Characterization of a Multiresistant Mosaic Plasmid from a Fish Farm Sediment Exiguobacterium sp. Isolate Reveals Aggregation of Functional Clinic-Associated Antibiotic Resistance Genes

Abstract

The genus Exiguobacterium can adapt readily to, and survive in, diverse environments. Our study demonstrated that Exiguobacterium sp. strain S3-2, isolated from marine sediment, is resistant to five antibiotics. The plasmid pMC1 in this strain carries seven putative resistance genes. We functionally characterized these resistance genes in Escherichia coli, and genes encoding dihydrofolate reductase and macrolide phosphotransferase were considered novel resistance genes based on their low similarities to known resistance genes. The plasmid G+C content distribution was highly heterogeneous. Only the G+C content of one block, which shared significant similarity with a plasmid from Exiguobacterium arabatum, fit well with the mean G+C content of the host. The remainder of the plasmid was composed of mobile elements with a markedly lower G+C ratio than the host. Interestingly, five mobile elements located on pMC1 showed significant similarities to sequences found in pathogens. Our data provided an example of the link between resistance genes in strains from the environment and the clinic and revealed the aggregation of antibiotic resistance genes in bacteria isolated from fish farms.

INTRODUCTION

The genus Exiguobacterium was first described in 1983, with characterization of the type species, Exiguobacterium aurantiacum, which was isolated from potato wastewater effluent (1). Since then, 14 new species have been added to the genus (2). The genus Exiguobacterium is related phylogenetically to the genus Bacillus. Exiguobacterium spp. have been isolated from diverse environments, including permafrost, hot springs, oceans, the rhizosphere of plants, and food processing plants (3). Exiguobacterium sibiricum 255-15 was isolated from a 2- to 3-million-year-old permafrost core in Russia and grew rapidly at temperatures as low as −6°C (4). Exiguobacterium strain AT1b, from a Yellowstone hot spring, grows well at 45°C (3). Isolates of Exiguobacterium can also survive in heavy metal-contaminated environments. Strain WK6 can reduce arsenate to arsenite (5). Exiguobacterium aurantiacum can reduce Cr(VI) anaerobically (6), and Exiguobacterium sp. strain TC38-2b harbors mercury resistance genes on a transposon (7).

Bacteria have mechanisms for gene exchange, and genomic and metagenomic analyses have revealed an unexpected number of horizontal gene transfer (HGT) events in bacteria (8). The most dramatic example of HGT is the dissemination of antibiotic resistance genes. Previous studies have shown that environmental bacteria are involved in the horizontal flow of resistance genes (9, 10), and a recent study provided evidence that soil has contributed to, or acquired, resistance determinants from pathogens (11). However, more examples are needed to elucidate the link between the environment and the clinic.

Plasmids often carry other mobile elements (such as transposons or insertion sequences), which capture different resistance genes and give rise to multidrug resistance. Resistance genes and associated mobile elements are often clustered on plasmids in Gram-negative bacteria, particularly the Enterobacteriaceae (12). However, our understanding of multiresistant plasmids in Gram-positive bacteria remains limited. Plasmids carrying multiple resistance genes have been identified in Staphylococcus and Enterococcus (13, 14), but examples of plasmids carrying multiple resistance genes from environmental Gram-positive bacteria are rare. One study has been performed on plasmids in Macrococcus caseolyticus harboring various antibiotic resistance determinants, including beta-lactams, aminoglycosides, and macrolides (15). The lack of characterized plasmids carrying multiple resistance genes has limited our understanding of the flow of resistance genes in Gram-positive bacteria. Therefore, there is considerable interest in characterizing more plasmids carrying resistance determinants in Gram-positive bacteria, especially in environmental samples.

