Sanger Dideoxy Sequencing
How do you read the exact order of four nearly identical chemical letters? These letters are strung together millions of times, along a molecule too thin to see. This is the problem of DNA sequencing. Its first great practical solution came from Frederick Sanger in 1977. His method, chain-termination sequencing, turned a fundamental limitation of DNA Polymerase into a tool. By starving a replicating strand of a single nucleotide's 3′-hydroxyl group, he could force synthesis to stop. This stop occurred at a defined, sequence-specific position, over and over, at every possible site. The result — a ladder of fragments differing by exactly one base — could be sorted by size. It could then be read off as a sequence. Sanger sequencing mattered because it converted DNA from an abstract carrier of information. It became a text that could be transcribed, compared, and archived. It made possible the first complete viral genome (bacteriophage φX174, 1977). It also enabled the reading of disease-causing mutations. Eventually, it led to the Human Genome Project. That project's reference sequence was assembled almost entirely with automated, fluorescent versions of Sanger's chemistry. For three decades it was simply "sequencing." It was the default technology every molecular biologist learned first. The big picture is that Sanger sequencing exploits a simple structural trick. It replaces the polymerase's normal substrate, a 2′-deoxynucleoside triphosphate (dNTP). In its place goes a dideoxynucleoside triphosphate (ddNTP) lacking the 3′-OH needed to add the next base. Each incorporation event is therefore a full stop. Doing this in four parallel reactions generates a nested set of fragments. In modern practice, one combined four-color reaction is used instead. These fragments' lengths encode the identity of every base in the template. This article works through the chemistry of chain termination. It covers the electrophoretic readout that turned fragment lengths into a sequence. It examines the classic 1977 experiments that established and validated the method. Two worked examples are included: reading an autoradiogram-style gel, and interpreting a chromatogram trace with a heterozygous peak. It also addresses the persistent misconceptions students bring to this topic. Finally, it considers the method's continuing relevance, even in an era dominated by next-generation sequencing.
- Sanger sequencing reads DNA by controlled, sequence-specific chain termination during in vitro DNA synthesis.
- Dideoxynucleotides (ddNTPs) lack a 3′-OH, so once one is incorporated, the polymerase cannot extend the strand further.
- A population of primed templates generates a nested set of fragments, one ending at every occurrence of each base.
- Fragments are separated by size using polyacrylamide or capillary gel electrophoresis, with single-nucleotide resolution.
- Modern "automated" Sanger sequencing uses four different fluorescent dyes (one per ddNTP) read by a laser in a single capillary lane.
- Sanger sequencing remains the gold-standard confirmatory method for single genes, clinical variants, and Sanger validation of NGS calls.
Chain-termination sequencing works because a dideoxynucleotide, once incorporated, chemically prevents the next phosphodiester bond from forming — turning random termination events into a precise, size-encoded readout of every base position in a DNA template.
The core idea
Sanger dideoxy sequencing (also called chain-termination sequencing) determines the base sequence of a DNA template by controlled interruption of DNA synthesis. A short primer anneals to a single-stranded (or denatured double-stranded) template adjacent to the region of interest, and DNA polymerase extends it using the four normal deoxynucleoside triphosphates (dATP, dCTP, dGTP, dTTP) — but the reaction also contains a small proportion of one dideoxynucleoside triphosphate (ddATP, ddCTP, ddGTP, or ddTTP). A dideoxynucleotide is structurally identical to a normal nucleotide except that its ribose sugar has a hydrogen atom at the 3′ carbon instead of a hydroxyl group. Because DNA Polymerase builds a new strand by forming a phosphodiester bond between the 3′-OH of the growing strand and the 5′-phosphate of the incoming nucleotide, any incorporation of a ddNTP is a dead end — synthesis is permanently terminated at that position on that particular template molecule. Because millions of copies of the primed template are present in the reaction, and each polymerase molecule "chooses" stochastically whether to incorporate a normal dNTP or the rarer ddNTP at each opportunity, the reaction produces a huge population of truncated fragments of every possible length, each one ending precisely at a position where that particular base occurs in the template. Running four separate reactions (one per ddNTP) and separating the products by high-resolution gel electrophoresis, then reading the four lanes side by side from smallest to largest fragment, reconstructs the complete sequence of the newly synthesized strand — which is complementary to the template. The elegance of the method lies in converting a chemical event (bond termination) into a physical, measurable property (fragment length), and then using electrophoresis — already familiar from restriction mapping — as a single-base-resolution ruler. Modern automated Sanger sequencing replaces radioactive labeling and four gel lanes with four different fluorescent dyes attached to the four ddNTPs, run together in a single capillary lane, and read by laser-induced fluorescence as fragments pass a detector in order of size.
