When Russian physician Ivan Petrovich Pavlov trained his dogs to associate the sound of a bell with an upcoming tasty morsel, he established one of the founding tenets of modern psychology—classical conditioning. Now, more than a century later, molecular biologists are demonstrating a similar anticipatory capability in the world of microorganisms.
A team of molecular geneticists at the Weizmann Institute in Israel has discovered two cases where microorganisms make predictions about how their environments might change in the future, based on how they have changed in the past. They first examined the bacteria Escherichia coli, one of several hundred microbial species that harmlessly inhabit the human gut. E. coli settle in the lower intestines after cruising through the rest of the human digestive tract—a movement that entails a switch from a lactose-rich environment to a maltose-rich environment, and thus a switch in food sources for the bacteria. The Weizmann team found that upon exposure to lactose, the E. coli immediately begin activating small amounts of genes for digesting maltose as well, even though none is yet present. In other words, the E. coli are able to somehow comprehend the presence of the lactose and predict an upcoming meal of maltose.
Princeton molecular biologist Saeed Tavazoie first began investigating this strange phenomenon after reading a 2002 paper demonstrating that yeast, when exposed to a stressful stimulus such as extreme temperature, would activate a large set of genes that served no function to combat the stress. “There seemed to be a disconnect.” Tavazoie says. “And so we decided to step back and rethink the whole idea of how microorganisms respond to their environments.”
What Tavazoie demonstrated in a study published last year, and what the Weizmann team argue in the new Nature paper, is that microorganisms are capable of interacting with their surroundings in a way that goes beyond simple reflexive behavior. Though pre-inducing genes diverts resources in the first environment and therefore temporarily hinders the organism, it confers a fitness advantage in the future. Prediction is a short-term energy expense for long-term gains in survival.
Yet, like the other millions of microorganismal species coating the biosphere, E. coli are unicellular and have no neural architecture capable of performing elaborate functions such as learning, anticipation, or prediction, let alone a cost-benefit analysis. How, then, are they able to form such complex associations and “prepare” for future events? The answer, simply, is that they don’t. The learning framework is only an analogy used to describe how these microorganisms model their environments. “A lot of our motivation and the way we started thinking about this problem arises from this analogy,” says Amir Mitchell, who conducted the Weizmann study. The real process occurs over millions of generations of rapidly evolving microbes, where the two processes of mutation and natural selection work together to generate highly fit networks that can interact most aptly with their environments—including the ability to anticipate future occurrences.
“Scientists usually don’t think of the temporal context in which microorganisms evolve or live. But these organisms actually make use of the consistency and repeating patterns in their environment in a quite fascinating way,” says Mitchell.
In further experiments the Weizmann team examined yeast, another unicellular creature that is ubiquitous in research as a “model organism.” Emulating the stresses that repeatedly occur in the process of alcoholic brew production, the researchers demonstrated that yeast show improved survival when the stresses were presented in their preserved order, as compared to an unnatural reversed order. In other words, yeast, too, can behave in a predictive fashion—a finding that expands the number of cases of genetic pre-conditioning into the realm of domesticated organisms. Unlike E. coli, which have coevolved with the digestive tracts of animals over millions of years, yeast’s metabolism has been shaped by the brewing process, an entirely man-made ecology only 7,000 years old. In the case of brewer’s yeast, man, not nature, could be playing Pavlov.
The implications of Tavazoie’s and Mitchell’s work are vast. For roughly a century, researchers have assumed that when they do something to an organism—say, subject their bacteria to heat or bathe them in sugars—the organism’s response is directly related to the stimulus. It’s a fundamental tenet in cellular biology called homeostasis. “We are proving that this idea is essentially wrong. This challenges interpretations of all experiments done on microorganisms now, ” Tavazoie says.
Audrey Gasch, a professor of genetics at the University of Wisconsin–Madison, studies similar stress responses in yeast. She agrees that an understanding of predictive gene induction will greatly alter how biological science is conducted. “Scientists really need to start thinking broadly about their results,” Gasch says. “We were duped based on a lot of assumptions in the literature, and no one had gone and tested this very basic question, which is: Are these gene expression changes really important to survive the condition that triggers their change?”
Beyond academic pursuits to understand the mechanisms of the phenomenon, these scientists have all expressed excitement about new frontiers in synthetic biology. There is now the possibility that scientists can actually force microorganisms to “learn,” programming their genetic networks for certain responses on the relatively brief timescale of months to years. Even further, scientists may soon be able to infer the ecological history of microbes based on their existing genetic networks—looking at what they have learned to find out where they have lived. Pavlov would be proud.
Originally published July 27, 2009
