DNA Should Repel DNA. These Images Show Two Helices Lining Up Anyway

DNA has an interpersonal problem.

The outside of each double helix is lined with phosphate groups carrying negative electrical charge. Bring two DNA molecules close together and, in the simplest picture, they should push apart, like matching poles of a magnet.

Yet researchers have now imaged separate DNA duplexes lying side by side with their spiraling grooves in remarkably precise alignment. Atom-by-atom computer simulations suggest how the apparently unfriendly encounter could happen: positively charged metal ions may settle between the molecules and bridge one helix to the other.

The result, published in Nucleic Acids Research, gives structural support to a physical model proposed roughly 25 years ago. It also requires a few careful distinctions.

The microscope did not see individual ions. It did not record two molecules moving together from opposite ends of a cell, recognizing a match and progressively zipping shut. It captured static structures on a surface. The proposed atomic mechanism came from simulations designed to reproduce and explain the geometry in those images.

That division between observation and inference is the heart of the story.

Reading the ridges of DNA

The team, led by researchers at the universities of Sheffield and York, used atomic-force microscopy, or AFM. Instead of forming an image with light, an AFM moves a very fine probe over a surface and records its topography.

The researchers deposited DNA on freshly cleaved mica and scanned it in liquid. Their best images resolved more than the general outlines of the molecules. They showed the repeating major and minor grooves that spiral around each double helix.

Where two duplexes made especially close contact, those grooves maintained nearly the same register. A major groove on one molecule sat beside a major groove on the other; the minor grooves lined up too. The contact continued for roughly 11 to 15 nanometers, equivalent to about three or four turns of the helix.

This was the kind of geometry predicted by a family of ideas known as the electrostatic zipper or helical-coherence model.

The model begins by complicating the familiar statement that DNA is negatively charged. Its phosphate backbones are negative, but DNA in solution is surrounded by positive counterions. If some of those ions collect in its grooves, the molecule's surface no longer resembles a uniformly charged cylinder. It acquires a spiral pattern of relatively positive and negative regions.

Two helices in the right orientation could fit those patterns together. Sequence-dependent variations in DNA's shape could then allow some alignments to remain in register more effectively than others.

Physicists Alexei Kornyshev and Sergey Leikin proposed in a 2001 Physical Review Letters paper that these variations in DNA's surface-charge pattern might let intact homologous duplexes recognize one another. Later experiments produced evidence for protein-free DNA pairing, but images resolving the predicted groove alignment had remained elusive.

The new AFM images supply that structural view.

What the microscope could not see

Even a high-resolution AFM image cannot identify the individual ions between the molecules. To investigate those invisible interactions, the researchers turned to all-atom molecular-dynamics simulations.

They modeled pairs of 32-base-pair DNA duplexes in virtual boxes containing water and ions. The duplexes began parallel, with their axes 3.5 nanometers apart. Across 40 simulations totaling more than 20 microseconds, the calculations tracked the movements of the modeled atoms.

In simulations containing nickel ions, the DNA molecules approached closely in almost every run. The tightest contacts formed when minor grooves faced one another. Nickel ions occupied sites where they could interact with both duplexes, forming DNA-ion-DNA bridges.

A 5′-GTAC-3′ sequence formed an especially persistent pocket in the nickel simulations. Two ions could occupy the opposing minor grooves, while a third, more mobile ion sometimes sat between the backbones. Some of the most stable modeled ions remained in place for more than 60 nanoseconds.

Those simulations provide a plausible mechanism for the geometry visible in the microscope images. They do not amount to direct observation of the bridges. The result depends on the accuracy of the force fields, the initial placement and concentration of the ions, the starting geometry of the DNA and the relatively short timescales available to atomistic modeling.

Magnesium and calcium are more relevant to normal biology than nickel, so the researchers modeled them as well. Under one set of solvent protocols intended to approximate physiological ionic conditions, parallel pairing occurred in two of six calcium simulations and one of six magnesium simulations. Similar groove-aligned structures could form, but less readily than in the nickel runs.

When the calcium or magnesium ions began randomly distributed at a concentration of 20 millimolar, none of those simulations produced parallel pairing within the available time. Pairing appeared when ions had been placed at energetically favorable sites or when enough divalent ions were present to neutralize the DNA.