In our study, Gram-positive bacteria, Exiguobacterium sp. strain S3-2 showed a multiresistance phenotype, and seven putative resistance genes were identified based on genome analysis. Five resistance genes showed significant similarities (83% to 100% similarity) with sequences found in pathogens. Two were considered novel resistance genes based on their low similarities to known resistance genes and functional experiments. These seven resistance genes were located on mosaic plasmid pMC1. Aggregation of resistance genes on plasmid is most likely due to the use of antibiotics in fish farms.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

A morphologically diverse collection of bacteria was isolated from sediments of a fish farm located in Xiangshan Harbor, Zhejiang Province, China, in November 2009. The isolates were screened on selection plates with trimethoprim-sulfamethoxazole (16 mg/liter), chloramphenicol (16 mg/liter), tetracycline (8 mg/liter), streptomycin (16 mg/liter), erythromycin (4 mg/liter), ciprofloxacin (0.25 mg/liter), and penicillin (0.25 mg/liter) separately. Strain S3-2 was selected for further study. The strain was cultured on seawater plates supplemented with 0.5% (wt/vol) fish peptone and 0.1% (wt/vol) yeast extract at 30°C. Escherichia coli K-12 JW0451-2 (16), a derivative of E. coli K-12, and DH5α were grown on Luria-Bertani agar plates at 37°C. The MICs for strains were determined using broth microdilution assays, as recommended by CLSI (Clinical and Laboratory Standards Institute [http://www.clsi.org/]) (17). Primers 27F and 1492R (18) were used for bacterial 16S rRNA gene amplification and sequencing for phylogenetic analysis of isolate S3-2.

Genome sequencing and annotation.

The genome of Exiguobacterium sp. strain S3-2 was sequenced using Illumina HiSeq 2000. Genomic DNA was used to construct the sequencing libraries according to the manufacturer's recommended protocols (Illumina), and paired-end shotgun sequencing was performed. After the shotgun stage, 1.85 gigabytes of raw data was obtained. Low-quality reads and adapters were filtered using Trim Galore software. Reads with sequencing quality scores greater than 20 and lengths greater than 80 bp were used for further analysis. The Velvet program was used to assemble genomes with an optimized k-mer of 81. The Glimmer software was used to predict open reading frames (ORFs) of the assembled genome with default parameters. The annotation of ORFs was based on BLAST searches against a number of bioinformatic databases, including Uniprot (http://www.uniprot.org/), KEGG (http://www.genome.jp/kegg/genes.html), Swissprot (http://www.expasy.ch/sprot/), and Refseq (http://www.ncbi.nlm.nih.gov/refseq/). In addition, the InterProScan software (http://www.ebi.ac.uk/Tools/InterProScan/) was used to search its domain architectures.

Genome analysis.

The contigs assembled by Velvet were subjected to a BLAST search against the nonredundant nucleotide database of NCBI (ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/) using BLASTn software (E-value cutoff of 10⁻¹⁰). Contigs greater than 50 kb were aligned with three completely sequenced Exiguobacterium spp. using VISTA (http://genome.lbl.gov/vista/index.shtml). Predicted plasmids were subjected to a BLAST search against the NCBI Plasmid Genome Database (ftp://ftp.ncbi.nlm.nih.gov/refseq/release/plasmid/) and identified by PCR amplification. Cumulative GC skew analysis data were retrieved using Lasergene software (19). In addition, BLASTP analysis with cutoff values of <10⁻⁵ (E value) and 30% similarity for orthologues against the Antibiotic Resistance Genes Database (ARDB) (http://ardb.cbcb.umd.edu/index.html) was performed.

Construction of recombinant plasmids carrying putative resistance genes.

To identify the functions of ORFs that shared similarities with reported resistance genes, the PCR product of each ORF was cloned into the multiple cloning sites of the pUC19 vector. The designed primers and the insertion sites of pUC19 are listed in Table S1 in the supplemental material. E. coli DH5α and K-12 JW0451-2 were transformed with these recombinant plasmids, and clones were selected by blue-white screening with Luria-Bertani agar (Oxoid, Wesel, Germany) plates supplemented with 100 mg/liter ampicillin. Transformants were confirmed to contain inserted fragments by PCR. Resistance of the transformants to trimethoprim, erythromycin, chloramphenicol, and streptomycin were determined by the microdilution method.

