Inc A/C Plasmids Are Prevalent in Multidrug-Resistant Salmonella enterica Isolates

Abstract

Bacterial plasmids are fragments of extrachromosomal double-stranded DNA that can contain a variety of genes that are beneficial to the host organism, like those responsible for antimicrobial resistance. The objective of this study was to characterize a collection of 437 Salmonella enterica isolates from different animal sources for their antimicrobial resistance phenotypes and plasmid replicon types and, in some cases, by pulsed-field gel electrophoresis (PFGE) in an effort to learn more about the distribution of multidrug resistance in relation to replicon types. A PCR-based replicon typing assay consisting of three multiplex PCRs was used to detect 18 of the 26 known plasmid types in the Enterobacteriaceae based on their incompatibility (Inc) replicon types. Linkage analysis was completed with antibiograms, replicon types, serovars, and Inc A/C. Inc A/C plasmids were prevalent in multidrug-resistant isolates with the notable exception of Salmonella enterica serovar Typhimurium. Cluster analysis based on PFGE of a subset of 216 isolates showed 155 unique types, suggesting a variable population, but distinct clusters of isolates with Inc A/C plasmids were apparent. Significant linkage of serovar was also seen with Inc replicon types B/O, I1, Frep, and HI1. The present study showed that the combination of Salmonella, the Inc A/C plasmids, and multiple-drug-resistant genes is very old. Our results suggest that some strains, notably serovar Typhimurium and closely related types, may have once carried the plasmid but that the resistance genes were transferred to the chromosome with the subsequent loss of the plasmid.

Bacterial plasmids are self-replicating extrachromosomal fragments of double-stranded DNA. They range in size from a few to several hundred kilobase pairs. Plasmids can contain a variety of genes that are beneficial to the survival of the host bacteria. These sequences can encode antimicrobial and heavy metal resistance, toxin production, or virulence factors that allow their bacterial host to adapt to changing environments (2, 12, 35). The classification and tracking of plasmids is beneficial because they are potentially a medium of horizontal gene transfer of drug resistance (5, 12, 24). Horizontal transfer of DNA in prokaryotes occurs in three forms: transformation, conjugation, or transduction (12, 30). Plasmids contain a replicon that consists of sequences that are necessary for self-replication in a host cell, which includes the origin of replication, control of initiation, and replication functions (5, 10). Plasmids can be classified according to incompatibility (Inc) types that are based on the inability of plasmids with the same replication mechanism to exist in the same cell (10, 22). Different plasmids of the same Inc type cannot coexist in one bacterial strain. In the Enterobacteriaceae, there are 26 known Inc types or replicon types. A PCR-based replicon typing assay has been developed to distinguish 18 replicon types (6, 18, 19).

