A Little Lab Tackles a Big Question

The Molecular Sciences Institute is located on a busy street in downtown Berkeley, Calif., steps away from the entrance to BART, the area's rapid transit system. On the second floor of a concrete building, the elevator doors open onto a teal and purple hallway lined with bicycles. In the rooms off this hallway, scientists concentrate on what's at their benches, looking up only briefly as I walk by.

Opening the door to a closet-sized microscope room, I inadvertently interrupt cellular biologist Kirsten Benjamin, who sits perched on a high microscope stool with a pile of papers in her hand. "I just wanted a quiet place to read this paper before the meeting," she says, with a look that tells me she wants to be left alone.

Benjamin is referring to the weekly modeling meeting, where the experimental biologists, physicists, mathematicians, and programmers at the Molecular Sciences Institute (MSI) convene to discuss their projects. Most of the 17 scientists here work on defining, in exquisite detail, the signal transduction pathway in yeast that triggers the first step toward sexual reproduction. By discovering the ins and outs of one pathway, the researchers hope to find that elusive model that predicts how higher eukaryotes respond to changes in their extracellular environment. It's a venture they've coined the "Alpha Project," after the sex pheromone that triggers the pathway.

MSI's rooftop patio where researchers often gather for lunch.

My visit comes one month before the deadline for the application to renew MSI's main grant, so I figure that the meeting - which included Benjamin, physicist Andrew Gordon, two other biologists, and two mathematicians - would focus on writing the grant. But instead, the meeting attendees want to discuss one of Gordon's manuscripts, which they hope to publish in Nature.

They start working out the kinks in a paper about the intricate regulation of the signal transduction pathway. Gordon, with a background in particle physics, is the primary author, and his writing is not what biologists are accustomed to. An example: "We scaled the variance of each response distribution by the ratio of gene-expression noise to cell-to-cell variation in expression power. We then calculated a distribution of pheromone doses that maximized the transinformation between dose and response for these adjusted distributions."

"Can I ask a stupid question?" says Benjamin. "Because I think it's one that biologists are going to ask." Mathematician Larry Lok's arms fold across his chest; he is listening. Gordon looks up from his copy of the draft and stops twirling his pen. "Why not take doses that are equally spaced?" asks Benjamin. Gordon thinks for a moment and agrees, "Yeah, I can do that, I can do that."

Biologist Pia Abola is more direct. "The way it's written, the conclusion here just seems trivial," she says. "It's probably obvious because it's so well written," Gordon offers, with a wry smile. Everyone bursts into laughter. "We can't just jump into all that math language," suggests Richard Yu. With a background in physics and biology, Yu tries to mediate when he can.

Only later did I learn why the meeting focused on the paper, not the grant: Both Gordon and Benjamin were leaving MSI for other positions by the end of the month, around the same time that the grant application is due, so there were many loose ends to tie up, in addition to the grant application. Even under normal circumstances, manuscript meetings are a frequent fixture of the organization; MSI scientists have already published 10 papers from the Alpha Project, half of them in Science or Nature journals, and they intend to publish 14 more next year.

MSI is approaching its 10-year anniversary and undergoing a reassessment of its past, present, and future. "The way [MSI] started out was conceptually simpler than the way it wound up," says Roger Brent, MSI's current president and research director. The original charter didn't focus on one particular pathway - quite the opposite, in fact. Now, as MSI researchers begin to understand the possibilities and limitations of their approach, they are taking a hard look at the organization's mandate, and wondering about the next move.

During a weekly meeting: (from left) Larry Lok, Richard Yu, and Andrew Gordon

I had been warned that this wasn't a great time to visit MSI. Leonore Reiser, MSI's internship coordinator who doubles as a press officer, tried to convince me to wait until after June 1, the deadline for its grant application. She agreed to have me come earlier, as long as I understood that what I see might not be the "real" MSI.

Despite its current focus on one particular pathway, MSI's original purpose was to create an environment where scientists could study whatever they wanted, within the field of what would become systems biology. In 1996, Sydney Brenner received a check for $10 million, to be allocated over five years, from Phillip Morris to conduct his research. He decided to create a place where bright scientists could pursue projects unfettered, says Brent. "Sydney was extremely frustrated with business as usual in biomedical research." (Brenner was unavailable for comment.)

At the time, Brent was a professor at Harvard who wanted to figure out how to better mine data collected by big undertakings such as the Human Genome Project. Brenner "told me, 'you're never going to be able to do that at Harvard, Harvard or any other [academic] institution.'" Brent agreed that it would be easier to bring together an interdisciplinary team of scientists at a separate entity, rather than painstakingly gather support and commitment from various academic departments. In late 1997, the two started to hunt for "a place to park" the Institute; they decided on Berkeley because of its proximity to the University of California, Berkeley. It wasn't until 1998 when Brenner moved his operations from The Scripps Research Institute in La Jolla that things really got started.

