Plasmid mediated antibiotic resistance in aquaculture: pRAS2
Plasmid mediated antibiotic resistance in aquaculture travels through conjugation between bacteria, not through fish that move. pRAS2, a conjugative plasmid found in Aeromonas salmonicida, encodes a complete type IV secretion system and uses it to mobilize a second, non conjugative plasmid, pRAS3, into new host cells, while pRAS2 itself stays behind in the donor. The event is too rare to emerge in routine culture, and it needs no antibiotic exposure at all to occur.
Why a single 1992 fish isolate still matters
Antibiotics are used at scale in aquaculture to treat bacterial infection, and that use is one of the strongest forces pushing antimicrobial resistance (AMR) upward in farmed aquatic animals. A salmon sector in northern Europe shows how fast the cycle completes. Through the 1980s and into the early 1990s the industry depended on antibiotics to hold bacterial pathogens down. Furunculosis, the disease caused by Aeromonas salmonicida subspecies salmonicida, dominated that period and turned into a major problem between 1988 and 1995, after infected smolts entered the region.
Control began with oxytetracycline, sulfadiazine and trimethoprim. Multidrug resistant strains of A. salmonicida followed. Once resistance to those three had established itself, the sector shifted to quinolones, first oxolinic acid and then flumequine, and point mutations in the gyrA gene appeared quickly. By the late 1980s, plasmids carrying oxytetracycline, sulfonamide and trimethoprim resistance had already been recovered from infected fish. A working vaccine arrived shortly before the middle of the 1990s, and the medical value of antibiotics against furunculosis largely disappeared.
The plasmids did not disappear with it. Here is where plasmid mediated antibiotic resistance in aquaculture stops being a laboratory question and starts being an asset question, because the resistance machinery outlasted the treatment practice that selected for it. A family of resistance plasmids, the pRAS family, came out of those outbreaks, and four types were described. pRAS1 carried a tetracycline efflux determinant plus a trimethoprim and a sulfonamide resistance gene. pRAS2 carried streptomycin, sulfonamide and tetracycline resistance. pRAS3 and pRAS3.2 carried a different tetracycline determinant alongside a set of mobilization genes. The three largest, pRAS1, pRAS2 and pRAS4, measured between roughly 45 and 50 kilobases and all transferred themselves. pRAS3 was the exception: small, and unable to conjugate alone.
Two gaps remained open. The complete conjugation machinery of pRAS2 had never been resolved, so it was unknown whether the plasmid carried a full type IV secretion system (T4SS) of the kind other conjugative plasmids encode. And although pRAS3 had been seen to move only when pRAS1 or pRAS2 occupied the same cell, nobody had worked out why that dependence existed.
Conjugation rewards attention because it is the most efficient route of horizontal gene transfer (HGT) between bacteria, and it is the single mechanism behind most plasmid mediated antibiotic resistance in aquaculture. A conjugative plasmid assembles a multi protein channel, the conjugation apparatus, and threads DNA through it from a donor cell into a recipient. Such a plasmid is self transmissible. A mobilizable plasmid holds only part of the transfer toolkit, usually an origin of transfer and relaxase functions, and leans on a conjugative element supplied in trans by a helper plasmid sharing the cell.
How the transfer experiments were set up
Testing how plasmid mediated antibiotic resistance in aquaculture behaves in practice called for a donor that carried the plasmid, a recipient that did not, and a way to separate the two after mating. Two A. salmonicida donors were used. One came from a typical furunculosis case and carried pRAS2 together with pRAS3. The other came from an atypical furunculosis case and carried pRAS3 alone, with no pRAS2 in the cell at all. The recipient was a laboratory Escherichia coli strain holding no resistance plasmids.
