Impact of Three Ampicillin Dosage Regimens on Selection of Ampicillin Resistance in Enterobacteriaceae and Excretion of bla_TEM Genes in Swine Feces

Eighteen 7-week-old, commercial, healthy piglets that had never received antibiotics were used. They were housed separately in individual pens throughout all the experiments. A meal was given twice daily, and water was provided ad libitum. Ampicillin was administered once a day at 20 mg/kg of body weight for 7 days (from day 0 to day 6) following three modalities: the intramuscular route, the oral route in fasted pigs, or the oral route in fed pigs. The design schedule consisted of three successive series of six animals receiving ampicillin treatments as follows: intramuscular (n = 2), oral route under fed conditions (n = 2), and control without treatment (n = 2) in the first series; intramuscular (n = 2), oral route under fasted conditions (n = 2), and control without treatment (n = 2) in the second series; oral route under fed conditions (n = 2), oral route under fasted conditions (n = 2), and control without treatment (n = 2) in the third series. Six pigs were used in the control group and four pigs in each ampicillin treatment group. Intramuscular injections of ampicillin sodium (Ampicilline Cadril; Laboratory Coophavet, Ancenis, France) were administered in the neck. For oral routes, a medicinal premix (Ampicilline 80 Porc Franvet; Laboratory Franvet, Segré, France) was dissolved in water and administered by gastric intubation. Fasted swine were starved 16 h before ampicillin administration and fed 4 h after ampicillin administration. Ampicillin was administered to fed pigs at the end of their morning meal.

For phenotypic evaluation of ampicillin resistance, fecal samples were taken from each pig, by digital manipulation or immediately after spontaneous defecation, at days 0 (before ampicillin administration), 1, 4, and 7. The samples were immediately transferred to the laboratory, and the Enterobacteriaceae were counted. For the quantification of bla_TEM genes in feces by real-time PCR, feces of each pig were collected two or three times before the treatment. The value given for day 0 is the mean of these samplings. Feces were then collected each day from day 1 to day 7. Samples were obtained as already described. Two hundred milligrams of feces from each sample was frozen in liquid nitrogen and stored at −80°C until assayed.

Feces (5 g) from each pig were homogenized with 45 ml of peptone water, including 30% glycerol, with a BagMixer (Interscience, St. Nom, France). Tenfold serial dilutions of the filtrate were prepared, and 100-μl samples of the dilutions were spread on MacConkey plates (AEB 151602; AES, Ker Lann, France) containing 0 and 16 μg/ml of ampicillin. MacConkey agar is classically used for selective growth of Enterobacteriaceae (7, 8, 11, 30). Enterobacteriaceae growing in the presence of 16 μg/ml of ampicillin were classified as resistant. This concentration corresponds to the MIC breakpoint value (MIC of ≥32 μg/ml) proposed by the CLSI (23) and the French Society for Microbiology (http://www.sfm.asso.fr ). The plates were incubated at 37°C for 24 h. Enterobacteriaceae counts from both plates were used to calculate the percentage of resistant Enterobacteriaceae at each sampling time.

For each sample, 20 colonies were randomly picked on the MacConkey plates without ampicillin and stored at −80°C until assayed. These colonies were considered to be E. coli on the basis of β-glucuronidase production using TBX agar (tryptone bile X-glucuronide agar; AES Laboratoire, Bruz, France) (14). Only a few colonies were β-glucuronidase negative. All β-glucuronidase-negative isolates and a portion of β-glucuronidase-positive isolates were tested by the API 20E Enterobacteriaceae identification system (bioMérieux, Marcy l'Etoile, France) to confirm their identification. For MIC determination, ampicillin susceptibility was tested by a microdilution broth dilution method according to the recommendations reported by the CLSI (22). The control strain was E. coli ATCC 25922.

E. coli JS238(pOFX326), the plasmid of which carries a monocopy of the target gene bla_TEM-1, was used to optimize real-time PCR, assess sensitivity, and generate quantification standards. The strain was cultured in Mueller-Hinton broth containing ampicillin at the concentration of 50 μg/ml at 37°C overnight.

pOFX326 was purified with the QIAprep Spin Miniprep kit (QIAGEN, Hilden, Germany). Quality was assessed by migration on gel electrophoresis in 1% agarose after digestion with HindIII, and concentration was assessed by spectrophotometry at 260 nm. The QIAamp DNA stool kit (QIAGEN, Hilden, Germany) was used to extract DNA from feces according to the manufacturer's recommendations. For each series of extractions, a positive control and a negative control were coextracted and subjected to real-time PCR.

