Association of ISVsa3 with Multidrug Resistance in Salmonella enterica Isolates from Cattle (Bos taurus)

Salmonella enterica is, globally, an important cause of human illness with beef being a significant attributable source. In the human patient, systemic Salmonella infection requires antibiotic therapy, and when strains are multidrug resistant (MDR), no effective treatment may be available. MDR in bacteria is often associated with the presence of mobile genetic elements (MGE) that mediate horizontal spread of antimicrobial resistance (AMR) genes. In this study, we sought to determine the potential relationship of MDR in bovine Salmonella isolates with MGE. The present study involved 111 bovine Salmonella isolates obtained collectively from specimens derived from healthy cattle or their environments at Midwestern U.S. feedyards (2000–2001, n = 19), or specimens from sick cattle submitted to the Nebraska Veterinary Diagnostic Center (2010–2020, n = 92). Phenotypically, 33/111 isolates (29.7%) were MDR (resistant to ≥3 drug classes). Based on whole-genome sequencing (WGS; n = 41) and PCR (n = 111), a MDR phenotype was strongly associated (OR = 186; p < 0.0001) with carriage of ISVsa3, an IS91-like Family transposase. In all 41 isolates analyzed by WGS ((31 MDR and 10 non-MDR (resistant to 0–2 antibiotic classes)), MDR genes were associated with carriage of ISVsa3, most often on an IncC type plasmid carrying blaCMY-2. The typical arrangement was floR, tet(A), aph(6)-Id, aph(3″)-Ib, and sul2 flanked by ISVsa3. These results suggest that AMR genes in MDR S. enterica isolates of cattle are frequently associated with ISVsa3 and carried on IncC plasmids. Further research is needed to better understand the role of ISVsa3 in dissemination of MDR Salmonella strains.


Introduction
Salmonella enterica subsp. enterica (S. enterica) is, globally, an important cause of human illness, and in the United States (U.S.), it is estimated to cause 1.35 million infections, 26,500 hospitalizations, and 420 deaths each year [1]. Although ranking behind seeded vegetables, eggs, and poultry, beef is a significant attributable source of S. enterica [2]. Specific to nontyphoidal Salmonella, beef ranks 14th, 8th, and 8th out of the top 37 pathogen-food pairs in burden of illness in the U.S. in terms of number of illnesses, basic cost, and economic cost, respectively [3]. In the human patient, systemic Salmonella infection requires antibiotic therapy [4,5], and when the strain is MDR (resistant to ≥3 antibiotic classes) [6], the case is particularly problematic. First-line antibiotic therapy for systemic Salmonella infections in humans includes third-generation cephalosporins (e.g., ceftriaxone), fluoroquinolones (e.g., ciprofloxacin), and macrolides (e.g., azithromycin) [4,5]. However, fluoroquinolones have adverse side effects in children and pregnant women [7,8], in which case ceftriaxone and azithromycin are the drugs of choice. Unfortunately, ceftriaxone and ciprofloxacin resistance in human Salmonella isolates has increased in recent years [9,10].
S. enterica is also a primary pathogen in cattle, mainly causing enteritis in calves between 2 and 6 weeks of age, but can also cause enteritis, pneumonia, and abortions in adult All S. enterica strains used in this study (n = 111) were isolates from 2000 to 2001 (n = 19) or 2011 to 2020 (n = 92). The 19 isolates from 2000 to 2001 were a subset of 530 isolates from feedlot beef cattle feces or their pen environments in Midwestern U.S. feedyards [25] ( Table 1). All 530 isolates had been serotyped and tested for antimicrobial susceptibility phenotype in 2006 using a standardized National Animal Resistance Monitoring System (NARMS) protocol, and also were tested by PCR for class 1 integron genes [26]. Of the 530 isolates from that study, 0 were positive for class 1 integron genes by PCR; however, based on the NARMS 2006 results, 13 were MDR (Table 1). These 13 MDR isolates were selected for inclusion in the present study; 6 other isolates that were resistant to only 1 antimicrobial that originated from the same sample or pen of cattle were also included for whole-genome sequence (WGS) and/or other test comparisons. Immediately prior to this study, all 19 strains were retested for antimicrobial susceptibility using the Sensititre BOPO7F veterinary plates (Thermo Fisher Scientific, Waltham, MA, USA) to provide additional data regarding their susceptibility to antimicrobials in current use for cattle, including those relevant for respiratory pathogens (Table 1). Based on the BOPO7F results, 2 of the 13 strains had become pan-susceptible in storage. Eleven of the 13 strains, including the 2 that had become pan-susceptible, and 6 non-MDR strains from the same study [26], were selected for WGS (Table 1). M. haemolytica strain 2308 was isolated from a bovine clinical sample by the NVDC from a diagnostic submission. This isolate had previously been determined by Sanger sequencing to contain eight AMR genes associated with ICEPmu1, namely aphA1, sul2, tetR(H), floR, erm (42), bla OXA2 , msr(E), and mph(E), with four of these genes also associated with ICEMh1; all eight genes had been detected by a mqPCR assay (described below) that had been co-developed by one of the authors (J.D.L.) [27], hence M. haemolytica strain 2308 was validated for use as positive control for this mqPCR assay. E. coli strain 25922 (American Type Culture Collection) is a laboratory strain that does not possess these genes, and was used as a negative control for the mqPCR assay (described below).

