The New York Times, August 16, 2005
Michael Ellison has a dream: to reconstruct a living thing inside a computer, down to every last molecule. It is, he said, “the ultimate goal in biology to be able to do this.”
It’s a dream that Dr. Ellison, a biologist at the University of Alberta, shares with other scientists, who have imagined such an achievement for decades.
Understanding how all of the parts of an organism work together would lift biology to a new level, they argue. Biologists would be able to understand life as deeply as engineers understand the bridges and airplanes that they build.
“You can sit down at a computer, and you can design experiments, and you can see the performance of this thing, and then you can figure out why it’s done what it’s done,” Dr. Ellison said. “You’re not going to recognize the full return of the biological revolution until you can simulate a living organism.”
In the past few years this fantasy has become plausible and now Dr. Ellison is part of an international team of biologists who are now trying to make it a reality. They have chosen to recreate Escherichia coli, the humble resident of the human gut that has been the favorite species for biology experiments for decades.
“We picked the simplest organism about which we know the most,” Dr. Ellison said.
Scientists may know more about E. coli than they do about any other species on earth, but that doesn’t mean that creating a virtual E. coli will be a snap.
Many mysteries remain to be solved, and at the moment even a single E. coli may be too complex to recreate in a computer.
But the effort is still worthwhile, some scientists argue, because it would become a powerful tool for drug testing, genetic engineering and for understanding some of life’s deepest mysteries.
Discovered in 1885, Escherichia coli soon proved easy to raise in laboratories. Its popularity boomed in the 1940’s when scientists figured out how to use it to pry open the secrets of genes.
In the 1970’s scientists figured out how to insert foreign DNA into E. coli, turning them into biochemical factories that could churn out valuable compounds like insulin.
“Everybody studies E. coli for everything,” said Gavin Thomas, a microbiologist at the University of York in England.
Research on E. coli accelerated even more after 1997, when scientists published its entire genome.
Scientists were able to survey all 4,288 of its genes, discovering how groups of them worked together to break down food, make new copies of DNA and do other tasks.
Some scientists speculated that before long they might understand how all of the pieces of E. coli worked together.
Such speculations were not new. In 1967, Francis H.C. Crick, the co-discoverer of DNA, and the Nobel Prize-winning biologist Sydney Brenner had called for “the complete solution of E. coli.”
But the call went unheeded for over 30 years. After all, E. coli contains an estimated 60 million biological molecules. Simulating all of them at once was an absurdly difficult task.
But by the late 1990’s, it began to look plausible, although not necessarily easy. Despite decades of research, many of E. coli’s genes still remain a mystery — “probably around 1,000 genes,” Dr. Thomas said. “There’s a lot more we need to know about E. coli before we can build a really solid model.”
To find out more, E. coli experts have been joining forces.
In 2002 they formed the International Escherichia Coli Alliance to organize projects that many laboratories could do together.
In one project, researchers have created over 3,900 different strains of E. coli, each missing a single gene. “It would have been foolish for two or three labs to carry this out at the same time and compete with each other,” said Barry Wanner of Purdue University, who led the project.
Soon scientists will be able to order the entire collection of these strains for their own research. “We’ve done a variety of simple tests, but we can’t do every conceivable experiment,” Dr. Wanner said. “But a hundred other laboratories can do hundreds of other ones.”
As knowledge of E. coli grows, scientists are starting to build models of the microbe that capture some of its behavior. “This field is moving forward very aggressively,” said Bernhard Palsson of the University of California, San Diego.
Dr. Palsson models E. coli’s metabolism. Like other living organisms, E. coli breaks down food with enzymes, whittling molecules down bit by bit.
It then uses other enzymes to refashion the fragments into new molecules. Dr. Palsson and his colleagues have reconstructed the interactions of over 1,000 metabolism genes.
They can predict how fast the microbe will grow on various sources of food, as well as how its growth changes if individual genes are knocked out. Based on experiments with real E. coli, the researchers find the model gives the right predictions 78 percent of the time. Now they are expanding their model to 2,000 genes.
Meanwhile, researchers at the laboratory of Philippe Cluzel at the University of Chicago have been focusing their efforts on making E. coli swim.
