The use of feed antimicrobials in livestock is controversial because of the potential selection pressure for resistant organisms and the subsequent risk to human health (Gustafson and Bowen, 1997; Barton, 2000). Monensin and tylosin are widely used in cattle diets to improve feed efficiency and reduce liver abscesses (Stock et al., 1995; Nagaraja and Chengappa, 1998). Because of concerns with antimicrobial resistance, there is interest in producing food animals without the use of antimicrobial feed additives. Wet distillers grains with solubles (WDGS) are a coproduct of ethanol production from cereal grains and are comprised principally of the bran, protein, and germ fractions (Spiehs et al., 2002). The high fiber content in distillers grain makes their primary use in ruminant diets (Ham et al., 1994; Lodge et al., 1997; Kleinschmit et al., 2006). During ethanol production, antimicrobials, such as penicillin and virginiamycin, are used to suppress bacterial contamination (Narendranath et al., 2000). It is generally believed that the distillation process destroys these antibiotics (Shurson, 2005). The high energy density in distillers grains provides an opportunity to reduce dependency on feed antimicrobials while achieving similar growth promotion.
Foodborne pathogens, such as Escherichia coli O157 and Salmonella, are important to the beef industry. The effect of antimicrobial additives and other feed ingredients on the prevalence of foodborne pathogens has implications for food safety.
Our objectives were to determine the effects of feeding WDGS with or without monensin and tylosin on the prevalence and antimicrobial susceptibilities of E. coli O157 and Salmonella and commensal E. coli and Enterococcus spp. in feces of finishing cattle. We hypothesized that feeding distillers grains and feed antibiotics to cattle will exert pressure for antimicrobial resistance in the gut bacteria and increase the concentration of antimicrobial resistance genes.
MATERIALS AND METHODS
Animals, Diets, and Sampling Schedule
This study and all the procedures in the care and management of cattle were approved by the Kansas State University Institutional Care and Use Committee.
The study was conducted in the summer of 2005. Three hundred seventy crossbred yearling heifers were allotted to 54 concrete-surfaced feedlot pens (5.2 m2 per animal) with 6 to 7 animals per pen. Diets were fed for 150 d and consisted of 6% (DM basis) alfalfa hay and steam-flaked corn (Table 1). The study was a randomized complete block design with a 2 × 3 factorial treatment arrangement (9 pens per treatment). Wet corn distillers grain with solubles was included at 0 or 25% (DM basis), and antimicrobial feed additives were the second factor, included as none, monensin (Rumensin, 300 mg • animal−1 • d−1; Elanco Animal Health, Greenfield, IN) alone or monensin and tylosin (300 mg • animal−1 • d−1 and Tylan, 90 mg • animal−1 • d−1, respectively; Elanco Animal Health). Pens were blocked in a series of 6 sequential pens, and treatments were randomly allocated within each block. There were common fence lines between adjacent pens. Fecal samples were collected rectally from each animal on d 122 and 136 of the finishing period, placed in sterile bags, and transported to the laboratory immediately.
Unless otherwise indicated, all culture media were Difco brand (BD, Sparks, MD). Each fecal sample was cultured for E. coli O157 (Greenquist et al., 2005). Briefly, 1 g of feces was enriched at 37°C for 6 h in 9 mL of Gram-negative broth with cefixime (0.5 mg/L), cefsulodin (10 mg/L), and vancomycin (8 mg/L). Immunomagnetic bead separation (IMS) was performed, followed by plating onto sorbitol-MacConkey agar with cefixime (0.5 mg/L) and potassium tellurite (2.5 mg/L). Plates were incubated overnight at 37°C, and sorbitol-negative colonies (up to 6) were picked and replated onto blood agar plates (BAP; Remel, Lenexa, KS). After an overnight incubation at 37° C, colonies were tested for indole production and latex agglutination for the O157 antigen (Oxoid, Remel). Species were confirmed using the API 20E kit (bioMe ′riux, Hazelwood, MO), and PCR was used to determine the presence of 2 Shiga toxin genes, stx1 and stx2 (Fagan et al., 1999). The multiplex PCR program to amplify targets was as follows: initial denaturation at 95° C for 3 min, 30 cycles of 95° C for 20 s, 58° C for 40 s, and 72° C for 90 s and a final elongation at 72° C for 5 min.