| Parameter | Detail |
| Donor strains | Two A. salmonicida isolates; one carrying pRAS2 and pRAS3, one carrying pRAS3 alone |
| Recipient strain | Escherichia coli with no resistance plasmids and no mobilization genes |
| Donor growth | Blood agar, 15 degrees Celsius, 72 hours |
| Recipient growth | Blood agar, 37 degrees Celsius, 24 hours, aerobic |
| Susceptibility testing | Disc diffusion on Mueller-Hinton agar |
| Discs used | Tetracycline 30 micrograms, trimethoprim 5 micrograms, streptomycin 10 micrograms, compound sulfonamides 300 micrograms |
| Compound sulfonamide disc | Sulfamerazine 78.6 micrograms, sulfathiazole 109.4 micrograms, sulfadiazine 110.5 micrograms |
| MIC panel | Tetracycline, sulfamethoxazole and trimethoprim by microdilution at 0.5 McFarland |
| Streptomycin MIC | Not determined; streptomycin was assessed by disc diffusion only |
| Mating | Donor and recipient mixed 1 to 1 by volume, spotted on non selective blood agar, 24 hours at room temperature, about 22 degrees Celsius |
| Counter selection | Donor cultured at 14 degrees Celsius, recipient at 37 degrees Celsius; the donor cannot grow at 37 degrees Celsius, so no chromosomal marker was required |
| Replication | Three independent biological replicates per mating condition |
| Sequencing | Paired end 2 x 301 bp reads; depth 61x for one donor, 36x for the other, 140x for the transconjugant |
| Resistance gene detection | Comprehensive Antibiotic Resistance Database at an 80 percent threshold, plus a plasmid finder tool at the same threshold |
| Statistical test | Unpaired two-tailed Student t test on log10 transformed frequencies; p below 0.05 taken as significant |
Two design choices carry more weight than they look. Streptomycin sat in the panel as a check on the helper plasmid rather than as a selection marker, since pRAS2 carries a functional transposon encoding streptomycin resistance. Had pRAS2 traveled with pRAS3, the recipient would have turned streptomycin resistant, and it did not. Splitting the donor and recipient growth temperatures removed any need for chromosomal markers, because a donor that cannot grow at 37 degrees Celsius cannot contaminate the selection plates. The study also drew on a set of conjugation trials run between 1993 and 1995, before whole genome sequencing existed, in which the same donors were mated with the same E. coli recipient and the transconjugants were characterized by agarose gel plasmid profiling and by acquired resistance testing. Read together, the two datasets describe one system twice, once as sequence and once as behavior.
Plasmid mediated antibiotic resistance in aquaculture: the three modules of pRAS2
Sequencing divided the pRAS2 backbone into three modules, and that division is the most useful frame in the work for anyone who has to treat plasmid borne resistance as an operating problem rather than a laboratory curiosity. Module A runs replication, maintenance and stable inheritance. Module B builds the conjugation machinery. Module C carries the resistance determinants and the mobile elements that shuffle them around. The three behave very differently over time and distance, and much of the confusion in this field comes from treating the plasmid as one indivisible object.
| Module | What it does | Genes identified |
| Module A | Plasmid replication, maintenance, partitioning and stable inheritance | tnpA (transposase), sprT, res (resolvase), korB (kil override protein B), soj (chromosome partitioning protein), trfB (transcriptional regulator), smc (structural maintenance of chromosome protein), klcA (kilA protein), repA (replication initiator), higA (antitoxin), traX (transfer associated protein), bin3, prt (phosphoribosyl transferase) |
| Module B | Type IV secretion system and DNA processing for transfer | virB2, virB3, virB4, virB5, virB6, virB8, virB9, virB10, virB11, virD4; plus topA, cwh, traI, traD, lpd7C, eexN |
| Module C | Resistance determinants and associated mobile genetic elements | zmp (zinc metalloprotein), aph(6)-Id (also called strB, streptomycin resistance), neo (aminoglycoside resistance), hin (site specific recombinase), tnpA (transposase), regX (transcriptional regulator), mobA, mobC, sul2 (dihydropteroate synthase), glmM (phosphoglucosamine mutase), tetA(31) and its regulator tetR(31) |
Module C matches a Tn5393-like transposon carrying the streptomycin resistance genes, which earlier work on this plasmid family had already flagged. The tetracycline determinant sits in the same region as tetA(31), paired with its own repressor tetR(31). Put plainly, Module C is the part of pRAS2 that a farm’s drug choice can influence. Modules A and B are the parts it cannot.
