β-Lactamases in Ampicillin-Resistant Escherichia coli Isolates from Foods, Humans, and Healthy Animals

Abstract

TEM-, SHV-, and OXA-type β-lactamases were studied by PCR with 124 ampicillin-resistant (AMP^r) Escherichia coli isolates recovered from foods of animal origin (n = 20) and feces of humans (n = 49) and healthy animals (n = 55). PCR showed that 103 isolates were positive for TEM and negative for SHV and OXA. Three E. coli isolates showed a positive reaction for OXA, and one showed a positive reaction for SHV. The remaining 17 E. coli isolates were negative for the three enzymes by PCR. Fifty-seven of the 103 bla_TEM amplicons were sequenced. Different molecular variants of bla_TEM-1 were found in 52 isolates: bla_TEM-1a (n = 9), bla_TEM-1b (n = 36), bla_TEM-1c (n = 6), and bla_TEM-1f (n = 1). Four inhibitor-resistant TEM (IRT) β-lactamase-encoding genes were also detected: bla_TEM-30c (IRT-2), bla_TEM-34b (IRT-6), bla_TEM-40b (IRT-11), and bla_TEM-51a (IRT-15). A new bla_TEM gene, named bla_TEM-95b, which showed a mutation in amino acid 145 (P→A) was detected. It was found in a food isolate of chicken origin (AMP^r, amoxicillin-clavulanic acid susceptible). The promoter region in 24 bla_TEM amplicons was analyzed, and the weak P3 promoter was found in 23 of them (bla_TEM-1 in 20 amplicons and bla_TEM-51a, bla_TEM-30c, and bla_TEM-95b in 1 amplicon each). The strong Pa/Pb promoter was found only in the bla_TEM-34b gene. No extended-spectrum β-lactamases were detected. Mutations at position −42 or −32 in the ampC gene promoter were demonstrated in 4 of 10 E. coli isolates for which the cefoxitin MIC was ≥16 μg/ml. Different variants of bla_TEM-1 and IRT bla_TEM genes were found among the AMP^r E. coli isolates from foods and the feces of humans and healthy animals, and a new gene, bla_TEM-95b (P3), was detected.

Escherichia coli is one of the main causes of nosocomial infections in humans. E. coli is also a common inhabitant of the human and animal gut and is considered an indicator of fecal contamination in food. β-Lactams are widely used in human and veterinary medicine to treat human and animal infections (31). This widespread use of antibiotics could be associated with the selection of antibiotic resistance mechanisms in pathogenic and nonpathogenic isolates of E. coli (46).

Resistance to β-lactam antimicrobial agents in E. coli is primarily mediated by β-lactamases, which hydrolyze the β-lactam ring and thus inactivate the antibiotic (30). Many different β-lactamases have been described (9, 30, 31). Over 200 β-lactamases have been classified into four main groups and eight subgroups according to their functional and structural characteristics (9, 10). The classical TEM-1, TEM-2, and SHV-1 enzymes are the predominant plasmid-mediated β-lactamases of gram-negative rods. Six different nucleotide sequences, called bla_TEM-1a, bla_TEM-1b, bla_TEM-1c, bla_TEM-1d, bla_TEM-1e, and bla_TEM-1f, have been described to codify the same TEM-1 β-lactamase (15, 20, 21, 28, 47). Weak (P3) and strong (P4 and Pa/Pb) promoters have been reported for bla_TEM genes (14, 20, 21, 28). Some variants of the TEM-1, TEM-2, and SHV-1 β-lactamases have emerged as a result of single amino acid substitutions in the sequences of the genes which render the extended-spectrum β-lactamases (ESBLs), which inactivate newer cephalosporins but which are still susceptible to β-lactamase inhibitors (e.g., clavulanic acid) (30).

β-Lactamase inhibitor-resistant strains emerged during the 1980s. Susceptibility to β-lactamase inhibitors could be affected in E. coli by different mechanisms. The most frequent one is the hyperproduction of classical β-lactamases or the synthesis of inhibitor-resistant TEM (IRT) β-lactamases by amino acid substitutions in TEM-1 or TEM-2. Other possible mechanisms are the hyperproduction of chromosomal AmpC β-lactamase (by gene amplification or the introduction of mutations at either the promoter or the attenuator of the structural gene) (11, 16, 25, 36, 37) and some types of OXA β-lactamases (30, 34), plasmidic cephalosporinase production (e.g., FOX) (2, 5, 19, 32, 40), or even changes in membrane permeability (33).

