Epidemiology and molecular characterization of the antimicrobial resistance of Pseudomonas aeruginosa in Chinese mink infected by hemorrhagic pneumonia

Abstract

Hemorrhagic pneumonia in mink is a fatal disease caused by Pseudomonas aeruginosa. Very little is known about P. aeruginosa in relation to genotype and the mechanisms underlying antimicrobial resistance in mink. A total of 110 P. aeruginosa samples were collected from mink from Chinese mink farms between 2007 and 2015. Samples underwent molecular genotyping using pulsed-field gel electrophoresis (PFGE) and multi-locus sequence typing (MLST), antimicrobial susceptibility and its mechanism were investigated at the molecular level. The PFGE identified 73 unique types and 15 clusters, while MLST identified 43 (7 new) sequence types (ST) and 12 sequence type clonal complexes (STCC). Sequence types and PFGE showed persistence of endemic clones in cities Wendeng (Shandong, China) and Dalian (Liaoning, China), even in different timelines. The MLST also revealed the gene correlation of the mink P. aeruginosa across different time and place. The ST1058 (n = 14), ST882 (n = 11), and ST2442 (n = 10) were the predominant types, among which ST1058 was the only one found both in Shandong province and Dalian (Liaoning, China). The MLST for P. aeruginosa infection in mink was highly associated with that in humans and other animals, implying possible transmission events. A small proportion of mink exhibited drug resistance to P. aeruginosa (9/69, 13%) with resistance predominantly to fluoroquinolone, aminoglycoside, and β-lactamase. Eight strains had mutations in the quinolone-resistance determining regions (QRDR). High proportions (65%; 72/110) of the fosA gene and 2 types of glpt deletion for fosmycin were detected. Furthermore, in the whole genome sequence of one multidrug resistant strain, we identified 27 genes that conferred resistance to 14 types of drugs.

Introduction

Pseudomonas aeruginosa (P. aeruginosa) is an invasive bacterium associated with a high fatality rate. This is not only a major nosocomial agent that is deadly to patients with cystic fibrosis (CF) (1,2), but is also a cause of fatal disease in mink affected with hemorrhagic pneumonia worldwide (3–6). Pseudomonas aeruginosa exerts serious economic effects on the mink farming industry, similar to those created by mink distemper, mink virus enteritis, and Aleutian disease (7). Pseudomonas aeruginosa was first reported in China in 1985 and there have been outbreaks every year since (3,8,9). Mink of all ages are affected, particularly during the warm humid days of August to early December. Typical symptoms include bleeding from the nose or mouth, pulmonary hemorrhage, and pleural effusions. The rate of mortality varies between 1% and 50% (4,5).

Mink farms are mainly spread across a dozen regions of 5 provinces on the eastern coastal area and northeast China. Most mink are reared by family-based farming with some middle- and small-scale breeding professionals; thus, these animals are situated near human habitats. When large and frequent outbreaks of P. aeruginosa occur, it is necessary to carry out epidemiological investigations. Multi-locus sequence typing (MLST) and pulsed-field gel electrophoresis (PFGE) are methods with which to characterize bacteria at the molecular level. The PFGE is more discriminatory than MLST in terms of determining genetic differences, while MLST is better for detecting population structure and genetic evolutionary relationships (10,11). Earlier studies confirmed that the outbreaks of hemorrhagic pneumonia in mink showed clonal diversity and were randomly isolated from the environment. However, while a close PFGE pattern was found in one particular region (4,8,12), relatively little is known concerning the association between PFGE and MLST data and the molecular epidemiology of P. aeruginosa infection in mink. In cases of nosocomial infection, MLST analysis can be used to investigate individual-to-individual P. aeruginosa transmission and to identify other high-risk clones that harbor drug resistant genes (11,13). Indeed, MLST has been used to investigate P. aeruginosa in dogs, horses, and cows and internationally widespread clones associated with human outbreaks, and sometimes, multi-drug resistance phenotypes have been identified (14).

Pseudomonas aeruginosa is known to have high rates of multidrug-resistance (MDR) even pan-drug-resistance (PDR), predominantly to β-lactamase, aminoglycoside, and fluoroquinolone isolated from patients hospitalized with CF and ocular infections worldwide (15,16). The plasmid-mediated quinolone resistance (PMQR) gene and the metallo-β-lactamase (MBL) genes bla_IMP, bla_VIM, and bla _DIM-1NDM are often detected in nosocomial infection. However, the mechanism of P. aeruginosa resistance in mink remains unknown (8,14).

