Edwardsiella tarda infection: what protects gill cells
Edwardsiella tarda infection damages gill epithelial cells through adhesion, entry, barrier loss and oxidative stress. In an RTgill-W1 gill cell model, an organic acid-based antimicrobial mixture (Aq) stopped growth at 1 percent and killed the pathogen at 2 percent, while a sub-inhibitory 0.5 percent level cut adhesion and intracellular bacterial recovery without measurable harm to the cells.
Why Edwardsiella tarda infection starts at the gill surface
Gills are a compromise, and an awkward one. The epithelium has to stay thin enough for gas exchange and ion regulation, but it sits right in the water column, exposed to everything living there. That is why gill barrier integrity governs the first contact between host and pathogen. Edwardsiella tarda makes the problem worse. It lingers in the environment, survives inside host cells, sidesteps immune clearance and causes disease across a wide range of fish hosts. Adhesion and internalization are not late complications. They decide whether colonization happens at all, and once inside, the bacterium has a route to persistence.
Production pays for this too. Modeled annual direct losses in one intensively farmed catfish sector ran from USD 5.2 million to 17.6 million, with a total negative economic impact of USD 15.5 million to 45.9 million. Set against reported annual sales of USD 437 million, that is roughly 1.2 to 4.0 percent of sales in direct losses and 3.5 to 10.5 percent in total impact. How much of it belongs to E. tarda specifically is harder to say, partly because historical disease records predate the taxonomic separation of fish-associated Edwardsiella.
Model and test conditions
RTgill-W1 is a rainbow trout gill epithelial cell line. It forms electrically resistive monolayers, expresses tight-junction-associated genes and mounts inducible innate responses, which is what makes it a usable stand-in for the gill interface. Cell culture, bacterial culture and infection conditions are listed below.
| Parameter | Setting |
|---|---|
| Cell model | RTgill-W1 rainbow trout gill epithelial cells |
| Culture medium | DMEM with 10 percent FBS and 1 percent penicillin/streptomycin |
| Culture temperature | 20 degrees Celsius in non-ventilated flasks |
| Passaging | 1:2 every 10 days |
| Seeding | 1 x 10^5 cells, 15 days before experiments, reaching over 80 percent confluence |
| Bacterium | A reference strain of Edwardsiella tarda |
| Bacterial culture | Tryptic soy broth or agar, or DMEM, at 28 degrees Celsius |
| Infection dose | Multiplicity of infection 100 (OD600 near 0.4) |
| Infection period | 3 days (72 hours) in DMEM held at pH 7.2 |
| Aq concentrations | 0, 0.1, 0.2, 0.5, 1 and 2 percent (v/v) |
| Replication | Three independent experiments |
Composition of the organic acid-based antimicrobial mixture
The blend is a dry one, and citric acid dominates it. Sodium citrate acts as its conjugate buffer, malic acid is a second acid, silica carries it, and two plant extracts sit alongside maltodextrin as filler. Shares are by weight.
| Component | Share by weight |
|---|---|
| Citric acid | 42 percent |
| Sodium citrate | 18 percent |
| Malic acid | 12 percent |
| Silica | 10 percent |
| Citrus extract | 9 percent |
| Maltodextrin | 5 percent |
| Olive extract | 3 percent |
| Sodium chloride | 1 percent |
Each part has a plausible job. Citric acid limits depletion of reduced glutathione and supports glutathione peroxidase activity in liver and brain tissue. Malate lifts total antioxidant capacity along with glutathione peroxidase activity, which works beside GSH to neutralize free radicals. Maltodextrin feeds in indirectly by supporting the amino acids that GSH synthesis needs. Running them together is deliberate: bacterial virulence, oxidative stress, inflammatory signaling and barrier disruption all arrive at once during an infection.
