This study was conducted to investigate the effects of in-feed encapsulated cinnamaldehyde (CIN) and citral (CIT) alone or in combination on antimicrobial resistance (AMR) phenotypes and genotypes of Escherichia coli isolates recovered from feces of 6-, 16-, 23-, and 27-day-old broiler chickens. The five dietary treatments including the basal diet (negative control [NC]) and the basal diet supplemented with 55 ppm of bacitracin (BAC), 100 ppm of encapsulated CIN, 100 ppm of encapsulated CIT, or 100 ppm each of encapsulated CIN and encapsulated CIT (CIN+CIT). Antimicrobial susceptibility testing of 240 E. coli isolates revealed that the most common resistance was to β-lactams, aminoglycosides, sulfonamides, and tetracycline; however, the prevalence of AMR decreased (P < 0.05) as birds aged. The prevalence of resistance to amoxicillin–clavulanic acid, ceftiofur, ceftriaxone, cefoxitin, gentamicin, and sulfonamide was lower (P < 0.05) in isolates from the CIN or CIN+CIT groups than in isolates from the NC or BAC groups. Whole genome sequencing of 227 of the 240 isolates revealed 26 AMR genes and 19 plasmids, but the prevalence of some AMR genes and the number of plasmids were lower (P < 0.05) in E. coli isolated from CIN or CIN+CIT birds than in isolates from NC or BAC birds. The most prevalent resistance genes were tet(A) (108 isolates), aac(3)-VIa (91 isolates), aadA1 (86 isolates), blaCMY-2 (78 isolates), sul1 (77 isolates), aph(3)-Ib (58 isolates), aph(6)-Id (58 isolates), and sul2 (24 isolates). The numbers of most virulence genes carried by isolates increased (P < 0.05) in chickens from 6 to 27 days of age. The prevalence of E. coli O21:H16 isolates was lower (P < 0.05) in CIN and CIN+CIT, and the colibacillosis-associated multilocus sequence type (ST117) was most prevalent in isolates from 23-day-old chickens. A phylogenetic tree of whole genome sequences revealed a close relationship between 25 of the 227 isolates and human or broiler extraintestinal pathogenic E. coli strains. These findings indicate that AMR and virulence genotypes of E. coli could be modulated by providing encapsulated CIN or CIN+CIT feed supplements, but further investigation is needed to determine the mechanisms of the effects of these supplements.
AMR genes were less prevalent in E. coli isolates from birds fed CIN or CIN+CIT.
AMR gene prevalence and plasmids numbers were low in E. coli from CIN and CIN+CIT birds.
AMR of E. coli isolates decreased as birds aged.
The distribution of E. coli virulence genes was affected by CIN, CIN+CIT, and bird age.
Some E. coli isolates had high potential for virulence in humans.
Global poultry production has increased from 250,000 to 1,250,000 tons annually in the last 50 years due to increased meat demands for human consumption (28). To promote growth and prevent or treat infectious diseases, antimicrobials have been widely used in poultry farms (11, 42). This practice decreases incidences of bacterial infections, promotes growth performance, and reduces the mortality of poultry (63). Bacitracin (BAC) at a subinhibitory concentration has been commonly added to feeds as a growth promoter to increase body weight and feed efficiency and to alter the intestinal microbiota of chickens (1, 38, 62). However, overuse or misuse of antimicrobials increases selective pressure on bacteria, increasing the prevalence of antimicrobial resistance genes (ARGs) and antimicrobial resistance (AMR) in general (45). ARGs can be transmitted horizontally from one bacterial species to another through mobile genetic elements (MGEs) such as plasmids (44). Because of the growing prevalence of antimicrobial-resistant bacteria, the group Chicken Farmers of Canada (18) decided to stop the preventive uses of category I to III antibiotics from 2014 to the end of 2020. Because the elimination of antimicrobials as feed additives may compromise the growth performance and increase the incidences of bacterial infections in poultry production, more work is needed to develop antimicrobial alternatives and investigate their effects on AMR prevalence and spread (70).
