A transistor is normally silicon, wires, and not much else. MIT's Department of Biological Engineering just built one out of bacteria. In a paper out August 17 in Nature Chemical Biology, a team led by Christopher Voigt describes engineered bacterial strains that behave like the switching components inside an electronic circuit — except instead of controlling electric current, they control the flow of small molecules that carry signals to the next part of the circuit. Print the right bacteria in the right pattern onto a petri dish, and you've built something that functions less like a science demo and more like an actual, if unconventional, circuit board.
The clever part isn't the switch itself — it's where the researchers decided to put the complexity. Most synthetic biology circuits try to cram an entire logic operation into a single cell, using proteins and transcription factors that interact with each other to sense something and produce an output. That approach runs into a wall fast: there are only so many transcription factors that won't interfere with each other, and pile too many into one cell and you overwhelm its protein-making machinery before the circuit gets anywhere near complex. Voigt's team split the problem up instead. Two engineered strains act as the transistors themselves; three more relay information between them. Five strains, deliberately kept simple on their own, but combined they're apparently enough building blocks to construct nearly any circuit you'd want — the team has already used them to build circuits that add two or three inputs together, or route a signal to a specific point in the layout.
The bacterium they picked for this isn't some lab-only specimen, either. It's Pantoea agglomerans, a species that naturally lives on plants. That's not incidental — it's the whole point. Voigt has been explicit that the goal isn't competing with silicon; it's giving biological systems computation they don't currently have access to at all. The team's stated target is coating plant leaves or roots with circuits like these, so a plant could sense drought stress or a pest attack and trigger its own response — producing a fungicide on demand, for instance — without a sensor, a battery, or a wire anywhere near it.
None of that agricultural version exists yet; what's been demonstrated so far is the underlying computing architecture, tested in a dish, not in a field. But the shift in approach is the real story here. Most efforts to blend biology and computing have gone one direction — bringing biological material into electronic systems. This is explicitly the opposite: taking computation, stripped down to switches and logic, and building it entirely out of living cells that could, in principle, just grow wherever you need them.