DNA replication depends on a coordinated team of enzymes that unwind the double helix, lay down new nucleotides, fix errors, and seal the finished strands. No single enzyme does the whole job. Instead, dozens of proteins work in sequence and in parallel at structures called replication forks, where the two parent strands are peeled apart and each serves as a template for a new complementary strand. Understanding which enzymes do what, and when, gives you a clear picture of how a cell copies roughly three billion base pairs of DNA in a matter of hours with remarkably few mistakes.
Getting Started at an Origin
Replication does not begin at random spots along the chromosome. Specific DNA sequences called origins of replication serve as launch pads. In eukaryotic cells, a multi-protein machine called the Origin Recognition Complex (ORC) identifies these sites. ORC is an ATP-dependent machine that recruits additional proteins to form what is known as a pre-replicative complex, setting the stage for DNA unwinding and copying once the cell enters S phase of the cell cycle.1PubMed Central. The origin recognition complex: a biochemical and structural view Bacteria, by contrast, typically have a single origin and use a different initiator protein called DnaA, but the basic principle is the same: mark the spot, recruit the machinery, then fire.
Once the origin is licensed, the real action starts with loading of the replicative helicase. In eukaryotes, this is the Mcm2-7 complex, a ring-shaped motor that encircles the DNA. Loading this complex onto DNA and then activating it are landmark regulatory events that the cell controls tightly to ensure each stretch of DNA is copied exactly once per cell division.2PubMed Central. The Mcm complex: unwinding the mechanism of a replicative helicase Two copies of the Mcm2-7 helicase are loaded head-to-head at each origin. When activated alongside a cofactor called Mcm10, these head-to-head helicases push against one another and pull on opposite strands of the double helix, generating enough force to melt the DNA open without needing any additional factors.3eLife. An explanation for origin unwinding in eukaryotes That melting creates two replication forks that travel in opposite directions away from the origin.
Unwinding the Helix and Relieving Tension
As the helicase barrels forward and splits the two strands apart, it creates a problem: the DNA ahead of the fork gets overwound and positively supercoiled, like a phone cord that twists tighter when you spin one end. If left unchecked, this tension would eventually stall the fork entirely. Topoisomerases solve this by cutting, rotating, and resealing the DNA to release that torsional stress.
Bacteria rely heavily on DNA gyrase, a type II topoisomerase that can remove positive supercoils with striking speed. In laboratory experiments, gyrase from Bacillus anthracis cleared all positive supercoils from a plasmid in under two minutes.4PubMed Central. Activities of gyrase and topoisomerase IV on positively supercoiled DNA Eukaryotes handle the same challenge with their own topoisomerases. Topoisomerase I and topoisomerase II both relax positive torsional stress ahead of the fork, though topo II does so faster and with a preference for positive over negative supercoils. This efficient relaxation of positive stress is thought to help both RNA and DNA polymerases advance smoothly.5PubMed Central. Chromatin regulates DNA torsional energy via topoisomerase II-mediated relaxation of positive supercoils
Once the two strands separate, exposed single-stranded DNA is fragile and prone to forming tangled secondary structures. Replication Protein A (RPA) in eukaryotes, or Single-Stranded DNA-Binding Protein (SSB) in bacteria, coats these exposed stretches. RPA is a three-subunit protein that binds single-stranded DNA and keeps it straight and protected until the polymerase arrives to copy it. Beyond just stabilizing the template, RPA also interacts with and influences many other proteins involved in replication and DNA repair.6PubMed. Replication protein A: a heterotrimeric, single-stranded DNA-binding protein required for eukaryotic DNA metabolism7PubMed Central. Replication protein A: single-stranded DNA’s first responder
Priming and Building New DNA Strands
Here is one of the quirks of biology that shapes the entire replication process: DNA polymerases cannot start a new strand from scratch. They can only add nucleotides onto an existing strand. To get things going, an enzyme called DNA primase synthesizes a short stretch of RNA, typically about ten nucleotides long, that serves as a starting point. DNA polymerases then extend from that RNA primer, replacing it later with DNA.8PubMed Central. Mechanism and evolution of DNA primases9PubMed. DNA primases
Because the two DNA strands run in opposite directions (antiparallel), the replication fork creates an asymmetry. One strand, the leading strand, points in the same direction the fork is moving, so it can be copied continuously as a single long stretch. The other, the lagging strand, points backward relative to fork movement. It has to be synthesized in short segments, each starting from its own RNA primer. These fragments, called Okazaki fragments, are later stitched together.10PubMed Central. High-accuracy lagging-strand DNA replication mediated by DNA polymerase dissociation
The workhorse polymerases differ between organisms. In bacteria, DNA Polymerase III handles the bulk of new strand synthesis. It does not act alone: a ring-shaped protein called the β-clamp encircles the DNA and tethers Pol III to the template. Without the clamp, the polymerase would fall off after copying just a handful of nucleotides. The clamp is loaded onto DNA by a dedicated clamp loader complex in an ATP-dependent process.11PubMed Central. The bacterial DNA sliding clamp, β-clamp: structure, interactions, dynamics and drug discovery Eukaryotes use an analogous clamp called PCNA (Proliferating Cell Nuclear Antigen) that serves the same function for their replicative polymerases, Pol δ and Pol ε.
