Skip to Content

Empty As Can Be

OK, drugs generally bind to some sort of cavity in a protein. So what’s in that cavity when the drug isn’t there? Well, sometimes it’s the substance that the drug is trying to mimic or block, the body’s own ligand doing what it’s supposed to be doing. But what about when that isn’t occupying the space – what is?
A moment’s thought, and most chemists and biologists will say “water”. That’s mostly true, although it can give a false impression. When you get X-ray crystal structures of enzymes, there’s always water hanging around the protein. But at this scale, any thoughts of bulk water as we know it are extremely misleading. Those are individual water molecules down there, a very different thing.
There seem to be several different sorts of them, for one thing. Some of those waters are essential to the structure of the protein itself – they form hydrogen bonds between key residues of its backbone, and you mess with them at your peril. Others are adventitious, showing up in your X-ray structure in the same way that pedestrians show up in a snapshot of a building’s lobby. (That’s a good metaphor, if I do say so myself, but to work that first set of water molecules into it, you’d have to imagine people stuck against the walls with their arms spread, helping to hold up the building).
And in between those two categories are waters that can interact with both the protein and your drug candidate. They can form bridges between them, or they can be kicked out so that your drug interacts directly. Which is better? Unfortunately, it’s hard to generalize. There are potent compounds that sit in a web of water molecules, and there are others that cozy right up to the protein at every turn.
But there’s one oddity that just came out in the literature. This one’s weird enough to deserve its own paper: the protein beta-lactoglobulin appears to have a large binding site that’s completely empty of water molecules. It’s a site for large lipids to bind, so it makes sense that it would be a greasy environment that wouldn’t be friendly to a lot of water, but completely empty? That’s a first, as far as I know. When you think about it, that’s quite weird: inside that protein is a small zone that’s a harder vacuum than anything even seen in the lab: there’s nothing there at all. It’s a small bit of interstellar space, sitting inside a protein from cow blood. Nature abhors a vacuum, but apparently not this one.

14 comments on “Empty As Can Be”

  1. dave says:

    From a thermodynamic perspective you would expect the protein to undergo transitions to expose this site in more lipophilic environments. But what might be most interesting is the amino acid composition around the site. Those residues must foster a “UN peace-keeping” team between the water adjacent to them, and the waters that gang up near the site. Had to keep the analogies going even if mine wasn’t very good:)

  2. such.ire says:

    Is a “vacuum” inside a protein really that unexpected? After all, there are plenty of crystals with gaps and voids. It’s not like there’s a law of nature dictating that all spaces must be filled with something; vacuum is just a statistical phenomenon, nothing more.

  3. Matt says:

    I haven’t had a chance to go through the publication, yet. However, one possibility to consider is that water molecules may actually reside in this pocket, but may be so disordered that they would never appear in the solved crystal structure. A very large, hydrophobic pocket would lend itself to having a large number of disordered water molecules, in my mind atleast.

  4. retread says:

    An intriguing paper all right. So much for the hydrophobic collapse theory of protein tertiary structure. My reading of their figure #1 shows at least half of the amino acids lining the cavity are polar (rather than polar or lipophilic). 17-0 and 2-H magnetic relaxation dispersion were used as well to rule out very mobile water inside the cavity which would escape detection on Xray crystallography. In some way the beta strands of the beta barrel (common to these lipid binding proteins) is holding the cavity open. Beta barrels are used to form pores in the outer membrane of gram negative bacteria.

  5. reckon says:

    I’ll bet that in solution there’s something in there. Even if that something is the proteins jostling around for space in the void until the ligand binds!! an solution NMR structure might be interesting as well…

  6. Shane says:

    Until we get a better handle on what water is and how it behaves computational chemistry will be a joke. Derek- any chance of a follow up piece on research from a few years back showing oil and water mix when all the dissolved gasses are removed? A group at ANU in Australia did the early work. Also Ive seen some discussion that the hydrogen bonding of water involves some weird quantum effects…not sure of the source of that though..

