Correlation analysis of Shiga toxin–producing Escherichia coli shedding and faecal bacterial composition in beef cattle

Authors


Correspondence

Luxin Wang, Department of Animal Sciences, Auburn University, Auburn, 210 Upchurch Hall, AL 36849, USA.

E-mail: lzw0022@auburn.edu

Abstract

Aims

The objectives of this study were to investigate the correlations between Shiga toxin–producing Escherichia coli (STEC) shedding and faecal microflora in beef cattle and to identify functional species that might be used for STEC control.

Methods and Results

Faecal samples were collected from 110 calves and 92 dams. The number and prevalence of STEC were determined using CHROMagar™ STEC; denaturing gradient gel electrophoresis (DGGE) was employed to analyse faecal bacterial composition. Six-month-old calves had the highest STEC shedding levels (3·03 ± 1·41 Log CFU g−1) and prevalence (95·5%). Both the number and prevalence decreased significantly as the calf age increased (P < 0·05). The DGGE analysis showed that faecal bacterial diversity increased, while cattle ages increased and STEC shedding levels decreased. Significant correlations between STEC shedding, cattle age and bacterial compositions were observed by redundancy analysis (P < 0·05). T-value biplots and sequencing results indicated that butyrate-producing bacteria (BPB) negatively correlated with STEC shedding.

Conclusions

Higher STEC shedding levels and prevalence were associated with younger cattle age, lower faecal bacterial diversity and lower BPB levels.

Significance and Impact of the Study

Butyrate-producing bacteria in GI tract might serve as an option for the future development of STEC shedding control strategy.

Introduction

Shiga toxin–producing Escherichia coli (STEC) is an important group of zoonotic pathogens, which can cause severe clinical syndromes in humans such as haemorrhagic colitis and haemolytic uraemic syndrome (HUS) (Karmali 1989). Annually, STEC strains cause approximately 176 000 illnesses, 2400 hospitalizations and 20 deaths in the United States (Scallan et al. 2011). Escherichia coli O157 has been the most frequently isolated and most well-characterized STEC. Current statistical analyses show that the most prevalent pathogenic non-O157 STEC serogroups in the United States are O26, O45, O103, O111, O121 and O145, causing more than 70–83% of the total non-O157 STEC illnesses (Brooks et al. 2005). Due to the increasing concerns of non-O157 STEC, the USDA Food Safety and Inspection Service (FSIS) listed these additional six STEC strains to the hazardous adulterant list in June 2012. These six non-O157 STEC serogroups are now included in the zero-tolerance policy concerning nonintact raw beef products (FSIS-2010-0023, USDA, 2012).

Cattle are considered to be the primary reservoirs of STEC, and the main route of STEC infections in humans is the consumption of contaminated foods. Cattle's insensitivity to shiga toxin and differential preference in colonization sites make them more tolerant hosts than human for enterohemorrhagic Escherichia coli (EHEC) (Griffin and Tauxe 1991; Nguyen and Sperandio 2012). It has been commonly accepted that a reduction in STEC shedding in cattle during the preharvest stage would contribute greatly to the postharvest safety of beef. Cattle can be infected with E. coli O157 early in life and continue to shed the bacteria for several months (Cray and Moon 1995; Shaw et al. 2004; Widiasih et al. 2004). Compared with dams, calves tended to shed more E. coli O157, and the shedding prevalence changed as the calf age changed (Hancock et al. 1994, 1997; Zhao et al. 1995). However, few studies have focused on the total STEC shedding and investigated the mechanisms for these age-related changes.

Cattle faeces contain complex bacterial communities (Dowd et al. 2008). Although up to 30% of cattle are asymptomatic carriers (Callaway et al. 2009), STEC are still considered to be pathogens, not indigenous bacteria, to cattle, which could induce diarrhoea and lesions of the small and large intestine sometimes (Dean-Nystrom 2003). The effects of indigenous micro-organisms on the growth of O157 and total STEC have been studied in vitro. It has been found that the presence of indigenous bacteria may inhibit O157 and STEC growth (Poole et al. 2003; Kim and Jiang 2010). Nutritional competition between the indigenous intestinal Escherichia coli and E. coli O157:H7 is reported to be one of the growth inhibition reasons (Momose et al. 2008). In addition, it has been found that acetic, propionic and butyric acids that are produced by intestinal microflora might also inhibit the growth of O157 (Shin et al. 2002). Unfortunately, in vivo studies that evaluate the effect of GI tract microflora on STEC shedding are limited. In-depth information regarding the composition and identity of the functional faecal microflora is still needed.

The detection and identification of non-O157 STEC are more challenging than O157 as most non-O157 strains are sorbitol-fermenting, limiting the effectiveness of traditional sorbitol-MacConkey agar (Wylie et al. 2012). CHROMagar STEC, of which the selective mechanism is not based on sorbitol fermentation, has been developed for the detection of STEC. This medium has been used and tested in several studies on human faecal samples, which showed more than 85% of sensitivity and more than 90% of specificity (Wylie et al. 2012; Gouali et al. 2013).

