Antimicrobial susceptibility, phylotypes, and virulence genes of Escherichia coli from clinical bovine mastitis in five provinces of China

ABSTRACT

The objectives of this study were to determine the antimicrobial susceptibility, phylotype, and virulence genes of Escherichia coli (E. coli) from cows with clinical mastitis. A total of 497 mastitis samples were collected and 92 E. coli isolates were identified. E. coli isolates were highly resistance to ampicillin (41.3%), followed by piperacillin (36.9%) and tetracycline (34.7%). Five intimin types, namely μR-ι2, λ, ξR/β2, νR-ε2, η, were observed among 75 EPEC strains with eae gene, with all the intimins were first detected in bovine mastitis milk. And 65 E. coli strains (70.7%) belonged to seven different O serotypes (O121, O91, O22, O26, O128, O111, and O113). For the phylogroup assignment, only 87 E. coli isolates could be assigned into phylogroup B1, A, C, and E. This study suggested the prevailing among EPEC strains isolated from mastitis in five provinces was important to understand the etiology of E. coli.

KEYWORDS:

• Escherichia coli
• bovine mastitis
• antibiotic resistance
• phylotype

1. Introduction

Bovine mastitis, which affects dairy farms throughout the world, reduces milk production, negatively impacts milk quality, and represents a source of contamination for raw milk and dairy products (Halasa, Huijps, Østerås, & Hogeveen, 2007). Several bacteria cause mastitis in dairy cows, including Escherichia coli (Olde Riekerink, Barkema, Scholl, Poole, & Kelton, 2010), one of the most predominant pathogens(Zhang et al., 2018), Staphylococcus aureus (Olde Riekerink, Barkema, Kelton, & Scholl, 2008), and Streptococcus uberis(Levison et al., 2016). E. coli, which has an isolation rate of 10.2% in bovine mastitis (Tenhagen, Hansen, Reinecke, Heuwieser, & Lam, 2009), can be classified into pathogenic and nonpathogenic bacteria, and pathogenic bacteria can be further classified into different types based on the pathogenic mechanism (Blum & Leitner, 2013)and virulence factor, such as enteropathogenic E. coli (EPEC), enterotoxigenic E. coli (ETEC), Shiga toxin-producing E. coli (STEC or VTEC), and enterohemorrhagic E. coli (EHEC (Fitzhenry et al., 2002; Nagy & Fekete, 1999)). STEC strains play important roles in mastitis (Güler & Gündüz, 2007; Kobori, Rigobelo, Macedo, Marin, & Avila, 2004) and EPEC strains are predominant in diarrheal patients, because different geographical location might influence the pathogenic molecular markers (Blanco, Blanco, Blanco, et al., 2004; Suardana, Artama, Asmara, & Dayono, 2010).

Bovine mastitis can be controlled and/or prevented by antimicrobial agents. Unfortunately, the misuse of antimicrobials has contributed to the emergence of antimicrobial-resistant bacteria (Mia et al., 2017; Van Boeckel et al., 2015; Zhang, Ying, Pan, Liu, & Zhao, 2015). For example, 20% to 33% of E. coli isolates from bovine mastitis were resistant to at least one antimicrobial agent (Fairbrother et al., 2015; Suojala et al., 2011), and 87.1% of E. coli isolates (61/70) are resistant to ampicillin and kanamycin according to a study performed in Beijing, China (Liu et al., 2014). Multidrug-resistant E. coli strains, which represent a significant public concern (Erb, Sturmer, Marre, & Brenner, 2007; Johnson et al., 2008), have been isolated from animals in several countries including China (Rao et al., 2014; Seni et al., 2016; Xu, An, Wang, & Zhang, 2015). Additionally, several virulence factors associated with E. coli infections and various pathogenic bacterial genes have been isolated from mastitis in dairy cows (Zhang et al., 2018). Virulence genes identified in E. coli from bovine mastitis include genes that code for hemagglutinin, aerobactin, P-fimbria, enterohemolysin, intimin, and autoagglutinating adhesion proteins (Bekal et al., 2003; Ewers et al., 2007; Johnson et al., 2008). The antimicrobial resistance and virulence genes of E. coli from bovine mastitis have been studied in Korea, Jordan, Iran, and Spain (Mora et al., 2005; Obaidat, Bani Salman, Davis, & Roess, 2018; Tark et al., 2017; Tavakoli & Pourtaghi, 2017; Zhao et al., 2018). A more thorough understanding of the resistance and virulence factors of E. coli isolates will assist in the implementation of appropriate treatments for the management of bovine mastitis.

