Shiv Grewal hasn't slept much in the past eight or nine years. Not since he found that in fission yeast, gene silencing depends on the machinery that carries out RNA interference (RNAi), a discovery that effectively tied together two of the hottest areas in modern cell biology. "I've never seen someone work so hard," says former postdoc Ken-ichi Noma of the Wistar Institute in Philadelphia. "I'm Japanese, and Japanese people usually try to work more than the boss. But I gave up. It was physically impossible. He was in lab by 9:00 every day and then he worked until 3:00 or 4:00 in the morning. I don't know how he did it."
"When we'd be writing a paper, I'd send him a draft at two in the morning and he would reply immediately. And the next morning at 9:00, there he'd be," adds Songtao Jia of Columbia University, another former postdoc. "A lot of scientists work weird hours when things get exciting or you need to take care of an experiment. But Shiv is always there. He has a tremendous amount of energy."
That energy has given rise to a dozen papers in Science and Cell and has helped shape our understanding of gene silencing, epigenetics, and chromatin biology. "Without question, Shiv is one of the uncontested leaders in the field," says Rockefeller University's David Allis. "And he didn't jump on the bandwagon once the field was red hot. He's been in the trenches, doing beautiful genetic dissections of the pathways that smelled important to him. So his work has really set the pace."
"Shiv has made a lot of breakthrough discoveries about the way that different proteins are controlling the formation and maintenance of heterochromatin," adds Steve Jacobsen of the University of California, Los Angeles. "There are maybe three or four labs at the very leading edge of this field of heterochromatin biology, using Schizosaccharomyces pombe as their experimental organism. And Shiv has consistently been right there at the forefront."
As a Cambridge Nehru scholar—India's equivalent of a Rhodes scholar—Grewal cloned the molecular switch that allows a mushroom-infecting pseudomonad to toggle between a pathogenic and nonpathogenic phenotype. When food is plentiful, the bug spits out a toxin that digests the mushroom on which it sits. When nutrients are scarce, it turns off its toxin and starts to move—presumably to find something to eat. For his thesis, which he completed in 1992, Grewal identified the genetic components that comprise that switch, work that primed him for his postdoctoral studies of mating type switching in fission yeast.
Like its cousin Saccharomyces cerevisiae, haploid S. pombe cells come in one of two "sexes." Whether a cell is of the plus or minus mating type is designated by information encoded at a transcriptionally active locus called mat1. However, the complete backup information for both mating types is stored remotely, at a pair of transcriptionally silent loci called mat2 and mat3: mat2 carries the instructions for making a plus cell, mat3 the instructions for a minus cell. To change its sex, a pombe cell copies the desired recipe from either mat2P or mat3M into the active mat1 locus. "And the process is nonrandom," says Grewal. In other words, the cell always switches: a minus cell always pulls the 'P' program from mat2P, and a plus cell reaches for the 'M' information at mat3M.
When he joined Amar Klar's lab at the National Cancer Institute as a postdoc, Grewal says, "they knew there was this nonrandom choice of donors, but they didn't know the mechanism." So he sequenced the 20 kilobase domain that contains mat2 and mat3, and discovered, sitting between the two genes, a highly repetitive bit of DNA, which he then removed. "To our delight," Grewal says, "deleting that element abolished not only the nonrandom choice of donors, but it also abolished the silencing across the extended mat2P-mat3M domain"—functions he would later show depend on the formation of heterochromatin at that repetitive stretch.
Furthermore, in experiments where the researchers inserted a marker gene into the region between mat2 and mat3, they determined that the chromatin state that marked that region as silent was heritable. "I think that was the highlight of his time here—finding that the silenced states of gene expression can be propagated as epigenetic states, both through mitosis and meiosis," says Klar. "The beauty of this is that it's very, very clear that it's not only DNA that's being inherited. It's DNA plus the state of the chromatin that make up what we call Mendel's gene."
"That was such a beautiful paper," says Noma of the resulting article, which came out in Cell in the summer of 1996. "It was very simple, only three or four figures. But the conclusions were completely solid."
After starting his own lab at Cold Spring Harbor, Grewal and his collaborators took off after a protein called Argonaute. In flies, Argonaute governs the asymmetric division of stem cells in the germline. The rules that pombe cells follow when they divide to produce M and P cells during mating type switching are remarkably similar to the rules that stem cells follow as they produce their progeny. "So we wanted to see if knocking out the Argonaute gene in pombe would affect this switching pattern," says Grewal. "It didn't." (Because, as Grewal later discovered, heterochromatin assembly is governed by redundant mechanisms.) But Grewal noticed something interesting about the mutants: they appeared to have trouble segregating their chromosomes. "While I was a postdoc, we had studied several other segregation mutants that were also defective in heterochromatin-mediated silencing at their centromeres," he says. "So it was natural to ask whether this mutant was also defective in centromeric silencing. When we looked, we were surprised to find that indeed it was."
