DNA is replicated during S phase, the synthesis phase of interphase. In a typical human somatic cell, S phase occupies roughly 6–8 hours of a 24-hour cycle and copies about 3.2 billion base pairs from each of 46 chromosomes. The copying is bidirectional, semiconservative, and begins at thousands of licensed origins that fire in a programmed order. Nothing about that order is random, and that is the point: S phase is not a photocopier running start to finish, but a scheduled construction project with staged permits.
The phase name matters less than the machinery. Licensing, origin firing, fork pacing, and checkpoint control all determine whether the copied genome is complete before mitosis begins. A cell that finishes S phase on time but with unreplicated gaps is not ready to divide, no matter what the clock says. The sections below work through the mechanism, the numbers, the exceptions, and the limits of what current evidence supports.
Mechanism: licensing, firing, and fork progression
Replication begins before S phase, in late mitosis and G1, with origin licensing. The origin recognition complex (ORC) binds replication origins, then loads Cdc6 and Cdt1, which together place two inactive Mcm2-7 hexamers around DNA. This produces a double-hexamer that encircles the duplex without opening it. Licensing is restricted to G1 by cyclin-dependent kinase (CDK) activity; once S phase begins, high CDK and Dbf4-dependent kinase (DDK) activity block re-licensing. That block is the core reason each origin fires once per cycle.
Firing requires DDK to phosphorylate Mcm4 and Mcm6, then CDK to phosphorylate Sld2 and Sld3 (in budding yeast; TOPBP1 and Treslin in humans). These events recruit Cdc45 and the GINS complex, forming the active CMG helicase. The helicase unwinds DNA, and three polymerases take over: Pol epsilon on the leading strand, Pol delta on the lagging strand. The lagging strand is synthesized as discontinuous Okazaki fragments, each roughly 150–200 nucleotides in human cells, later matured by RNase H, FEN1, and DNA ligase I.
Two replication forks move outward from each origin in opposite directions. In human cells, fork progression runs at roughly 1–2 kb per minute. At that pace, copying 3.2 billion base pairs from a single origin would take more than a century. The genome solves this by using tens of thousands of origins, firing in clusters across S phase. Late-firing origins tend to sit in heterochromatin and near telomeres; early-firing origins cluster in euchromatin and near highly transcribed genes. This temporal program is inherited with some flexibility, which is one reason replication timing is a usable cell-state marker.
Plainly: S phase is a distributed copying job with thousands of start points, strict one-time permits, and two polymerases doing different jobs on each fork.
Quantified context: how big, how fast, how long
Human cells replicate about 6.4 billion base pairs per division cycle when both copies of each chromosome are counted. In a diploid cell in G1, the genome is roughly 6 picograms of DNA; after S phase it is about 12 picograms, which is why flow cytometry can separate G1 and G2 populations by fluorescence intensity alone.
The origin count is large. Estimates for human cells range from roughly 30,000 to 100,000 licensed origins, with tens of thousands firing per S phase; the exact number depends on cell type and method. Each origin fires once, and each fired origin produces two forks. This redundancy creates a safety margin: unused dormant origins can fire if nearby forks stall, which is one reason low-level replication stress does not automatically halt the cycle.
Timeline of a typical 24-hour human cell cycle (approximate)
| Phase | Typical duration | Main event |
|---|---|---|
| G1 | 10–12 hours | Licensing of origins; growth; checkpoint at restriction point |
| G1/S transition | Minutes | CDK activation; licensing shutoff |
| S (early) | 2–3 hours | Euchromatic, gene-dense origins fire |
| S (mid) | 2–3 hours | Bulk replication; fork density peaks |
| S (late) | 1–2 hours | Heterochromatin, late domains, telomere-proximal regions |
| G2 | 3–4 hours | Repair window; decatenation; checkpoint before mitosis |
| M | 1 hour | Chromosome segregation and division |
Durations are culture-based approximations and vary by cell type, developmental state, and temperature. Embryonic stem cells can compress the entire cycle to under 12 hours, with S phase as short as 2–3 hours; some early embryonic divisions in other organisms are faster still.
Edge cases: where S phase breaks the textbook pattern
Not every cell copies DNA on a 24-hour schedule, and some copy more than once.