Finishing the Lagging Strand
Once each Okazaki fragment has been extended, the cell faces a cleanup job. Each fragment still has an RNA primer at its start that needs to be replaced with DNA, and then the fragments need to be joined into a continuous strand. This maturation process requires several enzymes working in sequence.
As a new Okazaki fragment grows, it eventually runs into the RNA primer of the fragment synthesized just before it. The primer gets displaced into a short flap of single-stranded material. FEN1 (Flap Endonuclease 1) recognizes and clips off this flap. FEN1 works by threading the free end of the flap through its active site, a design feature that prevents it from accidentally cutting the wrong strand.12PubMed Central. Flap endonuclease 1 In bacteria, the process involves DNA Polymerase I, which uses its own built-in nuclease to chew away the RNA primer while simultaneously filling the gap with DNA. RNase H enzymes also help by cleaving RNA that is paired with DNA, providing an additional pathway for primer removal.13PubMed Central. RNase HIII Is Important for Okazaki Fragment Processing in Bacillus subtilis
After the RNA is removed and replaced with DNA, a nick remains in the sugar-phosphate backbone between adjacent fragments. DNA ligase seals this nick by catalyzing a new phosphodiester bond. Ligase works fast: the rate of bond formation is comparable to the speed at which replicative polymerases themselves synthesize DNA.14PubMed Central. Dynamics of phosphodiester synthesis by DNA ligase Structurally, ligases wrap around the DNA like a C-shaped clamp, positioning the nick directly over the active site for catalysis.15PubMed Central. DNA binding with a minimal scaffold: structure–function analysis of Lig E DNA ligases
Catching Mistakes Before They Become Mutations
DNA polymerases are impressively accurate, but they are not perfect. A wrong nucleotide slips in occasionally. To deal with this, the major replicative polymerases have a built-in proofreading function: a 3′-to-5′ exonuclease activity that lets them back up and remove a mismatched nucleotide immediately after incorporating it.16PubMed Central. Proofreading exonuclease activity of human DNA polymerase delta and its effects on lesion-bypass DNA synthesis When researchers disabled this exonuclease in yeast DNA polymerase δ by introducing targeted mutations, the cells showed a drastically increased rate of spontaneous mutations, confirming that proofreading is essential for accurate replication.17PubMed Central. The 3′ to 5′ exonuclease activity located in the DNA polymerase delta subunit of Saccharomyces cerevisiae is required for accurate replication
After proofreading, a second safety net exists: the mismatch repair (MMR) system. This set of proteins scans freshly replicated DNA for base-pair mismatches and small insertion-deletion loops that escaped the polymerase’s proofreading. MMR proteins recognize the error, excise a short stretch containing the mismatch, and fill the gap correctly. Together, the base-selection accuracy of the polymerase, its proofreading exonuclease, and post-replication mismatch repair bring the overall error rate down to roughly one mistake per billion base pairs copied. That is part of why your cells can divide trillions of times over a lifetime without the genome falling apart.