Since MLST has not been carried out for P. aeruginosa in Chinese populations of mink and the antimicrobial resistance mechanisms involved have yet to be elucidated, we collected 110 isolates from mink farms from a large geographical region of China between 2007 and 2015. Our objective was to obtain MLST and PFGE data relating to the molecular epidemiology of P. aeruginosa in China and the structure of the P. aeruginosa genotype within the mink population. In addition, resistance genes were detected and an MDR strain was identified from whole genome sequence (WGS) data; this information is important as it can be used to investigate environmental threat and the potential for this disease to transfer to humans. To our knowledge, this is the first report to describe the use of MLST for investigating the molecular epidemiology and antimicrobial resistance of P. aeruginosa in mink.

Materials and methods

Identification and serological typing of P. aeruginosa

A total of 110 P. aeruginosa samples were included in this study, within which, 102 strains were isolated from mink that died from hemorrhagic pneumonia between 2007 and 2015. This sample cohort represented 78 outbreaks on 56 mink farms in 5 provinces, with 7 strains from soil collected from a mink farm in Dalian (Liaoning, China) in 2013, and 1 isolate from soil collected from a mink farm in Panjin (Liaoning, China) (Figure 1).

Figure 1.

Geographical distribution of Pseudomonas aeruginosa strains from 5 provinces in northeastern China. The letters in the figure represent the city within the province as follows: a — Harbin; b — Shenyang; c — Jinzhou; d — Dalian; e — Wendeng; f — Weihai; g — Weifang, h — Linyi.

Identification of P. aeruginosa was based on characteristic colony morphology when grown on nalidixic acid cetrimide medium (NAC agar; Hangzhou Binhe Microorganism Reagent Company, Hangzhou), Gram-negative staining, a positive cytochrome oxidase reaction, and a positive catalase reaction. The 16S RNA were amplified in accordance with a method described previously (17) and sequences were evaluated on NCBI by BLAST. Serotyping was done using the slide agglutination method and commercially available polyvalent I, II, and III group specific antisera against 14-O antigens (DenkaSeiken Company, Osaka Japan).

Pulsed-field gel electrophoresis

The PFGE procedure has been described elsewhere (18) and was modified for use with P. aeruginosa isolates. Slices (1-mm thick) of DNA embedded in 1% Seakem gold agarose (Lonza, Basel, Switzerland) were digested using SpeI (Takara Biotechnology Company, Dalian, Liaoning, China) at 37°C for 3 h. We used XbaI-restricted DNA of Salmonella enterica serovar Braenderup H9812 as a molecular size marker (Suzhou BeNa Biological Technology Company, Suzhou, Jiangsu, China). After restriction digestion, fragments were resolved in 1% agarose with 0.5 Tris-boric acid (TBE) buffer at 6 V/cm at 14°C using a CHEF Mapper XA pulsed-field electrophoresis system (Bio-Rad Laboratories, California, USA). Pulse times were 4 to 40 s for 18 h. Gels were stained for 30 min with Gelred (Biotium, Fremont, USA), de-stained in distilled water for 20 min and photographed in ultraviolet (UV)-light (Bio-Rad Laboratories). The resulting band profiles were analyzed using bionumerics computer software (Version 7.60; Applied Maths, Sint-Martens-Latem, Belgium) and were characterized using phylogenetic unweighted pair group method with arithmetic mean (UPGMA) dendrogram dice-band comparison and a position tolerance of 1.7%. Isolates were defined as belonging to the same strain if they had indistinguishable PFGE profiles. If the isolates differed by 1 to 5 bands, corresponding to similarities > 80%, they were regarded as belonging to a cluster of closely related strains.

