A slightly different wrinkle on the perennial "do we model unresolved sidechains" debate, I guess. I would argue that in the case of mutations, tags etc. those are things you know were there before you started firing x-rays at your sample. In the case of the sidechain decarboxylation we know it's an artifact of the data collection method, and we correct for known artifacts in other contexts all the time.
Tristan Croll Research Fellow Cambridge Institute for Medical Research University of Cambridge CB2 0XY > On 9 May 2017, at 17:49, Ian Tickle <ianj...@gmail.com> wrote: > > > Hi Tristan > > I'm not so sure. The co-ordinates are the result of the experiment. How > other people choose to interpret those results is their affair. Taking it to > its logical conclusion suppose that we 'damage' the protein by > mutating/deleting some residues or adding tags purely for the purpose of > getting it to crystallise, do we report the structure of the protein as it is > in the crystal, or do we report what it would have been if we hadn't messed > with it ? The choice is clear in that situation. > > Cheers > > -- Ian > >> On 9 May 2017 at 16:45, Tristan Croll <ti...@cam.ac.uk> wrote: >> Hmm... this is a bit of a philosophical pickle in my mind. Do we want to >> model the structure as what it looks like after radiation damage has had its >> way with it, or what it must have looked like *before* the damage? I can see >> arguments both ways (and can sympathise with the former if you want to make >> radiation damage a subject of your manuscript), but this is going to lead to >> headaches for people who want to make use of the resulting coordinates to >> study the actual biology of your protein. Personally, I'd strongly prefer >> the latter approach. >> >> Tristan >> >> >>> On 2017-05-09 16:06, Edward A. Berry wrote: >>>> On 05/09/2017 06:18 AM, Ian Tickle wrote: >>>> We have seen almost identical density to Ed's for GLU side-chains, with >>>> what looks like a linear molecule (yes exactly the size of CO2!) where the >>>> carboxylate group would be and absolutely no density for the CG-CD bond. >>>> So it's indeed very tempting to say that the CO2 is still there, and >>>> presumably making the same H bonds that the carboxylate was making to hold >>>> it there. It would not be hydrated to carbonic acid, according to >>>> https://en.wikipedia.org/wiki/Carbonic_acid : "The hydration >>>> <https://en.wikipedia.org/wiki/Hydrate> equilibrium constant >>>> <https://en.wikipedia.org/wiki/Equilibrium_constant> at 25 °C is called >>>> K_h , which in the case of carbonic acid is [H_2 CO_3 ]/[CO_2 ] ≈ >>>> 1.7×10^−3 in pure water^[5] >>>> <https://en.wikipedia.org/wiki/Carbonic_acid#cite_note-HS-5> and ≈ >>>> 1.2×10^−3 in seawater <https://en.wikipedia.org/wiki/Seawater>.^[6] >>>> <https://en.wikipedia.org/wiki/Carbonic_acid#cite_note-SB-6> Hence, the >>>> majority of the carbon dioxide is not converted into carbo >>> n >>> ic >>>> acid, remaining as CO_2 molecules.". >>> >>> It looks like this ignores subsequent ionization of H2CO3 which would >>> be quite spontaneous at neutral pH. However the Wikipedia article >>> also indicates the equilibrium is quite slow (which makes sense- >>> otherwise why would carbonic anhydrase exist?) and it would be a great >>> deal slower in vitreous ice at 100 K. Anyway, I had reached the same >>> conclusion and have modeled a number of the troublesome glutamates as >>> decarboxylated with CO2 hovering above. There is a problem that the >>> remaining CG tends to push the CO2 a little out of the density in some >>> cases, but not a severe clash and it may work itself out with further >>> refinement or manual assistance. >>> eab >
