The DNA Double Helix

Take a single strand of DNA and you have a floppy, chemically monotonous polymer. It's a sugar-phosphate rope with four kinds of flat rings hanging off it. Take two such strands and run them antiparallel. Pair every purine with its complementary pyrimidine. Then something extraordinary happens: the molecule folds itself into a rigid, self-repairing, information-bearing cable. This shape is essentially the same in every cell of every organism on Earth. That shape — the right-handed B-form double helix, its major and minor grooves, its 3.4 Å rise and 10.5 base-pair repeat — is not a decorative flourish. It is the geometric solution to a specific problem. That problem is storing digital information in a chemically stable, faithfully copyable, and physically compact form. This article is about why that particular geometry, and no other, is the one nature settled on. The double helix is not a single fact to memorize. It is the output of a set of constraints. These include bond lengths, glycosidic torsion angles, base-stacking geometry, sugar pucker, and backbone antiparallelism. All of them pull in the same direction. Given The Phosphodiester Backbone and Base Pairing and the Chemistry of Complementarity as building blocks, the helical parameters described here are essentially forced. Once you fix that A pairs with T and G with C via specific hydrogen-bonding faces, the rest follows. The two backbones must also run in opposite chemical directions. From these facts, the twist, rise, and groove asymmetry all follow from simple stereochemistry. The big picture is that DNA's information content (the base sequence) and its structural stability (the helix) are chemically decoupled. The bases carry the digital code. The sugar-phosphate backbone and the stacking/pairing geometry carry the structural burden. This separation is what lets any sequence — any "message" — adopt essentially the same physical scaffold. This is precisely what makes DNA a general-purpose information medium. Its shape does not depend on what it says. By the end of this article you will be able to derive the helical repeat from the twist angle per base pair. You will be able to explain mechanistically why the grooves are unequal in width and depth. You will also be able to compute rise and total contour length for a given number of base pairs. Finally, you will see exactly how antiparallel strand orientation, together with glycosidic bond geometry, forces right-handed B-DNA rather than some other fold. This sets up later treatments of Stability of the Double Helix: Stacking and Hydrogen Bonding, Alternative DNA Forms: A, B, and Z, and DNA Supercoiling and Topology.

Key takeaways

DNA is two antiparallel polynucleotide strands wound into a right-handed double helix stabilized by Watson–Crick base pairing and base stacking.
B-form DNA has a helical rise of 3.4 Å per base pair, a twist of ~36° per base pair, and ~10.5 bp per full turn (34 Å pitch).
Antiparallel strand polarity plus the anti glycosidic conformation of the bases forces an asymmetric backbone-to-base geometry, which creates unequal major (~22 Å) and minor (~12 Å) grooves.
The major groove exposes more unique hydrogen-bonding and steric information per base pair than the minor groove, which is why most sequence-reading proteins bind there.
Base pairs are perpendicular to the helix axis and stack directly atop one another, so the helix is stabilized mostly by stacking (van der Waals/dispersion + hydrophobic exclusion of water), not primarily by the hydrogen bonds themselves.
Right-handedness is not arbitrary: it is the low-strain solution once C1′–N glycosidic bonds, sugar pucker, and backbone antiparallelism are all satisfied simultaneously.
Key idea: the double helix is the unique low-energy fold that reconciles antiparallel backbones, Watson–Crick pairing, and base stacking into one repeating unit — its geometry is a consequence of chemistry, not an independent postulate.

Where this lives today

The specific groove and helical geometry described here is the working substrate for essentially every DNA-binding protein in molecular biology and biotechnology. Restriction endonucleases such as EcoRI dock into the major groove and read out a specific hydrogen-bonding and steric pattern from the edges of base pairs, a recognition strategy only possible because the major groove — as derived above — presents a wide, chemically rich face; the minor groove, by contrast, is the preferred binding site of small-molecule DNA-binding drugs such as netropsin, distamycin, and the AT-selective dye DAPI (used ubiquitously in fluorescence microscopy to stain nuclei), which exploit its narrow, electronegative, AT-rich-compatible geometry. The helical repeat number directly governs the design of CRISPR guide RNA spacing and of engineered zinc-finger and TALE arrays, all of which must be spaced and phased in integer or half-integer numbers of helical turns to present their recognition modules correctly. PCR primer and restriction-site engineering, nucleosome positioning (histones wrap ~147 bp = ~1.65 left-handed superhelical turns of B-DNA around the octamer core, a direct application of the contour-length relation derived above), and the entire field of DNA nanotechnology (DNA origami, which relies on precise helical-turn counting to design crossover points between duplexes) all use these same helical parameters as design inputs. Even the basic logic of Sanger and next-generation sequencing — reading a linear sequence off a double-stranded template — depends on the antiparallel, complementary geometry established here being essentially invariant across every sequence context.