The PCR primers were designed with Primer 3 and Oligo Analyser. The specificity of the sequence was further checked against all the available GenBank DNA sequences. The forward and reverse primers chosen for bla_TEM gene quantification were 5′-TTCCTGTTTTTGCTCACCCAG-3′ and 5′-CTCAAGGATCTTACCGCTGTTG-3′, respectively. These primers amplify a 112-bp segment of the bla_TEM-1D gene (GenBank accession number AF 1888200) from nucleotide positions 270 to 382. A 100% homology was demonstrated with 130 bla_TEM genes for which the nucleotide sequence was available, except for TEM-60.

The PCR amplification was performed in a 25-μl reaction mixture with a SYBR green PCR core reagent kit (PerkinElmer Biosystems, Foster City, CA). The reaction mixture contained 5 μl of test DNA solution, 2.5 μl of 10× SYBR green PCR buffer, 1.6 μl of a deoxynucleoside triphosphate solution (2.5 mM each of dATP, dCTP, and dGTP and 5 mM of dUTP), 0.25 μl of each primer (20 μM), 4 μl of 25 mM MgCl₂, 11.275 μl of ultrapure water (Qbiogene, Montréal, Canada), and 0.125 μl of AmpliTaq Gold DNA polymerase, LD (5 U/μl) (PerkinElmer Biosystems). Amplification was performed using a GeneAmp PCR system 5700 thermocycler (PerkinElmer Biosystems) with the following conditions: 95°C for 10 min followed by 45 cycles of 15 seconds at 95°C and 1 min at 60°C. A standard curve with three replicates of the control plasmid pOFX326 diluted in Tris-EDTA buffer was generated for each PCR assay. All sample PCRs were done in duplicate. The samples were checked for the absence of background levels of PCR-inhibiting compounds by spiking DNA extracted from the samples with target DNA and subjecting these spiked DNA samples to real-time PCR both undiluted and diluted (1:10).

The impact of DNA fecal environment on amplification sensitivity and performance was assessed by comparing standard curves obtained with the control plasmid diluted in Tris-EDTA or in swine fecal DNA. The accuracy and reproducibility of the entire assay (from DNA extraction to real-time PCR analysis) were measured by spiking 200 mg of feces with an overnight culture of E. coli JS238(pOFX326). Five aliquots per day were subjected to DNA extraction on three different days. The extraction recovery rate was calculated and checked to be the same for different concentrations of bla_TEM genes in feces by spiking fecal samples with 10-fold serial dilutions of an overnight culture of E. coli JS238(pOFX326). These samples were subjected to DNA extraction and then to real-time PCR.

Statistical analysis was performed using Systat 10 (Systat Software Inc., Richmond, CA). Changes in the level of ampicillin resistance were analyzed using a generalized linear mixed-effects model with the following equation:where Y_ijk is the measure of resistance for pig k undergoing ampicillin administration with modality i at day j, μ the overall mean, M_i the differential effect of treatment i, D_j the differential effect of day j, M*D_ij the corresponding interaction, A_k|_Mi the differential effect of animal k nested within treatment i, and ε_ijk an error term. Y, the measure of resistance, was monitored in various ways. For the phenotypic evaluation of resistance, Y was the log-transformed percentage of the resistant Enterobacteriaceae population or the log-transformed percentage of E. coli isolates with MICs of >16 μg/ml. For the genotypic evaluation, Y was the log-transformed quantity of bla_TEM genes. Multiple comparisons were performed using the Tukey test. The selected level of significance was a P value of <0.05.

