The pig production unit consisted of approximately 40 sows and their piglets and was located in Berlin, Germany. Pigs (hybrids of Deutsche Landrasse and Duroc) were stalled in groups and fed according to the National Research Council recommendations 18. The basal diet for sows and piglets mainly comprised barley and wheat, and wheat and soybean meal respectively. Sows were fed restrictively according to their body mass and litter size. Piglets had ad libitum access to feed. Sows and piglets had ad libitum access to water. The administration of antimicrobial substances was prohibited for both sows and piglets for at least 3 months prior to the Resistance Status Study trial. The sow of the Colonization Study received no antimicrobial substances 14 months prior to the trial except of two applications of Cefquinome (Cobactan^®, Veterinaria AG, Zürich, Switzerland) 10 months prior to the trial. The piglets of the Colonization Study received no antimicrobial substances during the examinations. Piglets were weaned at day 28 of age. Piglets were reared after weaning in segregated pens of flat-deck batteries. The health status of the pigs was monitored by control of the general animal condition and fecal consistency. The study was approved by the local animal welfare committee of the "Landesamt für Arbeitsschutz, Gesundheitsschutz und technische Sicherheit" Berlin, Germany (No. G0037/02).

To determine the general antimicrobial resistance status of intestinal E. coli from the pig production unit, E. coli were isolated from 15 clinically healthy piglets (age 56 days) from 12 different sows. This included the isolation from digesta as well as mucosa samples from the jejunum and colon. E. coli-like isolates [pink colonies on CHROM agar orientation plates 19, purple on MacConkey, blue or green on GASSNER agar plates] were assigned to clones by macrorestriction analysis (see below). Clones were finally identified as E. coli using standard methods 20. The detailed isolation protocol has recently been published 21. Forty nine isolates, each representing one E. coli clone, were further tested for antimicrobial susceptibility and resistance genes as mentioned below. These E. coli clones comprised dominant as well as minor clones 22 from the jejunum as well as colon.

To compare the intestinal colonization success of resistant and susceptible E. coli, the transfer of E. coli from one sow to five of her piglets was studied. The sow and the five piglets were from the same pig production unit like the piglets of the Resistance Status Study. The Colonization Study was carried out over a period of eight weeks, with samples being taken weekly. Sampling was started with sow rectal content taken on the day of birth of the piglets; sampling from the five piglets was started one week later (when piglets were seven days old); sow feces were sampled one week prior to weaning (when piglets were 21 days old); and the last piglet samples were taken at the age of 56 days. Additionally, teats and the bottom of the box were wiped off with sterile cotton swabs. In summary, a total of 1200 E. coli isolates had been isolated: 80 rectal isolates from one sow (over a period of four weeks), 80 isolates from the teats of the sow (four weeks), 80 isolates from the bottom of the sow box (four weeks) and 160 isolates from each of five piglets (eight weeks). Before the piglets were transported to their separation box after weaning, the bottom of the box was wiped off with sterile cotton swabs. Serial dilutions of samples were plated on MacConkey (Oxoid, Hampshire, UK) agar plates and incubated for 18 h at 37°C. Twenty colonies per sample and animal were randomly chosen and plated on GASSNER (Sifin, Berlin, Germany) and CHROM agar orientation (Chromagar, Paris, France) 19 plates. Enterobacteriaceae-like colonies (purple on MacConkey, blue or green on GASSNER and pink color on CHROM agar orientation plates) were pre-defined as E. coli.

Twenty E. coli colonies per sample were differentiated by macrorestriction analysis (PFGE) according to Hartley et al. (1979) who found that results were similar if they picked 20 colonies compared to 100 colonies to isolate both the predominant E. coli type and most of the minor types 23. PFGE was performed as previously described 24. Bacterial DNA was digested with 20 U XbaI (Promega, WI, USA) at 37°C overnight. DNA fragments were separated in a 1.2% pulsed-field agarose gel for 22 h at 6V, pulse 5-50. Evaluation of PFGE profiles for similarity was performed using Bionumerics software with the UPGMA method (unweighted pair group method with arithmetic mean) and Dice similarity indices (complete linkage; optimization, 1%; position tolerance, 1.3%; Applied Maths, Belgium). After macrorestriction analysis, each determined clone was additionally verified as E. coli using standard methods 20.