Multiple studies focused on the characterization of β-lactamases in human clinical E. coli isolates have been performed, but very few studies have been performed with E. coli isolates of other origins such as healthy animals or foods (23, 46), sick animals (6, 17; T. L. Teshager, L. Dominguez, M. A. Moreno, Y. Sáenz, M. Zarazaga, C. Torres, and S. Cardeñosa, Letter, Antimicrob. Agents Chemother. 44:3483-3484, 2000), or healthy humans (8, 22, 42). The objective of this study was to characterize the types of β-lactamases produced by 124 ampicillin (AMP)-resistant (AMP^r) nonpathogenic E. coli isolates recovered from foods and from the feces of humans and healthy animals.

MATERIALS AND METHODS

Bacterial isolates.

All 124 AMP^r E. coli isolates (MICs ≥ 32 μg/ml) recovered in a previous study (41) from samples of different origins (food products and fecal samples of humans and animals) were included in this work. The origins of the 124 AMP^r isolates were as follows (number of isolates): food products of chicken origin (n = 20), feces of healthy animals (broilers, n = 22; pigs, n = 20; pets, bulls, or horses, n = 13), and human fecal samples (patients, n = 40; healthy volunteers, n = 9). The healthy human volunteers from whom E. coli isolates were recovered had not been treated with antibiotics for at least 3 months preceding isolation of E. coli. The E. coli isolates from patients were not implicated in any kind of infection, and they were considered part of the normal microflora in these individuals.

Antibiotic susceptibility.

The antibiotic susceptibilities of the AMP^r E. coli isolates were analyzed by the NCCLS standard agar dilution method (35). The following antibiotics were tested: AMP, cefazolin, cefoxitin, cefotaxime, and ceftriaxone (Sigma Chemical Co., St. Louis, Mo.); amoxicillin-clavulanic acid (AMC) and ticarcillin (TIC; SmithKline Beecham, Madrid, Spain); ceftazidime (Glaxo, Madrid, Spain); imipenem (Merck Sharp & Dohme, Madrid, Spain); and aztreonam (Bristol-Myers Squibb, Madrid, Spain).

Detection of TEM, SHV, and OXA β-lactamase-encoding genes.

Microorganisms were grown on brain heart infusion agar plates (Difco, Detroit, Mich.) for 24 h at 37°C, and one colony was resuspended in 500 μl of sterile distilled water. The cells were lysed by heating at 95°C for 10 min, and cellular debris was removed by centrifugation at 16,000 × g for 2 min. The supernatant was used as the source of template for PCRs. PCR amplifications of the bla_TEM, bla_SHV, and bla_OXA genes were carried out for screening purposes, as described previously (3, 39, 45). The following primers were used in these reactions: for bla_TEM, primers tem-F (5′-TTCTTGAAGACGAAAGGGC-3′) and tem-R (5′-ACGCTCAGTGGAACGAAAAC-3′); for bla_SHV, primers shv-F (5′-CACTCAAGGATGTATTGTG-3′) and shv-R (5′-TTAGCGTTGCCAGTGCTCG-3′); and for bla_OXA, primers oxa-F (5′-TTCAAGCCAAAGGCACGATAG-3′) and oxa-R (5′-TCCGAGTTGACTGCCGGGTTG-3′). The conditions used for these reactions were as indicated in previous papers (3, 39, 45). The sizes of the bla_TEM, bla_SHV, and bla_OXA amplicons were 1,150, 885, and 702 bp, respectively. Positive controls (E. coli C282 [bla_TEM-1], E. coli C321 [bla_OXA-2], and E. coli EC98/4453-2 [bla_SHV-12] [Teshager et al., letter, 2000]) as well as negative controls were included in each type of PCR. Fifty-seven bla_TEM amplicons from isolates showing different phenotypes of resistance to the AMC association were sequenced (both strands) with the Rhodamine Dye Terminator Cycle Sequencing kit and analyzed in an automatic DNA sequencer (ABI 310; Applied Biosystems). The same set of primers used in the PCR analysis was used for sequencing purposes. DNA and deduced amino acid sequences were compared with those previously described for bla_TEM genes (EMBL database and the website of G. Jacoby and K. Bush [http://www.lahey.org/studies/webt.htm]). The nomenclature used for the bla_TEM variants and promoters was the one proposed previously (21, 28).