The type IV secretion system encoded by pRAS2
Module B settles the first open question. pRAS2 encodes ten T4SS proteins, landing inside the eight to twelve conserved proteins that form the nanomachinery of most conjugative plasmids. Those ten proteins divide into four subassemblies, and every one of them was present.
| Subassembly | Proteins in pRAS2 | Role |
| Extracellular outer membrane complex | virB2, virB5 | virB2 forms the bulk of the filamentous pilus; virB5 is the minor pilus component whose adhesive properties bring donor and recipient into contact |
| Inner membrane complex | virB3, virB6, virB8 | Provides the structural base of the secretion pore; virB6 controls the channel and modulates substrate secretion |
| Core secretory channel | virB9, virB10 | Spans the periplasm and the outer membrane; virB10 bridges the two membranes of the T4SS subassemblies |
| Cytoplasmic energy complex | virB4, virB11, virD4 | Fuels transport of substrate through the channel; virD4 and virB11 are the ATPases |
Two absent proteins are informative rather than alarming. virB7, normally part of the core channel, was not detected in pRAS2, and it was missing from every comparison plasmid as well, which suggests its loss does not block a functional secretion channel. virB1 was absent across the whole comparison set too. Set against that, the ATPase pair and the coupling protein virD4 were all present, and that combination is what supports the conclusion that pRAS2 encodes a canonical T4SS rather than a fragment of one.
Module B also carries the supporting cast for DNA handling. topA relaxes supercoiled DNA during replication and transcription and mediates the cutting and pasting that transposition needs. lpd7C belongs to the relaxase family that nicks plasmid DNA and forms a covalent link with single stranded DNA before transfer. cwh acts as a signal peptide for translocation, traI handles DNA processing during transfer, and traD works alongside it. eexN is the one worth pausing on, because it forms an entry exclusion system that blocks plasmid DNA from entering a host that already carries the same plasmid. That gives pRAS2 a control against depositing two or more similar plasmids in a single daughter cell.
How efficiently does pRAS3 move?
pRAS3 moved, and the rates were measurable and reproducible across mating formats. Frequency is reported two ways: transconjugants per donor colony forming unit (TF/donor) and transconjugants per recipient colony forming unit (TF/recipient).
| Mating condition | TF/donor | TF/recipient |
| Broth mating | 6.65 x 10^-7 (SD 1.98 x 10^-7) | 5.73 x 10^-9 (SD 2.26 x 10^-9) |
| Plate mating | 1.08 x 10^-6 (SD 7.79 x 10^-7) | 7.49 x 10^-9 (SD 1.56 x 10^-9) |
| Difference between conditions | p = 0.84 | p = 0.30 |
Three replicates per condition, and three independent experiments per format. Neither normalisation showed a significant difference between liquid and solid mating, so the physical setting did not change the outcome. Donor normalised frequencies came out roughly 100 fold above recipient normalised frequencies, which fits the expectation that only a fraction of donor cells successfully transfer at any moment.
The honest reading is that pRAS3 mobilization is a low frequency event with high reproducibility. At roughly one transconjugant per one to two million donor cells, it will never surface in a plate count a farm or a diagnostic laboratory runs as routine. It will surface in any experiment built to look for it. That gap between what is detectable by design and what is detectable by routine is the whole reason plasmid mediated antibiotic resistance in aquaculture gets underestimated in the field.
What the transconjugants were resistant to
What came out of the mating was more instructive than how often it happened. Disc diffusion produced these phenotypes, where R means resistant and S means susceptible.
| Strain | Tetracycline | Compound sulfonamides | Trimethoprim | Streptomycin |
| A. salmonicida donor with pRAS2 and pRAS3 | R | R | S | R |
| A. salmonicida donor with pRAS3 only | R | S | S | S |
| E. coli recipient | S | S | S | S |
| E. coli transconjugant | R | S | S | S |
The transconjugant picked up tetracycline resistance and nothing else. It remained susceptible to compound sulfonamides, to trimethoprim and to streptomycin. Because streptomycin resistance is a pRAS2 marker in this system, its absence is direct phenotypic evidence that the helper plasmid never traveled. The minimum inhibitory concentrations (MICs) tell the same story in numbers.
