Antimicrobial Resistance in Commensal Escherichia coli from Pigs during Metaphylactic Trimethoprim and Sulfamethoxazole Treatment and in the Post-Exposure Period

The prevalence of trimethoprim (TMP) and sulfamethoxazole (SMX) resistance in commensal E. coli from pigs was tested in this study. E. coli was derived from three groups of piglets in successive stages of metaphylactic therapy and from two groups of sows 10 and 18 weeks after the treatment. MIC values of TMP and SMX were determined for a total of 352 strains. The presence of resistance genes (dfrA1, dfrA5, dfrA7, dfrA12, dfrA17, sul1, sul2, sul3) and class 1 and 2 integron-associated dfrA gene cassettes was tested. Resistance to TMP was very high during the administration of the antimicrobial (from 97 to 100%) and amounted to 86% and 69% in the post-exposure period; MIC > 32 mg/L. The isolates from all groups of pigs were resistant to sulfamethoxazole, with MIC > 1028 mg/L. The dfrA1 and sul1 genes (as part of integrons) dominated in E. coli from piglets, but the dfrA12 and sul1 genes were prevalent in E. coli from sows. Coexistence of the different dfrA genes was detected in 71 isolates from all groups of swine. Transcription analysis revealed that most of these genes were not transcribed, particularly gene cassettes of class 1 integrons. The research revealed a high level of resistance associated with the metaphylactic treatment, persistence and circulation of resistance in bacterial populations. Diverse genetic background with multiple and not transcribed resistance genes was observed.


Introduction
The use of antimicrobials in the treatment of diseases of food-producing animals began in the mid-1940s. In the early 1950s, antimicrobials were introduced to commercial feed for cattle, pigs and poultry [1]. The worldwide use of antimicrobials in veterinary medicine is very high. For example, according to the recent Food and Drug Administration report, in 2012 in the United States, sales and distribution of antimicrobials for use in food-producing animals was over 14 thousand tons [2]. Although in European Union sales of veterinary antimicrobial agents is lower, it still occurs at a significant level [3]. The increase in demand for livestock products and the development of farming, nutrition and management have led to changes in the animal production systems. The intensification of food production systems rules out tolerance of disease outbreaks; thus, various antimicrobial drugs are administered for the prevention of diseases. In pig farming, treatment is often applied in young piglets. The mortality of piglets is high in the most critical periods of growth, during the first weeks of life and immediately after weaning piglets from a sow. The post-weaning stress and the related immunocompromised state promotes the development of bacterial infections such as respiratory infections, diarrhea and enteritis [4][5][6]. Antimicrobials are administered in the control treatment called metaphylaxis in order to avoid the spread of disease. Trimethoprim and sulfonamides in combination are antimicrobial agents that are commonly used in such cases. They are synthetic drugs that act as competitive inhibitors of the successive steps of the folic acid synthesis and are effective against a wide range of Gram-positive and Gram-negative microorganisms [7,8].
Due to the spatially limited environment and repetitive production scheme, large breeding farms form closed enclaves, where the pool of the resistant microorganisms and resistance genes can be accumulated and circulate between animals and their environment. It is interesting to what extent the supply of antimicrobials results in the development of resistance and whether after the cessation of antimicrobial pressure resistance decreases. This study concerns the relationship between TMP-SMX use in the metaphylactic treatment of post-weaning piglets and the resistance of commensal E. coli from pigs. The research include the determination of the phenotypic and genotypic TMP-SMX resistance prevalence, class 1 and 2 integrons and plasmids characterisation. It also concerns the study of resistance gene transcription and the correlation of the resistance genes transcription with integron and non-integron location of the genes.