Isolates from Nebraska Veterinary Diagnostic Center from 2011 to 2020
A second source of S. enterica strains was accessions from cattle systems to the Nebraska Veterinary Diagnostic Center (NVDC) during the period of 2011-2020. Of 98 Salmonella isolates identified from these accessions, 92 were viable from frozen stocks and included in the present study. Of these 92 isolates, 83 were from 42 of the 93 counties in Nebraska; 5 were from Missouri; and 1 each was from California, Colorado, Idaho, and Iowa. The 92 isolates had been serotyped previously as part of the diagnostic process, and included 27 different serotypes with S. Typhimurium (including 3 var. 5-) (n = 18), S. Newport (n = 13), S. Dublin (n = 10), S. Montevideo (n = 8), and S. Muenster (n = 7) constituting the 5 most prevalent and 60.9% (56/92) of the total (Table S1). A signalment (e.g., age, sex) and clinical history (e.g., diarrhea, abortion) and/or pathology data (e.g., enteritis, pneumonia) was provided in association with 84 (91.3%) and 78 (84.8%) of the cases, respectively. Salmonellae were most commonly isolated from accessions involving unweaned/neonatal calves (38.0%) and cows/heifers (29.3%) (Table S2). Overall, based on the clinical history and accompanying laboratory results, the Salmonella isolates were associated with disease (i.e., salmonellosis) in 81.5% of the accessions. Diarrheal disease (enteritis/colitis) and pneumonia were the most common manifestations, reported in 53.3% and 20.7% of the accessions, respectively.
The NVDC isolates were subjected to antimicrobial susceptibility testing either at the time of the accession or immediately prior to this study, if they had not been previously tested. Testing was conducted at the time of the accession either using the Sensititre BOPO6F or BOPO7F plate formats (Thermo Fisher Scientific Waltham, MA, USA), or immediately prior to this study using the Sensititre BOPO7F. Antimicrobial susceptibility testing was conducted using Clinical and Laboratory Standards Institute (CLSI, Annapolis, MD, USA) methods and recommended quality control strains for the broth microdilution assay [28]. Veterinary specific breakpoints were applied when available [29]. Of the 92 NVDC isolates, 22 (23.9%) were MDR, with serotypes S. Dublin (n = 10) and S. Newport (n = 5) combined representing 68.1% (15/22) of the MDR isolates ( Table 2). Of the 22 MDR isolates, resistance was most frequent to florfenicol and sulfadimethoxine (95.4% each), followed sequentially by 1 or more of the tetracyclines (chlortetracycline, oxytetracycline, and/or tetracycline; 90.9%), ceftiofur (77.3%), and a fluoroquinolone (danofloxacin and/or enrofloxacin; 40.9%). Resistance to macrolides (clindamycin, gamithromycin, tiamulin, tilmicosin, tildipirosin tulathromycin, and tylosin tartrate) was considered intrinsic and not reported. Aminoglycoside test results (gentamicin, neomycin, and spectinomycin) were also largely excluded since breakpoints and assessments of susceptibility or resistance for Salmonella are difficult to determine. Twenty of the 22 MDR strains were selected for WGS.

