DNA replication: how one DNA molecule becomes two identical copies

How to use this simulation

1. Start on the "Replication fork" view: the parent DNA is still joined on the right, helicase is unwinding it right at the fork, and two new strands are growing on the left.
2. Work through the "Step through the enzymes" list one at a time: watch where helicase acts, when primase lays a primer, how polymerase extends a strand, and when ligase flashes to seal a join.
3. Switch the "Template sequence" between the three samples and watch the base-pair colours change; then pick "Type your own" and type A, T, G, C letters of your own to see the complementary strand computed live.
4. Move to "Zoom on base pairing" and count the dashed lines between each pair — two for A–T, three for G–C.
5. Open "Meselson-Stahl experiment" and drag the Generation slider up from 0: watch the heavy, hybrid and light bands in the tube change, with the exact percentages printed on the right.
6. Watch the readings panel: bases copied and Okazaki fragments made keep climbing while it plays — press Reset any time to start those two counters back at zero.

How a cut heals: DNA gets copied every single day

A small cut on your hand closes up in a few days. Hair keeps growing after a cut. Blood cells you lose get replaced. Behind every one of these is the same event: old cells die and new cells are built, and each new cell needs a DNA copy that matches the original letter for letter. Get even one letter wrong in the wrong place and a cell can misbehave — sometimes badly enough to cause disease.

So how does a long DNA molecule inside a cell make an accurate copy of itself? That question is exactly what DNA replication answers. This page walks through DNA's structure, how it is unwound and copied, which enzyme does which job, and how two scientists proved the copying method is "semi-conservative" using nothing more than heavy and light nitrogen and a centrifuge.

The same copying principle, run in a test tube instead of a cell, is what lets forensic scientists identify a suspect from a single hair or a trace of saliva: a technique called PCR turns a few DNA molecules into billions of copies. The real-life section near the end of this page covers it in more detail.

Starting from zero: what DNA replication actually is

DNA (deoxyribonucleic acid) is a twisted ladder — a double helix — made of two strands wound around each other. Each strand is a chain of nucleotides, and every nucleotide has three parts: a deoxyribose sugar, a phosphate group, and one of four nitrogen bases. The sugars and phosphates link up to form the sugar-phosphate backbone, while the bases point inward, like the rungs of the ladder.

The four bases are adenine (A), thymine (T), guanine (G) and cytosine (C). A on one strand always pairs with T on the other, and G always pairs with C. This is the Watson–Crick base-pairing rule, from the double-helix structure James Watson and Francis Crick proposed in 1953 (built on Rosalind Franklin's X-ray images). The two strands run in opposite directions — one goes 5′ to 3′ while the other goes 3′ to 5′ right beside it — an arrangement called antiparallel.

Write the definition like this: DNA replication is the biochemical process in which the two strands of a DNA molecule separate, and each old strand is used as a template to build a new, complementary strand, producing two identical double-stranded DNA molecules. Each new molecule keeps one original strand and gains one brand-new strand — that half-and-half split is exactly what "semi-conservative" means.

Replication happens during the S phase of the cell cycle, right before a cell divides. The spot where the two strands first separate is called the origin of replication; a bacterium's single circular chromosome has one origin, and two replication forks race outward from it in opposite directions.

Key terms for this topic

These words come up in almost every question on this topic — learn them once and the rest reads easily.

Base pairing, hydrogen bonds and Chargaff's rule

In the 1950s, biochemist Erwin Chargaff measured the base composition of DNA from many different species and found something striking: in every sample, the amount of adenine equalled the amount of thymine, and the amount of guanine equalled the amount of cytosine — even though the overall A+T to G+C ratio varied a lot between species. This pattern, now called Chargaff's rule, became one of the clues Watson and Crick used to work out base pairing.

The reason sits in the number of hydrogen bonds. A and T form two hydrogen bonds; G and C form three. That extra bond makes a G–C pair a little more stable and more heat-resistant than an A–T pair. The "Zoom on base pairing" view in the simulation draws exactly these dashed hydrogen-bond lines.

The table below shows the simulation's three sample sequences. The first is perfectly balanced (25% of each base), the second is GC-rich and the third is AT-rich. Notice that %A = %T and %G = %C in every single one — that is Chargaff's rule holding regardless of the actual sequence. The GC-rich sample also carries more hydrogen bonds overall, so that stretch of DNA is more heat-stable.

The semi-conservative model and the Meselson–Stahl experiment

Once the double-helix structure was known, three models were proposed for how replication might actually copy it. Semi-conservative: each new molecule keeps one old strand and gains one new strand. Conservative: the two original strands would stay together, and a completely separate, all-new molecule would form. Dispersive: old and new segments would end up mixed together in both resulting molecules.

In 1958, Matthew Meselson and Franklin Stahl designed an elegant experiment to tell the three apart. They grew E. coli bacteria for many generations in a medium containing the heavy nitrogen isotope ¹⁵N, so both strands of every bacterium's DNA became heavy. They then switched the bacteria to an ordinary, light ¹⁴N medium and, after each generation, extracted the DNA and spun it in a caesium chloride (CsCl) density-gradient ultracentrifuge. In this setup heavy DNA settles lower in the tube, light DNA settles higher, and hybrid DNA (one heavy strand plus one light strand) bands exactly in between.

After exactly one generation, every molecule of DNA formed a single band at the intermediate (hybrid) position — no pure heavy band remained. That single result already rules out the conservative model, which would have kept a pure heavy band and produced a pure light band alongside it from generation one. After a second generation, half the DNA was hybrid and half was fully light, appearing as two sharp bands rather than one gradually shifting band — which is what a dispersive model would have produced instead. The results matched the semi-conservative model, and only that model.

