Info

Comparison of Simulated with Experimentally Measured Observables

The validation of simulated folding equilibria by comparison of simulated properties of the polypeptides with measured ones is not straightforward. First, the experimental data are generally averages over a conformational ensemble. Derivation of an ensemble from average values is impossible. On the other hand, the average of a particular observable, e.g. a NOE or a 3J-coupling constant, may be rather insensitive to the shape of the underlying conformational distribution over which the averaging is performed. For example, the folding equilibrium of the 7-b-peptide discussed before is rather different at 298 K, with 97% 314-helix present from the ensemble at 360 K with only 25% 314 helical content. Yet the agreement with the 21 measured 3J-values is as good for both quite different ensembles [53], as can be seen in Fig. 6.12. Yet, for another peptide, an 8-a-peptide, in DMSO, the 3J-values are sensitive to differences in the conformational distributions in solution on the one hand and in crystal on the other [42, 54]. In DMSO solution transient M-and P-helical fragments are present, leading to a broad conformational ensemble with <3J> = 6.8 Hz, the experimental value. In the crystal a rather narrow P-helical conformational ensemble is found with <3J> = 4.0 Hz close to the average 3J-value (4.2 Hz) of the X-ray structure.

Second, experimental data on folding equilibria are limited in number and accuracy. They may come from X-ray diffraction on crystals, or CD or NMR measurements in solution. The crystal data may only indicate that the fold that was adopted or preserved upon crystallization from a solution, is likely to be one of the dominant conformers in solution. However, the particular crystalline fold may also be induced by crystal contacts or particular co-solvents required for the crystallization. CD spectra may be very insensitive to the dominant conformers of an ensemble and may actually be determined by a fraction of the ensemble. An example of such a situation was reported in [19] where the CD spectrum was largely due to a conformer that constituted only 18% of the conformational ensemble. Regarding NMR-NOE spectra, it has been shown that different NOE peaks may show a very different sensitivity to the conformational ensemble [37].

Fig. 6.12 Comparison of the 21 experimental averaged 3J-coupling constants measured at 298 K with the corresponding averaged 3J-coupling constants calculated for the trajectory structures of 50 ns MD simulations of a b-heptapeptide (see Panel A, Fig. 6.1) in methanol at four different temperatures [53].

l-values (Hz) exp J-values i

Fig. 6.12 Comparison of the 21 experimental averaged 3J-coupling constants measured at 298 K with the corresponding averaged 3J-coupling constants calculated for the trajectory structures of 50 ns MD simulations of a b-heptapeptide (see Panel A, Fig. 6.1) in methanol at four different temperatures [53].

An example of quite different ensembles reproducing the same experimental NOE and 3J-value data for an 8-b-peptide in methanol can be found in [36]. These data appeared to be insufficient in number to uniquely determine the dominant conformer, a 2.512-P-helix or a 28-8-helix.

Was this article helpful?

0 0

Post a comment