Chemistry of Chain Elongation: DNA Polymerase Catalysis

The two-metal-ion mechanism is not a historical curiosity; it is a live target of modern biotechnology and medicine. Nucleoside reverse transcriptase inhibitors (NRTIs) used against HIV — zidovudine (AZT), lamivudine, tenofovir — are chain terminators: they lack a 3′-OH (or have a modified sugar ring), so once incorporated they leave no nucleophile for the next round of catalysis, halting viral DNA synthesis exactly as dideoxynucleotides do in Sanger sequencing. Understanding the precise geometry of metal-A coordination to the 3′ position explains both why these drugs work and why some resistance mutations in reverse transcriptase subtly reposition the active site to exclude the modified sugar while still accepting natural dNTPs. High-fidelity PCR enzymes (Phusion, Q5, KOD) are engineered variants of thermostable polymerases selected or designed to sharpen the geometric fidelity checks at the two-metal center, reducing mismatch incorporation rates for applications like clinical diagnostics and next-generation sequencing library preparation, where a single erroneous base can matter. Conversely, translesion and error-prone polymerases (family Y, such as human Pol η and Pol ι) have relaxed active-site geometry that tolerates non-Watson–Crick pairing — useful for bypassing DNA damage during replication stress, at the cost of higher intrinsic error rates, illustrating that the same two-metal chemistry can be tuned along a spectrum from extremely strict to deliberately permissive. Sequencing-by-synthesis platforms (Illumina and related technologies) directly exploit the chemistry described here: fluorescently labeled, reversibly terminated dNTPs are incorporated one at a time by DNA polymerase, imaged, and then chemically unblocked to expose a fresh 3′-OH for the next cycle — a technological reengineering of exactly the nucleophile-generation step central to Mechanism 1. In structural biology and drug discovery, transition-state mimics (nonhydrolyzable dNTP analogs, vanadate transition-state analogs) are used to trap and visualize the pentacoordinate phosphorane geometry directly, informing rational design of polymerase inhibitors for antiviral and anticancer chemotherapy (e.g., nucleotide analogs used in cancer treatment that terminate DNA synthesis in rapidly dividing cells). The chemistry taught here — one nucleophile, one electrophilic phosphorus, two metals — turns out to be one of the most therapeutically exploited reactions in all of biochemistry.

Ask the tutor: How does the fingers-domain conformational change fit into the overall kinetic scheme, and why does it act as an additional fidelity checkpoint before the chemical step even occurs?

Consider the primer–template junction sitting in a DNA polymerase active site. The primer strand ends in a free 3′-hydroxyl group; the template strand specifies, by Watson–Crick pairing, which dNTP should enter next. The incoming dNTP carries a triphosphate tail — Pα, Pβ, Pγ — attached to the 5′ carbon of the deoxyribose. The chemistry that must occur is transfer of the nucleotidyl group (base–sugar–Pα) onto the primer's 3′-oxygen, releasing PPi (Pβ–Pγ, still connected to each other) as the leaving group. The active site is built around a right-hand-shaped fold (fingers, palm, thumb) that is a conserved architecture across polymerase families despite low sequence identity. The palm domain houses the catalytic core: two or three strictly conserved aspartate (or glutamate) residues whose carboxylate oxygens coordinate two divalent metal cations, conventionally labeled metal A and metal B. In E. coli DNA polymerase I (the Klenow fragment) these are Asp705 and Asp882 (numbering varies by polymerase family, but the arrangement is universal). Both metals are usually Mg²⁺ in vivo, though Mn²⁺ can substitute in vitro with altered fidelity. Metal A sits adjacent to the primer 3′-OH. Its role is to lower the pKa of that hydroxyl (normally ~13 in bulk water) so that at physiological pH a substantial fraction exists as the reactive alkoxide, and to help align this nucleophile directly in line with the Pα–O(bridging) bond that will break. Metal B sits between the β- and γ-phosphates of the incoming dNTP. Its role is electrostatic shielding: the triphosphate carries four negative charges tightly clustered together, and without charge neutralization the developing negative charge on the pyrophosphate leaving group (and the repulsion between the approaching alkoxide nucleophile and the anionic phosphate) would make the reaction prohibitively slow. Metal B also helps position the triphosphate so that Pα presents its face correctly for backside attack, and it stabilizes the transition state as the Pβ–O–Pα bridging oxygen begins to carry substantial negative charge in the pentacoordinate intermediate.