The microbe swims with several spinning tails, each driven by a motor revolving 270 times a second.
If the tails turn counterclockwise, they all wrap into a bundle that can propel E. coli forward. If the microbe makes the motors turn clockwise, on the other hand, the tails fly apart and send E. coli into a tumble.
By alternatively swimming and tumbling, E. coli can navigate through its tiny world. It “decides” which way to spin its motors based on the information it gets from sensors that stud its outer membrane. “It’s a big information-processing network,” said Thierry Emonet, a research scientist at the University of Chicago.
To understand this microbial computer, Dr. Emonet and his colleagues have created a virtual E. coli that can sense its surroundings and decide how to swim. They simulate the chemical reactions that carry signals from sensors to motors, and then track the path a virtual E. coli takes through three-dimensional space.
Their virtual E. coli turns out to swim very much like the real thing. In one experiment, the scientists put virtual microbes in a gradient of food. “The higher you go, the more food there is, so their goal is to go up,” Dr. Emonet said.
And they do, although they don’t go in a straight line. “You see single cells go up, but then they lose track of their original direction, and they go down at some point,” Dr. Emonet said. “The sensory system reacts, and they tumble, and then they go up again. It’s pretty cool.”
Dr. Emonet said he hoped that his model would allow scientists to understand other sorts of decisions made by cells. Cells “decide” to divide in response to certain signals, for example, and runaway cell division can lead to cancer. Understanding the simple decisions of E. coli may help researchers understand the decisions of more complex cells like those in our own bodies.
“If you can’t understand how a single E. coli is able to find food by passing information from the outside to the inside, there’s very little hope for understanding a system like cancer,” Dr. Emonet said.
A full-blown model of E. coli would be able not only to swim, but eat food, fight off invading viruses, make copies of its DNA, and do many other tasks all at the same time. Scientists agree that building a multitasking model would be a daunting job. “Technically, that’s incredibly more difficult,” Dr. Thomas said.
Dr. Ellison and his colleagues have decided to take the first steps toward creating a full-blown model.
They want to begin by simulating a simplified E. coli. “We’re going to strip E. coli down to about one-quarter of its original size,” Dr. Ellison said.
Dr. Wanner is working with colleagues in Japan to make this minimal E. coli. “We’re trying to knock out groups of 100 genes at a time,” Dr. Wanner said. They hope to produce a stripped-down E. coli with only around 1,000 genes within two years. Dr. Ellison and his colleagues then hope to create a virtual twin, in an endeavor they have dubbed Project Gemini.
“Our approach is to track every biomolecule in that cell in space and time,” Dr. Ellison said.
As a proof of concept, he and his colleagues have simulated a bubble-shaped membrane made of 13,000 particles. The membrane acts a lot like a real one, swelling when it is filled with extra molecules. The researchers hope to use the model to recreate an entire E. coli, complete with genes, enzymes, membrane channels and other parts.
There is one major catch, however. Even a stripped-down E. coli is so complex that no existing program can simulate it. “Our gamble in this is that computers are getting more powerful, so we build the framework and within 5 or 10 years the computers will be able to deal with this,” Dr. Ellison said.
“Assuming the speed of computing keeps increasing, I don’t see why it’s not possible,” said Dr. Emonet, who is not involved in Project Gemini.
But, like some other scientists, he has some reservations about its usefulness. “Even if we could make a simulation of everything inside E. coli today, that does not mean we would understand it,” he said “The trick is to build the thing in steps and check that you understand the phenomena one at a time.”
A full-blown model of E. coli is still worth the effort, many scientists argue, because of its potential benefits. Scientists could adapt the E. coli model to more complex human cells to simulate how they react to different drugs.
“Then you can really do genetic engineering,” Dr. Ellison said. “I mean where you can actually design an organism or change it in massive ways. When people talk about genetic engineering today, it’s really kind of a joke because they mean, ‘I moved a gene from one organism into another organism, and I’m going to pray that it works.”‘
A virtual E. coli could allow scientists to see in advance how major changes to the microbe would affect it.
“That opens up a huge amount of opportunity,” Dr. Ellison said.
Copyright 2005 The New York Times Company. Reprinted with permission.