| Strain | Tetracycline (micrograms per mL) | Sulfamethoxazole (micrograms per mL) | Trimethoprim (micrograms per mL) |
| A. salmonicida donor with pRAS2 and pRAS3 | 32 | 8 | at or below 0.25 |
| A. salmonicida donor with pRAS3 only | 32 | at or below 0.25 | at or below 0.25 |
| E. coli recipient | at or below 0.25 | at or below 0.25 | at or below 0.25 |
| E. coli transconjugant | 32 | at or below 0.25 | at or below 0.25 |
A tetracycline MIC of 32 micrograms per mL in the transconjugant matches the donor exactly, so the transferred determinant is fully functional in its new host rather than merely present. Meanwhile the donor’s sulfamethoxazole MIC of 8 micrograms per mL and its streptomycin resistance both stopped at the donor. One mating event moved one determinant and left the rest behind, the pattern that makes plasmid mediated antibiotic resistance in aquaculture so hard to compress into a single resistance profile.
No transconjugants were recovered from matings using the donor that carried pRAS3 alone. That negative result is the experimental core of the work. A mobilizable plasmid in a cell with no helper does not move.
Why the same donor produced three different transconjugants
Taking the E. coli transconjugant as the whole story gives a misleading picture, because a donor carrying pRAS2 plus pRAS3 does not produce one transfer outcome. It produces a mixture, and the historical record from 1993 to 1995 shows how that mixture is structured.
| Trial date | Mating | Selection | Colony appearance | Frequency from donor 1 | Frequency from donor 2 |
| 18 June 1993 | Agar | Sulfadiazine | Not recorded | 1.83 x 10^-2 | Not recorded |
| 30 June 1993 | Agar | Sulfadiazine | Not recorded | 3.9 x 10^-2 | 4.35 x 10^-2 |
| 7 July 1993 | Agar | Sulfadiazine | Not recorded | 5.2 x 10^-2 | 7.0 x 10^-2 |
| 8 September 1993 | Agar | Sulfadiazine | Not recorded | 1.3 x 10^-1 | 1.3 x 10^-1 |
| 1 March 1995 | Broth | Tetracycline | Regular colonies | 1 x 10^-7 | 1.7 x 10^-7 |
| 7 March 1995 | Broth | Tetracycline | pRAS2 plus pRAS3, early, large colonies | 1.27 x 10^-7 | 2.4 x 10^-7 |
| 7 March 1995 | Broth | Tetracycline | pRAS2 alone, late, small colonies | 1.58 x 10^-4 | 2.2 x 10^-4 |
Three things fall out. Selection pressure changes what you recover. Under sulfadiazine selection, which matches a pRAS2 marker, transconjugants came back between 1.8 x 10^-2 and 1.3 x 10^-1, three to five orders of magnitude above anything recovered under tetracycline selection, and gel profiling repeatedly showed pRAS2 arriving alone. Tetracycline selection produced two visibly different populations on the same plate: an early wave of larger colonies carrying both plasmids, and a later wave of smaller colonies carrying pRAS2 by itself. And autonomous transfer of pRAS2 ran about three orders of magnitude above co transfer with pRAS3.
Across four independent conjugation trials, plasmid profiling showed pRAS2 moving alone in three and both plasmids moving together in one. In every trial across the whole series, pRAS3 entered the recipient only when the donor also carried a conjugative helper, either pRAS2 or pRAS1. It never moved from a donor where it sat alone. That is the definition of a mobilizable plasmid demonstrated rather than inferred from sequence.
One further point deserves stating plainly, because it is easy to misread in a dossier. The transconjugant that was sequenced carried pRAS3 without pRAS2. That does not mean the helper failed to transfer. It means the helper was either not maintained in the recipient or remained in the donor, and the colony that happened to be picked came from the rarer pRAS3 only class. With a three order of magnitude gap between classes and only one colony taken forward, recovering the rarer class fits the earlier literature rather than contradicting it.