Ask the tutor: Why does a small change in average twist angle (36° → 34.5°) produce a noticeably different bp/turn, and what does that imply about how sensitive helical phasing calculations are to the twist value you assume?

The physical picture — thorough prose

Picture the two backbones first, stripped of their bases. Each is a chain of alternating deoxyribose and phosphate, linked by 3′→5′ phosphodiester bonds (as built up in The Phosphodiester Backbone). Because the phosphodiester linkage is directional — a free 5′ end and a free 3′ end — every single strand has a chemical "arrow." In duplex DNA the two arrows point in opposite directions: one strand runs 5′→3′ left to right, its partner runs 5′→3′ right to left. This is the antiparallel arrangement, and it is not a stylistic choice. If you try to pair two strands running in the same direction (parallel), the geometry of the N-glycosidic bonds and the sugar-phosphate backbone cannot simultaneously satisfy Watson-Crick hydrogen bonding across the whole length of the duplex without severe steric strain — parallel-stranded DNA exists only in rare, specialized structural motifs, never as the generic double helix. Now hang the bases off each backbone, one per sugar, attached through the C1′–N glycosidic bond (N9 for purines, N1 for pyrimidines). Each base adopts the anti conformation about this bond — the bulk of the base points away from the sugar ring rather than folding back over it (syn). In the anti conformation, a purine on one strand can rotate into the same plane as a pyrimidine on the partner strand and form the Watson-Crick hydrogen-bonding pattern: two hydrogen bonds for A·T, three for G·C. Because A·T and G·C pairs have essentially identical overall dimensions — same C1′–C1′ distance (≈10.6 Å), same glycosidic bond angles relative to the pair axis — any sequence of base pairs can be threaded onto the same helical backbone without distorting it. This dimensional equivalence of the two Watson-Crick pairs is the deep reason DNA can encode arbitrary sequences while keeping a constant, predictable shape. Once a base pair is formed, it behaves as a rigid, roughly planar rectangular plate perpendicular to the helix axis. Stack many such plates directly on top of each other, twisted by a small angle relative to the plate below, and you generate a helix automatically — this is the same geometric principle as a spiral staircase made of stacked rectangular treads, each rotated slightly from the one below. The stacking interactions between adjacent base pairs (π-orbital overlap of the aromatic rings, van der Waals contacts, and the hydrophobic exclusion of water from between the flat faces) provide the dominant thermodynamic driving force holding the stack together, a point developed fully when treating Stability of the Double Helix: Stacking and Hydrogen Bonding. The two sugar-phosphate backbones spiral around the outside of this stack of plates, tracing out two intertwined helical ribbons. Because each base is attached to its sugar off-center — not through the middle of the base-pair rectangle but nearer one edge — the two backbone ribbons do not sit symmetrically opposite each other around the base-pair rectangle. Instead, they divide the cylinder's circumference into two unequal arcs. This asymmetry, projected outward from the twisting stack of base pairs, creates two spiral grooves of different width running along the helix: a wide, deep major groove and a narrower, shallower minor groove.

Figure

Antiparallel strands, base pairing, and the resulting major/minor grooves in B-DNA

Finally, consider the helix as a whole, viewed from the side. In canonical B-form DNA each base pair is rotated by a mean twist angle of 36° relative to its neighbor and translated 3.4 Å along the helix axis. Because 360°/36° = 10, one might naively expect exactly 10 base pairs per turn; the measured average in solution is closer to 10.4–10.5 bp per turn (a mean twist nearer 34.3–34.6°), a small but biologically important correction that we return to below. The result is a helix with a pitch (the axial rise for one complete 360° turn) of about 34 Å, a diameter of about 20 Å, and grooves whose dimensions — roughly 22 Å deep/12 Å wide for the major groove and 12 Å deep/6 Å wide for the minor groove — are set entirely by the stacking geometry and the off-center attachment of bases to backbone.

