A Marker That Never Burns out, and a Microscope That Can Say Where One Molecule Sits to 0.6 Angstrom

The whole trick of super-resolution microscopy is a kind of accounting. You cannot see a fluorescent molecule; you can only see the smeared blob of light it throws onto a camera, and that blob is hundreds of nanometres across no matter how good your lenses are. But the centre of a blob can be estimated, and the estimate gets better the more photons you collect. Catch the same molecule flashing twice and your guess sharpens. Catch it a thousand times and it sharpens again.

Which is why photobleaching is not a nuisance in this field. It is the ceiling.

Organic dyes, the standard markers, are consumed by the light that makes them glow. A given molecule will blink a few hundred or a few thousand times and then go dark forever, and whatever precision you had accumulated by then is what you get. Worse, coaxing a dye to blink at all usually means bathing the sample in a chemical buffer built for the purpose, and imaging in several colours means a separate laser for each, fired in sequence.

A team at MIT and the Broad Institute has now removed the ceiling by changing the marker. In a paper published in Nature Nanotechnology on July 27, Spontaneous and indefinite blinking in upconverting nanoparticles for ångström-precision multicolour super-resolution imaging, eight authors led by Saptarshi Mandal, in the lab of Chunte Sam Peng, describe particles roughly 10 nanometres across that blink by themselves, do not stop, and need no special buffer.

Upconversion, and why it changes the accounting

Upconverting nanoparticles run the usual fluorescence bargain backwards. A dye absorbs a high-energy photon and emits a lower-energy one. These crystals, doped with rare-earth ions in a carefully controlled core-shell arrangement, absorb several low-energy infrared photons and emit one photon of higher energy: infrared in, visible out.

Two things follow. Infrared light is far gentler on biological samples than the blue and ultraviolet that conventional dyes demand, and it penetrates deeper. And because the emission colour depends on the crystal's composition rather than on the excitation wavelength, particles of different colours can all be driven by the same beam. MIT's own account of the work notes that U-STORM, as the team calls the platform, "can operate with just one near-infrared laser" exciting several emission colours at once, with no sequential rounds of imaging.

The blinking is the part that had to be engineered. Compositional tuning of the crystals produced particles that switch on and off spontaneously, and, critically, keep doing it. Nothing is being consumed.

So the accounting changes. If a marker never dies, the number of flashes you can average is limited by your patience and by how still the sample holds, not by the marker's lifespan. The team took that to its logical end and collected more than 88,000 localisation events from a single particle. Their reported precision on that particle: 0.6 angstrom. The paper's figure for single-molecule imaging on cell membranes is 0.62 angstrom.

An angstrom is a tenth of a nanometre. A single covalent bond between two carbon atoms is about 1.54 angstrom long, so 0.6 angstrom is less than half a bond length.

What that number means, and what it does not

Here the press framing and the physics part company, and it is worth being exact.

The 0.6 angstrom is a localisation precision: how tightly repeated estimates of one marker's position cluster around their own average. It is not a resolution: how far apart two distinct objects must be before a microscope can report them as two things rather than one.

The distinction is not pedantry, and the arithmetic shows why. Localisation precision improves as the square root of the number of events. Start with a per-flash uncertainty of about 10 nanometres, average 88,000 flashes, and you arrive at roughly 0.34 angstrom. In other words, 0.6 angstrom is not an anomalous number. It is close to what the shot-noise floor predicts once you remove the constraint that killed the marker.

But that gain is bought by watching one unmoving particle for a very long time. Two different molecules in a living sample will not sit still for tens of thousands of frames, will not stop diffusing, and will not hold their separation while you accumulate statistics. No pair of objects 0.6 angstrom apart is being told apart here, and the paper does not claim otherwise.

The comparison to conventional markers needs the same care. The team describes its precision as three orders of magnitude beyond the nanometre-scale limits of standard fluorescent dyes, a figure carried in the MIT release and widely repeated. That is the team's own comparison against a nanometre reference. Working dye-based single-molecule localisation microscopy typically lands somewhere between 5 and 20 nanometres in practice, against which 0.6 angstrom is a gain of roughly 100 to 300 times. Still a large number. Not a thousand.

What it is actually for

The biological demonstration was modest and specific: mapping epidermal growth factor receptor dimers and multimers under physiological conditions, without the specialised imaging buffers that conventional blinking requires. Receptor clustering is a real question in cell signalling, and how many copies of a receptor sit together, and how far apart, is exactly the sort of measurement that sub-nanometre localisation can improve.

Whether the approach generalises is future work. Nanoparticles are large compared with dye molecules, and 10 nanometres of label attached to a protein you are trying to locate to a fraction of an angstrom is an obvious tension the field will have to argue about. Labelling chemistry, particle uniformity and how well the technique behaves inside a cell rather than on a membrane are all open.

In this experiment, the number that set the precision was no longer how long the marker survived. It was how many frames the team chose to record.