A droplet the size of a pea hangs in the air, held up by inaudible sound. Inside it, the protein ferritin, the very one that stores iron in our bodies, begins its dance. The water leaves, the dancers get crowded, and they line up in neat rows: that is how a crystal is born. But this dance has a conductor: polyethylene glycol, a long chain of repeating units. Its length decides exactly how the proteins come together and how long their formation holds.
The experiment at the Swedish synchrotron MAX IV was like filming a dance with two cameras. A droplet with ferritin and the polymer was suspended in an acoustic levitator running at a frequency of 100 kHz, the same as a mosquito-repelling ultrasound device. Every 0.1 seconds an X-ray beam passed through the droplet, and from its scattering one could tell how orderly the proteins had arranged themselves relative to one another. While there was plenty of water, the proteins moved chaotically, like a crowd in a square. When the concentration rose to 400-500 mg/ml, they suddenly froze into strict order: the characteristic signals of a crystal lattice appeared.
But from there the storylines diverged. With a polymer of long chains (6000 units on average), crystals flashed into existence quickly, but as evaporation continued their order collapsed. It seemed the polymer itself had started to crystallize and was getting in the protein's way, yet a check showed that the destruction of the crystals had nothing to do with the polymer. The proteins were attracting each other even before they lined up, and this attraction ultimately played a cruel trick on them: too tight a packing turned out to be fragile.
With short chains (1000 units) everything was different. Crystallization started later, when the droplet had almost stopped shrinking, but the crystals stayed stable right to the end of the experiment. The proteins repelled each other and lined up only when they literally had nowhere else to go, like passengers in a packed subway car who are forced to stand straight because otherwise they will not fit. The distance between proteins in such a crystal turned out to be larger than with the long polymer.
This difference is like two ways of organizing a party. Long polymer chains act like a pushy entertainer who grabs guests by the hand and quickly arranges them in a circle, but the moment he lets go, the circle falls apart. Short chains are more like rules of etiquette: they do not force anyone, they create the conditions under which the guests line up neatly on their own, and that line holds even when things get crowded. Before this work it was unclear why the same polymer with different chain lengths gives such different results. Now it is clear: short chains probably coat the protein surface and hold a little water around it, keeping the crystal from drying out and cracking.
In practice this means that by choosing a shorter polymer you can get protein crystals that will not crumble when taken out of solution. Such crystals are needed to determine protein structures by X-ray methods: it is like photographing a protein in the finest detail. And if today a crystal cracks during drying in the lab, tomorrow it may be enough to switch the polymer so that it survives both the trip to the synchrotron and the experiment itself. The protein's dance in the floating droplet goes on, but now we have the score.
A droplet with a volume of 10 microliters evaporates in the levitator in roughly 20 minutes, and over that time the protein concentration climbs from 50 to 500 mg/ml, that is, by a factor of 10.
In Stanislaw Lem's novel Solaris, the ocean creates structures that cannot be understood by earthly methods; here, by contrast, we see how a simple polymer governs the birth and death of protein crystals in a droplet suspended in midair.