Microbes Made to Order
Science, January 9, 2004
In the dead of a New England winter, 16 students worked day and night for a month trying to make Escherichia coli blink like a lighthouse. No one really expected a blinking bacterium to be all that useful. Instead, the exercise was meant to teach students--and their instructors--how to make reprogramming bacterial behavior more routine. The first class of its kind, held last January at the Massachusetts Institute of Technology (MIT) in Cambridge, also marked the emergence of the hot new field of synthetic biology.
Bacterial blinking circuits are just one element in the MIT researchers' "registry of standard biological parts," which is akin to an inventory that electrical engineers or basement tinkerers might consult when they design a new device, says class co- instructor Drew Endy of MIT. Researchers at MIT and elsewhere are working on sensors and actuators, input and output devices, genetic circuits to control cells, and a microbial chassis in which to assemble these pieces. If they're successful, the registry will help them reach one of the goals of synthetic biology: to allow researchers to "go into the freezer, get a part, hook it up," and have it work the first time, Endy says.The parts list is itself just one piece of a hugely ambitious plan: to engineer cells into tiny living devices. Some of the engineered devices these researchers envision will function as molecular-scale factories. Others will help detect chemical weapons, clean up environmental pollutants, make simple computations, diagnose disease, fix faulty genes, or make hydrogen from water and sunlight. "We're going to modify the whole behavior of the cell," says bioengineer Ron Weiss of Princeton University in New Jersey. Synthetic biologists aim to build cells from the ground up rather than tinkering with a handful of genes or tweaking a metabolic pathway or two, as do today's genetic engineers.
The fledgling field, which is attracting engineers and biologists in equal measure, means different things to different people. Engineers view it primarily as an engineering discipline, a way to fabricate useful microbes that do what no current technology can. But many biologists see it instead as a powerful new way to learn about cells. Unlike systems biologists, who analyze troves of data on the activity of thousands of genes and proteins (Science, 5 December 2003, p. 1646), synthetic biologists simplify and build. They create models of genetic circuits, build the circuits, see if they work, and adjust them if they don't--learning about biology in the process. "I view it as a reductionist approach to systems biology," says biomedical engineer James Collins of Boston University.
However it's defined, synthetic biology is catching on. A growing cadre is publishing in top journals. Researchers at Lawrence Berkeley National Laboratory (LBNL) in California established the world's first synthetic biology department last June. A European Commission program designed to support "unconventional and visionary research" has issued a request for synthetic biology research proposals. The inaugural synthetic biology conference (Synthetic Biology 1.0) is set for next June at MIT. "I think we're going to see some spectacular new science and engineering," says Eric Eisenstadt, a program manager who oversees synthetic biology funding for the Defense Advanced Research Projects Agency (DARPA). J. Craig Venter, who heads the Institute for Biological Energy Alternatives in Rockville, Maryland, predicts that "engineered cells and life forms [will be] relatively common within a decade."
Rewiring the cell
The library of such principles was inaugurated decades ago when microbiologists François Jacob and Jacques Monod of the Pasteur Institute discovered the first gene circuit--a set of genes that help E. coli digest lactose. A regulatory gene called a repressor is normally on, keeping the lactose-digestion circuit inactive. When lactose is present, however, the bacterium turns the repressor off. Such gene circuits can be diagramed with nodes representing genes and arrows indicating which other genes they regulate. "If you squint hard enough, it begins to look like [an electrical] circuit diagram," says bioengineer Jeff Hasty of the University of California (UC), San Diego.
The analogy falls down on the details, however. Electronics engineers know exactly how resistors and capacitors are wired to each other because they installed the wiring. But biologists often don't have a complete picture. They may not know which of thousands of genes and proteins are interacting at a given moment, making it hard to predict how circuits will behave inside cells.
