Novel Class A β-Lactamase Sed-1 from Citrobacter sedlakii: Genetic Diversity of β-Lactamases within the Citrobacter Genus

Abstract

Citrobacter sedlakii 2596, a clinical strain resistant to aminopenicillins, carboxypenicillins, and early cephalosporins such as cephalothin, but remaining susceptible to acylureidopenicillins, carbapenems, and later cephalosporins such as cefotaxime, was isolated from the bile of a patient treated with β-lactam and quinolone antibiotics. The isolate produced an inducible class A β-lactamase of pI 8.6, named Sed-1, which was purified. Characterized by a molecular mass of 30 kDa, Sed-1 preferentially hydrolyzed benzylpenicillin, cephalothin, and cloxacillin. The corresponding gene, bla_Sed-1, was cloned and sequenced. Its deduced amino acid sequence shared more than 60% identity with the chromosome-encoded β-lactamases from Citrobacter koseri (formerly C. diversus) (84%), Klebsiella oxytoca (74%), Serratia fonticola (67%), and Proteus vulgaris (63%) and 71% identity with the plasmid-mediated enzyme MEN-1. A gene coding for a LysR transcriptional regulator was found upstream from bla_Sed-1. This regulator, named SedR, displayed 90% identity with the AmpR sequence of the chromosomal β-lactamase from C. koseri and 63 and 50% identity with the AmpR sequences of P. vulgaris and Enterobacter cloacae, respectively. By using DNA-DNA hybridization, a bla_Sed-1-like gene was identified in two reference strains, C. sedlakii (CIP-105037) and Citrobacter rodentium (CIP-104675), but not in the 18 strains of C. koseri studied. Two DNA fragments were amplified and sequenced from the reference strains of C. sedlakii CIP-105037 and C. rodentium CIP-104675 using two primers specific for bla_Sed-1. They shared 98 and 80% identity with bla_Sed-1, respectively, confirming the diversity of the chromosomally encoded class A β-lactamases found in Citrobacter.

The genus Citrobacter, which was defined in 1932, initially encompassed seven species including Citrobacter freundii (type species) and Citrobacter koseri (formerly termed Citrobacter diversus or Levinea malonatica) (60). In 1993, Brenner et al. identified eight new DNA hybridization groups genetically distinct from C. freundii and C. koseri (13). These additional genomospecies included C. sedlakii (genomospecies 8) and C. rodentium (genomospecies 9, a bacterial pathogen of rodents).

In the Citrobacter genus, resistance to β-lactam antibiotics is mainly mediated by production of chromosomally encoded β-lactamases. C. freundii produces an inducible Ambler class C β-lactamase (60). In C. koseri (formerly C. diversus), which is naturally resistant to aminopenicillins and carboxypenicillins, resistance to β-lactams is mediated by a chromosome-encoded class A β-lactamase (4). The cloning and sequencing of the corresponding gene, which has been termed bla_CdiA, revealed a high degree of similarity (75%) between CdiA and the class A β-lactamase from Klebsiella oxytoca (33). Previous experiments carried out by DNA amplification with primers specific for bla_CdiA showed that this gene is not ubiquitous in C. koseri (32). Accordingly, different chromosome-encoded β-lactamases with distinct isoelectric points (pIs) varying from 4.8 to 9.5 have been identified in C. koseri, but the corresponding genes have not been characterized at the genetic level (27, 31, 41, 42, 47, 53, 55). The LysR-type transcriptional regulator (LTTR) protein CdiR, divergently transcribed from CdiA, has also been characterized. The AmpR regulator protein most closely related to CdiR is the Proteus vulgaris CumR protein, with an amino acid identity of 66%, whereas only 46% identity was found with the AmpR protein from C. freundii, strengthening the idea of a wide genetic diversity of the genes determining resistance to β-lactam antibiotics in the Citrobacter genus.

C. sedlakii has been rarely described since its first identification in 1993 (1, 2, 13, 23, 43). Strains isolated from human stools, blood, and wounds were studied by Brenner et al. (13), and a report of neonatal meningitis and brain abscess involving a strain resistant to ampicillin (MIC, 16 μg/ml), cefazolin (MIC, 16 μg/ml), and cefuroxime (MIC, 16 μg/ml), was made in 1997 by Dyer et al. (23), but the mechanism of resistance to β-lactams in these strains has not been investigated.

In the present study, we report the cloning and sequencing of the gene coding for the class A β-lactamase from C. sedlakii. The biochemical characteristics of this enzyme, named Sed-1, were studied. The transcriptional regulator associated with bla_Sed-1 was also cloned and sequenced. In order to address the issue of the genetic diversity of the β-lactamases found within the Citrobacter genus, DNA-DNA hybridization, high-stringency PCR, and DNA sequencing were used to study the distribution of bla_Sed-1 in 21 isolates of Citrobacter spp.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The clinical strain 2596 of C. sedlakii was isolated in 1997 from the bile of a patient treated with β-lactam and quinolone antibiotics. The strain was identified as C. sedlakii by biochemical tests using BIOTYPE 100 (Biomérieux) and the programs RECOGNIZER, ADANSON, and DENDOGRAPH of the Taxotron package (Institut Pasteur Taxolab). The identification was confirmed by amplifying and determining the nucleotide sequence of a PCR-amplified DNA fragment of 568 bp corresponding to the 16S RNA of the strain (the sequences of the primers are given in Table 1). The other bacterial strains and plasmids used in this work are listed in Table 2.

TABLE 1.

