The three steps of DNA replication are initiation, elongation, and termination. During initiation, the double helix unwinds at specific starting points. During elongation, enzymes build new DNA strands by matching nucleotides to each exposed template. During termination, the new strands are finalized, gaps are sealed, and two complete DNA molecules result. Each step involves a coordinated team of proteins working at remarkable speed, up to 1,000 nucleotides per second in bacteria and about 50 per second in human cells.
Step 1: Initiation
Replication begins at specific spots on the DNA molecule called origins of replication. These are stretches of DNA rich in adenine-thymine base pairs, which form only two hydrogen bonds (compared to three for guanine-cytosine pairs) and are therefore easier to pull apart. In bacteria like E. coli, there is typically a single origin on one circular chromosome. Human cells, with far more DNA spread across 46 linear chromosomes, fire thousands of origins simultaneously so the entire genome can be copied within a reasonable timeframe.
At each origin, a group of initiator proteins recognizes the sequence and recruits a helicase, the enzyme responsible for prying the two strands apart. Helicase is a ring-shaped motor that threads onto the DNA and travels along it, breaking the hydrogen bonds between base pairs as it goes. This creates a Y-shaped structure called the replication fork, where the two separated strands are exposed and ready to serve as templates.
Unwinding the helix creates a problem further down the line: the DNA ahead of the fork becomes overwound, forming tight coils called positive supercoils. If nothing relieves that tension, the fork stalls. An enzyme called topoisomerase solves this by cutting the DNA, allowing it to swivel and release the tension, then resealing the break. Meanwhile, another enzyme called primase lays down short RNA primers on the exposed strands. DNA polymerase, the main copying enzyme, cannot start building from scratch. It needs a small pre-existing piece to grab onto, and that is exactly what the primer provides.
Step 2: Elongation
Once primers are in place, DNA polymerase takes over and begins adding nucleotides to build new strands. It reads each template strand and matches incoming nucleotides by base-pairing rules: adenine pairs with thymine, cytosine pairs with guanine. Every DNA polymerase ever discovered works in only one direction, adding nucleotides from the 5′ end toward the 3′ end of the new strand. This one-direction rule has a major consequence for how the two strands are copied.
Because the two template strands run in opposite directions (they are antiparallel), only one strand, called the leading strand, can be copied continuously in the same direction the replication fork is moving. DNA polymerase latches on near the fork and simply races along, laying down a smooth, unbroken stretch of new DNA.
The other strand, called the lagging strand, faces the opposite direction. Polymerase can still only work 5′ to 3′, so it has to synthesize this strand in short, backward-facing segments. Each segment starts with a new RNA primer, then polymerase extends it until it bumps into the previous segment. These small pieces are called Okazaki fragments, named after the researchers who discovered them in the 1960s. In bacteria, each fragment is roughly 1,200 nucleotides long. In human cells, they are much shorter, only about 200 nucleotides, a difference driven by how DNA is packaged around proteins in eukaryotic cells.
Throughout elongation, DNA polymerase also acts as its own editor. Each time it adds a nucleotide, it checks whether the base pair fits correctly. If it detects a mismatch, it reverses, removes the wrong nucleotide, and tries again. This built-in proofreading is possible precisely because synthesis runs 5′ to 3′. The energy for adding each new nucleotide comes from the incoming nucleotide itself, which carries a cluster of phosphate groups that release energy when broken. If synthesis ran in the opposite direction, removing a mismatched nucleotide would also remove the energy source needed to continue building, making error correction impossible. Even with proofreading, polymerase makes about one mistake per 100,000 nucleotides. A separate mismatch repair system catches most of those errors after the fact, pushing the final error rate far lower.
Step 3: Termination
After elongation, the new DNA is nearly complete but not quite finished. The lagging strand is still a series of Okazaki fragments, each beginning with a short RNA primer that does not belong in the final product. Repair enzymes move along the strand, removing each RNA primer and filling the resulting gap with DNA nucleotides. Once the gaps are filled, an enzyme called DNA ligase seals the remaining nicks by joining the end of one fragment to the beginning of the next, creating a single continuous strand.
In bacteria, termination is relatively straightforward. The circular chromosome has defined termination sequences where the two replication forks, traveling in opposite directions from the single origin, eventually meet. Specialized proteins trap the forks at these sites, and the two new circular chromosomes are separated.
In human cells and other eukaryotes, termination is more complex because chromosomes are linear. The very tips of each chromosome, called telomeres, present a unique challenge known as the end-replication problem. Because DNA polymerase needs a primer to start, and because that primer is eventually removed, a small stretch of DNA at the end of the lagging strand cannot be fully copied. This means telomeres shorten slightly with every round of replication. Over many cell divisions, this progressive shortening can erode the protective telomere cap entirely. Most cells counteract this with an enzyme called telomerase, which extends the telomere by adding repetitive DNA sequences to the chromosome tip before conventional replication machinery fills in the rest.
How Bacteria and Human Cells Differ
The core three-step process is the same across life, but the scale differs dramatically. E. coli replicates its entire 4.4-million-base-pair genome from a single origin, with two forks racing in opposite directions at about 1,000 nucleotides per second. The whole job takes roughly 40 minutes.
Human cells have about 6 billion base pairs spread across 46 chromosomes, and each polymerase moves at only about 50 nucleotides per second. To compensate, the cell activates thousands of replication origins across all chromosomes at once, with each origin generating its own pair of replication forks. These individual replication units eventually merge as neighboring forks meet, and ligase stitches everything together into continuous daughter strands. The entire process takes roughly 6 to 8 hours during the S phase of the cell cycle.
Another key difference involves how origins are defined. Bacterial origins are specific, well-characterized DNA sequences that initiator proteins recognize directly. Eukaryotic origins, with the notable exception of budding yeast, are defined less by their DNA sequence and more by how accessible the DNA is. Regions that are actively transcribed or loosely packaged tend to serve as origins, giving the cell flexibility in where replication begins.
Why Accuracy Matters
Every time a cell divides, its entire genome must be faithfully duplicated. The combination of polymerase selectivity (choosing the right nucleotide), proofreading (immediately removing mismatches), and post-replication mismatch repair reduces the final error rate to roughly one mistake per billion nucleotides. For a human genome of 6 billion base pairs, that translates to only a handful of new mutations per cell division. This extraordinary accuracy is what keeps genetic information stable across trillions of cell divisions over a lifetime.