A virus's protease has two jobs. Slice its own polyprotein into the pieces it needs to build new copies of itself, and slice up the host's innate immune sensors so the cell can't yell for help. Seneca Valley Virus, a picornavirus that infects pigs and that humans are studying as an oncolytic agent in solid tumors, runs both jobs through a single enzyme called 3C protease. The same active site that processes the viral polyprotein also cleaves cGAS (the host enzyme that detects foreign DNA), gasdermin A (the pore-former that drives inflammatory cell death), and pro-IL-1 beta (the pre-cursor for one of the cell's main alarm signals). Block 3C and you block both jobs at once.
A new paper published online June 1 in PLOS Pathogens ↗ reports a ten-amino-acid peptide that does that. The lead, named P5, is a substrate-competitive decapeptide that binds the catalytic histidine at position 48 of 3C protease through hydrogen bonding. With P5 docked, 3C cannot cleave cGAS, gasdermin A, or pro-IL-1 beta. The host's DNA sensor reassembles into the droplet it needs to function, and downstream interferon signaling comes back online.
The screen
The group built a fluorescence biosensor whose two halves snap together when 3C protease slices a substrate sequence between them. When 3C is active, the two halves come apart and the signal drops. When 3C is inhibited, the signal stays on. They call it a dimerization-dependent red fluorescent protein, or ddRFP, biosensor system. It is a clean readout for a protease blocker because the signal scales with how much 3C activity is left in the well, not with how much peptide is in the well. False positives that bind the substrate without inhibiting the enzyme do not flicker the readout.
P5 was the hit. A decapeptide, meaning ten residues, derived from the natural substrate region 3C usually recognizes. That is the standard handle for designing a substrate-competitive inhibitor. Build something that looks enough like the cut site for the enzyme to grab it, but not enough like it to actually get cleaved. The peptide sits in the active site, takes up the room a real substrate would occupy, and refuses to leave.
The contact point is the catalytic residue itself. 3C protease is a cysteine protease whose chemistry runs through a Cys-His-Glu triad. P5's hydrogen-bonding network reaches His48, the histidine that has to be free to deprotonate the active-site cysteine before each cleavage event. With that histidine engaged, the catalytic cycle stops.
What the rescue looks like in cells
cGAS is the cell's main detector of cytosolic DNA. When DNA shows up where it shouldn't (a virus dumping its genome into the cytoplasm, for instance), cGAS binds it and the cGAS-DNA complex undergoes liquid-liquid phase separation. The cytosol forms small protein-rich droplets, like vinaigrette beading on a counter, that concentrate cGAS together with its DNA cargo. Inside those droplets, cGAS is catalytically active and makes the second messenger that turns on the interferon program.
SVV's strategy is to cleave cGAS before the droplets can form. A cleaved cGAS does not phase-separate. No droplet, no interferon, no antiviral response. The paper shows this directly. Adding 3C protease to porcine cGAS in the presence of DNA shut down droplet formation. Adding P5 alongside the protease let the droplets reform.
The downstream signal followed. P5 enhanced cGAS activity in cells under SVV pressure and restored type I interferon signaling. The same competitive block protected gasdermin A and pro-IL-1 beta from cleavage in parallel assays. One peptide, three substrates rescued, because all three are cut by the same enzyme at the same site.
What this is not
It is not a drug. P5 is a research-grade peptide tested in cells, with the standard early-stage notes: favorable membrane permeability, low cytotoxicity at the concentrations used, and stability good enough for the assay window. None of that survives unchanged into a porcine challenge model, let alone a human one. The team writes "translational potential" because the work was done in porcine systems with porcine cGAS, porcine gasdermin A, and porcine pro-IL-1 beta. Swine is the relevant species, since SVV is a swine pathogen, but the human-oncolytic-virology version of this story would need the same screen rerun against human substrates.
It is also not the first 3C protease inhibitor. The picornavirus field has been after this enzyme for decades. The interesting move here is the use of cGAS-DNA phase separation as a functional readout. Treating the droplet as the assay endpoint, rather than just the cleavage band on a gel, ties the protease inhibition directly to a measurable host-defense state. Other viral proteases that cleave host immune sensors (3CL pro in coronaviruses, NS3 in flaviviruses) are obvious next targets for the same readout.
Why peptides for this
The 3C active site is small and substrate-shaped. The enzymes evolved to recognize a few residues on either side of a cleavage point and to slice in between. A decapeptide built from that recognition region matches the geometry of the binding pocket the way a key matches a lock. Small molecules can hit this kind of site too, and the COVID era produced several. Peptides have the offsetting advantages of being faster to iterate (the team here screened a peptide library against the ddRFP biosensor in cells) and easier to predict from substrate sequence. The trade is the standard one. Better starting geometry, worse oral bioavailability. For a viral inhibitor intended for inhaled or injected delivery in animals, that trade is reasonable.
The substrate-mimic strategy is also clean about specificity. P5 was derived from the SVV 3C cleavage pattern, so its activity should be highest against 3C-like cysteine proteases that share the same substrate preferences. Whether it cross-reacts against host cysteine proteases (cathepsins, caspases) is the obvious follow-up assay the paper does not run. That is the next thing to look for.