Integron-Associated Antibiotic Resistance and Phylogenetic Grouping of Escherichia coli Isolates from Healthy Subjects Free of Recent Antibiotic Exposure

Abstract

The study of integrons in 181 Escherichia coli isolates from three groups of healthy subjects who lived in communities and had not taken antibiotics for at least 1 month showed that the presence of integrons was associated with antibiotic resistance and phylogenetic grouping of the bacterial host and dependent on a subject's living environment.

Integrons play an important role in antibiotic resistance of clinical Escherichia coli strains because they are able to capture, integrate, and express gene cassettes encoding antibiotic resistance (11). The prevalence of integrons ranges from 22 to 59% in clinical E. coli (20, 26) and increases with the resistance of the isolates (17). Integrons are also present in resistant intestinal E. coli isolates from subjects living in a community (18). However, the subjects studied (18) were those admitted to a neurology ward and thus were, strictly speaking, not “perfectly healthy.” Resistance in intestinal E. coli strains is promoted by recent exposure to antibiotics (23), diet (7), deficient hygiene, poor living conditions (28), and living in developing countries (16). E. coli populations are structured in four major phylogenetic groups, A, B1, B2, and D (13), and isolates from the B2 phylogenetic group appear to be the least resistant to antibiotics (14, 24).

We took advantage of collections of intestinal E. coli isolates from healthy adults free of recent direct exposure to antibiotics to investigate the epidemiology of integrons and the relationship between their prevalence and the isolates' phylogenetic grouping.

We studied a collection of 181 E. coli isolates of known phylogenetic groups (9) and known antibiotic susceptibilities (2, 10), originating from three age- and sex-matched groups of healthy subjects living in communities (25 subjects/group, four to five isolates/subject): (i) Wayampi Amerindians living in isolation in southern French Guyana (WA), (ii) pig farmers (PF), and (iii) bank or insurance workers (BIW); the last two groups were from western mainland France. Subjects had not taken antibiotics for at least 1 month before the fecal specimen was obtained. When several isolates from the same subject displayed indistinguishable antibiotic susceptibility patterns and belonged to the same phylogenetic group, they were considered replicates and only one was selected at random. It was noted that the prevalence of group B2 in these isolates (27/181 [15%]) was lower than that in isolates from healthy women in Michigan (42/88 [48%]) (30), a difference that has been discussed elsewhere (9).

DNA was extracted from each isolate, and plasmid DNA from E. coli K-12 DH1 containing plasmid PSU 2056 (21), PVC 2554 (6), or PSMB 731 (1) carrying, respectively, a class 1, 2, or 3 integron was used as a control. The intI1, intI2, and intI3 genes were detected by triplex real-time PCR on an ABI Prism 7000 sodium dodecyl sulfate thermocycler (Applera, Courtaboeuf, France) using specific primer pairs (Table 1). Dissociation temperatures were 85, 80, and 90°C for the amplification products, corresponding to intI1, intI2, and intI3, respectively. All PCRs were performed in a volume of 25 μl containing 1× SYBR green PCR master mix (Applera) with 250 nM of each primer and 100 ng of DNA. A denaturation cycle of 10 min at 95°C was followed by 30 cycles of 15 s at 95°C and 60 s at 60°C. Each PCR series included positive controls in triplicate ranging from 0 to 10⁷ copies of the targets.

TABLE 1.

