Living Patterns

For scientists interested in reaction networks, biological systems--both natural and artificial--provide excellent examples. In recent studies, researchers have used biological pattern formation as a model for complex reaction networks that could be helpful in applications such as tissue engineering or device fabrication.

Reaction networks are sets of interconnected chemical reactions that combine to perform a function. Rustem F. Ismagilov, an associate professor of chemistry at the University of Chicago, has been studying the development of fruit fly embryos to learn about these networks. "We're interested in reaction networks that are not uniform in space and time," he says. He acknowledges that Drosophila development is an unusual line of research for a chemist, but he believes that chemists can learn much from such systems. "Development happens through this orchestrated set of events where reactions take place exactly at the right place and the right time," he says. "Development is perhaps the pinnacle or most complex manifestation of these complex networks."

Ismagilov, collaborating with biologist Nipam H. Patel at the University of California, Berkeley, used microfluidics to perturb the development of fruit fly embryos (Nature 2005, 434, 1134). He and coworkers, including graduate student Elena M. Lucchetta, used a microfluidic device to subject Drosophila embryos to an unnatural environment in which the two halves of an embryo are held at significantly different temperatures. The researchers measured expression of a protein called Hunchback to determine whether development was normal. This early marker of development is expressed on only one side of the embryo, and that side ends exactly at the embryo's midpoint.

"If you think about an embryo as a bag of chemicals, which it is, then it is very amusing to think about how you would have a chemical system that finds its middle," Ismagilov says. "If you just have a beaker of chemicals, you don't see exactly half of the beaker turning blue suddenly."

Ismagilov hypothesized that embryos in the topsy-turvy temperature-differentiated environment would not find their middle accurately. "Changes in temperature affect the distribution of other proteins and should change how Hunchback is expressed," Ismagilov says.

HIS HUNCH was wrong. "The system somehow filters all these temperature perturbations, and the embryo can still find its own middle," Ismagilov says. Even though the "hot" side developed faster than the "cold" side, development occurred normally. When the embryo was removed from the microfluidic device and allowed to finish developing, the two sides caught up to form a normal larva. The system invokes some compensatory mechanism to ensure normal development.

Biologists had observed compensation in response to less extreme environmental fluctuations during Drosophila development. They thought it was due to a simple opposing chemical gradient; Ismagilov's results show that it is more complicated.

Ismagilov's group tried to pinpoint the compensation mechanism by perturbing the system not only in space but also in time. During development, they flipped the temperature distribution by switching the hot and cold ends. The reversal didn't affect compensation--the embryo still found its midpoint--except when reversal occurred between 65 and 100 minutes into development. The result "doesn't prove that's when compensation takes place, but it strongly suggests what the time window is," Ismagilov says.

The environmental perturbations revealed information about compensation that could not have been learned as easily by genetic manipulations, such as mutations or gene knockouts. With compensation, Ismagilov explains, identifying the genes is difficult because they are essential to survival only in limited circumstances. "If you're looking for a correction mechanism, you might miss it by doing only mutations," Ismagilov says. Ultimately, both genetic manipulation and environmental perturbation must be combined to identify how the network operates and what molecular components are involved, he adds.

The team's results show that the reaction networks in Drosophila development are robust and able to proceed normally, even when faced with unusual conditions that the organism has never experienced during the course of evolution.

Chemists have been proficient at making molecules and molecular assemblies but not with reaction networks, where multiple coupled reactions are running together, Ismagilov says. "A reaction network is a process, rather than a substance you can put in a vial. It would be exciting if we could build synthetic chemical reaction networks with the elegance of molecular synthesis." Chemistry could make a tremendous impact by providing a conceptual framework for reaction networks and the importance of space and time to those networks, he adds.

While Ismagilov is studying a naturally occurring reaction network, Ron Weiss of Princeton University is building a synthetic reaction network with engineered cells (Nature 2005, 434, 1130). Weiss, an assistant professor of electrical engineering and molecular biology, is interested in using reaction networks to make templates for tissue engineering and device fabrication. "By building networks from scratch, modeling them, and trying to understand them quantitatively, you can get some insight into the natural processes," he says.

In Weiss's network, engineered "sender" and "receiver" cells create various patterns based on the response of receiver cells to a chemical gradient set up by sender cells, which synthesize acylhomoserine lactone (AHL). The receiver cells, which are engineered with different sensitivities to AHL, respond over different concentration ranges. If the concentration of AHL is in the correct range, the cells fluoresce.

BY MIXING cells programmed to produce green fluorescent proteins with those wired to produce red fluorescent protein, Weiss generates different patterns from receiver cells uniformly distributed on the surface--including ellipses, circles, hearts, and clovers--simply by rearranging sender cells. No relationship exists among the network components other than what Weiss and coworkers have defined. "You want to keep things modular," he says. "You take parts that have not previously interacted with each other, and you put them together in determined ways. You can basically create any programmed response."

So far, Weiss has looked only at one-way communication from senders to receivers. Next, he would like to study communication flowing in both directions. "This work is pretty much the beginning of trying to make synthetic multicellular systems," Weiss says. "In nature, as well as what we would like to do, many processes are multiway, so you send messages back and forth." To achieve that, Weiss's team is working on many different networks, including feedback loops and those involving gradients of multiple signaling molecules.

A potential application is engineering stem cells to make patterns that can serve as templates for tissue engineering. Although the response of Weiss's current network is production of red or green fluorescent protein, a more sophisticated differentiation might create muscle, bone, or nerve cells in response to chemical cues.

Another application is device fabrication. "We want to harness cells to build structures and solid-state devices for us," Weiss says, "not necessarily to compete with the way we make computer circuits now but rather to be able to create materials and devices" that can't be made through conventional techniques.