Multi-locus sequence typing

The MLST was done according to previously published protocols (19). Briefly, genomic DNA was extracted from purified strains with a DNeasy kit (Qiagen, Beijing, China). Standard DNA amplification and sequencing of 7 housekeeping genes (acsA, aroE, guaA, mutL, nuoD, ppsA, and trpE) was done for all isolates. The nucleotide sequences of both strands were determined by using previously published primers and were compared to existing sequences in the MLST database (www.pubmlst.org/p.aeruginosa) for assignment of allelic numbers. The isolates were assigned a sequence type (ST) number according to their allelic profiles. Phylogenetic analysis was carried out using bionumerics software (Applied Maths, Version 7.40). Isolates that were identical at 5 or more alleles were considered to be part of an ST clonal complex (STCC).

Antimicrobial susceptibility testing

The minimum inhibitory concentration (MIC) of 69 non-repeated strains were detected by 9 drugs including polymycins (polymyxin B, polymyxin E), aminoglycosides (gentamicin, amikacin), cephalosporins (ceftazidime), fluoroquinolones (levofloxacin, ciprofloxacin, enrofloxacin), fosfomycin, spectinomycin, and florflenicol (Sigma-Aldrich Corporation, St. Louis, USA). Isolates were considered to be multi-drug resistant (MDR) if they were resistant to at least 3 of the antibiotic classes tested (cephalosporins, fluoroquinolones, aminoglycosides, polymycins) according to criteria published by the European Centre for Disease Prevention and Control (ECDC) (20). The MIC were determined by microdilution in cation-adjusted Mueller-Hinton broth, followed by the agar dilution method (fos-fomycin), in accordance with documents M100-S23 (2013) from the Clinical and Laboratory Standards Institute (CLSI, USA). Escherichia coli ATCC 25922 and P. aeruginosa ATCC 27853 were used as quality control strains.

Characterization of fosA and glpt

FosA was amplified by polymerase chain reaction (PCR) with primers 5′-GGCCGTGGCGTCCGTAGCTT-3′ and 5′-GCTTGCGCCAACTGATGGAAATG-3′ to produce a 532 bp product; PCR conditions were 95°C for 1 min, 63°C for 45 s, and 72°C for 45 s for 34 cycles. The PCR was also used to amplify glpt (primers: 5′-ATTTCCGACCAATGCACTCAACCTG-3′ and 5′-AAGCAGGGATGGACGCGCACCAGAT-3′), using the same PCR cycle conditions as for fosA. In order to identify mutations, the PCR products for glpt and fosA were compared with the known wild-type sequence (GenBank no: AE004091.2).

Amplification of metallo-β-lactamase (MBL) and plasmid-mediated quinolone resistance genes (PMQR)

The MBL genes (bla_IMP, bla_VIM, bla_SPM, bla_GIM, bla_SIM) and PMQR genes (qnrA, qnrB, qnrC, qnrD, qnrS, qepA, and aac(6′)-Ib) were amplified using established primers (33) and compared with the P. aeruginosa PA01 sequence (GenBank Accession Number: AE004091).

Alterations within quinolone-resistance determining regions (QRDR)

The target regions for mutations (gyrA, gyrB, ParC, and ParE) were amplified using established primers (21) and compared with the P. aeruginosa PA01 sequence (GenBank Accession Number: AE004091).

Sequencing of PCR products

The PCR products were sequenced (Sangon Biotech Company, Shanghai, China) and alignment was carried out using computer software (Lasergene MegaAlign, Version 10; DNASTAR, Madison, USA).

Whole genome sequencing for P. aeruginosa MDR PA59

Whole genome sequencing was done (Beijing Novogene Bioinformatics Technology Company, Beijing, China) and genomic DNA was subjected to cell sequencing (PacBio SMRT; Pacific Biosciences, California, USA). Low quality reads were filtered (SMRT 2.3.0) and filtered reads were then assembled to generate one contig without gaps. This contig was verified as being circular using computer software (Version 1.3.1; Gepard, London, UK). The circular contig was then reassembled to polish the assembly. The mean sequencing coverage was 118.8-fold with an accuracy > 99.99%. A whole genome Blast search (E-value < 1e-5; minimal 2 alignment length percentage > 40%) was done against the following databases: Gene Ontology (GO); Kyoto Encyclopedia of Genes and Genomes (KEGG); Clusters of Orthologous Groups (COG); Non-Redundant Protein Database databases (NR); Transporter Classification Database (TCDB); and Swiss-Prot, TrEMBL, and Antibiotic Resistance Genes Database (ARDB).

