Antimicrobial Drug Resistance in Fish Pathogens

ABSTRACT

Major concerns surround the use of antimicrobial agents in farm-raised fish, including the potential impacts these uses may have on the development of antimicrobial-resistant pathogens in fish and the aquatic environment. Currently, some antimicrobial agents commonly used in aquaculture are only partially effective against select fish pathogens due to the emergence of resistant bacteria. Although reports of ineffectiveness in aquaculture due to resistant pathogens are scarce in the literature, some have reported mass mortalities in Penaeus monodon larvae caused by Vibrio harveyi resistant to trimethoprim-sulfamethoxazole, chloramphenicol, erythromycin, and streptomycin. Genetic determinants of antimicrobial resistance have been described in aquaculture environments and are commonly found on mobile genetic elements which are recognized as the primary source of antimicrobial resistance for important fish pathogens. Indeed, resistance genes have been found on transferable plasmids and integrons in pathogenic bacterial species in the genera Aeromonas, Yersinia, Photobacterium, Edwardsiella, and Vibrio. Class 1 integrons and IncA/C plasmids have been widely identified in important fish pathogens (Aeromonas spp., Yersinia spp., Photobacterium spp., Edwardsiella spp., and Vibrio spp.) and are thought to play a major role in the transmission of antimicrobial resistance determinants in the aquatic environment. The identification of plasmids in terrestrial pathogens (Salmonella enterica serotypes, Escherichia coli, and others) which have considerable homology to plasmid backbone DNA from aquatic pathogens suggests that the plasmid profiles of fish pathogens are extremely plastic and mobile and constitute a considerable reservoir for antimicrobial resistance genes for pathogens in diverse environments.

INTRODUCTION TO ANTIMICROBIAL AGENT USE IN AQUACULTURE

With an increasing global population, aquaculture can provide a source for inexpensive protein. However, many of these sources are and will continue to be located in developing countries where medicated feeds are inexpensive and widely available for broad application on farms. These feeds can contain antimicrobial agents from classes generally considered unsafe for use in food-producing animals due to their critical importance to human medicine (e.g., fluoroquinolones).

The availability of antimicrobial agents over the counter in many developing countries with little to no veterinarian oversight can contribute directly to the emergence of antimicrobial resistance in bacterial pathogens of fish (Table 1) and of humans. Some environmental or commensal bacteria may also pass their resistance determinants to veterinary and zoonotic pathogens.

TABLE 1.

Frequently isolated bacterial pathogens of fish

Organism	Disease
Aeromonas caviae	Motile aeromonad septicemia
Aeromonas hydrophila
Aeromonas sobria
Aeromonas salmonicida	Furunculosis, ulcer disease, carp erythrodermatitis
Edwardsiella ictaluri	Enteric septicemia of catfish
Edwardsiella tarda-like organisms	Motile Edwardsiella septicemia (Edwardsiella piscicida), nonmotile Edwardsiella septicemia (Edwardsiella anguillarum)
Flavobacterium branchiophilum	Bacterial gill disease
Flavobacterium columnare	Columnaris disease
Flavobacterium psychrophilum	Cold-water disease, rainbow trout fry syndrome
Francisella spp.	Francisellosis
Lactococcus garvieae	Lactococcosis, Lactococcus septicemia
Moritella viscosa	Winter ulcer disease
Mycobacterium spp.	Mycobacteriosis
Photobacterium damselae subsp. damselae	Vibriosis
P. damselae subsp. piscicida	Photobacteriosis, fish pasteurellosis, pseudotuberculosis
Piscirickettsia salmonis	Piscirickettsiosis, salmonid piscirickettsial septicemia
Plesiomonas shigelloides	Winter disease
Pseudomonas spp.	Pseudomoniasis
Pseudomonas anguilliseptica	Red spot disease
Renibacterium salmoninarum	Bacterial kidney disease
Streptococcus agalactiae	Group B streptococcosis
Streptococcus dysgalactiae	Group C streptococcosis
Streptococcus iniae	Streptococcosis
Streptococcus phocae	Warm-water streptococcosis
Tenacibaculum maritimum	Salt-water columnaris, marine flexibacteriosis
Vagococcus salmoninarum	Cold-water streptococcosis
Vibrio salmonicida	Cold-water vibriosis, Hitra disease
Vibrio spp.	Vibriosis
Yersinia ruckeri	Enteric redmouth disease

In any given farm situation, many factors can contribute to the perceived need for chemotherapeutic intervention. The reality is, in some countries where medicated feeds can be obtained inexpensively, there is widespread and unrestricted use of antimicrobial agents to prevent bacterial infections. The use of a wide variety of antimicrobial agent classes, including nonbiodegradable antimicrobial agents useful in human medicine, ensures that they remain in the aquatic environment, exerting their selective pressure for long periods of time. This process has resulted in the emergence of antimicrobial resistant bacteria in aquaculture environments, the increase of antimicrobial resistance in fish pathogens, and the increasing potential for transfer of these resistance determinants to bacterial pathogens of terrestrial animals and humans.

Conversely, when a disease outbreak is believed to be imminent, improved water quality and/or metaphylactic treatment can be used efficiently to reduce animal suffering and economic losses. The timeliness of such interventions is absolutely critical, particularly in aquatic animal medicine, where declines in water quality and/or increases in pathogen shedding can quickly result in catastrophic losses and facility closures.

FACTORS CONTRIBUTING TO ANTIMICROBIAL RESISTANCE IN AQUACULTURE

Therapeutic Use on the Farm

Antimicrobial agents are most commonly administered in aquaculture settings using broadcasted medicated feeds. Often, the largest and strongest fish on farms consume the majority of the medicated feed, likely affording them a positive clinical outcome, whereas the smaller and/or sick fish are likely to be underdosed and considerably more prone to disease and ultimately death (1, 2). Alternative dosing strategies (i.e., pulse, increased dose) can circumvent some of these issues, but optimization of medicated feed dosages and durations in fish can be extremely challenging. Logistical difficulties confronted in pharmacokinetic investigations of fish include determination of both the amount of ingested antimicrobial agent in each animal studied and serum drug concentrations during the dosing regimen without introducing significant handling stress. Miller et al. used a gastric lavage technique and preweighed, homogenized oxytetracycline-medicated feed to standardize the amount of antimicrobial agent ingested per gram of body weight (3). Destructive sampling was used to reduce experimental bias and ensure sufficient blood volumes. Low standards of deviation were noted at all sampling time points. Environmental factors can also complicate therapy because temperature, water salinity, and the presence of inactivating divalent cations (e.g., Mg⁺² and Ca⁺²) can affect drug absorption and ultimately the amount of bioavailable free drug.

Agricultural Runoff and Human Pollution

Release of antimicrobial-resistant pathogens from an aquaculture facility or from agricultural runoff into surrounding areas can have an impact on the overall movement of resistance through microbial communities. Petersen et al. showed that integrated farming strongly favored antimicrobial-resistant bacteria in a pond environment (4). As the source, they attributed the selective pressure of the antimicrobial agents used or the direct introduction of antimicrobial-resistant bacteria from animal manure. Leaching of antimicrobial agents from fish farming operations into surrounding waterways may also contribute to the spread of antimicrobial-resistant animal and zoonotic pathogens. Additional research in this area is needed to fully characterize the extent of the issue at the national level, and then a determination can be made as to the breadth of the animal and human health risk. Due to the vast disparity between countries and regions in availability of antimicrobial agent classes and types of aquaculture farming practices, inherent risk profiles will be different.

ANTIMICROBIAL RESISTANCE GENES FOUND IN FISH PATHOGENS

The widespread use of antimicrobial agents to treat and control fish diseases has led to the emergence of antimicrobial resistance to single and multiple classes of antimicrobial agents in important fish pathogens. In one of the first reports of resistance, in fish pathogens, Aoki et al. investigated mobile R plasmids in three Aeromonas salmonicida strains (5), one of which was isolated in 1959 by S. F. Snieszko in the United States and has been considered the standard A. salmonicida strain of reference for many later genetic investigations. The R plasmids were transmissible to Escherichia coli, transferring two resistant phenotypes. The standard U.S. strain transmitted resistance to sulfathiazole and tetracycline, and the two Japanese strains transmitted resistance to chloramphenicol and dihydrostreptomycin. Aoki and collaborators, over a series of natural transformation and DNA-DNA hybridization studies in the 1970s and 1980s, were the first to fully characterize a number of fish-pathogenic bacterial species (Edwardsiella tarda, Vibrio anguillarum, A. hydrophila, and Pasteurella piscicida [since reclassified as Photobacterium damselae subsp. piscicida]) containing R plasmids and their associated antimicrobial resistance genes (6–9).

The rising concern regarding the development of antimicrobial resistance and the reporting of the discovery of antimicrobial resistance genes in fish pathogens and commensals in and around aquaculture systems has increased due to the occasional close proximity of human populations, recreational fisheries, and other animal production systems. These relationships have prompted a number of notable investigations, review articles, and opinion articles (10–13).