Nucleotide sequences of primers

Primer	Nucleotide sequence (5′ → 3′)	Genes	Annealing temp (°C)	Reference
AmpC div sens	AACGAGGTCGTCAGACG	blaCdiA	49	33
AmpC div A sens	CAGAATATCTTTACGCCC
Class A	AGCGAYAAYACGGCGATG	Class Aa	55	8
Class A rev	TGCKCCGGTYTTATCGCC
SHV rev	GCGTTGCCAGTGCTCGATCAG	blaSHV	62	6
SHV bis	ATGCGTTATATTCGCCTGTGTATT
TEM H	TGAGATCGAAGGGCCGTT	blaTEM	52	54
TEM rev	GGTCTGACAGTTACCAAT
OXA C	AGGTGCCATGAAAACATT	blaOXA group 1b	45
OXA C bis	TTAGCCACCAATGATGAA
OXA A	AAGGAAAAGTTAATGGCA	blaOXA group 2b	44
OXA A bis	TTTATCGCGCAGCGTCCG
OXA E	AGGACTTGGGACATCGAT	blaOXA18	47	48
OXA E bis	TGACTGGTCAGAAGTTTT
OXA B	AAGGAGGCTTCCTTGATAA	blaOXA20	49	38
OXA B bis	TTGGGTGGCAAAGCATTG
OXA C	As indicated above	blaOXA13	49	37
OXA O bis	TTATGTGCTTAGTGCATC
SmeA	CGGTCCTGA GGGGATGAC	blaSme-1	53	39
SmeB	CGTGATGCTTCCGCAATA
A1	GGAATTCCTWTGCTGCGCBCTGCTGCT	blaAmpCc	60	34
A2	CGGGATCCCTGCCAGTTTTGATAAAA
B1	GGAATTCCTCAFCGAGCAGACSCTGTT	blaAmpCc	60	34
B2	CGGGATCCCCCGCACMTKAYRTAGOTGTGG
CsA S	GCGCTGATTAATACCGC	blaSed-1	49	Present study
CsA AS	GCATCCTGCTGTGGCTGT
CUM 1	GGTCGTTTAGGTTAAAAC	blaCUM	48	22
CUM 1 bis	CCAGTGTTTTGGTAAACC
OXY 1	GTTCGTGGCGTAAAAC	blaOXY	49	25
OXY 1 rev	TAACACCTCTTTGCGGCT
RNA-S	AGAGTTTGATCCTGGYTCAG	16S RNA	56	61
RNA-AS	CTTTACGCCCARTAAWTCCG

^aDegenerated oligonucleotide primer designed from the consensus sequence of five class A β-lactamase genes (bla_CdiA, bla_OXY, bla_TOHO-1, bla_MEN, and bla_SHV). 

^bGroup 1 included oxacillinases OXA-5, -7, -11, -14, -16, and -17 and OXA-10 or PSE-2 (16, 18, 19, 21, 28, 29, 52). Group 2 included oxacillinases OXA-2, -3, -15, and -21 (17, 20, 50, 57). 

^cDegenerated oligonucleotide primers were designed from the consensus sequence of the ampC genes of E. coli, E. cloacae, and C. freundii for A1 and A2 and from MOX-1, FOX-1, and ampC of S. marcescens for B1 and B2. 

TABLE 2.

Bacterial strains and plasmids used

Strain or plasmid	Genotype or phenotype	Reference or source
Strains
C. sedlakii 2596a	Clinical isolate resistant to β-lactams	Present study
E. coli Top10	F−mcrA Δ(mrr-hsdRMS-mcrBC)φ80lacZ ΔM15ΔlacX74 deoR recA1 araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG
E. coli K12	Rifampin
C. koseri CKB	ESBLb	Clinical isolate
C. koseri CKC	ESBL	Clinical isolate
C. koseri CKD	ESBL	Clinical isolate
C. koseri CKE	ESBL	Clinical isolate
C. koseri CKG	ESBL	Clinical isolate
C. koseri CK1	Wild type	Clinical isolate
C. koseri CK2	Wild type	Clinical isolate
C. koseri CK3	Wild type	Clinical isolate
C. koseri CK4	Wild type	Clinical isolate
C. koseri CK5	Wild type	Clinical isolate
C. koseri CK7	Wild type	Clinical isolate
C. koseri CK10	Wild type	Clinical isolate
C. koseri CIP-82.87	Wild type (reference strain)	CIP
C. koseri CIP-76.3	Wild type (reference strain)	CIP
C. koseri CIP-72.8	Wild type (reference strain)	CIP
C. koseri CIP-72.14	Penicillinase	CIP
C. diversus CIP-82.94	Wild type (reference strain)	CIP
L. malonatica CIP-82.88	Wild type (reference strain)	CIP
C. sedlakii CIP-105037	Wild type (reference strain)	CIP
C. rodentium CIP-104675	Wild type (reference strain)	CIP
Plasmids
pBC SK+	Chloramphenicol	Stratagene
pBC 2596	pBC SK+ + 2.78-kb Sau3AI fragment from genomic DNA of C. sedlakii 2596	Present study

^aThe strain was deposited in the CIP under the number CIP-106793. 

^bPhenotype corresponding to penicillin and oxyminocephalosporin resistance by production of ESBLs. 

Antibiotics, media, and susceptibility testing.

The antibiotics were obtained from the following suppliers: ampicillin, oxacillin, and aztreonam from Bristol-Myers Squibb, Paris, France; chloramphenicol, cefuroxime, and cephalothin from Sigma Chemical Co., St. Louis, Mo.; cefoxitin and imipenem from Merck Sharp & Dohme, Chibret, France; cefotaxime, cefpirome, and rifampin from Hoescht-Marion-Roussel, Paris, France; ceftazidime and nitrocefin from Glaxo, Nanterre, France; ticarcillin, amoxicillin, and clavulanate, from SmithKline Beecham, Paris, France; piperacillin from Lederle, Paris, France; and benzylpenicillin from Sarbach, Suresnes, France.