The isolates were screened for ESBLs by the double disk diffusion method (30).

DNA sequence analysis of regulatory region of ampC gene.

A 191-bp fragment of the promoter region of the ampC gene was amplified and sequenced by using primers ampC-F (5′-AATGGGTTTTCTACGGTCTG-3′) and ampC-R (5′-GGGCAGCAAATGTGGAGCAA-3′) (11) and the conditions described previously (11). The mutations in the promoter and attenuator regions were studied by comparing the sequences with the sequence of the same region in the E. coli K-12 ampC gene (24).

Nucleotide sequence accession number.

The nucleotide sequence reported for bla_TEM-95b is included in the EMBL database under accession number AJ308558.

RESULTS

MIC determination.

Table 1 shows the MICs of the different β-lactams tested for the 124 AMP^r E. coli isolates from different origins included in this study (humans, 49 isolates; food and animals, 75 isolates). Most of the isolates (91%) were resistant to TIC (94% of the human isolates and 85% of the food and animal isolates). AMC resistance (MIC ≥ 32 μg/ml) was detected in 16% of the human isolates and 8% of the animal and food isolates. Thirty-five percent of the human isolates and 43% of the isolates of other origins were included in the intermediate category for AMC (MIC = 16 μg/ml). A cefoxitin MIC ≥16 μg/ml was observed for 10 isolates, 5 of which were from human samples (10%) and 5 of which were from the other origins (7%). Diminished susceptibility to extended-spectrum cephalosporins (MIC ≥ 2 μg/ml) was found for three isolates (one isolate each from a broiler, a pig, and a human patient).

TABLE 1.

Susceptibilities to different β-lactams of the 124 AMP^r E. coli isolates included in the study

Antibiotic and origina	No. of isolates for which MIC (μg/ml) was:
	≤0.015	0.03	0.06	0.125	0.25	0.5	1	2	4	8	16	32	64	128	≥256
Ampicillin
Humans												2			47
Animals or food												7	3	3	62
Ticarcillin
Humans										2		1			46
Animals or food							2	1	1	2		1	1	3	64
AMC
Humans									4	20	17	5	3
Animals or food									14	23	32	2	4
Cefazolin
Humans							2	25	13	6	2	1
Animals or food						1	7	32	28	3			2		2
Cefoxitin
Humans							1		23	21	2	1	1
Animals or food							2	7	22	38	1		1	4b
Ceftazidime
Humans			2	12	32	1	1			1
Animals or food	1		10	19	31	12		1		1
Cefotaxime
Humans		1	22	20	3	2			1
Animals or food	1	14	21	19	13	5		1	1
Ceftriaxone
Humans	2	19	19	7	1			1
Animals or food	8	27	27	11			2
Imipenem
Humans		4c	1	22	21	1
Animals or food				16	54	5
Aztreonam
Humans		19	17	10	1	1			1
Animals or food	5	24	29	14	1		1		1

^aHumans, n = 49; Animals or food n = 75.

^bMIC, ≥128 μg/ml.

^cMIC, ≤0.03 μg/ml.

Distribution of mechanisms of resistance.

PCRs with primers specific for the bla_TEM, bla_SHV, and bla_OXA genes were performed with the 124 AMP^r isolates included this study, and the results are shown in Table 2. Positive PCR results for the bla_TEM gene and negative PCR results for the bla_SHV or bla_OXA gene were found for 103 of the 124 (83%) AMP^r E. coli isolates. An SHV-type β-lactamase was detected by PCR in one additional isolate that was negative by PCRs for both bla_TEM and bla_OXA. An OXA-type β-lactamase was detected in three other isolates that were negative by PCRs for both bla_TEM and bla_SHV. The PCRs for all genes were negative for the remaining 17 isolates. For 7 of these 17 E. coli isolates, as well as 3 isolates with bla genes, the cefoxitin MICs were ≥16 μg/ml (Table 1). No ESBLs were found by the double disk test in any of our E. coli isolates.

TABLE 2.