A. salmonicida

Considerable antimicrobial resistance and corresponding resistance genes in A. salmonicida have been identified from clinical infections of fish, with one of the first reports of resistance being described in 1971 (5). Transmissible resistance was found conferring resistance to sulfathiazole and tetracycline on one R plasmid and a second R plasmid conferring phenotypic resistance to chloramphenicol and dihydrostreptomycin. In addition to this early report of resistance to four classes of antimicrobial agents (tetracyclines, sulfonamides, phenicols, and aminoglycosides), quinolone resistance has also been observed in A. salmonicida. Tetracyclines are commonly used to treat furunculosis caused by A. salmonicida. Genes conferring resistance to tetracycline are commonly associated with transfer via mobile genetic elements, and quinolones are of particular concern because they are used to treat Aeromonas infections in humans and are widely used in some parts of the world in aquaculture. Oxytetracycline-resistant strains (MIC ≥ 8 μg/ml) and oxolinic acid-resistant strains (MIC ≥ 1 μg/ml) from Korea were screened for multidrug resistance (MDR) phenotypes (i.e., resistant to more than two classes) and antimicrobial resistance determinants, revealing that a majority of the strains were considered MDR (14). One strain was found to be resistant to ampicillin, gentamicin, oxytetracycline, enrofloxacin, and oxolinic acid. The tetracycline resistance gene tet(A) was identified in most tetracycline-resistant isolates, and one isolate was positive for tet(E). Screening for mobile genetic elements was not conducted in this study; however, the sequenced tet(A) and tet(E) genes showed 100% homology to tet(A) in pRAS1 (GenBank accession no. AJ517790.2) and tet(E) in pAsa4 (GenBank accession no. CP000645.1). These plasmids have been associated with carriage in A. salmonicida. The quinolone-resistance-determining regions (QRDRs) consisting of the gyrA and parC genes were sequenced and, among the quinolone-resistant A. salmonicida strains, all harbored point mutations in the gyrA codon-83 which are responsible for the corresponding amino acid substitutions of Ser83→Arg83 or Ser83→Asn83. The authors detected no point mutations in other QRDRs, such as gyrA codons-87 and -92, and parC codons-80 and -84; the mobile qnr genes were not detected. Genetic similarity was assessed via pulsed-field gel electrophoresis, and the results indicated high clonality among the Korean antimicrobial-resistant strains of A. salmonicida (14). In agreement with this study, another investigation of the mechanism of quinolone resistance in seven oxolinic acid-resistant A. salmonicida isolates from Atlantic salmon was conducted by PCR amplification and sequencing of the gyrA gene. All strains with MICs ≥ 1 μg/ml (defined as resistant) were positive for a serine to isoleucine mutation at the 83 position in the gyrA- encoded amino acid. These mutations conferring quinolone resistance were identified solely at the 83 position in the gyrA-encoded protein and were nontransmissible (15).

Integrons carrying antimicrobial resistance genes have been widely identified in A. salmonicida and associated with multiple antimicrobial resistance phenotypes. In an effort to screen for class 1 integrons in A. salmonicida over a wide geographic range, PCR amplification of the class 1 integron structural component sul1 gene was utilized as a marker for integron carriage (16). Twenty-one sulfonamide-resistant A. salmonicida strains from France, Japan, Switzerland, Norway, Scotland, and the United States were determined to carry class 1 integrons. Of the strains positive for class 1 integrons, 19 possessed a single gene cassette and one possessed two resistance genes within the cassette. A variety of resistance genes were identified in the gene cassettes: aadA1 (three strains), aadA2 (eight strains), dfrA16 (five strains), dfrB2c (three strains), and qacG and orfD (one strain). Twenty strains were tetracycline-resistant and positive for either tet(A) (n = 16) or tet(E) (n = 4), with tet(A) being frequently associated with a Tn1721 transposon. Seven integron-negative sulfonamide-resistant strains possessed the sul2 gene. Geographic associations were reported, with aminoglycoside resistance being associated with the aadA2 gene in Scotland and with the aadA1 gene in Switzerland and France. The dfrA16 gene cassette was found only in Norwegian strains, whereas dfrB2c was identified as causing trimethoprim resistance in Scotland.

Mobile genetic elements associated with MDR phenotypes have been identified in A. salmonicida since the earliest reports of resistance by Aoki et al. (5). Early plasmids associated with drug resistance to sulfonamides, phenicols, and streptomycins were associated with the pAr-32 plasmid isolated in Japan in 1970 from an A. salmonicida isolate carrying a class 1 integron. Molecular characterization of a similar pRA3 plasmid from A. hydrophila in Japan revealed that these plasmids were nearly identical, suggesting transmission between A. hydrophila and A. salmonicida for some time (17). The plasmid pRA3 is the reference plasmid for the incompatibility group U (IncU). Atypical and typical A. salmonicida strains were resistant to tetracycline and potentiated sulfonamides and were found to carry the IncU pRAS1 plasmid, which is responsible for the resistance phenotypes. Molecular mapping of the pRAS1 plasmid from A. salmonicida showed that a 12-kb antimicrobial resistance-determining region was encoded, including a class 1 integron carrying the dfrA16, sul1, qac, and tet(A) genes conferring resistance to trimethoprim, sulfonamides, quaternary ammonium compounds, and tetracyclines. Molecular analysis of pAr-32 and pRAS1 showed that the backbone of pRAS1 is similar to that of pAr-32, indicating that pRAS1 may have evolved from pAr-32, with the addition of and rearrangement of antimicrobial resistance genes. Conjugation experiments showed the potential for the effective spread of the IncU pRAS1 plasmid to other aquatic pathogens such as Vibrio cholerae, Vibrio parahaemolyticus, and Yersinia ruckeri (17).

Since their identification in the United States in 1959, MDR A. salmonicida strains have been extensively studied for their rich and diverse plasmid repertoire. A. salmonicida was reported to carry three cryptic plasmids not conferring antimicrobial resistance genes [pAsa1 (pAsal2), pAsa2 (pAsal3), pAsa3], and four plasmids not conferring antimicrobial resistance genes (pAsal1, pAsal1B, pAsa5, pAsa6) (18). The cryptic plasmids [pAsa1 (pAsal2), pAsa2 (pAsal3), pAsa3] are ColE1- and ColE2-type plasmids and have been reported to encode genes responsible for replication, stability, and mobilization (19). Further investigation is ongoing with regard to the functions of the pAsal-type plasmids; however, pAsa5, pAsal1, pAsal1B, and pAsa6 are known to bear genes encoding type III secretion systems and effectors. Plasmids in A. salmonicida which carry antimicrobial resistance genes have been reported to include pRAS1 and pAr-32 (described above); pASOT, pASOT2, and pASOT3 [aadA2, tet(A)]; pSN254b [tet(A), floR, sul1, sul2, bla_CMY, aadA, strA, strB, sugE2, qacEΔ1]; pAB5S9b [tet(H), floR, sul2, strA, strB]; pAsa7 (cat); and pAasa4 [tet(E), sul1, aadA, cat] (18, 19). A new pAsa8 plasmid was characterized in Canada from A. salmonicida carrying a large number of antimicrobial resistance genes [tet(A), tet(G), floR, sul1, bla_PSE-1, and aadA2] and a complex class 1 integron with similarity to the Salmonella genomic island 1 (SGI132) found in Salmonella enterica (20). Plasmid exchange between terrestrial pathogens such as S. enterica serotypes has been shown previously (21), as exemplified by the IncA/C pSN254b plasmid, a variant of pSN254 found in S. enterica. The plasmids of the IncA/C group are known to be conjugative and found, in addition to A. salmonicida and S. enterica, in a broad range of bacteria such as Yersinia pestis, Klebsiella pneumonia, Aeromonas hydrophila, Photobacterium damselae subsp. piscicida, and E. coli. The IncA/C plasmids are also known to regulate the excision of SGI132 and to serve as helpers for its mobilization in trans (20).

While it is beyond the scope of this chapter to go into intimate detail of all known plasmids identified in A. salmonicida strains, it is obvious that the rich and varied plasmid profiles of this important fish pathogen have allowed strains to persist after antimicrobial agent therapy, and they may serve as a considerable reservoir for transmissible antimicrobial resistance elements.

Edwardsiella ictaluri and E. tarda

Despite the U.S. Food and Drug Administration approval of oxytetracycline, ormetoprim-sulfadimethoxine, and florfenicol in channel catfish, reports of antimicrobial resistance of the enteric septicemia of the catfish pathogen E. ictaluri are scarce. In 1993, ormetoprim-sulfadimethoxine-resistant E. ictaluri was isolated from treated catfish in Virginia and Mississippi (22). A 55-kb conjugative plasmid was found to be responsible for the resistance to ormetoprim-sulfadimethoxine. It also conferred resistance to tetracycline, oxytetracycline, streptomycin, trimethoprim, and trimethoprim-sulfamethoxazole.

In 2009, florfenicol resistance in E. ictaluri was found to be conferred via an MDR IncA/C plasmid, pM07-1 (23). The plasmid was self-transmissible and conferred resistance to florfenicol, chloramphenicol, tetracycline, streptomycin, ampicillin, amoxicillin-clavulanic acid, ceftiofur, and cefoxitin and reduced susceptibility to trimethoprim-sulfamethoxazole and ceftriaxone. The floR and bla_CMY-2 genes were detected on the plasmid via PCR and subsequently sequenced, revealing 100% nucleotide identity of the bla_CMY-2 gene from plasmids in the human pathogens S. enterica, E. coli, and K. pneumoniae. Discovery of IncA/C plasmids carrying floR and bla_CMY-2 resistance determinants in S. enterica serotype Newport and A. salmonicida highlight the mobility and potential promiscuity of these plasmids in aquatic and terrestrial environments (21, 24).

