Bacteria Prepare Themselves

Microbes react to environmental changes before they occur.

When we see dark clouds, we might grab an umbrella before heading outside.  We’ve long believed that showing such foresight requires a brain and complex information-processing capability. It turns out, though, that even microbes, which do not have brains or a nervous system, can learn to use cues from their surroundings to anticipate future events, according to a new research study based on both experimental and computational techniques.


Predictive behavior of a simulated microbe species at different points along an evolutionary trajectory. The resource (food) is always given shortly after giving either, but not both, of the two signals (environmental cues). Initially (subplot 1) the response seems random relative to the food and cues. Eventually, however (subplot 4), guided by the pattern of cues, the microbe evolves its feeding response to make it synchronize with the food availability. Courtesy of Ilias Tagkopoulos. Reprinted from the supporting online material for Predictive Behavior Within Microbial Genetic Networks, Ilias Tagkopoulis, et al.,  Science 320, 1313 (2008).“What we have shown is that microbes too have the intrinsic capacity for predictive behavior,” says Saeed Tavazoie, PhD, an associate professor of molecular biology at Princeton University who published the study in the June 6 issue of Science with co-authors and Princeton colleagues Ilias Tagkopoulos, PhD and Yir-Chung Liu, PhD. “Indeed, this may be essential for their survival.” The findings could have implications for infectious disease treatment and microbial applications in industry.


Escherichia coli (E. coli) normally adjusts its breathing to match the ambient oxygen level: In the open, the bacterium breathes oxygen; inside an animal's oxygen-poor gut, it doesn't. According to prevailing notions, this switch from aerobic to anaerobic respiration is a purely reflexive response to the drop in oxygen level.


But Tavazoie and his colleagues suspected the microbes wouldn’t survive if they responded only when they were already oxygen-deprived. They proposed that, instead, E. coli senses warmth when it enters an animal's mouth, and uses this as an early cue to switch to anaerobic breathing. In laboratory experiments, the researchers found this to be the case: When the temperature rises, E. coli turns off many genes needed for aerobic respiration. “By anticipating the subsequent lack of oxygen, it improves its chances of survival,” says Tavazoie. “This is clearly predictive behavior.” Moreover, when the researchers caused oxygen levels to rise shortly after an increase in temperature, E. coli evolved (over about 100 generations) to disregard warmth as a cue. “It rewires itself to forget the old association,” says Tavazoie.


To explain how a microbe could evolve such complex behavior, the researchers devised a computational framework that mimics the essential aspects of microbe ecology. Modeled as a network of genes and proteins, a virtual bug in this virtual ecology lives and breeds when it has enough energy, or dies when it runs out of it. To gain energy, it has to be ready to eat, biochemically speaking, when “food” is available. But if it gets ready to eat and no food arrives, it wastes precious energy.  


To help the virtual bugs, the researchers gave them different patterns of cues to indicate that food is coming. In one experiment, the bugs were fed shortly after they got one of two different cues—but not if they got both cues at once. “To predict mealtimes accurately in this case, the microbes would have to solve a complex logic problem,” says Tagkopoulos, an electrical engineer associated with the Lewis Sigler Institute for Integrative Genomics. Sure enough, after a few thousand generations, a gastronomically savvy—and ecologically fit—strain of microbe emerged. The feeding response of such a fit bug (see figure) illustrates how interacting genes and proteins could evolve complex behavior.


According to David Reiss, PhD, a computational biologist at the Institute for Systems Biology in Seattle, the researchers' computational framework is notable for incorporating more biological mechanisms than prior models did. He cautions, however, that even this model oversimplifies the behavior of real microbes.  Nevertheless, Reiss says, the study is interesting and novel for showing that anticipatory behavior is not restricted to higher systems with decision-making capability.


Post new comment

The content of this field is kept private and will not be shown publicly.
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Enter the characters shown in the image.