PCR detection of bla genes in the 124 AMP E. coli isolates included in this study

Origin and AMC resistance phenotype of isolates			No. of isolates in which bla genes for the following were detected by PCR:
Origin (no.)	Phenotypea	No. of isolates	TEM	SHV	OXA
Broiler (22)	S	15	12	0	0
	I	6	5	0	1
	R	1	0	0	0
Food (20)	S	10	5	1	1
	I	5	4	0	0
	R	5	1	0	0
Pig (20)	S	9	5	0	1
	I	11	11	0	0
	R	0	0	0	0
Pet, bull, or horse (13)	S	3	3	0	0
	I	10	10	0	0
	R	0	0	0	0
Human patient (40)	S	23	23	0	0
	I	11	11	0	0
	R	6	6	0	0
Healthy human (9)	S	1	1	0	0
	I	6	6	0	0
	R	2	0	0	0

^aS, susceptible; I, intermediate; R, resistant.

Fifty-seven of the 103 bla_TEM amplicons detected were sequenced. These 57 amplicons corresponded to the following E. coli isolates: (i) most of the isolates for which the AMC MIC was ≥16 μg/ml (intermediate or resistant phenotype; 48 of 54 isolates in which the bla_TEM gene was detected by PCR) and (ii) a group of isolates for which the AMC MIC was ≤8 μg/ml (susceptible phenotype; 9 of 49 isolates in which the bla_TEM gene was detected by PCR). Table 3 shows the bla_TEM sequences obtained. The bla_TEM-1 gene was found in 52 of the 57 sequences studied (91%). Different molecular variants of bla_TEM-1 were detected in the analysis of the nucleotide sequences of the structural gene and the promoter: bla_TEM-1a, bla_TEM-1b, bla_TEM-1c, and bla_TEM-1f. Neither bla_TEM-1d nor bla_TEM-1e variants were found. The most frequent molecular variant was bla_TEM-1b (69%), followed by bla_TEM-1a (17%). Curiously, bla_TEM-1c was mostly found in isolates from healthy humans (four of the six isolates in which this bla_TEM molecular variant was detected), and bla_TEM-1f was found in only one isolate recovered from a human patient. Other types of bla_TEM genes were identified (Table 3) in the remaining 5 of the 57 amplicons sequenced: bla_TEM-30c (which encodes the TEM-30 β-lactamase, also named the IRT-2 β-lactamase)_, bla_TEM-34b (which encodes the TEM-34 β-lactamase, also named the IRT-6 β-lactamase), bla_TEM-40b (which encodes the TEM-40 β-lactamase, also named the IRT-11 β-lactamase), bla_TEM-51a (which encodes the TEM-51 β-lactamase, also named the IRT-15 β-lactamase), and a new bla_TEM gene not previously described (see below). These four IRT β-lactamases showed a single amino acid substitution either at position 69 (TEM-34, Met→Val; TEM-40, Met→Ile) or at position 244 (TEM-30, Arg→Ser; TEM-51, Arg→His) and were detected in isolates recovered from a healthy pig (TEM-51) or from human feces (TEM-30, TEM-34, and TEM-40). The sequences of each of these IRT β-lactamase genes were obtained by sequencing both strands of the amplicons from two or three independent PCRs.

TABLE 3.

Types of bla_TEM genes detected in 57 AMP^r E. coli isolates in this study

Origin and AMC resistance phenotypes of isolates			blaTEM genes types
Origin	Phenotypea	No. of isolates	No. of isolates with the following blaTEM-1 molecular variant:				Other blaTEM genesb (promoter)
			blaTEM-1a	blaTEM-1b	blaTEM-1c	blaTEM-1f
Broiler	S	8	1	7
	I	3	1	2
Food	S	1					blaTEM-95b (P3)c
	I	4	1	3
	R	1		1
Pig	I	10		8	1		blaTEM-51a (P3)
Pet, bull, or horse	I	9	1	7	1
Healthy human	I	5		1	4
Human patient	I	10	5	5
	R	6		2		1	blaTEM-30c (P3), blaTEM-34b (Pa/Pb), blaTEM-40b

^aS, susceptible; I, intermediate; R, resistant.

^bEach of these bla_TEM genes was detected in a single E. coli isolate.

^cNew bla_TEM gene found in this study (EMBL database accession number AJ308558).