Mechanism
Step 1 — Template preparation and priming. The DNA to be sequenced is usually amplified first, often via PCR, and then denatured to single strands (or cloned into a single-stranded phage vector, as in Sanger's original work with M13). A short synthetic oligonucleotide primer, complementary to a known sequence just upstream of the region to be read, is annealed to the template. This primer supplies the free 3′-OH that DNA polymerase requires to begin synthesis — polymerases cannot initiate a new strand de novo.
Step 2 — The four (or one) reaction mixtures. Each reaction tube contains:
- Single-stranded template–primer complex
- DNA polymerase (originally the Klenow fragment of E. coli DNA Pol I; later a modified, highly processive Taq polymerase for automated cycle sequencing)
- All four normal dNTPs, one of them typically radiolabeled or fluorescently tagged (in classic 4-lane sequencing)
- A limited amount of one ddNTP (in classic sequencing, one type per tube; in modern automated sequencing, all four ddNTPs, each carrying a distinct fluorescent dye, together in one tube)
Step 3 — Chain elongation with stochastic termination. Polymerase extends the primer, adding dNTPs one at a time. At every position where the next correct base could be incorporated, there is a small, fixed probability that the polymerase instead incorporates the corresponding ddNTP (its concentration is kept low relative to the dNTP, typically ratios optimized so that termination events are spread across the whole readable range, often several hundred to ~1000 bases). Once a ddNTP is added, the 3′ carbon has no hydroxyl group, so no further phosphodiester bond can form — that particular polymerase molecule falls off, and that fragment is permanently finished.
Step 4 — Generating a nested fragment ladder. Because there are millions of template molecules in the reaction, and termination at any given base position is a probabilistic event, the reaction as a whole produces a full spectrum of fragment lengths — one ending at each occurrence of the relevant base along the template. In the four-tube version, the ddATP tube produces fragments ending at every A, the ddCTP tube at every C, and so on.
Step 5 — Size separation by electrophoresis. The fragments (which can differ in length by as little as one nucleotide) are separated by electrophoresis through a matrix fine enough to resolve single-base differences — historically a denaturing polyacrylamide slab gel, now a viscous polymer inside a very thin glass capillary. Because DNA is uniformly negatively charged along its phosphate backbone, its electrophoretic mobility in these sieving matrices depends almost purely on length: shorter fragments migrate faster.
Step 6 — Detection and base-calling. In classic radioactive/manual sequencing, the four lanes are exposed to X-ray film, and the sequence is read by eye from the bottom (smallest fragment, closest to the primer) to the top (largest fragment, farthest into the template). In automated (dye-terminator) sequencing, all four ddNTPs carry spectrally distinct fluorescent dyes; a single capillary run resolves fragments by size, and a laser excites each fragment as it passes a fixed detection window near the end of the capillary. A computer records the fluorescence color at each detected fragment and outputs a four-color chromatogram (an electropherogram), one peak per base, which base-calling software translates into a sequence of A, C, G, T letters.
Step 7 — Reading length and error profile. A single Sanger read typically yields high-quality, low-error sequence for roughly 700–1000 bases downstream of the primer. Quality degrades at both ends: very short fragments run together near the primer, and very long fragments lose resolution as they approach each other in size, producing declining signal quality and rising error rates toward the far end of the read.