Results

O-antigen serotyping of isolates

Of the 102 isolates from mink hemorrhagic pneumonia, there were 86 cases classified as ‘G’ serotype, 7 cases as ‘B’ serotype, 3 cases as ‘I’ serotype; there were also 6 cases which could not be serotyped. Of the 8 isolates acquired from soil samples, 4 samples were classified as ‘G’ serotype, 1 sample as ‘A’, 1 sample as ‘D’, and 1 sample as ‘E’; only 1 sample could not be classified.

Pulsed-field gel electrophoresis

The PFGE profiles were classified into 73 distinct PFGE patterns and 15 PFGE clone clusters (Figure 2). The phylogenetic tree showed that clusters 1, 2, 3 in Dalian (Liaoning, China), which had 90% similarity, were located on the same branch of the phylogenetic tree, cluster 12 which was also from Dalian (Liaoning, China) had 85% similarity. Clusters 4, 5, 6, 8, and 14 were all from Wendeng (Shandong, China) from different farms, showed 85% to 90% similarity. Clusters 10, 11, 13, and 15 were all from one outbreak in one farm from provinces of Shandong and Jilin, and city of Dalian (Liaoning, China). Data indicated that the PFGE profiles are diverse, while they appeared to be clustered to Dalian (Liaoning, China) and Wendeng (Shandong, China).

Figure 2.

Pulsed-field gel electrophoresis (PFGE) phylogenetic tree analysis for the 110 Pseudomonas aeruginosa strains obtained from mink using bionumerics computer software (Applied Maths, Version 7.6). The PFGE clusters 1 to 15 are shown beside the vertical line.

Multi-locus sequence typing

The MLST revealed 43 different ST from the 110 P. aeruginosa strains, among which were 7 new ST (ST2442, 2443, 2444, 2445, 2446, 2447, and 2501) and one new allele (aroE 217). The dominant types were ST1058 (strain number: 14, 4 outbreaks), ST882 (strain number: 11, 11 outbreaks), ST2442 (strain number: 10, 2 outbreaks), and ST266 (strain number: 8, 1 outbreak). Overall, 21 ST could be clustered into 12 different STCC, denoted as UM01 (ST471, ST934, ST167, ST2048, ST1058), UM02 (ST1661, ST244), UM03 (2443, ST882), UM04 (ST2444, ST1425), UM05 (ST2447, ST1207), UM06 (ST788, ST1748), UM07 (ST155), UM08 (ST399), UM09 (ST266), UM10 (ST1393), UM11 (ST189), and UM12 (ST2442). The remaining 22 ST were classified as singleton ST (Figures 3, 4).

Figure 3.

Multi-locus sequence typing (MLST) minimum-spanning tree for the 110 Pseudomonas aeruginosa strains obtained from mink using bionumerics computer software (Applied Maths, Version 7.6). Total number of isolates = 110; total number of ST = 43; number of loci per isolate = 7; number of identical loci for MLST clonal complexes definition = 5; total number of BURST groups detected = 6; number of re- samplings for bootstrapping = 1000. The MLST clonal complexes UM01 to UM12 are indicated by arrows.

Figure 4.

Multi-locus sequence typing (MLST) phylogenetic tree analysis for the 110 Pseudomonas aeruginosa strains obtained from mink using bionumerics computer software (Applied Maths, Version 7.6). Total number of isolates = 110; total number of ST = 43; number of loci per isolate = 7; number of re-samplings for bootstrapping = 1000. The MLST clonal complexes UM01 to UM12 are shown within the brackets.

The UM01 was responsible for 7 outbreaks in 6 different locations, including Dalian (Liaoning, China) in 2007, Linyi (Shandong, China) in 2007, Jilin province in 2008, Dalian (Liaoning, China) in 2009, WeiHai (Shandong, China) in 2010, Shandong province in 2012, and Wendeng (Shandong, China) in 2015. The UM02 was responsible for 3 outbreaks in 2 locations, including Wendeng (Shandong, China) in 2014 (2 outbreaks) and Jilin province in 2007. The UM03 was responsible for 11 outbreaks in one particular location, including Wendeng (Shandong, China) in 2013 (1 outbreak) and Wendeng (Shandong, China) in 2014 (10 outbreaks). The UM04 was responsible for 2 outbreaks in 2 locations, including Shandong province in 2012 and Jilin province in 2007. The UM05 was associated with outbreaks in one location, predominantly Dalian (Liaoning, China) in 2007, but was also detected in soil isolates from the same farm in 2013. The UM06 was identified in 2 locations, including Wendeng (Shandong, China) in 2014 and Dalian (Liaoning, China) in 2013 (soil samples). The UM08, UM09, and UM10 were associated with one outbreak each, from Dalian (Liaoning, China), Shenyang (Liaoning, China), and Shandong province, respectively. From these data it is evident that UM05, UM07, and UM12 belong to Dalian (Liaoning, China), and UM02 and UM03 belong to Wendeng (Shandong, China) from 2013 to 2015, which have the feature of regional aggregation, despite different timelines.