Certain replicon types are associated with multidrug resistance as well as bacteria implicated in disease outbreaks or found in food-producing animals. Multidrug-resistant (MDR) Salmonella strains are responsible for human outbreaks and may be acquired through food animals, which are a major source of zoonotic pathogens (3, 16, 23, 25, 35, 36). Twenty-two MDR Salmonella enterica serovar Heidelberg isolates from turkey-associated sources were found to have XbaI pulsed-field gel electrophoresis (PFGE) profiles that were indistinguishable from the most common profile associated with human infection. Conjugation experiments confirmed that the one tested representative isolate was able to transfer a large plasmid of approximately 120 kb (20). The Inc A/C plasmid backbone from MDR Salmonella enterica serovar Newport was found to have a backbone similar to that of plasmid pIP1202 from Yersinia pestis, the causative agent of the plague, as well as that of plasmid pYR71 from Yersinia ruckeri, a fish pathogen (34). MDR Salmonella enterica serovar Newport with plasmid-mediated extended-spectrum cephalosporin resistance was isolated from animals during a salmonellosis outbreak that led to the closure of the large animal hospital at the University of Pennsylvania's New Bolton Center (26). In clinical human isolates of the Enterobacteriaceae, Inc A/C or Inc N plasmids have been shown to be associated with extended-spectrum cephalosporin and carbapenem resistance determinants emerging in Greece, Italy, and the United States (7). Inc I1 and A/C replicon types are associated with plasmids carrying and disseminating extended-spectrum β-lactamase genes in animals and humans (9, 13, 18). The Inc HI1 replicon type is associated with an MDR plasmid, plasmid pHCM1, found in Salmonella enterica serovar Typhi, recovered during an MDR typhoid fever outbreak in Vietnam from 1993 to 1996 (33). Salmonella enterica serovar Paratyphi A was found to contain an MDR IncHI1 plasmid, plasmid pAKU_1, and, in some regions, can be the causative agent of enteric fever (17). To further study the Inc A/C plasmid, we characterized the antimicrobial resistance profiles and the presence of replicon types in 437 ampicillin- and tetracycline-resistant Salmonella enterica isolates recovered from ill animals in 2005. Amp- or Tet-resistant isolates were selected because they are more likely to have a plasmid.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Salmonella enterica strains in this study were collected from samples submitted to veterinary diagnostic laboratories in 2005 in which Salmonella was identified as being either the primary or secondary etiological agent associated with the illness, as previously described (http://www.ars.usda.gov/Main/docs.htm?docid=6750&page=1). Veterinary diagnostic laboratories obtained isolates from the state in which they are located; the isolates were sent to the National Veterinary Services Laboratory (NVSL), Ames, IA (www.aphis.usda.gov/animal_health/lab_info_services/about_nvsl.shtml); and we obtained the isolates from the NVSL. All strains tested for plasmid Inc types were resistant to ampicillin and/or tetracycline. The 437 isolates in this study (see Table S1 in the supplemental material) represent 55 serovars (including nonmotile and untypeable) from 17 different sources including canine (dog), cattle, chicken, dairy cattle, equine, environment, feline (domestic cat), parrot, reptile (turtle and lizard), swine, turkey, wild avian (pheasant and unknown spp.), wild mammal (alpaca, mongoose, and unknown spp.), and wild rodent (unknown spp.). Strains were obtained from 28 states in all five regions in the United States as defined for the NARMS program (http://www.ars.usda.gov/Main/docs.htm?docid=6750) and included region 1, the Northeast (Connecticut, Delaware, Indiana, Massachusetts, Maryland, Maine, Michigan, New Hampshire, New Jersey, New York, Ohio, Pennsylvania, Rhode Island, and Vermont); region 2, the Southeast (Alabama, Florida, Georgia, Kentucky, North Carolina, Puerto Rico, South Carolina, Tennessee, Virginia, and West Virginia); region 3, the Midwest (Iowa, Illinois, Kansas, Minnesota, Missouri, North Dakota, Nebraska, South Dakota, and Wisconsin); region 4, the Southwest (Arkansas, Louisiana, Mississippi, Oklahoma, and Texas); and region 5, the West (Arizona, California, Colorado, Idaho, Montana, New Mexico, Nevada, Oregon, Utah, Washington, and Wyoming).

Positive controls used in the replicon typing procedure were originally created in the laboratory of Werner K. Maas (6, 10) and generously provided by Alessandra Carattoli (Istituto Superiore di Sanita, Rome, Italy). All bacterial strains were stored at −80°C in LB lennox (Hardy Diagnostics, Santa Maria, CA) with 15% glycerol or stored at room temperature in tryptic soy agar (Hardy Diagnostics, Santa Maria, CA) slants until use.

Antimicrobial susceptibility testing.

Each Salmonella enterica isolate was tested for susceptibility to a panel of 15 antimicrobial drugs using the Sensititer system (Trek Diagnostic Systems Inc., Westlake, OH) that included amikacin, amoxicillin-clavulanic acid, ampicillin, cefoxitin, ceftiofur, ceftriaxone, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, sulfamethoxazole, tetracycline, and trimethoprim-sulfamethoxazole as defined by the NARMS program (http://www.ars.usda.gov/Main/docs.htm?docid=6750). Each isolate was classified as being susceptible, intermediate, or resistant using Clinical and Laboratory Standards Institute (CLSI; formerly National Committee for Clinical Laboratory Standards) breakpoints when available; otherwise, breakpoint interpretations from the National Antimicrobial Resistance Monitoring System were used (21). For linkage analyses described below, intermediate sensitivity was not distinguished from the susceptible trait.

