Teaching Bacteria to Behave

Single-celled organisms could be "trained" to deliver drugs.

A century after Pavlov's dog first salivated at the sound of a bell, researchers are saying that single-celled organisms such as bacteria can be "trained" to react in a similar way. Rather than use complex networks of nerve cells, or neurons, bacteria can "learn" to associate one stimulus with another by employing molecular circuits, according to a multidisciplinary team from Germany, Holland, and the United Kingdom.

This raises the possibility that bioengineers could teach old bacteria new tricks by having them act as sentinels for the human body, ready to spot and respond to signs of danger, the team says in the October issue of Journal of the Royal Society Interface (DOI: 10.1098/rsif.2008.0344). The basis for the claim is that single-celled organisms are able to associate stimuli that are applied simultaneously, according to the theoretical model produced by Chrisantha Fernando at the U.K.'s National Institute for Medical Research, in London, and his collaborators.

Bright bug: Chrisantha Fernando of the U.K.'s National Institute for Medical Research, in London, says that bacteria such as this E. coli can be trained through associative learning. Credit: Rocky Mountain Laboratories, NIAID, NIH

As with Pavlov's dog and all other examples of associative learning, the bacteria in the model learn to build stronger associations between the two stimuli the more they occur together. The Canadian neuropsychologist Donald Hebb established an underlying explanation back in 1945. Now called Hebbian learning, it's often expressed as a situation in which "neurons that fire together wire together." In the hungry dog's case, nerve cells triggered by the smell of food started to make physical links with the nerve cells simultaneously triggered by the sound of a bell. According to Hebb's theory, the more often the two stimuli are applied at the same time, the greater the link or "synaptic weight" between them.

Bacteria, of course, don't have synapses or nerve cells. Nonetheless, there are indications that single-celled organisms can learn. In the 1970s, Todd Hennessey claimed that paramecia, the single-celled pond dweller, could be conditioned in the lab. He electrocuted them and associated this with a buzzer. Following the simultaneous exposure to the buzzer and to electric currents, he claimed that the paramecia swam away from the buzzer when they had not done so before. The finding was never properly reproduced, but it raised the intriguing possibility that some sort of associated learning was possible for single-cell life forms.

Now Fernando's team has proposed a model for how bacteria might be trained. He has designed a cellular circuit that consists of several genes and their promoters, which produce proteins (transcription factors) that act to switch each other on and off like a digital electric circuit. The researchers' theoretical circuit consists of three fictional genes. Two of these genes, A and B, produce proteins pA and pB, which react with other transcription factors, iA and iB, to switch on the third gene, C.