Because of Brenner's renown and the kind of independent research environment it promised, the Institute attracted daring and motivated researchers. At the time, the goal of the Institute wasn't specific, but most of the staff wanted to sharpen the technology used to study biological systems, to take the "messiness" out of biology and make it more quantitative, more predictable - more like mathematics. How this should be done was anyone's guess.

A contingent of scientists that Brenner brought over from San Diego continued to work on the puffer fish (Fugu rubripes), which has a compact genome similar in function to the human genome but with at least 10 times less "junk" DNA. Alejandro Colman-Lerner, a cell biologist, investigated the use of aptamers to analyze proteomic data¹, while Ian Burbulis, a molecular biologist, looked at enzymatic complexes in Arabidopsis as a means of learning about metabolic pathways². Mathematician Lok plugged away at a computer modeling program that enabled scientists to enter the rules of protein interactions, rather than writing out each step³. The program, called Moleculizer, uses those simple rules to predict the ways the alpha pathway would respond to environmental changes.

Everyone had wild notions about the future of biology and a different plan about what needed to be done.

The researchers met weekly to discuss their projects. The gatherings had a very "think-tankish" feel to them, says Yu. Burbulis, who joined the lab as a postdoc in 1999 from Virginia Tech in Blacksburg, Va., remembers meetings where "ideas were just popping off like popcorn." Everyone had wild notions about the future of biology, and a different plan about what needed to be done.

The institute was fired up, but still without a single vision or direction. "The atmosphere was excellent, we could work on whatever we wanted," says Colman-Lerner, and most of the scientists did just that. But in 2001, the Philip Morris funding ended, and Brenner retired and took his Fugu scientists with him, leaving behind approximately 15 scientists. The Institute was at a crossroads.

MSI scientists began to yearn for a more cohesive mandate. Several came together and decided to pool their work into one project. Study one organism, one system, and use some of the tools and methods each had been designing separately. But which organism, which system? They all had been working on many model systems, including Arabidopsis, mammalian cells, and T7 bacteriophage.

Pia Abola and Richard Yu in the lab

Yeast "just hit the sweet spot," says Yu. The alpha mating receptor pathway was a relatively simple, well-characterized pathway, so the scientists already had many data to work with. "We own this organism, genetically," says Brent. But the system had never been quantified, leaving much still to learn, says Colman-Lerner, one of the principal investigators of the project. Several MSI investigators including Brent, Colman-Lerner, Lok, and Drew Endy (who later took a position at MIT), linked their projects together. They applied to the National Human Genome Research Institute (NHGRI) for a five-year $15 million grant to become a Center of Excellence for Genomic Science. The second-time around, the grant was accepted. The Alpha project was born.

The alpha pheromone drives the attraction that brings yeast together for sexual reproduction. Its release causes yeast to start growing a protrusion, or shmoo, in the direction of its potential mate. On a molecular level, the alpha pheromone triggers a G protein-coupled receptor, which leads to numerous conformational changes and phosphorylation events. The entire process culminates in the release of a transcription factor in the nucleus, turning on 100-200 genes and causing yeast to shmoo. MSI scientists focus on the signaling pathway that makes this all possible.

As more scientists came on board at MSI - including Benjamin, Abola, and Koichi Takahashi, who brought a bioinformatics background - each found an aspect of the pathway they wanted to work on, and envisioned novel tools to help tighten the resolution.

Lok's Moleculizer helped the researchers to imagine the ways that the alpha pathway would respond to environmental changes. Entering real data, however, instead of hypothetical values or estimates, would strengthen its predictions. So Benjamin set out to turn the Western blot into a quantitative test that could measure the relative amount of the 25 or so proteins involved in transmitting signal from the membrane to the nucleus. "I took several months to do experiments to identify the sources of quantitative error in immunoblotting," says Benjamin, "and there were so many things" that had to be changed.

The hallway leading to MSI's labs

Armed with the absolute quantities of most of the key proteins in the pathway that she laboriously obtained, Benjamin started to play with BioNetGen, a stochastic, rule-based modeling tool created by researchers at Los Alamos National Laboratories that is similar to Lok's Moleculizer. During experiments, the scientists saw that yeast cells contained lower than expected levels of the alpha pathway's scaffold protein, sterol 5, which holds together a number of coactivating proteins in the kinase cascade. Normally, increasing sterol 5 would increase the signal output.