Research Material
The research material was derived from a pig breeding farm in the western part of Poland, the Lubuskie Voivodeship. The farm is one of the largest in the region, with an annual production that amounts to 60,000 pigs. The farm works in a closed system. It includes the entire production cycle from sows to piglets to porkers. Five groups of pigs, which were subjected to the same metaphylaxis program, were chosen for the research. The groups consisted of individuals of different age. Fecal samples were taken from 6-week-old piglets, after the first week of treatment (Piglets 1; P1), from a 7-week-old cohort taking antimicrobials for 2 weeks (Piglets 2; P2), and an 8-week-old cohort taking antimicrobials for 3 weeks (Piglets 3; P3). The sampling also covered two herds of sows that were subjected to the same metaphylaxis program in the past. They were 10 (Sows 1; S1) and 18 weeks (Sows 2; S2) after the end of treatment and did not take any other antimicrobials before the time of sampling. Sows 1 were 4.5-months-old gilts, Sows 2 were 6-months-old first parity sows. The medical program was applied after the diagnosis of colibacillosis. The antimicrobials were administered in the form of medicated fodder which contained trimethoprim/sulfamethoxazole, 15-30 mg of active substance per 1 kg of body weight per day.

E. coli Isolation and Selection of Non-Identical Strains
Fecal samples were plated on mFC chromogenic agar (Merck, Darmstadt, Germany), and after incubation at 44 °C, 2 to 3 randomly selected blue colonies were cultured on MacConkey agar (Merck). Next, typical lactose fermenting colonies were identified by biochemical IMViC testing for identification of E. coli. Selection of non-identical E. coli isolates was performed via BOX-PCR fingerprint analysis [9][10][11] and phylogenetic grouping (identification of the main phylogenetic groups: A, B1, B2 and D) [12]. E. coli isolates were considered as non-identical (individual strains) when they demonstrated less than 80% genomic similarity in BOX-PCR fingerprint analysis and belonged to different phylogenetic groups (results not shown). A total of 352 E. coli strains (1 strain per animal) were selected for further investigations: 59 from Piglets 1, 72 from Piglets 2, 82 from Piglets 3, 77 from Sows 1 and 62 from Sows 2.

Antimicrobial Susceptibility Testing
E. coli isolates were tested for their antimicrobial susceptibility to the antimicrobials which were administered to the pigs in the metaphylactic treatment. Minimum inhibitory concentrations (MIC, mg/L) of trimethoprim and sulfamethoxazole were determined by the broth microdilution method using Sensititre plates for veterinary application (TREK Diagnostic Systems, Oakwood Village, OH, USA), in the range 0.5-32 mg/L for TMP and 8-1024 mg/L for SMX. The results were interpreted according to the epidemiological MIC cut-off values set by EUCAST (EFSA) with the breakpoint for resistance to trimethoprim set at > 2 mg/L, and sulfamethoxazole > 64 mg/L [13,14]. E. coli ATCC 25922 was used as a susceptibility control strain.

Resistance Genes Detection
Five TMP resistance genes: dfrA1, dfrA5, dfrA7, dfrA12, dfrA17, and three SMX resistance genes: sul1, sul2 and sul3, were chosen for the genotypic studies. The presence of all dfrA genes was detected by multiplex PCR designed by Grape [15]. Detection of sul genes was carried out using PCR as reported previously [16]. Bacterial thermal lysates were used as a DNA template. Positive and negative controls were included in all PCR arrays.

Detection and Characteristics of Class 1 and Class 2 Integrons
The presence of class 1 and class 2 integrons was analyzed by PCR amplification of integrase genes (int1, int2), variable regions containing gene cassettes (c1gc, c2gc) and the 3ꞌ conserved region of the class 1 integron (qacEΔ1, sul1) [17]. Randomly chosen, same-size PCR products of the variable regions of integrons were purified (Isolate II PCR and Gel Kit; Bioline, London, UK) and sequenced (Genomed, Warsaw, Poland). Characterization of the class 1 integron gene cassette promoter, including the Pc-P2 region, was performed by amplification and sequencing with the designed primers: PromI1-F 5ꞌ-AGGACAGAAATGCCTCGACT-3ꞌ and PromI1-R 5ꞌ-CCCGAGGCATAGACTGTACA-3ꞌ, located in intI1 and attI sequences, respectively.

