Just to clarify: Do you want to have the free energy difference between substates within the ensemble of the cross-linked product or the association free energy of your two short peptides? Or both?

Hi Ulf,

Essentially, I'm primarily after the free energy change of the crosslinking reaction itself:

(peptide-glutamine + lysine-peptide) (peptide-[gamma-glutamyl-lysine]-peptide + ammonia)

I suppose what I need overall is:

(free peptides in solution) (peptides associated in crosslinked-like conformation) (crosslinked peptides)

For now, I'm just interested in doing this for the one pair of peptides.

Dear Tristan,

  you are interested in a chemical reaction featuring the creation of covalent bonds: this cannot be easily studied within classical molecular dynamics, you should probably consider QM/MM simulations coupled with a technique for the reconstruction of free energy profiles, like umbrella sampling or metadynamics. A fully QM simulation on the other hand would be too expensive due to the size of your system (including water...).

Bests,

    Fabio Pietrucci

Hi Fabio,

That's what I find interesting about this case, though: the actual chemical reaction is simply the swap for one amide bond for another, the energy of which should be very close to zero. Therefore it seems to me that the quite strongly negative experimental free energy change for the reaction must be dominated by non-covalent interactions between the peptides: i.e. the crosslinking step must allow them to adopt an energetically favourable conformation that is much less favourable in the absence of crosslinking. Assuming a negligible contribution from the actual chemical reaction, the problem then boils down to the difference in solvation energy between (free peptides) and (crosslinked peptides + an ammonium ion). Am I naive in thinking that this should be accessible via similar alchemical methods to those used to estimate the effect of point mutations on solvation energy?

Go for it! I would try it:-) I would appear/disappear a molecular-mechanical bond between the two residues. You may face some problems with constraining algorithms and I assume poor convergence in the intermediate region. Maybe nonequilibrium approach would give you better results at lower cost. It means to repeat appearing the bond several (or many) times, do the same for disappearing and use e.g. Crooks fluctuation theorem for evaluation.

I would be happy to read an article on your simulations anyway.

I understand your point, maybe you are right but it really depdends if the "close to zero" in your first sentence is really close to zero. Or, if it is not zero, if your force field represents well enough this energy difference due to the change of chemical bonding, then I think you are right.

Thanks for your input, guys. It sounds like it's at least worth a try.

It's worth a try, indeed. But be aware - "small" peptides such as this can be trickier to deal with than a protein ( >= 60 residues) because the latter have a lot of intramolecular contacts and excluded volume (repulsion terms) that reduce the available conformational space. You don't have those constraints to help you in a short peptide. Also, since you are dealing with a chemical reaction (release of ammonia) you will definitely need QM methods. QM/MM as suggested by a previous respondent.

If you would like to use free energy as a primary argument then you would need to demonstrate that dG between your preferred conformation and any unfolded state is negative. This task is certainly out of your reach. You would need to use assumptions, such as "if the peptide shows negative dG between the preferred conformation and several representative unfolded states, same thing would occur with respect to other unfolded states". It would also allow you to make use of specific collective variables to calculate the potentials of mean force along each unfolded-folded path.