Polymerase Chain Reaction

The trick: a copy machine for one DNA sequence

Suppose you have a single drop of blood at a crime scene, a few cells from a tumor biopsy, or one virus particle in a nasal swab. Hidden inside is a specific DNA sequence you care about — but there is far too little of it to read, cut, or detect. The polymerase chain reaction solves this with brute-force exponential copying. You tell it which sequence to copy using two short primers, then let it run for an hour. One target molecule becomes two, then four, then eight, and after thirty rounds you are holding roughly a billion identical copies — enough to load on a gel, sequence, or clone.

The elegance is that PCR copies only the region you specify. Out of the three billion base pairs in a human genome, you can pluck and amplify a single 500-base-pair gene, leaving everything else untouched. That selectivity comes entirely from primer design, and the speed comes from a polymerase that survives near-boiling water.

The three-temperature cycle

A PCR reaction is a small tube — typically 10–50 microliters — containing the template DNA, the two primers, a thermostable DNA polymerase, the four deoxynucleotide building blocks (dNTPs: dATP, dTTP, dGTP, dCTP), magnesium ions (Mg²⁺, an essential cofactor for the polymerase), and a buffer. A machine called a thermal cycler drives the tube through a repeating temperature program. Each cycle has three phases.

1 · Denature (~95°C)

Heating to about 95°C breaks the hydrogen bonds between complementary bases — two per A–T pair, three per G–C pair — so the double helix unzips into two single strands. (This is why GC-rich templates are harder to melt and sometimes need a hotter or longer denaturation.) Ordinary enzymes would be cooked and destroyed at this temperature; PCR's secret is a polymerase that is not.

2 · Anneal (~50–65°C)

Cooling to roughly 50–65°C lets the two primers — short single-stranded DNA pieces, usually 18–25 bases — find and base-pair to their complementary sites on the now-separated strands. The forward primer binds one strand, the reverse primer the other, and they point toward each other, bracketing the target. The annealing temperature is tuned to the primers' melting temperature (Tm): too cold and primers stick to near-matches (non-specific products); too hot and they fall off entirely.

3 · Extend (~72°C)

Warming to ~72°C — the optimum for Taq — lets the polymerase latch onto each primer's free 3′ end and add nucleotides one at a time, reading the template 3′→5′ while synthesizing the new strand 5′→3′. Taq extends at roughly 1,000 nucleotides per second, so a 1-kilobase product is finished in about a second of pure synthesis (extension times are usually set to ~30–60 s for margin). When the cycle's heat returns, the new strands separate from their templates and the whole thing repeats.

Why the numbers explode

The pivotal idea is that every product strand becomes a template in the next cycle. Starting from one double-stranded target:

• After cycle 1: 2 copies
• After cycle 2: 4 copies
• After cycle 3: 8 copies
• After cycle n: 2ⁿ copies

So 20 cycles give 2²⁰ ≈ 1 million, 30 cycles give 2³⁰ ≈ 1.07 billion, and 40 cycles give over a trillion — in theory. Real reactions are not perfectly efficient, so the practical formula is N × (1 + E)ⁿ, where E is the per-cycle efficiency (often 0.9–1.0 early on). The reaction also cannot grow forever: once primers and dNTPs deplete and product strands start re-annealing to each other faster than primers can bind, the curve flattens into a plateau. This S-shaped (exponential then plateau) growth is exactly what real-time PCR exploits to measure how much template you started with.

There is a subtlety worth knowing: in cycle 1 the new strands run off the end of the template (they have no defined stop), but the primers fix the start. By cycle 3, products appear that are bounded by primers at both ends — the exact-length "amplicon." From then on, these fixed-length products dominate and accumulate exponentially, which is why the final band on a gel is a sharp size, not a smear.

Taq: the enzyme from a hot spring

The earliest PCR experiments used the polymerase from E. coli, which is destroyed at 95°C — so fresh enzyme had to be pipetted in after every single denaturation step, making the method tedious and expensive. The breakthrough was Taq polymerase, isolated from Thermus aquaticus, a bacterium that thrives in ~70°C Yellowstone hot springs. Taq survives repeated 95°C cycling, so it is added just once at the start. This single substitution turned PCR from a laboratory curiosity into an automatable, world-changing technique.