Primers used

Purpose	Primer	Target	Sequence (5′-3′)	Integron class	Reference
Real-time PCR	245 F	intI1	5′-TGAAAGGTCTGGTCATACATGTGA-3′	1	This work
	345 R	intI1	5′-CATTCCTGGCCGTGGTTCT-3′	1	This work
	312 F	intI2	5′-GCTAGGGCATTTAAAGCGATTTT-3′	2	This work
	412 R	intI2	5′-CAGACCATGGGCAGTGAAGA-3′	2	This work
	381 F	intI3	5′-TGCGCTCCAGTGCATGAG-3′	3	This work
	528 R	intI3	5′-GGCAAGGGCGACAAGGA-3′	3	This work
Standard PCR and sequencing	L1	5′Csa	5′-GGCATCCAAGCAGCAAG-3′	1	20
	R1	3′Csb	5′-AAGCAGACTTGACCTGA-3′	1	20
	Sul-833F	3′Csb	5′-TGGTGACGGTGTTCGGCATTC-3′	1	This work
	Sul-2602R	3′Csb	5′-GCGAAGGTTTCCGAGAAGGTG-3′	1	This work
	986R	5′Csc	5′-GTAGCAAACGAGTGACGAAATG-3′	2	3
	2021R	aadA	5′-TCTTCCAACTGATCTGCGGGC-3′	2	5
	1299F	aadA	5′-ATGAGGGAAGCGGTGATCGCC-3′	2	5
	TnsE-Tn7	3′Csd	5′-GAATTCGACATGTTTGGACGCCTTGGC-3′	2	3
	aadA1-283R	aadA1	5′-ATAACGCCACGGAATGATGTC-3′	1	This work
	dfr1-453F	dfr1	5′-CCAAATCTGGCAAAAGGGTTAA-3′	1	This work
	aadA2-283R	aadA2	5′-ATAACGCCACGGAATGATGTC-3′	1	This work
	dfr12-463F	dfr12	5′-TACACCCACTCCGTTTATGCG-3′	1	This work

^a5′ conserved class 1 integron segment.

^b3′ conserved class 1 integron segment.

^c5′ conserved class 2 integron segment.

^d3′ conserved class 2 integron segment.

To characterize the gene cassettes, a PCR was performed on isolates containing intI1 using primer pair L1/R1 and two amplifications on isolates containing intI2 using primer pair 986R/2021R and primer pair 1299F/TnsE-Tn7 (Table 1). If the PCR using L1/R1 was negative, amplifications were performed using primer pairs Sul-833F/Sul-2602R and L1/TnsE-Tn7 (Table 1) to detect two targets. PCR products were sequenced using the primers described in Table 1 and the ABI Prism sequencing kit (Applera).

Statistical analysis was done with SAS v.8.02 software (SAS Institute, Cary, NC) and with nQuery Advisor v.5.0 (Statistical Solutions, Boston, MA). Differences in the prevalences of integrons between phylogenetic groups and between groups according to antibiotic resistance were assessed using chi-square testing.

Twenty-seven isolates (15%) were integrase positive: 20 (11%) were positive for class 1 integrase, 6 (3%) for class 2, 1 (0.6%) for both classes 1 and 2, and none for class 3. All isolates harboring an integron were resistant to at least one antibiotic.

Integrons were more prevalent in isolates from PF than in those from BIW or WA (P < 0.05), and when each antibiotic was considered, integrons were more prevalent in the isolates resistant to each antibiotic than in the susceptible ones (Table 2). Isolates susceptible to all antibiotics were significantly more prevalent in the B2 phylogenetic group (15/27) than in the non-B2 groups (54/154) (P < 0.05). A single B2 isolate carried an integron. There was a trend towards lower integron prevalence among group B2 (Table 2) (P = 0.08).

TABLE 2.

Prevalence of intI1 and intI2 genes in commensal E. coli isolates

Characteristic	No. of isolates	No. (%) of isolates with indicated gene				P value
		None	Any	intI1	intI2
Source
Pig farmers	54	41 (75.9)	13 (24.1)a	10 (18.5)	4 (7.4)	<0.05
Bank/insurance workers	49	42 (85.7)	8 (16.3)	5 (12.2)	2 (4.1)
Amerindians	78	72 (92.3)	6 (7.7)	5 (6.4)	1 (1.3)
Resistance
Streptomycin
Susceptible	119	117 (98.3)	2 (1.7)	2 (1.6)	0 (0)	<0.001
Resistant	62	37 (59.7)	25 (40.3)a	19 (30.6)	7 (11.2)
Ampicillin
Susceptible	140	129 (92.1)	11 (7.9)a	7 (5)	5 (3.6)	<0.0001
Resistant	41	25 (61)	16 (39)	14 (34.2)	2 (4.9)
Co-trimoxazole
Susceptible	149	140 (94)	9 (6)	6 (4)	3 (2)	<0.0001
Resistant	32	14 (43.8)	18 (56.2)a	15 (46.8)	4 (12.5)
Tetracycline
Susceptible	103	98 (95.2)	5 (4.8)	1 (0.9)	4 (3.9)	<0.0001
Resistant	78	56 (71.8)	22 (28.2)a	20 (25.6)	3 (3.8)
Chloramphenicol
Susceptible	140	125 (89.3)	15 (10.7)	11 (7.9)	4 (2.9)	<0.01
Resistant	41	29 (70.7)	12 (29.3)a	10 (24.3)	3 (8.5)
Nalidixic acid
Susceptible	177	153 (86.4)	24 (13.0)a	19 (10.7)	6 (3.3)	<0.05
Resistant	4	1 (25)	3 (75)	2 (50)	1 (25)
Pefloxacin
Susceptible	177	153 (86.4)	24 (13.0)a	19 (10.7)	6 (3.3)	<0.05
Resistant	4	1 (25)	3 (75)	2 (50)	1 (25)
Phylogenetic groupb
A	74	62 (83.8)	12 (16.2)a	11 (14.8)	2 (2.7)
B1	41	34 (82.9)	7 (17.1)	5 (12.1)	2 (4.8)
B2	27	26 (96.3)	1 (3.7)	1 (3.7)	0	NSc (0.08)
D	39	32 (82.1)	7 (17.9)	4 (10.2)	3 (7.7)