MICs were determined on Mueller-Hinton (MH) agar by dilution technique (24) with a Steers multiple inoculator and an inoculum of 10⁴ CFU per spot. The plates were incubated at 37°C for 18 h. Brain heart infusion (BHI), Luria-Bertani (LB), and MH media were from Gibco-BRL.

Mating-out assays and plasmid content analysis.

Transfer of of β-lactam resistance to E. coli K-12 was attempted by liquid and solid mating-out assays. The recipient and donor cells were mixed into a ratio of 1:1 or 4:1 and were incubated in BHI with moderate shaking at 37°C for 3 h. After incubation, 200 μl of each mixture was plated out on a Millipore filter disk onto BHI plates and incubated for 18 h at 37°C. Transconjugants were selected on LB agar containing ampicillin (50 or 100 μg/ml) and rifampin (50 or 100 μg/ml). C. sedlakii 2596 was examined for its plasmid DNA content by the procedures of Birnboim (9) and Takahashi and Nagano (56).

DNA amplification.

DNA amplification of β-lactamase genes was carried out with the various specific primers (Eurogentec, Seraing, Belgium) listed in Table 1. The DNA amplifications were performed on 100-μl samples containing DNA (5 μl), deoxynucleoside triphosphate (250 μM), primers (0.4 μM concentrations each), Taq DNA polymerase (1 U), and its buffer. The following cycles were used: 10 min of denaturation at 94°C (1 cycle); 1 min of denaturation at 94°C, 1 min of annealing (see temperatures in Table 1), and 1 min of polymerization at 72°C (35 cycles), followed by 10 min of extension at 72°C. The amplified products were analyzed by electrophoresis of 5-μl aliquots on 1% agarose gels.

Nucleic acid techniques and sequence analysis.

Genomic DNA from C. sedlakii 2596 was extracted as described previously (49). For cloning experiments, the extracted DNA was partially digested with Sau3AI. The fragments were ligated into the dephosphorylated vector pBC SK⁺ previously digested with BamHI. The ligations were done at 4°C for 16 h with 100 ng of chromosomal DNA, 200 ng of digested plasmid vector pBC SK⁺, and 1 U of T4 DNA ligase (Amersham). After purification and concentration with the High Pure PCR product purification kit (Boehringer Mannheim), the ligation mixture was transformed by electroporation into Escherichia coli Top10. Transformants resistant to β-lactam antibiotics were selected on LB agar plates supplemented with 50 μg of amoxicillin/ml. Recombinant plasmid DNA was extracted using the rapid procedure of Birnboim (9) from 5-ml aliquots of overnight cultures grown at 37°C in BHI in the presence of amoxicillin (50 μg/ml) and chloramphenicol (50 μg/ml).

The inserted DNA fragment was sequenced on both strands by primer walking using the Taq DyeDeoxy Terminator cycle sequencing kit (Perkin Elmer) and an Applied Biosystems sequencer (PRISM 377). Sequence analysis was performed with the software available on the National Center for Biotechnology Information website. The program ORF Finder was used to determine all the putative open reading frames (ORFs), which were analyzed using the BLASTP program (3). The Sed-1 primary structure was analyzed with the software SignalP and ProtParam, which are available on the ExPASY tools website (Swiss Institute of Bioinformatics [5]).

For the hybridization experiments, genomic DNA was extracted from Citrobacter strains as described previously (49), except that phenol-chloroform was replaced by N-cetyl-N,N,N-trimethylammonium bromide (Merck). DNA was denatured by heating at 100°C for 10 min, and 5 μl of each sample was hybridized onto a nylon membrane (Hybond-N⁺; Amersham). The DNA was cross-linked on the membrane by 5 min of UV exposure. The hybridization was performed according to the manufacturer's instructions (ECL kit; Amersham) using as a probe an internal fragment of 668 bp obtained by PCR from the bla_Sed-1 gene.

Preparation of crude extracts and isoelectrofocusing.

Exponentially growing cells were harvested and resuspended in 600 μl of 50 mM phosphate sodium buffer (pH 7.0). The suspensions were disrupted by sonication and the crude extracts were used for β-lactamase detection. Isoelectric focusing was performed with an LKB Multiphor apparatus using polyacrylamide gel plates (Pharmacia Biotech, Saint Quentin en Yvelines, France) at pH 3.5 to 9.5 Gels were focused at 30 W for 90 min at 10°C. β-lactamase activity was revealed by staining the gel with the chromogenic β-lactam nitrocefin (40).

β-Lactamase purification and kinetic assays.