Virulence factors may be associated with phylogeny groups (Garcia-Aljaro, Moreno, Andreu, Prats, & Blanch, 2009). Furthermore, the determination of phylogroups (Clermont et al., 2008), intimin typing (Blanco et al., 2005), and O serotyping (Iguchi et al., 2015) of E. coli can be used for pointed potential health risks. Researchers have screened virulence genes in E. coli isolated from poultry in Canada with O serotypes, unknown intimin types, and phylogroups (Afset et al., 2006). However, few studies have evaluated virulence factors and typing in mastitic cows. Therefore, the objectives of this study were to assess (i) antibiotic resistance of E. coli isolates, (ii) virulence gene profiles, (iii) phylogroups, (iv) O serogroups, and (v) intimin typing.

2. Material and methods

2.1. Sample collection

Five provinces, including Hebei, Shandong, Hei Longjiang, Shanghai, and Inner Mongolia were selected as the sampling points (2017), which covered four geographical locations in China. Four types of farms were selected based on the population of cows in each sampling point. Among the selected farms, a total of 497 raw milk samples were collected twice from cows with clinical mastitis in 2017 (May and September): 100 samples from Hebei province, 100 samples from Inner Mongolia, 100 samples from Shanghai city, 100 samples from Heilongjiang province, and 97 samples from Shandong province. We selected two specific months which represent summer and fall for collection of mastitis milks, because the average temperature of the five sampled provinces and cities we selected in May and September is different. Therefore, we plans to analyze whether there was an impact on E. coli resistance in bovine mastitis. All samples were transferred into sterile plastic bottles (Corning, New York, USA) and transported to the laboratory at 4°C within 4 h.

2.2. Isolation and identification of E. coli

The collected samples (25 mL) were homogenized in 225 mL of modified E. coli broth. Aliquots (1.0 mL) of selected dilutions were streak-inoculated in MacConkey agar plates (Huankai, Guangdong, China) and incubated for 18–24 h at 36 ± 1°C. Presumptive colonies were selected from the nutrient agar and placed in prepared sterile saline, resulting in a 0.5 McFarland standard. Aliquots (1.0 mL) of the bacterial suspension were pipetted into a micro-biochemical tube of the IMVC biochemical identification kit (Huankai, Guangdong, China) and incubated at 36°C. The detected E. coli was stored in a seed preservation tube containing 20%–30% glycerol in brain-heart extract broth and stored in the refrigerator at −80°C.

The molecular identification of isolated strains was conducted by PCR (Bio-Rad S1000, Bio-Rad Laboratories, Hercules, CA, USA) of 16s rDNA and rpoB (Dahllof, Baillie, & Kjelleberg, 2000). PCR samples were sequenced (HuaDa, Beijing, China) and compared using the BLAST programme at the National Center for Biotechnology Information website (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The PCR reaction mixture (50 μL) consisted of 25 μL EmeraldAmp Max PCR Master Mix (Takara, Dalian, China), 19 μL dH₂O (Takara, Dalian, China), 2 μL of each primer (100 μM; Sangon Biotech Shanghai Co., Ltd., Shanghai, China), and approximately 2 μL of bacterial genomic DNA (Supplemental Table S1). A negative control (without DNA template) and a positive control (E. coli ATCC 25922 template) were included in all PCR assays.

2.3. Antimicrobial susceptibility test

Identified E. coli isolates were tested for phenotypic susceptibility to antimicrobials using the MIC broth microdilution method (Romney, 2018). The drugs included ampicillin (4−16 μg/mL), amoxicillin-clavulanic acid (4/2–16/8 μg/mL), cefoxitin (1–16 μg/mL), tetracycline (2–8 μg/mL), chloramphenicol (4−16 μg/mL), gentamicin (2–8 μg/mL), trimethoprim/sulfamethoxazole (9.5–38 μg/mL), piperacillin (4–64 μg/mL), ciprofloxacin (0.5−2 μg/mL), aztreonam (2−16 μg/mL), meropenem (1–8 μg/mL), ofloxacin (1−8 μg/mL), and kanamycin (4–16 μg/mL). E. coli ATCC 25922 was used as a reference strain. Antimicrobial resistance was defined as non-susceptible to one or more antimicrobials, whereas multidrug resistance (MDR) was defined as non-susceptible to three and more classes of antimicrobials.