Denaturing gradient gel electrophoresis (DGGE) is an effective technique for whole bacterial community analysis; it has been used for analysing cattle GI tract microflora in previous studies (Hook et al. 2009; Aldai et al. 2012). Combined with multivariate statistical analysis, correlations between the GI tract microflora and environmental factors can be determined by statistical significance and ecological interpretation (Kocherginskaya et al. 2001; Fromin et al. 2002).

To provide in-depth information to the future development of intervention strategies that can control STEC levels during preharvest stages, such as the weaning step, the objectives of this study were to (i) determine the prevalence and shedding levels of total STEC in young calves preweaning and dams, (ii) assess the effects of faecal microflora community on STEC shedding and (iii) identify micro-organisms that can generate an ‘inhibitory’ impact on STEC shedding.

Materials and methods

Animals

In this study, a total of 92 dams and 110 calves were sampled. Calves were born and reared with their dams in four different pastures at the E.V. Smith Research Center, Auburn University, Auburn, Alabama, USA. Dams and calves were grazed on warm-season perennial grass pastures throughout the growing season. Calves were sampled when they were approximately 6 months (C6M) and 9 months (C9M) old (May and August 2012). Dams (DAM) were sampled once in May. The age of the dams ranged from 2 to 11 years with an average age of 6·4 ± 2·6 years. The experimental protocol for care of the animals was approved by the Institutional Animal Care and Use Committee of Auburn University.

Faecal sample collection

Faecal samples were collected by rectal grab. Immediately after the sample collection, approximately 1 g of fresh faecal sample was transferred to a 15-ml sterile tube containing 5 ml of RNA later solution and 0·5 g of 3-mm glass beads. The tube was then oscillated and mixed manually for 10 s to preserve the samples for later DNA extraction. The rest of the sample was transferred into a 125-ml sterile specimen container (VWR, Radnor, PA, USA) and kept on ice. All samples were transported on ice to the laboratory. Samples in the RNA later tubes were stored at 4°C until analysed, and other samples were processed within 4 h.

Shiga toxin–producing Escherichia coli enumeration

To enumerate the total STEC in faeces, 25 g of each faecal sample was transferred from the specimen container into a sterile Whirl-Pak sample bag (Nasco, Atkinson, WI, USA) together with 112·5 ml of 0·1% (w/v) sterile peptone water (Difco, Franklin Lakes, NJ, USA). Samples were homogenized for 2 min in a Smasher (AES Chemunex, Cranbury, NJ, USA), and two 100-μl homogenates were taken and spread onto two CHROMagar™ STEC plates (CHROMagar™, Paris, France). The STEC colonies were counted according to the CHROMagar™ user's manual after incubating at 37°C for 24 h. To test the performance of CHROMagar STEC on cattle faecal samples, mauve colonies from randomly selected STEC plates were picked; both serotyping and multiplex PCR (for stx1, stx2 and eae genes) were employed to confirm their serotypes and the presence of shiga toxin–producing genes. To test the STEC in low concentrations, 112·5 ml of 2× lactose broth (Difco) was added to the rest of the homogenate in the Whirl-Pak bag, and mixtures were incubated at 35°C for 24 h. The enriched cultures were streaked onto CHROMagar™ STEC plates, and the presence or absence of STEC was determined the next day. The STEC concentration was reported as log CFU per gram of faecal sample, and the mean ± standard deviation for each faecal sample was calculated. The prevalence of STEC shedding in each group was calculated as the percentage of STEC shedding–positive cattle in relation to the total number of cattle in that group. To compare STEC shedding levels in different groups, Wilcoxon signed-rank test and Mann–Whitney U-test were performed. McNemar test and Fisher's exact test were also employed to compare STEC prevalence. Comparison between four pastures was made by Kruskal–Wallis test and chi-squared test. All tests were conducted by SPSS 16.0 (SPSS, Chicago, IL, USA) at a significance level of 0·05.

DNA extraction from faecal samples

To analyse total faecal microflora, 15-ml tubes containing faecal samples and RNAlater were mixed thoroughly by vortexing for 1 min, and 400 μl of homogenate was transferred into an ice-cold 2-ml screw-cap tube. DNA samples were extracted according to the method described by Matsuki et al. (2004), and the quality of DNA was determined by electrophoresis in 1% (w/v) agarose gels. Extracted DNA samples were stored at −20°C until use.

