Pyrosequencing

Pyrosequencing is a real-time bioluminescent sequencing-by-synthesis method of DNA sequencing that determines nucleotide order by detecting the stoichiometric release of inorganic pyrophosphate (PP_i) upon nucleotide incorporation via a four-enzyme cascade that converts PP_i into ATP and visible 560-nm chemiluminescent light. Unlike first-generation Sanger sequencing, which terminates strand elongation with 2′,3′-dideoxynucleotides and requires post-reaction electrophoretic separation, pyrosequencing uses native, unblocked deoxynucleotides dispensed sequentially in cyclic flows (A → C → G → T), monitoring nascent DNA synthesis optically in real time.

Conceived in 1985–1987 by Swedish biochemist Pål Nyrén and developed into a functional solid-phase and solution-phase sequencing chemistry in 1996–1998 by Mostafa Ronaghi, Mathias Uhlén, and Pål Nyrén at the Royal Institute of Technology (KTH) in Stockholm (Science, 1998), pyrosequencing established the sequencing-by-synthesis methodology. In 2005, Jonathan M. Rothberg and colleagues at 454 Life Sciences miniaturized pyrosequencing across 1.6 million picoliter fiber-optic wells on the GS20 / GS FLX instrument, launching the era of second-generation Next-Generation Sequencing (NGS)—which sequenced the first individual human genome by NGS (James D. Watson, 2008) and Svante Pääbo's draft Neanderthal genome (2010).

Pyrosequencing operates in an isothermal aqueous reaction chamber containing a primed single-stranded DNA template, two substrates—adenosine 5′-phosphosulfate (APS) and D-luciferin—and four cooperating enzymes: DNA polymerase, ATP sulfurylase, firefly luciferase, and apyrase. The four deoxynucleotides are pulsed into the chamber one at a time in a fixed cyclical order.

Step 1: Polymerase Incorporation and Pyrophosphate (PP_i) Release

When the dispensed deoxynucleoside triphosphate (dNTP) is complementary to the first unpaired base on the template strand downstream of the 3′ primer terminus, the exonuclease-deficient Klenow fragment of Escherichia coli DNA Polymerase I catalyzes nucleophilic attack by the primer 3′-OH on the α-phosphate of the incoming dNTP. Covalent phosphodiester bond formation elongates the DNA strand by n bases (where n = 1 for a single base, or n > 1 for a homopolymer stretch of identical bases) and releases an equimolar quantity of inorganic pyrophosphate (PP_i, P₂O₇⁴⁻):

(DNA)_m + n dNTP → (DNA)_m+*n* + n PP_i

If the dispensed nucleotide is not complementary to the template base, no polymerase incorporation occurs and zero PP_i is released.

Step 2: ATP Sulfurylase Conversion of PP_i to ATP

In the presence of Mg²⁺ ions, the enzyme ATP sulfurylase (recombinant Saccharomyces cerevisiae sulfate adenylyltransferase) quantitatively converts the released inorganic pyrophosphate and the co-substrate adenosine 5′-phosphosulfate (APS) into ATP and sulfate ions within milliseconds:

PP_i + APS → ATP + SO₄²⁻

Step 3: Luciferase Chemiluminescence and the dATPαS Substitution

The newly synthesized ATP functions as the rate-limiting energy cofactor for firefly luciferase (from Photinus pyralis), which catalyzes the oxidative decarboxylation of D-luciferin in the presence of molecular oxygen (O₂) into electronically excited oxyluciferin, AMP, PP_i, and CO₂. Relaxation of excited oxyluciferin to its ground state emits visible yellow-green photons at λ_max = 560 nm, detected by a charge-coupled device (CCD) camera or photomultiplier tube:

ATP + D-luciferin + O₂ → AMP + PP_i + oxyluciferin + CO₂ + *h*ν (560 nm)