Notice the trajectory drawn from the 3′-oxygen through Pα to the bridging oxygen: this is a straight line, not a bent path. Nucleophilic substitution at tetrahedral phosphorus proceeds through a pentacoordinate trigonal-bipyramidal transition state in which nucleophile and leaving group occupy the two apical positions, 180° apart. This in-line geometry is enforced by the polymerase active site — the fingers domain closes down around a correctly paired incoming dNTP and forces exactly this alignment, which is one reason a mismatched or wrongly-shaped nucleotide reacts far more slowly. The other essential piece of the picture is directionality. Because the nucleophile is always the primer's 3′-OH and the electrophile is always the incoming dNTP's Pα, chain growth can only occur by adding nucleotides to the 3′ end. There is no chemistry available for extending a strand from its 5′ end using free nucleoside triphosphates as monomers; the 5′ end of a growing strand carries a phosphate (or triphosphate, on the very first primer) rather than a nucleophilic hydroxyl. This is the chemical root of the rule that all known template-dependent polymerases synthesize DNA and RNA exclusively 5′→3′, which in turn dictates the entire architecture of the The Replication Fork and Its Machinery, including why one new strand is made continuously and the other in fragments.

The step chemistry performs — dNTP → dNMP incorporated + PPi — is only mildly exergonic on its own; what makes the overall process of chain elongation essentially irreversible in the cell is the rapid subsequent hydrolysis of pyrophosphate by inorganic pyrophosphatase, which pulls the equilibrium of the polymerase reaction far to the right by Le Chatelier's principle. This coupling is the same logic seen throughout biosynthesis: activation via NTP hydrolysis, followed by hydrolysis of the byproduct to make the overall process unidirectional.

The chemical transformation catalyzed at every step of chain elongation is:

where:

• $\text{DNA}_n$ — the primer/growing strand with $n$ residues already incorporated
• $\text{dNTP}$ — the incoming 2′-deoxynucleoside 5′-triphosphate (dATP, dGTP, dCTP, or dTTP)
• $\text{PPi}$ — inorganic pyrophosphate, $\text{P}_2\text{O}_7^{4-}$ (protonation state pH-dependent)

The rate law for a single-nucleotide incorporation step, under conditions where the enzyme–primer–template–dNTP complex is pre-formed, is well described by a two-step kinetic scheme:

where:

• $k_1, k_{-1}$ — dNTP binding and dissociation rate constants
• $k_{\text{pol}}$ — the chemical step: phosphoryl transfer at Pα (typically the rate-limiting, or near rate-limiting, step for a correct dNTP)
• $k_{\text{PPi}}$ — PPi release

For a correctly paired dNTP, $k_{\text{pol}}$ is fast (often 100–1000 s⁻¹ depending on the polymerase), while a mismatched dNTP suffers a large drop in $k_{\text{pol}}$ (often 10³–10⁵-fold) — this differential is the kinetic basis of fidelity, elaborated further in Fidelity and Proofreading in DNA Synthesis. The overall thermodynamics of nucleotide incorporation combine two coupled reactions:

where:

• $\Delta G^{\circ\prime}_{\text{phosphodiester bond formation}} \approx +0.5$ to $-1$ kcal/mol (near thermoneutral; phosphodiester bond formation from a nucleotidyl-phosphate and an alcohol is not strongly favorable by itself)
• $\Delta G^{\circ\prime}_{\text{PPi hydrolysis}} \approx -7.7$ kcal/mol (per mole PPi hydrolyzed by pyrophosphatase, physiological conditions)

Reading this: the chemistry at the polymerase active site alone would give a reaction with an equilibrium constant not far from 1 — reversible on the timescale of catalysis. It is the fast, essentially irreversible hydrolysis of the pyrophosphate byproduct, catalyzed by ubiquitous inorganic pyrophosphatase, that renders the net two-step process (nucleotide addition + PPi hydrolysis) strongly exergonic and unidirectional, with an overall $\Delta G^{\circ\prime}$ around $-7$ to $-8.5$ kcal/mol per nucleotide added.