Nucleotide sequence accession numbers.

The complete nucleotide sequences of 5 plasmids have been submitted to GenBank under accession numbers KF648874 (pMC1), KF648875 (pMC2), KF697250 (pMC3), KF697251 (pMC4), and KF697252 (pMC5).

RESULTS

Characterization of multiresistant Exiguobacterium sp. strain S3-2.

We isolated a morphologically diverse collection of bacteria from sediments of a fish farm located in Xiangshan Harbor, Zhejiang Province, China. A library of 181 isolates was screened against seven antibiotics. One isolate was resistant to five antibiotics, trimethoprim-sulfamethoxazole (MIC > 1,024 mg/liter), chloramphenicol (MIC = 32 mg/liter), tetracycline (MIC = 16 mg/liter), streptomycin (MIC = 128 mg/liter), and erythromycin (MIC = 8 mg/liter), and was susceptible to ciprofloxacin and penicillin. Amplification and sequencing of 16S rRNA gene from the isolate indicated that it shared 99.8% similarity with Exiguobacterium profundum strain 10C (AY818050.1) and 99.7% similarity with Exiguobacterium aestuarii strain TF-16 (AY594264.1) and the genome-sequenced Exiguobacterium sp. strain AT1b. We designated this isolate Exiguobacterium sp. strain S3-2.

Analysis of the Exiguobacterium sp. strain S3-2 genome.

The draft genome of Exiguobacterium sp. strain S3-2 was 2.8 Mb in length with an average G+C content of 48 mol%. A total of 2,880 protein-coding genes were predicted based on Uniprot, KEGG, Swissprot, and Refseq. Comparisons with other genomes of Exiguobacterium genera in the NCBI database identified ∼2,000 predicted proteins with similarities greater than 80% to proteins in Exiguobacterium strain AT1b. Exiguobacterium strain AT1b is a thermophilic, facultative anaerobic bacterium isolated from slightly alkaline, highly carbonate hot spring water samples. It contained 3,020 predicted proteins with a G+C content of 48.5 mol% (20). Only ∼200 proteins shared high similarities (>80%) with those of isolates of Exiguobacterium sibiricum and Exiguobacterium antarcticum.

Identification of plasmids in Exiguobacterium sp. strain S3-2.

Potential complete plasmids were identified from the assembled circular contigs. PCR was then used to identify these contigs as true complete circular plasmids, which were designated pMC1, pMC2, pMC3, pMC4, and pMC5 (Table 1). Plasmid pMC1 was 71,276 bp in size and contained 66 predicted ORFs (see Table S2 in the supplemental material). The average G+C content was 41.75%, which was lower than that of the chromosome (48%). However, the G+C content distribution was highly heterogeneous along the plasmid DNA, varying from 33.2% to 50.9%. The plasmid did not carry reported Rep initiator genes and could replicate using Rep proteins from other plasmids. Two ORFs corresponded to proteins predicted to be involved in bacterial conjugation. Seven ORFs shared various similarities with reported resistance genes, and three ORFs were homologs of known transposases.

TABLE 1.

Plasmids harbored by Exiguobacterium sp. strain S3-2

Characteristic	pMC1	pMC2	pMC3	pMC4	pMC5
Size (bp)	71,276	19,981	4,445	1,813	1,742
GC content (%)	41.75	41.24	42.14	42.25	42.77
No. of ORFs	66	22	1	2	2
Predicted replication genes	0	1	0	1	1
Predicted transposase genes	3	1	0	0	0
Predicted resistance genes	7	1	0	0	0