An additional refinement worth noting: because ddNTP incorporation is intentionally rare relative to dNTP incorporation (concentration ratios are tuned, historically often on the order of 1 ddNTP per 100 dNTPs, adjusted empirically per base), most polymerization events use a normal dNTP, and termination events are distributed roughly evenly across the readable length — this is what produces a full ladder rather than a pile-up of very short fragments.
The evidence
Sanger and Coulson's "plus-and-minus" method (1975) and dideoxy chain-termination method (1977). Frederick Sanger, working at the MRC Laboratory of Molecular Biology in Cambridge, first devised a "plus and minus" sequencing method using DNA polymerase and selective nucleotide omission, published with Alan Coulson in 1975. This was a proof of principle that controlled, incomplete polymerase reactions could be read out by gel electrophoresis to reveal sequence. It was informative but technically limited and error-prone. The decisive breakthrough came in 1977, when Sanger, Steven Nicklen, and Alan Coulson published "DNA sequencing with chain-terminating inhibitors" in Proceedings of the National Academy of Sciences. They showed that 2′,3′-dideoxynucleoside triphosphates, when included at low relative concentration alongside normal dNTPs and a radiolabeled tracer dNTP, acted as specific chain terminators during DNA polymerase-catalyzed synthesis from a primed single-stranded template. Four parallel reactions, each containing a different ddNTP, produced four sets of radiolabeled fragments. Electrophoresis on high-resolution denaturing polyacrylamide gels, followed by autoradiography, resolved fragments differing by a single nucleotide, and the four lanes read together gave the complete sequence.
Validation: the complete genome of bacteriophage φX174. The most powerful demonstration of the method's reliability was not a small test sequence but an entire genome. In 1977, Sanger's group used the chain-termination method (together with the earlier plus-minus method for cross-checking) to determine the complete 5,386-nucleotide sequence of bacteriophage φX174 — the first DNA genome ever fully sequenced. This was a rigorous test: the resulting sequence had to be internally consistent, had to correctly predict the previously known amino acid sequences of φX174 coat and other proteins from its open reading frames, and — most strikingly — revealed that some φX174 genes overlap in different reading frames, a genuinely novel and unexpected biological finding that could only have been discovered by directly reading the DNA sequence itself, not inferred any other way. That the method produced not just plausible-looking data but new, independently verifiable biology was decisive evidence that chain-termination sequencing was accurate and general.
Independent contemporaneous method: Maxam-Gilbert chemical cleavage. In the same year, Allan Maxam and Walter Gilbert published a chemically distinct sequencing method based on base-specific chemical cleavage of end-labeled DNA fragments, rather than enzymatic synthesis. The fact that two independently developed methods — one enzymatic (Sanger), one purely chemical (Maxam-Gilbert) — converged on the same sequences for shared test templates was strong cross-validation that both were correctly reading the underlying chemical sequence of DNA, not an artifact of either particular technique. Sanger's method eventually displaced Maxam-Gilbert sequencing almost entirely, mainly because it required less hazardous chemistry, was easier to automate, and scaled far better as radioactive- and later fluorescent-labeling and capillary electrophoresis were introduced.
Automation and the dye-terminator refinement (1980s). Leroy Hood's laboratory, together with Applied Biosystems, developed fluorescently labeled dideoxynucleotides (each ddNTP tagged with a spectrally distinct dye) in the mid-1980s, enabling all four termination reactions to be combined in a single tube and run in a single electrophoresis lane, detected by an automated laser-scanning instrument. This automated four-color Sanger method became the workhorse of large-scale genome sequencing, including the publicly funded Human Genome Project, whose finished reference sequence (announced substantially complete in 2003) relied overwhelmingly on capillary-based automated Sanger sequencing performed at industrial scale across many centers.