E. tarda has a broad host range and can be pathogenic to fish including economically important species such as catfish, turbot, flounder, and salmon (25) and to humans. Transferable, plasmid-mediated antimicrobial resistance has been identified in E. tarda going back to the 1970s (6) when naturally occurring R plasmids were found to confer resistance to up to five classes of antimicrobial agents (sulfonamides, tetracyclines, streptogramins, phenicols, and aminoglycosides). In China, a pathogenic E. tarda strain, TX01, from an outbreak at a fish farm was found to be resistant to kanamycin, tetracycline, and chloramphenicol and possessed associated antimicrobial resistance genes (25). The strain was positive for known kanamycin resistance determinants, catA3, and tet(A) genes which were cloned, sequenced, and compared to human and animal strains. These genes were 99 to 100% identical to genes on plasmids and transposons in human and terrestrial animal strains. Further, the carrier plasmid of the tet and catA3 genes, pETX, demonstrated its transferability among different genera of bacteria. Due to its transmissibility among a wide variety of genera and the high level of resistance gene identity between human, environmental, and fish strains, this plasmid has likely played a persistent role in the dissemination of antimicrobial resistance within and possibly between environments (25).

Plasmid typing, in addition to other molecular subtyping methods, can expand on our knowledge surrounding the potential for transmission between environments and geographic locations. Ninety-four E. tarda isolates from diseased eels were tested for their susceptibility to eight commonly used antimicrobial agents in Taiwan (26). The isolates were highly susceptible to ampicillin, amoxicillin, florfenicol, oxolinic acid, and flumequine. However, 21.3% were oxytetracycline resistant and positive for tet(A), and 90% also carried tet(M). Interestingly, the efflux pump inhibitor omeprazole reduced oxytetracycline MICs 2- to 8-fold, suggesting that an intact efflux pump, presumably encoded by tet(A), is required for high-level tetracycline resistance. Real-time PCR experiments showed that increased expression levels of tetA and tetR contribute to oxytetracycline resistance. Southern blot hybridization revealed that all oxytetracycline-resistant isolates carried the tet(A) gene on a 50- to 70-kb plasmid. Plasmid typing and restriction fragment length polymorphism analysis identified similarities across different farms in the same region, indicating that horizontal spread and local transmission of the plasmid in the oxytetracycline-resistant E. tarda population has occurred.

In addition to antimicrobial resistance determinants, plasmids and other mobile genetic elements can simultaneously carry virulence genes, giving pathogens even greater potential to cause disease and persist during and after antimicrobial agent exposure. A genetic analysis of a South Korean E. tarda strain (E. tarda CK41) with reduced susceptibility to kanamycin, ampicillin, tetracycline, and streptomycin revealed a 70-kb plasmid responsible for transferring kanamycin and tetracycline resistance genes (27). These authors hypothesized that virulence genes were concomitantly housed on plasmids in E. tarda and conducted survival experiments with a plasmid-cured strain, E. tarda CK108, and its parent strain, CK41. Of the fish infected with E. tarda, 80% survived infection with E. tarda CK108, while only 20% survived infection with the parent strain, CK41, which contained the antimicrobial resistance and virulence genes. Due to the difference in survival rate, it is clear that additional virulence genes can be transmitted with antimicrobial resistance genes on plasmids in Edwardsiella spp.

Y. ruckeri

Y. ruckeri causes the economically devastating enteric redmouth disease in several fish species, including rainbow trout (Onchorynchus mykiss) (28). Because of the economic impact, antimicrobial agents such as quinolones and potentiated sulfonamides are often used to combat infections, prompting investigations into the development of antimicrobial resistance to these and other antimicrobial agents.

MDR Y. ruckeri has been reported worldwide. Y. ruckeri isolates (n = 116) from geographically separated areas of Turkey were tested for their susceptibility to oxytetracycline, streptomycin, florfenicol, ampicillin, trimethoprim-sulfamethoxazine, and oxolinic acid (29). Of the isolates tested, 16.3% were MDR. Although the authors tested for class 1 and 2 integrons, none were identified, and the presence of plasmids was not determined. Oxytetracycline resistance was mediated by tet(A) (36.3%) and tet(B) (4.5%). Other variants of the tet genes were not screened. Ampicillin resistance was identified in 29% of isolates, but all were negative for bla_TEM and bla_SHV. Other beta-lactamases were not tested, and the authors proposed that a low-level AmpC-type enzyme may have been expressed in some Y. ruckeri isolates, as was reported by Schiefer et al. (30).

To characterize antimicrobial-resistant outbreak strains, 10 representative Y. ruckeri isolates with reduced enrofloxacin and nalidixic acid susceptibility were screened for resistance determinants (28). The authors sequenced the QRDR of the isolates and tested for plasmid-borne resistance in one isolate which displayed reduced susceptibility to the potentiated sulfonamides. Isolates with low-level resistance to enrofloxacin (MICs of 0.06 to 0.25 μg/ml) and nalidixic acid (MICs from 8 to 32 μg/ml) possessed a single gyrA mutation in the QRDR. These changes represented an 8- to 32-fold increase in the enrofloxacin MIC and a 32- to 128-fold increase in the nalidixic acid MIC. This finding is in agreement with other studies in which a single gyrA mutation conferred resistance to quinolones but only a slight increase for fluoroquinolones, which require more than one mutation to achieve clinical resistance (31). In one isolate with elevated MICs to potentiated sulfonamides, the sul2, strB, and dfrA14 gene cassette was identified to be housed on an 8.9-kb transmissible plasmid. This study confirmed that Y. ruckeri is able to develop mutations resulting in reduced fluoroquinolone susceptibility and can also acquire plasmid-borne resistance genes (28).

Plasmid acquisition and genomic content plasticity are significant in antimicrobial resistance development for Y. ruckeri, Aeromonas spp., and other important fish pathogens. Welch et al. investigated the genomic similarities of three MDR IncA/C plasmids identified in the human pathogen Y. pestis (pIP1202), the foodborne pathogen S. Newport (pSN254), and the fish pathogen Y. ruckeri (pYR1) (24). Sequencing analysis showed that the three plasmid backbones shared an IncA/C backbone with an overall backbone sequence identity of 99 to 100%. Similar codon usage was observed in the plasmid backbones encoding functions such as type IV secretion systems and plasmid maintenance/replication, further strengthening the assertion that these plasmids evolved from a common ancestor. These plasmids contained antimicrobial gene cassettes within Tn21 or Tn10 transposons, with the Y. ruckeri plasmid containing only the Tn10 transposon. The Tn10 transposon encoded the strA/B and tet(A) genes in this plasmid, conferring streptomycin and tetracycline resistance, respectively. The Y. ruckeri plasmid also encoded dfrA1 conferring trimethoprim resistance and a sul2 gene conferring sulfonamide resistance, which is common to all three plasmids and is believed to be the selective factor in the evolution of these plasmids. The Y. ruckeri plasmid was the smallest of the three, and the S. Newport plasmid was the largest and encoded the most numerous antimicrobial resistance genes including three cephalosporin resistance genes (bla_SHV-1, and two copies of bla_CMY-2), tetracycline resistance genes [tet(A)], aminoglycoside resistance genes (two copies of aadA), streptomycin resistance genes (strA/B), chloramphenicol resistance genes (cat), florfenicol resistance genes (floR), and mercury resistance genes (merRTPABDE), to name a few. Salmonellae carrying similar MDR IncA/C plasmids were also identified in retail meats between 2002 and 2005 (24). This highlights the dissemination ability of isolates harboring these MDR plasmids in aquaculture and food animals and in human health.

More-recent studies have included comparisons of these IncA/C plasmids with plasmids from Y. ruckeri, P. damselae, and terrestrial pathogens. Call et al. investigated the relatedness of IncA/C plasmids in S. Newport (pAM04528 from a human) and E. coli (peH4H and pAR060302 from cattle) against the published plasmid sequences of pSN254 from S. Newport, the fish pathogen Y. ruckeri (pYR1) (24), and the fish pathogen P. damselae (pP91278 and pP99-108) (32). While all of the plasmids shared a similar IncA/C backbone, the plasmids originating from fish pathogens were evolutionarily the most dissimilar. The Y. ruckeri plasmid pYR1 was the most dissimilar with the smallest number of antimicrobial resistance genes. Based on bootstrap analysis of all of the plasmids, those from E. coli and S. Newport clustered together, the P. damselae plasmids clustered together, and the Y. ruckeri plasmid clustered separately from the others, indicating evolutionary distance. The authors proposed that although the plasmids evolved from a common ancestor, those sourced from terrestrial animals and humans diverged a considerable time ago compared to those sourced from fish.

