ISKNV Antiviral Screening: How Molecular Docking Found 3 Compounds That Beat the Fish Virus

There’s a quiet killer moving through fish farms and ornamental tanks, and most people don’t notice it until the losses are already counted. It’s called Infectious spleen and kidney necrosis virus, or ISKNV. It’s a large DNA virus in the iridovirus family, and it shows up wherever fish are packed together. Researchers have known for years that ISKNV is dangerous. The hard part has always been doing something about it once it arrives.

This piece looks at a screening effort that took a practical shortcut around one of drug discovery’s worst bottlenecks. Instead of testing thousands of compounds in living cells, one at a time, the team used computer models of viral proteins to rank candidates first, then checked the winners in the lab. Two compounds, lycorine and fludarabine, kept beating ISKNV in fish cells. The workflow itself is the real prize, because other labs can reuse it for different aquaculture viruses.

What ISKNV is, and why fish farmers should worry

ISKNV is a double-stranded DNA virus in the genus Megalocytivirus. It infects many fish species and causes systemic disease: swollen spleen and kidneys, odd swimming, tissue death, and often heavy death tolls. It was first found in mandarin fish in China. Since then it has spread across Asia and turned up in India, Australia, Germany, the United States, and Ghana. Imported ornamental fish are a known way it crosses borders, which is why one farm’s problem can quietly become everyone’s problem.

The economic damage is real. Infected stock can die fast, and there is no approved antiviral to stop an outbreak once it starts. Vaccines exist for related iridoviruses and help some marine species, but they prevent infection. They work best before exposure, they depend on the fish and how you give them, and they do little for fish already sick during an active outbreak. That gap is exactly why hunting antiviral candidates is worth the trouble.

Why fish antivirals are slow to find

Any antiviral is slow and costly to discover. You take a library of candidate molecules, expose infected cells to each, and measure the outcome. Large DNA viruses like ISKNV make it worse. They need long incubation in fish cell lines, every compound must be toxicity-checked first, and you need several readouts (gene expression, infectious titer, cell damage) before you trust a result. Running that full pipeline on hundreds of compounds is out of reach for most labs.

That’s where virtual screening earns its keep. Molecular docking predicts how a small molecule might sit in a protein’s binding pocket, like a key in a lock. It doesn’t prove a drug works, but it’s fast and cheap, and it’s good at turning a long list into a short one worth testing for real.

Choosing the right viral targets

ISKNV carries over 120 predicted genes, so the first real question is which proteins to aim at. The team picked five that sit at the center of the virus’s life cycle:

  • DNA polymerase, which copies the viral genome
  • Transcription elongation factor (TFIIS), which helps read out viral genes
  • ATPase, an energy-handling enzyme tied to viral structure
  • An ankyrin repeat protein involved in host interaction
  • The major capsid protein (MCP), the shell that builds new particles

Hitting replication, transcription, structure, and host interaction together gives a wider shot than betting on a single protein.

Building the models and running the screen

ISKNV proteins aren’t all crystallized, so the team predicted their 3D shapes with AlphaFold2 (through ColabFold). They cleaned up those models, mapped the likely drug-binding pockets with SiteMap, and built docking grids for each target. Then they docked a library of 72 compounds with known or suspected antiviral activity, a mix of approved drugs, experimental molecules, and natural products, into each protein using Schrödinger’s Glide software.

For every target they pulled the top 20 compounds by docking score, then compared the lists across all five proteins. Only three appeared in the top 20 for every target: fludarabine, liquiritin, and lycorine. That consistency across targets, not one big score against a single protein, is what put them on the shortlist. A molecule that looks good against many viral proteins at once may work more reliably than one that merely tops a single chart.

Testing the shortlist in fish cells

Shortlists only matter if they survive the lab. The three candidates were tested in dwarf gourami fin (DGF) cells, a fish line that ISKNV infects. First the team checked toxicity, so any antiviral effect wouldn’t just mean the cells died. Lycorine was the fussiest: toxic at higher doses, so they used it at 0.5 µg/mL where cells stayed healthy through the long infection. Fludarabine and liquiritin were tested at 100 µg/mL.

They ran several assays side by side:

  • Viral gene expression (real-time PCR for MCP, ATPase, DNA polymerase)
  • Infectious titer (TCID50, measuring actual replicating virus)
  • Cytopathic effect over time (cell damage under the microscope)
  • Extracellular viral load over time (virus released into the medium)

What the lab actually showed

All three compounds lowered viral gene expression, but how deep and how long that lasted varied a lot.

Lycorine was the clear winner. It suppressed MCP, ATPase, and DNA polymerase expression by about 99.8%, 99.7%, and 85.5%. More telling, that genetic drop meant fewer real viruses: infectious titers fell roughly 81-fold versus infected untreated cells. Across the time course, lycorine kept virus release low and the cells looked almost normal, just mild enlargement even late in infection. Lycorine is a natural alkaloid from the daffodil family with a history against other DNA and RNA viruses, so a hit here fits what we already knew.

Fludarabine held up too. It blocked all three viral genes by 86 to 92% and cut infectious titers about 39-fold. The cell layer mostly held together through the window, though some cells looked enlarged or condensed. Fludarabine is a purine nucleoside analog already used in human medicine, and it interferes with DNA synthesis. That lines up with attacking a DNA virus that has to copy its genome.

Liquiritin was the weak one. A flavonoid from licorice, it lowered viral gene expression early (around 50 to 56% at 72 hours) but didn’t sustain it. By later points its viral load and cell damage looked close to the infected control. The docking model liked it; the biology wasn’t convinced. That gap is the point of the whole exercise: a good docking score is a hypothesis, not a verdict.

Why aiming at many targets paid off

The smartest move here was refusing to crown a winner per protein. By favoring compounds that ranked well across all five targets, the team cut the odds of being fooled by one fluke score. The lab results backed that logic. The two compounds that looked broadly promising in docking, lycorine and fludarabine, were also the two that performed across gene expression, infectious titer, virus release, and cell protection. Liquiritin, friendly docking profile and all, fell apart under sustained infection.

It also matters that they measured more than one thing. A single snapshot of gene expression can look reassuring while infectious virus keeps pouring out. Liquiritin shows the trap: the genetic readout and the functional one (does the cell still make infectious particles?) told different stories. Real antiviral activity has to survive that cross-examination.

The caveats

This is a promising screen, not a finished therapy. A few things weigh on the result:

  • The study never proved biochemically that the compounds bind the predicted proteins. Docking shows where they might fit; direct binding wasn’t shown.
  • Everything ran in one fish cell line, at a narrow range of doses, in a dish. A whole fish is a different question.
  • Even the best performers only reduced viral load. They didn’t eliminate it under these conditions. The effects were real but partial.
  • A direct comparison with compounds known to work (or known to fail) against ISKNV would help show how good the docking predictions really are.

None of that sinks the result. It just sets the next steps: dose-response work, confirmation at the protein level, and challenge tests in live fish.

What this means for aquaculture research

The lesson isn’t “we found two fish antivirals,” not yet. It’s that a cheap computational prescreen can shrink an unmanageable library down to a few candidates worth the expensive lab work, and doing it against several viral targets at once is a sound way to dodge false leads. For a field where testing is slow and fish-cell assays are a grind, that shortcut is the whole game.

Lycorine and fludarabine now lead the queue for deeper validation against ISKNV. If later studies confirm their activity and sort out safe dosing, they could become the first real anti-ISKNV treatments, something you reach for during an outbreak instead of only preparing for the next one.

Comments are closed