Essential oils (EOs) are aromatic and volatile liquids, and their potential uses as antimicrobial alternatives have been investigated (33, 49). Cinnamaldehyde (CIN) and citral (CIT) are two EOs that are extracted mainly from trees and that possess antimicrobial, antioxidant, and/or anti-inflammation properties (59, 73). Because of the instability of EOs during prolonged storage, feed processing, and gut transition, their use as feed supplements in poultry production has been limited (17, 69, 78). In our previous study (74), CIN and CIT were encapsulated with a soy protein–soy polysaccharide Maillard reaction product and fed alone or in combination (CIN+CIT) to broiler chickens. All three treatments had effects comparable to those of BAC for improving growth performance, reducing gut lesions, and modulating cecal microbiota. However, we did not determine whether in-feed CIN, CIT, or CIN+CIT could affect AMR genotypes and phenotypes of fecal bacteria in broiler chickens. The synergistic effects of CIN (an aldehyde) and carvacrol (a terpene) on inhibiting the growth of foodborne bacteria, including Escherichia coli and Staphylococcus aureus, have been reported (76). Although no significant synergistic effects of CIN and CIT on broiler chicken growth performance (74) have been found, no studies have been conducted to evaluate the synergistic effects of CIN and CIT (a terpene) on AMR profiles of bacteria from chicken feeding trials.
E. coli is a gram-negative bacterium that is ubiquitous in the chicken intestinal tract. Although most commensal E. coli strains are harmless to the host, extraintestinal pathogenic E. coli (ExPEC) strains can compromise chicken performance and increase mortality rate, with resulting economic losses (19). The pathogenicity of such E. coli strains is mediated by several virulence genes (VGs) (57). ARGs such as those conferring resistance to tetracycline (tet(A)), ampicillin (blaTEM), and sulfonamide (sul1), have been found in E. coli isolates from chicken ceca, cloacae, and feces (9, 72). MGEs such as plasmids carrying ARGs have been detected in chicken E. coli isolates, highlighting the transmissibility of these genes (64). Prolonged usage of antimicrobials as feed additives can promote the prevalence of ARGs and VGs in broiler chickens (53) and increase transmission of ARGs via plasmids (8). However, relatively limited studies have been conducted on the AMR profile of E. coli from broilers fed EOs. The objective of this study was to examine the longitudinal effects of feed supplemented with encapsulated CIN, CIT, and CIN+CIT compared with the effects of BAC on the AMR genotypes and phenotypes of fecal E. coli isolates from broiler chickens in a controlled trial.
MATERIALS AND METHODS
The CIN (≥95% purity; W228613, Sigma Aldrich, St. Louis, MO) and CIT (a mixture of cis and trans isomers, 95% purity; C83007, Sigma Aldrich) were encapsulated separately in a soy protein–soy polysaccharide Maillard reaction product by emulsification and spray drying technology as previously described (75).
Details of animal trial design and management were previously described (74). A total of 1,600 1-day-old male Cobb 500 broiler chickens were allocated into 20 floor pens (80 birds per pen) and vaccinated against coccidiosis with a commercial vaccine. Birds in each group were distributed into five dietary treatments (four pens per treatment): (i) basal diet as a negative control (NC); (ii) basal diet with 55 ppm of BAC methylene disalicylate; (iii) basal diet with 100 ppm of encapsulated CIN; (iv) basal diet with100 ppm of encapsulated CIT; and (v) basal diet with a combination of 100 ppm each of encapsulated CIN and CIT (CIN+CIT). Birds were fed the five diets during their three growth phases: day 0 to day 9 (starter phase), day 10 to day 19 (grower phase), and day 20 to day 28 (finisher phase). The basal diet contained primarily cereals and soybean meals and was formulated and pelleted according to Cobb 500 guidelines as previously described (74). All experimental procedures in this study were approved by the Animal Care Committee of the Centre de recherche en sciences animales de Deschambault (Deschambault, Quebec, Canada) based on the Canadian Council on Animal Care (Ottawa, Ontario, Canada) guidelines.
