Capture the cellular flag

For any game of capture the flag, you need patience. Many hours might be spent biding one's time and waiting for the right moment. Dr. Edus Houston ("Hootie") Warren has the makings of a champion — his hunt to find a leukemia flag took almost a decade.

Warren's study, published in the Sept. 8 issue of Science, describes how cells can patch together different sections of protein to create brand-new protein sequences. Commentaries in Science and the Sept. 8 online edition of Nature describe the protein products as "mix and match" or "scrambled and stitched." Other publications dubbed it the "Frankenstein protein." But behind the hoopla is a tale complete with twists, turns and a triumphant finish.

"We totally stumbled across it," Warren said of his work with colleagues in the Clinical Research Division's Immunology Program. "We never meant to study this."

The odyssey begins

The story starts in 1996, when Bill Clinton was campaigning for a second term against Sen. Bob Dole and Ross Perot, and "Twister" was in theaters. Warren set out to find a cellular flag associated with a mysterious immune reaction that helps cancer patients who undergo bone-marrow transplants. Known as the graft-vs.-leukemia effect, it happens when T-cells from the bone-marrow donor start destroying leukemia cells in the recipient. The process happens only in some patients, and is poorly understood.

"If we could understand what these [donors'] T-cells see, then maybe we can exploit their ability to kill leukemia cells," Warren said. "That's why we embarked on this odyssey."

Beginning as a postdoctoral researcher in Dr. Stanley Riddell's Clinical Research Division lab, Warren set out to learn how the transplanted cells recognized the patient's leukemia. He was looking for a peptide flag — one among tens of thousands poking out of the surface of a typical cell. Peptides displayed on the cell's surface are fragments of proteins normally found inside the cell. Just like flags or collage colors, they broadcast something about the bearer — whether it's one of the body's own cells, for instance, or whether the cell is infected.

Never give up

After a year and a half of work, they were close. The researchers had narrowed their target to a stretch of 20 amino acids of one protein. But surface peptides range from just eight to 11 amino acids long. Somewhere, inside the 20-amino-acid stretch, an answer was staring them in the face.

Except that it couldn't be found.

"We really pounded that one," Warren said. "If I had a nickel for every experiment I did trying to identify whether a shorter stretch of that 20-mer was recognized, I'd be a rich man."

By then, the researchers had spent years working on a seemingly fruitless project. One of the article's reviewers later wrote that at that point most sane people would have given up.

"We were really bummed," Warren said. But he didn't quit.

Instead, Warren tried a different tack. He sequenced DNA from 70 different leukemia patients, and showed that a single nucleotide difference in the section coding for that same 20-amino-acid stretch predicted perfectly whether that patient's cells would provoke a T-cell response.

"The other line of investigation had come to a dead end," Warren said. "But the genetic evidence that this was the right gene and the right protein was incontrovertible and convinced me this was not a fluke."

Then a phone conversation with a colleague at the National Institutes of Health provided a spark of inspiration. The government scientist happened to be studying a similar process and was on the cusp of making a discovery.

"I'll never forget his comment," Warren said. "He said, 'Well, Hootie, your own data show that you need both ends of that 20-mer, so what else can it be?' Basically, he was just challenging me to think outside the box."

It was unheard of to take two separate bits of a protein to make a peptide destined for the cell surface, but that seemed to be the suggestion. Warren walked down the hallway to Riddell's office. Riddell was skeptical, but agreed that Warren should give it a try.

So Warren took a few amino acids from the beginning of his 20-amino-acid sequence and a few from the end. Over the next two years, he used his knowledge of peptides to try to create something that would work.

"I know that 20-residue stretch so well, I can say it backward and forward and inside out," Warren said. "I spent hours and hours just thinking about it."

He also spent thousands of dollars synthesizing possible peptides, helped by grants from the Damon Runyon Cancer Research Foundation and the NIH. Finally, he got some weak response from his T-cells — enough to suggest they were onto something, though not strong enough to make them think they had their answer.

The magical 'aha'

Again, the trail went cold.

The final breakthrough took place across the Atlantic. In 2003, Warren took a seven-month sabbatical in Belgium, where he continued working on his elusive peptide. His last day is still the most memorable: He was giving a short recap seminar before catching a plane back to Seattle. During the presentation, he outlined his frustrations with the peptide search.

"One of the things that bugs me about that sequence, is that it doesn't satisfy the sequence motif," Warren told the audience, referring to a set of rules the peptide must obey.

A researcher named Pierre Coulie had been dozing at the back of the room. At that point, he looked up and asked Warren to repeat the sequence motif. Next, he casually inquired whether flipping the order of the two stretches wouldn't satisfy the requirements.

"That was the most magical moment," Warren said. "The light bulbs went on over everybody's heads — in a nanosecond, we all knew what it was."

The peptide is made from four amino acids at the beginning of his 20-amino-acid stretch, stuck to six amino acids from the end. But the six amino acids from the end come before the ones at the beginning.

"I had assumed that the two pieces would get spliced back together in the same order, which in retrospect is totally unwarranted, but it shows you how you can get limited by your assumptions," Warren said.

Back in Seattle, Warren quickly showed the new peptide provoked a T-cell response from the leukemia donor's T-cells. He and his colleagues suspected his peptide must be made in the proteasome, a barrel-shaped organelle that chops up proteins and is known as the cell's "chamber of doom." To prove it, they put purified proteasomes in a Petri dish, fed them the 20-amino-acid precursor peptide, and showed that indeed the barrels were able to spit out the smaller, rearranged peptide.

Another research group had shown in 2004 that some kidney cancer cells displayed spliced peptides on their surface. The new paper extends that to splicing and shuffling, and in healthy human cells. This implies all mammalian cells have the power to take two separate stretches of protein, shuffle them around and create a new protein.

It's not known how often cells make spliced proteins, or whether splicing proteins could somehow benefit the cell. Also unknown is whether a cell could splice together pieces from separate proteins. But the mere idea that a peptide can be made of two separate amino-acid stretches affects immunity.

"What if cells that are infected with HIV could take the HIV proteins and create spliced, rearranged peptides?" Warren said. "That would have enormous implications for making an HIV vaccine."

Riddell credits his colleague's tenacity when facing what seemed like an unsolvable problem. The obstacles he encountered, he said, make the result all the more valuable. "This tells us now is that there's a whole other class of potential antigens that could be targets for cancer immunotherapy, vaccine development or viral immunotherapy," Riddell said. "The potential array of targets is much larger than we had anticipated."