In order to construct calibration curves and determine the specificity and sensitivity of the primers in swine fecal DNA, the control plasmid pOFX326 was diluted in Tris-EDTA buffer and in swine fecal DNA. Each dilution was subjected to real-time PCR, and the amplifications were repeated four times. Melting curve analysis of the control plasmid, diluted either in Tris-EDTA buffer or in swine fecal DNA, showed specific amplification with a PCR amplicon at a melting temperature of 81°C (data not shown). Despite the use of highly purified AmpliTaq Gold DNA polymerase, analysis of the ultrapure water melting curves revealed contamination and thus restricted the PCR quantification limit (data not shown). Figure 1 shows the two standard curves: the relation between C_T (cycle threshold) values and the logarithm of bla_TEM concentration was linear from 10 to 10⁶ copies/μl. The determination coefficients (r²) were 0.996 in Tris-EDTA and 0.985 in swine fecal DNA. The closeness between these standard curves indicated that the complex fecal DNA environment did not affect amplification sensitivity or performance. The intra- and interday coefficients of variation of the entire assay (from DNA extraction to real-time PCR analysis) were 16.7% and 18.2%, respectively. The extraction recovery rate was 70 to 113% (mean, 98.5%). This was checked to be the same for different concentrations of bla_TEM genes in feces by spiking fecal samples with 10-fold serial dilutions of an overnight culture of E. coli JS238(pOFX326). The correlation between bla_TEM copy number/gram of feces and dilution factors of the JS238(pOFX326) solution was high (r² = 0.904). Thus, the extraction yields for different concentrations of E. coli JS238(pOFX326) in feces were similar. Overall, data demonstrated that this PCR analysis was suitable for quantification of bla_TEM genes in swine feces from 10 to 10⁶ copies/μl of eluate of extracted DNA, which corresponds to 10⁴ to 10⁹ copies/g of feces.

Average percentages of ampicillin-resistant Enterobacteriaceae for each treatment group are given in Fig. 2a. The average percentage of resistant Enterobacteriaceae ranged from 0.9% to 12% before ampicillin administration. On the first day of treatment, it rose to 26% for the intramuscular route and to 40% and 49% for the oral routes in fed and fasted pigs, respectively. By days 4 and 7, the level of resistance exceeded 50% in all treated groups. In contrast, the level of resistance in the control group remained below 13% at all times. Treated animals excreted significantly higher percentages of resistant Enterobacteriaceae than did the control group (P < 0.05). However, no significant differences were observed among the three modes of drug administration (P > 0.05). Furthermore, Fig. 2a shows the high interindividual variability within each group.

Ampicillin resistance was also monitored from the percentage of resistant E. coli isolates for each treatment group (Fig. 2b). The average percentage of resistant E. coli isolates ranged from 1% to 38% before ampicillin administration. At day 1 of treatment, about 70% of isolates were resistant, whatever the mode of drug administration. By days 4 and 7, nearly all the isolates, whatever the dosage regimen, were resistant. In contrast, the percentages of resistant E. coli isolates remained below 36% in the control group. Statistical analysis indicated that oral administration in fed pigs led to a higher fecal excretion of resistant E. coli than in control pigs (P < 0.05). Results for the two other dosage regimens did not differ significantly from those of the control group due to the great heterogeneity of the control group data (P > 0.05). High interindividual variability also existed within each ampicillin-treated group.

Ampicillin resistance in feces was measured by bla_TEM gene quantification using a validated PCR assay. bla_TEM gene copy numbers per gram of wet feces were measured on each day of treatment for each pig (Fig. 3). The baseline values for all pigs were below 10⁷ copies/g of feces. bla_TEM quantities increased after ampicillin administration. The between-day fluctuations for a given animal were large. The bla_TEM quantities for the oral routes fluctuated between 10⁷ and 10⁹ copies/g of feces, but only between 10⁵ and 10⁸ copies/g of feces for the intramuscular route. Two fed pigs treated by oral route excreted the highest bla_TEM quantities, with values above 10⁹ copies/g of feces. The bla_TEM quantities for the control group were lower than those of the three ampicillin-treated groups and fluctuated between 10⁴ and 10⁶ copies/g of feces.

Figure 4 shows the mean quantities of bla_TEM genes for each dosage regimen. Statistical analysis indicated that all ampicillin treatments had a significant effect on the excretion of bla_TEM genes compared to that in the control group (P < 0.001). Moreover, oral administration in fed pigs led to a significantly higher excretion of bla_TEM genes than intramuscular administration (P < 0.05).

We investigated the agreement between resistant Enterobacteriaceae counts and bla_TEM concentrations. Figure 5 shows a significant correlation (r² = 0.67) between the quantities of bla_TEM genes and the counts of ampicillin-resistant Enterobacteriaceae.