Worked examples
Example 1 — Reading a four-lane autoradiogram by hand. Suppose a student runs four dideoxy sequencing reactions on a template and obtains the following band pattern on a denaturing polyacrylamide gel (bands listed from the bottom of the gel, i.e., smallest fragment/earliest position, to the top, i.e., largest fragment/latest position). Lane order left to right is ddA, ddC, ddG, ddT. Bottom-to-top band lanes: G, T, T, A, C, G, A, A, T, C, G Because each band represents a fragment terminating with a particular ddNTP, and the smallest fragment corresponds to the first (5′-most) synthesized base after the primer, reading bottom to top directly gives the synthesized strand: 5′-G T T A C G A A T C G-3′ (newly synthesized strand, complementary to the template) To get the template strand's sequence in the same 5′→3′ orientation, take the reverse complement: Newly synthesized: 5′-GTTACGAATCG-3′ Template (reverse complement): 5′-CGATTCGTAAC-3′ This distinction — synthesized-strand sequence vs. template-strand sequence — is a classic point of confusion; the sequence you literally read off the gel (or chromatogram) is always the strand that was synthesized by the polymerase, complementary and antiparallel to the template that was actually sequenced.
Example 2 — Interpreting a chromatogram peak: detecting a heterozygous SNP by Sanger sequencing. A clinical genetics lab performs Sanger sequencing (dye-terminator, capillary electrophoresis) on PCR-amplified genomic DNA from a patient, covering a coding exon of a gene of interest, to check for a suspected pathogenic single-nucleotide variant. At one position in the chromatogram, instead of a single clean peak, the software calls two overlapping peaks of roughly equal height — a green peak (A) and a black peak (G) superimposed at the same base position, flagged automatically as a double peak, "R" in IUPAC ambiguity code (A or G). Reasoning: genomic DNA from a diploid human is a template mixture, in roughly a 1:1 ratio, of the two homologous chromosome copies (maternal and paternal alleles) after PCR amplification pools both alleles together into one sequencing reaction. If the patient is heterozygous at that position — one allele carrying the reference base (say G) and the other carrying a variant base (A) — then during the sequencing reaction, roughly half of all template molecules present are the reference allele and half are the variant allele. Chain-termination reactions on each template subpopulation independently produce fragments ending in G (from the reference-allele templates) or A (from the variant-allele templates) at that same position, migrating to the same size class, and so both dye colors are detected simultaneously in the capillary at that instant, superimposed as a single doubled peak of roughly half-height each relative to a homozygous peak. This is exactly how Sanger sequencing detects heterozygosity: a clean single-color peak indicates homozygosity at that position (both alleles the same), while a double peak of two colors at equal or near-equal height indicates heterozygosity. (If the two peaks were very unequal in height rather than roughly equal, that would instead suggest something like somatic mosaicism, unequal allele amplification bias, or contamination — a useful diagnostic subtlety, since true constitutional heterozygosity should give roughly 1:1 peak heights.)