^aIsolates with class 1 and class 2 integrons.

^bDetermined as described in reference 4.

^cNS, not significant (B2 versus non-B2 values).

Eighteen of 21 (85.7%) class 1 integrons carried either dfr (2/21 [9.5%]), aadA (8/21 [38.1%]), or both cassettes (8/21 [38.1%]). The remaining 3/21 (14.3%) integrons were resistant to both streptomycin and co-trimoxazole but probably carried truncated integrons lacking 3′-end-conserved segments since no class 1 or class 2 integron 3′-end-conserved regions were detected. All cassette sequences were identical to known GenBank sequences, except one that was closely related (96% identity) to aadA2. Four of the seven isolates containing class 2 integrons had dfr-1, sat, aadA, and orfX genes, in that order and were identical to the integron in the Tn7 transposon; two had sat, aadA, and orfX genes and were identical to the integron in Tn1826 (3). The remaining isolate contained two class 2 integrons, with three and four gene cassettes, respectively.

These results show that integrons can persist in commensal E. coli isolates, even in subjects who have not taken antibiotics for at least 1 month, a factor not analyzed previously (18). Antibiotic exposure is probably the major determinant of increased integron prevalence, as suggested by the prevalence of 22 to 59% in E. coli isolates from hospitalized patients (20, 26) who are more often exposed to antibiotics than subjects living in communities. Here, antibiotic exposure in the general population from which the included subjects were extracted was about twice as high in BIW as in WA (2, 10). This might explain why the prevalence of integrons was also twice as high in BIW as in WA (16.3% versus 7.7%) (Table 2). The observed differences in the integron prevalences between PF and BIW (Table 2) could be explained by differences in animal exposure. Indeed, PF and BIW were matched for age, gender, and country of residence, and their direct levels of antibiotic exposure were similar (2), but PF worked with animals receiving therapeutic and preventive antibiotics as well as antibiotics for growth promotion (8). As a result, it may be that the E. coli strains from swine harbored integrons and that these E. coli organisms were then transferred to the farmers, as has been demonstrated for enterobacteria and enterococci (19, 27).

As expected, class 1 integrons were more frequently observed than class 2 integrons (22). The rarely described class 3 integron was absent. All cassette sequences were identical to those of the most frequently described cassettes (20, 25), and integron structures were also as described previously (3, 12). We found only three unexpected structures corresponding to truncated class 1 integrons. Their role remains unclear. This low diversity of integrons and of their cassettes in commensal E. coli isolates from healthy subjects contrasts with the diversity previously found in clinical isolates (15, 22, 29) and may be due to the absence of direct recent antibiotic exposure of the source subjects. The dfr and aadA gene cassettes appeared stable, once inserted in integrons and even in the absence of selective pressure, for at least 1 month. This has not been tested for longer periods.

Finally, the phylogenetic group analysis may have revealed a new constraint to the dissemination of resistance. It has been suggested that clinical B2 E. coli strains were less resistant to antibiotics than non-B2 strains (14, 24). This was also true for commensal isolates in which a trend (P = 0.08) suggested that integrons were less frequent in B2 than non-B2 isolates, raising the hypothesis that the two phenomena are interrelated.