Denaturing gradient gel electrophoresis

For DGGE, the V3 variable region of bacterial 16S rDNA was amplified by PCR using the primers described by Muyzer et al. (1993), and the primers were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA, USA). A 50 μl PCR solution was prepared as follows: 2·5 μl bacterial genomic DNA, 0·25 μmol l−1 (final concentration) of each primer (341F-GC and 534R), 25 μl of Illustra Hot Start Master Mix (GE Healthcare, Piscataway, NJ, USA) and sterile Milli-Q water. A ‘touchdown’ PCR was performed in the Veriti Thermal Cycler (Applied Biosystems, Life Technologies, Grand Island, NY, USA) following the programme described by Lubbs et al. (2009). The products were determined by electrophoresis in 2% (w/v) agarose gels. Twenty microlitres of loading buffer was added to each 50 μl PCR product, and the mixture was stored at −20°C until use.

Denaturing gradient gel electrophoresis was performed using a D-GENE System (Bio-Rad, Hercules, CA, USA). PCR products were separated on 8% (w/v) polyacrylamide gels (generated from 40% acrylamide–bisacrylamide 37·5:1 stock solution; Amresco, Solon, OH, USA) in 0·5× Tris–acetate–EDTA (TAE) buffer along a 30–60% linear denaturing gradient (Muyzer et al. 1993). Electrophoresis was performed in 0·5× TAE at 150 V and 60°C for 7 h. Gels were stained by GelRed (Biotium, Hayward, CA, USA) for 15 min and photographed by a GelDoc system (Bio-Rad). In each gel, one control sample was loaded to the outside lanes as the marker, and all gel images were aligned and merged into one image according to the marker (Gafan et al. 2005).

Gel analysis

The DGGE gel image was analysed using Quantity One software (Bio-Rad) according to the user guide. The gauss trace quantity and relative intensity of each band (total intensity of all bands in each lane was defined as 100%) in the lane were calculated. The band type tables with relative intensity values and gauss trace quantities were exported to Excel (Microsoft, Mountain View, CA, USA). To visualize the similarity of the faecal microflora from samples, a dendrogram was constructed based on Dice's coefficient (UPGMA algorithm) implemented in the Quantity One software (Fuentes et al. 2008). The Shannon–Wiener diversity index (Shannon index) of samples was calculated by Shannon calculator (www.changbioscience.com/genetics/shannon.html) based on gauss trace quantities (Gafan et al. 2005). The effects of STEC prevalence and cattle age on Shannon index were tested by anova using univariate GLM of SPSS 16.0 (P < 0·05).

To assess the relationships between the composition of faecal microflora, STEC shedding and calf age, multivariate statistics were performed using Canoco 4·5 (Biometrics, Wageningen, the Netherlands). The band type table with relative intensity information was imported as species data; STEC shedding levels and cattle age were defined as ‘environmental’ variables (C6M, C9M and DAM were defined as 1, 2 and 3, respectively). Linear model of redundancy analysis (RDA) with the focus scaling on interspecies distances was employed (Janczyk et al. 2010). Unrestricted Monte Carlo permutation tests were applied to test the significance of the microflora response with two variables (499 random permutations, P < 0·05). To identify bacterial strains that are correlated significantly with STEC shedding and cattle age, t-value biplots for each variable were graphed based on the RDA by CanoDraw (one module of Canoco, Biometrics, Wageningen, the Netherlands). Species vectors (band types) enclosed in Van Dobben circles indicated significant positive/negative correlations with STEC shedding or age (regression coefficient < −2 or > +2) (Lepš and Šmilauer 2003).

Based on the analysis, the related bands were marked on the DGGE gel image and were excised with a sterile scalpel from the DGGE gels. The DNA was extracted from bands according to the procedure described by Simpson et al. (2000). The DNA was subjected to PCR amplification with 341F and 534R primers (Muyzer et al. 1993), and the resulting PCR products were cloned into E. coli 10G competent cells (Lucigen, Middleton, WI, USA) with the pGEM-T Easy Vector (Promega, Madison, WI, USA). The positive plasmid DNA was subjected to PCR (with 341F-GC and 534R primers) and was checked by DGGE to confirm the purity and size. The plasmids with desirable inserts were sequenced by Lucigen Corporation using the M13 primer. The obtained sequences were compared with known sequences in the GenBank database using the BLASTn algorithm (NCBI, Bethesda, MD, USA).

Real-time PCR

To better confirm the correlations between butyrate-producing bacteria (BPB) and STEC shedding levels, the concentrations of BPB in faecal samples from different cattle groups were determined. The gene copy numbers of butyryl-CoA/acetate CoA-transferase (BCoAT) were monitored with primers BCoATscrF/R (Louis and Flint 2007) in a 7500 real-time PCR system (Applied Biosystems). Each 20 μl of reaction mixture consisted of 1·0 μl bacterial genomic DNA, 0·25 μmol l−1 (final concentration) of each primer, 10 μl of PerfeCTa SYBR Green SuperMix (Quanta BioSciences, Gaithersburg, MD, USA) and sterile Milli-Q water. The real-time PCR was performed following the programme described by Louis and Flint (2007). To build standard curves, serial dilutions of purified and quantified BCoAT PCR products were generated by conventional PCR. The gene copies per gram of faecal sample were calculated. anova was employed to compare the mean values of gene copies between different cattle groups using univariate GLM of SPSS 16.0 (P < 0·05).