2.4. Detection of antimicrobial resistance genes

Phenotypically resistant E. coli isolates were characterized at the molecular level for their antimicrobial resistance mechanisms. PCR was performed as previously reported for the detection of drug-resistant genes associated with β-lactams (ampicillin, bla_TEM , bla_CTX-M, bla_NDM , bla_OXA and bla_KPC), including cefoxitin (bla_FOX). Amino acid changes in gyrA and gyrB proteins were studied by PCR and sequencing of the corresponding genes in quinolone-resistant isolates. Additionally, aminoglycoside (aac3, aacA4, and aac6-I), chloramphenicol (florR), and tetracycline (tetA and tetB) resistance genes were analyzed by PCR and sequenced. The amplified products were analyzed by electrophoresis in a 1.5% agarose gel stained with SYBR® Safe DNA gel stain (ThermoFisher Scientific, USA) and visualized using VersaDoc (Bio-Rad, Hercules, California, USA). The PCR reaction mixture (25 μL) consisted of 12.5 μL EmeraldAmp Max PCR Master Mix (Takara, Dalian, China), 9.5 μL dH₂O (Takara, Dalian, China), 1.0 μL of each primer (100 μM; Sangon Biotech Shanghai Co., Ltd., Shanghai, China), and approximately 1.0 μL of bacterial genomic DNA (Supplemental Table S1).

2.4.1. PCR for virulence genes

Virulence genes including eae, aggR, esp, elt, ipaH, stx1, and stx2 were amplified by PCR (Bio-Rad). The primer sequences of all virulence genes used for PCR and the cycling conditions are shown in Supplementary Table S2. The PCR reaction mixture (25 μL) consisted of 12.5 μL EmeraldAmp Max PCR Master Mix (Takara, Dalian, China), 9.5 μL dH₂O (Takara, Dalian, China), 1.0 μL of each primer (100 μM; Sangon Biotech Shanghai Co., Ltd., Shanghai, China), and approximately 1.0 μL of bacterial genomic DNA.

2.5. Typing methods

2.5.1. Phylogenetic grouping

The distributions of E. coli isolates in phylogroups were determined according to the phylogenetic grouping method for quadruplex PCR (Clermont et al., 2013a). The phylogenetic grouping method can assign E. coli strains into eight phylogroups (A, B1, B2, C, D, E, F, and clade-1). The data were used to adjust the primer sequences (chuA, yjaA, TspE4.C2, and arpA), and the sizes of PCR products were used for the quadruplex phylo-typing method. Analyses were performed using positive and negative controls. Primer sequences and cycling conditions for the phylo-typing method are shown in Supplemental Table S3.

2.5.2. Intimin typing

Complementary primers were used to identify 17 eae genes and intimin types by PCR. The primer sequences of intimin subtypes (α1, α2, β1, ξR/β2, δ/κ/β2O, γ1, θ/γ2, ε1, νR/ε2, ζ, η, ι1, μR/ι2, λ, μB, νB, and ξB) are shown in Supplemental Table S4 (Adu-Bobie et al., 1998; Blanco et al., 2004d; Blanco et al., 2005). The nucleotides sequence of the amplification products purified with a DNA purification kit (Qiagen) was determined by the deoxynucleotide triphosphate chain terminated method of Sanger using a BigDye Terminator v3.1 Cycle Sequencing kit and an ABI 3100 Genetic Analyzer (Applied Bio-Systems).

2.5.3. O-serogroups genes detection

The O-serotype genes of isolates were identified by PCR using seven pairs of primers to determine serogroups of E. coli (Supplemental Table S5).

2.6. Statistical analyses

Statistical analyses for each of the antimicrobial agents and resistance genes were performed. For phenotypic and genotypic analyses, data were analyzed by one-way ANOVA using SPSS (ver. 19.0, IBM Corp., Armonk, NY, USA). The results were presented as mean ± standard deviation. Paired t-tests were used for comparisons between groups. Statistical significance was set at P < 0.05.

3. Results

3.1. E. coli identification

There were 92 E. coli isolates in 497 raw milk samples. Among the isolates, 12 (13.0%) were from Shandong province, 18 (19.0%) were from Heilongjiang province, 19 (21.0%) were from Inner Mongolia, 19 (21.0%) were from Hebei province, and 24 (26.0%) were from Shanghai.

3.2. Antimicrobial susceptibility testing

The MIC broth microdilution method was used, and the interpretation of antibiotic breakpoint was carried out according to CLSI guidelines (2018). A total of 13 drugs were tested in this study, and 48 of the 92 E. coli isolates were non-susceptible. The isolates were resistant to ampicillin (38/92, 41.3%), piperacillin (34/92, 36.9%), tetracycline (32/92, 34.8%), trimethoprim/sulfamethoxazole (31/92, 33.7%), chloramphenicol (23/92, 28.3%), aztreonam (18/92, 19.6%), gentamicin (17/92, 18.5%), ofloxacin (12/92, 13.0%), ciprofloxacin (8/92, 8.7%), cefoxitin (6/92, 6.5%), and amoxicillin-clavulanic acid (4/92, 4.3%). None of the isolates were resistant to meropenem or kanamycin (Table 1). The 48 isolates resistant to one or more antibiotics presented 37 different resistance patterns (Table 2). MDR was observed in 32/92 (34.8%) of the isolates, and two of the isolates were resistant to eight classes of antibiotics (Figure 1). Furthermore, the level of antimicrobial resistance was higher in phylogroup B1 than in group A (Table 3).