A new bla_TEM gene was detected in this study and was named bla_TEM-95b. It was found in an E. coli isolate (E. coli Co52) recovered from a food sample. A comparison of the nucleotide and amino acid sequences of the new bla_TEM-95b gene and those of the bla_TEM-1a and bla_TEM-1b genes is shown in Table 4. The gene encoding bla_TEM-95b shows a mutation at nucleotide 635 (C→G) (numbering is according to Sutcliffe [47]), causing a change in amino acid 145 (proline→alanine) (numbering is according to Ambler and Coulson [1]). Both strands of the amplicons from three independent PCRs were sequenced to ascribe this sequence, and identical results were obtained in all three cases. This sequence is included in the EMBL database under accession number AJ308558. The MICs of β-lactam antibiotics for E. coli Co52 are as follows: AMP and TIC, >256 μg/ml; AMC, 8/4 μg/ml; cefazolin, 4 μg/ml; cefoxitin, 8 μg/ml; ceftazidime, 0.5 μg/ml; cefotaxime and imipenem, 0.25 μg/ml; and ceftriaxone and aztreonam, 0.125 μg/ml. As can be observed, E. coli Co52 shows a phenotype of resistance similar to that conferred in E. coli by the expression of a TEM-1-type β-lactamase.

TABLE 4.

Nucleotide mutation and amino acid substitution in the new β-lactamase-encoding gene bla_TEM-95b compared to the sequences of bla_TEM-1a and bla_TEM-1b

blaTEM gene (promoter)	Base at the indicated nucleotide positiona (amino acid position)b in the:
	Promoter region			Coding region
	32	162	175	226 (6)	436 (78)	604 (134)	635 (145)
blaTEM-1a (P3)	C	G	A	C	C	G	C (Pro)
blaTEM-1b (P3)	C	G	G	T	T	T	C
blaTEM-95b (P3)	C	G	G	T	T	T	G (Ala)

^aNumbered according to Sutcliffe (47).

^bNumbered according to Ambler and Coulson (1).

The promoter region was analyzed in 24 bla_TEM amplicons in which the electropherogram allowed us a clear reading of the entire region. The weak P3 promoter was found in 23 of these sequences. The AMC MICs for all isolates with bla_TEM genes and the P3 promoter were within the range of 8 to 32 μg/ml (for 17 of these isolates the AMC MIC was 16 μg/ml). Of the series of 23 isolates with the P3 promoter, 20 of them had the bla_TEM-1 gene, 1 had the bla_TEM-51a gene, and other 2 had the bla_TEM-30c and the bla_TEM-95b genes, respectively. The strong Pa/Pb promoter was found in only one isolate with the bla_TEM-34b gene, which encodes an IRT β-lactamase (the AMC MIC for this isolate was 64 μg/ml) (Table 3). The Pa/Pb promoter was identified by a mutation at nucleotide position 32 (C→T) of the bla_TEM promoter region.

In this study, the cefoxitin MICs for 10 E. coli isolates were ≥16 μg/ml (4 isolates from foods, 2 isolates from broilers, and 4 isolates from humans) (Table 5). A bla_TEM gene was detected by PCR in 3 of these 10 isolates. Amplification of the ampC gene promoter, including the −35 and −10 boxes as well as the attenuator, was carried out for all 10 isolates by using specific primers. The expected 191-bp fragment was obtained for six of these isolates. A negative PCR result was obtained for the remaining four isolates (the cefoxitin MIC was >64 μg/ml for all of them); these four isolates proved to be unrelated when they were studied by the pulsed-field gel electrophoresis methodology (data not shown). The six ampC amplicons detected were sequenced, and mutations were analyzed by comparing the nucleotide sequences with the nucleotide sequence of the same region in the E. coli K-12 ampC gene (24) (Table 5). Two promoters (E. coli Co100 and A99; cefoxitin MICs = 64 μg/ml) showed point mutations at positions −42, −18, −1, and +58. Two promoters (E. coli Co313 and Co321; cefoxitin MICs = 16 μg/ml) showed mutations at positions −32 and −28. The last two promoters (E. coli Co64 and Co210; cefoxitin MICs = 16 to 32 μg/ml) did not show mutations in the ampC promoter (Table 5). Mutations at position −32 or −42 were observed in four E. coli isolates recovered from three human samples and one broiler. The amplification of the ampC promoter was also performed with four additional E. coli isolates from our series, for which the cefoxitin MICs were 2 to 8 μg/ml (Table 5). Three point mutations (at positions −18, −1, and +58) were detected in two promoters (E. coli Co51 and Co61; cefoxitin MICs = 8 μg/ml). Two point mutations were detected (positions −28 and +17) in another ampC promoter (E. coli Co183; cefoxitin MIC = 2 μg/ml). No mutations were found in the last promoter (E. coli Co170; cefoxitin MIC = 2 μg/ml).