Core outcomes at a glance
The headline numbers, before the detail. Every row is a comparison against untreated infected controls at 3 days post-infection unless stated otherwise.
| Readout | Direction with Aq | Reported significance |
|---|---|---|
| Total bacterial adhesion | Reduced across 0.1 to 2 percent | P < 0.0001 |
| Intracellular bacterial recovery | Reduced across 0.1 to 2 percent | P < 0.0001 |
| Cell viability (MTT) | Not markedly reduced at tested levels | Not significant |
| LDH release, infected and uninfected | Lower than untreated infected controls | Stated as reduced |
| TEER during infection | Resistance maintained | Stated as preserved |
| cldn-10e, cldn-30, zo-1 mRNA | Higher in infected and uninfected cells | Stated as increased |
| SOD activity in infected cells | Lower | P < 0.0001 |
| H2O2 in infected cells | Lower at every concentration | P < 0.0001 |
| Intracellular GSH at 72 h | Higher, dose-dependent, strongest in infected cells | P < 0.0001 |
| NF-kB p65 activation | Lower, particularly during infection | Stated as reduced |
| IL-8, IL-1beta, TNF-alpha mRNA | Lower at every concentration | P < 0.0001 |
Inhibitory and bactericidal concentrations
Two-fold tube dilution in tryptic soy broth set the minimum inhibitory concentration and the minimum bactericidal concentration, running from 8 percent down to 0.015625 percent (v/v) with an inoculum of 1 x 10^5 CFU/mL held at 28 degrees Celsius for 24 hours. Any tube without visible growth sat above the MIC. From each of those tubes, a 100 microliter aliquot went onto tryptic soy agar to find the MBC. The minimum inhibitory concentration landed at 1 percent and the minimum bactericidal concentration at 2 percent. Growth curves ran separately, in 96-well plates from a starting culture of 2.0 log CFU/mL with 0, 0.1, 0.2, 0.5, 1 and 2 percent Aq, reading OD600 every 2 hours over 20 hours.
Two concentrations then carry the rest of the work. The 2 percent bactericidal level stayed out of the mechanistic assays, with the viability test as the exception. Wherever the question was what the mixture does below the point where growth stops, 0.5 percent served as the sub-inhibitory concentration.
Adhesion and intracellular recovery during Edwardsiella tarda infection
From 0 to 2 percent, Aq treatment pushed both total bacterial adhesion and intracellular bacterial recovery down, each at P < 0.0001. Adhesion was counted after washing the infected cells, breaking them open in 0.1 percent Triton X-100 in PBS for 15 minutes at 37 degrees Celsius, and plating tenfold dilutions onto tryptic soy agar at 28 degrees Celsius for 24 hours with 2 further days of incubation. The invasion readout inactivated attached bacteria first, using gentamicin (400 mg/mL in DMEM, 3 hours at 28 degrees Celsius), and only then broke the cells open.
Attachment to the epithelial surface is where Edwardsiella tarda infection begins, so a fall in total adhesion means the first step of colonization was interfered with. A fall in intracellular recovery means entry, or survival once inside, was interfered with, and both of those feed persistence and disease progression. What the data cannot settle is the route: direct pressure on the pathogen, altered host-cell surface properties, or modulation of host defense and barrier responses.
Cell viability cytotoxicity and barrier integrity
MTT went in at 0.5 mg/mL for a 3 hour incubation, the formazan was dissolved, and absorbance was read at 570 nm. Aq did not markedly reduce viability at any concentration tested, so the gill epithelial model tolerated it under these conditions. Cytotoxicity came from lactate dehydrogenase release at 3 days post-infection, read at 500 nm. LDH release was lower with Aq in infected and uninfected cells alike than in untreated infected controls, so infection-associated damage to membrane integrity went down rather than up.
Transepithelial electrical resistance was measured in transwells seeded with 1 x 10^5 cells (6.5 mm diameter, 0.4 micrometer pores) at 3 days post-infection, on an epithelial volt-ohm meter with an Endohm chamber. Resistance held up better during infection with Aq present, while uninfected monolayers showed limited disruption. TEER reports tight-junction function and monolayer permeability, so a fall would mean junction stability is being lost and holding it means the junctions took less damage. One detail matters here. The LDH trend looked similar in infected and uninfected cells across the concentration range, which suggests the effect on membrane integrity was broadly comparable in both conditions rather than tied only to infection.
Tight-junction gene expression
qPCR measured tight-junction gene expression using EF-1alpha and beta-actin as internal reference genes and quantification by the 2^-DDCt method, cycling at 95 degrees Celsius for 2 minutes followed by 45 cycles of 95 degrees Celsius for 30 seconds and 60 degrees Celsius for 60 seconds. In infected and uninfected cells alike, Aq raised relative mRNA for cldn-10e, cldn-30 and zo-1. Claudin 10e, claudin 30 and zonula occludens-1 are the junction proteins that keep this cell line physiologically intact, so higher transcript levels are molecular reinforcement of the same barrier TEER measured functionally. Having both readouts agree is worth more than either one in isolation, and during infection it separates a barrier that is holding from one being rebuilt after damage.