Figure 1. Antibiotic resistance of E. coli isolates from bovine mastitis in five major dairy-producing regions of China.

Table 1. Antibiotic resistance (%) of E. coli isolates from raw milk of mastitic cows in China.

Antibiotics	No. (%) of positive strains
	Shanghai	Inner Mongolia	Hebei	Heilongjiang	Shandong	Total
	(n = 24)	(n = 19)	(n = 19)	(n = 18)	(n = 12)	(n = 92)
AMP^a	15 (62.5)	4 (21.0)	7 (36.8)	8 (44.4)	4 (33.3)	38 (41.3)
PIP^a	11 (45.8)	3 (15.8)	6 (31.6)	10 (55.6)	4 (33.3)	34 (36.9)
TET^ab	12 (50.0)	4 (21.1)	6 (31.6)	6 (33.3)	4 (33.3)	32 (34.8)
SXT^ab	10 (41.7)	4 (21.1)	7 (36.8)	9 (50.0)	1 (8.3)	31 (33.7)
CHI	9 (37.5%)	4 (21.1)	2 (10.5)	4 (22.2)	4 (33.3)	23 (28.3)
ATM	3 (12.5)	1 (5.3)	5 (26.3)	5 (27.8)	4 (33.3)	18 (19.6)
GEN	9 (37.5)	1 (5.3)	2 (10.5)	3 (16.7)	2 (16.7)	17 (18.5)
OFLX^bd	4 (16.7)	1 (5.3)	2 (10.5)	2 (11.1)	3 (25.0)	12 (13.0)
CIP^bc	3 (12.5)	0 (0)	2 (10.5)	1 (5.5)	2 (16.7)	8 (8.7)
FOX	2 (8.3)	0 (0)	1 (5.3)	2 (11.1)	1 (8.3)	6 (6.5)
AMC^cd	1 (4.2)	2 (10.5)	1 (5.3)	0 (0)	0 (0)	4 (4.3)
MEM^e	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
KAN^e	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)

Note: No. (%): (numbers of positive strains)/n. Antimicrobials: AMP, ampicillin; FOX, cefoxitin; AMC, amoxicillin-clavulanic acid; SXT, sulfamethoxazole-trimethoprim; PIP: piperacillin; CIP, ciprofloxacin; ATM: aztreonam; MEM: meropenem; OFLX: ofloxacin; CHL, chloramphenicol; KAN, kanamycin; GEN, gentamicin; TET, tetracycline. (P ≤ 0.05).

Table 2. Antimicrobial resistance patterns of E. coli isolates.

Resistance pattern	No. of isolates
AMP	1
AMP + PIP	2
AMP + TET	1
AMP + SXT	1
OFLX + SXT	1
SXT + PIP	1
TET	1
PIP	3
AMC	1
OFLX	1
SXT	1
AMP + CIP + TET	1
AMP + PIP_ATM	1
AMP + AMC + TET	1
AMP + SXT + PIP	1
AMP + SXT + PIP + ATM	1
PIP + ATM + OFLX + CHI	1
AMP + SXT + CHI + TET	1
AMP + FOX + CHI + TET	1
AMP + SXT + PIP + GEN + TET	1
AMP + SXT + PIP + CHI + TET	2
AMP + SXT + CHI + GEN + TET	2
AMP + PIP + ATM + CHI + TET	1
AMP + SXT + ATM + GEN + TET	1
AMP + SXT + ATM + CHI + TET	1
AMP + SXT + PIP + ATM + TET	1
AMP + SXT + PIP + CHI++GEN + TET	2
AMP + SXT + PIP + CHI++ATM + TET	1
AMP + OFLX + PIP + CIP + ATM + SXT + TET	2
AMP + OFLX + PIP + CHI + ATM + SXT + TET	1
AMP + SXT + PIP + ATM + CHI + GEN + TET	4
AMP + AMC + FOX + AXT + PIP + CHI + GEN + TET	1
AMP + SXT + PIP + CIP + OFLX + CHI + GEN + TET	1
AMP + FOX + SXT + PIP + ATM + OFLX + CHI + GEN + TET	1
AMP + FOX + AMC + SXT + PIP + CIP + ATM + GEN + TET	1
AMP + SXT + PIP + CIP + ATM + OFLX + CHI + GEN + TET	2
AMP + FOX + PIP + CIP + ATM + OFLX + CHI + GEN + TET	1
Total	48

Table 3. Percentage of resistant strains based on phylogenetic group.

