Dry-eye damage in mice drained their corneas of a nerve-made peptide called CGRP, and putting the peptide back as an eye drop resealed the surface. The same peptide is what an entire class of migraine drugs is built to block.
That is the setup in a study published August 3 in Investigative Ophthalmology and Visual Science ↗, from the Eye Institute of Xiamen University and its affiliated Xiamen Eye Center in Fujian, China. The group led by Jintao Shi and Zhirong Lin was not testing a migraine drug. They were asking a narrower question: what does the sensory nerve of the cornea contribute to keeping the eye's outermost layer sealed, and what happens to that seal when the nerve is hurt.
CGRP, calcitonin gene-related peptide, is a small signaling molecule released by sensory nerve endings. It is best known outside the eye, where it dilates blood vessels and carries pain signals, and where blocking it is now standard migraine care. Inside the cornea, the densest sensory nerve supply in the body, CGRP is one of the messengers those nerves release into the surrounding tissue.
What the study did
The researchers damaged the ocular surface of mice with benzalkonium chloride, a preservative used in eye drops that is a standard way to induce dry-eye disease in the lab. They also stressed cultured human corneal epithelial cells, a line called HCE-T, with a salty, high-osmolarity solution. That is the cellular version of a dry, evaporating tear film.
Two things happened together. The corneal nerves were damaged, and the amount of CGRP in the tissue dropped sharply (a highly significant fall, at the strictest statistical threshold the paper reports). At the same time, the cells cranked up production of the receptor machinery that reads CGRP, the components named CRLR, RAMP-1, and CRCP. A tissue starved of a signal turning up its antennas for that signal.
Then they gave it back. Four days of a topical CGRP solution, dosed at 50 micromolar, restored the barrier. Less Oregon Green Dextran, a tracer dye, leaked across the treated corneas. That is a standard readout of how tight the surface is. The transepithelial electrical resistance across the cell layer rose too, another measure of a well-sealed sheet. The tight-junction proteins that act as the grout between surface cells stayed where they belonged instead of scattering.
The how had two arms. CGRP switched on the PKA and CREB signaling pathway, a growth-and-repair program, and switched down NF-kB, a master inflammation switch. Downstream, the inflammatory messengers IL-1-beta, IL-6, and TNF-alpha fell, along with MMP-9, an enzyme that chews up the tissue scaffolding. So the peptide both rebuilt the barrier and quieted the inflammation that was tearing at it, in mouse eyes and in human cells alike.
The tension worth naming
Here is what makes this more than one more preclinical wound-healing paper. The molecule doing the repairing is alpha-CGRP ↗, and blocking alpha-CGRP or its receptor is exactly how modern migraine prevention works. Erenumab, sold as Aimovig, sits on the receptor. Fremanezumab, galcanezumab, and eptinezumab mop up the peptide itself. The oral gepants like rimegepant and atogepant plug the receptor from the inside. Millions of people take these drugs to turn CGRP down.
The eye has already handed peptidemodel a signal pointing the same way and a signal pointing the opposite way. A large migraine cohort found that CGRP inhibitors were linked to a lower risk of glaucoma ↗, a disease of the eye's interior and its optic nerve. This new work says that on the ocular surface, the front window rather than the back, the peptide is protective and draining it is part of the damage. One molecule, two compartments, two directions.
That does not mean migraine patients are drying out their eyes. Real-world safety reporting on the anti-CGRP antibodies has not flagged dry eye as a prominent problem. And a systemic antibody is a very different exposure from a mouse cornea deliberately stripped of the peptide. CGRP is also a double-edged messenger elsewhere, driving the neurogenic inflammation that makes some tissues worse, so more is not automatically better. The honest read is that this is a mechanistic question the biology now raises, not a side effect anyone has measured.
It also lands in a small but real effort to treat the ocular surface with peptides directly. A thymosin beta-4 eye drop missed its Phase 3 in neurotrophic keratitis ↗ earlier this year, when the placebo arm healed better than expected. Corneal repair is easy to show in a dish and hard to prove against a control. A CGRP eye drop is a mouse result and a set of human cells, not a trial. The question it sharpens is worth a trial.