Marc,
Your reasoning about NMR at low pH and the fact to have a non-charged
terminal glutamic acid makes sense. From what I learned, when you have
an idea you follow it, see if this fits with the story you want to
tell and demonstrate it works. Moreover, a force field remains
empirical so once again you have to demonstrate your idea ;-)
regards, Francois
> Thanks again for your informed reply. I agree with the points you make re:
> environment and coupling effects on the pKa's of the carboxylates in a
> C-terminal glutamate / glutamic acid residue.
> It was probably not appropriate to cite Thurkill et al. specifically here -
> I just wanted to point out the likely order of pKa's of the side-chain and
> C-terminal carboxylate in a C-terminal Glu residue.
>
> Let me first say that I don't expect the residue to be important for
> interactions or dynamics and that the NMR bundle points to a quite solvent
> exposed (and 'disordered') C-terminal residue.
> Then, from a quantum-mechanical perspective, the side-chain carboxylate of
> a C-terminal glutamate would have a higher proton-affinity than the
> C-terminal carboxylate, wouldn't it? The side-chain carboxylate is more
> electronically 'isolated' than the C-terminal carboxylate, after all.
> Wouldn't the coupling between the two carboxylate pKa's result in the
> side-chain carboxylate shifting up and the C-terminal carboxylate shifting
> down a little?
> Predictions from propka3.0 on the models in my NMR bundle (which
> incorporates coupling between different groups as well as the environment,
> albeit in a relatively crude way of course) also point in this direction,
> with the average pKa for the C-terminal carboxylate around 3.3 and the
> side-chain carboxylate around 4.7.
>
> I'd like to stay as close as possible (with standard MM MD) to the species
> present in the NMR experiment; if I'd put a NME cap on the C-terminus, I
> think I'd be further away from that than if I model the C-terminal
> glutamate/glutamic acid as one of the possible (and probably most likely)
> protonation states.
>
> Thanks,
> Marc
> PS Francois, I saw your reply now too - thanks; I used 'extrapolate' for
> want of a better word, and understand your dislike of using extrapolation
> in relation to charge derivation! (hence the single quotes).
>
> On 5 June 2012 18:29, Jason Swails <jason.swails.gmail.com> wrote:
>
>> On Tue, Jun 5, 2012 at 11:46 AM, Marc van der Kamp <
>> marcvanderkamp.gmail.com
>> > wrote:
>>
>> > Dear Jason and Francois,
>> >
>> > Many thanks for your replies.
>> > Francois, I was thinking of taking charges from GLH in amino12.in, and
>> > adjusting the charges in line with the changes between GLU in
>> amino12.inand
>> > aminoct12.in. In this way, the issues you mention regarding differences
>> > between carboxylic acid and carboxylate don't apply for the side-chain. I
>> > already noticed that changes in GLU charges in amino12.in and
>> > aminoct12.inare quite small, and therefore thought that
>> > 'extrapolating' these changes
>> > for a C-terminal GLH might be OK.
>> >
>> > Jason, I want to run standard MD at a pH around 4 (not uncommon in NMR
>> > experiments, for example). Under those conditions, one would expect a
>> > (solvent exposed) Glu side-chain to be (partly) protonated (pKa ~4.25)
>> > whereas a C-terminal carboxylate may well be unprotonated (pKa ~3.67, see
>> > e.g. Thurkill et al., doi: 10.1110/ps.051840806).
>> > So, I didn't think that a C-terminal GLH is so strange, at least not
>> within
>> > the limitations of standard MM MD.
>> > But perhaps there is a good reason that C-terminal GLH is never present
>> in
>> > standard AMBER forcefields?
>> >
>>
>> I'll agree that this is a tricky situation, but I think the assumptions
>> you're making above are dangerous. The pKas that are quoted are for
>> solvent-exposed residues that are not shifted significantly compared to
>> free amino acids (since environment can have a strong perturbing effect on
>> pKas). Also, all of the amino acids they titrate (at least in the
>> Thurlkill paper) are capped amino acids, because they did not want the
>> termini affecting the pKas (and they would).
>>
>> The two carboxylates on a C-terminal GLU are coupled, and their pKas cannot
>> be estimated independently, IMO. Therefore, I would argue that the pKas
>> you cited for C-termini and for GLU do not apply specifically to a
>> C-terminal GLU.
>>
>> Ultimately, if the C-terminal residues is very important for dynamics or
>> some crucial interaction, you will probably need to determine a better way
>> of treating it (alternatively, can you just cap the C-terminus with a NME
>> group?). Otherwise, I would suggest just using CGLU. You may be able to
>> use your proposed CGLH also to show that the protonation state is not vital
>> to the behavior of your protein (or maybe a double-protonated C-terminal
>> GLU).
>>
>> Of course both Francois and I have our biases in this case -- he's an
>> expert on charge derivation and I study ionization equilibria and pKa
>> calculations in proteins.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 06 2012 - 02:30:03 PDT