Phylogroup	Total (n)	Isolates (n)	AMP	PIP	TET	SXT	CHI	ATM	GEN	OFLX	CIP	FOX	AMC	MEM	KAN
A	29	12	27.6	27.6	24.1	31	20.7	6.9	20.7	0	3.4	0	0	0	0
B1	33	15	39.4	33.3	27.3	18.2	9.1	18.2	9.1	9.1	9.1	6.1	3.0	0	0
C	21	13	42.9	42.9	47.6	42.9	28.6	28.6	28.6	23.8	14.3	4.8	9.5	0	0
E	4	2	0	50.0	50.0	0	25.0	50.0	25.0	50.0	25.0	25.0	0	0	0

Note: Antimicrobials: AMP, ampicillin; FOX, cefoxitin; AMC, amoxicillin-clavulanic acid; SXT, sulfamethoxazole-trimethoprim; PIP: piperacillin; CIP, ciprofloxacin; ATM: aztreonam; MEM: meropenem; OFLX: ofloxacin; CHL, chloramphenicol; KAN, kanamycin; GEN, gentamicin; TET, tetracycline; n, number of strains.

3.3. Detection of antimicrobial resistance genes

In our study, 13 resistance genes associated with β-lactam, sulfonamide, tetracycline, and aminoglycoside antibiotics were analyzed (Table 4). E. coli isolates resistant to β-lactam including cefoxitin carried no bla_KPC, bla_FOX, bla_NDM, or bla_OXA genes. Among the isolates, 97.8% (90/92) carried gyrA, and 95.6% (84/92) carried gyrB. Approximately 75% (24/32) of tetracycline-resistant strains possessed tetA; however, no tetB was detected. Isolates with resistance to gentamicin contained aac4A (13/92, 14.1%), aac6 (9/92, 9.7%), and aac4A + aac6 (8/92, 8.7%), but none of the isolates contained aac3. Among the chloramphenicol-resistant isolates, 18 (78.0%) contained floR. Furthermore, 50.0% of penicillin-resistant strains had bla_TEM, and 42.0% strains had bla_CTX-M.

Table 4. Antimicrobial resistance genes identified in E. coli isolates from raw milk of mastitic cows.

Antibiotic class	Antibiotic	resistance gens	positive strains
			No.	(%)
β-Lactam	AMP	blaTEM	19	20.6
		blaCTX-M	16	17.3
		blaTEM+blaCTX-M	10	10.9
	FOX	blaFOX	0	0
	AMC	blaNDM	0	0
		blaKPC	0	0
	SXT	blaOXA	0	0
Tetracyclines	TET	TetA	24	26
		TetB	0	0
Quinolones	CIP	gyrA	90	97.8
		gyrB	84	95.6
Aminoglycosides	GEN	aac4A	13	14.1
		aac6	9	9.7
		aac4A + aac6	8	8.7
		aac3	0	0
Chloramphenicol	CHI	floR	18	19.5

3.4. Virulence distributions of E. coli isolates

The distribution of virulence genes in the 92 isolates is shown in Table 5. In our study, the eae gene was identified in 75 isolates, which were classified as EPEC, while no stx genes were detected. All eae gene-positive E. coli strains were divided into seven O-serogroups. The remaining 57 isolates possessed ipaH and were classified as EIEC, whereas six strains contained elt genes and were classified as ETEC.

Table 5. Prevalence of virulence genes in E. coli isolates.

	Virulence gene	No. of isolates	% isolates
EPEC	eae	75	81.5
EAEC	aggR	0	0
STEC	Stx1	0	0
	Stx2	0	0
EIEC	ipaH	57	61.9
ETEC	elt	6	6.5

3.5. Typing results

3.5.1. Phylogenetic typing

All strains were assigned to five groups according to Clermont’s phylogenetic method (Clermont et al., 2000). The distribution of serogroups, intimin types, and virulence genes among the different phylogenetic groups were determined. The majority of the E. coli strains were assigned to phylogroup B1 (35.9%), while the other strains were assigned to phylogroup A (31.5%), C (22.8%), or E (4.3%). However, five strains (5.4%) were not assigned to any phylogroup. Approximately 41.7% of Shandong strains were in phylogroup A, while group B1 strains were more predominant (45.8%) in Shanghai. Phylogroup E consisted of four strains from Hei Longjiang, Hebei, and Inner Mongolia (Table 6).