Antimicrobial susceptibility of P. aeruginosa in mink

All 69 non-repeated isolates from mink were sensitive to polymyxin B and polymyxin E. Generally, resistance to fluoroquinolone, aminoglycoside, and ceftazidime was much lower (9/69, 13%). Six strains were resistant to levofloxacin, ciprofloxacin, and enrofloxacin. Two strains were resistant to levofloxacin and enrofloxacin. One strain was resistant to enrofloxacin, and one strain was resistant to both aminoglycoside and fluoroquinolone. One MDR strain was resistant to ceftazidime, aminoglycoside, and fluoroquinolone. A high incidence of resistance was found to spectinomycin (100% strains MIC ≥ 128 μg/mL), florfenicol (100% strains MIC ≥ 128 μg/mL), and fosmycin (51/69, 74%, with a MIC ranging from 32 to 1024 μg/mL; Table I).

Table I.

Minimum inhibitory concentration (MIC) of 9 P. aeruginosa-resistant drugs (μg/mL).

Strain number	Gentamicin	Amikacin	Ceftazidime	Ciprofloxacin	Enrofloxacin	Levofloxacin
99	—	—	—	—	64	32
114	—	—	—	—	16	—
27	—	—	—	—	128	16
11	—	—	—	—	16	—
100	—	—	—	128	256	128
95	≥ 512	≥ 512	—	—	8	—
59	≥ 512	—	≥ 512	128	512	128
101	—	—	—	128	256	64
60	256	0.5	—	128	256	128

— Sensitive value to drugs were not shown.

Alterations in glpt and fosA genes

Amplification of the glpt gene from 2 fosmycin resistant strains led to the identification of deletions at position T₂₂₀CGCCT₂₂₆ (ΔIle74-Ala75) in both strains; MIC was 1024 μg/mL. In 65% of samples (72/110); however, fosA was detected and no mutations were found. The MICs of fosA-positive strains to fosmycin ranged from 256 to 1024 μg/mL.

The PCR amplification of PMQR and MBL genes

Of the 9 quinolone resistant isolates tested, none were found to express either PMQR genes (qnrA, qnrB, qnrC, qnrD, qnrS, and qepA) or MBL genes (bla_IMP, bla_VIM, bla_SPM, bla_GIM and bla_SIM), except for one strain 95, which expressed aac(6′)-Ib.

The PCR amplification of QRDR

Six quinolone-resistant strains (MIC, ciprofloxacin 128 μg/mL, enrofloxacin 64 to 256 μg/mL, levofloxacin 16 to 128 μg/mL) were evaluated for QRDR mutations in both gyrA (Thr-83→Ile) and ParC (Ser-87→Leu), while 2 quinolone-resistant strains (MIC, enrofloxacin 8 to 16 μg/mL) were evaluated for QRDR mutations in only gyrA (Thr-83→Ile or Asp-87→Tyr). The MIC of strains that expressed 2 mutations was higher than that of strains expressing only one mutation (Table II).

Table II.

Minimum inhibitory concentration (MIC) and gene detection for fluoroquinoloneresistant strains.