C. sedlakii 2596 was grown overnight at 37°C in 6 liters of BHI broth supplemented with ampicillin (50 μg/ml) and cefoxitin (2 μg/ml) in order to induce β-lactamase production. After centrifugation at 5,000 × g for 10 min at 4°C, the bacterial pellet (30 g) was resuspended in 120 ml of 50 mM Tris (pH 8.0). Bacterial cells were lysed by ultrasonic treatment and the suspension was clarified by centrifugation at 38,000 × g at 4°C. The nucleic acids contained in the supernatant were precipitated by adding spermine (0.2 M) at 4°C, followed by centrifugation for 10 min at 12,000 × g and 60 min at 48,000 × g. The supernatant was then dialyzed overnight against 3 liters of 50 mM Tris (pH 8.0). After an additional centrifugation at 12,000 × g for 30 min, the supernatant was applied onto a 2.5- by 10-cm Q-Sepharose Fast Flow column (Pharmacia Co. Ltd., Uppsala, Sweden) previously equilibrated with the dialysis buffer. β-lactamase activity was detected in the unabsorbed fraction with the chromogenic cephalosporin nitrocefin (40). The active fractions were pooled, dialyzed overnight at 4°C against 2 liters of 40 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid [pH 8.0]), and loaded onto a 2.5- by 10-cm S-Sepharose cation exchange column (Pharmacia Co. Ltd.) previously equilibrated with the dialysis buffer. The protein was eluted by a linear gradient of 0 to 1 M NaCl in 40 mM HEPES (pH 8.0). Active fractions were pooled, dialyzed overnight at 4°C against 1 liter of 40 mM Tris (pH 9.0), and then loaded on a Mono Q anion exchange column previously equilibrated with the dialysis buffer. The enzyme was eluted by a linear gradient of 0 to 1 M NaCl in 40 mM Tris (pH 9.0). The active fractions were pooled and dialyzed overnight at 4°C against 1 liter of 40 mM HEPES (pH 8.0) and loaded on a Mono S cation exchange column equilibrated with the dialysis buffer. The β-lactamase activity was eluted in the nonadsorbed fraction. Enzyme purity was assessed by electrophoresis on sodium dodecyl sulfate (SDS)–12.5% polyacrylamide gels. The intensity of the β-lactamase band was measured using a computerized densitometer (Densylab; Bioprobe) from a gel stained with Coomassie blue. The enzyme concentration was determined in reference to a standard bovine serum albumin scale analyzed under the same conditions. β-lactamase activity was renatured by soaking an SDS-polyacrylamide gel in 100 mM Tris-HCl (pH 7.0) for 1 h and was detected by overlaying the gel with nitrocefin (1,000 μg/ml). N-terminal sequences were determined after protein purification using an Applied Biosystems sequencer. The purified protein was stored in 50% glycerol at −20°C.

Kinetic measurements and inhibition of β-lactamase activity.

The kinetic parameters K_m and k_cat were determined spectrophotometrically at 35°C in 50 mM phosphate buffer (pH 7.0) using an Uvikon 940 spectrophotometer. The absorption coefficients used were those previously described (11). Kinetic parameters were determined by fitting the Henri-Michaelis-Menten equation to the experimental data by using the regression analysis program LEONORA written by Cornish-Bowden (15). The values of k_cat and K_m were estimated using a nonlinear least-squares regression method with dynamic weights (15).

Enzyme inhibition was studied with benzylpenicillin (100 μM) as the substrate. The different inhibitors, at various concentrations, were preincubated with the enzyme for 5 min at 35°C prior to addition of the substrate. The concentration of inhibitor required to inhibit 50% of the β-lactamase activity (IC₅₀) was determined graphically for clavulanic acid, sulbactam, and cefoxitin.

Nucleotide sequence accession numbers.

The nucleotide sequences described in this report have been deposited in GenBank under accession numbers AF321607 and AF321608.

RESULTS

Antibiotic susceptibility.

Analysis of C. sedlakii 2596 by the conventional disk susceptibility assay indicated that the strain produced an inducible β-lactamase inhibited by clavulanic acid. Indeed, a synergy was detected between clavulanic acid on one hand and aztreonam, cefuroxime, or cefotaxime on the other hand, while an antagonism phenomenon was visible between cefoxitin or imipenem and extended-spectrum cephalosporins like cefotaxime or cefepime (data not shown). Determination of the MICs of β-lactams for strain 2596 showed that it was resistant to penicillins and to early cephalosporins such as cephalothin and cefuroxime, but it remained susceptible to acylureidopenicillins such as piperacillin and to later β-lactams such as cefoxitin, cefotaxime, and aztreonam (Table 3). The MICs of amoxicillin and cephalothin were significantly lowered by clavulanic acid, while that of cefuroxime was increased fourfold when cefoxitin was added at a concentration of 2 μg/ml. Such a phenotype closely resembled that of C. sedlakii CIP-105037, a reference strain from the Collection of the Institut Pasteur (CIP), but it was significantly different from that of the wild-type strain CK1 of C. koseri (Table 3).

TABLE 3.

MICs of β-lactam antibiotics for various strains studied

β-Lactam	MIC (μg/ml)
	C. sedlakii 2596	E. coli Top10 (pBC2596)	E. coli Top10a	C. sedlakii CIP-105037b	C. koseri CK1c
Amoxicillin	>512	>512	2	>512	64
Amoxicillin + CLAd	8	512	2	8	1
Ticarcillin	>512	>512	1	512	64
Piperacillin	8	128	2	2	8
Cephalothin	256	>512	1	256	2
Cephalothin + CLA	4	1	1	4	NDf
Cefuroxime	64	>512	4	32	4
Cefuroxime + FOXe	256	ND	ND	256	ND
Cefoxitin	4	2	2	4	2
Cefotaxime	0.25	2	<0.125	<0.125	<0.125
Aztreonam	2	16	<0.125	1	<0.125

^aReference strain. 

^bReference strain from CIP. 

^cA wild-type strain. 

^dCLA, clavulanic acid at a fixed concentration of 4 μg/ml. 

^eFOX, cefoxitin at a fixed concentration of 2 μg/ml. 

^fND, not determined. 

Transfer of resistance and cloning of the bla gene from C. sedlakii 2596.

No plasmid was detectable in C. sedlakii 2596, suggesting that the β-lactamase gene was located on the chromosome. Accordingly, repeated mating-out experiments failed to transfer the β-lactam resistance into E. coli. The negative DNA amplification tests obtained with 15 pairs of primers designed to amplify the most frequent class A, C, and D β-lactamases (listed in Table 1) suggested that C. sedlakii 2596 harbored a β-lactamase gene characterized by a nucleotide sequence substantially different from that of the most commonly encountered bla genes.