TABLE 5.

Mutations found in the promoter and/or attenuator region of the ampC genes of 14 E. coli isolates analyzed in this study for which cefoxitin MICs were different

E. coli isolate	Origin	Cefoxitin MIC (μg/ml)	bla gene detected	Positions of mutations at the ampC promotera
Co13	Food	>64	None	NDb
Co22	Food	>64	None	ND
Co23	Food	>64	None	ND
Co25	Food	>64	None	ND
Co100c	Broiler	64	None	−42, −18, −1, +58
A99c	Human patient	64	blaTEM	−42, −18, −1, +58
Co210	Human patient	32	blaTEM	None
Co64	Broiler	16	blaTEM	None
Co313	Healthy human	16	None	−32, −28
Co321	Healthy human	16	None	−32, −28
Co51	Food	8	blaOXA	−18, −1, +58
Co61	Broiler	8	blaOXA	−18, −1, +58
Co183	Dog	2	blaTEM	−28, +17
Co170d	Pig	2	None	None

^aMutations with respect to the E. coli K-12 sequence (24).

^bND, not determined. No PCR amplification was obtained with the primers for the ampC promoter.

^cThis isolate shows a diminished susceptibility to extended-spectrum cephalosporins.

^dThe ampicillin MIC for this isolate is 2 μg/ml.

The correlation between the β-lactam resistance phenotypes and the mechanism of resistance detected in our 124 AMP^r E. coli isolates are shown in Table 6. The following were the most frequent resistance patterns observed. Isolates of phenotype I were resistant to AMP and TIC but susceptible to the other β-lactams tested (48% of the isolates). Classic TEM and SHV β-lactamases were found in isolates of phenotype I, although a new TEM-type β-lactamase (TEM-95) was also detected in this group. Isolates of phenotype II were resistant to AMP and TIC and resistant or of intermediate susceptibility to AMC, in conjunction with susceptibility to the other β-lactams tested (41% of the isolates). The main mechanism of resistance found in isolates of phenotype II was the production of the TEM-1 β-lactamase, although different types of IRT enzymes (TEM-30, TEM-34, TEM-40, and TEM-51) were also detected. Six other phenotypes of β-lactam resistance were found in 12% of the isolates, and different mechanisms of resistance were demonstrated (Table 6). The two E. coli isolates from this study for which the ceftazidime and cefotaxime MICs were in the range of 4 to 8 μg/ml (diminished susceptibility) comprised the phenotype VI isolates. These isolates were recovered from a broiler (E. coli Co100) and from a human fecal sample (E. coli A99).

TABLE 6.

Phenotypes of β-lactam resistance and β-lactamases identified in 124 AMP^r E. coli isolates of different origins

Resistance phenotype	No. (%) of isolates	Phenotype of β-lactam resistancea								Mechanism detected (no. of isolates)
		AMP	TIC	AMC	FOX	CAZ	CTX	IMP	ATM
I	59 (48)	R	R	S	S	S	S	S	S	TEMb (39), TEM-1 (8), TEM-95 (1), SHVb (1), negativec (9), OXAb (1)
II	51 (41)	R	R	R, I	S	S	S	S	S	TEM-1 (42), TEMb (4), TEM-51 (1), TEM-30 (1), TEM-34 (1), TEM-40 (1), OXAb (1)
III	6 (4.8)	R	S	R, I	R, I	S	S	S	S	AmpCc,d (2), AmpCc,e (4)
IVf	2 (1.6)	R	R	I	R, I	S	S	S	S	TEM-1 (1), TEMb (1)
V	2 (1.6)	R	S	I	S	S	S	S	S	TEM-1 (1), OXAb (1)
VIg	2 (1.6)	R	R, I	R	R	S↓	S↓	S	S↓	TEM-1 + AmpCd (1), AmpCc,d (1)
VIIh	1 (0.8)	R	R	S	S	S↓	S↓	S	S	Negativec (1)
VIII	1 (0.8)	R	I	S	S	S	S	S	S	TEMb (1)

^aAMP, ampicillin; TIC, ticarcillin, AMC, amoxicillin-clavulanic acid; FOX, cefoxitin; CAZ, ceftazidime; CTX, cefotaxime; IMP, imipenem; ATM, aztreonam; S, susceptible, S↓, diminished susceptibility; I, intermediate susceptibility; R, resistant.

