Modification of the conventional chemical shift index (CSI) to distinguish between alpha and 3,10 helices

The conventional CSI plot is used to help determine protein secondary structure. The plot can be ambiguous when dealing with helical structure and often it is not clear whether or not the residues are in an alpha or 3,10 helix. The Ca and CO seconary shifts of residues in alpha helices are significantly higher than those of residues in 3,10 helices. This relationship was exploited to develop a new chemical shift index, which can be used to distinguish between these two helix subclasses.

The CSI equation was modified to account for the distinction between alpha and 3,10 helices as follows:

if fab > 0 and fao < 0
F = (fab² + fao²)/2
else if fab < 0 and fao < 0
F = -[(fab² + fao²)/2]
else if fab > 0 and fao < 0
if |fab| > |fao|
     F = (fab² + fao²)/2
else if |fab| < |fao|
     F = -[(fab² + fao²)/2]
else if fab < 0 and fao < 0
if |fao| > |fab|
     F = (fab² + fao²)/2
else if |fao| < |fab|
     F = -[(fab² + fao²)]

where

fab = Ca ss - Cb ss (ss = secondary shift)
fao = Ca ss - CO ss

CSI for residue i = 0.25*(F_i-1 + 2*F_i + F_i+1)

Conventional and modified CSI plots BMRB entries: comparison of CSI-determined secondary structure with crystal structure-defined secondary structure

accession # 4101 (X-ray 1CEX)

accession # 4193 (X-ray 1AKE)

accession # 4554 (X-ray 1RA1)

accession # 4836 (X-ray 1HAV)

Conventional and modified CSI plots for the above BMRB entries were compared to the PROMOTIF calculated secondary structure for the corresponding X-ray structures. In each CSI plot, alpha helices are shown in blue and 3,10 helices in red. Note that alpha helix maximum CSI values are systematically lower than those for 3,10 helices, providing a better distinction between these subclasses.

False positives in the CSI plot were typically localized to beta turns (see section 4).

Comparison with PROMOTIF defined secondary structure for NMR and crystal structures

accession # 4267 (NMR 1NGL, X-ray 1DFV)

Residues X-ray NMR modified CSI plot

13-15 3,10 3,10 (14-16) 3,10 (13-16)

25-28 - alpha 3,10 (25-27)

146-158 alpha alpha (149-161) alpha (147-159)

163-165 3,10 - 3,10 (163-166)

176-178 - 3,10 - (end of protein)

Yellow bars highlight the helical region (25-28) designated as 'alpha' in the NMR structure, but not identified in the x-ray structure.

accession # 4321 (NMR 5GCN, X-ray 1QSR)

Residues	X-ray	NMR	modified CSI plot
16-32	alpha	alpha (17-32)	alpha (16-20, 23-31)
38-45	alpha	alpha	alpha (38-46)
86-88	3,10	3,10	3,10 (85-87)
93-107	alpha	alpha (94-107)	alpha (93-109)
122-127	alpha	alpha (119-128)	alpha (122-127)
138-141	alpha	3,10	3,10 (138-140)

Purple bars highlight the final helix (residues 138-141) which is designated as alpha in the x-ray structure, but 3,10 in the NMR structure. The CSI plot suggests it is a 3,10 helix.

In applying the modified CSI to a protein for which both NMR and x-ray structures are available, it is shown that secondary structure prediction is generally consistent with secondary structure defined by the crystal structure. In the case of BMRB # 4321, the CSI and NMR data (PROMOTIF definition) agree on the subclass of helix which differs from the x-ray structure's PROMOTIF definition. However, breaks in helices and magnitude scaling issues indicate an optimal CSI equation has not yet been found (currently under development).

Introduction

I.	Average chemical shift values for helix and beta strand classes and subclasses
II.	Analysis of random coil backbone chemical shifts
III.	Modification of the conventional chemical shift index to distinguish between alpha and 3,10 helices
IV.	References