Results

Enumeration and prevalence of STEC

Ninety-two dams gave birth to 110 calves from November 2011, and the calves stayed with their dams for the whole study. Dams and their calves were grazed in four separate pastures. Faecal samples were collected from calves and dams, and the shedding levels (concentrations) of STEC in the faeces were determined. To check the efficiency of CHROMagar™ STEC, 49 mauve colonies from randomly selected STEC plates were picked and purified. They were sent for serotyping at the Escherichia coli Reference Center (Pennsylvania State University, University Park, PA, USA), and the presence of intimin (eae) and Shiga toxin–producing (stx1 and stx2) genes was checked by PCR. Among them, six serotypes (O120, O172, O41, O49, O157 and O176) were confirmed, and contained eae, stx1 and/or stx2 genes.

Based on the statistical analysis, no significant difference was detected between four groups of dams and calves on four pastures on both STEC shedding levels and prevalence (P > 0·05). CHROMagar™ STEC agar was used to enumerate the STEC levels in faecal samples. The detection limit of direct plate count was 1·74 log CFU per gram of faeces. Lower concentrations of STEC (<1·74 log CFU g−1) were tested after enrichment. As shown in Table 1, C6M had the highest STEC counts (3·03 ± 1·41 CFU g−1, 3·63 median), followed by C9M (2·30 ± 1·61 CFU g−1, 2·60 median) and the DAM group. Dams had the lowest STEC counts (0·46 ± 0·95 CFU g−1). Calves (C6M and C9M) had significantly higher numbers of STEC in their faeces than in their dams (DAM) (Table 1, P < 0·05). The overall prevalence of STEC shedding was 95·5% for samples collected when calves were 6 months old, which was significantly higher than that of the other two groups (C9M and DAM, P < 0·05). There was no significant difference of STEC shedding prevalence between samples collected when calves were 9 months old and samples from the dams (Table 1).

Table 1. Enumeration of Shiga toxin–producing Escherichia coli (STEC) from faecal samples
Cattle groupsNumber of STECaMinimumMedianMaximumSTEC prevalenceb, %
  1. a

    Reported as log CFU ± SD per gram of faecal sample. Wilcoxon signed-rank test (P < 0·05) was used for comparison between groups C6M and C9M; Mann–Whitney U-test (P < 0·05) was used for comparison between groups C6M/C9M and DAM.

  2. b

    Reported as the percentage of STEC shedding–positive cattle in relation to the total number of cattle in that group. McNemar test (P < 0·05) was used for comparison between groups C6M and C9M; Fisher's exact test (P < 0·05) was used for comparison between groups C6M/C9M and DAM.

  3. a,b,cIndicate significant differences between groups based on statistical analyses (P < 0·05).

C6M (n = 110)3·03 ± 1·41a0·003·634·6295·5a
C9M (n = 110)2·30 ± 1·61b0·002·604·6883·6b
DAM (n = 92)0·46 ± 0·95c0·000·003·3683·7b

Denaturing gradient gel electrophoresis analysis

To investigate the relationships among faecal microflora, STEC shedding and cattle age, 64 faecal samples (Table 2) were chosen for DGGE analysis according to the following rules: (i) samples that contained more than 4 log CFU g−1 of STEC when calves were 6 months old (named ‘high STEC C6M’); (ii) STEC-negative samples collected when calves were 6 months old (named ‘none STEC C6M’); (iii) samples from the calves that were STEC shedders when they were 6 months old but became STEC negative when they were 9 months old (named ‘shedder-to-none calves’); (iv) samples from calves that were STEC negative at 6 months old, but then become STEC shedders at 9 months old (named ‘none-to-shedder calves’); (v) samples from dams that were STEC shedders (named ‘STEC shedder dams’); (vi) samples from dams that were STEC negative (named ‘none STEC dams’). Based on software analysis, a total of 81 band types were identified by the Quantity One software (Bio-Rad).

Table 2. List of faecal samples chosen for denaturing gradient gel electrophoresis analysis
Cattle no.Cattle ear tag no.Number of Shiga toxin–producing Escherichia coli (STEC)a
  1. a

    Mean log CFU STEC per gram of faecal sample ± SD is reported.

  2. b

    4·14 log CFU g−1 is the upper limit of the detection by plating method.

  3. c

    Indicates none suspect colony presented on the CHROMagarTM STEC agar after enrichment.