Culture of Bacterial Strains and DNA Preparation
Frozen stock cultures (−80 • C) of Salmonella strains were streaked for isolation onto Luria Broth (Miller, Appleton, WI, USA; LB) Agar (Becton, Dickinson and Company, Sparks, MD, USA) and incubated 18-24 h at 37 • C. A single well-isolated colony was used to inoculate 5 mL LB, and this culture was incubated 24 h, stationary at 37 • C. A 2-mL aliquot of broth culture was moved into the GeneJET DNA Genomic Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA) to prepare the DNA template for mqPCR reactions. Extractions were performed according to the manufacturer's protocol. Frozen stock cultures (−80 • C) of Mannhemia haemolytica control strains were streaked for isolation onto Trypticase Soy Agar (TSA) with 5% Sheep Blood (BD) and incubated 18-24 h at 37 • C. A single well-isolated colony was used to inoculate 50 mL Brain Heart Infusion (BHI) in a 250 mL Erlenmeyer flask, aerated at 150 rpm for 24 h, at 37 • C. A 2 mL aliquot of broth culture was transferred into the GeneJET DNA Genomic Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA) to prepare the DNA template for mqPCR reactions. Extractions were performed according to the manufacturer's protocol. Purified DNA concentration for each isolate was determined via NanoDrop One (Thermo Fisher Scientific, Waltham, MA, USA) with M. haemolytica strain 2308 and E. coli strain 29522 as the positive and negative organismal controls, respectively, for the mqPCR assay (described below) and Invitrogen UltraPure Water (Thermo Fisher Scientific, Waltham, MA, USA) in place of DNA as the negative reaction control.

Endpoint PCR
Frozen stock cultures (−80 • C) of bacterial strains were streaked for isolation onto Luria Broth (Miller; LB) Agar (Becton, Dickinson and Company, Sparks, MD) and incubated 18-24 h at 37 • C. A single well-isolated colony was picked, suspended in 50 µL of UltraPure Water, and heated at 95 • C in the thermocycler for 10 min. A 2.0 µL aliquot of this DNA template was used in the 25-µL PCR reaction. Individual 25-µL reaction PCR assays were conducted using primer pairs as shown in Table S5 [27,30,31]. The PCR reaction consisted of 2.5 µL 10X ThermoPol Reaction Buffer, 1.0 µL of each forward and reverse primer (10 µM each, 2 µL total), 0.5 µL dNTP mix (

Sequencing Quality Control and Genome Assembly
Following sequencing, Illumina short read quality was assessed using FastQC v0.11.7 (Babraham Bioinformatics, 2018, Babraham Institute, Cambridge, UK). BBDuk (v37.62) was used to trim adapters from the lllumina short reads, and any short reads with average quality score (Q score) below 30 were discarded. The porechop (v0.2.4) was used to trim adapters from the Nanopore long reads. The average sequencing depth of each isolate was estimated by dividing the total length of its cleaned reads by the genome size. Additionally, Illumina reads and Nanopore long reads from each isolate were hybrid de novo assembled using Unicycler (v0.4.9). Isolates with unclosed genomes were reassembled by Raven (v1.5.1) with Nanopore long reads and then polished by Pilon (v1.24) with Illumina short reads. All isolates sequenced in this study had >95× depth and N50 > 4.6 Mb with genome sizes between 4.6 and 5.0 Mb (Megabases; million bases). All the genomes were closed.

GenBank Accessions
All WGS data on the 41 isolates is available under NCBI BioProject PRJNA929056.

Frequency of Salmonella Genomic Island 1 (SGI1) and SGI1 Variants
In our previous study [26], 0 of 530 beef feedlot pen S. enterica isolates was positive by endpoint PCR for the class 1 integron gene (intI1), suggesting that SGI1 was not involved in MDR. We again analyzed the 19 MDR isolates from that study (Table 1) and the 22 NVDC MDR isolates (Table 2) for intI1 by WGS. By WGS, consistent with the previous study [25], 0 of 9 MDR feedlot pen isolates tested were positive for intI1; however, 3 of 22 WGS NVDC isolates (13.6%) and 3 of 41 WGS isolates (7.3%) overall were positive for intI1; all 3 intI1 positive isolates were MDR (Table S6). These three isolates (RM055, RM074, and RM101), in addition to intI1, had qacE∆1 and sul1, and one (RM055) also had aadA2. Interestingly, none of these genes, which are markers of SGI1, were on the chromosome; all were on a plasmid that also carried ISVsa3 (Table S6).