For the determination of the general antimicrobial resistance status of E. coli from the pig population (Resistance Status Study), 49 E. coli strains from domestic piglets were tested for susceptibility to the following antimicrobial agents by the microdilution broth method as recommended by the Clinical and Laboratory Standards Institute 25 (breakpoints for resistance are indicated in parentheses): ampicillin (≥32 μg/ml), amoxicillin/clavulanic acid (≥32/16 μg/ml), cephalothin (≥32 μg/ml), cefazolin (≥32 μg/ml), chloramphenicol (≥32 μg/ml), enrofloxacin (no breakpoint available), gentamicin (≥16 μg/ml), kanamycin (≥32 μg/ml), neomycin (≥32 μg/ml, 26), spectinomycin (no breakpoint available), streptomycin (no breakpoint available), tetracycline (≥16 μg/ml) and sulfamethoxazole (≥512 μg/ml). Although there were no breakpoints available for spectinomycin and streptomycin for further analysis we defined E. coli resistant to these antimicrobial agents if the MIC of spectinomycin to this E. coli isolate was ≥128 μg/ml and to streptomycin ≥64 μg/ml. We assumed that these concentrations can not be clinically achieved in pigs. To perform the tests, commercially acquired microtiter plates (Sensititre, MCS Diagnostics, UK) were used.

For the Colonization Study one representative isolate of each E. coli clone from the sow and five of her piglets was tested for antimicrobial resistance by disk diffusion according to the Clinical and Laboratory Standards Institute 25. Only the antimicrobial agents were included for which resistance was detected during the Resistance Status Study: ampicillin, chloramphenicol, kanamycin, neomycin, spectinomycin, streptomycin, tetracycline, sulfamethoxazole.

Resistant E. coli strains were tested via PCR for the presence of the respective resistance genes bla_TEM (resistance to ampicillin), catA1 and cmlA/B (resistance to chloramphenicol), aph(3')-Ia (resistance to kanamycin and neomycin), aadA (resistance to streptomycin-spectinomycin), strA/strB (resistance to streptomycin), tet(A), tet(B), tet(C) (resistance to tetracycline) as well as sul1, sul2 and sul3 (resistance to sulfonamides). Primers and PCR reaction conditions were performed as previously described 15272829303132.

Tested VAGs are listed in Table 1. Primers and PCR conditions were previously described 1733.

As we sampled all isolates during a short period of time from rectal contents of animals of one small, well-defined pig production unit, E. coli isolates which showed in their PFGE patterns not more than one band difference were defined as members of the same clone 34. We defined a solitary clone of the sow as a clone found only in one rectal sample from the sow. Consequently, we defined a solitary clone of one piglet as a clone found only in one rectal sample from this piglet. A dominant clone was defined as a clone which represented ≥ 50% of typed colonies in one sample, a minor clone was defined as a clone which represented ≤ 10% of typed colonies in one sample according to Schlager et al22. The statistical analysis was performed using the software SPSS 12.0 (SPSS Inc., Chicago, IL, U.S.A.). Hypotheses on associations between nominal parameters (antibiotic resistance, resistance genes, VAGs) were tested by application of Fisher's exact test. Hypotheses on relationships between the number of resistance phenotypes and genotypes and the number of VAGs were tested by analyzing the non-parametric Spearman rank correlation coefficient. Relationships between nominal and numerical parameters were tested by application of the Mann-Whitney-U test. P-values below an alpha = 0.05 were considered significant.