Taq is fast (~1,000 nt/s at 72°C) and robust, but it lacks a 3′→5′ proofreading exonuclease, so it cannot correct misincorporated bases. Its error rate is roughly 1 mistake per 10⁴–10⁵ bases — fine for diagnostics and detection, but risky when the exact sequence matters (e.g., cloning a gene for expression). For high-fidelity work, proofreading enzymes such as Pfu (from Pyrococcus furiosus, ~1 error per 10⁶ bases) or engineered blends are used instead. Taq also has a quirk that became a feature: it adds a single untemplated adenosine to the 3′ end of products, enabling convenient "TA cloning."

Polymerases compared

Enzyme	Source	Speed	Proofreading	Error rate (per base)	Typical use
Taq	Thermus aquaticus	~1,000 nt/s	No	~1 × 10⁻⁴–10⁻⁵	Diagnostics, routine PCR, colony screening
Pfu	Pyrococcus furiosus	~500 nt/s	Yes (3′→5′)	~1 × 10⁻⁶	Cloning, mutagenesis, high-fidelity work
Phusion (engineered)	Fusion + Pyrococcus-like	Very fast	Yes	~5 × 10⁻⁷	Long, accurate amplicons
Reverse transcriptase	Retroviruses (e.g. MMLV)	Slow	No	High	RNA → cDNA before PCR (RT-PCR)

PCR's family tree

The basic reaction has spawned dozens of variants. Three matter most:

• qPCR (quantitative / real-time PCR). A fluorescent reporter — an intercalating dye like SYBR Green, or a sequence-specific TaqMan probe — brightens as product accumulates. The machine reads fluorescence every cycle. The cycle at which signal crosses a threshold (the Ct or Cq value) is proportional to the log of the starting amount, so qPCR quantifies DNA, not just detects it. Each whole Ct unit corresponds to roughly a doubling, i.e. a 2-fold difference in starting template.
• RT-PCR (reverse-transcription PCR). RNA cannot be a PCR template directly, so reverse transcriptase first copies it into complementary DNA (cDNA). This is how we measure gene expression and detect RNA viruses. The familiar "PCR test" for COVID-19 is RT-qPCR: reverse transcription, then real-time quantification.
• Digital PCR. The sample is split into thousands of tiny partitions, each with zero or one template molecule, and each is amplified independently. Counting the positive partitions gives an absolute molecule count without a standard curve — powerful for rare-mutation detection and liquid biopsy.

Where PCR changed the world

PCR is arguably the single most important technique in molecular biology, and its applications are everywhere:

• Diagnostics. Detecting pathogens — SARS-CoV-2, HIV, tuberculosis, HPV — from minute samples, often before symptoms or culture would reveal them.
• Forensics. A few cells of touch DNA can be amplified at standardized short-tandem-repeat (STR) loci to generate a profile matching one person in billions.
• Sequencing and cloning. PCR enriches a target before sequencing or inserting it into a plasmid; it is the indispensable front end of recombinant DNA work.
• Ancient and trace DNA. PCR can amplify degraded DNA from museum specimens, fossils, and Neanderthal bone, reconstructing genomes from fragments.
• Genotyping and research. Detecting mutations, confirming gene edits, measuring expression, and screening transgenic organisms.

Why PCR is also dangerous to trust blindly

The very power that makes PCR useful — billion-fold amplification of any matching sequence — makes it exquisitely vulnerable to contamination. A single stray molecule from a previous reaction, an aerosol, or a neighboring sample can be amplified into a convincing false positive. The worst offender is "carryover": amplified product from an earlier run is already enriched a billion-fold, so even a microscopic droplet swamps a clean reaction. Standard defenses include a no-template control (water instead of DNA, to catch contamination), filtered pipette tips, UTP/uracil-N-glycosylase decontamination systems, and physically separating the room where reactions are set up from where products are analyzed.

Specificity failures are the other risk: if the annealing temperature is too low or primers match similar sequences, you amplify the wrong thing, or primers bind each other to form primer-dimers. Good primer design (matched Tm, ~40–60% GC, no self-complementarity) and a melt-curve or gel size check guard against this.