pMC2 was 19,981 bp in size and contained 22 predicted ORFs (see Table S3 in the supplemental material). The gene encoding the putative plasmid replication protein RepB, which belongs to the CbiA (PF01656) domain family, was detected in the plasmid. Significant similarities (more than 99% similarity over 20% of the pMC2 genome) were found, mainly to plasmid pMCCL2 in Macrococcus caseolyticus JCSC5402 (NC_011996.1). The region contained four ORFs, encoding a putative resolvase, dihydrofolate reductase, thymidylate synthase, and transposase gene. Plasmid pMCCL2 harbored a mecA gene homolog, encoding a penicillin-binding protein similar to PBP2′ of methicillin-resistant Staphylococcus aureus (15). pMC3 was 4,445 bp in size, encompassing a gene for a putative recombinase involved in plasmid mobilization. pMC4 and pMC5 were less than 2 kb in size and contained genes for replication and hypothetical proteins. Both Rep proteins belonged to the Rep_trans (PF02486) domain family. The G+C contents of the four plasmids were ∼42%, which is similar to that of the large plasmid pMC1.

Resistance genes with high similarities to pathogens are present in Exiguobacterium sp. strain S3-2.

Five ORFs with high similarities to antibiotic resistance genes from pathogens were identified in plasmid pMC1 (Table 2). Macrolide phosphotransferase (MphB) shared 83% similarity with proteins found in E. coli plasmids and the Streptococcus uberis genome (21, 22). A tetracycline efflux pump (TetL) has more than 95% similarity with the corresponding protein sequences in many Gram-positive pathogens. The sequence of streptomycin aminoglycoside 6-adenyltransferase (AadE) was 100% identical to protein sequences from isolates of Streptococcus suis, S. aureus, Enterococcus faecium, and Enterococcus faecalis. The macrolide-resistant mefA-encoded efflux pump was more than 95% similar to those of isolates of Streptococcus, Staphylococcus, and Fusobacterium nucleatum. Florfenicol-chloramphenicol transporter (FexA) shared 99% similarity with those of isolates of Staphylococcus and E. faecium. Interestingly, we found that the five resistance genes were more prevalent in isolates of bacilli than other taxa.

TABLE 2.

Antibiotic resistance genes in Exiguobacterium sp. strain S3-2 with high similarities to known pathogens

ORF of pMC1	Antibiotic class	Annotation (mechanism)	Pathogen hit (GI no.)	% protein similarity
ORF_27	Macrolide	MphB (macrolide 2′-phosphotransferase)	E. coli (YP_006162251.1)	83
			S. uberis (ACE78320.1)	83
ORF_31	Tetracycline	TetL (efflux pump)	S. suis (YP_003028728.1)	97
			E. faecium (AEO13632.1)	96
			E. faecalis (AAB09024.1)	97
			S. aureus (YP_003084335.1)	97
			B. cereus (NP_043524.1)	97
			S. agalactiae (NP_040422.1)	97
ORF_59	Aminoglycoside	AadE (aminoglycoside 6-adenyltransferase)	S. suis (YP_001198323.1)	100
			E. faecium (WP_002294505.1)	100
			E. faecalis (AAL05549.1)	100
			S. aureus (AFM38043.1)	100
			S. parauberis (WP_003109014.1)	99
ORF_63	Macrolide	MefA (macrolide-efflux protein)	S. pneumoniae (CAI94891.2)	97
			S. suis (YP_006075994.1)	96
			F. nucleatum (ABI18324.1)	96
			S. pyogenes (ACG58974.1)	98
ORF_66	Chloramphenicol	FexA (florfenicol-chloramphenicol exporter)	S. simulans (CAJ31068.1)	99
			S. warneri (CAL64024.1)	99
			E. faecium (WP_002360219.1)	99
			S. cohnii (AEP69228.1)	99

We cloned the five resistance genes into the pUC19 vector, separately. Recombinant plasmids with fexA, tetL, and aadE were introduced into E. coli DH5α. E. coli carrying recombinant plasmids had increases in resistance to chloramphenicol of 4-fold, to tetracycline of 8-fold, and to streptomycin of more than 216-fold compared to E. coli with pUC19 alone (Table 3). Recombinant plasmids with mphB and mefA were expressed in E. coli K-12 JW0451-2, which is susceptible to erythromycin due to inactivation of the acrA pump. Transformants showed 32- and 8-fold, respectively, increased resistance to erythromycin compared to the control (Table 3). This result demonstrated that the five resistance genes can be functionally expressed in E. coli and that they may contribute to resistance in Gram-negative pathogens.