Plasmid Characteristics
Plasmids were isolated using the Plasmid DNA purification Kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer's instructions and separated electrophoretically. Plasmid size was estimated in relation to the size of the plasmids isolated from E. coli V517 reference strain, using the BioGene V.99.03 system (Vilber Lourmat, Collegien, France). PCR-based replicon typing was performed for the recognition of the plasmid incompatibility group replicon types: F, FIA, FIB, FIC, B/O, X, Y, N, P, W, T, A/C, HI1, HI2, I1-Ic, L/M, K and FII [18].

Real-Time PCR Assays with Reverse Transcription
The analysis of dfrA gene transcription in E. coli isolates with the co-existing dfrA genes was performed by the RT-PCR method. LB broth (Merck) was inoculated by overnight bacterial culture and grown at 37 °C to the logarithmic phase (OD600 = 0.8). Each strain was cultured in the presence of TMP (at a concentration under the individual MIC value: 32 mg/L) and in the absence of antimicrobial. The RNA was extracted using the SV Total RNA Isolation System (Promega, Madison, WI, USA) and quantitated based on absorption at 260 nm, and purity was determined by the ratio of the absorption values at 260 and 280 nm. Residual DNA was digested with DNase (Roche, Basel, Switzerland). The reverse transcription and real-time analysis were conducted with the DyNAmo SYBR Green 2-Step qRT-PCR Kit (Thermo Scientific) according to the manufacturer's recommendations. Random hexamers were used to prepare cDNA from 0.5 µg of total RNA for each sample. Real-time PCR reactions were performed on a Rotor-Gene 3000 (Corbett Research, Sydney, Australia) in 20 µL volumes, containing 10 µL of 2× Master mix buffer, 0.5 µM of each primer and 1 µL of cDNA template. In each analysis, a no-template control (NTC) and a no-reverse transcriptase control (-RT) were included, and no transcription profiles were observed before the onset of 40 cycles of PCR.
Positive controls of the transcription were performed by using primers specific for the gene encoding D-glyceraldehyde-3-phosphate dehydrogenase A (gapA). The positive RT-PCR signal was defined as a threshold cycle (Ct) value <35. All transcription assays were done in duplicate, using RNA from a single isolation.

Antimicrobial Susceptibility of Isolates
The majority of the isolates were resistant to the tested antimicrobials. The TMP resistance was 97.3% and 100% in the strains from Piglets 1, 2 and 3 during the metaphylaxis The prevalence of the resistant strains from pigs in the post-exposure period was lower: 86% and 69% for Sows 1 and Sows 2, respectively ( Table 1). All the strains from piglets during the medical program were resistant to SMX. Also all the tested E. coli from Sows 1 and 92% of strains from Sows 2 were resistant to SMX ( Table 1).
The MIC values were > 32 mg/L for TMP and > 1024 mg/L for SMX for all of the tested E. coli from pigs during the metaphylaxis and for most of resistant E. coli from pigs in the post-exposure period. Table 1. Resistance to trimethoprim and sulfamethoxazole of E. coli isolated from groups of pigs during metaphylaxis (P1-P3: antimicrobial pressure present) and in post-exposure period (S1, S2: antimicrobial pressure absent).

Prevalence of Resistance Genes
All 352 strains were tested for the presence of TMP and SMX resistance determinants. The dfrA1 gene was the most commonly detected trimethoprim resistance gene in E. coli from all groups of piglets. The dfrA1 and dfrA12 genes occurred with similar frequencies in strains from two groups of sows. The dfrA7 and dfrA5 genes were detected with a lower incidence in all groups ( Table 2).
The dfrA17 gene was not identified in any of the tested isolates. The sulfamethoxazole resistance gene sul1 dominated in E. coli from piglets and the sul3 gene dominated in E. coli from sows. The sul2 gene occurred with lower frequencies in E. coli from all groups ( Table 2). The coexistence of multiple resistance genes in single strains was also found (Table 3).