The genomic DNA from C. sedlakii 2596 was partially digested with the restriction endonuclease Sau3AI and was ligated to the BamHI site of pBC SK⁺ vector. After electroporation into E. coli Top10, recombinant clones were selected on plates containing amoxicillin (50 μg/ml). Analysis by disk susceptibility assay of E. coli Top10 harboring the recombinant plasmid pBC 2596 revealed an antagonism between cefoxitin and extended-spectrum cephalosporins, as was observed for C. sedlakii 2596 (data not shown). Overall, the β-lactam MICs determined for E. coli(pBC 2596) were higher than those for C. sedlakii 2596, but the resistance profiles of the two strains were similar (Table 3).

Sequence analysis of bla_Sed-1 and bla_SedR.

Restriction analysis of the recombinant plasmid pBC 2596 showed the presence of a 2.78-kb DNA insert. The sequencing of this fragment revealed two ORFs in opposite orientations, SedR and Sed-1 (Fig. 1). The nucleotide sequence of bla_Sed-1, which is 888 bp long, displayed 82% identity with the chromosomal gene coding for the CdiA β-lactamase from C. koseri and 80% identity with the bla gene from K. oxytoca. The bla Sed-1 gene was found to encode a 31.9-kDa protein comprising 295 amino acid residues. Analysis of the amino acid sequence with the program SignalP suggested that the cleavage site in the precursor protein is located between the first alanine and glutamine residues in the amino acid sequence LHAQATSDVQQ (Fig. 1). This putative site corresponds well to the N-terminal sequence determined experimentally for the mature β-lactamase purified from C. sedlakii 2596, which was QATSDVQQVQKKLAALEKQ. The pI and molecular mass values of 8.86 and 28.6 kDa, respectively, which were predicted from the sequence of the deduced mature protein Sed-1, were in agreement with the values measured by isoelectrofocusing (8.6) and by SDS-PAGE analysis (30 kDa).

FIG. 1.

Nucleotide sequences of bla_Sed-I and bla_SedR from E. coli(pBC2596). The deduced amino acid sequence is indicated in single-letter code above the nucleotide sequence. RBS indicates a potential ribosome-binding site. The Sed-1 signal peptide extends from amino acid 1 to 28, and the postulated cleavage site is indicated by a vertical arrow. The residues conserved in the active site of serine β-lactamases are boxed. Two numbering schemes are indicated: the one in boldface is for the Sed-1 amino acid sequence, and the other one is for the Sed-1 and SedR nucleotide sequences.

The degree of amino acid sequence identity found between Sed-1 and the TEM and SHV enzymes was rather low (about 40%), while the protein had high identity with the chromosome-encoded β-lactamases of C. koseri (84%), K. oxytoca (74%), Serratia fonticola (67%), P. vulgaris (63%), and the plasmid-mediated enzymes MEN-1 (71%) and TOHO-1 (70%). A multiple amino acid sequence alignment of these class A β-lactamases is shown in Fig. 2. Most of the residues known to be involved in the catalytic mechanism and in substrate binding are conserved in Sed-1, including the consensus sequences ⁷⁰SXXK⁷³, ¹³⁰SDN¹³², ²³⁴KTG²³⁶, and the highly conserved residue Glu-166 (based on the numbering system of Ambler et al. [4]) (Fig. 2). Other important conserved residues were identified in Sed-1, such as Arg-164, Asn-170, and Asp-179, which are located in the Ω-loop structure. It must be noted here that the sequence of this catalytically important loop is highly conserved in Sed-1, compared to the class A β-lactamases presented in Fig. 2. Regarding the other positions known to contribute significantly to the catalytic activity of class A β-lactamases, we found in Sed-1 a cysteine in position 69, an asparagine in position 104, and an alanine and a glycine in positions 237 and 238, respectively. No arginine was identified in Sed-1 in either position 220 or position 244, but a basic residue (a Lys for Sed-1) is present in position 276, a feature which is shared by the seven class A β-lactamases related to the chromosomal enzyme of K. oxytoca shown in Fig. 2 (OXY-1, OXY-2, CdiA, MEN-1, TOHO-1, CUV, and CUM).

FIG. 2.

Multiple alignment of amino acid sequences for nine class A β-lactamases. Dashes indicate gaps inserted in the alignment; asterisks indicate identical residues. Numbering is according to Ambler (R. P. Ambler et al., Letter, Biochem. J. 276:269–270, 1991). Boldface residues are the highly conserved residues involved in the catalytic site of the protein. The β-lactamases included in the alignment are Sed-1 from C. sedlakii (present study), CdiA from C. diversus (32), OXY-1 and OXY-2 from K. oxytoca (25), MEN-1 from E. coli (8), TOHO-1 from E. coli (30), CUV from S. fonticola (44), CUM from P. vulgaris (22), and TEM from E. coli (54).

The bla_SedR gene found upstream from bla_Sed-1 is an 861-bp ORF divergently transcribed from bla_Sed-1 which encodes a 32-kDa protein comprising 286 amino acid residues. This protein displayed an identity of 90% with CdiR, the transcriptional regulator of the class A β-lactamase CdiA from C. koseri, and 68% identity with CumR from P. vulgaris. The identity found with the transcriptional regulator AmpR of class C β-lactamases was lower (47% with the C. freundii AmpR protein). As shown in Fig. 3, the identity is mainly located at the level of the N-terminal part of the sequences, where the helix-turn-helix motif required for binding to the ampR-ampA intercistronic region is found. Regarding the 20 residues involved in this motif, 14 are strictly conserved among the five AmpR sequences of class A β-lactamases shown in Fig. 3.