TABLE 3.

Functional analysis of putative resistance genes in pMC1 compared to pUC19 of E. coli

Recombinant plasmid	Fold increase over pUC19 expression	Antibiotic resistance
pUC19::fexA	4	Chloramphenicol
pUC19::aadE	>216	Streptomycin
pUC19::tetL	8	Tetracycline
pUC19::mphB	64	Erythromycin
pUC19::mefA+mel	16	Erythromycin
pUC19::mph_like	64	Erythromycin
pUC19::dfr_like	>2,046	Trimethoprim

Identification of novel resistance genes in Exiguobacterium sp. strain S3-2.

The putative genes responsible for the resistance phenotypes were identified on plasmids. One of the ORFs located on pMC1 and pMC2 was 40% to 57% similar at the amino level to known dihydrofolate reductase genes and 100% identical to the gene encoding hypothetical proteins in the plasmid of Macrococcus caseolyticus JCSC5402 (15). Dihydrofolate reductase genes (dfr genes), which confer resistance to trimethoprim, have been identified in Gram-positive and -negative bacteria. We cloned the ORF into E. coli and found that E. coli carrying recombinant plasmids had resistance more than 2,046-fold that of the control (Table 3). Phylogenetic analysis at the amino acid level demonstrated that the ORF in Exiguobacterium sp. strain S3-2 belonged to the DfrA family and clustered with DfrA proteins found in E. coli, Vibrio cholerae, and Salmonella enterica (Fig. 1). This result suggested that the gene in Exiguobacterium sp. strain S3-2 may encode a novel dihydrofolate reductase. Until now, over 30 types of dfr genes have been identified and grouped into the families dfrA and dfrB. Proteins in the DfrA family (150 to 200 amino acids) are longer than those in the DfrB family (78 amino acids) (23).

FIG 1.

Neighbor-joining tree of aligned amino acid sequences of the dihydrofolate reductase constructed in this study. The alignment was performed with DfrA family proteins and one DfrB family protein (DrfB1, CAI56200) that was used as an outgroup. Genera, species, and strains are followed by their GenBank accession numbers. Bootstrap values were calculated as a percentage of 1,000 replicates, and those above 70% are shown at the branching points. The sequence obtained in this study is depicted in bold.

One putative macrolide 2′-phosphotransferase was detected in pMC1. The protein encoded by the ORF was 45% and 42% similar to MphB and MphC, respectively. The best match (99% similarity) to the protein was found in the plasmid pEspB of Exiguobacterium arabatum RFL1109 (24). Phylogenetic analysis showed that the protein was closely related to family C macrolide phosphotransferase (MphC) from Staphylococcus (Fig. 2). The gene functionally expressed in E. coli conferred 32-fold-increased resistance against erythromycin (Table 3). Based on the phenotype and phylogenetic relationship between the hypothetical protein and reported macrolide phosphotransferase, we predicted that the ORF encoded a novel plasmid-borne macrolide-resistant macrolide phosphotransferase.

FIG 2.

Neighbor-joining tree of aligned amino acid sequences of the macrolide 2-phosphotransferase constructed in this study. The alignment was performed with MphB and MphC proteins and one MphA protein (MphA, YP_002415417) that was used as an outgroup. Genera, species, and strains are followed by their GenBank accession numbers. Bootstrap values were calculated as a percentage of 1,000 replicates, and those above 70% are shown at the branching points. The sequences obtained in this study are depicted in bold.

Identification of mobile elements correlated with Tn1546.

