Chain Length Sends DNA Tiles Into Stars or Spheres

Hydrophobic chain length on DNA tiles decides whether they assemble into stars or spheres. Yihao Wu, Muhammad Ghufran Rafique, Christopher Saab and Hanadi Sleiman, working in McGill University’s Department of Chemistry, reported that result on 11 September 2026 in a study of the hierarchical assembly and self-sorting of DNA tiles.

The oily tails do more than decorate a particle. They pick which weak force fires first, and that order is what later shows up as serum-stable spheres, mixed-pot sorting, and a protocol that needs a laboratory heat ramp.

The Energetic Threshold Inside Each Tile

Sleiman is a chemistry professor at McGill and Canada Research Chair in DNA Nanoscience. Her group’s stated aim is to take DNA out of its biological context and build two- and three-dimensional materials from it, using synthetic add-ons as well as base pairing. Wu is first author. Rafique, listed with McGill on the paper, is now in biochemistry at the University of Cambridge. Natural Sciences and Engineering Research Council of Canada and the Canada Foundation for Innovation funded the work, which was received on 6 February 2026 and accepted on 27 August 2026.

THE TILE, IN FOUR NUMBERS

• Strand count: Each double-crossover tile is five distinct oligonucleotides, with two of those strands carrying C12 tails at their 5-prime ends.
• Chain lengths: The team varied hexaethylene (C12) units through 4 or 6 repeats on the short side and 12 or 16 on the long side.
• Heat cycle: Samples went to 95 °C for 5 minutes, then cooled from 85 °C to 25 °C at 1 °C per minute, a 60-minute ramp.
• Body-temperature control: The same mixes were also held at 37 °C with no heat ramp.

Double-crossover tiles are two parallel helices joined by two four-way junctions. The McGill design used an antiparallel motif with an even number of half-turns between those junctions, a layout that stays planar and relatively stiff. Solid-phase synthesis and phosphoramidite chemistry put a defined number of C12 units on the two modified strands, so hydrophobic strength was a counted length rather than a messy polymer blend.

Standard DNA origami still leans on a long scaffold plus hundreds of short staple strands. Sleiman’s review of sequence-defined DNA polymers argues for a DNA-minimal route: fewer unique DNA sequences to make complex structures by adding forces DNA does not already have. This paper puts that idea on a classic tile. The extra handle is hydrophobicity, and the claim is that its strength sets an energetic threshold against hybridization.

Short C12 Chains Let Base Pairing Win

When the tails were short, DNA pairing won the race. The tiles formed as ordinary double-crossover units first. Only later did the oily patches stick, and the objects that grew from those tiles were three-arm or four-arm nanostars.

HOW C12 LENGTH REDIRECTS ASSEMBLY

Atomic force microscopy was the main imaging tool. On the four-unit tiles, the stars also grew into extended networks. The authors propose that π-π stacking at blunt tile ends, not extra sticky sequences, stitched those nets together. That is another interaction DNA already knows, pressed into service once the tiles exist as rigid plates.

The same logic was copied onto three-point star tiles that carried two alkyl chains on one arm. Nile Red, a dye that brightens in oily pockets, tracked when hydrophobic cores appeared on the way down the temperature ramp. Ultraviolet absorbance at 260 nm gave melting curves for the DNA parts. Together those traces placed pairing above hydrophobic collapse for the short tails, which is why the stars are tiled objects first and clustered objects second.

Twelve and Sixteen Units Flip the Order

Lengthen the same tails to twelve or sixteen C12 units and the temperatures swap. Hydrophobic association now happens while the DNA is still unpaired. The modified strands collapse into micelles, packing their DNA on the outside as spherical nucleic acids. Further cooling then lets those surface strands pair, crosslinking the particles to one another and to themselves.

Spherical nucleic acids are DNA strands packed densely around a core. Chad Mirkin’s group at Northwestern University introduced them in 1996 on gold particles, and later work showed that architecture can enter cells without transfection agents by engaging scavenger receptors. The McGill particles are a related shape built without a pre-made gold seed. The oily block is the core. DNA hybridization is the glue that then locks the shell.

By varying hydrophobic strength and applying thermal annealing, we define an energetic threshold that dictates the hierarchical order of hydrophobic assembly versus DNA hybridization.