To simplify the problem, physicists Michael Elowitz of the California Institute of Technology (Caltech) in Pasadena and Stanislas Leibler of Rockefeller University in New York City built a genetic clock from scratch--the original blinking bacterium that last winter's MIT students were trying to improve upon. The two, then at Princeton University, designed a circuit of three repressor genes (call them genes A, B, and C), which they dubbed the "repressilator." It worked like the game "Rock, Paper, Scissors": Gene A turned off gene B, gene B turned off gene C, and gene C turned off gene A. Gene C also turned on a jellyfish gene that turned the cell green. In physicists' terms, the device was a limit-cycle oscillator:an oscillator that reestablishes the same behavior after it's perturbed. When they put the circuit into E. coli, the cells blinked. The work, reported in Nature in 2000, is "the high-water mark of a synthetic genetic circuit that does something," Endy says.
More recently, Michael Savageau of UC Davis, Alexander Ninfa of the University of Michigan, Ann Arbor, and their colleagues rewired some well-studied bacterial gene circuits to make a less noisy oscillator. With minor modifications, the oscillator also functioned as a toggle switch, turning a circuit on or off. The results, published in Cell in April, show that as researchers understand the design principles of genetic circuits, they learn to control their behavior, Savageau says.
So far, most designers have made gene circuits that mimic simple physical devices used routinely by engineers, including toggle switches, oscillators, and feedback loops. But evolution "might have come up with new designs that engineers never thought of," says Savageau. So they have begun drawing design principles from biology. By examining patterns of gene expression in E. coli, for example, physicist-turned-biologist Uri Alon of the Weizmann Institute of Science in Rehovot, Israel, and colleagues identified three widespread gene circuit designs. They synthesized a circuit using the most common design, called a feed-forward loop, and installed it in bacteria. As they reported in November in the Journal of Molecular Biology, it enables bacteria to turn genes on slowly but off quickly--a property that seems to help the cells filter out molecular noise and activate genes only when they're needed.
Biodesign will truly resemble engineering when researchers can construct models that accurately predict how a gene circuit will behave inside cells, says engineer-turned-biologist Harley McAdams of Stanford University. To do that, Collins, biomedical engineer Timothy Gardner, and their Boston University colleagues used a mathematical method from a branch of engineering called system identification to infer the design of a network--in this case, the SOS pathway that turns on genes in response to DNA damage--by monitoring parts of the network. Given messenger RNA levels produced by some genes, the algorithm correctly determined the entire circuit's wiring. The algorithm, reported in Science in July (4 July 2003, p. 102), also predicted which points in the network were blocked by a DNA-damaging drug. Pharmaceutical companies may be able to adapt the method to see if candidate drugs affect parts of the cell aside from their intended target, Collins says.
Despite the recent successes, it will take years for systems biologists to fully understand the logic of gene circuits, in part because there are so many of them. Engineers such as Endy, meanwhile, are happy to get information from any source about genetic modules that can direct the behavior of engineered microbes. They don't plan to wait around for the systems biologists, according to Endy: "Synthetic biology says, 'Screw it. You want modules? We'll build modules.' "
Like LEGO bricks
Each module was composed of parts from MIT's standard synthetic biology parts list, a data book dubbed "BioBricks." It's based on a parts list called the transistor-transistor logic data book, from which electronic circuit designers select compatible, LEGO-like modules for complex circuits. Each BioBrick is a piece of DNA; each can be spliced to any other BioBrick. Each either makes up or encodes a functional element familiar to any molecular biologist: promoters and terminators to start and stop transcription, antisense RNAs to block gene expression, ribosome-binding sites that spur cells to make protein from messenger RNA, and reporter genes that make cells glow green.
As often happens when engineers test electronic circuit designs, the MIT students were forced to improvise when their grand plans collided with reality. Each group was allowed to play with a budget of 5000 base pairs of DNA; all of them ran over budget. They learned to economize and share parts. Future synthetic biologists, like engineers, will also have to learn to specialize, says course co-instructor Gerald Sussman, with some designing circuits, others fabricating them, and still others making the larger components. "Eventually we'll be able to design and build in silico and go out and have things synthesized," says Jay Keasling, head of LBNL's new synthetic biology department.