Example 1 — Foundational: counting bonds broken and formed, and locating every charge. Write out the immediate chemical neighborhood before and after one incorporation event, using dATP as the incoming nucleotide added opposite a template T, onto a primer ending in ...–3′-OH. Before reaction: the primer terminal sugar carries a free 3′-OH (neutral, one lone pair used for the O–H bond, two lone pairs available). The incoming dATP carries Pα, Pβ, Pγ in a linear triphosphate chain; at physiological pH the triphosphate is almost fully ionized, carrying a net charge of approximately $-4$ (i.e., $\text{dATP}^{4-}$), each phosphate bearing one or two non-bridging oxyanions. Bond broken: the Pα–O(bridge to Pβ) bond (one P–O phosphoanhydride bond). Bond formed: the 3′-O–Pα bond (one P–O phosphodiester bond). Proton removed: one, from the 3′-OH, ultimately released to bulk solvent (or transiently held by an active-site base). Charge accounting: before, the leaving triphosphate oxygen was a bridging (non-charged, two-coordinate) oxygen within the anhydride; after, that same oxygen is now a terminal (single-coordinate) oxygen on free pyrophosphate, carrying a full formal negative charge, charge-balanced by loss of the proton from the nucleophile. Formal charge is conserved throughout: one proton leaves the system (to bulk water, well away from the transition state), and the negative charge that would otherwise reside on the 3′-alkoxide instead ends up stabilized on the departing pyrophosphate — consistent with metal B's role as an electrostatic sink for exactly this charge. Net result: primer is now $n+1$ residues long, terminating again in a free 3′-OH (on the newly added dA residue), ready for the next round; PPi ($\text{P}_2\text{O}_7^{4-}$, roughly, at physiological pH) has been released into solution, where it is rapidly hydrolyzed to 2 Pi.

Example 2 — Applied: energetics of extending a 37-nucleotide primer by one base, using real cellular concentrations. Suppose a replication fork is extending a primer, and the relevant cellular concentrations are: $[\text{dATP}] = 24\ \mu\text{M}$, $[\text{DNA}_n]$ (primer termini, treated as a reactant) effectively saturating (large excess, polymerase-bound), $[\text{PPi}] = 0.5\ \mu\text{M}$ (kept low by pyrophosphatase), and $[\text{Pi}] = 1\ \text{mM}$. Take $\Delta G^{\circ\prime}_A = -0.75$ kcal/mol for the polymerase step alone (Reaction A from the derivation above), $T = 310\ \text{K}$ (physiological, 37 °C), $R = 1.987\times10^{-3}$ kcal·mol⁻¹·K⁻¹. We want the actual free energy change of Reaction A alone under these non-standard conditions:

Since $[\text{DNA}_n]$ and $[\text{DNA}_{n+1}]$ are both DNA strands present essentially stoichiometrically (one polymerase, one primer, extended by one unit — treat their ratio as 1, since we are asking about the free energy of a single incorporation event on the same molecule, not a bulk concentration ratio), this simplifies to:

Compute the ratio:

Compute $RT$ at 310 K:

Compute the log term:

So:

Therefore:

Now couple in PPi hydrolysis at its own actual cellular free energy (using its own concentration dependence, $\Delta G_B \approx \Delta G^{\circ\prime}_B + RT\ln([\text{Pi}]^2/[\text{PPi}])$, with $[\text{Pi}] = 10^{-3}$ M and $[\text{PPi}] = 0.5\times10^{-6}$ M):

Net free energy for the coupled process under these actual cellular concentrations:

This is even more favorable than the standard-state estimate, driven mainly by the fact that dNTP concentration exceeds PPi concentration by roughly 50-fold in the cell (a consequence of active pyrophosphatase activity continually degrading PPi), which pulls Reaction A itself substantially forward even before PPi hydrolysis is counted. Limiting-case check: if pyrophosphatase were absent and PPi accumulated to equal dNTP concentration (ratio = 1), the $RT\ln(\text{ratio})$ term for Reaction A alone would vanish (since $\ln 1 = 0$), leaving $\Delta G_A = \Delta G^{\circ\prime}_A = -0.75$ kcal/mol — still slightly forward, but far less driven, exactly matching the qualitative claim that pyrophosphatase's job is to keep this ratio small.

