Catalytically active proteins come in many varieties, and you can classify them in many ways. When you look closely at their structures, one such scheme might be the “solid” ones versus the “delicately balanced” ones. In the first category would be things like carbonic anhydrase or acetylcholinesterase: they do their jobs more or less constantly and at their full (terrifying) catalytic speed. Their regulation is comparatively light; they’re like power tools whose “on” switch has been super-glued down. This is a big reason why you see carbonic anhydrase used as a model system in so many other applications – it’s cheap and easily available, for sure, but it also doesn’t need a lot of tricky conditions and cofactors. It can be treated pretty roughly and still function under a variety of conditions.
The other category are more delicate. They’re balanced on one or more knife-edges of activity, with regulatory factors that can easily throw them on or off. Often there are different subunits to these proteins, with binding sites for allosteric molecules that can modulate the active site’s behavior. And if that active site can be changed by something binding in a completely different part of the protein, that means that there must be a way for the fact of that binding to be transferred across the structure, a series of shifts and slippages that moves things in and out of their positions.
This new paper has some details on how that happens. The authors (a team from Göttingen, the Georg-August University and the Max Planck Institute there) are studying “low-barrier hydrogen bonds” in protein structures, which have been hypothesized as just the sort of easily-thrown structural elements needed. The idea of a “proton wire” has been proposed in some enzyme structures, a series of domino-like changes in the position of shared protons between various heteroatoms in the structure. This would be similar to the situation in bulk water, with its networks of hydrogen bonds stretching between long chains of individual molecules. It’s an appealing mechanism, but observing it in action is not so easy; tracking down the fine position of individual H atoms (particularly the slippery ones) in a protein structure is beyond what most X-ray crystallographic structures can tell you.
But this paper comes in with data for the transketolase enzyme at less than 1A resolution, which is way up on the sharp end of protein crystallography, for sure. The protein had been thought to have such a protein wire involving six glutamate residues (three on each subunit) and several bridging water molecules, allowing binding to one subunit’s active site to affect the activity of the other (over a distance of some 25A). The enzyme uses thiamine diphosphate (ThDP) as a cofactor, as do many others (pyruvate dehydrogenase, for example), and there’s a particular glutamate (E366′) that protonates this cofactor to prime it for activity. It’s also interacting with another nearby glutamate (E160), which is bridged via a water molecule to glutamate E165, which is involved via another water network with a glutamine residue (Q367) which is at the dimer interface.
E366′ and E160 are rather close together in the structure, and there’s a spot of electron density exactly halfway between them (see at right) which seems to be the proton involved in the low-barrier hydrogen bond. If the E160 glutamate is mutated to glutamine, though (E160Q), then the key protonation of the ThDP cofactor can still occur, but the proton-wire communication network is cut off. There’s still a hydrogen bond between that residue and the E366′ one, but it’s an ordinary one from the NH2 of the glutamine with a more fixed position (as seen in the X-ray structure of the mutant form). The mutant also has a five-fold lower Kcat constant, and stopped-flow kinetic analysis shows that the substrate-cofactor interaction is impaired in the mutant form and that it no longer has the positive cooperativity seen for substrate binding in the wild-type enzyme.
The two sites in the wild-type enzyme appear to be communicating with each other almost a thousand times faster than the Kcat value, and this is apparently happening by changes in the protonation states of the residues along the way. The low-barrier hydrogen bond between E366′ and E160 is the key link in this chain; it’s like the bubble in a carpenter’s spirit level, and can shift position depending on what’s happening out at either end of the network.
The team found a very similar effect in a proton-wire setup in a completely different enzyme (bacterial pyruvate oxidase) that also uses ThDP as a cofactor, so this is likely to be a general mechanism. The catalytic effects are not huge, but they’re real, and evolution seems to have found them to be worthwhile for a boost in enzyme activity. The hunt will now be for other examples of this effect in new structures and new bonding networks. You’d expect the glutamate-glutamate pairing to be used again, since having two identical residues participating sets up that low-energy-barrier balance well, but there could be other matched pairs as well, starting with Asp-Asp and working along from there. Tiny advantages like these add up across a large protein structure, across a metabolic network of enzymes, and across a billion years of time.