Chemical biology news from across RSC Publishing.
Instant insight: Living computers
15 March 2007
Hao Song, Cheemeng Tan, and Lingchong You, at Duke University in Durham, US, explain how cells can solve problems
Can we exploit and reprogram these networks to accomplish specific computational tasks? One may argue that basic programmed cellular computation has been achieved with engineered gene circuits, including logic gates and switches. Nevertheless, these examples are so far limited to fairly simple logic functions.
How about more complex computational tasks? To address this question, we may draw inspiration by examining some of the characteristics of natural cellular information processing, especially in comparison with digital computers.
Firstly, whilst cells can adapt to their environment to enhance fitness, integrated electronic circuits cannot change once constructed. Meanwhile, cellular information processing is naturally parallel: different types of membrane receptor proteins mean that each cell can receive multiple signals at the same time, and a large number of cells (up to 1010 per millilitre) can process information simultaneously.
Moreover, cells can move. By modifying the cell machinery, cell migration can be controlled for a specific purpose. This aspect will become more attractive if coupled with cells' ability to communicate with one another. Combining these features could allow scientists to program cells to form self-organised patterns and optimise their distribution over complex landscapes.
By reprogramming their genes, bacteria (red) could be used to solve computational problems. Here, the bacteria are shown migrating toward a point where a function (concentration of a ligand) is a maximum.
While some of these capabilities may also be realised with digital computers, heat accumulation in silicon chips becomes a significant constraint in accelerating their operational speed any further. This highlights some additional advantages of cellular computation: energy consumption is extremely small and heat generation is easily manageable.
By examining and exploiting cell signalling characteristics, it is possible to frame computational problems where cells might perform comparably well with or even better than digital computers. As a start, we proposed three commonly-encountered problems (integration, optimisation and Fourier transform) that could feasibly be solved by engineering cell gene circuits. These examples attempt to take advantage of the key information processing capabilities outlined above, in particular parallelism, communication, and motility.
Take the optimiser as an example. Inspired by a computational search algorithm, this circuit design attempts to use cell mobility controlled by both ligand sensing and cell-cell communication to effectively search a complex landscape with unevenly distributed ligand concentrations (see figure).
Addressing the challenges and limitations of programming cells to compute will enhance our understanding of the mechanisms involved in natural cellular information processing and is critical if we are to build functional cellular computers.
Read Tan, Song, Niemi, and You's review 'A synthetic biology challenge: making cells compute' in April's issue of Molecular BioSystems.
Link to journal article
A synthetic biology challenge: making cells compute
Cheemeng Tan, Hao Song, Jarad Niemi and Lingchong You, Mol. BioSyst., 2007, 3, 343