Figure

Helical parameters: rise, twist, and pitch defined along the axis

The core ideas — governing relationships

The helical repeat, pitch, and groove geometry all follow from three linked quantities: the axial rise per base pair, the twist angle per base pair, and the number of base pairs per full turn.

h \;=\; \frac{P}{n}

where: $h$ is the axial rise per base pair (Å), $P$ is the helical pitch — the axial distance for one complete 360° turn (Å) — and $n$ is the number of base pairs per turn. Reading: rise is simply the pitch divided among however many base pairs make up one turn; increasing $n$ (a "looser," more slowly twisting helix) proportionally decreases the rise contribution to a given pitch only if pitch itself is held fixed.

n \;=\; \frac{360^\circ}{\Omega}

where: $n$ is base pairs per turn, $\Omega$ is the mean helical twist angle (degrees) between adjacent base pairs. Reading: the twist angle and the base pairs-per-turn are reciprocal descriptions of the same rotational geometry; you can specify either one and the other is fixed.

L \;=\; N \cdot h

where: $L$ is the total contour length of a duplex of $N$ base pairs, $h$ is rise per base pair (3.4 Å for B-DNA). Reading: contour length grows linearly with base-pair number, which is why genome size (in bp) directly predicts physical chromosome length — this is the relation used, for instance, to compute that the human genome (~3.2 × 10⁹ bp per haploid set) is about 1 meter long when fully extended.

\Delta\phi_{\text{groove}} \;\propto\; \left(\theta_{\text{glyco}} - 180^\circ\right)

(qualitative relation) where: $\Delta\phi_{\text{groove}}$ represents the asymmetry between major- and minor-groove arc lengths around the helix circumference, and $\theta_{\text{glyco}}$ is the angle subtended at the base-pair center by the two glycosidic (C1′–N) bonds. Because this angle is not 180° (the two glycosidic bonds do not point in exactly opposite directions across a diameter), the backbone divides the circle unevenly, producing unequal grooves. This relation is qualitative — the point to retain is the sign and existence of the asymmetry, not a numeric proportionality constant.

Derivation AFrom twist angle to helical repeat and pitch

Goal. Starting only from the measured mean twist angle between adjacent base pairs in B-DNA, derive the number of base pairs per helical turn and the pitch, and show how a small deviation from the idealized 36° twist explains why real B-DNA in solution has 10.4–10.5 bp/turn rather than exactly 10.

Step 1 — Define the twist angle geometrically. Each base pair is modeled as a rigid rectangular plate stacked on the one below it, related by a rotation of $\Omega$ about the helix axis and a translation $h$ along that axis (a screw operation). X-ray fiber diffraction of B-DNA (Franklin and Gosling's Photo 51, and subsequent refined fiber and crystal structures) gives an average $\Omega \approx 36^{\circ}$ and $h \approx 3.4\ \text{Å}$ for the idealized repeating unit.

Step 2 — Apply the closure condition for one full turn. A full helical turn corresponds to a net rotation of exactly $360^\circ$ . If each base-pair step contributes $\Omega$ degrees of that rotation, the number of steps required to close one turn is

n = \frac{360^\circ}{\Omega} = \frac{360^\circ}{36^\circ} = 10 \ \text{bp/turn (idealized)}.

Step 3 — Compute the idealized pitch. Using $P = n h$ ,

P = 10 \times 3.4\ \text{Å} = 34\ \text{Å}.

This is the textbook B-DNA pitch, and it matches the layer-line spacing seen in B-form fiber diffraction patterns.

