"That's my second-favorite organ."

They were married two years later.

It was 1985. Still in school, Maguire was working in a lab that specialized in retinal degeneration, and he knew that many such diseases were the result of a single defective gene. With the practical, fix-it approach common to those in the surgical profession, Maguire wondered if matters could be addressed in the operating room.

"I thought: 'Simple problem: fix gene, fix disease,' " he said.

Of all the body's organs, the eye is one of the most accessible. So he asked his wife, who already had a Ph.D. in biology: Was it possible to inject a corrective gene?

Sure, she said - in theory. Researchers had been talking for years about the goal of inserting genes into living tissue.

Except that no one had identified any of the genes that cause retinal disease.

"Had I actually known how difficult and complex the problem really was, I probably would have dismissed the idea immediately," he says now. "Jean has a more long-term view and didn't tell me."

The two did a stint at Yale, where Bennett had grown up, the daughter of a famous physicist who co-invented the gas laser.

Later, Maguire interviewed for a fellowship with Robert Machemer, a pioneer in retinal surgery, and eagerly told him what he and his wife were trying to do. The response was not good.

"It'll never work," Machemer said.

Maguire, a Philadelphia native and Episcopal Academy graduate, didn't get the fellowship, but the rejection inspired the couple. They came to Penn's prestigious Scheie Eye Institute in 1992.

The dogs

The notion that eye problems could be inherited dates back to Aristotle. But the first gene for retinal degeneration was not discovered until 1989. Today, more than 400 eye-disease genes have been identified, aided by the Human Genome Project.

At Penn, Bennett studied the genetics of retinitis pigmentosa - a group of inherited diseases that gradually destroy sight.

In 1997, scientists at the National Eye Institute found a mutation that caused a rare subtype that strikes children, called Leber's congenital amaurosis (LCA). They began to try to induce these mutations in mice in order to figure out how to fix them.

Then, in 1998, Cornell University's Gregory Acland and Gustavo Aguirre - now at Penn - reported that the same mutation occurred naturally in a certain breed of dog, the Briard.

In 2000, Bennett and Maguire joined them in a pivotal experiment. They injected three blind dogs with a virus - one that had been modified to deliver the recipe for a missing enzyme in their eyes.

Within weeks, the dogs' vision improved more than anyone had hoped. They could see so well they could navigate a maze.

By 2006, Maguire had performed the surgery on 55 blind dogs, restoring vision in more than 90 percent of them.