In addition, 5 g of fecal sample from each animal was pooled by pen. Pooled samples were cultured for E. coli O157, Salmonella, and commensal E. coli by procedures adapted from Barkocy-Gallagher et al. (2002). Briefly, 10 g of feces was enriched in 90 mL of tryptic soy broth (TSB) for 2 h at 25° C, 6 h at 42° C, and overnight at 4° C, followed by anti-O157 IMS and identification as described above. For the isolation of generic E. coli, 50μ L of enriched TSB was plated onto MacConkey agar and incubated at 37° C for 24 h. Up to 2 morphologically distinct, lactose-fermenting colonies were picked and replated onto BAP. After overnight growth at 37° C, colonies were tested for indole production, and citrate tubes were inoculated to examine citrate utilization. For the isolation of Salmonella, 10 mL of enriched TSB was inoculated to 90 mL of tetrathionate broth and incubated 24 h at 37° C. One milliliter of tetrathionate broth was subjected to anti-Salmonella IMS, followed by enrichment of 100 μ L in 10 mL of Rappaport-Vassiliadis broth. The Rappaport-Vassiliadis broth was incubated at 42° C for 16 to 18 h, and 50 μ L was plated on both Hektoen enteric agar supplemented with novobiocin (15 mg/L) and brilliant green agar with sulfadiazine (Sigma-Aldrich, St. Louis, MO). The plates were incubated overnight at 37° C; up to 3 isolates from each plate were picked and replated onto BAP for overnight growth and latex agglutination. Species of Salmonella were serogrouped and sent to the National Veterinary Services Laboratories (Ames, IA) for serotyping.
Enterococcus isolates were obtained by diluting 1 g of pooled fecal sample in 10 mL of PBS and plating 50 μ L onto M-Enterococcus agar. After 24 h of growth at 37° C, characteristic colonies (metallic pink and pinpoint) were picked and plated on BAP. After overnight growth at 37° C, isolates were inoculated in 100 μ L of Enterococcosel broth and incubated at 37° C for 4 h to test for esculin hydrolysis. A commercial kit (20Strep API, bioMe ′rieux) was used for genus confirmation. All bacterial isolates were stored at − 80° C for further use.
Antimicrobial Susceptibility Testing
Antimicrobial susceptibility patterns of all isolates were determined by microbroth dilution using the Sensititre automated antimicrobial system (Trek Diagnostic Systems, Cleveland, OH). Minimal inhibitory concentrations (μ g/mL) were determined for antimicrobials in the standard bovine clinical panel (BOPO-1F) for Gram-negative organisms. The antimicrobials and the greatest concentrations evaluated in the Gram-negative panel were as follows: ceftiofur (8 μ g/mL), erythromycin (4 μ g/mL), chlortetracycline (8 μ g/mL), florfenicol (8 μ g/mL), penicillin (8 μ g/mL), ampicillin (16 μ g/mL), danofloxacin (1 μ g/mL), sulfadimethoxine (256 μ g/mL), neomycin (32 μ g/mL), sulfachloropyridazine (256 μ g/mL), tylosin tartrate (20 μ g/mL), sulfathiazole (256 μ g/mL), spectinomycin (64 μ g/mL), tilmicosin (32 μ g/mL), clindamycin (2 μ g/mL), tiamulin (32 μ g/mL), enrofloxacin (2 μ g/mL), trimethoprim/sulfamethoxazole (2/38 μ g/mL), gentamicin (8 μ g/mL), and oxytetracycline (8 μ g/mL). Enterococcus spp. were tested for antimicrobial susceptibility with the Gram-positive National Antimicrobial Resistance Monitoring System panel (CMV1AGPF), which included the following: bacitracin (128 μ g/mL), chloramphenicol (32 μ g/mL), erythromycin (8 μ g/mL), flavomycin (32 μ g/mL), penicillin (16 μ g/mL), daptomycin (16 μ g/mL), quinupristin/dalfopristin (32 μ g/mL), tetracycline (32 μ g/mL), vancomycin (32 μ g/mL), lincomycin (32 μ g/mL), tylosin tartrate (32 μ g/mL), ciprofloxacin (4 μ g/mL), linezolid (8 μ g/mL), nitrofurantoin (64 μ g/mL), kanamycin (1,024 μ g/mL), gentamicin (1,024 μ g/mL), and streptomycin (2,048 μ g/mL).