In a similar study, Fricke et al. compared the sequences of historical and contemporary MDR IncA/C plasmids from aquatic and terrestrial environments to investigate the evolution of these plasmids via gene cassette insertions and single nucleotide polymorphisms (SNPs) in the core plasmid backbones (33). Sequences of pRAx (the cryptic plasmid), pRA1 (A. hydrophila), pYR1 (Y. ruckeri), pIP1202 (Y. pestis), pP91278 and pP99-018 (P. damselae subsp. piscicida), and pSN245 (S. Newport) were compared. The authors found that the oldest plasmid in the study, pRA1, and the more contemporary plasmids shared a highly conserved plasmid backbone, with 80% identity. In addition, the core plasmid is largely identical between pRA1 and pYR1 (Y. ruckeri), pIP1202 (Y. pestis), and pSN245 (S. Newport). The pRAx plasmid was identified after conjugative experiments transferring pRA1 from A. hydrophila to E. coli, and despite the significantly reduced plasmid size and reduced antimicrobial resistance gene carriage compared to pRA1, no SNP difference was identified between the two plasmids. Data indicated that these plasmids evolved primarily by gene cassette rearrangement rather than mutational change. This assertion is supported by the observation that pIP1202, isolated from Y. pestis in 1995 from Madagascar, and pP99-018, isolated in 1999 in Japan from P. damselae subsp. piscicida, do not vary by SNPs in the 100-kb plasmid backbone. Further, pP91278 which was isolated from P. damselae subsp. piscicida in 1991 in the United States differs by one SNP. Even though these plasmids have almost identical plasmid backbones, they carry very different resistance islands, varying from as few as two resistance genes in the most evolutionarily primary plasmids in A. hydrophila [tet(A) and sul2 in pRAx and pRA1] to nine resistance genes in the S. enterica plasmid pSN254 [bla_CMY-2, strA/B, aacA/C, tet(A), floR, sul1/2]. IncA/C plasmids have evolved through gene cassette rearrangement from a common ancestor, which likely originated from an aquatic environment and subsequently spread to terrestrial environments.

Motile Aeromonads

Mobile genetic elements have played a role in the dissemination of antimicrobial resistance in Aeromonas spp. In 2012, 50 Taiwanese A. hydrophila isolates from tilapia were screened for class 1, 2, and 3 integrons; resistance gene cassettes; mutations in the QRDR; and their susceptibility to 15 antimicrobial agents (34). Nineteen isolates were resistant to enrofloxacin and/or ciprofloxacin (MIC of 4 to 16 μg/ml) and had mutations in their gyrA and parC genes in the QRDR. All 50 isolates were resistant to at least one antimicrobial agent, and 94% were MDR. Forty-six percent were positive for class 1 integrons, with six resistance gene profiles identified. The most common resistance gene cassette identified in 26% of the class 1-positive isolates contained dfrA12 and aad2. This was one of the first studies to report both a class 1 integron and more than one resistance gene cassette carried on a class 1 integron in an Aeromonas sp. in finfish (16, 35). High percentages of resistance were found to streptomycin (92%), trimethoprim (88%), and sulfamethoxazole (62%). The identification of multiple gene cassettes on integrons underscores the risk of resistance coselection following repeated use of a single antimicrobial agent class.

Another investigation of class 1 integrons was perormed in 112 Aeromonas spp. isolates from aquacultured food fish, ornamental fish, shrimp, turtles, and amphibians in China (36). Isolates positive for the int1 gene, indicating the presence of a class 1 integron, were subsequently screened for resistance gene cassettes and other resistance genes. Of the strains tested, 19.6% were positive for int1, and all of these carried the sul1 and qacEΔ1 genes, indicating that these were structurally complete integrons. Many of the class 1 integron-positive strains were A. hydrophila and Aeromonas veronii isolated from aquacultured fish. Similar to other studies, common class 1 integron gene cassettes were found to contain the dfrA17 gene alone (20%) or dfrA12-orfF-aadA2 gene cassettes (80%). All of the class 1 integron-positive strains from aquacultured food fish had mutations in the QRDR genes gyrA and parC, and two were positive for tet(A). Phenotypic resistance profiles varied for the class 1 integron-positive strains; however, every strain was MDR and showed reduced susceptibility to at least ampicillin, sulfonamides, trimethoprim-sulfamethoxazole, nalidixic acid, tetracycline, or streptomycin.

Class 1 integrons have also been identified in Aeromonas spp. from ornamental carp, with 100% of isolates from one study containing the sul1 gene (37). Gene cassettes were identified on the class 1 integrons containing either the aadA2 gene or dfrA12-aadA2 genes, in agreement with other reports (34, 36).

As discussed earlier in this article for A. salmonicida and Yersinia spp., Aeromonas spp. have long been identified to have conjugative plasmids carrying varied antimicrobial resistance gene cassettes (33, 38). The first floR-containing plasmid in an Aeromonas sp. (pAB5S9) was discovered in Aeromonas bestarium (39). Researchers identified and performed conjugation experiments with pAB5S9, showing the presence of genes that conferred resistance to phenicols, streptomycin, tetracycline, and sulfonamides. Sequencing revealed that pAB5S9 carried the floR, sul2, strA-strB, and a tetR-tet(Y) resistance determinant. This was the first description of the combination tetR-tet(Y) determinant. Further, in the region of the plasmid containing the floR gene, an ISCR2 insertion element, the sul2 and strA-strB genes matched with 100% identity to the SXT region, an integrating and conjugative element which has been reported via a BLAST search to carry resistance genes for potentiated sulfonamides and streptomycins in Vibrio cholerae.

P. damselae subsp. piscicida

Historically, P. damselae subsp. piscicida, the causative agent of photobacteriosis, has commonly been reported to carry R plasmids. One of the first reports in 1988 analyzed 168 MDR plasmids taken from P. damselae subsp. piscicida isolates from Japanese yellowtail farms (38). Plasmids conferred resistance to chloramphenicol, tetracycline, ampicillin, and sulfonamides; however, the resistance determinants housed on these plasmids were not identified. Later, investigators screened MDR isolates for R plasmids and used DNA probes to characterize and compare the DNA structure of detected R plasmids (40). These R plasmids included an amp gene for ampicillin; tet(A), cat1, and cat2 for chloramphenicol; a kanamycin resistance gene postulated to be aphA7; floR; and sul. Resistance to kanamycin, sulfonamides, and tetracyclines was most often transferred by the R plasmids, and DNA probe hybridization revealed that there was significant homology between nine plasmids, indicating persistence of plasmids in P. damselae subsp. piscicida in these environments. A complex transposon-like structure consisting of insertion element 26 (IS26), tet(A), and kan was identified in P. damselae subsp. piscicida, which conferred resistance to tetracycline and kanamycin. This study showed that the ease of genetic rearrangement due to flanking IS26 elements on the transposon-like structure in the plasmid was likely the cause of the carriage of a kanamycin resistance gene postulated to be aphA7 on this plasmid (41). Full sequencing has been conducted on two MDR plasmids from P. damselae subsp. piscicida, pP99-018 (Japan) and pP91278 (United States). Both plasmids were categorized as belonging to a novel incompatibility group closely related to incompatibility group IncJ. Multiple mobilizable elements, including IS26, were identified on pP99-018 as well as catA1, aphA7, tet(A), tet(R), and sul2. The plasmid from the United States, pP91278, was 99% identical to pP99-018, from Japan, except that it carried fewer resistance elements including tet(A), sul2, and dfrA8 (42).

A novel MDR IncP-1 plasmid, pP9014, was identified in P. damselae subsp. piscicida isolates from cultured yellowtail (Seriola quinqueradiata) in Japan (43). This MDR plasmid carried tet, bla, and catA2 and is the first IncP-1 group plasmid identified from a fish pathogen. The authors performed comparative sequencing analysis of the pP9014 plasmid with other plasmid sequences in fish pathogens and in the National Center for Biotechnology Information database and identified no other IncP-1 plasmids in fish pathogens. This plasmid, however, is closely related to IncP-1 plasmids isolated from genera from wastewater treatment plants, marine biofilms, soil, and humans. The authors postulated that IncP-1 plasmids in aquaculture pathogens could be a vehicle for transmission of resistance genes within and between these environments.

Other Halophiles

Infections in fish caused by resistant V. anguillarum have been reported in the literature since the 1970s, including resistance to quinolones. In a study of 25 V. anguillarum strains obtained from cultured Japanese ayu (Plecoglossus altivelis), characterization of the gyrA, gyrB, parC, and parE gene mutations compared to reduced susceptibility to oxolinic acid was investigated (44). Of the 25 strains, 10 were oxolinic acid-susceptible, and for those with reduced susceptibility, MICs ranged from 0.006 μg/ml to 25 μg/ml. No gyrB or parE mutations were detected in any of the strains showing reduced susceptibility. In strains with MICs > 12.5 μg/ml, point mutations were identified in both gyrA (conferring a change in amino acid composition at position 83 from serine to isoleucine) and parC (conferring a change in amino acid composition at position 85 from serine to leucine). An intermediate interpretive category was defined by an MIC of 6.25 μg/ml and correlated only with a mutation in gyrA (conferring a change in amino acid composition at position 83 from serine to isoleucine).