What whole genome sequencing confirmed
Sequencing closed the gap between phenotype and genotype.
| Strain | Genome size (Mb) | GC content (%) | Scaffolds | Proteins | Genes | R plasmids detected |
| A. salmonicida donor with both plasmids | 4.69 | 58.54 | 92 | 4,339 | 4,489 | pRAS3 and pRAS2 |
| A. salmonicida donor with pRAS3 only | 4.32 | 59.12 | 185 | 4,085 | 4,173 | pRAS3 |
| E. coli recipient (reference assembly) | 4.51 | 50.74 | 89 | 4,201 | 4,281 | None |
| E. coli transconjugant | 4.54 | 50.77 | 63 | 4,220 | 4,301 | pRAS3 |
The recipient reference assembly holds no sequence corresponding to either pRAS2 or pRAS3, which rules out the possibility that the resistance genes were already sitting in the recipient and simply switched on. The transconjugant genome contained a contig matching pRAS3 in gene content and gene order, and no contig matching pRAS2. Every mobilization gene carried by pRAS3 in the transconjugant matched the donor, and the recipient had none of them before mating. What moved was a complete, organized plasmid, not a fragment.
Where pRAS2-like plasmids have been found
The comparison set carries the widest implications, because it spans four countries and more than three decades of isolation dates.
| Plasmid | Host species | Region | Year |
| pRAS2 | Aeromonas salmonicida subspecies salmonicida | Norway | 1992 |
| pBY3 | Aeromonas salmonicida subspecies masoucida | South Korea | 2018 |
| pEIB202 | Edwardsiella piscicida | China | 2006 |
| pVIM-20710 | Enterobacter hormaechei | China | 2011 |
| pVb1978 | Vibrio alginolyticus | China | 1978 |
| p3PS10 and p3PS11 | Piscirickettsia salmonis | Chile | 2013 |
| p4PS8 and p4PS9 | Piscirickettsia salmonis | Chile | 2015 |
The closest relative of pRAS2 was pBY3, from a South Korean isolate of a different A. salmonicida subspecies, at 99.99 percent nucleotide identity across 82 percent of the query. That is a near identical plasmid backbone in a different subspecies on a different continent, isolated 26 years later. The Tn5393-like transposon and its resistance genes told the opposite story: 100 percent identity, covering only 7 to 16 percent of the pRAS2 sequence. The conjugation hardware is near identical worldwide. The resistance cargo is not.
Module B backs that reading. The T4SS gene set was conserved across plasmids from Edwardsiella piscicida, Piscirickettsia salmonis, Vibrio alginolyticus and Enterobacter hormaechei. Two gaps appeared anywhere in the comparison: one plasmid lacked traI and another lacked traD. virB2 was missing from three of the Piscirickettsia salmonis plasmids, which is a genuine limit on their conjugation capability rather than a detail.
Module C, the resistance region, was the most variable of the three. The table below shows which accessory resistance genes appeared where. A plus sign means the gene was detected and a minus sign means it was not.
| Gene or element | pRAS2 (Norway) | pBY3 (South Korea) | pEIB202 (China) | p3PS10 (Chile) | p3PS11 (Chile) | p4PS8 (Chile) | pVb1978 (China) | pVIM-20710 (China) |
| sul2 | + | + | + | + | + | – | + | – |
| aph(6)-Id | + | – | + | – | – | – | – | – |
| neo | + | – | – | – | – | – | – | – |
| tetA(31) and tetR(31) | + | + | + | + | + | – | – | – |
| cat or cat3 | – | – | + | – | – | – | + | – |
| cmlA | – | – | – | – | – | + | – | – |
| aacA4 | – | – | – | – | – | – | + | – |
| ant1 | – | – | – | + | + | + | + | – |
| emrE | – | – | – | – | – | – | + | – |
| mobA | + | + | + | + | + | – | – | – |
| mobC | + | + | – | – | – | – | – | – |
| regX | + | + | – | – | – | – | – | – |
| zmp | + | + | – | – | – | – | – | – |
| glmM | + | + | – | – | – | – | – | – |
| hin | + | + | – | – | – | – | – | – |
| tnpA | + | + | – | – | – | – | – | – |
| Insertion sequence | tnpA, hin | tnpA, hin | ISPrst3 | ISKpn13 | ISKpn13 | ISKpn13 | none | none |
Two patterns stand out. Tetracycline resistance as tetA(31) with its repressor was shared by the Norwegian, South Korean, Chinese Edwardsiella and both Chilean Piscirickettsia plasmids, and absent from the Vibrio and Enterobacter plasmids. And the mobile elements surrounding resistance genes differ by host and region: tnpA and hin in the Norwegian plasmid, ISKpn13 in the Chilean plasmids, a different transposase in the Chinese Edwardsiella plasmid, and no insertion element at all in one Vibrio plasmid. Insertion sequences and transposases are the machinery through which resistance genes get acquired and rearranged, so that distribution maps where resistance gene traffic has been happening.