Oxidative stress readouts: H2O2 SOD and CAT
Three readouts were taken in infected and uninfected cells at 0, 0.1, 0.2, 0.5, 1 and 2 percent Aq: catalase activity, superoxide dismutase activity, and hydrogen peroxide (H2O2) in the supernatant. Uninfected cells showed no marked concentration-dependent change in CAT with Aq, and both baseline SOD activity and baseline H2O2 were detectable. Infected cells behaved differently. CAT activity fell after Aq exposure, SOD activity fell significantly (P < 0.0001), and H2O2 fell significantly at every concentration used (P < 0.0001).
That split supports two conclusions. Aq by itself does not trigger an oxidative stress response in RTgill-W1 cells, since the uninfected cells show no concentration-dependent rise in any of the three. During infection it modulates the response instead of adding to it, which fits the higher viability, lower cytotoxicity and better-preserved barrier seen in the parallel assays. Reactive oxygen species such as H2O2 serve as regulated signaling intermediates at physiological concentrations and turn damaging only once production outruns local buffering and repair capacity, so lower H2O2 during infection reflects a change in load rather than a change in signaling tone.
The counterintuitive finding: lower antioxidant enzyme activity with better outcomes
Falling SOD and CAT activity reads like weakened antioxidant defense. Most infection writing would take it that way, and here that reading would be wrong. The same cells carried less H2O2 and more intracellular glutathione, so what fell was the oxidative load, not the defense. Enzyme activity is a demand signal. With less superoxide and less peroxide to clear, there is less reason to run SOD and CAT hard. On its own, a fall in enzyme activity tells you nothing; set beside H2O2 and GSH, it tells you the load dropped.
The same reasoning sets the bar for calling a buffering gain real. Intracellular GSH rises for reasons that have nothing to do with damage, because healthy cells build reserves as a preventive buffer and to support essential metabolism. A GSH rise on its own therefore proves little. GSH up while H2O2 is down is the combination that counts, and that is what appeared here. The 0.5 percent time course makes it explicit: GSH climbed over time with no matching rise in H2O2, which is why the GSH response was read as a preventive reserve rather than a reaction to oxidative stress caused by the mixture. Keep the two observations separate as well. GSH does not directly trigger or regulate CAT and SOD production, so the enzyme falls and the GSH rise are not one causal chain.
Glutathione NF-kB and cytokine output
By 72 hours post-infection, Aq had raised intracellular glutathione in a dose-dependent manner, strongest in infected cells (P < 0.0001) and smaller in uninfected cells. GSH transporter transcripts (mrp1, oatp1d1, sec61a1) were measured to probe the mechanism behind that shift. NF-kB p65 activation fell with Aq exposure, particularly during infection, and was expressed as absorbance at 450 nm per microgram of nuclear protein.
On the inflammatory side, IL-8, IL-1beta and TNF-alpha mRNA were quantified by qPCR and normalized to EF-1alpha, cycling at 95 degrees Celsius for 10 minutes followed by 40 cycles of 95 degrees Celsius for 30 seconds, 60 degrees Celsius for 30 seconds and 72 degrees Celsius for 30 seconds. All three fell significantly at every concentration tested (P < 0.0001). The order that fits these data runs from redox to transcription: H2O2 down and glutathione up, then NF-kB p65 activation down, then lower output of pro-inflammatory cytokines. Call that an association rather than a demonstration. Glutathione is not synonymous with NF-kB inhibition, and the direction and biological consequence of the redox and NF-kB interaction shift with context in other systems, so the defensible claim is that the two moved together here.
Where oxytetracycline and other aquaculture antibacterials fit
Antibacterials still carry most of the load when bacterial disease hits farmed fish, and oxytetracycline is one of the compounds that keeps appearing in that setting. It is a useful reference point precisely because it shows which layer the organic acid work touches. Treatment decisions usually rest on clinical signs and group-level risk rather than on a susceptibility result, so antimicrobial susceptibility testing tends to arrive after the first treatment instead of before it, and by then adhesion, entry and barrier changes of the kind measured here have already happened. Residue limits and withdrawal intervals sit inside every treatment decision, and that pressure is part of why options acting before bacteria establish attract interest.
