Why Can’t Adenine Pair With Guanine?

Adenine and guanine are both purines, meaning they share the same bulky, double-ringed molecular structure. The DNA double helix is built to fit one purine opposite one smaller, single-ringed pyrimidine at each rung, so pairing two purines creates a geometric clash. On top of the size problem, the chemical groups on adenine and guanine that form hydrogen bonds simply don’t line up in the standard Watson-Crick arrangement. The result is that A-G “pairs” are unstable, distort the helix, and are actively rejected by the molecular machinery that copies and maintains your DNA.

The Size Problem

DNA’s double helix has a remarkably consistent width of about 2 nanometers. That width is maintained because every rung of the ladder pairs a purine (a two-ring base) with a pyrimidine (a one-ring base). Adenine pairs with thymine. Guanine pairs with cytosine. In both cases, the combined width of one large base and one small base fills the space between the two sugar-phosphate backbones perfectly. Try to slot in two purines across from each other and the pair is simply too wide. The backbone strands would have to bow outward, straining the bonds that hold the helix together. Conversely, two pyrimidines would leave a gap, and the helix would pinch inward. This purine-pyrimidine sizing rule is not a loose guideline; it is a hard geometric constraint imposed by the structure of the helix itself.

The Hydrogen Bond Mismatch

Even if you could somehow tolerate the wider geometry, adenine and guanine still wouldn’t form a proper pair because their hydrogen-bond donors and acceptors face the wrong way. In a standard A-T pair, adenine’s hydrogen-bond donor on its 6-amino group lines up neatly with a hydrogen-bond acceptor on thymine, and a second complementary contact completes the pair. In a G-C pair, three hydrogen bonds snap into place because the donors and acceptors on each base are mirror images of each other across the helix. When adenine faces guanine in the Watson-Crick orientation, the donors on one base run into donors on the other, and acceptors face acceptors. It’s like trying to connect two magnets north-pole to north-pole. The atoms that need to participate in hydrogen bonds can’t reach each other in the right geometry, so no stable pair forms in the standard orientation.

What Happens When A and G Are Forced Together

Saying adenine “can’t” pair with guanine is a useful simplification, but the more precise truth is that they can pair, just not well and not in the standard way. X-ray crystallography of a DNA strand engineered with an A-G mismatch shows that the two bases manage to form two hydrogen bonds, but only after guanine flips into an unusual orientation called the syn conformation. In the normal anti orientation, guanine’s bonding face points inward toward its partner, but when paired with adenine, it rotates roughly 180 degrees around its sugar attachment to present a different edge. The resulting A(anti)·G(syn) pair is held together by those two hydrogen bonds and further stabilized by a network of water molecules clustered in both the major and minor grooves of the helix.

That pair comes at a real cost, though. Crystal structures of these mismatched duplexes show disorder in the sugar-phosphate backbone near the mismatch site, a sign that the helix is under strain and doesn’t settle into a single clean conformation.¹ NMR studies of mismatched duplexes confirm the backbone distortion: phosphorus signals near the mismatch site shift in ways that break the normal pattern seen in well-ordered DNA, and the backbone torsion angles around the mismatch diverge from their usual values.² The exact geometry of the G-A mismatch is also surprisingly sensitive to context. High-resolution NMR work on nine different self-complementary DNA sequences showed that changing the bases flanking a G-A mismatch completely switches the pairing scheme, the backbone shape, and even whether the duplex forms at all.³

The thermodynamic penalty is steep. Studies of short duplexes containing a single mismatch find that the melting temperature, the temperature at which the two strands separate, drops by roughly 15 to 20 degrees Celsius compared to a perfectly matched sequence, with a free-energy cost of about 15 kilojoules per mole per mismatched position.⁴ In practical terms, a mismatched duplex that would normally be rock-solid at body temperature becomes shaky and prone to falling apart.

How DNA Polymerases Reject the Wrong Base

Cells don’t rely on thermodynamics alone to prevent A-G mismatches. The enzymes that copy DNA, called polymerases, have their own shape-checking step. When a polymerase is about to add a new nucleotide to a growing DNA strand, it goes through a conformational change, a physical closing motion around the incoming base and its template partner. If the incoming base is correct, the enzyme snaps shut smoothly and catalysis proceeds. If the base is wrong, the closing motion stalls.

