by David Hart
Will Taylor hadn't planned on a gene hunt.
As part of his research into enzyme-blocking drugs that control HIV replication, he ran across a handful of signs that put him on the trail of a different quarry.
First, there were the pseudoknots. These twisting structures in RNA -- the genetic material of HIV and other retroviruses -- can cause hiccups in the process of translating the RNA code. These hiccups, called frameshifts, can even shift translation of the genetic code into an entirely different gene, which means a different protein will be constructed.
The challenge of finding potential pseudoknots and frameshift sites hidden in long stretches of genetic code led Taylor, a UGA associate professor of medicinal chemistry, to develop some innovative methods.
"My most useful tool in some cases is just colored pens," Taylor said. "Some of the early work was done at a pretty primitive level -- folding RNA in my head, looking for possible frameshifts or pseudoknots. It gets pretty complicated, and having different colored pens has been useful."
In the HIV code Taylor was examining, these pseudoknots cause shifts into RNA regions that, at first glance, look like genetic "junk," regions which don't contain instructions for building proteins. Scattered throughout these regions are numerous signs that normally halt the translation of RNA into proteins and tell geneticists to stop looking for genes. One particular stop sign, called a codon, reads "UGA" -- genetic shorthand for the three-nucleotide sequence of uracil, guanine and adenine.
"I guess it was my own naivete, or maybe just being kind of boneheaded or persevering," Taylor said. "Even though the stop codons were there, I took some fragments, translated them and looked at what the amino acids would be. I started hitting protein databases with those fragments, and they came back with some tantalizing things."
In the process, he came upon the final clue: rarely, and under specific conditions, the "UGA" codon is not a stop sign at all; instead, it tells the RNA to insert selenocysteine (SeC), a selenium-containing amino acid.
"As soon as I latched onto that idea, everything fell into place," Taylor said. By following the "UGA" signs like an alumnus to a home football game, he eventually located six genes or gene variants in HIV that potentially could construct selenoproteins.
He chased "UGA" down a trail that led to the genetic code for human T cells. There, Taylor saw familiar signs for inserting SeC into proteins.
In RNA, there are three possible stop codons, and each should appear randomly throughout true junk RNA -- an emphatic "No, really, this is junk." However, instead of a random mix of the three different stop codons, Taylor noticed that parts of these supposed junk regions contained significant clusters of "UGA" codons. Mathematically, the probability that a cluster of 10 "UGA" codons would overlap the CD4 gene by chance alone is one in 59,000.
In more recent work, Taylor has computerized the process of hunting for "UGA"-rich genes. His "gene viewer" program, developed with Dan Everett, a UGA assistant professor of computer science, has helped locate what may be selenium-encoding genes in the dreaded Ebola virus, as well as in several types of herpesvirus and a strain of coxsackievirus. The Ebola virus, for example, has a gene with 16 occurrences of the "UGA" codon, which has a chance probability of one in 43 million.
In the human genetic code, Taylor has found other "selenium genes," including several more in T cells. His research team also has identified a huge selenium gene containing 22 "UGA" codons that overlaps a human "proto-oncogene," which can cause cancer when mutated. "Needless to say, this has pretty radical implications related to the possible roles of selenium nutritional status in cancer," Taylor said.
He may not have intended a gene hunt, but Taylor has bagged some interesting trophies along an intriguing new trail.