In addition to chromosomal resistance determinant carriage, V. anguillarum strains were found to carry transmissible R plasmids containing tetracycline resistance genes. DNA-DNA hybridization analyses have differentiated the tetracycline resistance genes into four classes: A, B, C, and D. Aoki et al. analyzed tetracycline resistance determinants in R plasmids belonging to incompatibility types IncA/C and IncE in representative V. anguillarum isolates collected from 1973 to 1987 (8). The authors found that many strains of V. anguillarum obtained from 1973 to 1977 carried class B tetracycline resistance genes. However, a novel tetracycline resistance determinant, tet(G), which emerged in 1981, was identified on pJA8122 from V. anguillarum (8, 45). Interestingly, the authors also found that the DNA structure of the R plasmid changed in 1981 to a structure similar to that of pJA8122 and was different from R plasmids known at the time in other fish pathogens. Therefore, the origin of this new structure of plasmid similar to pJA8122 likely emerged in V. anguillarum with greater fitness over previous plasmids from sources not associated with other common fish pathogens (8).

Gliding Pathogens (Flavobacteria)

Flavobacterium psychrophilum causes bacterial cold water disease and rainbow trout fry syndrome in salmonids, ayu, and other species of finfish, resulting in major economic losses. Quinolone resistance has been reported in strains recovered from diseased fish. Izumi et al. characterized 244 isolates of F. psychrophilum from diseased fish using a genotyping method utilizing PCR restriction fragment length polymorphism and analyzing the gyrA gene to identify mutations in the QRDR (46). They amplified the gyrA gene and digested the product with restriction enzyme Mph1103I, resulting in the identification of quinolone-resistant isolates possessing an alanine residue at the GyrA 83 position. Overall, 62.7% of F. psychrophilum isolates were quinolone resistant and possessed the GyrA 83 mutation. Ayu isolates were found to have the same mutation at a rate of 82.5%.

Tetracycline resistance in F. psychrophilum has been noted as a result of six nucleotide polymorphisms in the 16s rRNA via 16s rDNA microarray experiments. Soule et al. reported that two genetic lineages of F. psychrophilum have been identified in diseased rainbow trout as defined by restriction fragment length polymorphism and microarray analysis, and tetracycline resistance is associated with one lineage (47).

A study to assess the variability in resistance genotypes of 25 F. psychrophilum isolates from Spain was conducted via plasmid profiling (48). Twenty isolates possessed 3.5-kb (n = 13) or 5.5-kb (n = 7) plasmids, but plasmid-mediated resistance was not found to be correlated with resistance to oxytetracycline or florfenicol, suggesting a chromosomal location for the source of the resistance genotype. Reduced susceptibility to oxytetracycline was identified in 80% of isolates, with MICs ranging from 2.4 to 9.7 μg/ml, and all isolates were susceptible to florfenicol.

Streptococcaceae

Streptococcus iniae, Streptococcus parauberis, and Lactococcus garvieae cause streptococcosis in a number of fish species, including olive flounder. Farmed olive flounder are common in Korea and can suffer heavy losses due to streptococcosis. This disease is usually treated with tetracycline and erythromycin. Park et al. investigated tetracycline and erythromycin susceptibility and corresponding resistance genes in S. iniae and S. parauberis isolates from Korean fish farms (49). The authors reported pan-susceptibility in S. iniae; however, a large number of the S. parauberis isolates were resistant to tetracycline (63%) and erythromycin (58%). PCR screening for resistance genes was conducted, and tet(M), tet(O), and tet(S) genes were found in most tetracycline-resistant isolates. Fourteen erythromycin-resistant isolates were positive for the ermB gene. The presence of both tet(S) and ermB genes was responsible for high-level resistance to both erythromycin and tetracycline in S. parauberis in olive flounder.

S. agalactiae (group B) is associated with high mortality of cultured tilapia worldwide. Dangwetngam et al. reported the antimicrobial susceptibilities of 144 S. agalactiae strains isolated from infected tilapia cultured in Thailand and screened oxytetracycline-resistant isolates for tetracycline resistance genes (50). Phenotypic resistance was observed against oxolinic acid, gentamicin, sulfamethoxazole, trimethoprim, and oxytetracycline. As determined by PCR and sequencing, 17 isolates were resistant to oxytetracycline and were positive for the tet(M) gene. All of these isolates were also resistant to trimethoprim, oxolinic acid, gentamicin, sulfamethoxazole, and oxytetracycline; however, investigation of any associations with mobile elements was not conducted.

Other Important Pathogens of Fish

Piscirickettsia salmonis causes piscirickettsiosis in a variety of species of salmonids, causing notable economic losses in Chile. Although oxytetracycline and/or florfenicol have been used to control P. salmonis for many years, few reports have characterized the antimicrobial resistance profiles of this key Chilean aquaculture pathogen. Cartes et al. conducted whole-genome sequencing of three Chilean P. salmonis strains to correlate oxytetracycline and florfenicol resistance genes with reduced susceptibility (51). After bioinformatic comparative genomic analysis, the authors identified resistance genes encoding transmembrane domains belonging to transporters and efflux pumps such as the major facilitator superfamily, which contributed to tetracycline resistance and concurrent tetracycline and florfenicol resistance. All antimicrobial resistance genes found were observed in the two field strains. One putative tetracycline resistance gene was found on a pathogenicity island, while another was postulated to be located on a transposon, due to the proximity of transposases detected. These genes have been identified in other fish pathogens (Photobacterium, Vibrio, Pseudomonas spp., and others). Membrane permeability, as studied by accumulated intracellular ethidium bromide after exposure to subinhibitory florfenicol concentrations, revealed that most of the accumulation occurred in the more susceptible strain. This demonstrated that the membrane transporters and pumps identified are capable of pumping out multiple classes of compounds and may be responsible for the MDR phenotype (51). This comprehensive study investigated the mechanism for oxytetracycline and florfenicol resistance, which may be transmissible or cohoused with virulence genes on pathogenicity islands.

Aliivibrio salmonicida, formerly known as Vibrio salmonicida, causes a bacterial septicemia called cold water vibriosis in farmed salmon and cod. Due to the long-term use of antimicrobial agents to combat this disease, resistance to multiple antimicrobial agents has been reported, in many cases mediated by transmissible mobile elements. Resistance plasmids in A. salmonicida have been reported to be associated with three separate transposon gene cassettes and resistance to trimethoprim, sulfonamides, and tetracyclines. Trimethoprim resistance has been associated with the dfrA1 gene cassette contained in a class II transposon, which does not contain other drug resistance genes. Tetracycline resistance was associated with the tet(E) gene and sulfonamide resistance with the sul2 gene. Both genes were located between two identical insertion elements (11).

While antimicrobial resistance has been identified for many of these key fish pathogens, at the writing of this arcticle, reports characterizing genetic determinants conferring resistance have not been identified for Tenacibaculum maritimum, Flavobacterium columnare, or Francisella noatunensis subsp. orientalis. One study described the complete genome sequencing of F. noatunensis subsp. orientalis, in which a putative resistance island was identified; however, further characterization of the underlying antimicrobial resistance genes and their correlation with phenotypes were not reported (52).

Rhodes et al. (53) investigated macrolide susceptibility and mechanisms of resistance in the bacterial kidney disease pathogen Renibacterium salmoninarum Due to the practice of using erythromycin or azithromycin in female brood stock to prevent vertical transmission of R. salmoninarum, antimicrobial resistance has been postulated to be a risk. The authors screened isolates from fish treated multiple times for their susceptibility to erythromycin and azithromycin. Four isolates displayed higher MICs than the R. salmoninarum control strain, ATCC 33209, with MICs for erythromycin ranging between 0.125 and 0.25 μg/ml and MICs for azithromycin ranging between 0.016 and 0.03 μg/ml. Sequence analyses of both the 23S rDNA gene and the open reading frames of ribosomal proteins L4 and L22 revealed identical sequences among all isolates. This finding indicates that the phenotype was not a result of mutations in the drug-binding site of 23S rRNA, and the mechanism of resistance was not identified. This study represents the first report of R. salmoninarum isolates from fish that have received multiple macrolide treatments and exhibit reduced susceptibility.

ANTIMICROBIAL SUSCEPTIBILITY TESTING OF FISH PATHOGENS

Historical Perspective

When adjustments to aquatic animal husbandry approaches fail, are economically impractical, or are unlikely to resolve a disease outbreak, veterinarians and farmers typically turn to chemotherapy. Selection of an appropriate treatment, in general, should be made by a veterinarian after considering the anticipated pharmacokinetics of the antimicrobial agent in the target population under the given conditions (i.e., water temperature, salinity, hardness), as well as which antimicrobial agent has shown prior success against the etiological agent. When such information is unavailable or when additional reassurance is desired, performance of a well-controlled, standardized antimicrobial susceptibility test (AST) (i.e., disk diffusion, broth microdilution) can be used as a surrogate. When AST data are used correctly in combination with validated clinical breakpoints (susceptible, intermediate, resistant), these data have a high likelihood of accurately predicting clinical outcome, especially if the normal dosage regimen recommended for the indicated disease is used (54). To date, the only fish pathogen with attributable and internationally accepted clinical breakpoints is A. salmonicida (55). In the absence of clinical breakpoints, AST data can be interpreted using epidemiological cutoff values (ECOFFs or ECVs; wild type, non-wild type) to determine if an isolate harbors antimicrobial resistance genes or resistance-mediating mutations. In Salmonella, the presence of resistance genes has been shown to be highly correlated (99%) with an increase in MIC (56). Although ECOFFs are most useful for the phenotypic detection of resistance mechanisms, veterinarians may be forced to select a therapy using available ECOFFs and/or susceptibility distributions from the published literature. However, effort should be made to select a therapy based on susceptibility results in conjunction with empirical evidence of effectiveness and knowledge of the pharmacokinetics and pharmacodynamics of the antimicrobial agent(s). At present, only A. salmonicida has published ECOFFs (55, 57), but data analyses are forthcoming for other key fish pathogens. In January 2017, the CLSI’s Subcommittee on Veterinary Antimicrobial Susceptibility Testing approved MIC ECOFFs for the key fish pathogen F. psychrophilum. These will be published in the next edition of the CLSI’s VET03/VET04-S2 (55).