Step 4 — Reconcile with the measured solution-average twist. Careful crystallographic and solution studies (cyclization kinetics, gel mobility of DNA minicircles, and high-resolution crystal structures) show the average twist per base pair in solution is closer to $\Omega \approx 34.3^{\circ}$ – $34.6^{\circ}$ , not exactly 36°, because twist varies substantially with local sequence (dinucleotide step) — AT-rich steps tend to twist more than the average, GC steps less, and the fiber-diffraction value of 36° is itself an average over a repetitive synthetic sequence used in the original diffraction experiments. Using the solution average,

n = \frac{360^\circ}{34.5^{\circ}} \approx 10.43\ \text{bp/turn},

consistent with the widely quoted value of 10.4–10.5 bp/turn for B-DNA free in solution (as opposed to 10.0 bp/turn sometimes used for the idealized fiber model, or 10.5 as a common rounded consensus value).

Step 5 — Recompute the pitch with the corrected turn number. Keeping $h = 3.4\ \text{Å}$ fixed (rise per bp is far less sequence-sensitive than twist),

P = n h \approx 10.43 \times 3.4\ \text{Å} \approx 35.5\ \text{Å}.

This is close to, but slightly larger than, the idealized 34 Å fiber value — consistent with the small systematic differences seen between the original fiber-diffraction B-form model and solution/crystallographic structures of specific sequences.

Step 6 — State the general formula and its limiting case. Combining Steps 1–3 into one closed expression for pitch as a function of twist,

P(\Omega) = \frac{360^{\circ}}{\Omega}\, h.

As a limiting-case check: if $\Omega \to 360^{\circ}/n_{0}$ for any integer $n_0$ , the helix closes exactly after $n_0$ base pairs with no cumulative phase drift — the defining condition for a strictly periodic (as opposed to only quasi-periodic) helical repeat. For $\Omega = 36^\circ$ exactly, $n_0 = 10$ is an integer and the helix would repeat after exactly one turn with no fractional base pair left over; the real, slightly smaller twist of ≈34.5° breaks this exact integer closure, which is why real B-DNA has an irrational-looking 10.4–10.5 bp/turn rather than a tidy 10.

Go deeper · Derivation B · 3 minWhy the backbone asymmetry forces unequal major and minor grooves

Derivation BWhy the backbone asymmetry forces unequal major and minor grooves

Goal. Show, from the glycosidic bond geometry alone, why the two sugar-phosphate backbones cannot divide the helix circumference into two equal grooves, and estimate the resulting groove width asymmetry.

Step 1 — Set up the base-pair rectangle. In a Watson-Crick pair, the two glycosidic bonds (C1′–N9 for the purine, C1′–N1 for the pyrimidine) are the attachment points where the backbone strands connect to the rigid base-pair plate. The measured C1′–C1′ distance across a Watson-Crick pair is essentially invariant between A·T and G·C, at approximately 10.6 Å — a critical fact, since it means the backbone attachment geometry (and hence the groove pattern) does not depend on sequence.

Step 2 — Identify the key angle. If the two C1′–N glycosidic bonds pointed directly away from each other (angle of 180° through the pair center), the backbone strands would divide the helix cylinder's circumference into two exactly equal arcs, giving two identical grooves — this is the (incorrect) naive expectation from pure symmetry. In reality, because of the specific hydrogen-bonding geometry that holds A·T and G·C pairs together (fixed N–H···N and N–H···O bond angles and lengths, discussed in Base Pairing and the Chemistry of Complementarity), the angle subtended at the pair center between the two glycosidic bonds is measurably less than 180°, roughly 120°.

Step 3 — Translate the angular asymmetry into arc-length asymmetry. Picture the circle traced by the backbone phosphates as viewed down the helix axis. The two glycosidic bond vectors from Step 2 split this circle into two arcs, one subtending roughly $360^\circ - 2\times(180^\circ - 120^\circ)$ -type geometry from the base-pair rectangle's uneven attachment — the qualitative upshot, confirmed by fiber diffraction and crystallography rather than by this simplified angle alone, is that one arc is distinctly larger (spanning the wider, deeper major groove) and the other is distinctly smaller (spanning the narrower, shallower minor groove).

Step 4 — Compare to measured groove dimensions. Crystallographic B-DNA structures give approximate major-groove dimensions of 22 Å depth and 11–12 Å width, and minor-groove dimensions of about 12 Å depth and 6 Å width (widths measured as the perpendicular phosphate-to-phosphate distance across the groove, corrected for backbone van der Waals radii). The major groove is both wider and deeper than the minor groove by roughly a factor of ~2 in width.