High STEC C6M
C1-6M-S10864·12 ± 0·02
C2-6M-S10894·38 ± 0·04
C3-6M-S1062>4·14b
C4-6M-S10714·23 ± 0·04
C5-6M-S1075>4·14
C6-6M-S1058>4·14
C7-6M-S10674·26 ± 0·11
C8-6M-S1081>4·14
C9-6M-S1065>4·14
C10-6M-S10024·05 ± 0·03
C11-6M-S1030>4·36
C12-6M-S10294·17 ± 0·02
C13-6M-S1035>4·14
C14-6M-S10974·47 ± 0·09
C15-6M-S11004·62 ± 0·14
None STEC C6M
C16-6M-NS11010c
C17-6M-NS10740
C18-6M-NS10070
C19-6M-NS10280
C20-6M-NS10430
Shedder-to-none calves
C21-6M-S1018>4·14
C21-9M-NS10180
C22-6M-S1025>4·14
C22-9M-NS10250
C23-6M-S11043·6 ± 0·01
C23-9M-NS11040
C24-6M-S1112>4·14
C24-9M-NS11120
C25-6M-S11133·65 ± 0·1
C25-9M-NS11130
C26-6M-S11222·94 ± 0·04
C26-9M-NS11220
C27-6M-S11234·2 ± 0·02
C27-9M-NS11230
None-to-shedder calves
C28-6M-NS10420
C28-9M-S10424·47 ± 0·06
C29-6M-NS10940·1
C29-9M-S10942·49 ± 0·21
C30-6M-NS10280
C30-9M-S10282·64 ± 0·01
C31-6M-NS10950
C31-9M-S10952·52 ± 0·25
C32-6M-NS10990
C32-9M-S10993·04 ± 0·01
STEC shedder dams
D1-S40762·88 ± 0·06
D2-S41063·36 ± 0·07
D3-S20282·72 ± 0·03
D4-S41062·74 ± 0·01
D5-S40983·08 ± 0·06
D6-S80413·27 ± 0·05
None STEC dams
D7-NS11610
D8-NS20640
D9-NS80500
D10-NS80590
D11-NS80580
D12-NS80890
D13-NS70630
D14-NS71020
D15-NS70980
D16-NS70120
D17-NS60460
D18-NS60340
D19-NS50390
D20-NS10640

Comparison of microflora profiles among samples

A dendrogram was obtained from DGGE analysis using the UPGMA clustering algorithm. As shown in Fig. 1, the faecal samples were grouped into four distinct clusters based on the similarity of the microflora. The Dice's coefficient range was between 0·55 and 0·93. All dam faecal samples, irrespective of whether they were STEC shedding or not, were grouped into one cluster (Cluster A), in which three 9-month-old non-STEC shedding calves were also included. An additional nine samples from 12 9-month-old calf samples belonged to cluster B, in which 9 of 10 samples from non-STEC shedding 6-month-old calves were also included. Nineteen of 22 samples collected from STEC-shedding 6-month-old calves were grouped in cluster C. One sample from an STEC-shedding 6-month-old calf and one sample from a non-STEC shedding 6-month-old calf were clustered in group D.

Figure 1.

The dendrogram of cattle derived from denaturing gradient gel electrophoresis analysis of faecal microflora based on Dice's similarity index and the UPGMA clustering algorithm. (a–d) four clusters are observed. Abbreviations C, D, S, NS, 6M and 9M represent calf, dam, Shiga toxin–producing Escherichia coli (STEC) shedder, non-STEC shedder, 6-month-old calf and 9-month-old calf, respectively.

Bacterial diversity analysis

The Shannon–Wiener diversity index (Shannon index) was calculated to evaluate bacterial diversity. Based on the anova analysis, results showed that both STEC shedding level and cattle age can affect the Shannon index significantly (P = 0·042 and P = 0·028, respectively), although the combined effect of STEC and age on the Shannon index is not significant (P = 0·567).

When age of cattle is considered as a variable, the mean Shannon indexes of C6M, C9M and DAM increased with cattle age. This indicates that the older the animals are, the higher the microflora diversity. When focusing on STEC shedding, the mean Shannon indexes of the no-STEC shedders are higher than STEC shedders, indicating that higher bacterial diversity is corresponding with non-STEC shedding.

Correlation analysis between microflora, STEC shedding and age

To assess the relationships between faecal microflora (the number of species and the abundance of each species) and STEC shedding/cattle age, RDA was employed. When both STEC shedding level and cattle age are treated as variables, the axis 1 of the RDA explains 18·2% of the total variation, while axes 1 and 2 together explain 25·4% (Fig. 2). When STEC or age is treated as one variable and another one treated as a covariable, significant species–variable correlation is detected (P = 0·010 and P = 0·004, respectively, Monte Carlo permutation tests). Results indicated that both STEC shedding and cattle age are correlated with the faecal microflora.

Figure 2.

T-value biplot analysis of bacterial species with Shiga toxin–producing Escherichia coli (STEC) shedding levels (a) and ages (b). Band types are represented as vectors, and the environmental variables (STEC shedding levels and ages) are represented as triangles. The circles are Van Dobben circles. Bands in transparent Van Dobben circles are positively correlated with STEC shedding/age, while bands in grey Van Dobben circles are negatively correlated with STEC shedding/age.