Computational studies of human DNA polymerase beta illustrate this vividly. When the correct nucleotide (dCTP) arrives opposite a template guanine, the enzyme moves through a relatively straightforward cascade of structural rearrangements to assemble the active site for the chemical reaction. When the wrong nucleotide (dATP) shows up instead, the enzyme follows a strikingly different pathway with a different series of transition states, and the active site never assembles properly for catalysis.⁵ Further simulations of the same enzyme with a G:A mismatch in the active site found that a “prechemistry” rearrangement step, a structural adjustment the enzyme must make before the chemical bond-forming reaction, requires considerably more energy for the wrong base than for the right one. That extra energy barrier is the dominant factor slowing down incorrect insertion, and it provides most of the enzyme’s error-filtering power.⁶ The polymerase is essentially using shape as a proxy for correctness: if the base pair doesn’t have the right geometry, the enzyme’s own structure can’t reach the configuration needed to catalyze the reaction efficiently.

The Repair Crew That Catches What Slips Through

Even with a highly selective polymerase, mistakes occasionally get through. When they do, cells have a post-replication patrol system called mismatch repair. The key player is a protein called MutS (or its human equivalent, MSH2-MSH6), which slides along freshly copied DNA scanning for irregularities. MutS doesn’t read the chemical identity of each base directly. Instead, it checks the physical shape and flexibility of the helix. A mismatch like A-G distorts the double helix and makes it easier to bend at that spot, which is exactly what MutS is looking for.

Molecular dynamics simulations show that MutS uses a multistep detection strategy. First, it makes contact with the sugar-phosphate backbone over a stretch of about two helical turns, checking the overall shape. Then it uses large-scale motions of one of its structural domains to bend and unbend the DNA, testing how flexible the helix is at each point. Finally, it probes the local stability of individual base pairs by pushing on them with specific amino acid residues. A mismatched base pair destabilizes more easily than a correct one, and MutS identifies it by that weakness.⁷ Once MutS finds the mismatch, it signals a series of downstream enzymes that cut out the incorrectly paired stretch and resynthesize it using the other strand as a template.

A key detail in this recognition process is a conserved phenylalanine residue that normally stacks directly against one of the mismatched bases, forming a critical contact. Structural modeling shows this stacking interaction is the single most important direct contact MutS makes with a mismatch.⁸

When Oxidative Damage Sneaks A Opposite G

There’s a sneaky way adenine ends up across from guanine that doesn’t involve a copying error at all. Under conditions of oxidative stress, reactive oxygen species can chemically modify guanine in the DNA, converting it to a damaged form called 8-oxoguanine (commonly written 8-oxoG).⁹ The problem with 8-oxoG is that it can flip into the syn conformation and mimic thymine’s hydrogen-bonding pattern, which tricks the polymerase into inserting adenine across from it. The result is an 8-oxoG:A mismatch that, if left unrepaired before the next round of copying, leads to a permanent G-to-T mutation, a type of change called a transversion.

Cells have a specialized repair enzyme for exactly this situation. The MUTYH glycosylase specifically recognizes 8-oxoG:A mispairs and removes the adenine, giving the cell a chance to insert the correct cytosine instead.¹⁰ Crucially, this enzyme must find and act on the mismatch before the next replication cycle, because once the DNA is copied again, the adenine that was placed across from 8-oxoG becomes the template for a thymine, and the mutation becomes permanent.¹¹ Inherited defects in MUTYH are linked to a hereditary colorectal cancer syndrome, underscoring how important this single repair pathway is for genome integrity.

Tautomeric Shifts and Fleeting Mimicry

One of the more subtle ways an A-G mismatch can escape detection involves a phenomenon called tautomeric shifting. The bases in DNA can exist in more than one chemical form. In their common forms, adenine and guanine don’t fit together in Watson-Crick geometry, as we’ve discussed. But in rare alternative forms called tautomers, the positions of certain hydrogen atoms shift, and the base’s hydrogen-bonding pattern changes. If both adenine and guanine happen to be in their rare tautomeric forms at the same moment, they can briefly form a pair that mimics the shape and hydrogen-bond pattern of a legitimate Watson-Crick pair. This was actually Watson and Crick’s original explanation for how spontaneous mutations might arise.

