Crystal structure prediction conformer selection

A careful study by McDermott and co-workers of the high-K+ and low-K+ states of the KcsA channel has focused exclusively on the selectivity filter.101 They discovered that low K+ also induces the "non-con-ductive" or "collapsed" structure of the selectivity filter at neutral pH, but only if the sample remains well hydrated this state is lost if the bulk buffer is removed from the NMR rotor after spinning. Comparison of the measured chemical shifts with predictions by the SHIFTX and SPARTA programs identified the crystal structures that are most consistent with the selectivity filter conformations in the high-K+ and low-K+ proteolipo-some samples. Titrations of the chemical shift changes were used to measure site-specific affinities for K+. Based on the slow exchange rate between these conformations (<500 s 1), the authors suggest that the low-K+ conformation is relevant to channel inactivation rather than to conduction.101... 

The experimentally determined (S)/(R)-ratio of 18/82 was compared with the relative stabilities of the two diastereomeric products ([Co((S),(S)-ppm)((R)-ala)] / [Co((S),(S)-ppm)((S)-ala)] ), calculated by strain-energy minimization. The reported strain energies, based on a single conformer for each of the two diastereomeric products (identical to the crystal structure of the complex with coordinated (R)-alanine [328]), are in good agreement with the experimentally determined data (23/77 versus 18/82). A full conformational analysis led to a ratio of 30/70 when only conformational flexibility is allowed, or 33/67 when other isomers were also included in the analysis [294]. The assumption in the original report was that the enantio-selectivity is based on the relative energies of the diastereomeric forms of the cobalt(III) products [327]. Fortunately, a qualitatively similar result is expected if the stereoselectivity is controlled by the deprotonated intermediates. However, a quantitatively accurate prediction of the product ratio is not expected in this case. 

At the start of a calculation, one or more molecular conformations must be selected. If the simulation is carried out to test a novel prediction method, usually a known crystal structure from the CSD is used. Although it is tempting to use the known, solid state molecular conformation, a more rigorous test... 

Now we arrived at the problem of choice of Ri and R2. We selected for Ri the sec-butyl group, for a number of reasons. First, this group is bulky so that the monomer will avoid a ]8-type structure, which would lead to formation to mirror-symmetric dimers. Second, we did not consider it to be desirable to leave it to chance as to whether or not the monomer crystallizes in a chiral structure. Rather we decided to encourage crystallization in such a structure by use of this chiral handle . We emphasize that this is the sole role of the handle after the polymerization the handle can be removed, and if the asymmetric synthesis has been successful, the polymer will be chiral non-racemic because of its backbone conformation. It is true that an ambiguity arises because of the possibility of direct asymmetric induction by the sec-butyl group this point will be treated later. The third factor which led to our choice of this substituent is its prediction for chiral disorder, which we shall also return to. 

---