A central biochemical obstacle discovered by Ronaghi and Nyrén during early development is that native 2′-deoxyadenosine 5′-triphosphate (dATP) functions directly as a weak substrate for firefly luciferase, triggering false background light emission as soon as dATP is dispensed regardless of whether DNA polymerase incorporates it. To eliminate this background luminescence, pyrosequencing substitutes natural dATP with the phosphorothioate analog 2′-deoxyadenosine-5′-O-(1-thiotriphosphate) (dATPαS) (specifically the resolved S_p stereoisomer). Klenow DNA polymerase incorporates dATPαS opposite template thymine (T) with normal kinetics and releases unmodified PP_i (consisting of the β and γ phosphates), whereas firefly luciferase is sterically unable to utilize dATPαS as an energy substrate.

Step 4: Apyrase Nucleotide Degradation

Before the next nucleotide in the cycle can be dispensed, any unincorporated dNTPs remaining in solution and all ATP generated during the light flash must be removed. The fourth enzyme present in the reaction mixture—apyrase (a nucleotide-degrading diphosphohydrolase purified from potato, Solanum tuberosum)—hydrolyzes unincorporated dNTPs and residual ATP sequentially into deoxynucleoside monophosphates (dNMPs), AMP, and orthophosphate (P_i, PO₄³⁻), which neither ATP sulfurylase nor luciferase can use:

dNTP / ATP + 2 H₂O → dNMP / AMP + 2 P_i

Once apyrase returns the chamber to darkness (~60 seconds per flow cycle), the next nucleotide is pulsed into the reactor.

Emulsion PCR and Fiber-Optic PicoTiterPlates (2005)

While 96-well microtiter plate pyrosequencing (Qiagen PyroMark) reads short 30- to 80-bp stretches for clinical CpG DNA methylation and KRAS/BRAF mutation quantification, Jonathan M. Rothberg and colleagues at 454 Life Sciences (Margulies et al., Nature 2005) scaled pyrosequencing into the first commercial Next-Generation Sequencing platform (454 GS20 and GS FLX Titanium).

In the 454 workflow, sheared genomic DNA fragments (400–800 bp) ligated to A and B adapters are captured at one molecule per 28-μm streptavidin-coated Sepharose microbead inside aqueous micro-droplets suspended in mineral oil (emulsion PCR, emPCR). Thermal cycling amplifies each single template molecule into ~10 million clonal copies immobilized on its bead. Approximately 1.6 million DNA-carrying microbeads are then centrifuged into a honeycombed PicoTiterPlate (PTP)—a fused fiber-optic faceplate etched with millions of 44-μm diameter, 75-picoliter wells (sized so that one 28-μm DNA bead fits per well), packed alongside smaller 3-μm packing beads coated with immobilized ATP sulfurylase and luciferase. As laminar fluidic flows sweep T, A, C, and G reagents across the open tops of the wells, a cooled CCD camera bonded to the base of the fiber-optic bundle records flashes across all 1.6 million wells simultaneously, generating 400–600 Mb of 400–700 bp reads in a 10-hour run.

Pyrogram Interpretation and the Homopolymer Indel Limitation

The output of a pyrosequencing run is a quantitative time-intensity trace called a pyrogram, where the horizontal axis represents sequential nucleotide flows and the vertical axis represents integrated photon intensity. Because the dispensed dNTPs carry native, unblocked 3′-OH groups, whenever the template strand contains a homopolymer run of n identical contiguous bases (for example, 5′-TTTT-3′), a single flow of complementary dATPαS causes DNA polymerase to incorporate all four adenines in succession during that single flow cycle, releasing 4 equivalents of PP_i, generating 4 equivalents of ATP, and emitting a light peak 4 times higher than a single-base (1-mer) peak.