All animals were clinically healthy which was proved by monitoring the general animal condition and the fecal consistency. To define existing antimicrobial resistance, 49 E. coli clones from 15 piglets were tested to 15 antimicrobial agents by the broth microdilution method. Thirty-three clones (67.4% of clones) were resistant to at least one antimicrobial agent, with resistance to ampicillin (20.4%), chloramphenicol (2.0%), kanamycin (10.2%), neomycin (10.2%), spectinomycin (34.7%), streptomycin (51.0%), tetracycline (65.3%) and sulfamethoxazole (36.7%). A total of 13 different resistance profiles were detectable with six clones (12.2%) being resistant to one antimicrobial agent, five clones (10.2%) being resistant to two antimicrobial agents, five clones (10.2%) being resistant to three antimicrobial agents, eight clones (16.3%) resistant to four antimicrobial agents, six clones (12.2%) being resistant to five antimicrobial agents, one clone (2.0%) resistant to six antimicrobial agents, and two clones (4.1%) being resistant to seven antimicrobial agents. Sixteen clones (32.6%) were susceptible to all tested antimicrobial substances. Resistant E. coli carried the resistance genes bla_TEM, catA1, aph(3')-I, aadA, strA, strB, tet(A), tet(B), sul1, sul2, and/or sul3. In a single sulfonamide resistant isolate no gene coding for resistance to sulfamethoxazole was detected. Up to seven different resistance genes in one strain and a total of 18 different resistance gene profiles were detectable (see Additional file 1).

All animals remained clinically healthy throughout the Colonization Study which was proved by monitoring the general animal condition and fecal consistency. During the Colonization Study a total of 1200 E. coli isolates had been affiliated to clones via macrorestriction analysis. E. coli were not detectable in feed and drinking water.

A total of 44 clones were isolated from the five piglets, 34 clones from the sow rectal content, 28 clones from the teats and 23 clones from the bottom of the sow box. PFGE patterns varied considerably between single E. coli clones and indicated that clones were unrelated 34. These results have recently already been published in more detail 17. With one exception, all isolates of one clone did not differ in their PFGE band pattern. In the one exceptional case there was one clone which comprised eight isolates (one isolate from the teats of the sow, six isolates from one piglet, one isolate from another piglet) one of which had an additional 97 kb fragment. Isolates of several clones were found in samples of different sampling sizes resulting in a total of 68 different E. coli clones.

All 68 E. coli clones were tested to the 8 antimicrobial agents against which there was resistance present in this pig population as revealed by our Resistance Status Study. Tests were done by the disk diffusion method 25. Of the 68 clones, 49 clones (72.1% of all clones) were resistant to up to seven of the antimicrobial agents whereas 19 clones (27.9%) were susceptible to all tested antimicrobial agents. Clones were resistant to ampicillin (19.1%), chloramphenicol (7.4%), kanamycin (4.4%), neomycin (4.4%), streptomycin (32.4%), spectinomycin (20.6%), tetracycline (70.6%) and sulfamethoxazole (33.8%). A total of 17 different resistance profiles were detectable with 9 clones (13.2%) resistant to one antimicrobial agent, with 17 clones (25.0%) resistant to two antimicrobial agents, 13 clones (19.1%) resistant to three antimicrobial agents, four clones (5.9%) resistant to four antimicrobial agents, four clones (5.9%) resistant to five antimicrobial agents, one clone (1.5%) resistant to six antimicrobial agents and one clone (1.5%) resistant to seven antimicrobial agents (see Aditional file 2). Resistant E. coli carried the resistance genes bla_TEM, catA1, cmlA, aph(3')-I, aadA, strA, strB, tet(A), tet(B), sul1, sul2, and/or sul3. No respective resistance gene was detected for two isolates resistant to sulfamethoxazole and one isolate resistant to spectinomycin. Up to seven different resistance genes in one strain and a total of 31 different resistance gene profiles were identified. The detected resistance genes are included in Figure 1 (see also Additional file 2).