Quantification of Antimicrobial Resistance Genes
Real-time PCR was used to quantify specific antimicrobial resistance genes, ermB and tetM. To accurately determine the concentration of these genes, standard curves were first generated using known concentrations of plasmid vectors with the antimicrobial resistance gene inserted. The quantification of genes in the pen floor fecal samples was conducted using the standardized real-time PCR after DNA had been extracted with a commerical kit.
Bacterial Strains and Plasmids.
Enterococcus faecalis MMH594 (L. Zurek, Kansas State University) was used as a template to amplify an erythromycin-resistant gene (ermBCT) and a 16S rDNA gene (EUB). These fragments were cloned into pCRII-TOPO vector (Invitrogen, Carlsbad, CA) and transformed into One Shot Top10 E. coli cells (Invitrogen). An E. coli isolate (L. Zurek) containing pFD310, which carries tetracycline-, erythromycin-, and ampicillin-resistant markers (Smith et al., 1992) was used to amplify the tetM gene.
Primers were designed or modified using Integrated DNA Technology PrimerQuest software (Coralville, IA; Table 2). The degenerate primer used to amplify ermBCT (product size 404 bp; Jost et al., 2004) for cloning procedures was redesigned for real-time PCR analysis to more precisely target the ermB gene (product size 175 bp; this study).
PCR Running Conditions.
The running conditions for PCR reactions were as follows: initial denaturation of 3 min at 94° C, 36 cycles of 30 s at 94° C, 30 s of annealing (Table 2), 30 s at 72° C, and a final elongation of 2 min at 72° C with Taq DNA polymerase (Promega, Madison, WI). Real-time PCR running conditions were identical for each gene, with the addition of a 15-min, 95 ° C initial activation and an increase to 40 cycles.
Products from ermBCT, tetM, and EUB PCR were purified with Wizard SV Gel and PCR Clean-up System (Promega) and cloned into pCRII-TOPO. Transformation was performed by heat shock into E. coli with plating onto Luria-Bertani agar with ampicillin (100 μ g/mL) and 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (40 mg/mL). Plasmids were purified from the white colonies using the Wizard Plus SV Minipreps DNA Purification System (Promega). The concentration of DNA was determined using a nanodrop spectrophotometer (Nanodrop ND 1000, Nanodrop Technologies, Wilmington, DE) and confirmed with EcorI (Promega) digestions and sequencing (Beckman-Coulter CEQ8000 Genetic Analysis System, Fullerton, CA).
Real-Time PCR Standardization.
Purified plasmid samples were serially diluted 10-fold. Five microliters of plasmid sample was added to Absolute QPCR SYBR Green Mix (Abgene, Epsom, UK) for a 25-μ L reaction. Real-time PCR was performed according to the instructions of the manufacturer. Reliable gene products were available between 1 × 108 molecules/μ L and 1 × 103 molecules/μ L for EUB and between 1 × 109 molecules/μ L and 1 × 103 molecules/μ L for ermB. Dilutions of tetM were reliable between 7.35 × 108 molecules/μ L and 7.35 × 102 molecules/μ L.
Extraction of DNA from Fecal Samples.
Pooled fecal samples (180 to 220 mg) that were frozen at − 80° C immediately after collection were thawed on ice, and DNA was extracted using the QIAamp DNA stool mini kit according to the directions of the manufacturer (Qiagen, Valencia, CA).
Real-Time PCR on Fecal Samples.
Real-time PCR was performed on 108 fecal samples to target ermB, tetM, and EUB genes. Melting curves were generated (0.5 increments) to test uniformity and singularity of products. Electrophoresis on 1.2% agarose gels, cloning into pCRII-TOPO, and sequencing were used to confirm the composition and correctness of randomly selected PCR end products. The specific log copy number of genes was computed using the standardization curves.
The prevalence of E. coli O157, estimated as the proportion of positive samples per pen, was analyzed with a pen-level logit model to test for effect of diet (with or without distillers grains), antimicrobial feed additive (no antimicrobial, monensin only, or monensin and tylosin), sampling day, and interactions using PROC GEN-MOD (SAS Inst. Inc., Cary, NC; Agresti, 1996). The probability of detecting foodborne pathogens (E. coli O157 and Salmonella spp.) from pooled pen samples was analyzed using pen-level logistic regression models in PROC GENMOD with diet, antimicrobial, and day as effects.