Sample collection and E. coli isolation and identification
Fresh fecal samples were collected on days 6, 16, 23, and 27 from all 20 pens (three birds per pen = 60 samples) in sterilized Whirl-Pak plastic bags (Nasco, Ft. Atkinson, WI). All fresh feces were immediately placed in an insulated foam container with dry ice, transported to the laboratory, and stored at −20°C until further analysis. Fecal E. coli were isolates were recovered on Chromocult coliform agar (CCA). A 1-g portion of fecal sample was weighted into 15-mL sterilized plastic centrifuge tubes, 9 mL of 0.85% saline was added, and the tube was vortexed for 5 min to obtain a homogenous solution. Approximately 100 μL of 10-fold serial dilutions of the homogenate were spread on CCA and incubated aerobically at 37°C for 16 to 24 h. Three presumptive colonies were purified from each analyzed sample and stored at −80°C in 25% glycerol in tryptic soy broth. API 20E strips (bioMérieux, St. Laurent, Quebec, Canada) were used to confirm E. coli identity according to the manufacturer's specifications.
Antimicrobial susceptibility test
MICs of the antimicrobials were determined by the broth-dilution method with an automated system (Sensititre, Trek Diagnostic Systems, Cleveland, OH) as previously described (65). The following 14 antimicrobial agents were included in the test panel: amoxicillin–clavulanic acid, ampicillin, ceftiofur, ceftriaxone, cefoxitin, chloramphenicol, ciprofloxacin, gentamicin, azithromycin, nalidixic acid, streptomycin, sulfisoxazole, tetracycline, and trimethoprim-sulfamethoxazole. The MICs were interpreted according to the breakpoints of the Clinical Laboratory Standards Institute (20) and the Canadian Integrated Program for Antimicrobial Resistance Surveillance (50).
Whole genome sequencing and phylogenetic analysis
To get a better insight into the AMR genotype, all E. coli isolates were submitted for whole genome sequencing as described previously (52, 65). DNA was extracted from each E. coli isolate with the DNeasy Blood & Tissue Kit (QIAGEN, Toronto, Ontario, Canada) and quantified with a Qubit fluorometer (Invitrogen, Thermo Fisher, Waltham, MA). The sequencing libraries were established with a DNA sample preparation kit (Nextetra XT, Illumina, San Diego, CA), and a 600-cycle MiSeq reagent kit (v3) was used to perform paired-end sequencing on the MiSeq platform (Illumina). For genome sequence analysis, the reads were checked with FastQC and combined with FLASH, v. 1.2.9 (40). The high-quality reads were assembled with the SPAdes genome assembler, v. 3.9.0 software (6). The assembled genomes were annotated with Prokka, v. 1.11 (61). The genome sequences were assembled in SPAdes v. 3.0 with the Integrated Rapid Infectious Disease Analysis platform (https://www.irida.ca). The Resistance Gene Identifier, v. 4.0.2 (21) platform and the PlasmidFinder v. 1.3 (14) and VirulenceFinder (15) databases were used to identify ARGs, plasmid replicons, and VGs in sequenced genomes.
The phylogenetic analysis was conducted with core genome single nucleotide variant phylogenomics pipeline, with single nucleotide variants calling at a minimum mapping quality of 30, minimum base quality of 30, minimum alternate fraction of variant bases in agreement 80%, and minimum depth of coverage of 15× (48). The reference genome used in this analysis was obtained from GenBank (accession NZ_CP008957.1). A maximum likelihood phylogeny with bootstrap supports was built in FigTree v. 1.4.4 (https://github.com/rambaut/figtree/releases).
The number of antibiotics to which isolates were resistant (resistance spectrum), MGEs found in each isolate, multilocus sequence types (MLSTs), and VG categories were analyzed with a randomized complete block design using the general linear mixed model procedure of the Statistical Analysis System v. 9.4 (2016, SAS Institute, Cary, NC). Treatments and day (bird's age) were used as sources of variation, and the individual pens were the experimental units. Least significance differences were used to separate treatment means whenever the F value was significant. The association Cochran-Mantel-Haenszel test was used to determine the relationship between treatment or day and the prevalence of AMR phenotype and genotype and specific MGEs (plasmids) using the FREQ procedure. The difference between treatments was considered significant at P < 0.05.
No significant differences in E. coli concentration (CFU per gram of sample) (P > 0.05) were noted between dietary treatments. From chickens 6 to 27 days of age, 240 E. coli isolates were obtained (48 isolates per treatment, 12 isolates per sampling day) and characterized.