Example 3 — Exam/competition level: distinguishing correct from incorrect base pairing using $k_{\text{pol}}/K_d$ and relating it to the two-metal mechanism. A DNA polymerase is measured (pre-steady-state kinetics, single-turnover conditions) incorporating dGTP opposite a template C (correct, Watson–Crick pair) and dTTP opposite the same template C (incorrect, a G·T-type mismatch is not applicable here — take it as a genuine mispair, e.g., T opposite C). The measured parameters are: Correct (dGTP·C): $k_{\text{pol}} = 300\ \text{s}^{-1}$, $K_d = 5\ \mu\text{M}$. Incorrect (dTTP·C): $k_{\text{pol}} = 0.02\ \text{s}^{-1}$, $K_d = 2000\ \mu\text{M}$. The specificity constant for insertion is $k_{\text{pol}}/K_d$ (formally analogous to $k_{\text{cat}}/K_M$ for a simple enzyme, since dNTP binding is fast and reversible relative to the chemical step under these conditions). Compute each:

The discrimination factor (fidelity contributed by this insertion step alone) is:

Convert this to an apparent free-energy discrimination between correct and incorrect transition states, using $\Delta\Delta G^{\ddagger} = -RT\ln F$ at 310 K:

So the polymerase active site discriminates against the mismatch by roughly 9.6 kcal/mol at the level of the transition state for the chemical step — an enormous effect that cannot be explained by Watson–Crick hydrogen-bonding energy differences alone (a single hydrogen bond is worth only 1–2 kcal/mol in this context; mismatches often retain at least some hydrogen bonding). The extra discrimination comes directly from the two-metal-ion geometry discussed in Mechanism 1: a mispaired base distorts the sugar–phosphate backbone geometry at the primer terminus just enough to misalign the 3′-O nucleophile relative to Pα and to metal A, dramatically slowing $k_{\text{pol}}$ even when binding (reflected in $K_d$) is only modestly weaker. This single-nucleotide insertion fidelity of order $10^5$–$10^6$ combines multiplicatively with a subsequent proofreading exonuclease step (typically another 10²–10³-fold) to give the overall replication fidelity of roughly $10^9$–$10^{10}$ observed in cells — the subject taken up fully in Fidelity and Proofreading in DNA Synthesis.

Arthur Kornberg discovered DNA polymerase I in E. coli in 1956, the first enzyme shown to synthesize DNA in vitro, for which he shared the 1959 Nobel Prize in Physiology or Medicine with Severo Ochoa. Kornberg's five "rules" for DNA polymerization — the requirement for template, primer, all four dNTPs, Mg²⁺, and 5′→3′ synthesis — set the empirical foundation on which mechanistic chemistry was later built. But it was the crystallographic era, decades later, that revealed the atomic mechanism. Thomas A. Steitz, working with Chi-Wang Ho and Catherine Joyce among others at Yale in the late 1980s and 1990s, solved the crystal structure of the Klenow fragment of DNA polymerase I bound to primer-template DNA and observed two divalent metal ions positioned exactly where mechanistic chemistry required them: one juxtaposed to the primer 3′-OH, the other bridging the triphosphate. Steitz proposed the general two-metal-ion mechanism for all polymerases (and, by extension, many other phosphoryl transfer enzymes) in a landmark 1993 review, a proposal since validated by dozens of subsequent structures across DNA polymerases, RNA polymerases, reverse transcriptases, and even ribozymes. Steitz shared the 2009 Nobel Prize in Chemistry (for ribosome structure, a related but distinct body of work), and his polymerase mechanism papers remain among the most cited in structural enzymology.