A team of structural biologists has taken the first atomic-scale picture of the protein clumps that slowly cloud the cornea in an inherited eye disease, and used that picture to design short peptides that jam the clumps' growing ends.
The work, published in the Proceedings of the National Academy of Sciences ↗ by a UCLA group including Michael Sawaya and David Cascio, goes after a family of corneal dystrophies. These are inherited diseases in which the clear front window of the eye gradually turns hazy and vision fades with it. A well-studied subset, lattice corneal dystrophy, is caused by mutations in a protein called TGFBIp, short for transforming growth factor beta-induced protein. In these patients the mutated protein misfolds and stacks up in the cornea as amyloid, the same class of insoluble, fibrous protein deposit seen in diseases like Alzheimer's, except here it collects in the eye. Until now no one knew what those corneal amyloid fibrils actually look like at the level of individual atoms.
To find out, the team used cryo-electron microscopy, a method that flash-freezes molecules and images them by the thousand to reconstruct a three-dimensional shape. They solved the structure of fibrils grown from a fragment of TGFBIp carrying a dystrophy-linked mutation named V624M. The fibril turned out to be built from a specific stretch of the protein, the residues numbered 569 to 609. That stretch winds into two mirror-image strands, called protofilaments, with a narrow channel of water running down the middle where the two strands meet.
One detail matters more than the rest. The ordered core of the lab-grown fibril includes a segment, running from residue 571 to residue 588, that earlier work found enriched in the amyloid actually scraped from patients' corneas. That overlap is the anchor. It is the difference between a shape a protein happens to form in a tube and a shape that resembles the disease. It tells the researchers they were looking at something worth designing against.
So they designed. Amyloid fibrils grow by adding units onto their ends, the way a crystal lengthens, so the ends are the vulnerable point. Because the two mirror-image strands in this fibril do not grow at the same rate, the team could aim inhibitors at the faster end. They built two peptides, called G1 and H4, meant to bind onto a fibril end and cap it, blocking further elongation. In the test tube the two peptides cut aggregation in a dose-dependent way, meaning more peptide gave more effect. That was measured three ways. One used thioflavin T, a dye that lights up when it binds amyloid. Another separated soluble protein from clumped protein. The third looked directly under the electron microscope.
The caveats are large and worth stating plainly. Every result here is in a tube, using a recombinant fragment of the protein rather than a whole eye, a cell, or an animal, let alone a person. Thioflavin T is a screening dye, and slowing fibril growth in a cuvette is many steps removed from clearing deposits from a cornea or restoring sight. The structure comes from one mutation, V624M, while the corneal dystrophies span dozens of TGFBIp mutations, and a fibril that forms in a lab over days may not match one that builds in an eye over decades. Nobody has shown how you would deliver a peptide into the dense, avascular tissue of the corneal stroma. G1 and H4 are research tools, not drugs.
What the paper delivers is a method and a target, not a treatment. Corneal dystrophy is currently managed by scraping or lasering the surface, a procedure called phototherapeutic keratectomy, or by transplanting the cornea outright. The deposits tend to come back even in a graft because nothing stops the underlying protein from clumping. A molecule that binds the fibril itself would be a different kind of tool. Peptidemodel does not host a card for G1 or H4, and it should not, they are one-off experimental sequences rather than clinical candidates. But the logic on display, solve the structure of a disease-causing aggregate and then draw a peptide to fit its weak point, is the same structure-guided design that is starting to turn up across peptide therapeutics. Here it is pointed at an eye disease that has had no molecular option at all.