Strain number	CIP	ENO	LEV	ParC	gyrA	ParE	gyrB	aac(6′)-Ib
95	—	8	—	—	—	—	—	+
114	—	16	—	—	87 (D-Y)	—	—	—
11	—	16	—	—	83 (T-I)	—	—	—
27	—	128	16	87(S-L)	83 (T-I)	—	—	—
100	128	256	128	87(S-L)	83 (T-I)	—	—	—
99	—	64	32	87(S-L)	83 (T-I)	—	—	—
59	128	512	128	87(S-L)	83 (T-I)	—	—	—
101	128	256	64	87(S-L)	83 (T-I)	—	—	—
60	128	256	128	87(S-L)	83 (T-I)	—	—	—

S — ser; L — Leu; T — Thr; I — Ile; D — Asp; Y — Tyr; CIP — ciprofloxacin; ENO — enrofloxacin; LEV — levofloxacin; ParC — DNA topoisomerase IV subunit A; gyrA — DNA gyrase subunit A; ParE — DNA topoisomerase IV subunit B; gyrB — DNA gyrase subunit B; aac(6′)-Ib — Aminoglycoside N(6)-acetyltransferase.

+ detected; — not detected or mutation.

Whole genome sequencing of P. aeruginosa 59

Automated genome annotation of the MDR strain, P. aeruginosa 59, was carried out using the PacBio RSII server, which predicted 6.973 million coding sequences, including 6519 genes in the chromosomes and 46.627 kb of coding sequence, including 65 genes in one plasmid. The anti-drug genes were used in a BLAST search [blastp, e-value ≤ 1e-5, identity ≥ 40%, coverage ≥ 40%], in the Antibiotic Resistance Genes Database (http://ardb.cbcb.umd.edu/). Several multi-drug efflux pumps, MexAB-OprM, MexEF-OprN, and MexIH-OpmD of the resistance nodulation cell division (RND) family, which were previously reported in P. aeruginosa strains (22–25), were also detected in the P. aeruginosa PA59 genome.

Twenty-seven types of genes were identified as resistant to 14 classes of antimicrobial agents, including chloramphenicol, sulfonamide, aminoglycoside, kasugamycin, β-lactamase, penicillin, polymyxin, bacitracin, streptomycin, tetracycline, fosfomycin, fluoroquinolone, spectinomycin, and penicillin. The resistance genes for β-lactamase were bl1_pao while the resistance genes of MBL genes for β-lactamase were not detected. The resistance genes for fluoroquinolone were mdtk and the QRDR mutations in both gyrA (Thre-83→Ile) and parC (Ser-87→Leu). The resistance genes for aminoglycoside were aph3vb, aph3ia, pur8, aac6iia, and aac3iia. No resistant genes were detected in the one plasmid tested (Table III).

Table III.

Drug-resistant genes found in the Pseudomonas aeruginosa59 genome.

Antimicrobial agents	Antimicrobial resistance genes found in the PA59 genome
Chloramphenicol	catb3, catb3, catb4, cml_e3
Sulfonamide	dfra26, sul1
Efflux pump	MexAB-OprM, MexIH-OpmD, MexEF-OprN
Aminoglycoside	aph3vb, aph3ia, pur8, aac6iia, aac3iia
Kasugamycin	ksga
β-lactamase	bl1_pao
Penicillin	pbp2, pbp1a
Polymyxin	arna
Bacitracin	baca
Streptomycin	aph6ic, ant3ia
Tetracycline	tet41, tetg
Fosfomycin	fosA
Fluoroquinolone	mdtk
Spectinomycin	ant3ia

BLASTp in Antibiotic Resistance Genes Database (ARDB), e-value ≤ 1e-5. Identity ≥ 40%, coverage ≥ 40%.

Discussion

Since P. aeruginosa in mink occurred frequently in recent years in Shandong, Liaoning, and Jilin provinces in China (3,8), MLST and PFGE assays were used to compare the epidemiological relationships of P. aeruginosa isolates in mink recovered from different sources. Our results indicate that typing could be an essential assay to determine the strain genotyping relatedness.

The PFGE clusters and MLST clone complexes showed characteristics associated with regional clustering appeared in Wendeng (Shandong, China) and Dalian (Liaoning, China) despite different timelines. Also, we found the same cluster in one outbreak in one farm. The study further strengthens the evidence linking strain epidemic with small range. Several closely related PFGE patterns were found for P. aeruginosa in Weihai (Shandong, China) and throughout Shandong province (8), and that the same clone appeared to have spread, possibly via feed, sinks, equipment, personnel, air, or food production personnel (4,12).