Importance of Understanding the Scope of an AST Testing Program

When conducting investigations of the AST of fish pathogens, it is important for workers to understand the scope and goals of their work. If a project intends to monitor for clinically relevant resistance, then clinical breakpoints should be used when available. If the surveillance and detection of emerging low-level resistance is the goal, then ECOFFs should be used. If one of the program’s goals is to provide meaningful susceptibility comparisons across borders and even worldwide, then serious consideration should be given to whether clinical breakpoints should be used. This is because clinical breakpoints are established based partly on pharmacokinetics and pharmacodynamics using the normal dosage regimen (58), which may differ greatly by country, thus altering the practical definitions of susceptible, intermediate, and resistant, and ultimately the clinical usefulness of the clinical breakpoints. Also, in most cases use of clinical breakpoints as the sole interpretive criteria in an antimicrobial resistance surveillance program of bacterial pathogens of fish or any other animal species may not be as sensitive in the detection of emerging resistance. This is because clinical breakpoints are often one to three 2-fold dilutions higher than the corresponding ECOFFs (Fig. 1).

FIGURE 1.

Distribution of MICs and categorization by clinical breakpoints contrasted to ECOFFs.

Standard AST Methods

Prior to 2001, aquatic animal disease researchers commonly used AST methods and clinical breakpoints developed in their own laboratories. The use of different methods and breakpoint values limited accurate interlaboratory comparisons and correlation to clinical cases. The magnitude of the issue was brought to light by Smith (59) when he surveyed 32 fish health laboratories in 18 countries and found that 25 laboratories used predominantly disk diffusion testing procedures and only 4 used the standard CLSI procedure found in the original M42-A guideline published in 2003 and again in 2006 as VET03-A (60). Alarmingly, 22 of the 25 laboratories reported that for any given antimicrobial agent they employed the same clinical breakpoints for all nonfastidious pathogens. Due to outreach efforts over the past decade by Smith and other members, advisors, and observers of the CLSI’s Subcommittee on Veterinary Antimicrobial Susceptibility Testing, the authors have confidence that the percentage accordance with standardized CLSI methods has risen significantly. However, it is clear that additional educational opportunities are warranted on what constitutes use of a standardized AST method and quality control, data interpretation, clinical breakpoints, and ECOFFs.

Currently, there are two fish pathogen-specific CLSI guidelines: one for disk diffusion AST (60) and an updated guideline for broth microdilution AST (61). These guidelines provide standardized methods and quality control procedures that allow internationally harmonized testing of nonfastidious aquaculture pathogens (group 1 organisms) on unsupplemented Mueller-Hinton media, and F. columnare and F. psychrophilum in diluted cation-adjusted Mueller-Hinton broth (Table 2). For other fastidious species pathogenic for fish, recommendations are provided to develop laboratory-specific methods until a method can be standardized.

TABLE 2.

Antimicrobial susceptibility testing methods of aquatic bacterial pathogens^a

Organism	Mediumc	Incubation
Group 1: nonfastidious bacteria
Enterobacteriaceae	MHA	22°C (24–28 h and/or 44–48 h or 28°C (24–28 h) or 35°C (16–18 h)
Aeromonas salmonicida (nonpsychrophilic strains)	CAMHB	22°C (24–28 h and/or 44–48 h or 28°C (24–28 h) or 35°C (16–20 h)
Aeromonas hydrophila and other mesophilic aeromonads
Pseudomonas spp.
Plesiomonas shigelloides
Shewanella spp.
Group 3: gliding bacteriab
Flavobacterium columnare	Diluted CAMHB (4 g/liter)	28°C (44–48 h)
Flavobacterium psychrophilum	Diluted CAMHB (4 g/liter)	18°C (92–96 h)
Group 2: strictly halophilic Vibrionaceae
Vibrio vulnificus, V. parahaemolyticus, V. alginolyticus, V. anguillarum, V. fischeri, V. metschnikovii, V. ichthyoenteri	CAMHB + NaCl (1%)	22°C (24–28 h and/or 44–48 h) or 28°C (24–28 h and/or 44–48 h)
	MHA + NaCl (1%)	22°C (24–28 h and/or 44–48 h) or 28°C (24–28 h and/or 44–48 h)
Group 4: Streptococci
Lactococcus spp., Vagococcus salmoninarum	MHA + 5% sheep blood	22°C (44–48 hours) + CO2
	CAMHB + LHB (2.5–5% vol/vol)	22°C (44–48 h) + CO2
Streptococcus spp., Carnobacterium maltaromaticum, and other streptococci	MHA + 5% sheep blood	28°C (24–28 h and/or 44–48 h) + CO2
	CAMHB + LHB (2.5% to 5% v/v)	28°C (24–28 h and/or 44–48 h) + CO2
Group 5: other fastidious bacteria
Atypical Aeromonas salmonicida	CAMHB	15°C (44–48 h)
	MHA	15°C (44–48 h)
Aliivibrio salmonicida (formerly V. salmonicida) and Moritella viscosa	MHA + NaCl (1.5%)	4°C (A. salmonicida) or 15°C (M. viscosa) (6 days)
Tenacibaculum maritimum	FMM or diluted MHB (1:7) + inorganic ion supplementation	25°C (44–48 h)
	FMM or diluted MHA (1:7) + inorganic ion supplementation	25°C (44–48 h)
Francisella spp.	CAMHB + IsoValitex (2%) + 0.1% glucose	28°C (44–48 h)
Renibacterium salmoninarum	KDM	15°C (with agitation for 96 h)
Piscirickettsia salmonis	AUSTRAL-SRS	18°C (92–96 h and/or 116–120 h)

^a Bold text indicates that the method has been standardized and quality control guidelines are available (60, 61).

^b Disk diffusion testing has proven difficult. Some modification may be necessary for testing F. branchiophilum, which may include cations, horse or fetal calf serum, or NaCl.

^c CAMHB, cation-adjusted Mueller-Hinton broth; FMM, Flexibacter maritimum medium; KDM, kidney disease medium; LHB, lysed horse blood; MHA, Mueller-Hinton agar.

Non-fastidious bacterial pathogens

Nonfastidious bacterial pathogens of fish (termed group 1 by the CLSI) are those that grow adequately in unsupplemented Mueller-Hinton media at 22°C for 24 to 28 hours and/or 44 to 48 hours, 28°C for 24 to 28 hours, or 35°C for 16 to 20 hours (16 to 18 hours for disk diffusion). These organisms typically include members of the Enterobacteriaceae, some members of the Vibrionaceae, A. salmonicida, A. hydrophila and other mesophilic aeromonads, Pseudomonas spp., and Plesiomonas shigelloides. These organisms should be tested using CLSI guidelines (60–62) with quality control.

Investigations of A. salmonicida using standardized AST methods far outnumber those of any other fish pathogen. Miller and Reimschuessel published oxytetracycline, ormetoprim-sulfadimethoxine, oxolinic acid, and florfenicol MIC and zone diameter distributions for 217 A. salmonicida isolates (57). They used the data to establish the first consensus-approved, CLSI-endorsed ECOFFs for a fish pathogen (Table 3). Later, the distributions were used in conjunction with available pharmacokinetics, pharmacodynamics, and clinical effectiveness data to establish MIC and zone diameter clinical breakpoints for oxytetracycline and oxolinic acid (55). Douglas et al. determined trimethoprim-sulfamethoxazole, trimethoprim, and sulfamethoxazole zone diameters for 106 A. salmonicida isolates (63). Their analysis showed that testing trimethoprim-sulfamethoxazole together was unreliable for predicting a susceptibility phenotype for trimethoprim and sulfamethoxazole if tested alone. In a similar study by the same group of investigators, Ruane et al. determined oxolinic acid, flumequine, and enrofloxacin zone diameters for 106 A. salmonicida isolates (64). They proposed laboratory-specific ECOFFs for each species and argued that oxolinic acid is the best quinolone representative to monitor for the presence of resistance mechanisms. Other A. salmonicida susceptibility testing studies employed similar testing methods but lacked available quality control parameters at the time, thus making direct comparisons with current data more difficult (65, 66).

TABLE 3.