  7. Very ineteresting. If such large cavity can be completely empty even in the “frozen” crystal, that would imply that significant empty gaps should occur in hydrohopic pockets quite frequently during the dynamic movement of the proteins. That in turn would have significant implications for scoring, i.e. binding energy estimation with regards to estimating de-solvation energy for the binding site.
    ZZ

  8. Morten says:

    Cool – thanks Derek. We’re working on a structure where we see something similar. It gives some weird negative density in the difference map. But I do believe our’s contains very disordered water (it’s a hydrophobic pocket with some polar functional groups at the very bottom). Personally I think it has something to do interaction partner binding and some kind of rearrangement of the protein.
    Oh… just checked the abstract properly (instead of just the title). Looks very convincing but not like what I’m looking at =]. It reminds me of a talk in Strasbourg… I think it was John Ladbury (might have been Gerhard Klebe who talked right after). Something about modeling the water at a hydrophobic surface as a gas/vacuum. It was hard to evaluate whether it was brilliant or what but it looked like a better description than what “they” teach at (most) universities.

  9. Retread says:

    Remember the old high school demonstration of air pressure, where a metal can collapsed when the air was sucked out? Probably the same thing happens to the protein in solution (e.g. in vivo). I’m just starting to (re)read P-chem after a gap of 46 years, but I do recall that the velocity of a water molecule at room temperature is quite high (because there is little other outlet for total energy), the mean free path of water in solution isn’t very high, etc. etc. So the protein is getting clobbered from all sides many times/second like the hapless can in the high school experiment. Does anyone know the actual numbers? Forgive my ignorance.

  10. burt says:

    “Remember the old high school demonstration of air pressure, where a metal can collapsed when the air was sucked out?”
    The physics of scaling suggests to me that a tiny evacuated space would be more stable than a big evacuated space.
    Think– a bridge. A small span versus a big span with the same proportions.

  11. Nick K says:

    I loved the analogy of people in a lobby stuck to the walls, holding the building up. Sounds like Edinburgh on a Saturday night.

  12. rosko says:

    “My reading of their figure #1 shows at least half of the amino acids lining the cavity are polar (rather than polar or lipophilic).”
    This is not true–ALL of the amino acids in the channel are nonpolar. There are a few lysines and asparagines right outside the “mouth” of the cavity, which is quite solvent-exposed anyway.
    There are several reasons I thought of why it makes sense there is no water in the cavity, whereas there is in carbon nanotubes:
    1) Locating a chain of water molecules inside the cavity would reduce the entropy of the side chains around it, assuming there exist fluctuations that change the shape of the cavity. This is not true in rigid nanotubes.
    2) Water molecules in the nanotube have the aromatic quadrupole moment to stabilize them (think OH-benzene or NH-benzene interactions).
    3) A nanotube is open on both ends, whereas the cavity is open only on one end. This means the chain of waters is broken at one end, and also it reduces the translational entropy of the chain (by preventing “flow” along the long axis).
    It seems the authors already thought of 1) and 2) as well.

  13. Wavefunction says:

    Yes, it’s not that unusual to find a “vacuum” inside a protein active site. A good example is the Cox-2 cavity.
    Post:
    http://ashutoshchemist.blogspot.com/2007/02/nature-adores-vacuum.html

  14. Retread says:

    Rosco:
    Thanks — I should have read the caption of figure 1 more carefully. However, the point stands. The authors color coded ATOMs not amino acids for polarity vs. nonpolarity. Red stands for a polar atom, brown stands for nonpolar — the polar/nonpolar proportion still looks about 50/50 to me. A ligand probably cares only for the surface (e.g. the atoms) of the cavity into which it fits, not the larger structures (amino acids) to which the atoms are attached.
    It would be interesting to see if the cavity persists in solution. Could NMR tell us this? I don’t know enough about it to say.

Comments are closed.