Benjamin and her collaborator, Ty Thomson, in Drew Endy's lab at MIT, created a page in Wikipedia that catalogues their work as they build and refine their BioNetGen-based model with Benjamin's data. Using Wikipedia.com also lets them link their assumptions about the pathway to supporting data. After improving the model using real-world data, the model showed that keeping sterol 5 at suboptimal levels allowed for a more optimal response to pheromone. If this proves true by experimentation, it would be "a fairly strong example of something that came from modeling first," says Brent. The two are currently doing the biological experiments to confirm their model's conclusion.

The Wikipedia page (under the title YeastPheromoneModel.org) lets other researchers working on the alpha pathway "build models automatically from the documentation in the Wiki," says Benjamin. As of press time, no outside scientists have contributed to the MSI Wiki page; if they do, that extra data could improve the model's accuracy even more. "It's not our tool, but it's everybody's," says Thomson. (At least one other lab is also using Wikipedia to catalogue how the scientists built their model.)

As part of another project, biologist Colman-Lerner and physicist Gordon worked side-by-side to determine why thousands of genetically identical yeast cells behave quite differently from one another in response to the same concentration of alpha pheromone. Above a certain threshold concentration, the majority of cells switched on the signal transduction pathway to initiate transcription of mating genes. Among cells receiving lower hormone concentrations, however, there was much variability.⁴

Roger Brent

The scientists tracked individual molecules with fluorescent markers and reporter genes, and they used an open-source computer program that Gordon wrote called Cell-ID⁵ to quantify proteins as they where transcribed. They found that, in cells exposed to a low concentration of alpha pheromone, the initiation of a signaling pathway that would lead to a shmoo was tightly regulated by the MAP kinases Fus3 and Kss1, which were in turn likely regulated by cell health, age, or phase in the cell cycle. In other words, the signal transduction pathway is much more than just an on/off switch that alpha pheromone triggers. Already representatives from industry have shown interest in the free software, which they hope will help facilitate high-throughput drug screening, says Brent.

The scientists are starting to see patterns that may have important implications. When tracking the precise quantities of each type of protein along the pathway, Yu and his collaborators saw a spike in some of the proteins and then a rapid falling-off, which suggested a fast-acting negative-feedback mechanism that lowered the concentration of a protein when it had completed its task. Understanding the extent of this regulation could help biomedical researchers develop simple-to-manufacture inhibitors of the negative feedback, in order to increase the signal.

The information that MSI has gathered about the parts that make up this one system could, in theory, help show what happens when the parts are broken in disease states, and build new synthetic biologic systems that perform similar functions.

At least four other research centers collaborate with MSI towards these goals: Jehoshua Bruck's lab at California Institute of Technology, Endy's lab at MIT, and Richard Smith's group at the Pacific Northwest National Lab. The smallness of MSI makes collaborations easy, almost inevitable. Still, with few resources to spare for lab technicians, all scientists do their own experiments, and dishes.

"Here I am making up buffer," says biologist Abola, as I aim the camera at her performing one of the more mundane laboratory tasks. Yu does Abola one better by holding up and peering quizzically at a Petri dish, striking what he calls "the quintessential scientist pose." Brent watches the researchers' antics while sitting in a relaxed slouch at one of the lab benches.

The scientists here expect a lot from each other. Geneticist Gustavo Pesce describes the discussions that go on between scientists at MSI as "high level." Part of what separates MSI from other institutes is its insistence on bridging the language and cultural barriers of different scientific disciplines, just like in the meeting I attended. "We struggle with it every day," says Orna Resnekov, MSI's deputy director. Every paper the institute publishes contains a note detailing each author's contribution. "It's what we do to make team-based research viable," says Resnekov.

MSI scientists are hunting for a model that predicts how higher eukaryotes respond to environmental changes.

Now, MSI finds itself at another crossroads. Its NHGRI grant is up for renewal. And, it's not just Gordon and Benjamin who are leaving: Colman-Lerner is starting a lab in Argentina to continue his MSI work in mammalian cells, though he plans to stay on at MSI, at least part-time. Many of the others, such as Abola, Burbulis, and Yu, say they would like to lead their own labs one day.

If renewed for 2008, the next generation of Alpha - Colman-Lerner calls it Alpha II - will focus on more applied problems, such as learning how pathways similar to alpha are involved in disease and malignancy states. One of his goals will be to look at "how tiny changes in the genome" change the "quantitative behavior of a complicated pathway," and how those changes affect phenotype, says Colman-Lerner.

Other MSI researchers plan to continue to study the mechanics of the "information transmission system," as Brent calls it. The original mission, to find that elusive model that predicts how higher eukaryotes respond to changes in their extracellular environment, is clearly still unanswerable, for the time being. "There's still a lot to be done," says Brent. Much of what MSI has accomplished over the past five years has been "to learn how to understand the behavior of one thing," and to show that zooming in on one pathway can help answer much bigger questions. "I think the people [at MSI] did it."