Multiplex PCR for plasmid replicon typing.

Salmonella isolates were examined by PCR using three multiplex primer panels for the presence of 18 plasmid replicons as described previously by Johnson et al. (19). This replicon typing procedure is a modified version of eight PCR multiplex and simplex reactions described previously by Carattoli et al. (6). Primers were obtained from a Eurofins MWG operon (Huntsville, AL). Inc-type reference plasmid DNA was extracted from 2 ml LB broth cultures grown overnight with appropriate antibiotics and processed with a Qiaprep spin miniprep kit (Qiagen, Valencia, CA). Template DNA for PCR from the 437 Salmonella isolates was prepared by suspending a single colony in 200 μl sterile water and treating in a boiling water bath for 10 min. PCRs were performed according to the polymerase manufacturer's instructions, with a final volume of 25 μl: 5 μl of boiled lysate or 10 ng of each reference plasmid, 1× AmpliTaq buffer 1, 0.50 μM of each primer, 4 mM MgCl₂, 0.2 mM of each deoxynucleoside triphosphate, and 1.25 units of AmpliTaq polymerase (Applied Biosystems, Foster City, CA). Positive controls as well as a negative control without DNA were run with each multiplex primer panel. PCR cycle conditions were as follows: 5 min at 94°C; 30 cycles of 30 s at 94°C, 30 s at 60°C, and 90 s at 72°C; and a final extension step of 5 min at 72°C. Amplicons were visualized on 1× Tris-borate-EDTA 2% agarose gels run for 2 h at 80 V alongside a TrackIt 1-kb Plus DNA ladder (Invitrogen Corporation, Carlsbad, CA). An isolate was considered to be positive for a particular gene if an amplicon of the expected size was observed.

PFGE.

A total of 223 isolates of the original 437 strains were analyzed by a 24-h Salmonella PFGE protocol as described by PulseNet (8) at the USDA VetNet Laboratory (Athens, GA). Briefly, genomic DNA was digested with 10 U of XbaI (Roche Molecular Biochemicals, Indianapolis, IN) and separated with the CHEF-Mapper XA PFGE system (Bio-Rad, Hercules, CA) in 0.5× Tris-borate-EDTA buffer at 14°C at 6 V for 18 h with a ramped pulse time of 2.16 to 63.8 s. The BioNumerics software program (Applied Maths Scientific Software Development, Sint-Martens-Latem, Belgium) was applied for cluster analysis using the Dice-based coefficients with a 1.5% band tolerance and 1.5% optimization with coefficient and the unweighted pair-group method.

DT104 analysis.

Pentaresistant (ampicillin, chloramphenicol, streptomycin, sulfonamides, and tetracycline [ACSSuT] resistance) Salmonella enterica serovar Typhimurium isolates were phage typed at the National Veterinary Services Laboratory, Ames, IA (www.aphis.usda.gov/animal_health/lab_info_services/about_nvsl.shtml).

Statistical analysis.

Linkage disequilibrium (LD) was calculated as an extension of Fisher's exact probability test on contingency tables (27) as instituted by the program Arlequin (11). Standard settings were used, 10,000 steps in the Markov chain and 1,000 dememorization steps; and calculations of D, D′, and r² coefficients were made with a significance level of 0.05.

RESULTS

Characterization of Salmonella enterica isolates.

The total number of Salmonella enterica diagnostic isolates in the collection was 1,548; 842 (54%) of these were Amp or Tet resistant. Strains that were identical in all of three traits, serovar, state/region, and host animal, were removed from the set to decrease redundancy. We characterized the remaining 437 Salmonella enterica isolates for their antimicrobial resistances and replicon types. These isolates represent 55 serovars from 17 different host sources and were obtained from 28 states in five regions of the United States (see Table S1 in the supplemental material). When the original parameters of Amp or Tet resistance were examined in the characterized strains, 224 (51%) were both Amp and Tet resistant, 47 (11%) were only Amp resistant, and 166 (38%) were only Tet resistant. The top four prevalent antimicrobial resistances by percent in the population are tetracycline (89.2%), streptomycin (74.8%), sulfamethoxazole (73.2%), and ampicillin (62.0%) (Table 1). Ciprofloxacin is the only tested drug that did not show resistance in any of the strains, and it was not included in further analyses. Sixty-five percent of the strains analyzed show resistance to four or more of the antimicrobials on our panel.