Step 5 — Connect groove asymmetry to protein recognition. Because the major groove is wider, it exposes a larger, more chemically differentiated set of hydrogen-bond donor/acceptor atoms and methyl groups projecting from the edges of each base pair (for example, the N7 and O6 of guanine, the N6 amino of adenine, or the C5-methyl of thymine, all distinguishable from one groove but not reliably from the narrower one). This is the chemical basis for why the overwhelming majority of sequence-specific DNA-binding proteins (transcription factors, restriction enzymes) read the major groove rather than the minor groove — the minor groove presents a much less sequence-discriminating pattern of atoms (mostly just N3 of purines and O2 of pyrimidines, whose pattern is similar for A·T and T·A, and for G·C and C·G).

Step 6 — Limiting case check. If the glycosidic bond angle in Step 2 were exactly 180° (a hypothetical, symmetric attachment), Step 3 predicts two identical grooves of equal width and depth — a structure with no major/minor distinction at all. Since real B-DNA unambiguously shows two grooves differing by roughly a factor of two in width, this limiting case confirms that the observed groove asymmetry is a direct, necessary consequence of the non-180° glycosidic geometry, not an independent structural feature bolted on separately.

Worked examples

Example 1 (foundational) — Contour length of a defined plasmid. A common cloning plasmid, pUC19, is 2686 bp long (double-stranded, circular). Compute its contour length in nanometers if fully relaxed into B-form. Using $L = N h$ with $N = 2686$ bp and $h = 3.4\ \text{Å/bp}$ :

L = 2686 \times 3.4\ \text{Å} = 9132.4\ \text{Å}.

Converting to nanometers ( $1\ \text{nm} = 10\ \text{Å}$ ):

L = 913.24\ \text{nm} \approx 0.91\ \mu\text{m}.

Sanity/limiting check: for a much shorter, non-round fragment — say a 147 bp nucleosome-core-length segment — the same formula gives $L = 147 \times 3.4 = 499.8\ \text{Å} \approx 50.0\ \text{nm}$ , matching the widely cited ~50 nm length of DNA wrapped around one histone octamer (before accounting for the ~1.65 left-handed superhelical turns of wrapping, which is a separate topological effect treated under DNA Supercoiling and Topology).

Example 2 (applied) — Number of helical turns in a gene and phasing of a regulatory element. A bacterial promoter element requires two protein-binding sites to be on the same face of the helix, separated by 31 bp, for a transcriptional activator to loop DNA correctly and contact RNA polymerase cooperatively. Using $n = 10.5$ bp/turn (a standard working value for B-DNA in solution), determine whether 31 bp places the second site on the same face as the first, and find the angular offset if not. Number of turns spanned:

\frac{31\ \text{bp}}{10.5\ \text{bp/turn}} = 2.952\ \text{turns}.

The fractional part, 0.952 turns, corresponds to an angular offset from perfect same-face alignment of

\Delta\theta = (1 - 0.952) \times 360^{\circ} = 0.048 \times 360^{\circ} \approx 17.3^{\circ}.

So a 31 bp spacer places the two sites on nearly, but not exactly, the same face of the helix — offset by about 17°, which is close enough that many DNA-bending or flexible-linker proteins can still bridge the gap, but a spacer of exactly $10.5 \times 3 = 31.5$ bp would phase the sites perfectly (illustrating why natural regulatory spacers cluster near multiples of the helical repeat, and why insertions of a non-integral number of turns between two protein sites are a classic tool for testing helical phasing requirements in molecular genetics). Limiting check: if the spacer were exactly $21$ bp $= 2 \times 10.5$ , the fractional turn count would be exactly $2.000$ , giving $\Delta\theta = 0^{\circ}$ — perfect same-face phasing, confirming the formula degenerates correctly at exact integer-turn spacings.