Identification of significant DGGE bands

Once the correlations between microflora, STEC shedding and cattle age were confirmed by RDA, t-value biplots were used to identify significant DGGE bands. As shown in Fig. 2, a total of 30 bands were found to be correlated significantly with STEC shedding level and/or cattle age (Fig. 2). In Fig. 2a, bands B40, B43, B66 and B72 are positively correlated with STEC shedding (enclosed in transparent Van Dobben circle), whereas B39, B44, B60, B79 and B81 are negatively correlated with STEC shedding (enclosed in grey Van Dobben circle). In Fig. 2b, B4, B13, B22, B29, B39, B41, B49, B53, B58, B60, B61, B64, B66, B67, B72, B74, B77 and B80 are positively correlated with cattle age (enclosed in transparent Van Dobben circle); B21, B27, B42, B55, B59, B62, B73 and B81 are negatively correlated with cattle age (enclosed in grey Van Dobben circle). The positions of these bands in a DGGE gel are labelled in Fig. 3, and these bands were extracted and sequenced.

Figure 3.

Positions of the bands that were significantly correlated with Shiga toxin–producing Escherichia coli shedding and ages in a denaturing gradient gel electrophoresis gel.

Characterization of identified bands

The characterizations of these bands are summarized in Table 3. The abundance of five bacterial strains is identified to be correlated with both STEC shedding and age. The abundance of one uncultured bacterial strain (B66) and one Clostridiales strain (B72) increased as STEC shedding level and cattle age increased. The quantity of one Firmicutes strain (B81) decreased as the STEC shedding level and age increased. The quantity of one Anaerostipes butyraticus strain (B39) and one uncultured bacterial strain (B60) decreased as STEC shedding level increased but increased as age increased.

Table 3. Summary of band characterizations and their correlations with Shiga toxin–producing Escherichia coli (STEC) shedding and cattle ages
Band no.Closest relativeSTECAge% identityAccession no.
  1. +: higher amount of this particular strain corresponding with higher STEC shedding levels and/or cattle ages; −: higher amount of this particular strain corresponding with lower STEC shedding levels and/or cattle ages.

B39Anaerostipes butyraticus strain JCM 17466+98 AB616134·1
B60Uncultured bacterium clone Hdb1-92+99 JX095573·1
B81Uncultured Firmicutes bacterium clone TF2-64100 GU957992·1
B44Uncultured Alistipes sp. clone EMP_M11 99 EU794135·1
B79Uncultured Firmicutes bacterium clone PA_424·33-2 99 GU939450·1
B66Uncultured bacterium clone UM2-8-23++100 JN857827·1
B72Uncultured Clostridiales bacterium clone EMP_I21++90 EU794242·1
B40Uncultured Bacteroides sp. clone M40-88+ 99 JN167622·1
B43Uncultured Clostridium sp. clone GAS2+ 100 JX826396·1
B4Uncultured bacterium clone Hma1-68 +99 JX096093·1
B13Uncultured Porphyromonadaceae bacterium clone SL34 +100 JN680579·1
B22Uncultured Alistipes sp. clone EMP_U8 +94 EU794272·1
B29Uncultured Lachnospiraceae bacterium clone EMP_P12 +99 EU794306·1
B41Uncultured bacterium clone Hmb2-65 +100 JX096363·1
B49Uncultured bacterium clone RL188_aah17h02 +100 DQ802446·1
B58Uncultured Ruminococcaceae bacterium clone SH40-32 +93 JN253632·1
B53Uncultured bacterium clone Hmb2-41 +90 JX096339·1
B61Uncultured Ruminococcaceae bacterium clone EMP_F40 +97 EU794170·1
B64 Clostridium sartagoforme  +98 AB610554·1
B67Uncultured bacterium clone EMIRGE_OTU_s8b4e_17298 +97 JX225938·1
B74Uncultured bacterium clone CR_66 +99 JX457189·1
B77Desulfovibrio desulfuricans strain E056 +99 JX267085·1
B80Uncultured Clostridiales Family XIII bacterium clone EMP_E28 +94 EU794083·1
B21Uncultured Bacteroidales bacterium clone cow 174 98 HQ201837·1
B27Uncultured Bacteroidales bacterium clone EMP_V8 97 EU794171·1
B42Uncultured bacterium clone T2C122 100 JQ265475·1
B55Uncultured rumen bacterium clone Bovine_70 86 GU222532·1
B59Uncultured firmicutes bacterium clone O2-76 98 GU955851·1
B62Uncultured bacterium clone DE06454A10 95 JQ695531·1
B73Uncultured bacterium clone TU2_aaa01c06 99 EU470175·1

Four bacterial strains are correlated with STEC shedding levels: they are one Bacteroides strain (B40) and one Clostridium strain (B43), which increased as STEC shedding level increased, and one Alistipes strain (B44) and one Firmicutes strain (B79), which decreased as STEC shedding level increased. In addition, 21 bacterial strains were identified to be correlated only with cattle age: the quantity of one Porphyromonadaceae strain (B13), one Alistipes (B22) strain, one Lachnospiraceae strain (B29), two Ruminococcaceae strains (B58 and B61), one Clostridium sartagoforme strain (B64), one Desulfovibrio desulfuricans strain (B77), one Clostridiales Family XIII strain (B80) and six uncultured bacterial strains (B4, B41, B49, B53, B67 and B74) increased as cattle age increased; two Bacteroidales strains (B21 and B27), one rumen bacterial strain (B55), one Firmicutes strain (B59) and three uncultured bacterial strains (B42, B62 and B73) decreased as cattle age increased.