FIG. 3.

Alignment of LysR-type proteins. The AmpR sequences involved in the class A β-lactamase regulation systems were from C. diversus (CdiR), E. cloacae (NMCR), P. vulgaris (CumR), and Serratia marcescens (SmeR). LysR, the type regulator protein, was from E. coli. The predicted helix-turn-helix domain is boxed. Residues identical in SedR and the AmpR family sequences are indicated in boldface, whereas those identical in SedR and the LysR family sequences are indicated with asterisks. The residues highly conserved in LTTRs are shaded in gray (the amino acid numbering is indicated below the multiple alignment).

Purification and kinetic study of the β-lactamase Sed-1.

After four purification steps, two bands of 30 kDa (major band) and 26 kDa (minor band) were observed on SDS-PAGE analysis of the enzyme preparation. By soaking the polyacrylamide gel in Tris-HCl (pH 7.0), the β-lactamase activity could be renatured and was associated with the 30-kDa band. Isoelectric focusing performed from the partially purified enzyme confirmed the pI of 8.6 initially found for Sed-1 from C. sedlakii 2596.

As shown in Table 4, the highest catalytic activities were measured for benzylpenicillin, ampicillin, cefpirome, cephalothin, and cloxacillin (k_cat values between 300 and 165 s⁻¹). Nevertheless, the high K_m values found for ampicillin (565 μM) and cefpirome (1,000 μM) reduced the corresponding catalytic efficiencies of these two drugs (k_cat/K_M = 425 and 215 s⁻¹ · mM⁻¹, respectively) when compared to those of benzylpenicillin (6,650 s⁻¹ · mM⁻¹), cephalothin (6,000 s⁻¹ · mM⁻¹), and cloxacillin (1,270 s⁻¹ · mM⁻¹). Cefuroxime had a relatively low k_cat (65 s⁻¹) but a high apparent affinity (K_M = 20 μM), resulting in a catalytic efficiency (k_cat/K_M = 3,250 s⁻¹ · mM⁻¹) higher than that observed for cloxacillin. Ticarcillin, piperacillin, oxyimino-cephalosporins, and aztreonam were hydrolyzed with a lower catalytic efficiency (2 to 625 s⁻¹ · mM⁻¹; Table 4). Imipenem and cefoxitin hydrolysis was not detectable. The IC₅₀s determined with benzylpenicillin as a substrate showed that Sed-1 was well inhibited by clavulanic acid but poorly inhibited by sulbactam (0.065 and 2.5 μM, respectively). These IC₅₀s are comparable to the values of 0.09 μM (for clavulanic acid) and 6.1 μM (for sulbactam) obtained for the TEM-1 β-lactamase (14). Sed-1 is also weakly inhibited by cefoxitin, with an IC₅₀ of 16 μM.

TABLE 4.

Kinetic parameters of various β-lactam antibiotics for the Sed-1 β-lactamase from C. sedlakii 2596

Substrate	kcat (s−1)	Km (μM)	kcat/Km (s−1 · mM−1)	IC50 (μM)
Benzylpenicillin	300 ± 3	45 ± 2	6,650 ± 230
Ampicillin	240 ± 20	565 ± 75	425 ± 30
Ticarcillin	25 ± 1	40 ± 4	625 ± 50
Piperacillin	20 ± 0.05	165 ± 1	120 ± 0.04
Cephalothin	180 ± 0.3	30 ± 2	6,000 ± 30
Cefoxitin	<0.01	NDb	ND	16
Cefuroxime	65 ± 5	20 ± 0.3	3,250 ± 100
Cefotaxime	80 ± 5	170 ± 20	470 ± 40
Ceftazidime	5 ± 0.1	2,380 ± 200	2 ± 0.1
Aztreonam	30 ± 3	390 ± 80	75 ± 8
Cefpirome	215 ± 15	1,000 ± 100	215 ± 15
Oxacillin	85 ± 8	340 ± 60	255 ± 20
Cloxacillin	165 ± 8	130 ± 10	1,270 ± 40
Imipenem	<0.01	ND	ND
Clavulanic acid				0.065
Sulbactam				2.5

^aMean ± standard deviation values are indicated. 

^bND, not determined. 

β-Lactamase diversity within the Citrobacter genus.

Twenty strains representing three species (C. koseri, C. sedlakii, and C. rodentium) within the Citrobacter genus were studied. The two reference strains, C. sedlakii CIP-105037 and C. rodentium CIP-104675, were from the CIP. For C. koseri, 12 clinical isolates were studied in addition to three CIP reference strains. All the strains were identified using Api 20E galleries, the identification being confirmed by PCR and sequencing of the hypervariable region in the 16S rRNA. In order to characterize the β-lactamases present in each isolate, the pIs of the β-lactamases produced by each strain were determined by isoelectrofocusing. Genetic characterizations by PCR with primers specific to bla_TEM, bla_CdiA, and bla_Sed-1 were also carried out. The results are shown in Table 5.

TABLE 5.