Example 3 (exam/competition level) — Reconstructing rise and twist from a crystallographic sequence with non-uniform base-pair steps. A crystal structure of a 10-bp oligonucleotide duplex reports the following individual base-pair step parameters (twist, rise) for each of the 9 dinucleotide steps: (38.2°, 3.2 Å), (31.5°, 3.5 Å), (36.9°, 3.3 Å), (29.8°, 3.6 Å), (40.1°, 3.1 Å), (33.4°, 3.4 Å), (37.7°, 3.3 Å), (30.9°, 3.5 Å), (35.5°, 3.4 Å). Compute (a) the average twist and rise, (b) the implied bp/turn, and (c) the total axial length spanned by the duplex, and compare to the idealized B-DNA prediction. (a) Average twist (sum of 9 values divided by 9):

\Omega_{\text{avg}} = \frac{38.2+31.5+36.9+29.8+40.1+33.4+37.7+30.9+35.5}{9} = \frac{314.0^{\circ}}{9} = 34.89^{\circ}.

Average rise:

h_{\text{avg}} = \frac{3.2+3.5+3.3+3.6+3.1+3.4+3.3+3.5+3.4}{9} = \frac{30.3\ \text{Å}}{9} = 3.367\ \text{Å}.

(b) Implied base pairs per turn:

n = \frac{360^{\circ}}{34.89^{\circ}} = 10.32\ \text{bp/turn}.

This is close to, but slightly tighter (fewer bp/turn, more twist per step) than, the 10.5 bp/turn solution-average value — consistent with this particular sequence being somewhat more tightly wound than the genomic average, well within the natural sequence-dependent range (roughly 10.0–10.6 bp/turn depending on local base-pair steps, with pyrimidine-purine steps such as CA/TG generally under-twisted and purine-purine steps often over-twisted relative to the mean). (c) Total axial length spanned by 9 steps (10 base pairs have 9 connecting steps):

L = \sum_{i=1}^{9} h_i = 30.3\ \text{Å}.

Compare to the idealized prediction using the canonical $h = 3.4\ \text{Å}$ for 9 steps:

L_{\text{ideal}} = 9 \times 3.4\ \text{Å} = 30.6\ \text{Å}.

The crystallographic sum (30.3 Å) is within 1% of the idealized fiber-diffraction prediction (30.6 Å), demonstrating that although individual base-pair steps vary by as much as ±0.3 Å in rise and ±5° in twist depending on local sequence context, the average over even a short 9-step sample converges quickly toward the canonical B-DNA parameters — the idealized $h=3.4$ Å, $\Omega=36°$ values are genuine statistical averages, not fictions, even though no single real base-pair step exactly equals them. Limiting check: if all nine steps had been assigned the exact canonical values (36.0°, 3.4 Å) instead of the scattered crystallographic values, both averages in (a) would trivially reduce to $\Omega_{\text{avg}}=36.0°$ and $h_{\text{avg}}=3.4$ Å, and part (b) would give $n = 360/36 = 10$ exactly — recovering the idealized fiber-diffraction model as the special uniform-sequence limit of the more general sequence-averaged treatment.

Limiting cases, scope & common traps

Situation	What happens	Why
Idealized fiber-diffraction B-DNA	Exactly 36° twist, 3.4 Å rise, 10 bp/turn, 34 Å pitch	Homogeneous repetitive sequence used in original diffraction fibers; an average model, not a per-base-pair truth
B-DNA in solution, mixed sequence	~34.3–34.6° average twist, 10.4–10.5 bp/turn, pitch ≈35–36 Å	Local sequence-dependent twist variation averages to a slightly smaller mean twist than the idealized fiber value
Very AT-rich tract	Often narrower minor groove, altered twist locally	A·T steps stack differently than G·C steps; propeller twist and base-pair roll vary with sequence
Highly negatively supercoiled DNA in vivo	Local unwinding, reduced effective twist	Torsional stress redistributes between twist and writhe — full treatment under DNA Supercoiling and Topology
Low water activity / high salt (A-form conditions)	Shorter, wider helix, deep narrow major groove, shallow wide minor groove	Sugar pucker shifts from C2′-endo (B-form) toward C3′-endo (A-form) — covered under Alternative DNA Forms: A, B, and Z
Alternating purine-pyrimidine sequence, high salt	Can adopt left-handed Z-DNA	Alternating syn/anti glycosidic conformation changes the entire backbone zig-zag topology — a genuinely different helix, not a B-DNA variant
Parallel-stranded synthetic constructs	Do not form the standard Watson-Crick duplex	Antiparallel geometry is required for the standard base-pairing register; parallel duplexes need non-Watson-Crick pairing schemes