Table 6. Distribution of O serogroups among different phylogenetic groups.

phylogenetic group	Strains (No.)	Data for O serotype/strains in:					Total
		HLJ	SH	HB	SD	NM
A	29	4(13.8),O121,UT(3)	9(31.0),O111(2),O121(2),O128(3),O91,UT	5(17.2),O121(2),O91,UT(2)	5(17.2),O121(2),O91,UT(2)	6(20.7),O121,O91,UT(4)	O121(8),O91(4),O128(3),O111(2),UT(12)
B1	33	7(21.2),O121(2),O128,O91,UT(3)	11(33.3),O121(6),O128(3),O22,UT	8(24.2),O121,O91(2),O22(3),O113,UT	4(12.1),O91,UT(3)	3(9.1),O121(2),UT	O121(11),O91(4),O113,O22(4),O128(4),UT(9)
B2	ND
C	21	3(14.3),O121(3)	3(14.3),O121(3)	4(12.1),O113,O26,O91,O121	3(14.3),O91(2),UT	8(24.2),O121(3),O91(3),UT(2)	O91(6),O121(10),O113,O26,UT(3)
D	ND
E	4	1(25.0),O121		2(50.0),O121(2)		1(25.0),O22	O121(3),O22
F	ND
UNKNOWN	5	2(40.0),UT(2)	1(20.0),O121			2(40.0),O91,UT	O91,O121,UT(3)
Total	92	17(18.5)	24(26.1)	19(20.7)	12(13.0)	20(21.7)

Note: ND: not detected.

3.5.2. O serotype grouping

A total of 65 of the 92 E. coli strains (70.7%) belonged to seven O serotypes (O121, O91, O22, O26, O128, O111, and O113), and the remaining 27 strains were untypeable because the serotypes could not be determined (UT; Table 6). The prevalent serotypes were O121 and O91, accounting for 35.9% (33/92) and 16.3% (15/92), respectively. Other O genotypes obtained were O128 (seven strains, 7.6%), O22 (five strains, 5.4%), O111 and O113 (two strains in each genotype, 2.1%), and O26 (one strain, 1.1%). The isolates in phylogroups A and B1 had several O genotypes, whereas the C and E isolates were divided into four and two serogroups, respectively (Table 6).

3.5.3. Typing of intimin

Among the 92 E. coli strains, 75 (81.5%) were positive for the eae gene. Intimin-type eae-μR-ι2 was highly prevalent (48.9%,45/92) and belonged to six different O serogroups, additionally, 12 strains were designed UT, it was predominant (17/45, 15.6%) in phylogroup C compared other groups (A, B1, and E) (Figure 2). Intimin type λ (16.3%, 15/92) belonged to four serogroups. Ten strains were detected in type ξR among E. coli strains subjected to three serogroups, three and two strains showed positive reactions using two groups of primers (νR-ε2 and η). The remaining E. coli strains were UT because amplicons were not generated using the typing primers (Table 7).

Figure 2. Distribution of intimin types among phylogroups in E. coli isolates from bovine mastitis in China.

Table 7. Distribution of O serogroups among different intimin types.

Intimin type	Strains (No.)	O serotype(s) in:					Total
		HLJ	HB	SD	NM	SH
eae-μR-ι2	45	O121(5),O128,UT(6)	O121(2),O113(2),O91,UT(2)	O121,O91(3),UT	O121(5),O91(2),UT(3)	O121(6),O128(2),O22,O111(2)	O121(19),O113(2),O91(6),O128(3),O22,O111(2),UT(12)
eae-λ	15	O121	O91,O121,O22(2),UT	O121	O91,UT(2)	O121(2),O128,O91,UT	O91(3),O121(5),O22(2),O128,UT(4)
eae-ξR/β2	10		O26,O121	UT	O91(2),UT(3)	O121(2)	O26,O121(3),O91(2),UT(4)
eae-νR-ε2	3	O121	O91			O128	O91,O121,O128
eae-η	2			UT	O121		O121,UT
ND	17	O91,UT(2)	O121(2),O91,O22	O91,UT(3)	O22	O121(2),O128(2),UT	O22(2),O91(3),O121(4),O128(2),UT(6)

Note: UT: untypeable; ND: not detected.

4. Discussion

A recent report of large Chinese dairy farms indicated that E. coli is involved in the pathogenesis of bovine mastitis (Gao et al., 2017). In our study, there were 92 E. coli isolates (18.5%) in 497 raw milk samples. The prevalence of E. coli isolates was higher in our study than in a study performed in Turkey (3.2%, 8/250) (Akkaya, Cetinkaya, Alisarli, Telli, & Gök, 2008) and in Liaoning province, China (20.1%,79/389) (Zhang et al., 2018). Our results revealed that E. coli is prevalent in raw milk from mastitic cows in all five regions of China.