Isoelectric points and results of the hybridization and amplification experiments carried out on the different Citrobacter isolates

Strain	Phenotype	Isoelectric point(s)	PCR result			Hybridization with blaSed-1 probe
			blaTEM	blaCdiA	blaSed-1
C. sedlakii 2596	Wild type	8.6	−	−	+	+
C. sedlakii CIP105037	Wild type	8.7	−	−	+	+
C. rodentium CIP104675	Wild type	8.7	−	−	+	+
C. koseri CKB	ESBL	6.8, 6.1, 4.8	+	−	−	−
C. koseri CKC	ESBL	6.1, 5.7, 4.8	+	−	−	−
C. koseri CKD	ESBL	6.8, 6.1, 4.8	+	−	−	−
C. koseri CKE	ESBL	6.8, 6.1, 4.8	+	−	−	−
C. koseri CKG	ESBL	6.1	+	−	−	−
C. koseri CIP-72.14	Penicillinase	5.4	+	−	−	−
C. koseri CK1	Wild type	7, 4.7	−	−	−	−
C. koseri CK2	Wild type	8.3	−	−	−	−
C. koseri CK3	Wild type	5.6, 5.4	−	−	−	−
C. koseri CK4	Wild type	5	−	−	−	−
C. koseri CK5	Wild type	7	−	−	−	−
C. koseri CK7	Wild type	5	−	−	−	−
C. koseri CK10	Wild type	8.2, 5.6, 5.3	−	−	−	−
C. koseri CIP-82.87	Wild type	5.7	−	−	−	−
C. koseri CIP-76.3	Wild type	5	−	−	−	−
C. koseri CIP-82.94	Wild type	7.4	−	−	−	−
L. malonatica CIP82.88	Wild type	8.1, 5, 4.5	−	−	−	−

The clinical strain 2596 and the reference strain CIP-105037 of C. sedlakii displayed very similar phenotypes: they were resistant to aminopenicillins, carboxypenicillins, and early cephalosporins but remained susceptible to piperacillin (Table 3). An antagonism between imipenem and extended-spectrum cephalosporins could be observed for both strains, indicating inducible β-lactamase production in C. sedlakii. The hybridization experiment carried out with a bla_Sed-1-specific probe gave a positive result for both strains. Accordingly, sequencing of the DNA products amplified with primers specific for bla_Sed-1 showed that C. sedlakii CIP-105037 harbored a β-lactamase gene, the sequence of which is very similar to that of bla_Sed-1 (98% identity for 640 bp sequenced). The differences found at the amino acid level between the two β-lactamases accounted for the difference in pI values observed for C. sedlakii 2596 (8.6) and for C. sedlakii CIP-105037 (8.7).

The C. rodentium CIP-104675 strain produced a β-lactamase characterized by a pI value of 8.7 and yielded a positive hybridization signal with the internal probe specific for bla_Sed-1. The DNA fragment of 640 bp obtained by PCR from C. rodentium CIP-104675 was sequenced and was found to share 80% identity with bla_Sed-1.

Regarding C. koseri, three different phenotypes of resistance to β-lactams (wild type, penicillinase, and extended-spectrum β-lactamase [ESBL] phenotypes) could be identified among the 17 strains included in the present study. As shown in Table 5, very heterogeneous pI values were found for these different strains of C. koseri which produced different combinations of noninducible β-lactamases, each strain being characterized by one, two, or three distinct pI values. PCR amplification and sequence analysis were made in an attempt to identify the corresponding β-lactamase genes. Amplification with the primers specific for bla_TEM confirmed the presence of this gene in the six strains showing a penicillinase or an ESBL phenotype. By contrast, the amplifications with the bla_CdiA and bla_Sed-1 primers remained negative in all the strains tested, and none of the C. koseri strains hybridized with the bla_Sed-1 internal probe.

DISCUSSION

C. sedlakii 2596, which produces the chromosomally encoded class A β-lactamase Sed-1, was resistant to aminopenicillins, carboxypenicillins, and early cephalosporins but not to acylureidopenicillins such as piperacillin. This resistance profile can be readily explained by the hydrolytic properties of Sed-1 (high catalytic efficiency against benzylpenicillin, cloxacillin, and early cephalosporins such as cephalothin and cefuroxime, but low efficiency for piperacillin), which closely resemble those of the cefuroximases, such as the chromosome-encoded β-lactamase CUM from P. vulgaris or the plasmid-encoded β-lactamases FEC-1, FPM-1, and FUR from E. coli, Proteus mirabilis, and Klebsiella pneumoniae, respectively (36, 45, 58, 59). Cefotaxime, cefpirome, and aztreonam were also hydrolyzed by Sed-1 but with low catalytic efficiencies and moderate apparent affinities (Table 4). Clavulanic acid and sulbactam inhibited the β-lactamase activity of Sed-1 with IC₅₀s similar to those encountered among class A β-lactamases highly susceptible to these inhibitors, such as TEM-1.

With the aim of determining whether the unusual substrate profile of Sed-1, including its significant activities with cefuroxime, cefotaxime, and cefpirome, could be related to the presence of specific amino acid residues in the sequence of the enzyme, we compared its amino acid sequence with those from other class A β-lactamases. A high percentage of identity was found between Sed-1 and the chromosome-encoded β-lactamases of C. koseri (84%), K. oxytoca (74%), S. fonticola (67%), P. vulgaris (63%), and the plasmid-mediated enzymes MEN-1 or CTX-M (71%) and TOHO-1 (70%), suggesting that these class A enzymes may be derived from a common ancestor. All the conserved residues considered to be important for catalysis in class A β-lactamases (⁷⁰SXXK⁷³, ¹³⁰SDN¹³², ²³⁴KTG²³⁶, and E¹⁶⁶) were found in Sed-1, as was the Ω-loop, an important structural element including the amino acid residues 161 to 179 in class A enzymes.