Yihao Wu, Muhammad Ghufran Rafique, Christopher Saab and Hanadi Sleiman, Department of Chemistry, McGill University, in Angewandte Chemie International Edition

Cyanine-3-labeled indicator strands were added to already-built particles and left for 12 hours. Native agarose gels showed that the crosslinked spheres refused those extra strands, a sign that the surface was tightly paired and slow to exchange. Weak hydrophobicity never took that path, because pairing had already spent the sticky sequences inside discrete tiles.

Human Serum Albumin Breaks the Loose Spheres

Blood is a hostile bath for naked DNA. Nucleases cut strands. Human serum albumin, the most abundant protein in plasma, binds oily patches and can unpack a hydrophobic assembly. The McGill tests put those two pressures on the particles in the tube, not in an animal.

WHAT THE SERUM TUBE SHOWED

• Nuclease mix: Crosslinked spherical nucleic acids held together in medium with 10% fetal bovine serum.
• Albumin: Excess human serum albumin pulled non-crosslinked spheres apart and left the crosslinked ones intact, as read by size-exclusion chromatography.
• Strand exchange: After 12 hours, the Cyanine-3 indicators did not join the crosslinked particles.
• Left untested: Cargo loading, release, uptake into cells, and any in vivo run.

Crosslinking is the difference that survived those assays. A sphere held only by a greasy core is a guest albumin already knows how to unbind. A sphere whose DNA shell has paired across the surface is a different object, and that object is the one that stayed in one piece. The biomedical reading writes itself, and the paper does not overclaim it. Nothing was loaded. Nothing was released. No cell saw these particles.

Why the Heat Ramp Matters for Self-Sorting

The sharpest result is not a pretty AFM frame. It is what happened when the same DNA sequences, carrying different C12 lengths, were mixed in one pot. After the 95 °C heat and the slow cool, the mix split into distinct nanostar and spherical populations. The authors call that narcissistic self-sorting: each chain length preferred partners with the same hydrophobic length rather than a mixed hybrid. Base pairing, by contrast, is social. Complements want each other even when they come from different molecules.

Hold that same mixed sample at 37 °C with no ramp and the split disappears. One population forms. Thermal history is part of the code. The stars-versus-spheres switch is easy to sell as a chain-length trick, and that is the line that travelled with the paper. The sorting, though, is a pathway fact. It needs the two forces to fire in sequence on the way down a temperature gradient a bloodstream does not provide.

That does not make the long-tail spheres useless at body temperature. It does mean the one-pot sorting, the clean separation of two morphologies from identical DNA, is a laboratory annealing result. Anyone who wants both shapes from one mixture has to heat the sample first. Anyone who wants only the crosslinked spheres still has to get hydrophobic collapse to beat pairing, which on this design is a long-tail, high-temperature event.

A Trigger Strand Can Undo the Crosslinks

The crosslinked spheres were not welded shut forever. The team left single-stranded toehold domains on the design. A trigger oligonucleotide bound that toehold and displaced a backbone strand used in surface crosslinking, shifting the particle toward a non-crosslinked state. In the abstract’s wording, those crosslinked spherical nucleic acids remain fully reversible via strand-displacement-triggered decrosslinking. Release of a drug was not measured. The handle is there for a later experiment that would ask whether a disease marker, or an aptamer, could pull the same strand.

WHAT WE KNOW

• The switch: Short C12 tails yield nanostars after pairing; long tails yield crosslinked spheres after a hydrophobic collapse.
• The sort: Identical DNA with different tail lengths separates into those two forms when annealed together.
• The serum gap: Crosslinking, not the oily core alone, is what survived albumin and the 10% serum nuclease mix.

WHAT IS UNCONFIRMED

• Delivery: No cargo, no release curve, no cell uptake, no animal data.
• Body-temperature sorting: The mixed-length split was not seen at 37 °C.
• Targeting: Aptamers in the toehold and backbone are a proposed next step, not a result in this paper.

Sleiman’s lab already talks about DNA structures for cancer therapy and about expanding the DNA toolbox with polymers, metals and lipids. This paper is a design rule from that program, not a carrier ready for a clinic. Chain length now sits beside sequence as a bit you can set on a synthesizer. Set it short and DNA writes first. Set it long and oil writes first. The shapes follow that order, and so does everything the tube later proves, or fails to prove, about life in serum.