Other Mobile Genetic Elements, Their Genomic Locations, and MDR Association
By WGS, all 41 isolates (100%) had the following AMR genes on the chromosome: aac(6 ), aac(6 )-Iaa, and aadA (all involved in aminoglycoside resistance); ampH (a penicillinbinding protein; PBP); bacA (involved in bacitracin resistance); and mrdA (a PBP known to confer reduced susceptibility to carbapenems) (Table S6). However, the presence of these genes was not associated with AMR for the respective antibiotic classes. Instead, resistance was associated with the following: aminoglycoside with aph(3")-Ib and aph(6)-Id; phenicol with floR; tetracycline with tet(A); sulfonamide with sul2; and β-lactam with bla CMY-2 (Table S6). A total of 25 out of 41 isolates were positive for all 5 genes: floR, tet(A), aph(6)-Id, aph(3")-Ib, and sul2 (Table 3), typically arranged in that order and flanked by ISVsa3 (Figures 1 and 2). Hence, an MDR phenotype was predominantly associated with carriage of ISVsa3 in which the genes were most often located on an IncC type plasmid that also carried bla CMY-2 (Table 4; Figure 1). Based on the combined results of endpoint PCR and WGS, 31/33 (93.9%) MDR isolates were positive for ISVsa3, whereas 6/78 non-MDR isolates (7.7%) were positive for ISVsa3 (OR = 186.00, p < 0.0001; Table 5). In addition, based on combined data from endpoint PCR and WGS, of the 111 isolates, the number (percentage) positive for the above-mentioned genes was: 38 (34.2%) for ISVsa3; 27 (24.3%) for bla CMY-2 ; 33 (29.7%) for floR; 30 (27.0%) for tet(A); 38 (34.2%) for sul2; and 31 (27.9%) for both aph(3")-Ib and aph(6)-Id (Tables S6 and S7). The serotypes in which ISVsa3 was found included S. Derby, S. Uganda, S. Typhimurium, S. Newport, S. Dublin, S. Muenster, S. Heidelberg, and S. Saintpaul (Table S6).  (6)     In column, data in the first cell refer to number of isolates positive for ISVsa3/number of isolates tested (percentage). In the remainder of the column, the data refer to the number of ISVsa3-positive isolates positive for gene/number of ISVsa3-positive isolates tested (percentage). 2 For the entire column, the data refer to number of ISVsa3-positive isolates in which the gene was located on a plasmid/number of ISVsa3-positive isolates that were positive for the gene (percentage). 3 For the entire column, data refer to number of ISVsa3-positive isolates in which the gene was located on an IncC plasmid/number of ISVsa3-positive isolates positive for the gene on a plasmid (percentage). 4 Positive WGS results for either the aph(6)-Id or aph(6)-I family genes as shown in Tables S6 and S7 were counted as positive for aph(6)-Id. 5 Isolate RM043 has 2 copies of blaCMY-2, but only 1 copy was counted in the total, as shown in Table S6.

Core Gene Alignment and Metadata
A phylogenetic tree was constructed from an alignment of core genes and selected metadata, depicting the evolutionary relationship of the isolates sequenced ( Figure 2). The core gene alignment reflected the conservation and diversion in these genomes.

ISVsa3 and Associated AMR Genetic Segment Alignment
An alignment of the genetic segment of the 36 sequenced isolates carrying ISVsa3 and associated AMR genes is shown in Figure S1.