CLSI-approved MIC and zone diameter CBP and ECOFFs for A. salmonicida (55)^a

Antimicrobial agent	MIC CBP			MIC ECOFF	Zone diameter CBP			Zone diameter ECOFF
	S	I	R	WT	S	I	R	WT
Gentamicin								≥18
Florfenicol				≤4				≥27
Oxytetracycline	≤1	2-4	≥8	≤1	≥28	22–27	≤21	≥28
Oxolinic acid	≤0.12	0.25-0.5	≥1	≤0.12	≥30	25–29	≤24	≥30
Erythromycin								≥14
Ormetoprim-sulfadimethoxine				≤0.5/9.5				≥20
Trimethoprim-sulfamethoxazole				b				≥20

^a S, susceptible; I, intermediate; R, resistant; WT, wild type.

^b MIC ECOFF for ormetoprim-sulfadimethoxine can be used for interpretations of this antimicrobial.

Čížek et al. evaluated the antimicrobial susceptibility of Aeromonas spp. from ornamental (koi) carp and common carp in the Czech Republic to investigate differences in resistance patterns resulting from different on-farm usage practices (37). While it is permitted in the Czech Republic to treat common carp with oxytetracycline, the regulation of antimicrobial use is not enforced on ornamental carp production, resulting in more widespread use. Significant differences in reduced susceptibility were identified in isolates from ornamental carp versus common carp, and 60% of isolates were resistant to one or more antimicrobial agents. Ornamental carp isolates were significantly more resistant to more than one antimicrobial agent and more resistant to ciprofloxacin than those from common carp. Resistance to oxytetracycline was detected in 50% of aeromonads from ornamental carp and slightly less (41%) in common carp.

Despite E. ictaluri being a major pathogen of catfish, few susceptibility investigations using standard or nonstandard methods are available. Hawke et al. used standard broth microdilution testing methods for eight E. ictaluri isolates from laboratory-held zebrafish and reported all as lacking resistance mechanisms to florfenicol, enrofloxacin, and trimethoprim-sulfamethoxazole (67). Dung et al. used nonstandard testing conditions to measure the MICs of 15 antimicrobial agents for 64 E. ictaluri isolates from Vietnam (68). Study-specific ECOFFs for streptomycin, oxolinic acid, flumequine, oxytetracycline, and trimethoprim were calculated using the current “gold-standard” statistical method published by Turnidge and Paterson (69). All isolates were highly susceptible to florfenicol, with MICs ≤ 0.25 μg/ml. McGinnis et al. (70) noted similar findings for florfenicol, using nonstandard conditions. Dung et al. were the first to report widespread resistance emergence to streptomycin, with 82.8% of isolates identified as non-wild type (68).

E. tarda is primarily isolated from diseased eels and flounder. In a Taiwanese study, 94 isolates were tested at 25°C for 18 hours in broth microdilution panels (26). Susceptibility to five classes of antimicrobial agents was determined, with decreased susceptibility to potentiated sulfonamides and tetracyclines reported. It is important to note that these investigators stated that the methods used were in accordance with CLSI guidelines; however, they used a nonstandard incubation temperature of 25°C, and the use of human-specific interpretive categories are only reliable for data generated at 35°C and when the data are generated for human medical application.

Enteric redmouth disease, caused by Y. ruckeri, is an important disease in rainbow trout aquaculture. Huang et al. used standardized broth microdilution testing to obtain MICs of 24 antimicrobial agents for 82 clinical isolates from Germany (28). Using wide test concentration ranges and quality control parameters when possible, this study accurately captured the current antibiogram for Y. ruckeri. Researchers noted that measurements after 24 hours and 48 hours often revealed an increase in calculated MIC₅₀ and MIC₉₀ values with time. Due to the adequate growth of Y. ruckeri at 22°C after 24 to 28 hours, investigators recommended this incubation time for future research. For most of the antimicrobial agents tested, a unimodal distribution of MICs was observed. All isolates that showed low enrofloxacin MICs also showed low nalidixic acid MICs. Seventy-two isolates had elevated nalidixic acid and enrofloxacin MICs of 0.06 to 0.25 μg/ml and 8 to 64 μg/ml, respectively. Only one isolate had elevated MICs for sulfamethoxazole (≥1,024 μg/ml) and for trimethoprim-sulfamethoxazole (≥32/608 μg/ml). Three isolates had colistin MICs of ≥32 μg/ml. It is envisaged that these data will be combined with others derived from the same standardized methodology to establish ECOFFs that can be useful to detect emerging resistance concerns in this economically important fish pathogen. Balta et al. obtained zone diameters for 116 Y. ruckeri isolates from infected rainbow trout in commercial farms in Turkey (29). The authors stated that CLSI methods, quality control, and clinical breakpoints were used, but it is unclear if this is, in fact, the case since detailed methods were not provided and no clinical breakpoints are available for Y. ruckeri at this time. The authors did find the presence of resistance genes for oxytetracycline in 35.3% of isolates, ampicillin in 29%, oxolinic acid in 11.1%, streptomycin in 10.2%, sulfamethoxazole in 9.4%, trimethoprim-sulfamethoxazole in 9.4%, florfenicol in 4.2%, and enrofloxacin in 2.3% of the isolates.

At least one member of the family Vibrionaceae (V. cholerae) does not require NaCl supplementation for in vitro susceptibility testing, but others such as V. parahaemolyticus, V. alginolyticus, V. vulnificus, V. anguillarum, V. fischeri, V. metschnikovii, and V. ichthyoenteri have been shown to need NaCl at least in agar media (71, 72). The methods, quality control parameters, and interpretive categories described in the CLSI’s M45-A3 guideline (62) should be used for tests of V. cholerae, but additional investigation is clearly warranted for tests of other vibrios to ensure accurate phenotypic characterization.

Motile aeromonad septicemia caused by mesophilic motile Aeromonas species is probably the most common bacterial disease in freshwater aquarium fish (73). A. hydrophila has often been the primary implicated pathogen, but others including A. veronii and Aeromonas caviae have been shown to play important roles. Jagoda et al. used disk diffusion methods described in CLSI M31-A3 (74) to test 53 motile aeromonads at 35°C and found resistance to tetracycline (58.5%), erythromycin (54.7%), trimethoprim-sulfamethoxazole (26.4%), nitrofurantoin (22.6%), neomycin (9.4%), chloramphenicol (7.5%), and enrofloxacin (7.5%) (75). A total of 49% of the isolates were found to be resistant to more than one antimicrobial agent class. Deng et al. tested 112 motile Aeromonas spp. by disk diffusion against 14 antimicrobial agents (36). Tests were conducted under standard conditions at 35°C. Although not all antimicrobial agents had interpretive criteria available, the authors noted that 49.1% of isolates were MDR. Most isolates (≥90%) were susceptible to cefotaxime, amikacin, ciprofloxacin, norfloxacin, and doxycycline. Interestingly, A. hydrophila and A. veronii displayed greater levels of resistance compared to A. caviae, Aeromonas sobria, Aeromonas dhakensis, Aeromonas jandaei, Aeromonas trota, and Aeromonas media. Čížek et al. conducted nonstandardized agar dilution testing at 28°C with 72 motile Aeromonas spp. isolated from common carp and koi carp (37). The authors noted that they cautiously used clinical breakpoints appropriate for tests conducted at 35°C (62). Resistance rates of 50% to oxytetracycline, 15% to ciprofloxacin, and 7% to chloramphenicol were noted. A significantly higher incidence of class 1 integrons and ciprofloxacin resistance in motile aeromonads from koi carp was ascribed to a distinct Czech policy on antimicrobial use in ornamental fish farms (koi) versus in traditional aquaculture (common carp). Antimicrobial resistance linked to class 1 integrons has also been shown by other researchers (34).

Gliding pathogens (group 3)

Fish health specialists have a standard broth microdilution method for the nutritionally fastidious aquatic gliding bacteria family, Flavobacteriaceae (61, 76). Recently, ECOFFs have been proposed for F. columnare (77) (Table 4) and F. psychrophilum (78, 79) (Table 5). Gieseker et al. proposed provisional ECOFFs for ampicillin, enrofloxacin, erythromycin, florfenicol, flumequine, oxolinic acid, and oxytetracycline using MIC distributions of 120 geographically diverse F. columnare isolates (77). ECOFFs were calculated using the ECOFFinder method (69). These ECOFFs combined with existing pharmacokinetic and clinical data may be used in the future to establish clinical breakpoints. Van Vliet et al. recently published 10 MIC distributions for 50 Michigan F. psychrophilum isolates (79). Their data revealed a unimodal, wild-type distribution for all antimicrobial agents tested except oxytetracycline, for which there were some non-wild-type isolates. Using normalized resistance interpretation and the ECOFFinder method to calculate an ECOFF, these researchers proposed provisional ECOFFs that largely agree with those presented by Smith (78) in 2017 and were approved by the CLSI Subcommittee on Veterinary Antimicrobial Susceptibility Testing (Table 5). Earlier, using nonstandard agar dilution and disk diffusion methods to generate MIC and zone diameter data distributions, Miranda et al. calculated ECOFFs using normalized resistance interpretation for amoxicillin, florfenicol, oxytetracycline, and three quinolones (80). These data from 125 F. psychrophilum isolates from farm-raised salmonids in Chile revealed that agar dilution testing was more precise than disk diffusion at determining a non-wild-type phenotype for all antimicrobial agents tested. The authors concluded that the disk diffusion testing method is not suitable for investigating the susceptibility of F. psychrophilum isolates. Henríquez-Núñez et al. came to a similar conclusion when they used nonstandard broth microdilution and disk diffusion methods on 40 F. psychrophilum isolates from Chilean salmonid farms (81). They determined susceptibility to oxytetracycline, florfenicol, and oxolinic acid and noted very high levels of resistance to all three antimicrobial agents, and again found that the disk diffusion testing method was difficult with this species.