Meanwhile, 11 outbreaks in 6 regions were linked to STCC UM01 between 2007 and 2015. ST1058 appeared in 5 outbreaks which occurred in Dalian (Liaoning, China), Jilin province, and Shandong province. The MLST clonal complexes identified were spread over different areas and timepoints, the clonal transmission and gene correlation has occurred between Dalian (Liaoning, China), Jilin province, and Shandong province over time. The authors believe that animal trade was an important factor. There are several large mink farms in Dalian (Liaoning, China), which provide excellent mink sources to other mink farms. In addition, the same STCC in mink isolates were identified in soil samples isolated from Dalian (Liaoning, China) mink farms in 2013. This situation implies that some strains, or their genetic variants, may persist in the environment for years with recurrent outbreaks. Moreover, P. aeruginosa is a highly variable strain in which alleles are at least 3.5 times more likely to change by recombination among a subset of environmental isolates (11,26). This may explain the variation in ST genotype variation and the observed genetic correlation within STCC.

The ST diversity within mink isolates were not distinct from that of global isolates when compared among 43 ST in this study. Eleven ST were identified as the same as those in humans, other animals, and water isolates (ST260, ST266, ST155, ST144, ST244, ST399, ST882, ST116, ST167, ST189, ST655, and ST1228), of these ST116 was found in soil from a mink farm, while the remaining ST were identified in mink isolates (Table IV; 27). These specific clones have previously been identified in patients without CF (ST260, n = 1; ST244, n = 3), patients with CF (ST155, n = 5; ST244, n = 1; ST399, n = 1; ST882, n = 1; ST116, n = 1; ST655, n = 2), canine ear isolates (ST266, n = 2; ST144, n = 4; ST244, n = 2; ST167, n = 1; ST189, n = 1; ST1228, n = 1), bovine mastitis (ST244, n = 1), and river water (ST266, n = 1; ST155, n = 2; ST144, n = 1) (14). ST155 was also linked to internationally widespread clones, outbreaks, and sometimes to multi-drug resistance (14) in P. aeruginosa outbreaks in mink. An earlier study showed that ST260, isolated from a patient with CF in Thailand was not only MDR, but also carried bla_IMP-14 (28). However, in mink isolates, this was the only MDR strain that was resistant to 14 different types of drugs but did not carry bla_IMP-14. This implies that the genotype of P. aeruginosa in the mink population is diverse and highly related to humans, animals, and the environment, and further suggests the possible risk of mink-to-human transfer.

Table IV.

A list of same Pseudomonas aeruginosa multi-locus sequence typing (MLST) from mink isolates acquired from other sources, as reported by Kidd et al (27).

ST from the mink	Same ST reported	Isolate site	Number
260	Non-CF human (maxillary)	Swab: Sinus	1
266	Water	Brisbane River	1
	Non-CF human	Urine	1
	canine	Ear	2
155	CF patient	Sputum	5
	Swab: Air-water interface	Brisbane River	2
	Non-CF human	Ear	4
144	Air-water interface	Brisbane River	1
244	CF patient	Sputum	1
	Canine	Respiratory	1
	Canine	Swab: Ear	1
	Bovine mastitis	Swab	1
	Non-CF human	Blood	1
		Wound (heel)	1
		Urine	1
399	CF patient	Sputum	1
882	CF patient	Sputum	1
116	CF patient	Sputum	1
655	CF patient	Sputum	2
167	Canine	Ear	1
189	Canine	Ear	1
1228	Canine	Otitis	1

CF — cystic fibrosis; ST — sequence type.

In contrast to the high resistance rates and frequent discovery of PMQR and the MBL gene in humans (1,2,29,30), low resistance rates have been reported in mink (8,31) and other animals, such as dogs, cows, horses, mice, rabbits, and guinea pigs (14). Other studies have reported a lack of multidrug resistance in dairy cows and horses, along with 14% to 34% resistance to multiple antimicrobials in soft tissue infections from dog isolates (14,32); 13.3% of isolates were also shown to be resistant to ticarcillin/clavulanic acid, ciprofloxacin, and levofloxacin in mink (8). Our results were consistent with these earlier reports, in that a lower proportion of strains were resistant to ceftazidime, aminoglycosides (gentamicin, amikacin), and fluoroquinolones (enrofloxacin, ciprofloxacin, levofloxacin), and that all of the strains were sensitive to polymycin, spectinomycin, and florfenicol. Furthermore, 74% of strains were resistant to fosmycin because P. aeruginosa shows innate resistance to most antimicrobials, such as β-lactamase (ampicillin/sulbactam sodium, cefazolin, ceftriaxone and ampicillin, cefuroxime sodium and cefuroxime axetil), trimethoprim/sulfamethoxazole, spectinomycin, florflenicol, and fosmycin (33). Collectively, these results provide an indication of the relative efficacy of these drugs on mink farms.

