I got interested in cell-free systems the wrong way around. I was reading about a biosensor — a strip that detects Zika virus RNA, freeze-dried onto paper, activates in twenty minutes when you drop a sample on it — and I assumed there were live cells in the strip. There weren’t. The strip was running transcription, translation, and a reporter circuit using nothing but the molecular machinery extracted from E. coli and lyophilized into a wafer. I had been carrying around a stale mental model of what “biology” needed in order to happen.
The basic idea of cell-free protein synthesis (CFPS) is older than I expected — Marshall Nirenberg used a crude E. coli extract in 1961 to crack the first codon, UUU coding for phenylalanine. What was new, in the 2010s, was treating the extract not as a research curiosity but as a production substrate. You crack open cells, spin out the membranes and the genome, and what’s left in the supernatant — the “lysate” — is essentially the protein factory minus the factory walls. Add DNA, the right cofactors (ATP, GTP, amino acids, an energy regeneration system), and the lysate will transcribe and translate the DNA for several hours before it runs out of fuel.
What I find appealing about this is what it does for iteration speed. In a living cell, every change you make to a circuit gets filtered through whether the cell is willing to grow with it. Build a metabolic pathway that produces a slightly toxic intermediate — even one the cell tolerates poorly — and you spend a week debugging growth before you know whether the pathway worked. In a cell-free reaction, the cell doesn’t need to survive. It is, in a sense, already dead. So you can express toxic proteins, run pathways that would never reach steady state in vivo, and try fifty variants in an afternoon by pipetting DNA into well plates.
The first time I tried to think through the costs, the obvious one was that cell-free is expensive per microgram. You’re paying for ATP regeneration, you’re paying for the cofactor mix, you’re not getting the cell’s self-maintenance for free. For a manufacturing run of insulin, you’d be insane to use it. For prototyping — for figuring out whether a circuit composes the way you hoped before you commit it to a strain — it’s a different equation. James Collins’s group at MIT has been pushing this hardest, both for the diagnostics work and for what they call “freeze-dried cell-free” reactions, which sit at room temperature indefinitely and activate on rehydration.
There’s a more interesting question lurking here, which is what gets harder, not easier, when you remove the cell. Membrane proteins are the obvious one — without lipid bilayers, you’re stuck unless you supply liposomes or nanodiscs. Anything that depends on compartmentalization (cofactor channeling, pH gradients, signaling cascades) loses its scaffold. The cell wasn’t just a container; it was a topology. The lysate is well-stirred and homogeneous, which is exactly the wrong starting condition for a lot of what biology actually does.
I’m still puzzling over how much of the field’s enthusiasm will translate to large-molecule manufacturing. The diagnostic-strip story is real. The “we’ll make our biologics in a tube” story feels more aspirational. But the prototyping argument seems durable: when you want to know whether your circuit composes, you no longer have to ask a cell for permission.
The natural follow-up is the synthetic-cell project — building lipid vesicles that contain a minimal cell-free system, which is the field’s attempt to put the topology back. Magdalena Bezanilla and Petra Schwille have nice recent reviews.