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "Sanger sequencing reads the template strand directly." | It reads the newly synthesized strand — complementary and antiparallel to the template — so the raw output must be reverse-complemented to get the template sequence. |
| "ddNTPs stop the reaction entirely, all at once." | Each ddNTP terminates only the one strand into which it happens to be incorporated; other template molecules in the same tube keep extending until they, in turn, incorporate a ddNTP at a different position, producing the whole ladder. |
| "You need four separate tubes to sequence anything." | Only the original manual radioactive method required four tubes; modern automated dye-terminator sequencing runs all four ddNTPs together in one tube/one capillary lane, distinguished by dye color, not by physical separation. |
| "A double peak in a chromatogram always means a sequencing error." | A double peak at one position, especially near 50:50 height, is the expected and correct signature of a true heterozygous variant in diploid genomic DNA — it is data, not noise. |
| "Longer reads are always better/more accurate." | Read quality declines toward both ends of a Sanger read (poor resolution very close to the primer, and declining signal/resolution far from it), so the highest-confidence bases are typically in the middle few hundred bases, not the extremes. |
The scientist behind it
Frederick Sanger (1918–2013) is the only person to have won two Nobel Prizes in Chemistry for two entirely separate biochemical achievements 22 years apart. His first Nobel (1958) recognized his determination of the complete amino acid sequence of insulin — the first protein ever fully sequenced — using chemical fragmentation and chromatographic methods, work that established that proteins have a precise, genetically specified sequence rather than being some loosely repeating polymer. Having solved the sequencing problem for proteins, Sanger spent much of the 1970s tackling the analogous, and at the time far harder, problem for nucleic acids. He was famously modest and hands-on, reportedly happiest at the laboratory bench rather than in administrative roles, and he twice turned down offers of a knighthood before eventually accepting the Order of Merit, a rare British honor with no title attached — reportedly because he did not wish to be called "Sir." Sanger's second Nobel Prize (1980), shared with Walter Gilbert (for the independent Maxam-Gilbert chemical sequencing method) and Paul Berg (for recombinant DNA), recognized his 1977 chain-termination method. The Wellcome Trust Sanger Institute, one of the world's leading genome-sequencing centers and a major contributor to the Human Genome Project, is named in his honor.
Origins · 1 minFrederick Sanger and how this was discovered
Where this matters today
Even though massively parallel next-generation sequencing platforms have overtaken Sanger sequencing for whole-genome and whole-transcriptome work — offering vastly higher throughput at far lower per-base cost — Sanger sequencing has not disappeared. It remains the clinical and research gold standard for targeted, single-region confirmation, precisely because of its high per-base accuracy, long, contiguous, unambiguous single reads, and low setup cost for one-off targets.
- Clinical variant confirmation. When a next-generation sequencing panel flags a candidate disease-causing variant (in genes such as BRCA1, CFTR, or TP53, for example), diagnostic laboratories routinely re-sequence that exact region using Sanger sequencing before reporting the result to a patient, because Sanger chemistry has an extremely low false-positive rate for single-base substitutions and gives an easily human-interpretable chromatogram as documentary evidence.
- Forensic and paternity/identity testing of short, specific loci, where a single unambiguous confirmatory read is required.
- Plasmid and construct verification in molecular cloning and synthetic biology, following restriction digestion and ligation (building directly on Restriction Enzymes and DNA Cutting), to confirm insert identity, orientation, and absence of unwanted mutations before an engineered construct is used further.
- Sequencing PCR products directly, connecting back to The Polymerase Chain Reaction: an amplicon generated by PCR is a natural, abundant substrate for immediate Sanger sequencing without cloning.
- Microbial strain typing and species identification via targeted sequencing of marker genes (such as 16S rRNA in bacteria), where a single clean Sanger read of a few hundred bases is sufficient to place an isolate on a phylogenetic tree.
- Newborn screening follow-up and prenatal diagnosis, where speed, cost for a single locus, and interpretability outweigh the throughput advantages of massively parallel platforms.
- Teaching and training, since the chain-termination logic is the conceptual ancestor of, and pedagogical bridge to, sequencing-by-synthesis methods used in modern high-throughput platforms.
Ask the tutor: How exactly does a modern automated capillary sequencer distinguish four different fluorescent dyes in real time as fragments pass the detector?
What is the actual chemical difference between a dNTP and a ddNTP?
A deoxynucleoside triphosphate (dNTP) has a hydroxyl group (-OH) on the 3′ carbon of its deoxyribose sugar. A dideoxynucleoside triphosphate (ddNTP) has a hydrogen atom (-H) at that same 3′ position instead, because both the 2′ and 3′ positions lack an oxygen (hence "dideoxy"). DNA polymerase forms the sugar-phosphate backbone by linking the 3′-OH of the last incorporated nucleotide to the 5′-phosphate of the next one. Without a 3′-OH, there is no chemical group available for that bond, so once a ddNTP is added, the polymerase cannot extend that strand any further — synthesis stops permanently at that exact nucleotide.
