How DNA Replicates Itself

Every time one of your cells divides, it must first copy roughly three billion letters of DNA, and it must do so with astonishing fidelity — typically fewer than one error per billion bases. The mechanism that makes this possible was already half-suggested by the structure of DNA itself. When James Watson and Francis Crick proposed the double helix in 1953, they ended their famous paper with a sentence whose understatement has become legendary: it had not escaped their notice that the pairing they proposed immediately suggested a copying mechanism. Each strand, they realized, carries the information needed to specify its partner.

DNA is built from two strands wound around each other, held together by hydrogen bonds between paired bases. Adenine pairs with thymine; guanine pairs with cytosine. This complementary base pairing is the key. If you separate the two strands, each one becomes a template: wherever the old strand has an A, the new strand needs a T; wherever the old has a G, the new needs a C. The result is two daughter molecules, each containing one original strand and one freshly built strand. This pattern is called semiconservative replication, and it was confirmed in 1958 by Matthew Meselson and Franklin Stahl in an experiment now considered one of the most elegant in biology.

In practice, copying the helix requires a small crowd of enzymes working in coordination. Replication begins when an enzyme called helicase pries the two strands apart, unwinding the helix at sites called origins of replication. The unwound region forms a Y-shaped junction known as the replication fork. Single-strand binding proteins coat the exposed strands to keep them from snapping back together.

The enzyme that actually builds the new strands, DNA polymerase, has a quirk that complicates the process: it can only add nucleotides to an existing chain, and it can only extend that chain in one direction, conventionally called 5' to 3'. It cannot start from nothing. So before polymerase can act, another enzyme called primase lays down a short RNA primer — a starter sequence — to give polymerase something to build onto.

The directional constraint creates an asymmetry at the replication fork. One of the two template strands is oriented so that polymerase can follow the fork continuously as it opens. This is the leading strand, and it is synthesized in one smooth run. The other template runs the wrong way relative to the moving fork. Polymerase cannot work backward, so it copies this strand in short bursts, each starting with its own primer and heading away from the fork. These fragments, called Okazaki fragments after their discoverers, must later be stitched together. This is the lagging strand.

Finishing the job requires more enzymes. The RNA primers are removed and replaced with DNA. An enzyme called ligase seals the nicks between fragments, joining the lagging strand into a single continuous molecule. Along the way, DNA polymerase proofreads its own work, checking each newly added base against the template and excising mistakes. Additional repair systems catch errors the polymerase missed.

The overall picture is worth holding in mind. Replication is not a single step but a choreography: unwinding, priming, continuous synthesis on one side, fragmented synthesis on the other, and final stitching and proofreading. The reason for the choreography traces back to a structural fact — the two strands of DNA run in opposite directions, and the enzyme that copies them works in only one. Much of what looks like baroque complexity in molecular biology turns out, on inspection, to be the consequence of a simple constraint meeting an unforgiving demand for accuracy.