Three genes encoding thymidylate synthase, putative dihydrofolate reductase, and resolvase were identified on both pMC1 and pMC2. In plasmid pMC2, the transposase-encoding gene was adjacent to the three genes, and these four genes were flanked by 38-bp imperfect inverted repeats (IRs). A resolution site was found between tnpA and tnpR, which faced in opposite directions (Fig. 3). Structural analysis indicated that the mobile element was a transposon belonging to the Tn3-like subgroup (25). A similar transposon has also been found in plasmid pMCCL2 from Macrococcus caseolyticus which contains an extra insertion between tnpA and tnpR (15). Interestingly, the transposase and resolvase shared 94% and 91% similarity with Tn1546 proteins, respectively. Tn1546 carries a vancomycin resistance cluster that is responsible for the majority of spread of vancomycin resistance among enterococci (26). The imperfect IRs showed more than 95% similarity with IRs of Tn1546. The shared backbone of the transposon has been found in the chromosome of Bacillus amyloliquefaciens LL3 and the plasmid of Bacillus thuringiensis BMB (Fig. 3). In Bacillus amyloliquefaciens LL3, an 8,847-bp sequence encoding hypothetical proteins and proteins involved in sporulation was inserted into the putative backbone, and the transposon was duplicated in the chromosome in opposite directions. In the Bacillus thuringiensis plasmid BMB, the transposon carried genes encoding hypothetical and spore germination proteins.

FIG 3.

Mobile elements correlated with Tn1546. This diagram shows structure of mobile elements with transposase, resolvase, and repeat sequences highly similar to those of Tn1546. ORFs are shown as arrows, and similar genes are depicted in gray. Tn1546 is from E. coli (GenBank number M97297.1). Other GenBank numbers are as follows: B. amyloliquefaciens LL3 complete genome, CP002634.1; B. thuringiensis BMB, CP001904.1; pMCCL2 from M. caseolyticus, AP009486.1; pMC2 (this study), KF648875.

Mosaic mobile genetic elements embedded in the plasmids of Exiguobacterium sp. strain S3-2.

Using BLASTN, several blocks with high similarities to other mobile elements were identified (Fig. 4). A total of 28% of the pMC1 genome showed significant similarities (92% to 94% similarity) to 38.9-kb plasmid pEspB from Exiguobacterium arabatum (24). According to analysis of plasmid pEspB, the majority of the block was a putative transfer region, and included conjugal proteins. The gene encoding a putative macrolide 2′-phosphotransferase was adjacent to the transfer region. The G+C content of the region was 50.9%, similar to the mean G+C content of the host. The G+C content of the remainder of the plasmid was ∼35%, much lower than that of the host. These results suggested that the region homologous to pEspB could serve as a backbone and that other regions could be acquired by horizontal gene transfer from other bacteria.

FIG 4.

Genetic organization of pMC1 (A) and pMC2 (B). This diagram shows the circular genomes of pMC1 and pMC2. ORFs are shown as arrows. The deviation of the G+C content from the mean, calculated with a window of 500 nt, is indicated in the inside track (yellow, G+C > mean; purple, G+C content < mean).

The first gene within the low-G+C region was ORF27, which encodes macrolide 2′-phosphotransferase with 83% similarity to MphB in the E. coli plasmid and Streptococcus uberis genome (Fig. 4A). The adjacent block included ORF28 to ORF30, which encoded an ATP-binding protein, transposase, and resolvase, showing 89% similarity to plasmids pH308197, pPER272 and pE33L466 from various Bacillus cereus isolates (see Table S2 in the supplemental material). Downstream, tetL resistance genes were found. The next block (ORF32 to ORF34) was 100% identical to the mobile element identified in pMC2 (Fig. 4B), but was not a complete transposon and lacked TnA and IR. Downstream, the block encompassing 23 putative ORFs did not match any sequence in the NCBI database. The adjacent block included ORF59 to ORF61, which encode streptomycin aminoglycoside-6-adenyltransferase AadE and two hypothetical proteins. The 2.3-kb block showed 98% similarity to the sequences from S. suis SC84 and S. aureus and 96% similarity to the E. faecalis pEF418 plasmid. Next, the 3.1-kb block included mefA and mel, conferring macrolide resistance, and aligned with the macrolide efflux genetic assembly (mega) element with 58% coverage. The mega element, which was 5.5 kb in size, was inserted at several chromosomal sites in Streptococcus (27). The presence of a mega element is the most prevalent mechanism of Streptococcus pneumoniae macrolide resistance in the United States, Canada, and the United Kingdom (28). The last block, with a length of 7 kb, showed 99% similarity with Tn558 and included the florfenicol-chloramphenicol transporter FexA, transposase, and a putative oxidoreductase. Tn558 is prevalent in Staphylococcus and has been identified in Bacillus isolates (29).