The prevalence of resistance to antimicrobials in the susceptibility panels was evaluated for Enterococcus and generic E. coli isolates using pen-level logistic regression models as described above. For the analysis, isolates reported as intermediate were considered susceptible. Initially, univariate models were used to screen for the effect of day on resistance to each antimicrobial compound. Day was included in final multivariate regression models containing treatment effects if P ≤ 0.15 (Dohoo et al., 2003). Type 3 likelihood ratio statistics were used to test for treatment effects (P < 0.05), and least squares means tests were used to separate probability estimates when significant effects were observed. The MIXED procedure of SAS was used to evaluate mean differences in the log copy number of ermB and tetM, with the copy number of the EUB control target as a covariate in the model. Identical effects as above were examined.
Prevalence of E. coli O157 and Salmonella
The overall prevalence of E. coli O157 in fecal samples collected from individual animals was 9.1% (67 of 738). Prevalence on d 122 was 8.4% (31 of 371) and d 136 was 9.8% (36 of 367). The stx2 gene was present in 94.0% of isolates (63 of 67), whereas the stx1 gene was present in 22.4% of isolates (15 of 67). Comparison of treatment groups revealed a significant WDGS effect (P = 0.02); however, a significant WDGS × sampling day interaction was also observed (P = 0.02). The WDGS effect was significant (P < 0.001) on d 122 but not on d 136 (P > 0.2; Figure 1). There was no feed antimicrobial effect (P > 0.8) or feed antimicrobial × WDGS interaction (P = 0.19) on the fecal prevalence of E. coli O157. Prevalence of E. coli O157 in pooled fecal samples collected from pens was 19.4% (21 of 108) with 85.7% (18 of 21) containing the stx2 gene and 23.8% (5 of 21) containing the stx1 gene. There were no effects of WDGS (P > 0.4), feed antimicrobial (P > 0.4), or day (P > 0.4) on prevalence of E. coli O157 in pooled pen samples. Salmonella prevalence in pooled pen samples was 19.4% (21 of 108) with no significant treatment effects. Sero-types recovered were Montevideo (16 isolates) and Muenster (5 isolates).
Antimicrobial susceptibility patterns of all bacterial isolates are shown in Table 3. All Salmonella isolates (n = 21) were susceptible to gentamicin, neomycin, tetracyclines, and ampicillin, whereas they were resistant to clindamycin and macrolides. Escherichia coli O157 isolates (n = 21) were susceptible to aminoglycosides with only 1 isolate resistant to neomycin. The most frequent resistance in E. coli O157 isolates was to tetra- cyclines and sulfonamides. Chlortetracycline resistance was displayed in 33.3% of isolates, whereas 47.6% of isolates were resistant to oxytetracycline. Sulfonamide resistance ranged from 38.1% for sulfachlorpyridazine and sulfadimethoxine to 42.9% for sulfathiazole.
Similar to E. coli O157 and Salmonella serotypes, all generic E. coli isolates were largely susceptible to aminoglycosides. A total of 188 generic E. coli isolates were tested, and only 1 isolate exhibited resistance to gentamicin, whereas 7 isolates were resistant to neomycin. Feed antimicrobial treatment affected the susceptibility of E. coli. Cattle fed monensin and tylosin had a decreased proportion of isolates resistant to chlortetracycline (P = 0.008) and oxytetracycline (P = 0.002) than cattle fed no antimicrobials. Cattle fed monensin and tylosin also had a decreased proportion (P = 0.04) of isolates resistant to oxytetracycline than cattle fed monensin only. The resistance patterns to sulfonamides in generic E. coli were impacted by the inclusion of 25% WDGS. Cattle without WDGS had a decreased proportion of resistance in sulfachloropyridazine (P = 0.02), sulfathiazole (P = 0.02), and sulfadimethoxine (P = 0.02) than cattle fed diets with WDGS.