All 240 E. coli isolates were susceptible to ciprofloxacin, azithromycin, and nalidixic acid. Isolates from CIN birds had a lower prevalence of AMR (P < 0.05) to 7 of the 14 antimicrobials (amoxicillin, ampicillin, ceftiofur, ceftriaxone, streptomycin, sulfonamide, and tetracycline) than did isolates from NC and BAC birds (Fig. 1A). Isolates from CIN+CIT birds had a lower prevalence of AMR (P < 0.05) to 5 of the 14 antimicrobials (amoxicillin, ceftiofur, ceftriaxone, cefoxitin, and sulfonamide) compared with isolates from BAC birds, but a higher prevalence of resistance to streptomycin and trimethoprim-sulfamethoxazole was noted in isolates from CIT birds (P < 0.05) than in isolates from NC and BAC birds. Of the 240 isolates, 125 (52.08%) were resistant to gentamicin regardless of the treatment group. An effect of bird age on the prevalence of resistance to this antibiotic (P < 0.05) was also noted. Prevalence of resistance to 6 of 14 antimicrobials (amoxicillin, ceftiofur, ceftriaxone, cefoxitin, gentamicin, and sulfonamide) decreased (P < 0.05) with bird age (Fig. 1B).
Multidrug resistance was observed among all 240 E. coli isolates; >70% were resistant to three or more antibiotics, with a mean of three to five. Isolates from CIN birds had the lowest resistance (P = 0.06); >41% of these isolates were pansusceptible or resistant to only one antibiotic compared with 27% of isolates from CIN+CIT birds, 21% from CIT birds, 21% from NC birds, and 19% from BAC birds (Fig. 2A). The highest number of multidrug-resistant isolates was found in 6-day-old birds, with the BAC and CIN treatment groups having the highest and lowest numbers (P < 0.05), respectively. Resistance to nine antimicrobials was found in 40% of E. coli isolates recovered on day 6, 20% on day 16, 7% on day 23, and 7% on day 27 (Fig. 2B). Except in BAC birds, the lowest numbers of multidrug-resistant isolates were found in 27-day-old birds (P < 0.05).
Feed with CIN or CIN+CIT and increasing bird age were two factors that reduced the prevalence of resistance to antimicrobials. Of the 240 E. coli isolates submitted for sequencing, 13 isolates (4 from NC birds, 1 from BAC, 3 from CIN, 2 from CIT, and 3 from CIN+CIT) had sequences of poor quality, so only 227 isolates were included in the final analyses (MLSTs, ARGs, VGs, and plasmids).
Serotypes and MLSTs
The 227 E. coli isolates belonged to 66 serotypes, and the most common were O21:H16 (22 isolates), O99:H15 (16 isolates), O8:H19 (14 isolates), O77:H18 (10 isolates), O91:H7 (9 isolates), O?:H21 (8 isolates), O39:H7 (7 isolates), and O15:H6 (7 isolates) (Table 1). The distribution of 28 serotypes was significantly affected (P < 0.05) by dietary treatments and/or days (Table 1). Serotype O21:H16 was less prevalent in isolates from CIN or CIN+CIT birds compared with those from NC, BAC, or CIT birds but more prevalent in younger birds (6 and 16 days of age) than in older birds (23 and 27 days). Two isolates of serotype O78 were recovered from BAC birds on day 23. A total of 40 MLSTs were identified, and the distribution of 20 MLSTs was affected (P < 0.05) by dietary treatment, day, and/or their interactions (Table 2). Each MLST represented one or more serotypes, and some colibacillosis-associated MLSTs such as ST38 and ST117 had higher numbers of isolates on day 23 (P < 0.05) than on days 6, 16, and 27. A significantly lower number of ST38 isolates was observed in CIN birds (P < 0.05) than in control birds, CIT, and CIN+CIT birds.
These data indicate that CIN or CIN+CIT in feed could decrease the prevalence of avian colibacillosis-associated MLSTs and serotypes compared with birds fed the NC or BAC diets. Bird age also could be a factor influencing the prevalence of MLSTs and serotypes.