Antimicrobials are commonly used in the control and/or prevention of mastitis. Unfortunately, the misuse of antimicrobials contributes to the emergence of bacterial strains with antimicrobial resistance (Metzger & Hogan, 2013). In this study, a high percentage of strains were resistant to ampicillin (41.3%), followed by tetracycline (34.8%; Table 1), possibly due to the overuse of drugs in the provinces we selected. The resistance rate towards ampicillin was higher in our study than in South Korea in 2012–2015 (16.6%) (Tark et al., 2017) and some provinces of China (25.3%) (Zhang et al., 2018). Antimicrobial resistance to ampicillin can be used to predict the susceptibility to amoxicillin (Romney, 2018). Therefore, it can be speculated that the E. coli strains showed a high resistance to amoxicillin. Resistance rates towards tetracycline were 9.0% in the European Union (Thomas et al., 2015), 14.3% in Canada (Saini et al., 2012), and 55.1% in Southwestern Nigeria (Adesoji, Ogunjobi, Olatoye, & Douglas, 2015). Moreover, 34.8% isolates had MDR (Figure 1). A previous study reported that 45.8% of 48 E. coli strains were resistant to 13 antibiotics (Todorovic et al., 2018). The continuous use of antibiotics may cause MDR in dairy herds (Suojala et al., 2011). Therefore, it is important to control antibiotic use to prevent the risk of MDR. In Korea, researchers have concluded that the decline in tetracycline resistance may be due to the decreased use of the antibiotic in cattle (Tark et al., 2017).

Reports on antimicrobial resistance are useful for understanding the pathogenesis of E. coli infection in bovine mastitis (Blum et al., 2008). Broad-spectrum antibiotics, especially tetracyclines and quinolones, are frequently used in the clinical treatment of mastitis (Chajęcka-Wierzchowska, Zadernowska, & Łaniewska-Trokenheim, 2016; Zurfluh et al., 2014). It has been reported that resistance genotypes against tetracycline are correlated with resistance phenotypes in S. aureus and E. coli (Gomi et al., 2016; Yu et al., 2018). Some studies have shown that the tetA gene is more prevalent than the tetB gene in E. coli strains (Gomi et al., 2016; Hu et al., 2008; Rebbah et al., 2017), and our results supported this finding. In our study, 24 tetracycline-resistant isolates (75%) carried the tetracycline gene tetA; however, no tetB was detected. Our results revealed a discrepancy between genotype and phenotype. Quinolone resistance genes, gyrA and gyrB, were detected in more than 95% isolates, and only eight isolates had resistance to ciprofloxacin (Table 4), according to CLSI (2018). The results revealed the differences between genotype and phenotype in ciprofloxacin-resistant E. coli strains. The misuse of antimicrobials is a contributing factor, but probably not the main or single factor involved in antimicrobial resistance (Miika, Nyberg, Pentti, Pirkko, & Hakanen, 2009; Oliver, Murinda, & Jayarao, 2011). Future studies should evaluate the differences between phenotype and genotype.

Another factor correlated with the pathogenesis of E. coli infections in bovine mastitis is virulence profile. In the present study, none of the E. coli isolates were STEC (Table 5). Similar findings were reported in Iran (Mansouri-Najand & Khalili, 2007) and Belgium (Vivegnis, Ellioui, Leclercq, & Lambert, 1999). However, low prevalence rates were found in milk from US (Jayarao et al., 2006; Van Kessel, Karns, Lombard, & Kopral, 2011) and Europe (Claeys et al., 2013; Little et al., 2008; Ombarak et al., 2016; Pradel, Bertin, Martin, & Livrelli, 2008). In this study, there were 75 EPEC strains, in agreement with the findings of Momtaz (Momtaz, 2010). However, EPEC prevalence in bovine mastitis was lower in Brazil (40.0%) than in our study (Schroeder et al., 2002). There might be a direct relationship between the presence of eae and the capacity of EPEC to cause disease.

Intimin type is an important E. coli virulence determinant because intimin is necessary in the pathogenesis of EPEC (Dean-Nystrom, Bosworth, Moon, & O’Brien, 1998). The intimin variation could cause specificity for different host tissues, and intimin type could be a key factor in animal-host specificity (Fitzhenry et al., 2002; Torres, Zhou, & Kaper, 2005; Zhang et al., 2002). Overall, the 75 EPEC strains had five different intimin types (μR-ι2, λ, ξR/β2, νR-ε2, and η). To the best of our knowledge, there have been no reports on intimin types in milk from mastitic cows (Blanco, Blanco, Dahbi, Mora, et al., 2006; Osman, Mustafa, Aly, & AbdElhamed, 2012). These five intimins have been isolated from fecal samples of adults and children with clinical diarrhea or other gastrointestinal alterations and from neotropical non-human primates with diarrhea (Blanco, Blanco, Blanco, et al., 2004; Blanco et al., 2005; Blanco, Blanco, Dahbi, Alonso, et al., 2006). Furthermore, intimin η was observed in bovine typical EPEC strains (Blanco et al., 2005; Blanco, Blanco, Dahbi, Alonso, et al., 2006). The findings reveal that different intimin EPEC strains are important pathogens associated with human diarrhea and bovine mastitis, which suggests that intimins may play a role in the transmission of infections between animals and humans.