Module A, the replication and maintenance module, kept its core functions across the whole set, with individual accessory and stability genes variably retained. One plasmid lacked trfB, higA, bin3 and prt together, another lacked bin3 and trfB, and a third lacked higA, traX and bin3.
The pRAS3 comparison agrees with the same picture. Both donors, and the transconjugant, shared a conserved set of mobilization genes plus a repressor and an antitoxin gene, along with a tetracycline efflux pump and its regulator. The transconjugant’s pRAS3 was nearly identical in organization to a pRAS3 sequence from a British isolate obtained 23 years later. Two Canadian variants differed by carrying three copies of an antitoxin gene where the donor plasmids carried one.
The counterintuitive part: the conserved half of pRAS2 is not the resistance genes
Most people who think about plasmid borne resistance assume the resistance genes are the interesting part and the delivery system is only a vehicle. The comparative data says the reverse. What traveled intact across four countries and 26 years was the conjugation machinery, conserved to 99.99 percent nucleotide identity. What differs everywhere is the resistance cargo. The delivery system is globally distributed and stable. The payload gets assembled locally.
That inverts the usual risk model, with direct operational consequences. A farm that has never imported fish, never shared equipment and never seen a resistant isolate on site can still sit in a watershed where pRAS2-like backbones already circulate as part of the normal aquatic microbiota. The plasmid does not have to arrive with a shipment. The resistance gene has to arrive, and it only has to arrive once, because a mobilizable plasmid sharing a cell with a helper plasmid will pick up the payload and carry it onward.
A second counterintuitive result is mechanical, and it comes from the historical trials. The plasmid doing the moving is not the plasmid that shows up. pRAS3 was mobilized into the recipient while pRAS2 was not detected there, even though pRAS2 supplied the entire transfer apparatus. Across the trial series, the helper transferred alone about three orders of magnitude more often than it co transferred with the mobilizable plasmid. The most common event detectable from such a donor is therefore the helper alone, and the rare event carrying the extra determinant is exactly the one a single colony pick is most likely to miss.
A third point concerns selection pressure, and it cuts against the intuition that more drug exposure simply means more resistance transfer. Under sulfadiazine selection, which matches the resistance the helper plasmid carries, transconjugants came back between 1.8 x 10^-2 and 1.3 x 10^-1. Under tetracycline selection, which matches the resistance carried by the mobilizable plasmid as well as the helper, they came back between 1.0 x 10^-7 and 2.4 x 10^-4. The agent you select on changes not only how much you recover but which plasmid class you recover, and so it changes which picture of plasmid mediated antibiotic resistance in aquaculture you end up believing.
What this changes on a fish farm and in an export dossier
None of this translates into “use less oxytetracycline”, and none of it translates into “our strain is clean, so we are fine”. It translates into specific decisions about what to measure and what to document.
- Make the conjugative backbone the surveillance unit, not just the resistance phenotype. A site antibiogram shows which resistances are present. It does not show whether the machinery to move them is present. Plasmid level screening is the only thing separating a stable local resistance profile from one about to shift.
- Stop treating a single positive culture as a risk assessment. The same donor in these experiments produced three transconjugant types at frequencies spanning three orders of magnitude. One colony pick samples a distribution, it does not summarise it.
- Record which molecule was applied, not only how much. Selection agent determined both recovery frequency and plasmid class in the historical trials. A usage log that records total tonnage without recording which active ingredient was applied at which point in the cycle cannot support any inference about which resistance determinants were being favored.
- Keep two question sets apart in customer and regulator conversations. “What resistances does your isolate carry” is a phenotype question and it is cheap to answer. “What is the probability those resistances move to another host” is a plasmid question needing sequence level data. Blending them produces claims the evidence cannot support in either direction.
- Treat a farm’s stocking and receipt history as a plasmid risk factor, not only a disease risk factor. pRAS1, a related conjugative plasmid, shares an identical backbone with a plasmid isolated from aquaculture environments decades earlier in a different country, and related backbones have also been reported in Edwardsiella piscicida. Backbones circulate over long timescales and long distances.