Endoreplication: megakaryocytes and many placental trophoblast cells enter repeated S phases without mitosis, producing polyploid nuclei. This is programmed, not pathological. Gene amplification: some cancer cells re-fire selected origins, producing extra copies of specific loci; methotrexate resistance via DHFR amplification is the classic example. Replication stress: oncogene activation, nucleotide depletion, or polymerase-blocking lesions slow forks and can cause fork reversal. Stalled forks may restart, but persistent stalling leads to double-strand breaks. Re-replication: loss of licensing control, for example through Cdt1 overexpression or geminin loss, allows origins to fire twice, producing local copy-number gains. Dormant origin firing: when a fork stalls, nearby licensed origins can fire to rescue the gap, a normal backup system that becomes essential under stress. Mitochondrial DNA: mtDNA replicates independently of S phase, by a displacement-loop mechanism, so "DNA replication" in a cell is not confined to nuclear S phase.
The counterintuitive caveat: S phase length and genome size are weakly coupled across species. Some amphibians with genomes 10–20 times larger than humans replicate them in comparable or longer S phases using larger replicons and slower forks.
Misconceptions that survive textbooks
Several errors recur in thin explainers and in professional reviews.
First, DNA is not replicated "during mitosis." Mitosis separates already-duplicated chromosomes; the copying is complete well before prophase in normal somatic cells. Second, S phase does not begin at a single clock time for every origin. Origin firing is staggered across hours, and the same origin can fire at slightly different times in different cells. Third, "once per cycle" is a rule enforced by licensing, not an intrinsic property of DNA; when licensing fails, re-replication happens. Fourth, semiconservative replication does not mean both strands are copied symmetrically: leading and lagging strand synthesis differ in mechanism, error rate, and Okazaki fragment processing. Fifth, S phase is not repair-free. Mismatch repair and base excision repair operate during and after replication, and the G2 window exists partly to finish repair before mitosis.
A professional caveat worth stating: replication timing data from cell populations average over cells, so a "late-firing domain" is a statistical property, not a fixed time stamp in any single cell.
Practical actions: measuring and interpreting S phase
These are bench and analysis steps, not clinical instructions.
Measure S phase fraction by labeling with a thymidine analog (BrdU or EdU) for 30–60 minutes, then detecting incorporation by flow cytometry or microscopy. Typical proliferating culture cells show S fractions of roughly 20–40 percent; quiescent cells show near zero. Distinguish G1, S, and G2 by DNA content: 2n (G1), intermediate (S), 4n (G2/M). Combine with a mitotic marker such as phospho-histone H3 to separate G2 from M. Use dual-pulse labeling (for example, sequential IdU and CldU pulses) to measure fork rate and origin firing density in single molecules. Detect replication stress with markers: phospho-RPA32, phospho-ATM/ATR substrates, and gamma-H2AX. Treat elevated gamma-H2AX as a red flag for fork collapse, not proof of it. Report cell-cycle stage with the method used. Flow DNA content, thymidine incorporation, and single-molecule fiber assays answer different questions and are not interchangeable. Red-flag warning: an apparent S-phase fraction can be inflated by DNA repair synthesis after damage, by apoptosis-related DNA fragmentation, and by polyploid cells in G1-like DNA content.
One reconciliation point: doubling times, S-phase durations, and S-phase fractions are separate quantities. A 24-hour doubling time does not mean a 24-hour cell cycle, and an S-phase fraction of 30 percent does not imply a 3-hour S phase unless the cycle length is known.
Limitations: what current evidence does not settle
The core model, licensing followed by CDK- and DDK-dependent firing once per cycle, is well supported across eukaryotes. Several details remain unsettled or method-dependent. Human origin counts and firing efficiencies vary widely by study and technique. The rules that set replication timing are partly understood but not fully predictive from sequence alone. Fork rate estimates depend on labeling method and cell type, with published values spanning roughly 0.5–3 kb per minute in human cells. Whether specific dormant origins are selected deterministically or stochastically under stress is still debated.
Current evidence supports the mechanism and the approximate numbers above, but not precise per-cell timings or universal origin counts. Any claim that a specific cell type replicates a specific locus at a specific minute is a population average, not a single-cell fact.
For laboratory or clinical decisions about cell-cycle analysis, proliferation markers, or replication stress in patient samples, consult a qualified local pathologist or laboratory director. This article is general educational information about cell biology, not a diagnostic or treatment protocol.