DISCUSSION

The genus Exiguobacterium can adapt readily to, and survive in, diverse ecosystems and is prevalent in marine environments (3). The ocean is a natural habitat for antibiotic-producing bacteria (30), and marine aquaculture introduces antibiotics into the ocean to treat infections and improve aquaculture production. We characterized seven plasmid-borne antibiotic resistance genes in an Exiguobacterium sp. strain S3-2 isolated from marine fish farms, two of which were identified as novel resistance genes.

The seven resistance genes are located on plasmid pMC1. According to the mean G+C content, plasmid pMC1 was divided into two regions. The 19.8-kb region shared significant similarity with plasmid pEspB from Exiguobacterium arabatum. It had a 50.9% G+C content, which fit well with the mean G+C content of the host. The remainder of the plasmid was composed of mobile elements with markedly lower G+C ratios than the host. Except for a novel mph gene, the six other resistance genes were located on six mobile elements in a lower-G+C region. Interestingly, five mobile elements in this region showed significant similarities (83% to 100% similarity) to nucleotide acid sequences found in pathogens, including S. aureus, S. pneumoniae, and E. faecalis; which like Exiguobacterium belong to the class Bacilli. For instance, Tn558 carrying fexA was found in found in three genera of Bacillales and two genera of Lactobacillales. Resistance gene flow can often overcome the barrier of taxonomy (31). However, HGT occurs more frequently between close relatives (32). These results demonstrate that the active flow of mobile elements can occur in isolates of Bacilli. They enhance understanding of the horizontal flow of resistance genes and provide insight into the shared resistome of environmental bacteria and human pathogens.

Multiresistance plasmids have been investigated extensively in Gram-negative bacteria, especially Enterobacteriaceae. For example, multiresistant E. coli and Klebsiella pneumoniae strains isolated in India, which were highly resistant to all tested antibiotics with the exception of tigecycline and colistin (33), carried plasmids containing dozens of resistance genes, including blaNDM-1. However, multiresistance plasmids do not appear to spread in Gram-positive bacteria as readily as in the Enterobacteriaceae. Plasmids carrying multiple resistance genes have been identified in Staphylococcus and Enterococcus (13, 14), but examples of multiresistant plasmids from environmental Gram-positive bacteria are rare.

Antibiotics are used in marine aquaculture to treat infections and improve aquaculture production. Oxytetracycline, florfenicol, sarafloxacin, erythromycin, and Tribrissen, which is a combination of sulfonamide and trimethoprim, are authorized antibiotics for use in aquaculture (34). Tetracycline resistance genes have been identified in fish farms throughout the world (35). Five different trimethoprim resistance genes, dfrA1, dfrA5, dfrA7, dfrA12, and dfrA15, were prevalent in fish farms in Pakistan and Tanzania (36), and dfr16 (dfrA16) and dfrIIc (dfrB3) were detected in Aeromonas salmonicida from fish farms in Norway and Scotland, respectively (37). The mosaic plasmid pMC1 carried both tetracycline and trimethoprim-resistance genes; as well as fexA, conferring florfenicol resistance and mph and mefA, conferring erythromycin resistance.

Although antibiotic resistance genes are native to marine sediment microbiomes, we hypothesize that the aggregation of resistance genes on plasmid pMC1 is most likely due to selective pressure propagated by exposure of Exiguobacterium sp. strain S3-2 to antibiotics. The use of antibiotics in aquaculture may accelerate the horizontal transfer of resistance genes and influence the composition of the local resistome.