- Expect the resistance region to vary, and design documentation around it. A conserved backbone means a clean record at one site does not predict a clean record at the next, and the same species can carry different resistance loads in different regions, as the Chilean, Chinese and South Korean comparison plasmids show.
- Read a negative result correctly. Two donors here carried the same mobilizable plasmid; only the one that also carried the helper moved it. Absence of transfer in a trial says as much about the donor’s plasmid content as about the recipient.
Limitations
The work carries real limits, and they should travel with the conclusions.
Only one transconjugant colony was taken forward for sequencing, which is why the three class mixture is visible in the historical record rather than in the sequencing data itself. The recipient E. coli strain was not sequenced here, so a reference assembly rather than a matched genome confirmed that the recipient carried no related plasmid sequences. Streptomycin MIC was not determined because it was not in the custom panel, leaving that one marker on disc diffusion alone. There are no standardized interpretive breakpoints for A. salmonicida, so donor susceptibility was read against published guidelines for aquatic pathogens rather than a validated species specific standard, and cross laboratory comparison of those phenotypes needs care.
The stability question is explicitly open. Serial passage of the transconjugants over successive generations is required to establish how fast pRAS3 is lost from the E. coli background, particularly given that a plasmid from a cold adapted host is expected to replicate and segregate less stably in a host grown at 37 degrees Celsius. Whether a mobilized plasmid persists is a different question from whether it transfers, and neither this work nor the historical trials answers it. That distinction matters for plasmid mediated antibiotic resistance in aquaculture, because a plasmid that transfers rarely but persists for years is a different problem from one that transfers often and is lost quickly.
The 1993 to 1995 trial record is a historical comparison, and the original annotation for the older sibling plasmids was weaker, which the source notes as a likely explanation for differences seen during comparison. The selection concentrations logged in that record are also in different units from the MIC panel, so the frequencies in that table are best read as relative comparisons between conditions rather than as absolute calibration for any other system.
One internal inconsistency deserves flagging for anyone reading the primary record. The methods description lists all three A. salmonicida isolates as resistant to tetracycline, compound sulfonamides and trimethoprim, while both the disc diffusion results and the MIC panel show one donor resistant to tetracycline only. The tabulated and results section data are used in this article. A mismatch of this kind usually reflects a summarising sentence that was not updated once the individual datasets were finalised, but it means donor phenotypes should be quoted from the tables rather than from the methods paragraph.
The geographic interpretation also needs a caution. The wide distribution of near identical plasmid backbones across Norway, South Korea, Chile and China is a sequence observation. The explanation offered, that these plasmids may circulate through broad environmental reservoirs such as large scale aquatic connectivity rather than only through direct human transfer, is a hypothesis that fits the data and is not tested by it. Movement of fish stocks, movement of equipment and ocean scale dispersal are not separated in this dataset.
Bottom line
Resistance in aquaculture is not only a question of which bacteria are resistant. It is a question of which plasmids are in the water. Plasmid mediated antibiotic resistance in aquaculture depends on a delivery system far more conserved and far more widely distributed than the resistance genes it carries. Here, one conjugative plasmid supplied a complete type IV secretion system and moved a non conjugative plasmid into a new host while staying behind itself, at roughly one event per one to two million donor cells, and the donor’s other resistances did not travel with it. A conserved global delivery system, a locally variable cargo, and a transfer rate invisible to routine culture but reproducible on demand: that combination is what any serious resistance management plan has to be built around.
FAQ
What is plasmid mediated antibiotic resistance in aquaculture?
Plasmid mediated antibiotic resistance in aquaculture is resistance carried on conjugative plasmids rather than inherited by descent. A conjugative plasmid builds a protein channel and passes a copy of its DNA into a recipient during contact. The recipient need not be related to the donor, so resistance genes cross species boundaries in a single step and can reach bacteria never exposed to that drug.
How does pRAS2 transfer resistance genes into a new bacterial host?
pRAS2 encodes ten type IV secretion system proteins forming an outer membrane pilus, an inner membrane complex, a core channel and a cytoplasmic energy complex. After contact, the relaxase machinery nicks the plasmid DNA and a single stranded copy crosses into the recipient. Frequencies ran from 6.65 x 10^-7 transconjugants per donor cell in broth to 1.08 x 10^-6 on plates.
