Doctoral Research · École Normale Supérieure

Reading More From a Strand of DNA

Innovative Strategies for DNA Analysis using Supramolecular Assemblies and qPCR Multiplexing — my PhD thesis, rebuilt as three interactive models.

Every DNA test works the same way: apply a stimulus, read a response. My doctoral work asked a simple question — what if we read that response more carefully, or used a different stimulus altogether? Each model below is a working piece of that thesis. Move the controls; the science responds in real time.

The Through-Line

DNA is information — four letters, read by base-pairing. A diagnostic test turns that information into a signal you can measure. Conventionally, one target gets one fluorescent colour and one melting temperature, and that is all the information the test returns.

The thesis followed two strategies for getting more out of the same biology. The first: extract richer information from a conventional assay — this produced high-content PCR, which reads several targets from a single optical channel. The second: use a conventional tool in an unconventional way — this turned the humble molecular beacon into a self-assembling nanostructure and a sharper detector of single-base mutations.

A molecular beacon is the recurring character here, so it is worth meeting now. It is a short single strand of DNA folded into a hairpin: a fluorescent dye on one end, a quencher on the other. Closed, the dye sits beside the quencher and stays dark. When the loop of the hairpin finds its target sequence, the hairpin springs open, the dye separates from the quencher, and the beacon lights up. That open-or-closed switch is the raw material for everything below.

Thesis · Chapter 3

Programmable DNA Self-Assembly

Because base-pairing is predictable, DNA can be designed to fold and assemble into shapes — the principle behind DNA origami, where a long scaffold strand is stapled into nanoscale patterns. My thesis used a smaller-scale version of the same idea. When a molecular beacon captures a target that is shorter than its probe, the stem arms of the hairpin are left dangling — free sticky ends. Those sticky ends pair with the stems of neighbouring beacons, and the beacons chain themselves into a supramolecular network. We imaged that network by atomic force microscopy — to our knowledge, the first time such an assembly had been seen.

Set the sample below and drag the temperature. Watch when the beacons stay apart, when they lock into a network, and when heat melts the structure back down.

DNA target in the sample
A field of molecular beacons. Schematic — beacon and network geometry are stylised to illustrate the mechanism.

What the thesis showed

  • Molecular beacons bound to a short target self-assemble through sticky-end pairing — demonstrated, and imaged by AFM as an interconnected honeycomb-like network.
  • Stem-blocking experiments confirmed the assembly forms through the stem arms: block either stem and the network — and its melt signal — disappears.
  • The assembly is a genuine supramolecular structure, but it is an artefact of the assay (it needs a target shorter than the probe) and AFM could not resolve its fine geometry — both flagged as open questions.

Further reading

Thesis · Chapter 3 · Centerpiece

Molecular Beacons & SNP Detection

Heat a molecular beacon bound to its target and it eventually melts apart. Plot the melting as a curve and its steepest point gives the melting temperature, Tm — a fingerprint of how well probe and target match. A single-nucleotide polymorphism, one wrong base, lowers Tm in a way that is predictable from DNA thermodynamics.

The thesis found a second fingerprint. With short targets the beacons self-assemble (the model above), and that assembly melts at its own low temperature — Tm*. Crucially, a SNP shifts Tm* in a way standard thermodynamics cannot predict — a separate, independent signal. Plot one against the other and mutations that are invisible on Tm alone pull apart on the map.

Place a mutation in the target — choose where, and which base — and watch both fingerprints move.

Target base at that position wild-type: —
The beacon and its target. The mutated base is marked.
Derivative melt curve: a peak at Tm* (assembly) and at Tm (probe–target duplex).
The two-dimensional Tm*–Tm map. Grey points are a reference SNP library; the bright point is your current mutation.

What the thesis showed

  • Tm* is a real, reproducible second melting signal — demonstrated across 5-base and 7-base stems, mutated and deleted targets, even "sloppy" beacons.
  • A SNP shifts Tm predictably (it follows the SantaLucia nearest-neighbour model) but shifts Tm* unpredictably — the effect is steric, not thermodynamic, and depends on where the mutation sits and which base it is.
  • Plotting Tm* against Tm resolves mutations with nearly identical Tm — a proof-of-concept for sharper SNP discrimination. A predictive model for Tm* itself remains future work.

Further reading

Thesis · Chapter 2

hcPCR — One Channel, Many Targets

Conventional qPCR multiplexing is capped by optical crowding: only four or five fluorophore colours fit the visible spectrum, so n detection channels read n targets. High-content PCR (hcPCR) breaks that ceiling. Under fast thermal cycling each probe traces a distinct fluorescence-versus-temperature waveform — different again once its target is found and the probe is cleaved. Several probes then share one colour channel, and a least-squares deconvolution against each probe's known signature separates the targets back out.

Targets present — all three probes share one optical channel
Three stacked probe waveforms sum into the measured single-channel signal (white). Curves are stylised from the thesis data to illustrate the method.

What the thesis showed

  • Fluorophores have characteristic, reproducible temperature-dependent waveforms, and a probe's waveform changes once it is cleaved by the polymerase — demonstrated across a panel of dyes.
  • Two targets were resolved on a single optical channel — including two targets labelled with the same fluorophore — and up to four targets in total were detected as a proof-of-concept.
  • The principle scales to 2n targets per n channels; robust quantitative deconvolution at equal target concentrations is the main open challenge.

Further reading

  • Rajagopal et al. (2019), Significant expansion of real-time PCR multiplexing — amplitude-modulation multiplexing, the closest prior art.
  • SantaLucia & Hicks (2004), The thermodynamics of DNA structural motifs, Annu. Rev. Biophys. doi.org/10.1146/annurev.biophys.32.110601.141800

Where It Goes Next

Temperature is one stimulus. The later chapters of the thesis ask what happens when you swap it for another. A photoswitchable DNA-binding molecule, AzoDiGua, lets light — not heat — stabilise or release a DNA duplex, which means a molecular beacon can be controlled with a torch instead of a thermocycler. Combined with the supramolecular assembly above, that opens light-driven control of DNA nanostructures and, eventually, an isothermal route to both amplifying and detecting DNA in one step.

Those chapters are described in full in the thesis itself.

Read the full thesis on HAL