^bThis result was obtained by PCR but not by sequencing.

^cPCRs for TEM, SHV, and OXA were negative.

^dA point mutation was detected at either position −32 or −42 in the ampC promoter.

^ePutative mechanism of resistance.

^fOther mechanism of cefoxitin resistance could be involved, such as through a efflux pump or permeability.

^gFor the two E. coli isolates with this phenotype (Co100 and A99), cefotaxime, ceftazidime and aztreonam MICs were in the range of 4 to 8 μg/ml.

^hFor the single E. coli isolate with this phenotype (Co129), the cefotaxime and ceftazidime MICs were 2 μg/ml.

DISCUSSION

As shown in Table 6, almost half (48%) of the AMP^r E. coli isolates recovered from foods and fecal samples of humans and healthy animals showed a β-lactam resistance phenotype that included resistance only to aminopenicillins (AMP and TIC) and not to other β-lactams or β-lactamase inhibitors (phenotype I), and 41% of the isolates showed the AMP^r TIC^r AMC^i,r phenotype (phenotype II). Phenotype I (AMP^r TIC^r) was less frequently found (15%) in a study performed by other investigators with consecutive AMP^r E. coli isolates of human clinical origin (49), and a large group of the isolates in that series (62%) showed the AMP^r TIC^r AMC^r phenotype. The widespread use of β-lactams alone or in combination with β-lactamase inhibitors in hospitals could account for the higher frequency of phenotype II among clinical isolates.

It has been shown in this study that the most frequent mechanism of AMP resistance in E. coli isolates recovered from foods and feces of healthy animals and humans was by a TEM-type β-lactamase (83%). SHV- or OXA-type β-lactamases were detected in only a few isolates (3%). A bla_TEM-1 gene was found in 91% of the bla_TEM amplicons sequenced. We also found a number of molecular variants: bla_TEM-1a, bla_TEM-1b, bla_TEM-1c, and bla_TEM-1f. The most frequent variant among our isolates was bla_TEM-1b (69%). The first description of the bla_TEM-1b molecular variant was provided by Chen and Clowes (15), although it was called blaT1b. Goussard and Courvalin (20) completed the sequence of this variant. Recently, a large variety of other bla_TEM genes have been reported: bla_TEM-1c (21) and bla_TEM-1d, bla_TEM-1e, and bla_TEM-1f (28). We found bla_TEM-1c in E. coli isolates recovered from animals (one isolate from a pig and one isolate from a dog) and from healthy humans (four isolates). The bla_TEM-1c gene was previously described in a clinical E. coli isolate (21). In our study the bla_TEM-1f gene was found in only one nonpathogenic AMP^r E. coli isolate recovered from the feces of a human patient. The bla_TEM-1f gene was very recently described (29) in human clinical E. coli isolates. Very few studies on the molecular characterization of the bla_TEM genes in isolates of food or animal origin have been performed to date. The reason why some molecular variants of the bla_TEM genes are mainly found in human E. coli isolates and not in animal isolates could be linked to the use of different β-lactams in humans and animals or, probably, to a lack of studies with animal isolates. More extensive studies with animal E. coli isolates are required to obtain conclusive results.

The weak P3 promoter was found in all 20 bla_TEM-1 genes (molecular variants bla_TEM-1a, bla_TEM-1b, bla_TEM-1c, and bla_TEM-1f) analyzed for their promoter sequences. The P3 promoter corresponding to molecular variants bla_TEM-1a, bla_TEM-1b, bla_TEM-1c, bla_TEM-1d, and bla_TEM-1e has previously been found in E. coli isolates (15, 20, 21, 29, 47). Nevertheless, the strong P4 promoter was detected in the only studies in which an E. coli isolate containing the bla_TEM-1f gene was described (28, 29).