Which resistance genes does pRAS2 carry?
pRAS2 carries sulfonamide resistance through sul2, streptomycin resistance through aph(6)-Id on a Tn5393-like transposon, aminoglycoside resistance through neo, and tetracycline resistance through tetA(31) with its repressor tetR(31). In the donor used here that produced a tetracycline MIC of 32 micrograms per mL, a sulfamethoxazole MIC of 8 micrograms per mL, and susceptibility to trimethoprim.
Can a non conjugative plasmid transfer on its own?
No, and this was demonstrated rather than assumed. Two donors carried the same mobilizable plasmid, pRAS3, but only the donor that also carried the conjugative helper pRAS2 produced transconjugants. Matings using the donor with pRAS3 alone recovered none at all. A mobilizable plasmid encodes only part of the transfer toolkit and needs a conjugative element supplied in trans.
What is helper dependent plasmid mobilization?
Helper dependent plasmid mobilization is the process where a conjugative plasmid builds the transfer apparatus and a second, non conjugative plasmid uses it to move. The helper supplies the secretion channel and DNA processing, while the mobilizable plasmid supplies the origin of transfer and its own mobilization genes. Here the mobilizable plasmid reached the recipient while the helper was not detected there.
How often does pRAS3 move into a new host?
Mobilization ran at 6.65 x 10^-7 plus or minus 1.98 x 10^-7 transconjugants per donor cell in broth mating and 1.08 x 10^-6 plus or minus 7.79 x 10^-7 on solid surfaces, with three replicates per condition and no significant difference between formats. That is roughly one successful transfer per one to two million donor cells, far too low to show up in routine culture counting.
Why did the same donor produce three different types of transconjugant?
Because such a donor does not generate one outcome but a mixture. Under tetracycline selection, an early wave of larger colonies carried both pRAS2 and pRAS3 at about 1.3 x 10^-7 to 2.4 x 10^-7, while a later wave of smaller colonies carried pRAS2 alone at about 1.6 x 10^-4 to 2.2 x 10^-4. Autonomous transfer ran roughly three orders of magnitude above co transfer.
Why is the conjugation machinery conserved across countries while the resistance genes are not?
Because the machinery is inherited as a stable backbone while resistance genes are acquired as an accessory. The closest relative of pRAS2 here came from a different subspecies on a different continent, isolated 26 years later, at 99.99 percent nucleotide identity across 82 percent of the query. The resistance region matched at 100 percent identity but covered only 7 to 16 percent of the plasmid.
Does oxytetracycline use drive resistance transfer in fish farming?
The data show selection agent choice changes both the frequency and the class of transconjugant recovered, but they do not test whether drug exposure increases transfer. Under sulfadiazine selection, recovery ran from 1.8 x 10^-2 to 1.3 x 10^-1. Under tetracycline selection it ran from 1.0 x 10^-7 to 2.4 x 10^-4. The wider claim about antimicrobial use remains a hypothesis here.
Can resistance genes from fish pathogens reach human pathogens?
The mechanism allows it and the plasmid distribution is consistent with it, but this study did not demonstrate it. Conjugative pRAS plasmids transfer across species boundaries, including into E. coli, and near identical backbones appear in phylogenetically distant aquatic species. The authors suggest aquatic microbiota may be a platform where resistance is recruited before reaching clinical settings.
Why does a resistant donor not pass on all of its resistances at once?
Because the determinants sit on different plasmids and only one is mobilizable. The donor carried tetracycline, sulfonamide and streptomycin resistance, yet the transconjugant acquired tetracycline resistance only, holding an MIC of 32 micrograms per mL against the donor’s 8 micrograms per mL for sulfamethoxazole and its retained streptomycin resistance. Genome sequencing confirmed the pRAS2 determinants stayed behind.
How should a fish farm or an aquatic drug supplier manage resistance gene movement?
Shift the surveillance unit from phenotype to plasmid. Record which active ingredient was applied, not just tonnage, because the selection agent decides which resistance class is favored. Treat a single isolate as one sample from a wide distribution: this donor produced transconjugant classes spanning three orders of magnitude. Keep cheap phenotype questions separate from transfer questions needing sequence data.
