Living cells can be exploited to sense and process environmental stimuli, including poorly defined microenvironments, biological markers of disease, defects in materials and complex small molecules. But obtaining a reliable signal from individual cells has proved a challenge. Prindle et al report a solution to this problem: a sensor composed of millions of bacterial cells that communicate with each other over long distances (up to 2.4 centimetres). The cells respond to the presence of arsenic by altering the rate at which they produce synchronized pulses of fluorescence.

Cells have been engineered to sense many environmental signals, including light, chemicals, touch, metal ions and pH. For example, sensors have been made in which human olfactory receptors are expressed in yeast cells. Most cellular sensors are based on a protein or messenger RNA that responds to a signal by causing the expression of a gene. Such genetic sensors often suffer from low dynamic range (that is, there is little change in output between the absence and presence of a signal) and nonspecificity (they are activated by multiple signals). Furthermore, because cells are living systems, individual responses may vary because of stochastic effects or differences in growth states.

Prindle et al. have addressed the problem of dynamic range by applying the principles of signal processing to a biosensor based on genetic circuits. Such circuits use biochemical interactions to produce functions analogous to their electronic counterparts. Previously, the same group built a robust genetic oscillator - a network of genes and proteins that produced regular pulses of molecules  and used this as a time-keeping mechanism to control cell-cell communication between bacteria. This yielded populations of bacteria that expressed a fluorescent protein in unison, and so produced synchronized pulses of light.

Theoretically, such an oscillator would enable a sensor to use the frequency of oscillations as signals, making the sensor less sensitive to environmental noise and exposure time than systems based on steady-state signals. A problem with the previous oscillator, however, was that cell–cell coupling relied on the diffusion of a small molecule through cellular media, a process that is too slow to allow rapid, long-range coupling of millions of cells. Molecular diffusion in the gas phase is much faster, so Prindle et al. used this mechanism to accelerate the coupling between separate colonies of bacteria.

In this way, the authors were able to couple 2.5 million cells of the bacterium Escherichia coli, which were arranged as an array of colonies across a distance of 5 millimetres. As in the previously reported oscillator, the output of the system was the coordinated, oscillating expression of a fluorescent protein, which the authors detected using a microscope. The period of the oscillations was quite long (more than an hour), but the degree of synchronization was high and the colonies produced light pulses within 2 minutes of each other.

To demonstrate a potential application of their system, Prindle et al. ‘rewired’ their network to incorporate elements that respond to arsenic. The resulting system acted as an arsenic sensor: once the concentration of arsenic reached a threshold value, the amplitude and period of the oscillations increased significantly. This produced a device with a large dynamic range. What’s more, because the device averaged the outputs of a population of cells, noise was reduced and the sensor’s response was decoupled from the growth state of individual cells. The authors scaled up their device so that it included more than 12,000 communicating bacterial colonies, covering an area of 2.4 × 1.2 centimetres.

Several advances, yet to be achieved, would improve the ability to connect this living sensor to an electronic system. One problem is that the minimum response time of the sensor to an input signal is slow, because gene expression  which takes about 20 minutes to occur  is required. Another issue is that the output involves fluorescence, which is awkward for electronic devices to use; the ideal output would be a direct electrical signal. To this end, cells have been metabolically engineered so that they can be induced to release electrons, which can then be read by an electronic sensor. The electron-transport system found in bacterial nanowires (extracellular appendages that conduct electricity) has also been harnessed to link cells to an electronic system. Nevertheless, these strategies still require the expression of a gene that triggers electron flux, so the resulting sensors are relatively slow to respond to signals.

More broadly, there are several collaborative research efforts aiming to develop better toolboxes for building interfaces between cellular and electronic components. One such project is to build a millimetre-scale robot that swims like a lamprey, using a combination of human muscle cells, yeast-based sensors, an electronic brain and flexible materials (nicknamed ‘cyberplasm’). Another project is to develop genetic sensors, along with genetic circuits to apply signal processing within the cell and new approaches to link cellular outputs to an electronic system, with the ultimate objective of controlling robots. As the integration of cellular and electronic systems matures, it will be interesting to see how circuitry in future devices is divided between biological and electronic components.

Source: Nature