TABLE 1.

Antimicrobial resistance of strains in this study^a

Antimicrobial resistance trait	No. (%) with profile
Tetracycline	390 (89.2)
Streptomycin	327 (74.8)
Sulfamethoxazole	320 (73.2)
Ampicillin	271 (62.0)
Chloramphenicol	172 (39.4)
Kanamycin	170 (38.9)
Ceftiofur	132 (30.2)
Amoxicillin-clavulanic acid	132 (30.2)
Cefoxitin	131 (30.0)
Gentamicin	82 (18.8)
Trimethoprim-sulfamethoxazole	54 (12.3)
Ceftriaxone	14 (3.2)
Nalidixic acid	6 (1.4)
Amikacin	2 (0.5)
Ciprofloxacin	0 (0.0)

^aShown are the numbers and percentages of isolates exhibiting resistance to at least Amp and/or Tet and each of the other antimicrobials tested.

Replicon typing, a PCR-based assay consisting of three multiplex primer panels, was conducted on all Salmonella isolates for the presence of 18 different plasmid based Inc types (19). The most prevalent replicon types by percent in the population that we characterized were Inc A/C (26.3%), I1 (23.6%), HI1 (6.4%), and FIIA (2.5%) (Table 2). Forty-seven percent of our strains did not show a positive reaction for any replicon type, but our assay detects only 18 of the 26 known replicon types in the Enterobacteriaceae. We did not identify any isolates that carried the FIC, K/B, T, or W Inc replicon type.

TABLE 2.

Plasmid replicon typing of strains in this study^a

Inc/Rep type	No. (%) with profile
None	208 (47.5)
A/C	115 (26.3)
I1	103 (23.6)
HI1	28 (6.4)
FIIA	11 (2.5)
Frep	11 (2.5)
FIB	4 (0.9)
N	4 (0.9)
P	2 (0.5)
Y	2 (0.5)
X	2 (0.5)
B/O	2 (0.4)
FIA	1 (0.2)
HI2	1 (0.2)
L/M	1 (0.2)
FIC	0 (0.0)
K/B	0 (0.0)
T	0 (0.0)
W	0 (0.0)

^aShown are numbers of isolates and the percentages of our total sample population with each replicon type tested.

Inc A/C was present in 115 strains and 21 out of 53 serovars (not including nonmotile and untypeable) (Table 3). Inc A/C was not found in isolates from two states in separate regions, and each of these consisted of only one isolate. Inc A/C was found in all source species except for wild rodent, parrot, pheasant, and unknown avian spp.; each of these groups consisted of only one isolate (Table 3). Recognizing the selection for the study of isolates that are Amp or Tet resistant, in the 93 Inc A/C-positive strains, 88 (95%) were both Amp and Tet resistant, none were only Amp resistant, and 5 (5%) were only Tet resistant. When the Amp or Tet resistance was examined in the 81 Inc I1-positive strains, 21 (26%) were Amp and Tet resistant, 11 (14%) were only Amp resistant, and 49 (60%) were only Tet resistant. In the 22 isolates that were both Inc A/C and Inc I1 positive, 21 (95%) were both Amp and Tet resistant, 0 (0%) were only Amp resistant, and 1 (5%) was only Tet resistant. In the remaining 241 isolates that were not Inc A/C or Inc I1 positive, 94 (39%) were both Amp and Tet resistant, 36 (15%) were only Amp resistant, and 111 (46%) were only Tet resistant.

TABLE 3.