Quantification of butyrate-producing bacteria by real-time PCR

The detection limit of the real-time PCR assay was 103 gene copies g−1 faeces. The gene copy numbers of BCoAT were 4·15 ± 0·21 for 6-month-old calves, 4·48 ± 0·41 for 9-month-old calves and 4·55 ± 0·49 for the dams (Fig. 4). DAM showed higher numbers of BCoAT genes than C6M (P < 0·05) and had no difference with C9M. Although no significant difference was found between STEC shedders and non-STEC shedders from the C9M and DAM groups, for C6M calves, non-STEC shedders had significantly higher copy numbers of BCoAT than STEC shedders (P < 0·05).

Figure 4.

Gene copy numbers of butyryl-CoA/acetate CoA-transferase (log gene copy numbers g−1 of faecal sample). a,b,cDifferent letters indicated significant differences existed between cattle groups (anova, P < 0·05).

Discussion

Cattle faeces are considered to be the main contamination source of STEC in foods especially beef products. In this study, faecal samples collected from 92 cows and their calves (110 calves) were used. For STEC detection and enumeration, CHROMagar™ STEC was chosen based on published literature (Wylie et al. 2012; Gouali et al. 2013). According to the users' manual, both Escherichia coli O157 and non-O157 STEC can grow on this agar. More than 85% of sensitivity and more than 90% of specificity on human faecal samples have been observed in previous studies (Wylie et al. 2012; Gouali et al. 2013). Growth on CHROMagar™ STEC is associated with the resistance to tellurite, and the presence of tellurite resistance gene (terB sequences) has been found in 87·2% of 235 EHEC but only in 12·5% of 567 non-EHEC strains (Tzschoppe et al. 2012). To check the efficiency of CHROMagar™ STEC, preliminary tests were performed, and the data revealed that the specificity of CHROMagar™ STEC for this study was promising. However, future in-depth studies are still needed to better confirm the presumptive colonies on the agar and to investigate the efficiency of this agar on STEC recovery and detection.

In this study, no relationship was found between STEC shedding dams and their calves. However, a significant age-related STEC shedding level was found (P < 0·05). Dams and 9-month-old calves shed lower levels of STEC and had lower STEC shedding prevalence. The prevalence of STEC shedding was between 80 and 95% in this study; higher STEC prevalence (compared with E. coli O157:H7 prevalence) has been found in the previous studies: the presence of the stx gene has been detected frequently from both dairy cattle (at 82%) and beef cattle (at 53%) (Cerqueira et al. 1999); Shinagawa et al. (2000) reported that levels of stx genes in faeces from healthy Japanese cattle ranged from 39·4% to 78·9%, depending on the age of the cattle, with calves 2–8 months old having the highest prevalence; Kobayashi et al. (2001) used a nested PCR approach and detected stx genes from 100% of the cattle faecal samples; high prevalence of non-O157 STEC in healthy cattle faecal samples was also reported in Argentina (44%; Parma et al. 2000) and Spain (35%; Blanco et al. 1997). Seasonal effect can be another reason for high STEC shedding prevalence. Higher faecal prevalence of E. coli O157:H7 in ruminants was found in summer months than in winter (Edrington et al. 2006). The two sampling times in this study were May and August; the consistent hot and humid weather in Alabama might be a contributing factor for the high STEC shedding levels/prevalence.

Age-related STEC shedding changes were found in this study; highest STEC shedding levels were found in the youngest calf group. Similar findings have been reported by others. Wells et al. (1991) isolated E. coli O157:H7 from 5 of 210 calves (2·3%) and 12 of 394 heifers (3·0%), but only 1 of 662 adult cows (0·15%). This and later studies (Wilson et al. 1992, 1993) revealed that E. coli O157:H7 is more frequently carried by calves and heifers than by adult cattle. However, opposite conclusions were also found; a study conducted by Zhao et al. (1995) found that weaned calves (4·9–5·3%) shed E. coli O157:H7 in faeces more frequently than preweaned calves (1·5–2·9%). The discrepancy between these previous studies might be explained by the following reasons: (i) when survey studies were conducted, the on farm- and individual animal-level management practices were different, and the age of the animals varied among farms; (ii) if the calves were weaned during the time of sampling, the diets used during the weaning period and the weaning stresses could influence the shedding prevalence. In this study, to minimize the effects of different management practices, various ages and dietary stresses, the ages of the calves and the management practices were all well controlled.