Common trapMisconception: "the hydrogen bonds hold the helix together"

Hydrogen bonds between paired bases provide specificity — they are what makes A pair with T and G pair with C rather than some other combination — but they are not the dominant source of helix stability. Melting thermodynamics and single-strand-stacking experiments show that base-stacking interactions (dispersion forces between stacked aromatic rings, plus the hydrophobic release of ordered water from between stacked bases) contribute more free energy of stabilization than the hydrogen bonds themselves. A helix of unpaired but stacked bases is measurably more stable than one with paired but unstacked bases. This distinction matters directly for melting-temperature prediction and is developed quantitatively when treating Stability of the Double Helix: Stacking and Hydrogen Bonding.

The chemist behind it

Rosalind Franklin generated the sharp, information-rich X-ray fiber diffraction pattern — "Photo 51" — of the B-form of DNA at King's College London in 1952, working alongside Raymond Gosling. Franklin's meticulous control of hydration conditions let her cleanly separate the A-form (drier) and B-form (wetter) diffraction patterns, and her analysis already established the helical symmetry, the 34 Å repeat, and the position of the phosphate backbone on the outside of the molecule, based on rigorous application of Patterson-function and helical-diffraction theory. James Watson and Francis Crick, at the Cavendish Laboratory in Cambridge, combined this diffraction evidence (seen partly through an internal Medical Research Council report and partly through Maurice Wilkins) with Erwin Chargaff's base-composition rules (A = T, G = C in amount) and physical model-building using accurately scaled atomic models, to propose the complementary, antiparallel, right-handed double-helix structure in April 1953. The achievement was fundamentally an act of structural chemistry: reconciling bond lengths, van der Waals radii, and hydrogen-bond geometry into a single physically consistent, chemically sensible three-dimensional model that simultaneously explained base-pairing specificity, X-ray diffraction data, and a natural mechanism for replication.

Origins · 1 minRosalind Elsie Franklin and how this was discovered

Rosalind Elsie Franklin

1920–1958 · King's College London, X-ray crystallography of DNA fibers

Franklin's rigorous physical chemistry training (she had done doctoral work on the microstructure of coals and carbons in Paris before turning to nucleic acids) gave her an unusually quantitative command of diffraction theory. Her 1953 manuscript, submitted essentially simultaneously with the Watson-Crick paper, independently derived the space group and helical parameters of the A-form of DNA from first-principles diffraction analysis, and her B-form data provided the decisive experimental constraints — the 34 Å repeat and the helical diffraction cross pattern diagnostic of a helix — that made the double-helix model buildable.

History

- 1868 — Friedrich Miescher isolates "nuclein" (crude DNA) from white blood cell nuclei.

- 1944 — Oswald Avery, Colin MacLeod, and Maclyn McCarty show DNA, not protein, is the transforming principle carrying genetic information.

- 1950 — Erwin Chargaff reports that in DNA from many organisms, %A ≈ %T and %G ≈ %C.

- 1952 — Rosalind Franklin and Raymond Gosling obtain Photo 51, the definitive B-form X-ray fiber diffraction pattern.

- 1953 — James Watson and Francis Crick publish the antiparallel, complementary, right-handed double-helix model in Nature.

Why is DNA right-handed and not left-handed?

Right-handedness in canonical B-DNA emerges from the specific stereochemistry of the deoxyribose sugar (its C2′-endo pucker in B-form), the anti glycosidic conformation of the bases, and the antiparallel backbone arrangement, all combined with the steric and electronic preferences of stacked base pairs. Given fixed D-deoxyribose chirality (a biochemical constant, not a free parameter) and anti glycosidic torsion, right-handed stacking is the lower-energy, less strained solution compared with left-handed stacking under the same constraints. Left-handed DNA does exist — Z-DNA — but it requires an alternating purine-pyrimidine sequence with alternating syn/anti glycosidic conformations and a fundamentally different, zig-zag backbone path, making it a distinct structural family rather than a simple mirror image of B-DNA.

