How DNA Copies Itself Before a Cell Divides

DNA is shaped like a twisted ladder. The sides are made from repeating sugar and phosphate units, while the rungs are pairs of chemical bases. Adenine pairs with thymine, and cytosine pairs with guanine. These matching rules are the key to copying DNA.

When copying begins, an enzyme called helicase moves along the DNA and separates the two strands. It breaks the weak attractions between paired bases, creating an open region called a replication fork. Each original strand can now act as a template for building a new partner.

The cell does not unzip an entire chromosome in one dramatic movement. Copying starts at many points and spreads along the DNA. This lets the cell finish the enormous job more efficiently, instead of waiting for one tiny copying team to crawl from one end to the other.

An enzyme called DNA polymerase adds new DNA building blocks beside the exposed bases. Wherever the template has adenine, the new strand receives thymine. Wherever it has cytosine, the new strand receives guanine. The order on the old strand therefore determines the order on the new one.

The two template strands point in opposite directions, and DNA polymerase can build in only one direction. One new strand is made smoothly. The other is built in short pieces that are later joined by another enzyme called ligase. The result looks tidy at the end, although the work was carried out in two slightly different ways.

Each finished DNA molecule contains one original strand and one newly built strand. This is called semiconservative replication. The old strand is not thrown away. It remains as half of the new double helix and guides the construction of its matching partner.

DNA polymerase checks much of its own work as it moves. Other repair systems inspect the DNA afterward and fix many remaining errors. A few changes can still escape, and these changes are called mutations. Most have little effect, some are harmful and a small number can create useful variation. Cells also pause at checkpoints before division, giving repair systems more time to detect damage that could cause larger problems later.

You can remember the copying process as a calm sequence:

• •The double helix opens.
• •Each old strand becomes a template.
• •Matching bases are added.
• •Short pieces are joined where needed.
• •Repair systems check the finished copy.

When the double helix was described in the 1950s, there were several possible ways it might copy itself. One idea was that the original molecule stayed intact while a completely new one was made. Another was that the old and new material ended up scattered in patches. In 1958, Matthew Meselson and Franklin Stahl designed an experiment to decide between these ideas. They grew bacteria in a medium containing a heavy form of nitrogen, so the bacteria's DNA became slightly denser than normal.

They then moved the bacteria to a medium with ordinary, lighter nitrogen and let them divide. Using a centrifuge, they separated DNA by density. After one round of copying, all the DNA had an intermediate density, halfway between heavy and light. After a second round, there were two bands: one intermediate and one light. This is exactly what semiconservative replication predicts, because each new molecule keeps one old strand. The experiment is often admired as a clear example of a well-designed test that rules out competing explanations.

Scientists can copy a chosen stretch of DNA outside a cell using a method called the polymerase chain reaction, or PCR, developed by Kary Mullis in the 1980s. It borrows the cell's logic but replaces some of its tools. Instead of helicase, heat separates the two strands, usually at around 95 degrees Celsius. The mixture is then cooled so short pieces of DNA called primers can attach either side of the target region. A DNA polymerase then builds new strands from the template, just as it does in a cell.

Ordinary polymerase would be destroyed by the repeated heating, so PCR uses an enzyme from a bacterium called Thermus aquaticus, found in hot springs at Yellowstone National Park in the United States. Each cycle of heating and cooling roughly doubles the number of copies of the target region, so about thirty cycles can turn a few molecules into many millions or more within a couple of hours. PCR is used in forensic science, medical diagnosis and research, and was the basis of many tests for viral infections.

Copying DNA has an awkward catch at the very tips of each chromosome. DNA polymerase cannot start a new strand from nothing; it can only extend a short starter piece called a primer, which in cells is made of RNA by an enzyme called primase. When the primers are later removed, each gap is filled by extending the neighbouring piece of new DNA. At the very end of the strand built in short pieces, there is no neighbouring piece to extend from, so a small stretch stays uncopied. As a result, a linear chromosome tends to lose a little from its end each time it is copied.

Chromosomes cope with this using telomeres, stretches at each end made of one short sequence repeated over and over. In humans the unit is TTAGGG, present hundreds to thousands of times, so the losses eat into spare repeats rather than genes. In many body cells telomeres shorten gradually with each division, and very short ones can signal a cell to stop dividing. Cells that make eggs and sperm, along with some stem cells, use an enzyme called telomerase, which carries its own RNA template and rebuilds the repeats. Many cancer cells switch telomerase back on, which helps them keep dividing far beyond the normal limit.

The figures involved show why copying must start at many places at once. A human cell holds about 6 billion base pairs of DNA, counting both sets of chromosomes. In human cells, a replication fork adds roughly 50 nucleotides each second. Chromosome 1 alone is about 250 million base pairs long, so if its copying began at a single point and spread in both directions, it would take around a month. Instead, the cell fires tens of thousands of starting points, called origins of replication, across its chromosomes, and the whole genome is usually copied within several hours.

Accuracy comes in layers. When DNA polymerase chooses which building block to add, it picks the wrong one only about once in every hundred thousand or so. Its proofreading ability, which lets it step back and remove a mismatched base, catches most of those errors, improving accuracy around a hundredfold. Separate mismatch repair enzymes then scan the new strand and fix most of what remains. Together, these steps bring the error rate down to roughly one mistake in a billion bases or better. Across a whole human genome, that means each round of copying leaves only a few new uncorrected changes.

DNA replication works because each base has a reliable partner. The strands separate, enzymes build matching strands and repair systems check the result. Slow, careful copying gives dividing cells the instructions they need.