From his earliest days at the bench, Jack Szostak has shown a penchant for "tackling difficult projects where nobody has succeeded," says Ray Wu of Cornell University, who oversaw Szostak's graduate work on cloning cytochrome c. "It was 1973 and no genes had yet been cloned," says Wu. So Szostak set out to chemically synthesize an oligonucleotide primer that would help him pull down the cytochrome c gene - an effort that required a collaboration with an organic chemist and a year-and-a-half of his time, all to accomplish something Szostak notes "you could now do in an hour." Szostak was successful: "We could light up the gene on Southern blots and detect a message and so on," he says, although another group beat him to publication.
That was just the first of Szostak's long list of successes. His identification and characterization of yeast telomeres, carried out in the 1980s in collaboration with Liz Blackburn, last year earned him a Lasker Award. His work on chromosome composition and dynamics in the late 1970's and early 1980's led to the enumeration of the double-strand break model of homologous recombination and to the construction of the first artificial chromosome. His longstanding interest in catalytic RNA - launched soon after Tom Cech discovered the first ribozyme in 1982 - led to the development of the in vitro selection technique that allowed Szostak and his colleagues to evolve nucleic acids that can carry out the kinds of chemical reactions they'd need to be able to replicate themselves.
"It's very rare for someone to do pioneering work in fields that are totally unrelated," says Kevin Struhl, a colleague at Harvard Medical School. Szostak continues to churn out important papers. "Not just little important papers, but really important papers," every couple years, says Gary Ruvkun, another colleague at Harvard Med and the Massachusetts General Hospital (MGH), where Szostak is an HHMI investigator.
FROM PLASMIDS TO TELOMERES
While still in Wu's lab as a postdoc, Szostak worked with fellow postdoc Rodney Rothstein and began dabbling in yeast genetics. It was 1977, and Gerald Fink (also at Cornell at the time) had just worked out how to introduce genes into yeast. "We were playing around with that; everybody was," says Szostak. "I don't really remember what we were doing, but it became obvious that when you cut circular plasmids with restriction enzymes, funny things started happening." Sometimes the plasmids would integrate into the genome, and sometimes they would remain separate.
Szostak continued to pursue the observation when he set up his own lab at the Harvard Medical School in 1979, and it wasn't long before he and Rothstein, then a faculty member at the New Jersey Medical School, and Terry Orr-Weaver, one of Szostak's first graduate students, determined that cutting the plasmid in a part of the DNA that was homologous to a yeast sequence boosted its integration into the genome. "Following up on that, figuring out the mechanism, and figuring out that it could be something that's happening naturally in yeast," Szostak says, led to the development of the double-strand break-repair model of homologous recombination. "Every molecular biology textbook in the entire universe has the double-strand break-repair model in it," says former postdoc Andy Ellington of the University of Texas, Austin. "That's Jack's baby."
"And it's one of the major conceptual ideas in the field of recombination," says Struhl.
The find also primed Szostak for the discovery of yeast telomeres. "We were looking at what happens to restriction enzyme-cut linear pieces of DNA when they go into a cell," says Szostak. "Those DNA ends are very reactive. They love to recombine. If there's nothing they can recombine with, they ligate to each other. They do anything except stay as ends. That suggests that there's something special about normal linear chromosomes" that prevents them from behaving the same way.
With that in mind, Szostak went to the 1980 Gordon Conference on nucleic acids, where he heard Blackburn talk about her work on telomeres in Tetrahymena. "Here was a piece of DNA that could act as a stable chromosome end," says Szostak, "but it was in this weird ciliate, so nobody knew if it was something completely different or if the system would be the same in a standard model like yeast. So it seemed like a really simple experiment to take those Tetrahymena telomeres, put them on the ends of cut plasmids, move them into yeast, and see if they would act as stable ends. And they did." The result was published by Szostak and Blackburn in 1982.
Szostak and his postdoc Vicky Lundblad also showed that as telomeres shrink, as they do after each round of cell division in mutants that lack the enzyme telomerase, the "cells got sicker and sicker." They started losing chromosomes, sustaining chromosomal rearrangements, and undergoing cell death. "It was a senescence phenotype," says Szostak, "so the thing that was really exciting for us was the first link between telomerase and senescence" - work that was published in 1989.
"I think there's a reasonable shot he'll win a Nobel Prize for this," says Struhl. The work opened up the possibility of doing genetic experiments on telomeres and telomerase, because yeast are easier to manipulate than Tetrahymena, says Wu. It also led to Szostak's engineering, along with then-student Andrew Murray, of the first artificial chromosome: a segment of DNA, with a centromere and telomeres and origins of replication, that behaved like a chromosome when its host cell divided, an accomplishment that Ruvkun says "knocked my socks off."