TABLE 4.

Provisional MIC ECOFFs for F. columnare (77)

Antimicrobial agent	MIC ECOFF (ECOFFinder)
	Wild type
Ampicillin	≤0.25
Florfenicol	≤4
Oxytetracycline	≤0.12
Oxolinic acid	≤0.25
Flumequine	≤0.12
Enrofloxacin	≤0.03
Erythromycin	≤4

TABLE 5.

MIC ECOFFs and zone diameter ECOFFs for F. psychrophilum (78–80)

Antimicrobial agent	MIC ECOFF (ECOFFinder + NRI)a	MIC ECOFF (ECOFFinder)	MIC ECOFF (NRI)	Zone diameter ECOFF
	Wild type	Wild type	Wild type	Wild type
Amoxicillin				≥44
Florfenicol	≤2	≤2	≤2	≥46
Oxytetracycline	≤0.12	≤0.06	≤0.12	≥47
Oxolinic acid	≤0.25	≤0.25	≤0.25	≥29
Flumequine	≤0.12	≤0.12	≤0.12	≥47
Enrofloxacin	≤0.03	≤0.015	≤0.03	≥49
Erythromycin	≤8	≤8	≤8
Ormetoprim-sulfadimethoxine		≤0.5/9.5	≤1/19
Trimethoprim-sulfamethoxazole		≤0.12/2.38	≤0.5/9.5

^a Accepted by the CLSI Subcommittee for Veterinary Antimicrobial Susceptibility Testing in January 2017.

Nonstandard, Recommended AST Methods

Strictly halophilic Vibrionaceae (group 2)

As mentioned earlier, some noncholera vibrios should be tested in vitro with NaCl present. Israil et al. used agar dilution testing to show that resistant phenotypes were observed when tests were conducted in Mueller-Hinton agar supplemented with 1%, 2%, and 3% NaCl, but not in unsupplemented media (71). Until a standardized susceptibility testing method can be developed for noncholera vibrios, the methods and quality control parameters described in the CLSI’s M45-A3 guideline (62) should be employed, but with caution. Tang et al. tested fluoroquinolone activity against 46 human clinical V. vulnificus isolates by agar dilution in unsupplemented Mueller-Hinton agar (82). They found a lack of decreased susceptibility in all isolates to all fluoroquinolones. Similarly, Hsueh et al. investigated 19 V. vulnificus isolates from Taiwanese patients and found them to be pan-susceptible to 17 antimicrobial agents tested and resistant only to clindamycin (83). All isolates were tested on unsupplemented media. A recent Iranian investigation of two P. damselae subsp. damselae isolates tested on unsupplemented Mueller-Hinton agar showed zone diameters indicative of susceptibility to ciprofloxacin, chloramphenicol, and nalidixic acid (84). Zones suggesting a clinically resistant phenotype were noted for ampicillin, amoxicillin, gentamicin, streptomycin, erythromycin, clindamycin, and novobiocin. Clearly, additional research is needed to determine not only whether NaCl is required for testing all halophilic vibrios, but also if an incubation temperature lower than 35°C may be required for some fish-pathogenic vibrios.

Streptococcaceae (group 4)

Most streptococcal fish pathogens can likely be grown and tested under standard testing conditions as described in the CLSI’s VET01 standard (58). The recommended agar medium for testing streptococci is Mueller-Hinton agar supplemented with 5% defibrinated sheep blood. Incubation conditions are at 35°C ± 2°C in an atmosphere of 5% ± 2% CO₂ for 20 to 24 hours. For fish pathogens that cannot grow satisfactorily under these conditions, adjustments may be necessary, including longer incubation times and probably lower incubation temperatures. Broth microdilution testing should be performed using cation-adjusted Mueller-Hinton broth supplemented with 2.5 to 5% lysed horse blood. Incubation conditions are 35°C for 20 to 24 hours without CO₂. Some streptococcal fish pathogens likely require lower incubation temperatures (22°C for Lactococcus spp. and Vagococcus salmoninarum; 28°C for some Streptococcus spp. and Carnobacterium maltaromaticum) and CO₂ supplementation (55).

Other fastidious bacterial pathogens (group 5)

Those organisms that do not grow in standardized Mueller-Hinton media or under standardized conditions have been classified by the CLSI as group 5 organisms (55). These included atypical A. salmonicida, psychrophilic members of Vibrionaceae, T. maritimum, Francisella spp., R. salmoninarum, and P. salmonis.

Atypical A. salmonicida isolates have optimal temperatures in vitro of <22°C but can be tested in cation-adjusted Mueller-Hinton broth. Disk diffusion results of these isolates should be read with caution because extended incubation times may be required and drug stability can be an issue.

Some psychrophilic vibrios including Moritella viscosa and A. salmonicida (formerly V. salmonicida) may have optimal temperatures in vitro of between 4 and 15°C. At these temperatures extended incubation times will almost certainly be required, and drug stability can be an issue. Coyne et al. conducted agar dilution and disk diffusion susceptibility testing on clinical M. viscosa isolates in Mueller-Hinton media at 15°C and used a 6-day incubation period (85).

Similar to the group 3 organisms, T. maritimum is a gliding bacterium and a member of the class Flavobacteriaceae. However, this species has been shown to require a specialized Flexibacter maritimum medium prepared with seawater (86) and an in vitro growth temperature of 24°C (87). Zone diameter and MIC determinations on this media for seven antimicrobial agents appeared to provide reproducible results when conducted in duplicate. However, the lack of a source of commercial grade seawater has precluded the standardization of a method for this organism.

F. noatunensis subsp. orientalis is an emerging warmwater fish pathogen and the causative agent of francisellosis in tilapia. Soto and colleagues are close to standardizing a broth microdilution testing method for this important species (88). Testing at 28°C for 48 hours in cation-adjusted Mueller-Hinton broth supplemented with 2% IsoVitalex and 0.1% glucose tested at a pH of 6.4 ±1 provided optimal growth and reproducible results. The authors noted that the lower pH did affect the inhibitory properties of the potentiated sulfonamides.

The Gram-positive fish pathogen R. salmoninarum is the causative agent of bacterial kidney disease. R. salmoninarum can be transmitted horizontally as well as vertically, and fish can be susceptible to disease throughout their life cycle. For veterinarians, the windows of opportunity for chemotherapy can be limited, and macrolides are typically the most common choice. Rhodes et al. used a specialized medium (kidney disease medium) to conduct broth micro- and macro-dilution tests at 15°C with 150 rpm agitation for 4 days (measurements every 24 hours) (53). The authors found that broth microdilution did not provide optimal growth conditions due to a generation time twice that calculated in the broth macrodilution tests. Reduced susceptibility to erythromycin and azithromycin was noted in isolates obtained from fish that received multiple macrolide treatments.

Piscirickettsiosis caused by P. salmonis is by far the most damaging infectious disease in the Chilean salmon industry. The recommended AST conditions for this pathogen can be found in CLSI VET03/VET04-S2 (55) and are based on the research of Yañez et al. (89). AUSTRAL-SRS medium was found to perform well when inoculated and incubated at 18°C for 92 to 96 hours and/or 116 to 120 hours. More recently, Henríquez et al. conducted a large-scale field study and obtained susceptibility profiles of 292 P. salmonis isolates (90). They used a new medium called ADL-PSB to conduct broth microdilution tests at 19°C ± 2°C for 5 to 7 days. The authors analyzed the MIC data using a method similar to the ECOFFinder method (69) to establish laboratory-specific epidemiological cutoff values for oxolinic acid, flumequine, and oxytetracycline.

FUTURE RESEARCH

Whether it is routine culture and susceptibility testing in a diagnostic laboratory or epidemiological surveillance of antimicrobial resistance, incorporation of a standard AST method when available should always be the priority. In the absence of a standard method, researchers can optimize their testing methods and eventually lead a multilaboratory method standardization trial (74). Alternatively, laboratory-specific methods, quality control, and interpretive criteria can be developed. Although this route precludes a lab’s data from being internationally harmonized, it may be appropriate in some situations.

Aquatic animal disease diagnosticians need additional standard testing methods, as well as internationally harmonized criteria for interpreting zones of inhibition and MICs. Clinical breakpoints are notoriously difficult to establish for fish for a multitude of reasons. Pharmacokinetic and pharmacodynamic data are very difficult to ascertain because antimicrobial exposure data during the dosing interval can be extremely challenging. Clinical effectiveness data are difficult to interpret due to on-farm predation and cannibalism. Lastly, available susceptibility data in large numbers is traditionally hard to locate because of the lack of a centralized database and the general lack of standardized methods until recently and because there are not standardized, reproducible methods for all fish pathogens. To compensate for the lack of clinical breakpoints, laboratories should instead use ECOFFs to determine whether their isolate harbors resistance genes that may confer clinically relevant resistance.