Resistance Genes Characteristics
Considering the prevalence of the dfrA1 gene both in integrons and not associated with integrons, the integron location of the dfrA1 gene predominated in E. coli from piglets during the antimicrobial treatment. The dfrA1 integron gene cassette was detected in 42/76%, 47/66% and 60/73% of strains from Piglets 1, Piglets 2 and Piglets 3, respectively. In isolates from Sows 1 the dfrA1 integron and non-integron location was similar (15/19% vs. 14/18%). In isolates from Sows 2 the dfrA1 gene not associated with the integrons was more frequently found (15/24% integron and 25/40% non-integron location). The dfrA7 and dfrA12 genes occurred as an integron gene cassette only in seven strains from piglets and one from sows.
The coexistence of 2 to 4 TMP resistance genes (associated or not associated with integrons) in various combinations was found in 73 (21%) of all tested isolates. The most frequently appearing genes coexistence concerned the most commonly identified genes: dfrA1 and dfrA12 ( Table 3). The dfrA1 gene coexisted also with dfrA5 or dfrA7 and dfrA12 genes. In Piglets 3, combinations of all identified dfrA genes were observed (Table 3). Strains carrying more than one SMX resistance gene were detected exclusively in E. coli from piglets during medical treatment. Three strains from Piglets 1 and eleven strains from Piglets 3 carried both sul1 and sul2 genes. Six strains from Piglets 3 had sul1 and sul3 genes.

Plasmid Characteristics
The plasmid characterization was performed for the strains that included multiple TMP resistance genes (n = 71). The plasmid profile for a single strain contained from 1 to 5 plasmids. The plasmid profiles with 2 to 4 plasmids dominated in E. coli from animals during the antimicrobial treatment (61%). Overall, plasmids with sizes ranging from 1.5 kb to 100 kb were identified. Large plasmids, of approx. 100, 60 and 12 kb, occurred only in E. coli from animals during treatment. The most frequently identified plasmid had size approx. 56 kb, followed by 7 kb and 1.5 kb, and they were isolated from both groups. Plasmids of small size, approx. 8 kb, 6 kb, 4.5 kb and 3.8 kb, were detected only in isolates from pigs after treatment. The replicon typing revealed the presence of eight different plasmid incompatibility groups: FIB, FIIA, K and I1, P, HI1, B/O, T in the set of 71 tested isolates. Replicons FIB, FIIA, K and I1 were detected both in isolates from pigs during the metaphylaxis and in the post-exposure period. Replicons P, HI1, B/O and T were detected only in isolates from pigs during the treatment.

Transcription Analysis of dfrA Genes
In order to determine if all coexisting in a single strains dfrA genes contribute to the phenotypic resistance, the transcription analysis of dfrA genes was performed. Generally transcription of at least one of the dfrA genes was revealed in 31 from 71 tested isolates (Table 4). The presence of the transcript of the dfrA12 gene was found in 25 out of the 46 strains carrying the gene. In four out of the 22 strains, transcription of the dfrA7 gene was detected. Also transcription of the dfrA5 gene was detected rarely, and was revealed in three out of the 18 strains carrying this gene. Transcription of the dfrA1 gene was not detected in any of the strains carrying this gene although in 28 strains dfrA1 constituted class 1 integron gene cassette. Only in two cases were the transcripts of two trimethoprim resistance genes in a single strain detected: dfrA5, dfrA7 and dfrA7, dfrA12. The dfrA gene transcription was detected in strains derived both from animals during the antimicrobial therapy and in the post-exposure period (in 16 and 15 strains, respectively).
In order to verify whether the transcription of tested genes can be induced by the direct presence of trimethoprim, Real-time PCR analysis was conducted afresh but with the presence of the antimicrobial in the culture prior to the isolation of RNA. Regardless of the presence of TMP pressure during bacterial growth, the transcription score for each gene in the strain was the same.

