Tricking the Light Fantastic: How to Use Photons to See Something Smaller Than a Photon

A while back I compared visualizing something on the nanometer scale (like, for instance, a DNA strand) with optical microscopy to trying to see what a strand of yarn looked like by throwing beach balls at it. But because engineers can get around almost any rule, they’ve found a way to do just that. Read on for more about how it works, why it’s cool, and what it means.

This article proposes a way to simultaneously dye numerous components of a cell, and pinpoint their location to within 10 nanometers, using a combination of proteins, DNA, and fluorescent molecules (summary sans paywall here). Reading it sent me to this article, which uses a technique called Fluorescence Imaging with One-Nanometer Accuracy to see a protein move a couple of nanometers using a fluorescent dye and traditional fluorescent microscopy. A summary of FIONA can be found here, for those of you who are technically inclined.

A Distribution is More Accurate than any Data Point

So here’s the big secret of FIONA: you can’t tell where a molecule is with more accuracy than about 200 nanometers if you consider only one photon, because photons have a wavelength around the order of 200 nanometers. But if, instead, you capture a large number of individual photons, they’ll tend to form a cluster, a little “mountain” of intensity. That mountain is centered at the “true” position of the fluorescent dye. And if you use the proper mathematics to map out that mountain and find its center, you can determine the position of a fluorescent molecule to stunning accuracy — more accurate than any individual photon would allow you to be.

Basically, with the yarn/beach ball analogy, you consider the cluster of beach-balls that get bounced back by the strand of yarn. Something that skates just barely by the strand will pass through, something that hits it on edge will scatter and might be picked up, something that hits it dead on will bounce straight back and be sure to be picked up. So if you consider the distribution of signals, rather than any individual signal, and you pull out just the center of that distribution, you can probably find your piece of yarn.

The reason this is an engineering feat isn’t just because it’s a new way to analyse images. It’s a crucial part of this technique that we be able to capture the signal from exactly one molecule, or else the mountain gets lost in a mountain range.

DNA Gives Specificity

This requirement to see discrete signals from individual molecules is why the researchers in the first paper turned to DNA. DNA is a great tag — short sequences of DNA will bind to complementary or near-complementary sequences and not others. You can fine-tune how “sticky” you want your tag to be by fine-tuning how exactly the sequence matches: add a few mismatches in your sequence and you get a molecule that will stick for a moment and then wash away. So while there may be a bunch of tethers around waiting to be fluorescent, you only see fluorescence from one of them at a time.

Plus, with a relatively short tag (say, 10 base pairs), you have 10^4 different possibilities for matching or mismatching sequence. Which means you can put in a huge number of tether sequences tethered to a huge number of different things, image with different sticky sequences sequentially, and get lots of images of the same sample — identifying the location of a huge number of different cellular or molecular components. The researchers in the first study I mentioned call this ten-color imaging, even though they only really use one color, with ten different tags, used sequentially.

The short answer, though, is that with enough creativity and knowledge of biology and physics, you can, in fact, look at a strand of yarn using a bunch of beach balls.