The End Replication Problem and Telomerase
Linear chromosomes pose a unique challenge that circular bacterial chromosomes do not face. Because the lagging strand relies on RNA primers to start each Okazaki fragment, the very tip of a chromosome cannot be fully replicated. When the final primer is removed, there is no upstream fragment to fill the gap. This means that every round of replication leaves the chromosome slightly shorter. The cell buffers against this with telomeres, repetitive DNA sequences capping each chromosome end.18PubMed Central. Telomere Replication: Solving Multiple End Replication Problems
Most telomere DNA is still copied by the standard replication machinery. But to compensate for the inevitable shortening, cells that need to keep dividing, such as stem cells and germ cells, activate an enzyme called telomerase. Telomerase is a reverse transcriptase: it carries its own RNA template and uses it to add new telomeric repeats to the chromosome ends.19PubMed. Reverse transcriptase motifs in the catalytic subunit of telomerase A protein complex called shelterin helps recruit telomerase to telomeres and regulates how many repeats get added, maintaining a balance between extension and erosion so that telomere length stays within a functional range.20PubMed Central. How long does telomerase extend telomeres? Regulation of telomerase release and telomere length homeostasis In most normal somatic cells, telomerase is turned off, which is why telomeres shorten with age and eventually contribute to cellular senescence. Cancer cells, on the other hand, frequently reactivate telomerase, giving them the ability to divide indefinitely.
How the Cell Prevents Replication From Happening Twice
Copying a genome once per cell cycle is essential. Copying part of it a second time could produce extra gene copies, trigger chromosomal rearrangements, or destabilize the genome entirely. The cell prevents this through a licensing system controlled by protein kinases tied to the cell cycle.
The pre-replicative complex that loads the helicase onto origins can only assemble during G1 phase, when cyclin-dependent kinase (Cdk) activity is low. Once the cell enters S phase, rising Cdk activity triggers the firing of licensed origins but simultaneously prevents new pre-RCs from forming.21PubMed Central. Cell cycle regulation of DNA replication In budding yeast, strains missing all B-type cyclins cannot enter S phase at all, underscoring how tightly kinase activity is linked to replication initiation.22Current Biology. DNA replication: Controlling initiation during the cell cycle Then, the kinases that drive cells into mitosis prevent any re-licensing until the cell divides and kinase levels drop again, resetting the system for the next round.23PubMed. Cell cycle control of DNA replication This two-step control, license in G1 and fire in S, is the main reason each origin fires only once per cell cycle.
When Replication Forks Stall
Replication forks do not always glide smoothly along the template. They frequently hit roadblocks: chemical damage to a base, proteins covalently stuck to the DNA, collisions with the transcription machinery, or unrelieved supercoiling that the topoisomerases have not yet resolved.24PubMed Central. Lesion Bypass and the Reactivation of Stalled Replication Forks A stalled fork is dangerous because the exposed single-stranded DNA can break or recombine inappropriately.
Cells have evolved an elaborate toolkit for dealing with stalled forks. One strategy is fork reversal: the fork backs up and reanneals the two newly synthesized strands with each other, forming a four-way junction that protects the fork from collapse. Specialized motor proteins catalyze this reversal, and other proteins guard the reversed fork against being chewed up by nucleases.25PubMed Central. Making Choices: DNA Replication Fork Recovery Mechanisms Another strategy is damage bypass: the fork skips past the lesion entirely and restarts downstream, leaving a gap that gets filled later. Or, if the damage is on the template strand, specialized translesion synthesis polymerases can be swapped in temporarily. These polymerases are sloppier than the replicative ones, but they can copy across damaged bases where the main polymerase would stall. The trade-off is a higher chance of introducing a mutation at that specific spot.26PubMed Central. Mechanisms for stalled replication fork stabilization: new targets for synthetic lethality strategies in cancer treatments Under normal conditions, though, cells manage to complete replication without catastrophic fork failures, thanks to the redundancy of these rescue pathways.
Bacteria Versus Eukaryotes
The core logic of DNA replication, unwind, prime, polymerize, proofread, seal, is shared across all domains of life. But the specifics diverge considerably between bacteria and eukaryotes, and these differences have real consequences, particularly for medicine.
Bacterial chromosomes are usually circular and have a single origin of replication. One pair of replication forks can copy the entire genome in about 40 minutes in a fast-growing bacterium. Eukaryotic genomes are vastly larger, spread across multiple linear chromosomes, and use many origins, tens of thousands in human cells, to get the job done within a reasonable S phase. The challenge of coordinating all those origins so that none fires twice is a regulatory problem bacteria simply do not face at the same scale.27PubMed Central. Principles and concepts of DNA replication in bacteria, archaea, and eukarya
The enzymes also differ in identity even when their functions overlap. Bacteria use DNA gyrase for supercoil management; eukaryotes do not have gyrase and rely on topo I and topo II instead. Bacterial replication uses DNA Pol III as its main polymerase, with a β-clamp for processivity; eukaryotes use Pol δ and Pol ε with PCNA. These differences create exploitable targets for antibiotics that hit bacterial enzymes without touching human ones.