That is both support and warning. The proposed structure is possible with more biologically relevant ions, but whether it appears during a simulation can depend strongly on the starting conditions.

The AFM comparison came with an important wrinkle

The experiment also compared large sets of 339-base-pair DNA fragments prepared with nickel, calcium or magnesium salts. The reported mean fraction of molecules in a paired state was 14 percent for nickel-prepared samples, 11 percent for calcium-prepared samples and 10 percent for magnesium-prepared samples. Given the variation among three replicates, the differences were not statistically significant. Average contact lengths were similar too.

But these were not three exclusive, single-ion imaging environments.

The DNA was initially immobilized and washed in buffer containing 10 millimolar nickel, calcium or magnesium chloride. After the final wash, the researchers added 10 millimolar nickel chloride to every surface to ensure stable AFM imaging, including the surfaces initially prepared with calcium or magnesium.

The comparison can therefore show that contacts persisted after different initial preparation conditions. It cannot cleanly demonstrate what AFM would find during imaging with calcium or magnesium alone.

The study's clearest high-resolution images, the ones in which individual grooves could be distinguished, were obtained in nickel chloride.

Recognition may not begin with recognition

The large-area images produced another revealing complication. DNA molecules did not always meet their partners in the tidy geometry expected from complete sequence matching.

The researchers classified pairings according to where the two fragments touched. Configurations incompatible with full homologous alignment were the majority under every preparation condition: 63 percent for nickel, 57 percent for calcium and 64 percent for magnesium.

Divalent ions, in other words, can help produce local contact without first demanding that two complete DNA sequences match.

That does not eliminate sequence recognition. It suggests a possible two-stage picture. Ions may help nearby helices overcome their general repulsion and form tentative contacts. Sequence-dependent pockets could then make certain arrangements more stable, allowing a favorable alignment to persist or spread.

A separate PNAS study published in June 2026 fits that interpretation. Researchers used tethered DNA nanosensors to compare homologous and nonmatching duplexes in solutions containing magnesium or calcium. They measured a homology-associated recognition free energy of roughly −0.01 kilocalorie per mole for each paired base-pair position, about one percent of typical Watson-Crick base-pairing energy.

That is a weak preference. Spread across a longer stretch of DNA, however, weak contributions can add up. Repeated encounters in a crowded environment could also give a small physical bias opportunities to matter.

Biological evidence has hinted that the helical rhythm itself is relevant. In the fungus Neurospora crassa, matching units just three base pairs long could trigger a genome-defense process when repeated every 11 or 12 base pairs, approximately once per turn of relaxed DNA. The effect persisted without MEI3, the fungus's only member of the RAD51/DMC1 recombinase family.

That result is consistent with some form of helix-sensitive recognition, but it does not prove that the ion bridges in the new simulations produced the effect inside the fungus.

A physical bias, not a replacement for biology

None of these experiments shows naked DNA independently sorting an entire genome.

Inside a cell, DNA is bent, crowded and associated with proteins. Homologous recombination depends on specialized molecular machinery that processes damaged DNA, searches for a suitable template and exchanges strands. The new study does not replace that machinery, nor does it establish that simulated ion bridges control recombination, gene silencing or chromosome organization in living organisms.

The connection to cancer is more distant still. Mistakes in DNA repair and genome organization can contribute to cancer, making the basic physics of DNA pairing worth understanding. But this study did not test cancer cells, identify a treatment or show that disrupting an ionic bridge causes disease.

DNA nanotechnology may offer a more direct testing ground. If researchers can establish reliable rules connecting sequence, ion type and side-by-side alignment, those rules might eventually provide another way to assemble DNA structures. The GTAC pocket identified in the nickel simulations is a lead, not yet a general design principle.

For now, the achievement is more fundamental. Researchers have resolved two negatively charged double helices fitting their grooves together over several turns. Simulations show how a few positive ions might stabilize that otherwise unlikely arrangement.

It is not a complete explanation of how chromosomes find their partners. It is something subtler: evidence that before cellular machinery confirms the right introduction, the shape and charge of DNA may already be influencing who gets close enough to meet.