Oxytetracycline belongs in the treatment layer, not the prevention layer. What was measured here sits upstream of where an antibacterial acts, so a feed or water additive that lowers the infectious load at the epithelium works on a different step. That is the case for studying organic acid mixtures as complements to drug therapy rather than substitutes, and for measuring adhesion below the inhibitory concentration even though oxytetracycline and related compounds remain the tools once an outbreak is underway.
Practical implications for gill health programs
- Judge barrier outcomes with a functional readout (TEER) next to a molecular one (cldn-10e, cldn-30, zo-1). Agreement between the two is stronger evidence than either alone, and disagreement is a signal to check the model before drawing a conclusion.
- Treat sub-inhibitory exposure as a distinct question from growth inhibition. A 0.5 percent level that does not stop growth still cut adhesion and intracellular bacterial recovery, so screening only at inhibitory levels would miss the effect.
- Do not read a fall in SOD or CAT as damage until H2O2 and GSH are on the table. Enzyme activity tracks demand, and a load reduction looks identical to a capacity loss if only the enzyme is measured.
- Pair every barrier claim with a cytotoxicity readout (LDH) and a viability readout (MTT). Improved resistance alongside rising LDH is not protection, and improved resistance alongside falling viability is not either.
- Keep susceptibility testing ahead of treatment decisions where practical, and set withdrawal intervals before treatment starts rather than after, because residue limits apply regardless of what the susceptibility result shows.
Limitations
The findings come from a single rainbow trout gill cell line studied in vitro, so they describe one epithelial background with no circulation, no mucus turnover and none of the systemic host response. One pathogen was tested. Concentrations are reported as volume percent rather than as mg/L, which makes comparison with bath or feed dosing awkward, and in some cases the effective water concentration will differ from the nominal figure. No in vivo challenge was run, so the step from monolayer to fish is not yet verified. Aq is a multi-component mixture, so no effect can be attributed to a single acid or extract, and the mixture may behave differently from any of its parts taken alone. The SOD and CAT reduction appeared in infected cells, and the study notes that glutathione does not directly regulate SOD or CAT production, so the enzyme changes and the GSH rise are best treated as separate observations. Which route produced the adhesion effect also remains open.
Bottom line
In this gill cell model, Edwardsiella tarda infection was blunted at its earliest steps: less adhesion, less intracellular bacterial recovery, better-preserved barrier resistance and lower cytotoxicity, all without measurable harm to the cells. The host-side signature was a redox and inflammatory shift (glutathione up, H2O2 down, NF-kB p65 activation down, IL-8, IL-1beta and TNF-alpha down) rather than broad immune stimulation. The result worth carrying away is interpretive: lower antioxidant enzyme activity went with a better outcome because the oxidative load had been reduced, and the glutathione gain counts as real buffering because it arrived as H2O2 was falling, not rising.
FAQ
What is Edwardsiella tarda infection
Edwardsiella tarda infection is disease caused by Edwardsiella tarda, a bacterium that persists in water and sediment, survives inside host cells, evades immune clearance and produces systemic disease across a wide range of farmed fish. At the gill, the events that matter earliest are adhesion to the epithelial surface, entry into cells and the loss of barrier integrity that follows. Its economic weight is real but less securely quantified than its pathogenic weight, partly because historical records predate the taxonomic separation of fish-associated Edwardsiella.
How does Edwardsiella tarda infection damage gill epithelial cells
Damage starts with adhesion and internalization and then shows up as loss of membrane integrity, measurable as LDH release, loss of viability by MTT, and a drop in transepithelial electrical resistance, which is the functional sign that tight junctions between cells are failing. Oxidative and inflammatory responses follow: H2O2 accumulates, NF-kB p65 activation rises, and IL-8, IL-1beta and TNF-alpha expression increase. In this model all of those readouts moved in the infected untreated controls and were reduced when the organic acid-based antimicrobial mixture was present.
What inhibitory concentration did the organic acid mixture reach
Using two-fold tube dilution in tryptic soy broth from 8 percent down to 0.015625 percent (v/v) with an inoculum of 1 x 10^5 CFU/mL at 28 degrees Celsius for 24 hours, the minimum inhibitory concentration was 1 percent and the minimum bactericidal concentration was 2 percent. A level of 0.5 percent was taken as sub-inhibitory and used for the mechanistic assays, while the 2 percent bactericidal level was excluded from those assays except in the viability test.
