Phylogenetic Groups and Pathogenicity Island Markers in Fecal Escherichia coli Isolates from Asymptomatic Humans in China

Abstract

The study of phylogenetic groups and pathogenicity island (PAI) markers in commensal Escherichia coli strains from asymptomatic Chinese people showed that group A strains are the most common and that nearly half of all fecal strains which were randomly selected harbor PAIs.

Escherichia coli is a well-diversified commensal species in the intestine of healthy humans but also includes intestinal or extraintestinal pathogens. It has been reported that pathogenic E. coli may be derived from fecal strains by acquisition of virulence determinants (11). The relationship between the E. coli genetic background and the acquisition of virulence factors is now better understood (1, 5). Extraintestinal E. coli strains may harbor several virulence factors, such as adhesins, fimbriae, and hemolysin, which can contribute to bacterial pathogenesis. These traits are usually encoded on pathogenicity islands (PAIs), which have been studied in pathogenic E. coli previously (15). The E. coli population includes 4 major phylogroups (A, B₁, B₂, and D) (2). Pathogenic strains belong mainly to groups B₂ and D, while most fecal isolates belong to groups A and B₁. Strains of groups B₂ and D often carry virulence factors that are lacking in group A and B₁ strains (3, 9, 13).

In this study, we examined the distribution of phylogroups and the prevalence of PAIs in commensal E. coli strains isolated from asymptomatic persons in one region of China.

Bacterial strains.

The asymptomatic individuals (174 males and 151 females, with an age range from 18 to 75 years) were recruited from those who underwent annual personal physical examination in one hospital in Fuzhou, China, from February to May 2009. All had no confirmed diagnosis of digestive tract diseases. No information on antibiotic history was available. All of the participants gave their informed consent. Samples were cultured on MacConkey agar plates directly. One E. coli isolate was collected per person. All isolates were identified by biochemical methods (indole-methyl red-Voges-Proskauer-citrate [IMViC] tests and urease production, H₂S production, and various sugar fermentation tests).

Phylogroup analysis.

Phylogroups were examined by a PCR-based method (2). The results showed a slightly greater number of strains in group A than in the other three groups (Table 1). Statistical analysis demonstrated that sex and age factors had no effect on distribution of phylogroups (data not shown). Only two studies showed concern about the influence of these two factors on distribution of phylogroups, but they did not reach a consensus (4, 6). Further studies are required to better describe the disparity.

TABLE 1.

Prevalence of the four major E. coli phylogenetic groups in fecal samples from different human populations^a

Population	No. of samples	Prevalence (%) of phylogenetic group:
		A	B1	B2	D
French people
Paris area residents Ib	56	61.0	12.5	10.5	16.0
Paris area residents IIc	27	29.6	11.1	37.1	22.2
Brittany (BIW)e residents	25	24.0	24.0	32.0	20.0
Brittany (PF)f residents	25	32.0	28.0	16.0	24.0
Brest University students	21	14.3	23.8	33.3	28.6
Tours residents	24	25.0	21.0	29.0	25.0
Michigan residents	88	20.5	12.5	47.7	19.3
Tokyo, Japan, residents	61	28.0	0.0	44.0	28.0
Bogota, Colombia, residents	28	57.1	3.6	14.3	25.0
Cotonou residents	46	50.0	32.6	17.4	0.0
Guyana Amerindians	93	63.4	20.4	3.2	12.9
Malian people	55	23.6	58.2	1.8	16.4
Croatian people	57	35.1	31.6	19.3	14.0
Australian people	266	19.5	12.4	45.1	22.9
Korean people	141	29.8	34.0	0.0	36.2
Chinese peopled	325	43.7	23.4	16.0	16.9

^aData from studies by Duriez et al. (3), Escobar-Paramo et al. (4), Gordon et al. (6), and Unno et al. (18) and from this study.

^bData from study by Duriez et al. (3).

^cData from study by Escobar-Paramo et al. (4).

^dData from this study.

^eBIW, bank and insurance workers.

^fPF, pig farmers.

It was reported that geographic and climate factors might affect the distribution of phylogroups of commensal E. coli (4). The prevalence of group A isolates among persons in temperate regions was half that among persons in tropical regions in one study (4). In contrast, such strains predominated in this study, though China lies in the temperate belt. However, it would be prudent to analyze the difference between these studies, because no background data of hosts, such as genetic factors, diet, etc., were included in any of this research. As discussed by Duriez et al. (3), this difference may result from factors related to the external environment or may be due to cultural differences in diet or food processing and preparation practices.