Inc A/C prevalence by population^a

S. enterica serovar or source species	No. of Inc A/C-positive isolates in population/total no. of isolates (%)
S. enterica serovar
Agona	12/18 (67)
6,7:-:l,w	1/2 (50)
Bardo	3/3 (100)
Bredeney	4/5 (80)
Choleraesuis variant kunzendorf	3/15 (20)
Derby	5/41 (12)
Dublin	1/3 (33)
Havana	1/3 (33)
Heidelberg	4/27 (15)
Infantis	1/4 (25)
London	1/1 (100)
Mbandaka	2/6 (33)
Newport	40/44 (91)
Nonmotile	1/2 (50)
Ohio	3/3 (100)
Reading	9/9 (100)
Typhimurium	9/38 (24)
Typhimurium variant 5−	2/55 (4)
Uganda	8/9 (89)
Yovokome	1/1 (100)
Source species
Alpaca	1/1 (100)
Canine	7/11 (64)
Cattle	38/80 (47.5)
Chicken	3/24 (12.5)
Dairy cattle	9/26 (11.5)
Environmental	2/8 (25)
Equine	17/30 (57)
Feline	1/4 (25)
Lizard	1/1 (100)
Miscellaneous mammal	2/3 (67)
Swine	25/178 (14)
Turkey	6/61 (10)
Turtle	2/2 (100)

^aShown are numbers of Inc A/C-positive isolates/total numbers of isolates per population, with the percentages in parenthesis.

Pairwise LD between the antimicrobial resistance profile and Salmonella enterica isolate serovar showed a linkage (P value of <0.05) with all the antimicrobials tested except amikacin and nalidixic acid (data not shown). Resistance to these two antimicrobials showed low representation in our population, with nalidixic acid at 1.4% and amikacin at 0.5% (Table 1). Pairwise LD between antimicrobial resistance profiles and Inc A/C replicon type of Salmonella enterica isolates shows linkage (P value of <0.05) with all antimicrobials tested except amikacin, ceftriaxone, ciprofloxacin, nalidixic acid, streptomycin, and sulfamethoxazole (data not shown). Pairwise LD between 14 replicon types and Salmonella enterica serovars show linkage (P value < 0.05) of serovar and Inc A/C, B/O, Frep, HI1, and I1 (Fig. 1). A nonrandom association is seen between strain serovar and the Inc A/C, B/O, Frep, HI1, I1, and A/C subsets (Fig. 1).

FIG. 1.

Pairwise linkage disequilibrium among 14 replicon types, serovars (ST) of Salmonella enterica strains, and Inc A/C types based on a PFGE dendrogram. A “+” indicates a P value of 0.05 or less, indicating significant linkage, and a “−” indicates a P value of greater than 0.05. ¹, 18 subsets were used for LD. Seventeen subsets contained two or more Inc A/C-positive isolates in clades with 80% or better identity by PFGE. One subset consisted of seven individuals of serovar Derby, all Inc A/C negative, and showed 100% identity by PFGE. The remaining isolates were designated members of a null group, and the LD was calculated and listed as a subset.

PFGE analysis was conducted on 216 of the 437 total isolates including 98 of 104 Inc A/C-positive isolates (Fig. 2). Six strains were not included in the cluster analysis because their PFGE patterns were inconsistent with serovar results (E. McGlinchey, personal communication). PFGE analysis was conducted on 118 Inc A/C-negative isolates that were randomly selected, including 49 Inc I1-positive strains (51 additional isolates containing Inc I1 were identified but were not included in PFGE analysis) as well as 69 other strains that were picked as a reference set. Cluster analysis based on PFGE showed 155 unique types, suggesting a variable population. The source animal of the isolate did not appear to have a correlation to cluster analysis. Inc A/C-positive strains have a high degree of multidrug resistance compared to strains without any detected replicon type, which have a reduced antimicrobial profile. However, Inc A/C-positive strains of Salmonella enterica serovar Choleraesuis variant Kunzendorf were resistant to sulfamethoxazole and tetracycline only. Inc A/C-positive strains form clusters, whereas Inc I1 appears random, and not enough information was available to detect linkage patterns among the B/O, Frep, and HI1 Inc types (Fig. 2).

FIG. 2.