The hybrid fraction at any generation follows a simple rule: after n rounds of replication, the fraction of duplexes that are hybrid equals (½)^(n−1), and the rest are fully light. At generation 3, for example, 25.00% of the DNA is hybrid and 75.00% is fully light; by generation 5 the hybrid fraction has fallen to just 6.25%, with 93.75% fully light. No generation from n = 1 onward keeps a fully heavy duplex, because every original heavy duplex splits into two separate daughter molecules the very first time it replicates. The table below runs the numbers from generation 0 through 6.

The replication enzymes and their jobs

No single enzyme carries out replication alone — it is teamwork, and each member has one specific job. The "Step through the enzymes" list in the simulation lets you pick each one and watch it act on its own; the table below summarises what each does.

Two more helper enzymes are worth knowing beyond these five. Topoisomerase (gyrase in bacteria) relieves the extra twisting (supercoiling) that builds up just ahead of the fork, which would otherwise stop helicase from unwinding any further. And in prokaryotes, DNA polymerase I removes each RNA primer and fills the gap with DNA before ligase makes the final seal.

Leading strand, lagging strand and Okazaki fragments

DNA polymerase has one strict rule: it can only extend a new strand 5′→3′, never the other way. But the two old template strands run in opposite directions (antiparallel), so at the very same fork, the two new strands have to be built by two different strategies.

Proofreading and mutation

DNA polymerase is remarkably accurate, but not perfect. Its raw error rate — a wrong base slipping in — is roughly 1 mistake in every 100,000 bases. But polymerase carries its own 3′→5′ exonuclease activity, which recognises a mismatched base right after it's added, removes it, and replaces it correctly. This proofreading step brings the error rate down to about 1 in 10,000,000.

After replication, a separate mismatch repair system catches almost all of what proofreading missed. Combined, the final error rate falls to roughly 1 mistake in every 1,000,000,000 bases copied. Whatever tiny fraction slips past both proofreading and repair can remain as a permanent change to the DNA sequence — a mutation.

Mutations are not automatically bad. Most are neutral, some are harmful (disabling a working gene, for instance), and some are even beneficial, supplying the raw material evolution works with. But when the proofreading or repair machinery itself breaks down, the mutation rate climbs sharply — one of the routes to cancer.

DNA vs RNA: the differences that matter here

DNA and RNA are both nucleic acids, but they differ in structure and job in ways this topic keeps returning to.

Try these experiments in the simulation

Press Reset before each experiment if you want the bases-copied and Okazaki-fragment counters to start back at zero.

Solved problems

Every number below is computed straight from the simulation's own model, so it can never drift from what the animation shows.

Real-life uses of DNA replication

PCR (the polymerase chain reaction) copies the exact logic of DNA replication inside a lab tube, turning a tiny amount of DNA into billions of copies. Kary Mullis invented it in 1983 and won a Nobel Prize for it. It relies on a heat-tolerant DNA polymerase (Taq polymerase, first isolated from bacteria living in hot springs), primers, and repeated heating-and-cooling cycles that double the amount of DNA every cycle.

DNA fingerprinting (DNA profiling) uses PCR to amplify a trace sample — a hair, a drop of blood, a swab of saliva — enough to examine a handful of short, repeated regions called STRs. The length pattern of those regions is unique to almost everyone except identical twins, which is why DNA profiling is central to forensic identification and paternity testing today.

Whole-genome sequencing, gene therapy, crop and livestock breeding, disease-diagnosis kits, and the RT-PCR test used throughout the COVID-19 pandemic all rest on the same underlying ability: copying DNA (or converting RNA into DNA first) accurately, at scale.

Common mistakes to avoid

Run through this list before you write an exam answer.

• Writing "replication is conservative" — wrong. The Meselson–Stahl experiment (1958) proved it is semi-conservative.
• Assuming DNA polymerase can start a brand-new strand from nothing — wrong. It can only extend an existing primer or strand, which is exactly why primase has to act first.
• Assuming the leading and lagging strands are made by different enzymes — wrong. The same DNA polymerase builds both; only the direction and the number of primers differ.
• Thinking Okazaki fragments only occur in bacteria — wrong. Both eukaryotes and prokaryotes build the lagging strand in Okazaki fragments; only the typical fragment length differs.
• Saying a light band appears at generation 1 in Meselson–Stahl — wrong. Generation 1 is entirely hybrid; a light band only appears from generation 2 onward.
• Mixing up which pair has more hydrogen bonds — remember G–C has three, A–T has two, which is also why GC-rich DNA is more heat-stable.

Exam corner

DNA replication is a staple topic across school and college biology syllabi worldwide. Expect questions on matching each enzyme to its job, applying Chargaff's rule to a given percentage, interpreting the Meselson–Stahl bands, and defining an Okazaki fragment — all of which this page covers with computed, checkable numbers rather than memorised facts.

One-screen revision summary

Read this the night before the exam.

• Definition: one DNA molecule makes two identical copies, each with one old strand and one new strand (semi-conservative).
• Chargaff's rule: %A = %T, %G = %C. A–T has 2 hydrogen bonds, G–C has 3.
• Meselson–Stahl (1958): ¹⁵N/¹⁴N labelling proved replication is semi-conservative; generation 1 is all hybrid, and the hybrid fraction halves every generation after that.
• Enzyme order: helicase unwinds → SSB proteins hold the strands apart → primase lays a primer → polymerase extends 5′→3′ → ligase seals the joins.
• Leading strand: continuous, one primer. Lagging strand: built in Okazaki fragments, many primers.
• Proofreading is polymerase's own job; mismatch repair catches the rest, bringing the final error rate below one in a billion bases.
• Real-world uses: PCR, DNA fingerprinting, forensic identification, whole-genome sequencing.

Frequently asked questions