Among Enterococcus isolates (n = 96), 57.3% were resistant to macrolides (erythromycin and tylosin). Enterococcus isolates from monensinfed (P = 0.01) or monensin and tylosinfed (P = 0.01) cattle were more resistant to macrolides compared with isolates from cattle fed no antimicrobials (Figure 2). In cattle fed monensin and tylosin, 75% of Enterococcus isolates displayed resistance to tylosin and erythromycin, and 66% of Enterococcus isolates from cattle fed only monensin were resistant to these antimicrobials. Only 36.8% of Enterococcus isolates were resistant in cattle fed no antimicrobials. There was no WDGS effect on the resistance or susceptibility of fecal enterococci to macrolides (data not shown). However, the resistance to several antimicrobials appeared to be influenced by WDGS treatment. Fewer Enterococcus isolates showed resistance to flavomycin in animals fed WDGS than those fed steam-flaked corn only (P = 0.01). However, with quinupristin/dalfopristin there was a tendency for a greater proportion of resistant Enterococcus in cattle fed WDGS (P = 0.08).
Quantification of Fecal ermB and tetM Genes
The tetM gene was detected in every fecal sample, whereas the ermB gene was detected in 94 of 108 fecal samples (87%). Including 0 or 25% of WDGS did not affect the concentration of either ermB or tetM genes in the feces. There was no antimicrobial effect on the concentration of either ermB or tetM in the feces.
Distillers grains with solubles with highly digestible bran, high ruminal escape protein, and energydense germ fractions of the corn are well-suited as a ruminant feed (Lodge et al., 1997; Kleinschmit et al., 2006). In the production of distillers grains, antimicrobials, such as penicillin G, streptomycin, tetracycline, monensin, and virginiamycin, are used to suppress the bacterial growth (Day et al., 1954; Aquarone, 1960; Narendranath et al., 2000). It is generally believed that temperatures achieved during the distillation are sufficient to destroy these antimicrobials (Shurson, 2005). There are no data to suggest that the coproduct has antibiotic residue. However, even if there is no residue, there is a possibility of genetic elements for antibiotic resistance being present in the coproduct. There is evidence that stored wet distillers grains have a dense population of lactobacilli (Pederson et al., 2004), which could be a source of antibiotic resistance genes.
In our study, both susceptibility and real-time PCR results implied that WDGS was not associated with antimicrobial resistance. The proportion of Enterococcus isolates resistant to quinupristin/dalfopristin, a streptogramin antimicrobial compound like virgin-iamycin, was greater in cattle fed WDGS; however, this was only a trend and was not significant. No evidence of increased penicillin resistance was observed in Enterococcus isolates, which may have occurred if residual antibiotics remained. In addition, quantification of 2 resistance elements, ermB and tetM, by PCR revealed no differences in concentration between WDGS-fed cattle and cattle fed no WDGS.
Additional antimicrobial susceptibility data from our study showed that monensin and tylosin use were associated with increased macrolide resistance in Enterococcus species, which was consistent with other studies reporting tylosin use in pigs (Aarestrup et al., 2001; Jackson et al., 2004). The most common mode of resistance to macrolides in enterococci of animal origin is the ermB gene (Jensen et al., 1999), which causes a 23S rRNA methylation (Roberts et al., 1999), thereby rendering the ribosomes tolerant to erythromycin. Furthermore, a previous study has shown that ermB and tetM can coexist in a single transposon (De Leener et al., 2004). Although our results showed the presence of both ermB and tetM resistance genes in every pooled pen sample tested, the concentration of each gene varied in samples.
Antimicrobial feed additives did not have a significant effect on ermB genes in feces; however, the results did not correlate with the macrolide susceptibility in Enterococcus isolates. Although macrolide resistance in enterococci isolates was more prevalent, the concentrations of ermB genes from pooled fecal samples were unchanged compared with cattle fed no antimicrobials. Because other genes can provide resistance to macrolides (Schwarz et al., 2006), all enterococci isolated in this study were screened by PCR and confirmed to contain the ermB gene. These results suggest that there might be additional mechanisms affecting macrolide resistance in these isolates.