Notably, certain phylogroups were absent in certain populations, such as B₂ types in South Korea (18), B₁ types in Japan, and D types in Cotonou (Benin, Africa) (Table 1). These data suggested a strong genetic influence on the phylogroup distribution in commensal E. coli strains. Curiously, the levels of prevalence of group A strains were quite different in Paris, France, in different studies (Table 1). As discussed by Escobar-Paramo et al. (4), this was probably the result of different dietary habits and hygiene factors.

Factors such as those discussed above may influence the distribution of phylogroups in fecal isolates in humans. It would be of interest to identify strains with genotypes that may possess particular characteristics which could help them to survive in and adapt to various environments. Moreover, since groups A and B₁ are the most prevalent among E. coli strains in the environment (19), it is possible that human populations that acquire more bacteria from the environment will also have an overabundance of these types of E. coli. Further study is needed to confirm these hypotheses.

However, the Clermont method is not perfect in classifying E. coli strains into phylogroups; in particular, group A strains (with genotype “−−−” [i.e., lacking chuA, yjaA, and Tspe4.C2]) are seldom classified correctly (7). In this study, the majority of strains belonged to group A, and 9.5% (31/325) of those were the “−−−” genotype. To validate our results, PCR with E. coli-specific lacZ primers was performed (8), and all of the strains generated the predicted amplicon. Although the strains in this study were relatively rare, as reported by other researchers (7), they should be characterized further using a multilocus sequence typing method.

Distribution of PAIs.

PAIs have been investigated widely in pathogenic bacteria, but little attention has been paid to commensal strains. In this study, PAI markers were detected by PCR, as described previously, for 220 E. coli strains randomly selected from each group (14). Overall, 46.8% (103/220) of these strains carried PAIs, which was a little higher than that determined in Spain (14), and 161 PAIs were detected in total. PAI IV₅₃₆ was the most common one (38.2%), followed by PAI I_CFT073 (20.9%), PAI II_CFT073 (10.9%), PAI II₅₃₆ (1.8%), and PAI I₅₃₆ (1.3%), respectively. Meanwhile, three PAIs (III₅₃₆, I_J96, and II_J96) were not detected.

PAI IV₅₃₆ was reported to be the most ubiquitous PAI in Enterobacteriaceae (16, 17) and is supposed to be a fitness island rather than a pathogenic one (10, 12). Interestingly, it was previously always detected together with PAI I_CFT073 (14). However, PAI IV₅₃₆ was frequently detected alone in this study.

Among all of the strains, 20.5% (45/220) had multiple PAIs, which was higher than the results for strains from another nation (Table 2). The percentages of fecal E. coli strains with PAIs and with multiple PAIs were higher in group B₂ than in the other three groups (Table 3). Because data on the prevalence of PAIs in human intestinal flora are so far quite limited, the difference may be of limited significance from the pathogenicity standpoint. The majority of group B₂ strains had more than two PAIs, which revealed that these strains were highly virulent even in the human gut.

TABLE 2.

Number of pathogenicity islands (PAIs) detected in 220 commensal Escherichia coli isolates, classified according to phylogenetic group

Phylogenetic group	Sourcea	No. (%) of isolates with the following number of PAIs detected:
		0	1	2	3	4	5	6	7
A	E	73 (73)	21 (21)	5 (5)	1 (1)
	O	7 (70)	3 (30)
B1	E	35 (53)	20 (31)	4 (6)	5 (8)	1 (2)
	O	14 (82)	3 (18)
B2	E	5 (17)	4 (14)	11 (38)	7 (24)	1 (3)	1 (3)
	O			2 (22)	3 (34)		2 (22)	2 (22)
D	E	4 (15)	13 (50)	9 (35)
	O	9 (64)	3 (22)	2 (14)

^aE, our work; O, another work (14).

TABLE 3.

Distribution of pathogenicity islands (PAIs) among 220 commensal isolates of Escherichia coli, classified according to phylogenetic group

Phylogenetic group	% of isolates (no./total tested) with:		No. of PAIs
	Single PAI	Multiple PAIs	I536	II536	III536	IV536	IJ96	IIJ96	ICFT073	IICFT073
A	27 (27/100)	6 (6/100)	0	1	0	23	0	0	5	6
B1	46 (30/65)	15 (10/65)	0	1	0	28	0	0	10	8
B2	83 (24/29)	69 (20/29)	3	2	0	22	0	0	21	8
D	54 (14/26)	35 (9/26)	0	0	0	11	0	0	10	2

In summary, to our knowledge, this is the first report to analyze the phylogroups of fecal E. coli from asymptomatic humans in China. The results indicate that commensal E. coli strains from group A predominate in the gut flora. Moreover, some fecal E. coli strains appeared to be potentially virulent.