As Arthur Kornberg never tired of pointing out, cells are gels. It’s too easy for biologists and chemists to imagine cells as sort of like liquid-filled plastic bags – and while that’s an OK picture as far as it goes, it tends to make you picture the cytoplasm as a lot more dilute than it really is. Something about the consistency of gumbo is more like it – maybe even the first batch of gumbo I ever made, to the whole pot of which I cluelessly added a good strong dose of filé powder, causing it all to set up into something could be nearly be sliced like a meat loaf.
The point is, there’s nowhere near as much bulk solvent in a cell as there is in anything you’d willingly work with in a lab. And that means that a lot of things behave differently than you expect – proteins, for example. Spherical proteins are the easiest to deal with, since they’re probably going to stay that way no matter what happens to them, short of outright denaturation. (Spheres are good choices for extreme environments). But most proteins aren’t spherical, and their shapes are extremely important – how well do we understand their behavior under real-world conditions?
It’s not an easy question to answer. The standard ways to study protein structures are (1) X-ray crystallography, a rather artificial state of affairs, since proteins are rarely found in the crystalline state in vivo, (2) NMR spectra, which can be very informative but are usually taken from purified proteins in a clean buffer solution, and (3) molecular modeling. That last technique’s relation to reality depends (among other things) on the patience, skill, and computational resources of the people using it. But just making sure that you’re modeling a protein’s structure in the presence of water molecules, rather than in some sort of ideal mathematical vacuum, can be enough of a challenge. Including a stew of other proteins right around the one of interest just isn’t feasible, even if we knew which ones to put in.
There’s a recent open-access paper in PNAS that does a good job calling attention to this problem. The authors studied a roughly football-shaped protein, VIsE, which comes from Borrelia burgdorferi, the Lyme disease organism. Diagnostic tests for Lyme recognize one stretch of this protein - but the odd thing is, that region appears to be buried inside the hydrophobic core of its structure, which makes you wonder how anything could recognize it at all.
The team studied the protein under different levels of denaturing agents and non-denaturing additives, and found several different structures seem to present themselves under different conditions. To their evident surprise, this even agreed with their molecular modeling of the process. Both the speed of protein folding and the courses the folding takes are altered - and under cellular levels of crowding, it turns out the protein may well adopt a spherical state that exposes much more of that antigen sequence. That's shown in the illustration, where C is the structure that's suggested for real-world conditions, as opposed to A, and the antigen sequence is shown in green. (The Y axis relates to the volume fraction taken up by various crowding agents).
Drug discovery people have always been wary of the structures of membrane-bound proteins, because we don't understand much about them. We should be wary of the structures of the free-swimming ones, too - after all, they're certainly wary of us.