Does the organic acid mixture kill Edwardsiella tarda or only slow it
Both, depending on concentration. Growth was inhibited at 1 percent and cells were killed at 2 percent, so the mixture is bactericidal at the higher level. The more interesting result sits below inhibition: at levels where growth is not stopped, total bacterial adhesion and intracellular bacterial recovery still fell significantly (P < 0.0001). That separation matters practically, because a product used continuously in feed or water will often sit below its inhibitory level rather than above it.
Why did antioxidant enzyme activity fall during infection
Because the oxidative load fell, not because antioxidant capacity was lost. In infected cells treated with the mixture, superoxide dismutase and catalase activity were lower, while H2O2 was lower at every concentration tested (P < 0.0001) and intracellular glutathione was higher. SOD and CAT activity track how much superoxide and peroxide there is to clear, so less substrate means less enzyme activity is required. Read next to H2O2 and glutathione, the drop indicates reduced demand. Read alone, it is ambiguous.
How does glutathione protect gill cells in this model
Glutathione is the main endogenous antioxidant buffer in these cells, and gill tissue needs high levels of it to neutralize excess reactive oxygen species and to protect its own cells from collateral damage while fighting invading bacteria. At 72 hours post-infection the mixture raised intracellular glutathione in a dose-dependent way, with the strongest rise in infected cells (P < 0.0001). The 0.5 percent time course showed glutathione climbing over time without a matching rise in H2O2, which is why the increase was read as a preventive reserve rather than a response to oxidative stress caused by the mixture.
Does the organic acid mixture damage the gill epithelial barrier
No evidence of barrier damage was seen at the concentrations tested. Viability by MTT was not markedly reduced, LDH release was lower rather than higher in both infected and uninfected cells, and uninfected monolayers exposed to the mixture showed limited disruption of transepithelial electrical resistance. Tight-junction transcripts cldn-10e, cldn-30 and zo-1 were higher, not lower, in both infected and uninfected cells, which points toward molecular reinforcement of the barrier rather than erosion of it.
Which inflammatory markers change during Edwardsiella tarda infection
The three measured were IL-8, IL-1beta and TNF-alpha, and all three were significantly reduced by the mixture at every concentration tested (P < 0.0001) in infected cells. NF-kB p65 activation also fell, most clearly during infection. The sequence that fits the data runs from redox to transcription: less H2O2 and more glutathione, then less NF-kB activation, then lower cytokine output. Glutathione is not the same thing as NF-kB inhibition, so the two should not be collapsed into a single mechanism.
Can organic acids replace antibiotics such as oxytetracycline
Not on this evidence. The mixture acted on steps that sit upstream of where an antibacterial such as oxytetracycline works, namely adhesion, entry and barrier integrity, rather than on bacteria that have already established. It was also tested in a cell model with no in vivo challenge, so no claim about replacing drug therapy can be supported. The reasonable reading is complementary use, with antimicrobial susceptibility testing, residue limits and withdrawal intervals still governing treatment decisions.
What does NF-kB p65 activation tell us here
NF-kB p65 activation is the readout that links redox conditions to inflammatory transcription. In this model it fell with mixture exposure, most clearly during infection, and was measured as absorbance at 450 nm per microgram of nuclear protein. Its value is as a bridge: oxidant signals can feed into NF-kB and drive IL-8, IL-1beta and TNF-alpha, so lower activation alongside lower H2O2 and higher glutathione forms a coherent picture. The direction of that interaction is context dependent elsewhere, so the link here is an association rather than proof.
How should susceptibility testing guide Edwardsiella tarda treatment
Susceptibility testing should decide which compound is used and at what dose, and ideally it runs before the first treatment rather than after a failure. In practice results often arrive late, because treatment is triggered by clinical signs and group-level risk. Where testing is available it also builds the resistance record that makes later empirical choices defensible. Compounds such as oxytetracycline remain treatment tools, and withdrawal intervals and residue limits apply regardless of what the susceptibility result shows.
What are the main limits of this Edwardsiella tarda infection model
Four limits stand out. It is a single rainbow trout gill cell line studied in vitro, so circulation, mucus turnover and the systemic host response are absent. One pathogen was tested. Concentrations are reported as volume percent rather than as mg/L, which complicates translation to bath or feed dosing. No in vivo challenge was performed, so the step from monolayer to fish is not yet verified, and because the mixture is multi-component, no effect can be assigned to a single acid or extract.