PFGE-based dendrogram, replicon typing, resistance profile, and strain serovar. The dendrogram is based on PFGE analyses using BioNumerics software. Epidemic clones are labeled A to G. Replicon typing results are to the right of the dendrogram; black indicates that the trait is present, and gray indicates that the trait is absent (columns 2 to 15 indicate replicon type). For the antimicrobial resistance phenotype (columns 16 to 28), black indicates resistance; white indicates intermediate or susceptible and the isolate serovar (column 29). ¹, antimicrobial resistance phenotype: amoxicillin-clavulanic acid (AMO), ampicillin (AMP), cefoxitin (FOX), ceftiofur (TIO), ceftriaxone (AXO), chloramphenicol (CHL), gentamicin (GEN), kanamycin (KAN), nalidixic acid (NAL), streptomycin (STR), sulfamethoxazole (SUL), tetracycline (TET), and trimethoprim-sulfamethoxazole (TRI).

This study included 38 isolates of serovar Typhimurium and 55 isolates of serovar Typhimurium variant 5−. Fifty percent of the population of Salmonella enterica serovar Typhimurium or serovar Typhimurium variant 5− isolates were pentaresistant (ACSSuT) (or more) and phage typed. Thirty-four percent of the isolates of these serovars were DT104 positive (Table 4). Ten percent were Inc A/C-positive and pentaresistant (or more), and only 1% were Inc A/C and DT104 positive (Table 4).

TABLE 4.

Prevalence of pentaresistance, DT104, and Inc A/C plasmids in strains in this study^a

Prevalence of pentaresistant plasmids	Prevalence of:		No. (%) of positive isolates
	DT104	Inc A/C	Serovar Typhimurium	Serovar Typhimurium variant 5−	Serovars Typhimurium and Typhimurium variant 5−
+	−	−	4 (10.5)	5 (9)	9 (9.7)
+	+	−	13 (34)	18 (33)	31 (33)
+	+	+	0	1 (2)	1 (1)
−	+	+	0	0	0
−	−	+	1 (2.6)	0	1 (1)
−	+	−	1 (2.6)	2 (3.6)	3 (3.2)
+	−	+	8 (21.5)	1 (1.8)	9 (9.6)
−	−	−	11 (29)	28 (51)	39 (42)

^aThe three leftmost columns indicate the presence (+) or absence (−) of the phenotype. The remaining columns contain the numbers of positive individuals and the percentages of our total serovar population in parenthesis. The column to the far right contains the combined totals of serovar Typhimurium and serovar Typhimurium variant 5− columns.

DISCUSSION

The objective of this study was to characterize a large population of Salmonella enterica isolates from different animal sources for their antimicrobial resistance phenotypes and plasmid replicon type and, in some cases, by PFGE in an effort to learn more about the distribution of multidrug resistance in relation to Inc A/C as well as the other replicon types which we examined. PFGE was conducted on approximately one-half of the study population and was used for cluster analysis. Linkage analysis was completed with antibiograms, replicon types, serovars, and Inc A/C groups.

Cluster analysis based on PFGE of 216 isolates showed 155 unique types. Cluster analysis showed that Inc A/C-positive strains form groups based on multidrug resistance and serovar, whereas Inc I1-positive isolates did not appear to be clonally distributed. This indicated that Inc A/C plasmids have been stably associated with clones for a very long time (however, molecular clocks cannot be accurately constructed with a phylogenetic analysis of PFGE) (29), while Inc I1 plasmids are much more mobile. Inc A/C plasmids are large, >150 kb, while Inc I1 plasmids are frequently approximately 100 kb (9, 34). Larger plasmids usually transfer at lower frequencies than smaller plasmids and can be expected to be more stable (14). It can also be inferred from the stable association with clones with similar resistance phenotypes that the plasmid itself is very old and that the linkage of the resistance genes does not represent a recent accretion.

The cluster analysis shows seven groups that may be considered to be epidemic clones, highly successful clones, which, for the sake of this discussion, were considered to be PFGE types (28) with three or more identical individuals (labeled clones A to G) (Fig. 2). There is some replicon type diversity in these clones; two have diversity with regard to Inc A/C (clones A and B), three have diversity with regard to Inc I1 (clones C, D, and E), two have diversity with regard to Inc F1B (clones B and C), and one does not have any detected replicon types associated with it (clone G). We also found diversity in the antimicrobial resistance profiles associated with these seven clones; one clone has no diversity with regard to antimicrobials (clone G), two clones have diversity with regard to one antimicrobial (clone A and F), three clones have diversity with regard to two antimicrobials (clone B, C, and E), and one clone has diversity with regard to three antimicrobials (clone D).

