Chapter 7: Junk DNA and a Demolition Tool
Lenski's Bacteria
A second real example, and this one has an added virtue: it shows two separate, small-scale genetic changes building directly on each other — the second one only becoming useful because the first one had already happened, quietly, with no payoff of its own toward the final result.
In 1988, a biologist named Richard Lenski started twelve identical flasks of E. coli, fed them a daily diet of glucose, and let them run. He's still running them today. Each day, every flask grows from about five million cells to about five hundred million; then one percent is carried over into fresh broth to start the next day, and the rest is discarded or frozen. The growth medium also happened to contain citrate — a molecule ordinary E. coli cannot use as food whenever oxygen is present, which describes every one of these open flasks. It was there for an unrelated historical reason, an untouched food source, sitting in the broth the whole time.
Then, in one of the twelve flasks, and only one, something changed. Around generation 31,500 — about fifteen years into the experiment — a lineage of E. coli appeared that could finally use that citrate. The flask, which had been fairly clear, turned cloudy: a sign that a huge new food source, sitting untouched through several trillion cell divisions, had finally been unlocked.
Look closely at what actually changed, and the story turns out to be exactly the kind of two-step process this chapter is interested in.
E. coli was never missing the ability to use citrate once it was inside the cell. What it was missing was a way to bring citrate across the cell membrane in the first place, under oxygen. A molecule the size of citrate can't simply slip through a cell's membrane; it needs a gate, a protein built to let that particular molecule in. The gene that builds that gate — citT — was already there, fully functional, and the bacterium already used it whenever oxygen was absent. Under oxygen, though, the gene simply wasn't switched on. The capability was sitting right there the whole time; what was missing was a way to activate it under the wrong conditions.
Two separate genetic events, of very different character, had to happen before that activation took place.
The first, arising sometime before generation 20,000, was some ordinary change to the bacterium's background genetics — the kind of small-scale change every population in the experiment was racking up constantly, just from adapting to life in the flask. On its own, this change did essentially nothing toward using citrate. Researchers still haven't identified its exact molecular details with full confidence; one leading candidate involves a gene called gltA, sitting at an early step of the cell's core metabolism, though the full picture isn't settled. What this change did do, invisibly, was alter the bacterium's internal chemistry just enough to set up the genetic background — "potentiation," in the researchers' own term — that a much rarer, later event could actually build something useful on top of.
Surviving long enough to be built on was not automatic. A new mutation starts out in a single cell, and every day ninety-nine cells in a hundred are thrown out. A lone mutant, or a small handful of its descendants, can easily be lost in the daily cull, taking the mutation with it. What seems to have helped this one is that it carried a modest benefit of its own, unrelated to citrate: the gltA candidate, at least, has been shown to improve growth on acetate, a byproduct the bacteria were dealing with every day. Carriers of the change grew a little better, multiplied, and held their ground long enough for something else to happen to one of them. It may well have arisen before, more than once, and simply been lost. There's no way to know.
That second event didn't happen until generation 31,500, and it wasn't a simple point mutation at all. It was a duplication — an existing stretch of DNA containing the citT gene got copied, landing next to a different piece of DNA capable of switching a gene on even with oxygen present. That accidental pairing is what finally let citT run under conditions it had never operated in before. Put that same rearrangement into a cell without the first change, and it produces only a feeble trace of citrate use, too weak to matter. Nothing new was invented in either step. No new protein, no new chemical machinery came into existence. An already-functional gene was simply handed a new "on" switch — and only because an earlier, invisible change had already prepared the ground for that switch to matter at all.
The numbers give some sense of what even this took. In the one flask where it happened, something like a trillion and a half bacterial divisions had already taken place by generation 20,000, when the first change was in place — about eighteen trillion, summed across all twelve flasks. The second change took nearly another trillion divisions in that one flask, over almost five more years of daily growth. (The arithmetic: Lenski's Flasks.) And E. coli's genetic code runs to about four and a half million letters, so even the "easy" first change had to land on one of a comparatively tiny number of spots capable of producing the right effect. As for how rare the second change was, the clearest measure is the other eleven flasks: given the same span of time and a comparable number of chances, they never managed it at all.
There's an outside check on this as well. Other researchers later ran a harsher test — starving E. coli directly on citrate, rather than the milder daily glucose limitation of Lenski's original setup — and evolved citrate-using bacteria in a matter of weeks, not years. The two groups disagree about whether those bacteria needed anything like the quiet first step. But sequencing the winners turned up the same underlying class of change doing the work: a duplication that placed citT next to a switch able to turn it on with oxygen present. Turn the selective pressure up hard enough, and the decisive step can happen fast. But it still took the same kind of genetic event to get there — a duplication, not an ordinary point mutation — and it was still building on existing, already-functional machinery, not creating anything new.
Last revised 2026-11-19