"If I hadn't been doing the crosses myself, I sometimes feel we might not have made that connection," says Grewal. Jia agrees: "I think a lot of people would have missed it completely because it's not a massive defect"—perhaps one out of 20 mutant spore tetrads showed the segregation problem. "When I first started, if I'd gotten that result, I would have thought I didn't do the experiment right," says Jia. "But Shiv knew the system so well, he realized that something real was going on—something he should pursue."
Around the same time, Craig Mello, a Howard Hughes Medical Institute investigator at the University of Massachusetts Medical School, showed that Argonaute proteins are involved in RNAi. "So one thing led to another, and that's how we ended up finding out that this repetitive DNA element in the middle of the mat2-mat3 cassette is actually an RNAi-dependent heterochromatin nucleation center," says Grewal. Centromeres, telomeres, and the mat locus all contain similar, repetitive DNA sequences. So in fission yeast, RNAi machinery contributes to heterochromatin assembly at the mating type locus, and at the centromeres and telomeres, as well.
"The idea that you might somehow direct this heterochromatin silencing through small noncoding RNAs took the field by complete surprise," says Allis. Science hailed Grewal's discovery as its 2002 'Breakthrough of the Year,' and it prompted "an explosion of additional studies," says Allis, including many from Grewal himself. By 2004, he and his colleagues had shown that a protein complex they call RITS, which contains two proteins in addition to Argonaute, recognizes the methylated histones that are characteristic of heterochromatin. This RNAi-induced transcriptional silencing complex sits atop heterochromatin and stays on the lookout for any transcripts that might be produced from the repetitive DNA beneath it. It then directs those stray transcripts to the RNAi machinery for degradation. The resulting fragments (through mechanisms that are still being worked out) attract histone methyltransferase—which methylates more histones, thus attracting more RITS complexes, and so on. The mechanism is not completely self-perpetuating, although Grewal's success in unraveling it seems to be.
"Pretty much without exception, every one of the gene products that Shiv and his colleagues have identified has turned out to be a central player in heterochromatin-based gene silencing in pombe," says Allis. And the best part is, the work is done in vivo.
"For Shiv, the starting point is always the genetics," says Carl Wu, a colleague at NCI, where Grewal now heads the chromosome biology section. "That's very powerful, because you can't lose if you have a phenotype. The thing that's really impressive, though, is that Shiv uses these mutant phenotypes in a very rigorous and thorough way to discover new and surprising molecular connections."
He's also been able to move from classical genetics into genomics and proteomics, says Wu. "Shiv was one of the first people to study whole-genome binding of heterochromatin proteins. That has opened up the field of heterochromatin biology, and allowed him to gain tremendous insight into how heterochromatin is assembled and how it functions."
For one thing, Grewal's high-resolution maps of the entire pombe genome confirm that where there's heterochromatin, there's RITS, along with cohesins, histone deacetylases, and other components of the RNAi machinery. "All these factors are coating entire heterochromatin domains and silencing those sequences," says Grewal. These studies suggest that "there's another whole level of genome organization beyond the gene-by-gene view the field had been focused on before," says Allis. "Now there are all these papers in mammalian cells looking at epigenetic landscapes in different cell types. And if you think about where that idea comes from, again, you go back to some of the remarkable pombe experiments that Grewal and his colleagues conducted."
Grewal's combination of genetics, biochemistry, and genomics is "killer," adds Jacobsen. Take Swi6, for example. This adaptor protein, the pombe equivalent of the mammalian protein HP1, binds to histones that are methylated on lysine 9. Grewal has shown that knocking out the swi6 gene disrupts heterochromatin. He's purified its binding partners and found that they, too, interact with heterochromatin. And he's shown that across the genome, Swi6 is found associated with heterochromatin. "That makes a pretty convincing argument that Swi6 is critical for heterochromatin," says Jacobsen. "Any one of those approaches would give you a pretty good idea, but when you use all three, it's a very strong confirmation. And he's done that again and again, for a variety of proteins that are involved in the system."
Like a piece of DNA taking cues from the environment, Grewal says being at NIH helps him to get it all done. "Here we can focus on the scientific questions without feeling like we're in a pressure cooker," he says. "We can pursue high-risk research without necessarily worrying about whether it will lead directly to the next high-profile publication." Of course, in Grewal's case, it probably will.