Physical Biology: From Gene Regulation to Living Matter

# Why does life spend energy to avoid mistakes? Watch the video first. Use this companion to revisit the reasoning and its evidence limits. Call the probability of an error epsilon. Fidelity is the probability of getting it right, one minus epsilon. Lower epsilon means higher fidelity. Next, equilibrium: no net driving. Every reaction runs forward as often as backward, and nothing flows on average. Cells are different. They constantly burn fuels like ATP and GTP and maintain gradients across membranes. That doesn't violate thermodynamics: cells take in energy and release heat, like any engine. Suppose recognition used binding energy alone, at equilibrium. Right and wrong substrates both bind; the right one more tightly, by an energy difference delta. The best error rate is then a Boltzmann factor: epsilon is about e to the minus delta over k B T. If delta is a few k B T, errors are around one in a hundred. Nowhere near one in ten billion. In the 1970s, John Hopfield and, independently, Jacques Ninio proposed how energy could buy accuracy: kinetic proofreading. A substrate binds and is checked. Then an energy-consuming, nearly irreversible step, like hydrolyzing GTP, moves it to a second state, where it can still fall off before the final reaction. A wrong, weaker-binding substrate gets a second chance to leave. Proofreading has costs. Fuel is burned every cycle, even when correct substrates are discarded by mistake, and each check takes time. So there's a triangle of trade-offs: speed, accuracy and energy cost. Improving one corner usually costs another. Two cautions. Spending ATP does not automatically produce accuracy; energy has to be coupled to the right steps. And not all biological fidelity is kinetic proofreading. Induced fit, structural checks and repair after copying all contribute. Now a research twist. A 2026 Science paper by Ravasio, Husain and colleagues, a collaboration led by Arvind Murugan's group in Chicago with Rob Phillips as a co-author, asked why error correction evolves at all. The key ingredient is stalling: after a polymerase adds a wrong nucleotide, the next steps can slow dramatically. Can correcting errors make copying faster? When stalls are long, yes: cutting out the wrong base and retrying beats waiting out the stall. In their models and simulated evolution, selecting for speed alone drove the evolution of proofreading, with higher fidelity as a by-product. And they found stalling in real systems, from non-enzymatic copying to polymerases. The theme reaches gene regulation too. A 2024 Phillips Lab theory paper by Mahdavi, Salmon and colleagues showed that promoter networks driven out of equilibrium can produce responses forbidden at equilibrium: for example, one transcription factor acting as a repressor at some concentrations and an activator at others. ## Evidence guide PHYSICAL-BIOLOGY BACKGROUND: classic kinetic proofreading is a mechanism, not an explanation for all fidelity. COLLABORATIVE PRIMARY RESULT: the Murugan-led Ravasio work combines models, selection simulations, and consistency with polymerase measurements. MODEL PREDICTION: nonequilibrium promoter flexibility is theoretical. The collaborative evidence does not show universal proofreading or automatic error correction from ATP. Sources: [hopfield1974], [ninio1975], [ravasio2026], [mahdavi2024]. See the course bibliography and claim audit.