But Szostak wasn't satisfied. Although his lab was still working on telomeres and chromosomes in the early to mid-1980s, he says, "the yeast field was getting very crowded: There were probably more labs than genes. So I started to think it might be more fun to find some other field that was less crowded, where there would be an opportunity to do something new and interesting. Then the ribozyme story burst on the scene."
"Jack thought it was cool and he wanted a new toy to play with," says Ellington. "And he thought, ?Damn, Tom Cech is having all the fun.'" His vision from the start was clear, says Jennifer Doudna of the University of California, Berkeley, who joined Szostak's lab as a graduate student in 1986. "Jack's idea was to reengineer a ribozyme" - which had the ability to catalyze a splicing reaction - "to make it function more like a polymerase."
Their efforts culminated in a 1989 Nature paper in which he and Doudna "showed for the first time that a ribozyme could use short RNA oligonucleotides as substrates and ligate those oligos together in a template-directed way," she says.
"It was actually pretty exciting to see that kind of copying reaction, but it was not very efficient," says Szostak, "so we started thinking, well, maybe we can evolve our own ribozymes and find something better. So that led to the in vitro selection technology." Not only would the selection allow Szostak and his colleagues to generate better replicators, but they could also use the technique "as a way to explore the kind of functions that RNA molecules are capable of without being limited by what one sees currently in biology," says Doudna, potentially providing a window on the RNA world that might have existed before proteins took over as catalysts.
The approach, which Doudna considered "sheer genius," is theoretically simple: Generate very large pools of random nucleic acid sequences and then select for the ones that have the desired function. Amplify and repeat. "It sounds easy," says Szostak, "once you know it works."
Ellington, who made the system work and in 1990 used it to isolate RNAs that could bind to a small dye molecule, agrees. "Once you've done it, you know what the numbers are. So you can say, OK, it's one part in 1,010, and I've got 1,013 things, so I'm good." Of course, if it had turned out that only one in every 1,020 molecules would have the desired activity, he says, "I could have done the experiment every day for the rest of my life and I would never have gotten a damn thing. That's where it was useful to have Jack's vision. Whereas I would worry that it might be one-in-1,020, Jack would say, ?It's going to work.' Maybe he was lying through his teeth, but he seemed to have the ability to see the future. They used to talk about Barbara McClintock and her feeling for the organism. Jack Szostak has a surer feeling for the molecule than any human I've ever met."
When David Bartel joined the lab as a graduate student, he and Szostak used the method to generate the first completely artificial ribozyme, guaranteeing that "Jack will forever be known as the guy who evolved catalysts from random sequence," says Ellington.
RETOOLING THE LAB - AND LIFE
That's no small feat for someone who started his career as a yeast geneticist. "You'd have to be out of your freakin' mind to go from yeast genetics to ribozyme biochemistry and think it was gonna work out for you," says Ellington. "But I don't think Jack even flinched. It demonstrates his intellectual boldness and his capability."
"It's as dramatic a retooling of a lab as I'm aware of in all of science, let alone biology," says Ruvkun, and that retooling has paid off. "Look at his body of work over the years," says Doudna. "In every field he's worked in, he's had a really major impact, and that's really unusual in science - to find somebody who's that diverse and that accomplished."
Szostak credits a supportive environment. When he joined the faculty at MGH in 1984, the molecular biology department was supported entirely by the German pharmaceutical company, Hoechst. "For me it was fantastic," says Szostak. "Everything was paid for. I could do anything I wanted. It allowed me to start the RNA work without having to get funding from NIH, where reviewers might say, ?You can't do that because you've never done that before.'" The same isn't true of HHMI, which expects Szostak as an investigator to "get into new things."
Now, there's one more big new thing on Szostak's to-do list. "He wants to create artificial life," says Ruvkun. For the past 10 years, Szostak has been laboring to encase increasingly competent self-replicating RNAs inside lipid vesicles that would be able to grow and divide. If these protocells were then able to evolve - to adapt to changing conditions in the lab by, say, increasing their rate of replication - he will have succeeded.
"That would be a cell," says Ellington. "You could say, ?Before Jack mixed these chemicals in a test tube, it was just a pile of crappy chemicals. After Jack mixed these chemicals in a test tube, they turned into a self-organizing system capable of evolution. Bingo. That's a cell. That's life."
"If anybody else said that you'd laugh," says Doudna. "But it's Jack, so you say, jeez, maybe he'll do it."