ARGs and plasmids
Among the 227 sequenced genomes, 26 ARGs and 19 plasmids were detected. In agreement with the AMR phenotypes, E. coli isolates from CIN and CIN+CIT birds had a low prevalence of ARGs, including those conferring resistance to aminoglycosides (aac(3)-VIa, aadA1, and aadA2b), β-lactams (blaCARB, blaCMY-2, and blaCTX-M-1), sulfonamides (sul2 and sul3), and tetracycline (tet(A)). The prevalence of 18 (69.2%) of the 26 ARGs was significantly affected (P < 0.05) by dietary treatment, sampling day, or their interaction (Table 3). These 18 genes included 7 for aminoglycosides (aac(3)-VIa, aadA1, aadA2b, aadA23, aph(3)-Ia, aph(3)-Ib, and aph(6)-Id), 3 for β-lactams (blaCARB, blaCMY-2, and blaCTX-M-1), 3 for sulfonamides (sul1, sul2, and sul3), 2 for trimethoprim (dfrA15 and dfrA16), and 1 each for erythromycin (ereA), fosfomycin (fosA7), and tetracycline (tet(A)). The lowest prevalence (P < 0.05) of aac(3)-VIa, aadA1, aadA2b, blaCARB, dfrA15, dfrA16, ereA, tet(A), and sul3 was found in CIN and/or CIN+CIT birds. The prevalence of some ARGs, including aac(3)-VIa, aadA1, blaCMY-2, sul1, and tet(A), decreased (P < 0.05) from day 6 to day 27, whereas the prevalence of aadA2b, aph(3)-Ib, aph(6)-Id, blaCARB, dfrA16, and ereA increased (P < 0.05) from day 6 to day 27. Significant interactions between dietary treatment and day (P < 0.05) were found for aadA1, aadA2b, aadA23, aph(3)-Ia, dfrA16, ereA, and tet(A).
The distribution of ARGs was influenced by serotype. The aac(3)-VIa gene was not detected in the 16 and 14 isolates of serotypes O99:H15 and O8:H19, respectively, whereas all 10 O77:H18 isolates and 21 (95.45%) of the 22 O21:H16 isolates carried this gene. The 86 aadA1-positive isolates included 20 (90.91%) of serotype O21:H16, none of O77:H18, and none of O99:H15, whereas the 17 aadA2-positive isolates were the 16 of serotype O99:H15 and the single O99:H38. All O99:H15, all O77:H18, 21 O21:H:16, and none of the O8:H19 isolates carried the blaCMY-2 gene.
The distribution of 15 plasmids was affected by dietary treatment (Fig. 3A) and/or sampling day (Fig. 3B) (P < 0.05). Compared with E. coli isolates from NC or BAC birds, a lower prevalence of plasmids IncFIB, IncI1, IncY, and IncA/C2 (P < 0.05) was observed in isolates from CIN and CIN+CIT birds. The prevalence of eight plasmids (IncFII, IncFIIA, IncFIB, IncH, IncI1, IncX1, ColRNA1_1778, and ColRNAl_1857) decreased (P < 0.01) from day 6 to day 27, whereas the prevalence of ColRNAl_1993, ColRNAl_1885, and ColRNAl_1291 increased (P < 0.05) from day 6 to day 23 but decreased (P < 0.01) thereafter from day 23 to day 27. Significant effects of the interaction between dietary treatment and day (P < 0.05) were found on the prevalence of nine plasmids (IncFII, IncFIIA, IncFIB, IncH, IncX1, IncY, IncA/C2, ColRNAl_1857, and ColRNAl_1885). For AGRs, the prevalence of plasmids was also serotype dependent. A clear relationship between some plasmids and some ARGs was found. For example, 74 (81.31%) of the 91 isolates harboring aac(3)-VIa, 72 (83.72%) of the 86 isolates harboring aadA1, 63 (80.77%) of the 78 isolates harboring blaCMY-2, all 5 of the isolates harboring blaCTX-M-1, and 34 (85.0%) of the 40 isolates harboring blaTEM also carried the plasmid IncFII.