Does every organism's DNA have exactly the same helical parameters?

No. The canonical values (3.4 Å rise, 36° twist, 10 bp/turn) describe an idealized, sequence-averaged B-form model derived from fiber diffraction of a repetitive sequence. Real genomic DNA shows local sequence-dependent variation in twist, rise, roll, and slide at each base-pair step; average bp/turn in solution is closer to 10.4–10.5. Additionally, environmental conditions (hydration, salt, and specific sequence composition) can shift DNA toward the A-form (common in RNA-DNA hybrids and dehydrated DNA) or, for alternating purine-pyrimidine tracts under high salt, toward left-handed Z-DNA. The B-form described here is the dominant, biologically default form under normal cellular conditions.

Why do most sequence-specific DNA-binding proteins read the major groove rather than the minor groove?

The major groove is wider and deeper, exposing a larger and more chemically distinguishable set of functional groups projecting from the edges of each base pair — including atoms that differ between A·T and T·A, and between G·C and C·G, which is essential for genuinely sequence-specific recognition. The minor groove, being narrower, presents a more repetitive pattern of hydrogen-bond acceptors that looks similar for several different base-pair identities, making sequence discrimination harder from that face alone. Transcription factors, restriction enzymes, and most site-specific recombinases dock helices or loops into the major groove for this reason, though a smaller set of proteins and drugs specifically exploit minor-groove geometry (often for AT-rich recognition, exploiting the narrower, more electronegative minor groove found in A-tract DNA).

How was the antiparallel orientation of the two strands established experimentally, rather than just assumed?

The antiparallel model was strongly motivated by stereochemical necessity — Watson and Crick found that a parallel arrangement could not simultaneously satisfy correct Watson-Crick hydrogen-bonding geometry and reasonable backbone bond angles/lengths across an extended helix — but it was subsequently confirmed by direct experiments, including chemical mapping of restriction enzyme cut sites (showing 5′ and 3′ overhangs consistent with antiparallel strands), enzymatic sequencing that reads defined 5′→3′ polarity on each strand, and ultimately atomic-resolution X-ray crystal structures of short DNA oligonucleotides that directly resolve the opposite chemical polarity of the two backbones.

What is the difference between helical twist, rise, and pitch, and how are they related?

Twist ( $\Omega$ ) is the rotation angle between adjacent base pairs about the helix axis, typically averaging ~34.5°–36° in B-DNA. Rise ( $h$ ) is the axial translation between adjacent base pairs, about 3.4 Å. Pitch ( $P$ ) is the total axial distance for one complete 360° turn of the helix, related to twist and rise by $P = (360^\circ/\Omega)\, h$ . Twist and rise are local, per-base-pair-step quantities that can vary with sequence context; pitch is a derived, whole-turn quantity computed from their (often sequence-averaged) values.

Why does the double helix matter for replication and repair, beyond just storing information?

The antiparallel, complementary structure means each strand is a complete template for reconstructing its partner — this is the direct structural basis for semiconservative replication, in which each daughter duplex retains one parental strand and synthesizes one new strand by Watson-Crick pairing. The same complementarity underlies most DNA repair pathways (mismatch repair, base-excision repair, nucleotide-excision repair), which use the intact strand as a template to correct damage or errors on the other strand. Without strict, geometrically enforced complementary pairing, none of these template-copying mechanisms would be chemically possible.

Is the B-form the only double-helical form of DNA, and how common is it in cells?

B-form is the dominant form under normal intracellular ionic strength and hydration, and it is the form implied whenever "the DNA double helix" is discussed without qualification. However, DNA is structurally polymorphic: A-form arises under dehydrating conditions and is the standard form of RNA-DNA hybrid duplexes and double-stranded RNA (which cannot adopt B-form due to the 2′-OH group), while Z-form is a left-handed alternative accessible to alternating purine-pyrimidine sequences, particularly under high salt or negative supercoiling, and has been implicated in specific regulatory and immune-recognition contexts in cells. These variants are structurally related but chemically and geometrically distinct families, treated separately under Alternative DNA Forms: A, B, and Z.