Why are four separate reactions (or four dye colors) needed rather than just one?
Each ddNTP can only terminate synthesis at positions matching its own base — ddATP-terminated fragments end only where an A belongs in the synthesized strand, ddCTP-terminated fragments end only at C positions, and so on. To determine the full sequence, you need termination information for all four bases across the entire template. Classically this meant four separate reaction tubes and four gel lanes, read side by side. Modern automated sequencing achieves the same information in a single tube and single capillary lane by tagging each of the four ddNTPs with a spectrally distinguishable fluorescent dye, letting a laser detector read base identity by color as each fragment passes, rather than by which physical lane it occupies.
How long can a single Sanger sequencing read reliably go, and why is there a limit?
A typical high-quality Sanger read covers roughly 700 to 1000 bases of reliable sequence downstream of the primer. The limit arises from the physics of gel/capillary separation: resolution depends on the relative size difference between fragments, and as fragments get longer, a one-base difference becomes a smaller and smaller fraction of total length, so very large fragments migrate too similarly to be cleanly resolved. Near the primer, very short fragments also tend to run together and give poor separation. This produces a characteristic quality profile: low confidence near both ends of the read, high confidence in the middle few hundred bases — a key reason genome-scale sequencing needs many overlapping reads rather than one long one.
How does Sanger sequencing relate to and differ from next-generation sequencing?
Sanger sequencing analyzes one template (or one pooled PCR product) at a time, in a single capillary lane, producing one sequence read per reaction — its throughput is intrinsically limited, even though each individual read is highly accurate and fairly long. Next-generation sequencing platforms instead sequence millions to billions of DNA fragments simultaneously by immobilizing them on a solid surface (a flow cell) and using massively parallel, often shorter, cycles of synthesis-and-detection. NGS achieves far greater throughput and lower cost per base for large-scale sequencing (whole genomes, transcriptomes, metagenomes), but individual reads are typically shorter and slightly more error-prone than a well-run Sanger read, which is why Sanger sequencing is still used to confirm specific NGS-detected variants.
Can Sanger sequencing be used directly on genomic DNA, or does the DNA need to be amplified first?
In practice, genomic DNA must first be amplified, almost always via The Polymerase Chain Reaction, to produce enough copies of the specific region of interest for a robust sequencing signal — a single genomic DNA molecule does not provide nearly enough template material for reliable chain-termination chemistry and fluorescence detection. The PCR product (amplicon) is then used directly as the sequencing template, with a sequencing primer that anneals within or flanking the amplified region. This PCR-then-sequence workflow is the standard approach for confirming point mutations, screening candidate variants, and verifying cloned inserts.
Why does the newly synthesized strand read out in the chromatogram need to be reverse-complemented sometimes?
Sanger sequencing always reports the sequence of the strand actually built by the polymerase during the reaction, read in the 5′-to-3′ direction of synthesis. If the sequencing primer was designed to bind the "top" strand of a double-stranded template and synthesize along the "bottom" strand's information, the raw output corresponds to the bottom strand, not the top. To compare that raw read against a reference genome sequence, database entry, or a colleague's data reported in the conventional (say, sense-strand) orientation, you may need to computationally reverse-complement it, swapping A ↔ T and G ↔ C and reversing the order, so that everyone is describing the same physical DNA in the same orientation.
How does Sanger sequencing detect a heterozygous single-nucleotide variant, and how confident can you be in that call?
Because a diploid organism's PCR-amplified DNA contains roughly equal numbers of template molecules from each of the two homologous chromosomes, a position where the two alleles differ produces two competing termination products of the same length but different terminal base, superimposed in the same electrophoretic peak. The chromatogram shows two overlapping colored peaks of roughly equal height at that single position. Confidence is generally high for a clean roughly 1:1 double peak, though a laboratory would typically also confirm the call from an independent PCR/sequencing reaction (to rule out PCR or amplification artifacts) and check for consistency with any related clinical or family data before reporting it diagnostically.