Discussion
In this study, the first objective was to address the hypothesis that ICEMh1, ICEPmu1 or other ICEMh1-like elements occur in Salmonella isolates. We found no evidence of these BRD pathogen-associated ICE elements in our collection of 111 isolates. It is possible that one or more of these ICE elements might have been detected if more isolates were tested, and if we had tested isolates representing a more widespread area of the country, different production settings, and cattle that had been subjected to metaphylactic treatment for BRD. A number of studies have assessed the effects of metaphylactic regimens for BRD on the prevalence and selection for AMR Salmonella in field trials. A randomized controlled longitudinal study that followed cattle through the entire feeding period to harvest found that one dose of tulathromycin administered to healthy cattle at feedlot arrival did not result in an increase in the prevalence or AMR of Salmonella [41]. We hypothesize that detection of ICEMh1, ICEPmu1, or other ICEMh1-like elements in Salmonella would be more likely if one were to culture large numbers of fecal or other samples from BRD high-risk calves having respiratory colonization with ICEMh1-positive M. haemolytica [21][22][23][24], and especially following metaphylactic treatment, but this was beyond the scope of our study. Information concerning antimicrobial treatment was not available, and that concerning the signalment and clinical history was also limited.
Our second objective was to address whether SGI1 or its variants were associated with MDR, and we found that only 3 MDR isolates (7.3%) had SGI1 genes, and in these isolates, the genes were carried on a plasmid that also carried ISVsa3. One of these three isolates (RM101) carried SGI1 genes on an IncC plasmid. SGI1 cannot transfer itself into a new host because it does not carry a full set of conjugation genes, but it is mobilizable, and can be transferred if an IncC plasmid is present in the donor [19]. SGI1 only excises from the chromosome in the presence of a helper plasmid [42], and although SGI1 is known to modify and use the conjugation apparatus encoded by IncC, the two (SGI1 and IncC) are incompatible [42]. SGI1 destabilizes IncA and IncC plasmids after a few generations and, conversely, the presence of an IncC plasmid enhances the recombination rate within SGI1, leading to the generation of SGI1 deletion variants [42]. Interestingly, in our study, 20 of 29 WGS isolates (69.0%) that, collectively, had 4 or 5 AMR genes carried them on an IncC plasmid.
Our third objective was to determine the frequency of other MGE and their association with MDR, which yielded the main finding of the study: MDR was strongly associated with the presence of ISVsa3 (IS91-like Family transposase). This appears to be a novel finding, although other investigators recently reported similar genetic and phenotypic AMR profiles in 15 MDR S. Dublin isolates from retail meat and human patients [43]. In that study, nothing was stated about ISVsa3, but when we searched the associated NCBI BioProject PRJNA357723 sequence data, we found ISVsa3 (also listed as IS91-like element) in 12 of the 15 isolates. Besides our finding that ISVsa3 was strongly associated with MDR in Salmonella, our results extend these findings in that ISVsa3 was found in MDR isolates in seven other serotypes besides S. Dublin.
Previous studies have shown that ISVsa3 was first identified in the fish pathogen Vibrio salmonicida [44], and made its way into other fish pathogens, e.g., Edwardsiella piscicida [45], and further into pathogens isolated from other animals and humans, e.g., Salmonella Choleraesuis and Acinetobacter baumannii [44], carrying with it high-level resistance to antibiotics such as the tetracyclines. ISVsa3 is frequently found on conjugative plasmids [44] and poses a significant threat to spread AMR.
ISVsa3 is an IS91-like MGE, referred to as an Insertion Sequence Common Region (ISCR) [46,47]. In our study, ISVsa3 was found to be in a conserved relationship with floR, tet(A), aph(6)-Id, aph(3")-Ib, and sul2 in 72.2% (26 of 36 ISVsa3-positive) of the WGS S. enterica isolates. Due to this insertion sequence families' unique method of transposition, they are capable of and frequently responsible for mobilizing many classes of AMR genes, and are considered an evolutionary feature of IncC plasmids [47,48]. In most positive strains in our study, this ISCR was located on plasmids, and particularly IncC, while only 3 of those carrying ISVsa3 and the associated AMR genes were found on the chromosome. The strong association of ISVsa3 with MDR in the Salmonella isolates in our study and the knowledge that ISCR frequently assemble multiple AMR genes and transpose them into conjugatable plasmids suggests this particular transposon poses a significant threat to increasing MDR. Further research is needed to better understand the role of ISVsa3 in dissemination of MDR in Salmonella.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms11030631/s1, Table S1. Serotypes of Salmonella isolates from cattle in accessions submitted to the Nebraska Veterinary Diagnostic Center from 2011 to 2020; Table S2. Frequency of association of Salmonella isolation with bovine source and disease manifestation in Nebraska Veterinary Diagnostic Center accessions; Table S3. List of Salmonella isolates from Nebraska Veterinary Diagnostic Center arranged in chronological order of isolation with accompanying information concerning signalment, clinical history, sample type, serotype, and antimicrobial resistance phenotype; Table S4. Primers, probes, and products of multiplex qPCR assays; Table S5. Primer pairs and products used in endpoint PCR assays; Table S6. Antibiograms, mqPCR, endpoint PCR, and whole-genome sequencing results of Salmonella enterica isolates (n = 111); Table S7. Combined endpoint PCR and WGS results for ISVsa3, bla CMY-2 , floR, tet(A), sul2, aph(3")-Ib, and aph(6)-Id on Salmonella enterica isolates (n = 111); Figure S1. Alignment of the genetic segment of the 36 sequenced isolates carrying ISVsa3 and associated AMR genes.