Quantum-mechanical calculations have mapped out the trajectory of these double-proton-transfer events and confirmed that the A*·G* tautomeric pair (the asterisks denoting rare tautomers) can indeed adopt Watson-Crick-like geometry.¹² Related computational work has explored how homo-purine mispairs can toggle between wobble geometry and Watson-Crick-like geometry through sequential proton transfers within the pair, providing a plausible microstructural mechanism for how spontaneous transversion mutations, where a purine replaces a pyrimidine or vice versa, actually happen at the atomic level.¹³ These tautomeric events are extremely rare and short-lived, which is why they cause only a tiny fraction of base insertions to go wrong. But they represent a fundamental limit on copying fidelity that even the most accurate polymerase cannot fully overcome.

G·A Pairs Are Actually Common in RNA

Everything discussed so far applies to DNA, where the double-helical structure demands strict purine-pyrimidine pairing. RNA plays by different rules. Unlike DNA, RNA is usually single-stranded and folds back on itself to form complex three-dimensional shapes. In those folded structures, G·A pairs are not only tolerated but functionally important. They show up frequently in internal loops, hairpins, and junctions, where they help stabilize specific RNA architectures.

In RNA, G and A commonly form what’s called a “sheared” pair, where the two bases slide past each other rather than facing head-on. In this arrangement, the hydrogen bonds form between different atoms than the ones involved in Watson-Crick pairing: guanine’s 2-amino group bonds with adenine’s N7, and guanine’s N3 bonds with adenine’s 6-amino group. Both bases stay in the anti conformation, and the pair nestles into the wider A-form helix that RNA adopts.¹⁴ When multiple sheared G·A pairs stack consecutively, the guanines and adenines form separate columns of stacked bases, creating a distinctive structural motif that can serve as a platform for interactions with other parts of the RNA molecule or with proteins and small molecules.¹⁵

A broad survey of G·A pair geometry across known RNA structures reveals that these pairs appear in multiple modes of association and in a wide variety of secondary structural settings. They often occur as higher-order arrangements involving two or more G·A pairs clustered together, and these clusters play a role in bridging different secondary structural elements within a larger folded RNA.¹⁶ The ribosome, the enormous molecular machine that translates RNA into protein, is full of G·A pairs that contribute to its structural stability. So while G·A pairing is a mistake in DNA, it is a feature in RNA. The difference comes down to context: DNA needs a uniform helix for faithful information storage, while RNA needs structural diversity for its many functional roles.

Expanding the Genetic Alphabet Beyond A, T, G, and C

The strict pairing rules that exclude A-G pairs have inspired synthetic biologists to ask whether new, artificial base pairs could be designed from scratch. The idea is to create molecules that obey the same geometric and hydrogen-bonding logic as natural pairs but carry different information, effectively adding new “letters” to the genetic alphabet. Several research groups have succeeded. One approach uses ring-expanded purines paired with ring-expanded pyrimidines, maintaining the size-complementarity principle while introducing four non-standard hydrogen bonds. One such pair, called ImNN:NaOO, works not only for single nucleotide insertion but can be faithfully copied by PCR alongside the natural A-T and G-C pairs.¹⁷

Another strategy developed a two-unnatural-base-pair system for expanding the genetic code in protein synthesis. One pair, between a thienyl-substituted purine (called “s”) and pyridin-2-one (called “y”), allows site-specific incorporation of unnatural components into RNA during transcription. Combined with a second unnatural pair for tRNA synthesis, the system enables translation of proteins containing amino acid analogs encoded by codons that don’t exist in nature.¹⁸ These synthetic base pairs have been shown to replicate faithfully alongside natural A-T and G-C pairs in PCR and to be transcribed into RNA, demonstrating that the four-letter genetic alphabet is not a biological necessity but an evolutionary settlement.¹⁹

The success of these synthetic pairs actually reinforces why adenine and guanine don’t pair with each other. Every working artificial pair obeys the same core principle that the natural pairs do: one large base opposite one small base, with hydrogen-bond donors lined up across from acceptors. Two purines facing each other violate both rules simultaneously, which is why no amount of clever engineering has produced a functional purine-purine pair for DNA replication. The constraint isn’t arbitrary; it’s structural, and it appears to be universal across both natural and synthetic genetic systems.