Replication Inside Mitochondria
Your nuclear chromosomes are not the only DNA your cells replicate. Mitochondria carry their own small, circular genome, and they copy it using a distinct set of enzymes that are more similar to those of bacteriophages than to the nuclear replication machinery.28PubMed. Replication and Transcription of Human Mitochondrial DNA The main polymerase is DNA polymerase gamma (Pol γ), which handles both synthesis and proofreading for mitochondrial DNA. A dedicated mitochondrial helicase called Twinkle unwinds the template ahead of Pol γ, and a mitochondrial single-stranded DNA-binding protein stabilizes the exposed strands.
Each mitochondrion carries multiple copies of its genome, and these are packaged into protein-DNA structures called nucleoids by mitochondrial transcription factor A (TFAM). The degree to which TFAM compacts the DNA influences whether replication and transcription can initiate.29PubMed Central. Mitochondrial DNA replication in mammalian cells: overview of the pathway Mutations in Pol γ or Twinkle are linked to a range of mitochondrial diseases, including progressive external ophthalmoplegia and mitochondrial depletion syndromes. Because mitochondrial DNA is maternally inherited and present in hundreds of copies per cell, the genetics of these conditions can be complex: a mixture of normal and mutant genomes can coexist, and disease symptoms often appear only once the proportion of mutant copies crosses a certain threshold.
Drugs That Target Replication Enzymes
Many of the enzymes described above are drug targets in clinical medicine. The most direct example in infectious disease involves topoisomerases. Fluoroquinolone antibiotics, one of the most widely prescribed antibiotic classes, work by trapping bacterial DNA gyrase and topoisomerase IV in complexes with cut DNA. Instead of allowing the enzymes to pass a strand through and reseal the break, the drug stabilizes the broken intermediate. The result is double-strand breaks that kill the bacterium.30PubMed Central. Topoisomerase Inhibitors: Fluoroquinolone Mechanisms of Action and Resistance This mechanism is possible because fluoroquinolones target bacterial-specific enzymes, leaving human topoisomerases largely unaffected.
In cancer therapy, the situation is reversed. Because cancer cells divide rapidly and rely heavily on DNA replication, drugs that interfere with eukaryotic topoisomerase II are effective chemotherapeutics. Agents like etoposide and doxorubicin stabilize double-strand breaks created by human topo II, selectively damaging cells that are replicating their DNA most aggressively.31PubMed. Topoisomerase II inhibitors design: Early studies and new perspectives Other cancer strategies target the replication stress response: tumors with defective fork-stabilization pathways can be pushed over the edge by drugs that increase replication stress, a concept behind several synthetic lethality approaches currently in development.26PubMed Central. Mechanisms for stalled replication fork stabilization: new targets for synthetic lethality strategies in cancer treatments
Antiviral drugs also exploit replication enzymes, though their targets are viral polymerases rather than human ones. Nucleoside analogs like acyclovir (for herpes viruses) and remdesivir (developed for RNA viruses) mimic natural nucleotides and get incorporated by viral polymerases, then stall or terminate the growing strand. The selectivity comes from the viral polymerase being much more willing to incorporate the decoy than the human polymerase is.
Why Redundancy Matters
One theme running through all of this is how many backup systems exist. There are multiple pathways for removing RNA primers. Proofreading catches errors during synthesis, and mismatch repair catches whatever proofreading misses. Stalled forks can be reversed, bypassed, or restarted through recombination. Dormant origins that normally do not fire can activate if nearby forks fail. This layered redundancy is not an accident. Faithful genome duplication is so critical that evolution has stacked safeguard on top of safeguard, making complete replication failure a genuinely rare event under normal conditions even though individual forks encounter problems constantly.26PubMed Central. Mechanisms for stalled replication fork stabilization: new targets for synthetic lethality strategies in cancer treatments When those safeguards break down, the consequences are familiar: genome instability, accelerated aging, and cancer.