In our study, the bla genes encoding IRT-type β-lactamases were found in only four isolates recovered from one pig fecal sample (bla_TEM-51a [P3]) and three fecal samples from human patients (bla_TEM-30c [P3], bla_TEM-34b [Pa/Pb], and bla_TEM-40b). These IRT β-lactamases showed a single amino acid substitution at either position 69 (TEM-34 and TEM-40) or position 244 (TEM-30 and TEM-51). The bla_TEM-34 and bla_TEM-40 genes were described for the first time in 1994 and 1995, respectively, from clinical E. coli isolates (44, 52). Variants bla_TEM-34b (Pa/Pb) and bla_TEM-40b found in our study have previously been reported by Leflon-Guibout et al. (28) in clinical E. coli isolates. The TEM-30 β-lactamase was first described by Vedel et al. (50), and its sequence was published 2 years later (3). The bla_TEM-30c variant did not appear in the studies in which the bla_TEM-30 gene was reported by using the new nomenclature (28). The TEM-51 β-lactamase in an E. coli clinical isolate was described for the first time in 1997 by Bret et al. (7). In that paper the molecular variant was not mentioned, but from a comparison of its sequence with those of all the variants, it can be concluded that it corresponds to bla_TEM-51c or bla_TEM-51d. To our knowledge, our paper represents the first description of the variants bla_TEM-30c (P3) and bla_TEM-51a (P3). Leflon-Guibout et al. (28) reported that the IRT-encoding genes show strong promoters most of the time, whereas in our study two of the three IRT-encoding genes studied showed weak promoters and only one showed the Pa/Pb strong promoter (TEM-34).

The new bla_TEM-95b (P3) gene found in this study showed one mutation at position 635, which led to the change of the amino acid proline-145 to an alanine. The roles played by specific mutations in the action spectra of TEM β-lactamases were analyzed previously and described in different papers (4, 13, 27, 43, 48). The new bla_TEM-95b (P3) gene renders a β-lactam resistance phenotype similar to those of other isolates in this study with a TEM-1-type β-lactamase. In fact, the isolate was resistant only to aminopenicillins (AMP and TIC) but was susceptible to the other β-lactams tested. Amino acid 145 is positioned at the beginning of the 5α helix. This change could affect the tertiary structure, but according to our results, this substitution does not affect the inhibition profile. None of the important substitutions found in the IRT variants (amino acid positions 69, 165, 182, 244, 275, and 276) occurs at position 145 (4, 13, 27, 48). The purification of this new enzyme is in process in our laboratory.

For 10 of the E. coli isolates tested the cefoxitin MIC was ≥16 μg/ml. The hyperproduction of the AmpC chromosomal β-lactamase might be a possible mechanism of cefoxitin resistance. The mutations at positions −42 and −32 of the ampC promoter region were previously reported to be important in increasing the level of ampC transcription (11, 12, 26, 36, 37, 38). Among our isolates, four E. coli isolates for which the cefoxitin MICs were within the range of 16 to 64 μg/ml showed mutations at position −42 or −32. Mutations at both positions were not found together in the same isolate in our study, and similar results were documented previously (11, 37). Jaurin and Grundström (24) reported that a mutation at position +24 was important to increase the level of transcription due to the modification of the attenuator loop structure. This specific mutation was not found in our isolates. We detected three mutations (at positions −18, −1, and +58) in two isolates for which the cefoxitin MIC was 8 μg/ml. These three changes are frequent in resistant isolates but are usually combined with the mutation at position −42 (11, 37). As expected, no mutation was found in an E. coli isolate for which the cefoxitin MIC was 2 μg/ml. Nevertheless, in another isolate for which the MIC was the same, two mutations were identified (at positions −28 and +17). According to our results, it might seem that these mutations are not related to an increase in the level of ampC transcription. In some of our isolates with resistance or diminished susceptibility to cefoxitin and in which no mechanisms of cefoxitin resistance were identified (phenotype IV of Table 6), other mechanisms of resistance could be involved, such as the presence of plasmidic AmpC β-lactamases (2, 5, 18, 19, 32, 40, 51) and even the presence of efflux pumps or altered porins (33).

It is important to emphasize the spread of different variants of bla genes, including IRT β-lactamase-encoding genes, among nonpathogenic E. coli isolates recovered from foods or from the intestinal environments of humans and healthy animals. Further research should be carried out to study in depth the distribution and evolution of the bla genes in isolates from different ecosystems.