Of the 9 fluoroquinolones-resistant strains tested, we identified 6 strains that had a high rate of resistance (MIC: 16 to 256 μg/mL) to fluoroquinolone and have the mutation site gyrA (Thr-83→Ile) and ParC (Ser-87→Leu). These 2 mutations were most common in QRDR in P. aeruginosa from minks, dogs, and humans, and lead to a higher MIC (21,32,34). While integrons contained different gene cassettes (aacA4-aadA1, blaOXA-31-aadA2, aadA1-arr-3-catB3, and cmlA5-cmlA-aadA1) to all of the fluoroquinolone strains tested in soft tissue infections from dogs in China (24), we did not find this situation in our current study. However, in one isolate, we identified the antibiotic resistance gene aac(6′)-Ib for fluoroquinolone, which is known to be mediated by an encoded plasmid (30,35,36). This implies that transfer could occur between resistant conjugative plasmids that encode the relevant function to promote cell-to-cell DNA transfer (35).

This study discovered a high proportion of fosfomycin-resistant isolates that were induced by the fosA gene, and existed in chromosomes, although the use of this antibiotic is not currently authorized in veterinary medicine. Furthermore, we identified T₂₂₀CGCCT₂₂₆ deletion (ΔIle74-Ala75) of the glpt gene in 2 strains, which have also been reported to be associated with fosfomycin and fosfomycin-resistant (Fos-R) mutants in vitro (37). The high MIC to fosfomycin have also been found in humans, dogs, horses, and cows (14,38).

One MDR strain, P. aeruginosa PA59, was sequenced and 27 resistant genes were identified for 14 types of drugs in the chromosomes, including β-lactamase, chloramphenicol, sulfonamide, polymyxin, streptomycin, fluoroquinolone, and spectinomycin, along with multi-drug efflux pumps MexAB-OprM, MexIH-OpmD, and MexEF-OprN. The results are consistent with a drug-resistant phenotype. Pseudomonas aeruginosa is intrinsic resistant to chloramphenicol, spectinomycin, tetracycline, and fosfomycin, which were related to the intrinsic resistant genes catb3, catb3, catb4, cml_e3, tet41, tetg, fosA, and ant3ia. The QRDR were detected in fluoroquinolone resistant strain PA59. We also detected MdtK, belonging to the multidrug and toxic compound extrusion family C, which contributed to the reduced susceptibility to fluoroquinolone (39). Our study has revealed that aminoglycoside co-occurrence in genes encoding aminoglycoside-modifying enzymes [aac(6′)-Ih, aac(3)-Ia, aac(3)-IIa, aac(6′)-Ib, aph(3′)-Ia, aph(3′)-VI] contributes to the resistance to gentamicin and had resistant/intermediate susceptibility to amikacin. This explained the susceptibility to amikacin and the resistance to gentamicin of PA59 in our study; it appears to be a consequence of the accumulation of mutations and the acquisition of resistance determinants by the transfer of transposons and integrons, which may lead to the formation of clusters of resistance genes, termed “resistance islands” (40). Additionally, we found that one plasmid was detected in the genome but did not contain resistance genes.

Our work proved that P. aeruginosa isolates from mink are genetically diverse and endemic in Dalian (Liaoning, China) and Wendeng (Shandong, China). Many anti-drug genes were also detected. This disease may become a potential risk factor to human health. This may be especially important for patients with CF, resulting from unique P. aeruginosa strains acquired in the environment of the patient. In the future, further studies on the genetic relationship of P. aeruginosa isolates from mink farms and the detection of other, clinically important virulence factors are needed.

Nucleotide sequence accession number

The complete genome sequence of P. aeruginosa PA59 has been deposited in GenBank under the accession numbers CP024630 and CP024631.