Humanizing Protein Splicing

JEWELS FROM BOWLING BALLS

The road to the discovery wasn't a clear one for Yang and Hanada, however. They spent months trying to identify the epitope mounted on the major histocompatability complex I (MHC I) on the surface of the regressing tumor cells. Yang's lab had isolated the CD8⁺ T lymphocytes that had been attacking the tumor, but the usual bioinformatics tools that they used to determine the sequence of the epitope weren't working. They knew that the T cells recognized a peptide fragment of a protein created from the fibroblast growth factor 5 (FGF-5) oncogene, but no matter how hard their computer crunched the numbers, they couldn't determine the exact sequence of the antigen.

So Hanada fragmented the FGF-5 gene and transfected those fragments into an HLA3⁺ tumor not originally recognized by CD8⁺ cells. They found that the smallest chunk of FGF-5 recognizable by the CD8⁺ cells was a peptide of 60 amino acids that could create the epitope for his cancer sample.

The size of that chunk presented a basic problem. As Yang describes it, an epitope is like a jewel and the MHC I molecule is like a ring setting, and it is these two joined entities that must be recognized by a T-cell's receptor. But most epitopes are no longer than 13 amino acids. "Having an epitope with a string of 60 amino acids is like trying to mount a bowling ball on a wedding ring," says Yang. Because the T-cell clone could also recognize a synthetic version of the 60-amino-acid peptide processed in vitro, the final epitope processing had to be occurring posttranslationally. "We looked at every known processing method, from phosphorylation [to] glycosylation, but there was no evidence for any of that," Yang says.

The next step involved determining which of the 60 amino acids were necessary for epitope recognition. To do that, Hanada conducted an alanine scan. He substituted every non-alanine amino acid with alanine and every alanine with glycine, inspecting each amino acid for its role in epitope construction. The alanine scan revealed that of the 60 original building blocks, only seven were required, three at the amino terminus and four at the carboxyl end. That's when coauthor Jonathan Yewdell suggested that some sort of splicing was going on.

SPLICING'S ROOTS

To get a better understanding of protein splicing, Hanada studied the yeast literature. Saccharomyces cerevisiae researchers have identified dozens of examples of protein splicing.² The event Yang and Hanada had observed, however, didn't mesh with the yeast model. Most yeast protein splice events are autocatalytic: Proteins do the cutting and re-ligation in a self-regulated process that's more analogous to the exon/intron motif of RNA splicing. Generally, a nonfunctional protein is made functional by excision of a so-called intein and ligation of flanking exteins without cofactors. "We were trying to fit our sequence into that same model, and it just didn't work no matter how hard we pushed it," says Yang.

Then he found a 10-year-old plant proteomics paper.³ A protein in the jackbean plant was spliced by the proteasome, which normally destroys proteins. In this case the enzyme was working backwards to splice them. Instead of lopping out nonsense segments as in the yeast proteins, the plant was using splicing as a form of transpeptidation, or the creation of a new protein. "It was like driving a peptide bond hydrolization in reverse," says Yang.

The transpeptidation model dovetailed with the evidence they saw from their alanine scan. "The real genius of this work is that they were able to take a plant paper and apply it to human immunology," says van Endert.

Most scientists would have given up on the project much earlier, says Nilabh Shastri, a researcher and professor at the University of California, Berkeley. "It was the clever thought process and the extremely hard work that made their paper so impressive."

UNEXPECTED PLAYER

A few months after the breakthrough, Hanada presented some preliminary results at the Keystone conference in Utah early in 2003. Pierre Coulie at the Ludwig Institute, Brussels, brought back word to fellow principal investigator Benoit Van den Eynde and graduate student Nathalie Vigneron. They had a thought: Maybe the mechanism that Hanada described was at work with a gp100 melanoma epitope, a sequence that had eluded all their efforts to identify. Van den Eynde and Vigneron followed the same steps as Hanada and soon found their epitope's sequence. Then they went further and put a fragment of the precursor protein into a test tube filled with proteasomes. The peptide immediately spliced itself just as it did in the cell.

That experiment proved an essential point about this new protein-splicing process: It occurs in the proteasome, the massive barrel-shaped protease often called the garbage disposal of the cell.⁴ "This opens up the possibility that the proteasome, which was only discovered about a decade ago, is not only degrading proteins but it's also remodeling them," says Van den Eynde.

Topping the list of outstanding questions is the matter of significance. Derek Lowe, a senior researcher for a major pharmaceutical company and the editor of In the Pipeline, a Web site about drug research, says, "Right now it's of interest to cancer immunologists and people working on cancer vaccines. But I'm sure that every pharmaceutical executive read the paper and filed it away in the 'could be important, see what develops' folder."

Meanwhile, Yang is in the process of establishing a Phase I/II cancer vaccine trial based on the CD8⁺ cells he isolated from the original regressing tumor. The trial will enlist only about a dozen patients with renal cancer. Its commercial prospects, he says, are dim as it will probably work only among a tiny population of patients with cancer.

The impact of the research, however, is being felt more in basic research than in the clinic. Yang cautions that the significance of his discovery is still to be determined. "It might just be an interesting footnote," he says. "But if it turns out to be much more common, it could alter our basic understanding of the biochemistry of proteins."

Sam Jaffe can be contacted at Sjaffe@the-scientist.com.