It doesn’t take much energy to denature a protein. About .4 kiloJoules/amino acid, so that a protein of 100 loses its function (denatures) with an energy input of 40 kiloJoules/Mole or about the energy required to break two measly hydrogen bonds [ Voet and Voet Biochemistry Ed. 3 p. 258 ]. Covalent bonds are a lot stronger, with carbon carbon single bonds and C – H bonds ten times stronger. All you have to do to denature chymotrypsin is pull apart its catalytic triad of histidine at position #57, aspartic acid at #102 and serine at #195. Clearly to get these 3 amino acids together the protein backbone has to turn and twist in space. Separating them doesn’t take much energy.
Amazingly, denature many of them and they spontaneously reform the active structure. Certainly the first such protein studied this way (ribonuclease by Anfinsen) did just that, leading to the idea that the 3 dimensional structure of a protein was determined by linear sequence of its amino acids along the backbone.
Over the decades crystal structure of protein after protein was solved by Xray crystallography, and everyone came to think of proteins as having ‘a’ structure. It was quickly found that there are parts in many proteins that won’t sit still even for crystallography, and it is now estimated [ Proc. Natl. Acad. Sci. vol. 103 pp. 12353 - 12358 '06 ] that 30% of all proteins have stretches of over 30 amino acids that are intrinsically disordered.
Now sight your eye at the alpha carbon of one of the amino acids of a protein, looking toward the carbonyl carbon. There are 3 conformational energy minima the carbonyl can adopt. That’s potentially 3′^99 = 10^48 conformations. This is clearly an overestimate because of self intersection, but still quite large. Yet to be crystallizable the protein must choose just one of them and it must be lower in energy by 2 hydrogen bonds than all the rest.
Now think like a chemist and think about the side chains of the amino acids. The hydrocarbon types (alanine, glycine, valine, leucine, isoleucine and perhaps methionine) can dissolve in each other. Hydrogen bonding is possible between the serine and threonine and any carbonyl on the side chain or any of the amines. Salt bridges are possible between the two acids and 3 of the bases. The list goes on and on. Yet somehow the 195+ amino acids of chymotrypsin spontaneously form this one shape. As a chemist I find this incredibly strange and unlikely. Among the 10^48 conformation of a 100 amino acid protein are there none within 40 kiloJoules of ‘the’ structure? If there are, are the energy barriers so high that it is never found?
We’ve seen this happen so often we’ve gotten used to it, but speaking as a former chemist, I find this behavior incredibly strange. I probably know enough math now to really delve into the physical chemistry of protein folding, but haven’t gotten around to it yet,. But saying that proteins fall down a potential energy funnel seems (to me) like just a fancy way of saying they fold into one shape.
My guess is that an incredibly small fraction of the possibilities in protein space have these properties. There is an experiment which could possibly prove me wrong. See http://luysii.wordpress.com/2010/08/08/a-chemical gedanken-experiment/.
I mean you don’t even have to be a chemist to see what I’m talking about. Back in the day, girls used to wear charm bracelets, with little charms hanging of the chain. Some of the chains attract each other, others have the opposite effect. Make one with 100 charms of 20 different types, throw it into a pail of oil and agitate the pail so it doesn’t sink. Do you think just one shape would result?
I think our biochemical sense of wonder has been dulled by what we’ve found so far. For some thoughts on this see http://luysii.wordpress.com/2009/09/25/are-biochemists-looking-under-the-lampost/. Just this month [ Proc. Natl. Acad. Sci. vol. 107 pp. 17710 - 17715 '10 ] A new player in bone formation was found. It’s oleic acid esterified to the hydroxyl group of serine. How many more things are there like this out there?
This has nothing to do with mutation, or the evolution of protein structure by natural selection. That’s for next time. But if proteins with one or a few structures are as rare as I think them to be, it’s going to be tough to get new proteins with this property from old ones by mutation. Once obtained, natural selection can go to work on them. The problem is getting to them in the first place.