An interesting observation of this study was the association between feeding WDGS and fecal prevalence of E. coli O157. The prevalence of E. coli O157 in cattle fed WDGS was greater compared with those not fed WDGS when individual cattle were sampled; however, this was only statistically significant on 1 sample day. Pooled pen samples did not reveal any difference in the prevalence of E. coli O157 or Salmonella species. The difference between individual animal samples and pooled pen samples is not surprising, because the sensitivity in detection methods decreases when E. coli O157-positive fecal samples are pooled with E. coli O157-negative samples (Sanderson et al., 2005). In addition, the enrichment and isolation procedures were slightly different between the 2 samples. The role of diet on fecal shedding of E. coli O157 in cattle has been well studied (Buchko et al., 2000; Callaway et al., 2003; Berg et al., 2004; Van Baale et al., 2004). Dewell et al. (2005) reported that fecal samples from feedlot pens fed brewers grains, a similar fermentative product to distillers grains, were 6 times more likely to be positive for E. coli O157 than feedlot pens not fed brewers grains. Our study is the first report of an association between feeding distillers grains and prevalence of E. coli O157 in cattle. The reason for this possible association is not known. We hypothesize that the greater prevalence associated with distillers grains or brewers grains is possibly due to the different hindgut environment created by the diets. Research has shown that the primary colonization site of E. coli O157 is the hind-gut of cattle (Grauke et al., 2002; Naylor et al., 2003; Van Baale et al., 2004). The feeding of a coproduct with less starch results in greater ruminal pH than corn diet without distillers grain (Firkins et al., 1985; Lodge et al., 1997). Additionally, replacing corn with a highly digestible fiber source in distillers grain may cause a shift in digestion from the rumen to the hindgut (Ham et al., 1994). The greater fat content in distillers grain, in addition to providing more energy, could have an effect on ruminal fermentation and rumen microbial population. Montgomery et al. (2005) showed that feeding corn germ to cattle reduced the incidence of liver abscess; one reason may be that fatty acids in the germ have an antibacterial effect, thereby suppressing the growth of Fusobacterium necrophorum, the causative agent of liver abscesses. Increased supply of fiber and possibly protein and germ in cattle fed distillers grain could have a significant impact on hindgut fermentation. Possibly, the altered hindgut environment is more conducive to E. coli O157 colonization.
Monensin and tylosin in cattle diets had no effect on E. coli O157 prevalence. The effect of monensin on the prevalence of E. coli O157 has been studied, and the results are conflicting. A positive association between feeding ionophores and the prevalence of E. coli O157 was noted in dairy cattle (Herriott et al., 1998). This was later contradicted by 2 studies that showed no effect of short-term monensin feeding on fecal shedding of E. coli O157 or Salmonella Typhimurium in lambs (Edrington et al., 2003) and a decreased duration of E. coli O157 shedding in forage-fed steers supplemented monensin (Van Baale et al., 2004). Recently, McAllister et al. (2006) showed no association among monensin, tylosin, or monensin and tylosin on the prevalence of E. coli O157 in fecal samples of orally challenged animals.
Several limitations to our study include the small number of sample collection days, low statistical power to detect differences, and lack of baseline prevalence and antimicrobial susceptibility estimates. Effects that may be attributable to study duration may not have been seen with only 2 collection days. Additionally, the low prevalence of E. coli O157 and Salmonella in pooled-pen samples reduced our ability to statistically analyze antimicrobial susceptibility in these isolates, although our resistance patterns to tetracyclines and sulfonamides were in agreement with previous studies (Meng et al., 1998; Galland et al., 2001; Schroeder et al., 2002; Fitzgerald et al., 2003).
In conclusion, additional work should be conducted to confirm the increased prevalence of E. coli O157 when animals are fed WDGS. Feeding antimicrobials to cattle was associated with a greater prevalence of resistance toward related antimicrobials in commensal enterococci when individual isolates were characterized. However, quantification of resistance genes in cattle feces, regardless of bacterial species, may provide a better means to assess the impact of production practices on the dynamics of antimicrobial-resistant bacterial populations. Antimicrobial feed additives did not appear to increase the presence or concentration of either ermB or tetM elements in cattle.
|———% of DM———|
|Corn wet distillers grains with solubles||—||25.0|
|Corn steep liquor||5.0||—|
|Urea, 46% N||1.0||—|
|ermBCT||F 5′ GAAATTGGAACAGGTAAAGG 3′||43||404||Modified from|
|R 5′ TTTACTTTTGGTTTAGGATG 3′||Jost et al., 2004|
|ermB||F 5′ GAATCCTTCTTCAACAATCA 3′||45||175||This study|
|R 5′ ACTGAACATTCGTGTCACTT 3′|
|tetM||F 5′ CTGTTGAACCGAGTAAACCT 3′||48||156||This study|
|R 5′ GCACTAATCACTTCCATTTG 3′|
|EUB||F 5′ TGGAGCATGTGGTTTAATTCGA 3′||50||159||Yang et al., 2002|
|R 5′ TGCGGGACTTAACCCAACA 3′|