Regarding positions 104, 164, 179, 205, 237, 238, and 240, which are involved in the extended substrate specificity of the mutants of the class A β-lactamases TEM and SHV (35), it is striking that an Asn was found at position 104 in Sed-1. Indeed, according to Petit et al. (46), residue 104 contributes to the precise positioning of the ¹³⁰SDN¹³² loop, which is a crucial structural and catalytic element for the binding and hydrolysis of β-lactams. Accordingly, amino acid modifications are found at this position in a large number of ESBLs derived from TEM-1 and TEM-2. Moreover, an asparagine residue is also found at position 104 in the nine CTX-M variants described to date and also in related enzymes such as TOHO-1 from E. coli and CUV from S. fonticola (Fig. 2), which, like Sed-1, are enzymes that efficiently hydrolyze cefuroxime and cefotaxime but not ceftazidime (8, 10, 12, 26). Note here that the residue found at position ABL 237 in Sed-1 is an alanine, whereas a serine has been found at this position in CUV, CUM, MEN-1 (CTX-M), and TOHO-1. In TEM-type ESBLs, as in other class A enzymes with an extended spectrum of activity, the presence of a serine residue in position 237 has been clearly shown to increase the level of β-lactamase activity against expanded-spectrum cephalosporins. Consequently, the presence of Ala 237 in Sed-1 could explain the relatively low level of activity the enzyme displays against cefotaxime, compared to the activities reported for the related enzymes MEN-1 (CTX-M), CUV, CUM, and TOHO-1, which all confer a high level of resistance to this drug (7, 22, 30, 44).

Other amino acid residues of interest, which are not strictly conserved in class A βlactamases, were found in Sed-1. Among them, there is a cysteine at position 69 in Sed-1, which neighbors the active-site serine found at position 70. Cys 69 is present in all the β-lactamases belonging to the K. oxytoca subgroup (Fig. 2), so that this residue could have a significant role in the substrate profile of these enzymes. Also note that there is no arginine residue at positions ABL 220 or 244 in Sed-1. Matagne et al. (35) have suggested that a basic residue is found in position ABL 276 when Arg is absent at positions 220 and 244. This is the case in Sed-1, which presents a Lysine at position 276 that is highly conserved in the cefuroximases group and perfectly aligned with the corresponding arginine residue found in MEN-1 and TOHO-1 (Fig. 2).

The negative results obtained from the conjugation and plasmid extraction experiments strongly suggested a chromosomal location for the bla_Sed-1 gene. The presence, upstream from bla_Sed-1, of a gene coding for a transcriptional regulator belonging to the LysR family reinforces this hypothesis. In C. sedlakii, the regulator protein SedR induces the production of Sed-1 when the bacteria are grown in the presence of an inducer β-lactam antibiotic such as cefoxitin and imipenem, as illustrated by the increase of the MIC of cefuroxime observed in the presence of cefoxitin for C. sedlakii 2596 (Table 3). Regarding its amino acid sequence, SedR is more related to the regulators of the class A β-lactamases from C. koseri and P. vulgaris (90% identity with CdiR and 68% identity with CumR) than to the regulators of the class C β-lactamases (47% identity with the C. freundii AmpR protein), suggesting that the relationships among LysR proteins are related to the type of β-lactamase they regulate. The similarities are the highest around the N-terminal region, where the consensus sequence for the helix-turn-helix motif is found. In this region, the residues conserved among the LTTRs which are Ala(Gly)-27, Ser(Thr)-33, Gln-34, Pro-35, Phe(Leu)-44, and Glu-45 (51), are all present in SedR, except for Pro-35.

β-Lactamase distribution within the Citrobacter genus is complex. C. freundii produces an inducible chromosome-encoded class C β-lactamase (62), whereas C. sedlakii, C. rodentium, and C. koseri produce class A β-lactamases (27, 41, 47, also, the present study). On the basis of the sequencing of the DNA fragments amplified in this study, C. sedlakii CIP-105037 and C. rodentium CIP-104675 harbor β-lactamase genes sharing 98 and 80% of identity to bla_Sed-1, respectively. The expression of bla_Sed-1, and also that of the closely related gene found in C. sedlakii CIP-105037, is inducible, whereas that of the bla_Sed-1-like gene found in C. rodentium CIP-104675 seems to be constitutive. Regarding C. koseri, DNA amplification experiments with bla_Sed-1-specific primers showed that this β-lactamase gene was not present in the 17 C. koseri isolates tested in this study. Moreover, the results obtained by PCR were confirmed by dot-blot hybridization with a bla_Sed-1-specific probe, strongly supporting the idea that this gene is not present in C. koseri. More surprisingly, no DNA amplification could be obtained using the set of primers designed to amplify specifically the bla_CdiA gene initially reported in the strain ULA27 of C. koseri, suggesting that the latter gene is not ubiquitous in this species or, alternatively, that the strain used for the initial characterization of bla_CdiA was not a C. koseri strain. Such a hypothesis is confirmed by the fact that the constitutive resistance phenotypes observed for the clinical isolates of C. koseri included in the present study were all clearly different from the inducible resistance phenotype initially reported for the strain ULA27 of C. koseri, which produces CdiA under control of the regulator protein CdiR (33). Moreover, we have characterized 13 distinct isoelectric point values among the 17 C. koseri strains studied, with one to three different pI values detected per strain. The strains with a wild-type phenotype had pI values varying from 4.7 to 8.2. Such values are in agreement with those found in the literature, which vary from 4.8 to 9.5 (27, 47). For the strains displaying an ESBL profile, three pI values of 6.8, 6.1, and 4.8 were found, one corresponding to the pI value of a TEM variant, the identification of which was confirmed by PCR with bla_TEM primers. Therefore, it is likely that the β-lactamase genes found in C. koseri encompass a series of genes that have significantly diverged, as previously described by Jones et al. (32), and which are characterized by a nucleotide sequence different from that of bla_Sed-1 and bla_CdiA. Further studies on the genetic characterization of these genes must be done to elucidate the origin of the bla genes in C. koseri.