Cattle faeces contain complex bacterial communities (Dowd et al. 2008). As shown in this study, all DAM samples were grouped into cluster A and calf samples distributed in three clusters. This indicated the dams had higher similarity of bacterial communities than calves. The results showed bacterial diversity increased as cattle age increased, which corresponded with lower STEC shedding levels and prevalence. Non-STEC samples showed the highest bacterial diversity. This indicates that the high-diversity bacterial community might be one factor that influences STEC survival, attachment and shedding in the cattle intestine. Although the inhibitory effects of indigenous micro-organisms on the growth of E.  coli O157:H7 have been studied in vitro (Poole et al. 2003; Momose et al. 2008; Kim and Jiang 2010), the mechanism is still not well defined.

To investigate the mechanisms of the inhibitory effects of high-diversity bacterial communities on STEC shedding, the microflora populations of 64 cattle faecal samples were analysed by PCR-DGGE. The RDA was used to evaluate correlations between microflora diversity, STEC shedding and cattle age. The first and second axes explained up to 25·4% of the variance, indicating that other factors may also influence cattle faecal microflora (Fuentes et al. 2008; Janczyk et al. 2010). A total of 30 bands correlated with STEC shedding and age were identified in t-value biplots based on RDA. Those bands were sequenced and characterized.

Among all identified strains, one A. butyraticus strain (B39) was recognized as a potential STEC controller in this study, especially in the dams' GI tract. T-value biplot showed that this bacterium was negatively correlated with STEC shedding. The presence of a high abundance of A. butyraticus was found in low-STEC-shedding calves and cattle. Anaerostipes butyraticus is a novel butyrate-producing species belonging to the family of Clostridium cluster XIVa (Eeckhaut et al. 2010), and its butyrate-producing metabolism could contribute to STEC growth suppression (Shin et al. 2002). Meanwhile, this A. butyraticus strain showed a positive correlation with cattle age, which can better explain the phenomena of having low STEC shedding levels and prevalence in dams. In addition to A. butyraticus, 5 other BPB, including one Porphyromonadaceae strain (B13) (Sakamoto et al. 2009), one Lachnospiraceae strain (B29) (Carlier et al. 2004), two Ruminococcaceae strains (B58 and B61) (Collins et al. 1994; Makivuokko et al. 2010) and one Clostridium sartagoforme strain (B64) (Simunek et al. 2001), were positively correlated with cattle age. Higher concentrations of BPB may be one characteristic of the dams' intestine that is important to the STEC attachment/shedding control.

To monitor and confirm the dynamic changes of the BPB concentration, butyryl-CoA/acetate CoA-transferase (BCoAT) genes were quantified using real-time PCR. BCoAT is the last step of butyrate formation, and its gene has been used to quantify BPB in the gastrointestinal tract (Louis and Flint 2007). In this study, dams had the highest numbers of BCoAT genes than younger calves (C6M), and in younger calves (C6M), higher number of BCoAT genes found in non-STEC calves. The data confirm the negative correlations found between BPB and STEC shedding in cattle.

Besides the BPB, Alistipes (B22 and B44) and Desulfovibrio desulfuricans (B77) can also be potential STEC shedding ‘inhibitors’ in cattle intestine. Meanwhile, several potential STEC shedding ‘promoting’ bacteria, like one Bacteroides strain (B40), one Clostridium strain (B43) and one Clostridiales (B72) strain, were identified in this study. Together with 15 uncultured bacterial strains, further research should focus on the role of these bacteria to better understand their correlations with STEC shedding and cattle age.

To our knowledge, this study is the first study that has attempted to relate STEC shedding to faecal microflora diversity and cattle age. Using an overall faecal microflora fingerprinting method, this study was able to demonstrate the correlations between the faecal microflora diversity, STEC shedding levels as well as cattle age. High STEC shedding levels and high STEC shedding prevalence were observed from 6-month-old calves. Both the shedding levels and prevalence decreased as the calf age increased and the microflora diversity index increased. The higher abundance of BPB is the main group of bacteria identified to be directly related to the lower STEC shedding levels, and the negative relationship between BPB and STEC was confirmed by real-time PCR. Previous studies showed that diets could change the numbers of BCoAT in pig and human GI tract (Metzler-Zebeli et al. 2011; Kabeerdoss et al. 2012). Based on the results, preharvest intervention strategies such as diet modification should be explored in the future to control STEC shedding. Ongoing research in our laboratory includes the gene expression monitoring of BPB during the weaning with different diets and evaluating the transportation stress on STEC shedding.

Acknowledgements

The authors thank the Alabama Beef Forage Initiative grant and the Alabama Agricultural Experimental Satiation for their support of this research.

Ancillary