In general, no synergistic effects of CINCIT on the total number of ARGs, MGEs, and MLSTs were observed (data not shown). However, the interaction of CIN and CIT affected AMR (P < 0.05) and ARGs (P < 0.01) on day 6. A significant interaction effect between CIN and CIT was observed on the number of ST69 isolates (P < 0.05) on day 16 and the number of ST38 isolates (P < 0.01) on days 16, 23, and 27.
The 430 VGs belonged to 11 virulence factor categories: flagella, chemotaxis, fimbriae, pili, curli, iron intake proteins, toxins, secretion systems, enzymes, regulators, and miscellaneous (Table 4). No significant differences (P > 0.05) in the mean number of virulence factor categories were observed between dietary treatments. However, the mean number of VGs in each virulence category was lower (P < 0.01) in isolates collected on day 6 than in those collected on days 16, 23, and 27. The effect of the interaction between dietary treatment and day on five virulence categories (pili, toxins, secretion systems, enzymes, and regulators) was significant (P < 0.05).
A phylogenetic tree for all 227 sequenced E. coli genomes was based on single nucleotide polymorphisms (Fig. 4A). Three avian pathogenic E. coli (APEC) strains and 11 human ExPEC strains were used for comparison with the reference gene obtained from GenBank (accession NZ_CP008957.1). Twenty E. coli isolates in five dietary treatments (1 NC isolate, 5 BAC isolates, 7 CIN isolates, 5 CIT isolates, and 2 CIN+CIT isolates) on three sampling days (7 isolates on day 16, 5 isolates on day 23, and 8 isolates on day 27) had a high level of genetic relatedness (single nucleotide variants < 30) with human ExPEC. Five E. coli isolates in two dietary treatments (three BAC isolates and two CIN+CIT isolates) on two sampling days (three isolates on day 23 and two isolates on day 27) had a high level of genetic relatedness (single nucleotide polymorphism distance < 30) with both human and broiler chicken ExPEC. To clearly show the AMR and virulence genotypes of the 25 isolates with high relatedness to human ExPEC and broiler APEC, a reduced phylogenetic tree was constructed (Fig. 4B), and detailed information of their AMR phenotypes, genotypes, and virulence are shown in Table 5. Among 24 E. coli isolates, eight serotypes (O15:H6, 6 isolates; O16:H48, 3 isolates; O21:H21, 2 isolates; O25:H18, 4 isolates; O78:H4, 2 isolates; O81:H39, 2 isolates; O103:H21, 3 isolates; and O184:H4, 2 isolates) had some level of relationship with human or chicken ExPEC. Among 20 E. coli isolates, four MLSTs (ST10, 5 isolates; ST69, 6 isolates; ST101, 5 isolates; and ST117, 4 isolates) were related to human or broiler chicken ExPEC. The 25 E. coli isolates that were phylogenetically related to human or chicken ExPEC were found in all five dietary treatments (NC, 1 isolate; BAC, 9 isolates; CIN, 6 isolates; CIT, 5 isolates; and CIN+CIT, 4 isolates) and on three sampling days (day 16, 7 isolates; day 23, 8 isolates; and day 27, 10 isolates).
APEC is one of the most significant pathogenic bacteria in broiler chickens, causing colibacillosis characterized by acute septicemia or subacute airsacculitis, high mortality, compromised bird performance, and economic loss (34). Antimicrobial-resistant E. coli isolates from chicken gastrointestinal tracts and feces have the potential to cause human infections through contaminated chicken meats and carcasses, which represent a food safety risk (51). The present study was conducted to evaluate the longitudinal (6, 16, 23, and 27 days of age) effects of in-feed encapsulated EOs (CIN, CIT, and CIN+CIT) compared with the effects of BAC on AMR phenotypes and genotypes of fecal E. coli isolates recovered from broiler chickens. Because various factors, including age and diet, can affect gut microbiota and AMR (4, 24), isolates recovered from isolates collected on these four days were investigated based on the nutritional phase (starter, days 0 to 10; grower, days 11 to 20; and finisher, days 21 to 28). Days 6 and 16 were in the middle of the starter and grower phases, and days 23 and 27 represented the beginning and end of the finisher phase close to harvest.