Characteristics of Class 1 Integron Promoters
Gene dfrA1, for which no transcript was detected, constituted a part of class 1 integron gene cassettes. Therefore, for all 28 strains carrying class 1 integron with gene cassettes dfrA1, the strength revealed in 20 of the tested integrons (compared to gb:KF921585.1). The hybrid weak PcH1 promoter (-35 TGGACA/-10 TAAACT) was detected in eight tested integrons (compared to gb: KF921569.1) detected in E. coli from piglets during antimicrobial administration. The simultaneous presence of the second P2 promoter (-35 TTGTTA/-10 TACAGT) was detected in all tested sequences, but without the GGG base sequence in the spacer region [19].

Discussion
The presented research constitutes an assessment of the impact of trimethoprim (TMP) and sulfamethoxazole (SMX) administration in the metaphylactic program on the prevalence and diversity of TMP-SMX resistance in commensal E. coli from pigs during the production cycle. Antimicrobial usage in food animals contributes to the development and the spread of resistant microorganisms in the environment. Although antimicrobial growth promoters have been forbidden in the EU since 2006 [20], antimicrobials can be used, apart from the direct treatment of diseases, in control treatment called metaphylaxis. After the diagnosis of infection in a part of a group of animals, antimicrobials are administered to the whole group of animals or to a herd, in order to prevent the spread of disease to animals in close contact and at substantial risk of infection [21,22].
In our study in piglets during the first week of antimicrobial administration (P1) the high prevalence of E. coli TMP and SMX resistance was detected. Also all of the tested strains from piglets (P2) taking antimicrobials from 2 and 3 weeks (P3) were resistant. The level of resistance, expressed as MIC values, for all resistant isolates was very high. The high resistance in piglets in the first week of treatment may be due to the rapid development of resistance but also due to the selection of resistant clones. Strains present both in the environment and from the sow can colonize the piglets' intestine [23,24]. Our study did not cover pigs in the pre-exposure period, but it is highly probable that since the commensal flora of sows is already resistant to antimicrobials, piglets after the birth were colonized by the resistant clones. Horizontal transfer of resistance genes through numerous plasmids detected in the tested strains, may also be responsible for the development of the resistance.
Lower prevalence of resistant E. coli was observed in sows that were 10 and 18 weeks after treatment, but the percentage of resistance was not less than 69%. The persistence of resistance in spite of absence of the antimicrobial pressure in the population contributes to the high prevalence of resistance. Different mechanisms can be responsible for the stability of antimicrobial resistance in bacterial populations [25]. In the tested set of isolates from pigs in the post-exposure period, all TMP and SMX resistance genes and integrons that were detected in isolates from pigs during antimicrobial treatment were also observed. It may be the result of adaptation, circulation and maintenance of some set of resistant strains regardless of antimicrobial pressure, in the environment of the studied farm. The differences between the two groups (A+, A−) concerned mainly the number of the detected resistance genes, and more frequent detection of the estX-sat2-aadA1 gene cassette array in class 2 integrons in E. coli from pigs in the post-exposure period. The plasmids of strains from pigs during and after the treatment differed in their sizes and classification to incompatibility groups. Greater diversity was observed in E. coli from animals at the time of the metaphylaxis. This indicates the high dynamics of the resistance transfer in the environment during antimicrobial pressure.
In the studies of the genetic background of resistance, the presence of four different TMP resistance genes and three SMX resistance genes was revealed. The dfrA1, dfrA12, dfrA7 and dfrA5 TMP resistance genes were detected. The general prevalence of the different dfrA genes in E. coli was in agreement with the earlier studies from different clinical and commensal E. coli [26,27]. An exception was the absence of the dfrA17 gene, because it has been observed frequently by others in E. coli isolates from animal origin [26][27][28]. Higher prevalence of the dfrA1 gene was related to the significant incidence of class 1 and class 2 integrons carrying the dfrA1 gene cassette detected in this group. The sul1 and