MurJ is a bacterial flippase — a protein that moves peptidoglycan precursors across the cell membrane, providing the building blocks for the cell wall. Without MurJ flipping these molecules from the inner leaflet to the outer leaflet, the wall cannot be built, and the bacterium dies. The protein works by alternating between inward-facing and outward-facing conformations, shuttling its cargo through a structural rocking motion.
Li et al. (Nature, February 2026) show that three unrelated bacteriophages — viruses that infect bacteria — have independently evolved small proteins that disable MurJ in the same way. The proteins, called single-gene lysis proteins (SglM, SglPP7, and SglCJ3), each attach to a groove in MurJ and lock it in its outward-facing conformation. The rocking stops. No cargo crosses. The wall stops being built, and the cell dies.
What makes this remarkable is the convergence. The three phage proteins share no sequence homology — they evolved independently in unrelated viral lineages. Yet cryo-electron microscopy reveals that all three bind essentially the same site and freeze the transporter in essentially the same state. Three different evolutionary trajectories, separated by vast phylogenetic distance, arrived at the same structural answer.
This is evolution performing target validation. A human drug discovery program identifies a promising target, tests compounds against it, and hopes the target matters enough to kill the pathogen reliably. Evolution did the equivalent — independently, three times, in organisms under the most intense selection pressure imaginable (replicate or die within minutes). When three unrelated lineages converge on the same vulnerability, the signal is stronger than any single experiment could produce. The convergence is the evidence.
The outward-facing conformation in which MurJ is locked also happens to be the one exposed to the cell's exterior environment. A drug targeting this state would be attacking an accessible surface, not a buried interior. The phages found not only the right target but the right conformation of the right target — the one most amenable to external intervention.
The general principle: when you want to know which targets are real, check where evolution has aimed more than once. A single adaptation is a data point. Convergent adaptations from unrelated lineages are a consensus. The lock that three independent keys were built to turn is, by that evidence alone, the lock worth picking.