More than 85% of the E. coli isolates in this study were resistant to at least one of the tested antimicrobials. A higher prevalence of resistance to ampicillin, gentamicin, and sulfonamide was observed in isolates from birds fed BAC than from birds in the other dietary treatments. BAC has been commonly used in poultry farms to prevent or treat necrotic enteritis caused by Clostridium perfringens (23), but increases in AMR among E. coli (a gram-negative bacterium) in birds fed BAC need further investigation because the targets of this antibiotic are generally gram-positive bacteria. However, the effects of BAC supplementation on antimicrobial resistance in E. coli via possible plasmid transfer or altered AMG expression have been reported (41, 67). These findings highlight the urgency of reducing the common use of antimicrobials in poultry production and exploring alternatives. In the present study, E. coli isolates from birds fed CIN or CIN+CIT as alternatives to BAC had a lower prevalence of resistance to most of the tested antimicrobials, such as ceftriaxone, cefoxitin, and sulfonamide, indicating the potential of CIN to reduce the spread of AMR. Successive in vitro exposures of gram-negative bacteria, such as Serratia marcescens and Proteus mirabilis, isolated from clinical human samples to CIN and oregano oil resulted in increased MICs of antimicrobials such as ampicillin, nalidixic acid, and tetracycline (7). The present findings are the first reported for the effects of encapsulated CIN, CIT, and CIN+CIT in chicken feed on antibiotic susceptibility of E. coli from chicken feces. A higher prevalence of resistance to streptomycin was detected in E. coli isolated from CIT birds. This phenomenon may be related to terpenes as the active compounds in CIT because increased MICs of streptomycin have been found against E. coli isolates from fecal samples of broiler chickens fed oregano oil, which contains high concentrations of terpenes, similar to CIT (31). The mode of action of CIN or CIT against bacteria has been reviewed, and researchers have suggested that these compounds interact with the E. coli cell membrane, inducing rapid inhibition of energy metabolism and leakage of small ions without leakage of larger components of ADP (22, 29). However, the mechanisms of action by which CIN, CIT, and CIN+CIT modulate the prevalence of AMR are still unknown.
In agreement with our present study, a high prevalence of AMR in young birds has been previously reported in broiler chickens or laying hens (10, 24, 43). The possible explanations for the decline of AMR with increasing bird age include decrease in the fitness of the gut with increased bird body mass (10, 36). The decrease of AMR levels with increasing bird age also could be due to colonization of young birds with some resistant strains that are then replaced by susceptible strains with as the birds age (36). In a similar study, AMR levels were affected by bird age (24). Dietary changes among bird stages (starter, grower, and finisher), changes in facility temperatures, and changes in the richness and diversity of gut microbiota at different ages also could be factors that affect AMR in bacteria (37, 39, 46, 58). However, these possibilities need further study.
A clear correlation between AMR phenotype and genotype has been found (56). The low level of AMR found in birds fed CIN could be due to the damage of bacterial cell membranes the interaction of CIN with membrane proteins or interference with cellular energy production, which could then result in the disruption of efflux pumps and the expression of ARGs (71, 77). In the present study, four of the E. coli isolates recovered from fecal samples collected on day 6 harbored a previously described fosfomycin resistance gene with 94% similarity, highlighting the importance of monitoring the emergency of fosfomycin resistance (13).
MGEs such as plasmids play an important role in the spread of AMR via horizontal gene transfer. Therefore, new strategies able to decrease this transfer are needed. In the present study, a low prevalence of the plasmids IncFIB, IncI1, and IncA/C2 carrying ARGs was observed especially in CIN-fed birds, suggesting potential antiplasmid activity of this EO in E. coli. Little is known about the antiplasmid activity of CIN; however, this activity has been reported in compounds such as promethazine and menthol (60). Because plasmids can be horizontally transferred from one bacterial species to another, our study results indicate that in-feed CIN may help to slow down the transmission and spread of AMR. These data also suggest that the efficacy of EOs in chickens could be altered by bird age, but the mechanism is still unknown (12).
