PROTEIN SECONDARY AND TERTIARY STRUCTURE STUDIES USING VIBRATIONAL SPECTROSCOPY AND HYDROGEN EXCHANGE 

Baello, Bernoli

The optical spectra measured for a training set of 23 globular proteins using equilibrium hydrogen-deuterium exchange difference Fourier transform IR (FTIR) spectroscopy method presented in the previous chapter were further tested for structural information. The systematic variation of the band intensity with the solvent deuterium content was assessed for quantitative correlation with protein tertiary structure using factor analysis and restricted multiple regression scheme. For this purpose, a protein tertiary structure descriptor was developed that represents the interactions between well-defined secondary structural elements such as helices and sheets. It accounts for the packing of the secondary structures as well as their distribution in space. The contact profiles were calculated from the contact maps as distributions of the different interactions between these structural elements as functions of distance. They were reconstructed using the H/D exchange data and yielded decent qualitative agreement.

The contact descriptor developed here represents an additional quantitative information on protein structure. Previously, a well-established secondary structure description provided by the fractional components of secondary structure (FC values) was used to characterize proteins from their optical spectra through the percentage amounts of the different secondary structure elements. Determination of the number of segments of these elements differentiates the structures that contain relatively the same FC values as shown by the work on ribonuclease A and T₁ (16). Finally, the current tertiary descriptor offers a means to further describe the possible arrangement in space of these segments and rule out those that are inconsistent with the rest of the experimental data.

Qualitative and quantitative correlation was observed with the amide III VCD spectral response to variations in secondary structure, commonly expressed as fractional contents of secondary structure content (FCs). This work provided the first measurement and systematic analysis of amide III VCD. Additionally, a similarity algorithm was developed for correlation studies that showed, together with the results from a factor analysis-restricted multiple regression (FA/RMR) approach, that the newly accessible vibrational mode contains similar degree of secondary structural information as the more established amide I VCD and that this information rests primarily in the bandshape. Amide III VCD measurement presented difficulties not encountered in the other modes. This was shown to be surmountable by proper mathematical filtering. Other vibrational modes occurring at even lower frequencies which are not currently investigated using VCD appear to present greater challenges due to solvent interference and instrumentation requirements. Since the information content of amide III has been demonstrated to be similar to amide I, the latter would probably be the preferred vibrational region for study unless investigations beyond this level is desired. One area for continued study is the amide III VCD of cytochrome c, which yielded an atypical, intense spectral response that possibly arise from coupling to the covalently bound heme moiety.

Chemical perturbation through hydrogen exchange has afforded greater sensitivity due to the structural significance of the labile hydrogens in the protein. As demonstrated in this work, with H/D exchange monitored via FTIR spectroscopy, greater predictive accuracy of the helix and sheet contents were achieved on the secondary structure level. Further, information on the packing and distribution of these secondary structures was obtained by another treatment of the same set of data. With respect to the descriptor that was proposed in this thesis, different levels of structure are now expressed as separable mathematical quantities that would be useful for future structural studies beginning with the fractions of the secondary structure, to the number of segments of these elements and to their packing and distribution in space. Analysis with the tertiary structure descriptor would be most profitable when used in consort with determination of the other descriptors as these rely to some extent on information about the secondary structure content.

For all of these efforts on protein structure-spectra correlations to be of greater use, a bigger training set is necessary to make the generalizations more widely applicable, as well as to provide greater confidence in their predictions. In a bigger basis set of proteins, greater structure variations would be taken into account.

Hydrogen exchange traditionally has been coupled with other techniques in the temporal regime and this would be a logical extension of this work. As has been shown by the available amount of literature, significant structural insights can come from such work. Other perturbations, such as temperature and pH effects could be investigated with this technique as these are issues that have not been addressed here. However, one has to be prepared to meet the consequent greater demands on the analysis methods as more variables are considered.