05. The Upgrade to DNA

A single strand of RNA sits in warm, faintly alkaline water, roughly four billion years ago, in the same window when the RNA world was still running and LUCA's population was assembling the frozen code from the last leg of this journey. At one point along its backbone, a small chemical group, a hydroxyl, one oxygen and one hydrogen sitting on the second carbon of the sugar ring, reaches around and attacks the very bond holding the backbone together. Nothing hostile does this. It is built into the molecule's own chemistry: give that hydroxyl group a little heat, a stray metal ion, or alkaline water, and it will cut its neighbor's throat. Every RNA molecule, then and now, carries this flaw as standard equipment.

DNA's fix is to remove the flaw at its source. DNA's sugar, deoxyribose, is missing exactly that one hydroxyl group, which is what the "deoxy" in its name means: literally, without that oxygen. No hydroxyl, no self-inflicted knife. Tomas Lindahl's classic 1993 review quantifies the payoff: DNA's backbone bonds are roughly two hundred times more stable than RNA's, measured at the neutral pH and physiological magnesium concentrations a living cell actually runs. That is not a trivial improvement. It is the difference between an archive that can hold its shape for a lifetime and one that cannot.

The logic that follows is simple and unforgiving. Any lineage able to move its master archive onto the sturdier molecule would suffer fewer spontaneous backbone breaks per generation, a standing chemical pressure favoring the switch the moment it became biochemically possible. DNA does not eliminate copying errors; it only slows the rate of spontaneous decay enough to buy time between them, time for repair systems to catch mistakes, time for a genome to grow larger without collapsing under its own damage. The gamble still shows up today in the handful of lineages that chose not to make the switch: influenza, coronaviruses and HIV all still carry their genomes as RNA, and all three pay for it in mutation rates far higher than any cellular genome could survive carrying in its own chromosomes.

A second clue to the same transition hides in a single altered letter. DNA uses thymine, essentially uracil with a small methyl tag bolted on, where RNA uses plain uracil, and the reason reads like a forensic signature system. Cytosine, one of the four genetic letters, spontaneously degrades into uracil at a slow, steady background rate through a chemical process called deamination, in which the molecule loses a nitrogen-and-hydrogen group and turns into a different base entirely. If DNA used plain uracil the way RNA does, there would be no way to distinguish a correct original letter from that kind of damage. By making the real letter thymine instead, any uracil found sitting in a DNA strand is automatically recognizable as a mistake rather than a legitimate letter.

Cells exploit that signature with dedicated machinery. Four distinct human enzymes, UNG2, SMUG1, TDG and MBD4, each one a uracil-DNA glycosylase, a molecular proofreader whose only job is spotting uracil where thymine belongs, recognize and cut out that uracil, triggering a cut-patch-and-reseal cycle called base-excision repair. Hans Krokan and Magnar Bjørås's 2013 review calls this pathway evolutionarily highly conserved. No independent date is attached to that phrase, but conservation this deep, across such distantly related organisms, usually means old, plausibly tracing back to the transition itself.

A third clue sits in the machinery that builds DNA's raw materials, and it is the most telling of the three: there is no known biological route to make a DNA building block from scratch. The only known path runs through an enzyme called ribonucleotide reductase, which takes an RNA building block and strips off that same troublesome hydroxyl group, converting it directly into a DNA building block. Every dividing cell alive today still makes its DNA out of chemically modified RNA parts, every single time it divides. That is not a metaphor; it is current, ongoing biochemistry, and it is a fossil record hiding in plain sight. DNA synthesis is still mechanically parasitic on RNA chemistry, exactly as it would be if DNA arrived second.

The enzyme's own history reinforces the point. In 1997, Jordi Riera and colleagues cloned and sequenced this reductase from Pyrococcus furiosus, a heat-loving archaeon, and found it to be a class II, adenosylcobalamin-dependent enzyme carrying two intein insertions, a structurally distinct form that nonetheless shares strong sequence resemblance with the two other known classes of the same enzyme running in aerobic and strictly anaerobic organisms today. In 2015, Daniel Lundin and colleagues synthesized this into a single hypothetical ancestral enzyme, nicknamed the urRNR, that diversified into the three classes still dividing the tree of life: Class I in oxygen-breathing organisms, Class II in archaea and bacteria such as P. furiosus, Class III in strict anaerobes. Lundin's own team is candid that the true ancestral enzyme has never been identified, and calls it an open enigma whether an RNA-based enzyme, a ribozyme, doing a job usually left to proteins, could ever have managed this particular, radical-driven chemistry unassisted.