Most of the Salmonella enterica isolates in this study that carried the Inc A/C plasmids were multidrug resistant, which is consistent with the findings described previously by Welch et al. (34). We observed that Inc A/C-positive strains carry more antimicrobial resistance than isolates with other replicon types, so it was of interest to note a clone with three Inc A/C-positive individuals of serovar Choleraesuis variant Kunzendorf, which were resistant to sulfamethoxazole and tetracycline only. It is possible that these three isolates carry other resistance genes that are not expressed. All of the Salmonella enterica serovar Choleraesuis variant Kunzendorf strains in this study originated from swine in region 3, and they had antimicrobial resistance profiles of two to five drugs. We infer from the PFGE analysis that the serovar Choleraesuis variant Kunzendorf clone is a stable clone that is being amplified and is undergoing little change despite the lack of important antimicrobial resistance genes.

It is interesting to compare the clades that include serovars Typhimurium and Heidelberg with the clade that includes serovar Newport or the clade that includes serovar Derby. Pentaresistance (ACSSuT resistance) was seen to be common among Salmonella enterica serovar Typhimurium isolates, although Inc A/C plasmids were not prevalent in the clade, and other plasmid types were scattered throughout the clade. Pentaresistance in serovar Typhimurium is often associated with phage type DT104, which has multiple resistance genes located in an island on the chromosome (4, 15, 31, 32). In this study, serovar Typhimurium and serovar Typhimurium variant 5− ACSSuT (or more) resistance was more frequently associated with carrying the DT104 phenotype than having the Inc A/C plasmid (Table 4): 34% of the isolates of these serovars were DT104 positive and pentaresistant, while 10% were Inc A/C positive and pentaresistant (or more). In contrast, the clade that includes serovar Newport isolates has a high prevalence of Inc A/C-type plasmids associated with resistance to eight or more antimicrobials, and the clade that includes serovar Derby has Inc A/C plasmids infrequently and little resistance to beta-lactam antimicrobials. From these considerations, we hypothesize that the serovar Newport clade had an Inc A/C plasmid at its origin, while the serovar Derby isolates with Inc A/C plasmids may represent recent acquisitions. The serovar Typhimurium clade may have had an Inc A/C plasmid early in its genesis, but it was largely lost, perhaps after transferring genes that helped to stabilize the plasmid, including antimicrobial resistance genes, to the chromosome. Welch et al. (34) previously compared the DNA sequences of three Inc A/C plasmid backbones: pSN254 from an MDR Salmonella enterica serovar Newport isolate; pIP1202 from Yersinia pestis, the causative agent of the plague; as well as plasmid pYR71 from an isolate of the fish pathogen Yersinia ruckeri that was found to be resistant to a large number of antimicrobials. When the sequence of the DT104 resistance island (GenBank accession number AF071555) (4) was compared to these plasmids by BLAST (1), there was a very high degree of similarity of resistance genes, supporting the hypothesis that the genes have a common origin. An alternative hypothesis is that the MDR genes originated in serovar Typhimurium and were transferred to an Inc A/C plasmid that was not amplified with serovar Typhimurium. With PFGE data, it is difficult to distinguish between the two hypotheses because proper rooting of the tree cannot be established (29). However, when PFGE data were reanalyzed, including 15 isolates of serovar Typhimurium that were pansusceptible and Inc A/C negative, they fell into separate clades that appear more ancient than the clades representing the MDR strains in several alternative rootings (data not shown).

In conclusion, the present study showed that the combination of Salmonella species, the Inc A/C plasmids, and MDR genes is very old. Our results suggest that some strains, notably serovar Typhimurium and closely related types, used to carry the plasmid and had a transfer of resistance genes to the chromosome with a subsequent loss of the plasmid. This means that MDR Salmonella strains have been around in numbers too low to be found for much longer than humans had any influence, that there has not been substantial new accretion of resistance by a recent acquisition of existing genes into new strains, and that any increase in levels of these types of strains represents clonal expansion that may or may not be driven by modern practices.