Synthetic biologists eventually aim to make bacteria into tiny programmable computers. Like electronic computers, the live ones would use both analog circuits and digital logic circuits that perform simple computations. Rudimentary components are already taking shape. Princeton's Weiss, MIT's Tom Knight, and their colleagues made an amplifier and other analog circuits. They also made a set of genetic on-off switches that can perform basic Boolean computations and used them to fashion eight genetic circuits that work as logic gates, including a NOT gate and an AND gate. And chemist Milan Stojanovic of Columbia University and computer scientist Darko Stefanovic of the University of New Mexico in Albuquerque created a digital logic circuit made of DNA that's unbeatable at ticktacktoe. Such computers would never rival the raw computing power of their electronic cousins, Weiss says, but they'd be able to direct the operation of engineered cells.
Peripheral components are being developed, too, such as engineered cells that can sense chemicals in their environment and respond with a signal. Bacteria are already good at sensing the molecules they care about, but biochemist Homme Hellinga of Duke University in Durham, North Carolina, and colleagues have devised an algorithm to direct natural biosensor proteins to bind whatever chemical the designers want. The method, reported in Nature in May, allowed the team to reengineer a single E. coli sugar-binding protein to bind the explosive TNT; a metabolite called lactate; or serotonin, a compound brain cells use to communicate. The team members plugged the redesigned protein into an engineered gene circuit, which they stuck into a bacterium to create a bug that glows green when it sees its target chemical. Similar microbial biosensors could detect underwater ordnance or environmental pollutants or be used in medical diagnostics. "We'd be the ones who give you different components for an electronic breadboard, beyond what nature offers," Hellinga says.
Designers are working on programmable cells that would need the equivalent of a keyboard to receive input. So far researchers have used chemicals to send signals to reengineered microbes, but the MIT team is exploring ways to use light as well. And to provide a readout on what's going on inside cells--the equivalent of a monitor--the researchers will need something beyond the fluorescent jellyfish protein that's been used until now. Ideally, engineered cells would communicate with one another, allowing them to act in concert, Weiss says. To do that, he and his colleagues are rigging E. coli with quorum-sensing proteins, which other bacteria use to send and receive signals.
All together now
But getting installed functions to work reliably--and safely--will be a tremendous challenge (see sidebar on p. 159). For starters, devices put together in the lab may not work well in cells, where they'll sit in close quarters with hundreds or thousands of other biological parts, says DARPA's Eisenstadt. But Weiss, Caltech chemical engineer Frances Arnold, and their colleagues have come up with a possible way around that problem. Last year in the Proceedings of the National Academy of Sciences, they reported fine-tuning the connections of a genetic circuit using directed evolution, a test tube method by which researchers mutate microbial genes until the bugs perform the way the researchers want. The method allowed two previously incompatible parts of the circuit to work together.
Making the components compatible isn't enough, McAdams says. Synthetic biologists will have to test their prototypes for robustness--whether they work under a wide range of conditions. They'll need computer programs that predict how designed circuits would behave in the cell, he adds. And once circuits are installed, Savageau says, researchers will need the equivalent of a voltmeter to test how well the circuit is working. "What you'd like is some magic instrument where you could look at the concentration" of proteins, RNAs, and metabolites, he says. That doesn't exist, but with microarrays and other technology, "it's coming," he says.
Despite many early successes, synthetic biologists might be getting ahead of themselves. Much more needs to be known about the basics of cellular "device physics"-- including where proteins are located, how fast they turn over, and what other proteins they talk to, says Eisenstadt. "We'd like to be building life forms from first principles," says Venter, "but it's kind of hard when you don't know all the first principles." And after all is said and done, researchers may never be able to make a synthetic cell at all, Venter says: "People should not accept as a fait accompli that this will work."
Back at MIT, it's still not clear whether last year's bacterial class projects will blink. The modules were made, and Endy and co-instructor Knight's teams are still installing them in E. coli to test them. But whether they work or not, the MIT engineers are pressing on. For the second annual synthetic biology class, which kicked off this week, they'll challenge the students to make bacteria communicate with their neighbors on a petri dish to turn genes on or off. The goal this time: genetically encoded polka dots.