E. coli exclusively found in the sow (n = 17) were not significantly more or less resistant or carried significantly more or less resistance genes than E. coli exclusively isolated from piglets (n = 28). To test for the correlation between resistance phenotypes and genotypes and the ability of intestinal colonization of piglets, we included all 44 clones which were isolated from the piglets and analyzed the correlations between the sum of all resistance phenotypes and genotypes of each clone and numerical criteria describing colonization success. Additionally, effects of single resistance/resistance genes on nominal and numerical criteria relevant for colonization were tested. Nominal colonization criteria included the detection of one clone exclusively before or after weaning or at both time points (as weaning is thought to massively change the intestinal milieu), dominance and solitariness of a clone. Numerical criteria covered the maximum numbers of isolates of one clone in one sample (as we included 20 isolates of one sample there was a theoretical range between 0 and 20 isolates per clone); total numbers including all isolates from all samples of one clone (as we included 800 isolates from all piglets over the whole eight week period and found 44 clones a theoretical range between 1 and 757); numbers of colonized piglets with a specific clone (as we included five piglets a theoretical range between 0 and 5); numbers of time points of the detection of a specific clone (as we sampled over an eight week period a theoretical range between 1 and 8). As there were only very few clones positive, the following resistance phenotypes and genotypes were excluded from detailed statistical analysis: resistance to kanamycin, neomycin and chloramphenicol and the resistance genes cmlA, catA1 and aph(3')-Ia.

There were no significant associations between resistance/resistance genes and the time points of isolation of clones (before or after weaning or at both time points), the dominance of E. coli in one sample or the probability to be a solitary clone. Additionally, resistance phenotypes and genotypes did not significantly correlate with the numerical criteria "maximum numbers of isolates of one clone in one sample", "total numbers including all isolates from all samples of one clone", "numbers of colonized piglets with a specific clone" and "numbers of time points of the detection of a specific clone". Thus, resistant clones obviously had no significant advantage or disadvantage in competition with susceptible clones and clones with resistance to several antimicrobial agents had no significant advantage or disadvantage in competition with clones showing resistance to only one or two antimicrobial agents. Clones exhibiting antimicrobial resistance and carrying resistance genes had often even a slightly better colonization success than susceptible clones. Examples are pictured in Table 2. Values of numerical criteria were higher for clones with at least one resistance compared to susceptible clones and values of numerical criteria were higher for clones with resistance to three or more antimicrobial agents compared to clones with less than three resistance properties (Table 2). Clones which exhibited resistance to a specific antimicrobial agent had no significant colonization advantage compared to susceptible clones, but values of all four numerical criteria could also be higher for such resistant clones compared to susceptible clones as seen for ampicillin-resistant clones (Table 3).

As it was recently shown that VAGs typical for extraintestinal pathogenic E. coli might play a role for intestinal colonization 16 and thus might play a role for "cross-selection" we analyzed dependencies between antimicrobial resistance/resistance genes and ExPEC-typical VAGs. All clones had been already tested for the presence of 32 VAGs 17 (Figure 1). In our present study we related the occurrence of resistance/resistance genes with the occurrence of VAGs. There were no significant correlations between the absolute number of resistance/resistance genes and the absolute number of VAGs in one clone. Only the occurrence of the group of resistance genes bla_TEM and thus the occurrence of the resistance to ampicillin was positively associated with the absolute numbers of VAGs. Here, ampicillin-resistant E. coli clones had significantly more VAGs compared to ampicillin-susceptible E. coli clones (p < 0.05). Regarding single resistance genes, the following significant positive associations were observed: the occurrence of bla_TEM was associated with the occurrence of fyuA, irp2, iucD, sitD ep., cva, iss; the occurrence of tet(A) was associated with the occurrence of tsh; the occurrence of strA was associated with the occurrence of sitD ep There was one significant negative association: the occurrence of tet(B) was negatively associated with the occurrence of mat. Here, tet(B)-positive clones carried significantly less frequent the gene mat compared to tet(B)-negative clones (p < 0.05).

Both data from the Resistance Status Study as well as from the Colonization Study were pooled. There were significant positive associations between ampicillin and sulfamethoxazole (p < 0.01), spectinomycin and streptomycin (p < 0.001), tetracycline and streptomycin (p < 0.001), tetracycline and spectinomycin (p < 0.01) and tetracycline and sulfamethoxazole (p < 0.005). Additionally, there were significant positive associations between the genes bla_TEM and sul2 (p < 0.01) and tet(B) and strA/B (p < 0.05) and a significant negative association between the genes tet(A) and tet(B) (p < 0.05).