Genotyping is a useful technique in epidemiology. In this study, we coupled serotyping with other molecular typing methods. Most of the seven O serogroups have not been detected in bovine cows. Approximately 77.4% of E. coli isolates belonged to four different O serogroups (O26, O86, O111, and O127) in Egypt (Osman et al., 2012). The high frequency of eae found in this study may be related to the high frequency of certain serogroups, as it has been reported that the presence of eae is associated with specific O serogroups, such as O157, O145, O103, O26, and O111 (Miika et al., 2009). In this study, 45 intimin μR-ι2 isolates were in six serotypes (O121, O91, O113, O128, O22, and O111), 15 intimin λ strains were in four serotypes (O121, O91, O22, and O128), 10 intimin ξR/β2 strains were in three serotypes (O26, O91, O121), and intimin η was in serotype O121. O91 and O121 were frequently observed in these five intimin types. Our results showed that serogroups, such as O26, O91, O111, O113, O121, and O145 may cause diarrhea (Kappeli, Hachler, Giezendanner, Beutin, & Stephan, 2011; Schmidt, Beutin, & Karch, 1995). Furthermore, intimin types (β1 and γ1) present in O157 and O26 were similar in human and bovine tissue (Blanco, Blanco, Mora, et al., 2004). Therefore, isolates from milk, which belong to serogroups associated with diarrhea, are potentially pathogenic to humans.

In this study, E. coli strains associated with mastitis belonged to groups B1 (35.9%) and A (31.5%). In Brazil, most mastitis E. coli isolates were assigned to phylogroups A (52%) and B1 (38%) (Tomazi, Coura, Goncalves, Heinemann, & Santos, 2018). The major antibiotic-resistant isolates belonged to phylogroups A and B1, and phylogroup B1 had a higher level of antimicrobial resistance (Table 3). Additionally, 32 isolates of the EPEC strains were resistant to one or more antibiotics, and 23 isolates were non-susceptible to three or more antibiotic classes (Supplemental Figure 1). In Beijing, 58.6% of antibiotic-resistant E. coli strains were in group B1 and 35.7% were in group A (Garcia-Aljaro et al., 2009). Additionally, E. coli isolates with MDR were classified in phylogenetic groups A or B1 (Ramos et al., 2013). The combination of different virulence determinants and the phylogeny of bacteria may improve the recognition of new subgroups of virulent bacteria (Picard et al., 1999). The 75 EPEC strains in our study harbouring intimin genes could be separated into groups A (30.7%, 23/75), B1 (30.7%, 23/75), C (26.7%, 20/75), and E (4.0%, 3/75) (Figure 2). Similarly, researchers reported that intimin type β1 in the majority of E. coli from poultry fecal samples belonged to phylogroup B1 and virulence group I (Parvej et al., 2019). The intimin μR-ι2 isolates in this study were most frequently found in phylogroup C.

5. Conclusions

The antimicrobial resistance of E. coli strains in the five major dairy-producing regions in China should be a public concern. E. coli strains were resistant against antibiotics especially ampicillin, but susceptible to meropenem and kanamycin. Future studies should evaluate the relationship among antimicrobial susceptibility, phylogroups, and virulence genes of E. coli strains associated with human and bovine mastitis. Additionally, studies should assess whether EPEC isolated from bovine mastitis may cause disease in humans.

Ethics approval and consent to participate

All experimental procedures involving the care and management of dairy cows, and the collection of bovine mastitis milk in the current study, were approved by the Animal Care and Welfare Committee of Institute of Animal Sciences, Chinese Academy of Agricultural Sciences.

Acknowledgement

We sincerely thank all the participants who took part in this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The research was financially supported by the Project of Risk Assessment on Raw Milk [grant number GJFP2019008], the Agricultural Science and Technology Innovation Program [grant number ASTIP-IAS12], the Modern Agro-Industry Technology Research System of China [grant number CARS-36], Fundamental Research Funds for the Central Non-profit Research Institution [grant number 2018-YWF-RW-18], and the Scientific Research Project for Major Achievements of the Agricultural Science and Technology Innovation Program (ASTIP) [grant number CAAS-ZDXT2019004].