sul3 genes were the most frequently detected SMX resistance genes among E. coli from all groups of pigs, but differences in the proportion of these genes were observed depending on the source of the strains. The sul1 gene was the most common in E. coli from piglets during the antimicrobial therapy, and the sul3 gene among isolates from animals after the antimicrobial treatment. More frequent occurrence of the sul1 gene than sul3 in the E. coli from piglets during metaphylaxis was connected with the prevalence of the sul1 gene in the 3' conservative segment of class 1 integrons. Class 1 integrons dominated among E. coli from this group of pigs. The sul2 gene was detected with lower incidence in all but the Piglets 3 group. According to many reports, sul1 gene (as part of an integron) and sul2 dominate among sulfonamide resistant E. coli in humans and animals [17,27,29,30] but also some data show the prevalence of the similar gene sul3 [16].
The coexistence of multiple TMP and SMX resistance genes in single strains is described in the literature usually with very low frequency [26,28]. In our studies, the coexistence of dfrA genes occurred frequently (71/20%) both in E. coli from pigs during the antimicrobial treatment and in the post-exposure period. Multiple SMX resistance genes were detected with lower frequency, and concerned 20/9% E. coli, exclusively from piglets during treatment. Transcription analysis of the dfrA genes revealed that in most cases, only one of the multiple genes underwent transcription. Only in two isolates was the simultaneous transcription of two genes found. The dfrA12 gene was the most frequently expressed, whereas the dfrA5 and dfrA7 genes were expressed in few of the tested isolates. Transcription of these genes was detected in the isolates both from piglets during the treatment and from sows in the post-exposure period. Unexpectedly, the most commonly prevalent dfrA1 gene was not transcribed in any of the tested isolates. Some of the genes that were not transcribed were associated with integrons. The strength of the promoter is responsible for the level of integron gene cassette transcription [31]; therefore promoter analysis of class 1 integrons was performed. Two different Pc variants were found in combination with the P2 promoter: PcW-P2 and PcH1-P2. Both PcW and PcH1 promoters are classified as weak promoters. It has been demonstrated that when weak Pc and the second P2 promoters are present, the P2 promoter is responsible for 90% of the total promoter activity [32][33][34]. However, the P2 promoter is active when carrying a fragment of 3 bp insertion GGG in its 14-bp spacer. The lack of this GGG base sequence of the P2 promoter in all integrons analyzed in our study was detected. This may be the reason for the absence of the dfrA gene cassette transcript in the set of the tested E. coli. The presence of the weak promoter PcW, but without a second promoter P2, was previously observed by other researchers of the dfrA1-aadA1 gene cassette array of class 1 integrons [35]. The reason for the lack of transcription of TMP resistance genes not associated with integrons is not known and requires further detailed research.
It is widely accepted that integrons are a very efficient mechanism for the dissemination of antimicrobial resistance [36][37][38][39][40]. Also the dfrA1 gene has commonly been detected by many researchers in animal and human E. coli isolates [26,30,39], often as an integron gene cassette. However, the reports of unexpressed bacterial resistance genes are rare. Enne and co-workers [41] showed that the expression of plasmid-borne resistance genes could be silenced in the pig gut, while the intact resistance genes and their promoters are retained.

Conclusions
Great diversity of the genetic background of resistance to TMP and SMX was observed in E. coli from animals at the time of metaphylaxis. High resistance levels in the first week of antimicrobials administration confirmed the effect of antimicrobial pressure on the rapid development of resistance and selection of resistant isolates. Also, the persistence of resistance in the absence of antimicrobial pressure in sows, 10 and 18 weeks after treatment, confirmed the statement that once acquired antimicrobial resistance is difficult to lose. In our study, multiple combinations of dfrA genes occurred in E. coli from all the tested groups. However, a significant portion of these genes was not transcribed, including the most frequently identified dfrA1 gene in class 1 and 2 integrons. These findings raised the question whether commonly detected genes are responsible for phenotypic resistance.