (RBS) and 10 and 35 regions are underlined, and potential int recombination sequences (gttrrry) are indicated in boldface in the DNA sequence. Horizontal arrows indicate the direction of transcription; vertical arrows identify the cutting sites of the restriction endonucleases.

View this table: [in this window] [in a new window]   TABLE 2.   DNA homology between -lactamases of group 2c genes

Sequence of bla_AER-1 encoding AER-1. In pMON511 (Fig. 1C), the PvuII restriction site was mapped outside the AvaI fragment, in contrast to earlier reports (25), and the location is now supported by sequencing results. The entire nucleotide sequence obtained from the Sau3A DNA fragment shown in Fig. 2B is 1,727 nts long and encompasses the bla_AER-1 structural gene with flanking sequences. From nts 1 to 83, we noted 100% homology with the 5' region of sulII, as underlined in Fig. 2B (41). This sequence homologous to the upstream region of the dihydropteroate synthase (sulII) gene found in pLS88 and in the broad-host-range plasmid RSF1010 suggested that the bla_AER-1 gene is integrated in a transcriptional region and would have been acquired by the 7711 mobile element (41). We did not find homology to known genes in databases when we used sequences from nts 84 to 324. The bla_AER-1 gene from nts 325 to 1239 in Fig. 2B was localized by translation of the 1,727 nts in all six frames. BLAST searches identified a single ORF of 304 amino acids encoding an AER-1 whose sequence was homologous to those of known -lactamases, primarily PSE-3 found in class A. The sequence 100% homologous to the 5' region of sulII had a G+C content of 48%, the 5' region flanking bla_AER-1 had G+C content of 54%, bla_AER-1 had a G+C content of 54%, and the 3' region flanking bla_AER-1 had a G+C content of 52.7%.

Homology and multiple alignment of CARB enzymes. Since the CARB-4 and AER-1 -lactamases hydrolyze carbenicillin at a high rate, we focused our analysis by comparisons of the CARB-4 and AER-1 enzymes with known class A and class D carbenicillin-hydrolyzing enzymes (10). Comparisons of CARB-4 and AER-1 with class D enzymes having carbenicillin-hydrolyzing activity, such as LCR-1 and PSE-2, gave low levels of identity for CARB-4 (19.3 and 12.3%, respectively) and AER-1 (20.2 and 19.7%, respectively) (data not shown). In contrast, both CARB-4 and AER-1 had consistently higher (more than 30%) pairwise identities with class A enzymes (Table 3). As shown in Table 3, the BRO-1 enzyme was included in group 2c but has an identity of between 31.07 and 33.45% compared with other enzymes of group 2c. To better assess the degrees of relatedness between -lactamases having so-called carbenicillin-hydrolyzing activities, defined as group 2c (7), we aligned CARB-4 and AER-1 with PSE-4, PSE-3, GN79, BRO-1, and TEM-1, as shown in Fig. 3. In the alignment the seven canonical amino acid boxes representing conserved motifs of PRPs are well conserved in CARB-4 as well as in AER-1.

View larger version (57K): [in this window] [in a new window]   FIG. 3.   Alignment of the CARB-4 and AER-1 amino acid sequences with the sequences of class A enzymes with carbenicillin-hydrolyzing activity. PSE-1 is from P. aeruginosa RPL11 (17); PSE-4 is from P. aeruginosa Dalgleish (4); CARB-3 is from P. aeruginosa Cilote (23); N-29 is from P. mirabilis N-29 (47); PSE-3 is from P. aeruginosa ps142 (8; unpublished data); GN79 is from P. mirabilis GN-79 (38); BRO-1 is from Branhamella catarrhalis (unpublished data and GenBank accession no. U49269); TEM-1 is from pBR322 (45). Identical residues in eight proteins are indicated as a consensus sequence. The numbering of the amino acids follows standard nomenclature (1).