The Match Between: Pairing, the Four-Letter Alphabet, and Why the Genetic Line Was Never Flat

Flatland as a starting point

The most useful abstraction in modern biology is a line. Write the genome as a string of four letters — A, T, G, C — and it can be stored, compared, searched and copied by any machine that handles text. Dawkins put the point memorably: the genetic code is digital, and a gene is information written in a linear alphabet. As a description of what a reading enzyme meets along one strand, the abstraction is excellent. It is also the reason the rest of this chapter has to exist.

A line has one dimension and one direction. The molecule has neither. The flat reading is a projection — the shadow the molecule casts when only one kind of question is put to it — and like every projection in this volume, it is honest about what it keeps and silent about what it drops. The discipline is to keep the projection and refuse to mistake it for the object.

The first break: two strands, one match

The line breaks at the very first step past the string. DNA is two strands, and they run in opposite directions: one from its 5′ end to its 3′ end, the other the reverse. Each letter on one strand faces a partner on the other — A with T, G with C — so that either strand is sufficient to rebuild its companion. That is the observation Watson and Crick closed their 1953 paper on, with deliberate understatement: the specific pairing immediately suggests a possible copying mechanism.

Consider what that does to the notion of where the information is. On a line, a letter means what it is. In the duplex, a letter means what it can match. The sequence on one strand is fully determined by the other, which is to say that the content is not stored in either strand alone but in the relation between them. Remove the partner and the content is still recoverable — which is exactly how replication and repair work — but only because the rule of the match was never in any single letter. It belongs to the pair.

Chargaff had seen the shadow of this before the structure was known: in double-stranded DNA the amount of A equals the amount of T, and G equals C, across species whose overall composition differs widely. A ratio that holds everywhere a line would not require it is the fingerprint of a second dimension.

The alphabet is a pairing system, not a list

The flat view treats A, T, G and C as four interchangeable symbols. They are not interchangeable, and the differences are what make them work. A and G are purines, two-ringed; T and C are pyrimidines, single-ringed. Every correct pair joins one of each, which is why the rungs of the ladder are all the same width — about two nanometres across the helix — and why the molecule can wind smoothly regardless of what it says.

The pairs also differ in how tightly they hold. A–T is joined by two hydrogen bonds, G–C by three. A stretch rich in G–C is harder to pull apart; a stretch rich in A–T opens more easily, and the places where the duplex is opened to be read are, in many organisms, A–T rich. So the same four letters that spell a message along the strand also write a map of mechanical stability across it. A single symbol carries two kinds of information at once: what it says, and how firmly the saying is held.

This is the cross-dimensional relationship the chapter is named for. The alphabet works along the strand (sequence), across the strand (complementarity and bond strength), and — as the next sections show — around and outside it. The flat reading keeps the first of these and discards the rest, not because they are small but because a line has nowhere to put them.

The turn: stacking, water and the helix

Pairing supplies specificity, but it is not what chiefly holds the molecule together. Yakovchuk, Protozanova and Frank-Kamenetskii measured the contributions separately in 2006 and found that stacking — the flat faces of neighbouring bases lying against one another like a column of coins — dominates the stability of the duplex, with the pairing contributing much less than the textbook picture implied. The rungs recognise; the stack holds.

The stack, in turn, is set by water. The sugar-phosphate backbone is charged and faces outward into the surrounding water; the bases, which avoid water, turn inward and shelter against each other. The helix is what that arrangement looks like when it is allowed to settle: in the common B form, about ten and a half base pairs to each full turn, each step rising roughly a third of a nanometre. The turn is not decoration. It is the geometry that lets a charged outside and a sheltered inside coexist — the hydrophilic turn this book has followed since its opening essay.

And the turn depends on the letters. Different neighbouring pairs stack with different energies and prefer slightly different twists and tilts, so the local shape of the helix varies with the sequence. The line does not merely sit on the helix. It bends it.

The grooves: reading without opening

Because the two backbones are not spaced evenly around the helix, the surface carries two grooves of unequal width: a wide major groove and a narrow minor one. In 1976 Seeman, Rosenberg and Rich pointed out that each of the four possible base-pair orientations — A–T, T–A, G–C, C–G — presents a distinct pattern of hydrogen-bond donors and acceptors on the floor of the major groove. A protein can therefore read the sequence from the outside, by touch, without opening the duplex at all.

Rohs and colleagues added a second channel in 2009: proteins also read the shape of the minor groove, whose width narrows in certain sequence contexts and concentrates negative electrostatic potential there. That is a reading of geometry and charge, not of letters. The same four symbols, once paired and turned, become a landscape of shape and field that a binding protein recognises the way a hand recognises a key.

At this point the one-dimensional description is not merely incomplete; it is describing a different act. What the cell's own readers do most of the time is not scan a line. It is fit to a surface.

The fold and the long reach

Beyond the helix the relations keep extending. The duplex winds around histone proteins, is twisted and untwisted by enzymes that change its supercoiling, and is folded into loops that bring sequences far apart on the line into contact in space. A regulatory stretch can act on a gene thousands of letters away because, in the folded molecule, it is not far away. Distance on the string and distance in the cell are two different measures, and the cell uses the second.

The same string is also read in both directions, from both strands, sometimes in overlapping frames, and its transcripts are cut and rejoined before use. None of this contradicts the digital description. All of it shows that the digital description is the part of the molecule a line can hold.

What the chapter carries, and what it does not

Read through this part's governing principle, DNA is the most familiar instance of a form whose content lives in a relation. The drumhead carries field at its folds (Chapter 80); the altermagnet carries information in modes its bulk cancels (Chapter 81); the axon carries conduction speed in a shape its preparation had smoothed (Chapter 82). The genetic molecule carries its sequence in a match between strands, its stability in the difference between two and three bonds, its readability in groove geometry, and its regulation in folds that make the far near. Each of these is a relation. None is an ingredient added to the line.

What it does not carry should be said as plainly. Nothing here overturns the digital code or the central role of sequence; the linear reading remains the right tool for the questions it answers. The claim is about dimension, not about priority: that the flat reading is a projection of a molecule that is structurally cross-dimensional from its first pair onward, and that a theory of heredity built only from the projection will keep being surprised by the object.

The checkpoint question this chapter leaves visible is the one it opened with. When a description works as well as the line does, what is it quietly declining to see — and which of those things is the molecule using?