The dates bound each other. The RNA world is conventionally placed at roughly four billion years ago; Paul Higgs and Niles Lehman's 2015 review, treated in the field as the standard modern synthesis on the question, draws together comparative genomics, ribozyme-catalysis experiments and molecular-clock data to reach that figure, and Lehman's own companion editorial states it without hedging: the RNA world is "now some four billion years behind us." The switch to DNA-and-protein genomes is squeezed into the same narrow window, essentially complete by the time LUCA's population was already running its frozen code, roughly 4.2 billion years ago by the previous leg's dating. How long the RNA world itself lasted before that handover is still argued over, anywhere from perhaps a hundred million years to considerably longer.

Two molecules, two jobs

Once DNA existed as an option, life did not simply trade one molecule for the other. It split one job into two. DNA became the permanent archive, filed away and copied only when a cell divides. RNA became the disposable working copy, transcribed fresh off the DNA archive whenever a particular protein is actually needed, then broken down and discarded. This division is the core of what molecular biologists later named the Central Dogma: information flows outward from a stable archive to disposable working copies, not the reverse.

The scale of that disposability has been measured directly, not merely assumed. A 2003 genome-wide study tracked the decay of every messenger RNA transcript in Escherichia coli, a workhorse gut bacterium, across 1,036 genes and 329 operons, an operon being a set of genes read off the DNA together as a single unit, and found the average message survives just 6.8 minutes before it is broken down. Compare that to the DNA archive itself, which persists across a cell's entire lifetime and is copied faithfully into every descendant. Two molecules, two wildly different lifespans, doing two very different jobs in the same cell.

That split is not treated as a weakness today. Cells now exploit short RNA lifespans on purpose, switching genes on and off quickly because the messenger evaporates fast once production stops. Whether that speed was the original reason short-lived RNA persisted, or whether it is an inherited holdover from RNA-world chemistry that later proved useful for an unrelated reason, remains an open question rather than a settled one.

The living fossil of the whole transition runs in the opposite direction from normal. Reverse transcriptase is an enzyme that reads RNA and writes DNA, the reverse of the usual archive-to-messenger flow, and it is not a curiosity confined to retroviruses like HIV. It turns up, repeatedly borrowed and repurposed, across every domain of life: in LINE-1 retrotransposons, mobile stretches of DNA that copy-paste themselves through an RNA go-between; in bacterial retrons and CRISPR-associated defense systems; in self-splicing group II introns; and, in eukaryotes with linear chromosomes, in telomerase, the enzyme that rebuilds the protective caps on chromosome ends.

A phylogenetic analysis posted in October 2025, still a preprint and not yet through formal peer review, traced telomerase's catalytic core to a specific bacterial antiviral defense system named DRT10, a system that already used a noncoding RNA template to write short repeated stretches of DNA, sharing structural and sequence parallels in how that RNA template dictates the pattern of repeat addition, long before there were chromosome ends left needing protection. If that finding survives review, reverse transcription was not invented to solve the telomere problem at all; it was inherited from a far older bacterial toolkit and put to a new use. It is flagged here as provisional, since preprint evidence has not yet cleared the same bar as the rest of this leg's sources.

Every one of these clues points the same direction. DNA reads less like a replacement for RNA than a repair job built on top of it, still wearing RNA's fingerprints in its own raw materials and in enzymes that run the whole relationship backward. What DNA bought the lineage was time: an archive that could hold its shape for longer between copies. What it could not do on its own was read itself back out into a working protein. That reading job belonged to a separate piece of machinery, one already ancient by the time DNA took over the archive, and it is still running, essentially unchanged, in every cell alive.

Overturned if: a natural cellular pathway were found that synthesizes DNA building blocks directly, without routing through an RNA precursor and ribonucleotide reductase, breaking the molecular fossil record tying DNA's own chemistry back to an RNA origin.

Sources

• 1993 · Lindahl, T., "Instability and decay of the primary structure of DNA" — Nature 362:709–715, PMID 8469282
• 2013 · Krokan, H.E. & Bjørås, M., "Base Excision Repair" — Cold Spring Harbor Perspectives in Biology 5:a012583
• 2015 · Lundin, D. et al., "The Origin and Evolution of Ribonucleotide Reduction" — Life 5:604–636, PMC4390871
• 2015 · Higgs, P.G. & Lehman, N., "The RNA World: molecular cooperation at the origins of life" — Nature Reviews Genetics 16:7–17
• 2003 · Selinger, D.W. et al., "Global RNA Half-Life Analysis in Escherichia coli Reveals Positional Patterns of Transcript Degradation" — Genome Research 13:216–223
• 2025 · (preprint) "Antiviral reverse transcriptases reveal the evolutionary origin of telomerase" — bioRxiv / PMC12632851