Virulence factors in bacteria allow attachment to and colonization of host cells (3) and iron uptake (2), resulting in host cell and tissue damage (25, 30). In the present study, the mean numbers of VGs in the E. coli isolates recovered from birds' fecal samples were not impacted by dietary treatments. The expressions of VGs could be affected by several factors such as pH, environmental conditions, and oxygen tension (68, 79). Data from the present study revealed that the number of VGs encoding flagella, chemotaxis, fimbria, pili, curli, iron intake proteins, toxins, secretion systems, enzymes, regulators, and miscellaneous factors was lower in young birds than in old birds. These results are inconsistent with those of a previous study in which pulsed-field gel electrophoresis revealed that the frequency of VGs was higher in young birds than in old birds (35). Both study results indicate that age is an essential factor that could modulate the number of VGs and that further study of this factor is needed.
E. coli is classified into serotypes based on somatic (O) and flagellar (H) antigens. Serotypes such as O1, O2, O4, O8, O15, O35, O78, O88, O109, and O115 have been associated with colibacillosis in poultry (32). In the present study, 66 E. coli serotypes were detected among 227 sequenced E. coli isolate genomes. Isolates of serotype O21:H16, which frequently have been associated with zoonotic infections (54), were the most prevalent (22 isolates: 6 in NC birds, 6 in BAC birds, 6 in CIT birds, 3 in CIN+CIT birds, and 1 in CIN birds). The virulence and pathogenicity of O21:H16 isolates were not evaluated in this study. However, 159 to 167 VGs were detected in these isolates, which were classified as APEC according to criteria described previously (9). The low number of isolates of this serotype in CIN birds suggests the potential of CIN to decrease the number of the potential pathogenic E. coli isolates. Two isolates of serotype O78:H4, which is closely associated with avian colibacillosis in poultry (27, 55), were detected in BAC birds on day 23. These two APEC isolates carried six ARGs and >199 VGs. Serotypes associated with bacteremia and urinary tract infections, including O4 and O15 (49), were more prevalent in 16- and 23-day-old birds.
Some MLSTs such as ST38 and ST117 have been associated with avian colibacillosis (16, 55). The MLSTs ST69, ST73, ST95, and ST131 are predominant in ExPEC isolates from human infections (66). In the present study, a high number of ST38 and ST117 were found in E. coli isolates from day 23, which indicates a risk of colibacillosis. CIN reduced the number of ST38, which suggests its potential for preventing colibacillosis. The phylogenetic tree also showed E. coli serotypes O15:H6, O16:H48, O25:H18, O78:H4, O81:H39, O103:H21, and O184:H4 clustered closely with clinical E. coli isolates used as controls in this study, indicating their potential pathogenicity in human or broiler infections. Multidrug-resistant E. coli of ST10 or ST101 have been reported to carry β-lactamase (blaCTX-M, blaNDM-1) or colistin resistance (mcr-1) genes (5, 26, 47). The present study revealed that AMR E. coli ST10 or ST101 from chicken feces may be pathogenic.
In conclusion, dietary CIN affected AMR phenotypes and genotypes of E. coli isolated from chicken feces. The underlying mechanisms need to be elucidated; however, antiplasmid activity could not be excluded. Bird age also affected the AMR phenotypes and genotypes of E. coli in broilers. Although no significant dietary treatment effect was observed on the number of detected VGs, bird's age appeared to influence the prevalence of VGs, serotypes, and MLSTs. Dietary CIN affected the presence of various plasmids harboring ARGs in these E. coli isolates, indicating the ability of this EO to decrease horizontal transfer of ARGs between E. coli and other chicken gut bacteria. More studies are necessary to determine the in vitro and in vivo activity of CIN against antibiotic-resistant bacteria before CIN can be used as an alternative to antibiotics to decrease food safety risks in broiler production.
This work was financially supported by Agriculture and Agri-Food Canada (AAFC) through the Genomics Research and Development Initiative to Mitigate Antimicrobial Resistance (project PSS 1858, J-001262). The authors acknowledge the staff of Centre de recherche en sciences animales de Deschambault who helped during the chicken trial, Paul A. Manninger for technical help with sequencing, Dr. Hai Yu (AAFC) for help with E. coli isolation, and Dr. Paula Azevedo (University of Manitoba) for assistance with manuscript revision.