For short runs of 1 to 4 identical bases, pyrogram peak height scales linearly with homopolymer length n. However, for homopolymer tracts of ≥6 identical nucleotides (e.g., AAAAAA or TTTTTT), local substrate depletion of APS/luciferin, apyrase competition, and CCD detector saturation cause light intensity to lose linearity. Consequently, distinguishing a 7-mer from an 8-mer becomes statistically ambiguous, making homopolymer insertion/deletion (indel) errors the primary accuracy limitation of both pyrosequencing and Ion Torrent semiconductor sequencing—leading to the phase-out of 454 sequencers between 2013 and 2016 in favor of Illumina's 3′-blocked reversible terminator chemistry.

Why must pyrosequencing use dATPαS instead of natural dATP during Adenine nucleotide flows?

In the pyrosequencing cascade, firefly luciferase emits a 560-nm light flash whenever it binds adenosine triphosphate (ATP). Because natural 2′-deoxyadenosine triphosphate (dATP) differs from ribonucleoside ATP only by a single oxygen atom at the 2′ carbon of the sugar ring, firefly luciferase recognizes natural dATP directly as a substrate, producing non-specific chemiluminescence every time dATP is injected into the reaction well—even when no DNA synthesis occurs. Substituting the sulfur-modified analog dATPαS (carrying a sulfur atom on the α-phosphate) allows DNA polymerase to incorporate the base normally and cleave off native β,γ-PP_i, while blocking direct recognition by luciferase.

Why are homopolymer insertion and deletion errors the dominant error mode in pyrosequencing?

Unlike Illumina sequencing-by-synthesis, which uses nucleotides blocked at the 3′-OH position so only one base can be added per cycle, pyrosequencing uses unmodified 3′-OH nucleotides. When the template contains a homopolymer run of n identical nucleotides (such as six consecutive thymines, TTTTTT), all n complementary adenines are incorporated within a single reagent flow, and the basecaller must infer the integer count n from the analog height of the emitted light flash. Above n = 5–6 bases, variance in photon emission (~10%–15% CV) causes the Gaussian intensity distributions of n = 6, n = 7, and n = 8 to overlap, producing 1-bp insertion or deletion calls and lowering the Phred quality score.

Why is short-read pyrosequencing (PyroMark) still used in clinical pathology for CpG DNA methylation analysis?

When genomic DNA is treated with sodium bisulfite (bisulfite sequencing), unmethylated cytosines are deaminated to uracil (which amplifies as thymine, T, in PCR), whereas methylated cytosines (5mC) remain cytosine (C). Pyrosequencing quantifies the percentage of methylation at each individual CpG site within a 50-bp clinical amplicon (such as the MGMT promoter in glioblastoma or MLH1 in colorectal cancer) by sequentially dispensing C and T at the target CpG position and calculating the stoichiometric ratio of the C light peak height to the (C + T) light peak heights with ±1% to 2% quantitative precision in under 30 minutes.

1. Nyrén, P. (1987). "Enzymatic Method for Continuous Monitoring of DNA Polymerase Activity." Analytical Biochemistry, 167(2), 235–238. https://doi.org/10.1016/0003-2697(87)90158-890158-8)
2. Ronaghi, M., Uhlén, M., & Nyrén, P. (1998). "A Sequencing Method Based on Real-Time Pyrophosphate." Science, 281(5375), 363–365. https://doi.org/10.1126/science.281.5375.363
3. Margulies, M., Egholm, M., Altman, W. E., Attiya, S., Bader, J. S., Bemben, L. A., Berka, J., Braverman, M. S., Chen, Y.-J., Chen, Z., et al. (2005). "Genome Sequencing in Microfabricated High-Density Picolitre Reactors." Nature, 437(7057), 376–380. https://doi.org/10.1038/nature03959
4. Wheeler, D. A., Srinivasan, M., Egholm, M., Shen, Y., Chen, L., McGuire, A., He, W., Chen, Y.-J., Makhijani, V., Roth, G. T., et al. (2008). "The Complete Genome of an Individual by Massively Parallel DNA Sequencing." Nature, 452(7189), 872–876. https://doi.org/10.1038/nature06884