The ultimate goal of structural studies of proteins is to gain insight into protein three-dimensional structure at highresolution level. This can often be accomplished by the application of techniques such as X-ray crystallography or multidimensional nuclear magnetic resonance (NMR) spectroscopy. However, high-resolution studies of proteins are not always feasible. For example, crystallographic studies require high-quality single crystals which for many proteins (e.g., the vast majority of membrane proteins) are not available. Furthermore, the question arises as to whether the relatively "static" structure in single crystals adequately represents the protein conformation in a complex and dynamic environment of living cells. There is a growing realization [e.g., Martinek et al. (1989)] that in vivo most proteins act in an interfacial environment where they form dynamic complexes with biological membranes, nucleic acids, polysaccharides, or other proteins. Aqueous buffers, from which protein crystals are usually grown, do not necessarily mimic well the conditions of protein functioning in vivo. NMR offers a somewhat better flexibility in studying protein structure in "biologically relevant" environments. However, the interpretation of NMR spectra of larger proteins is very complex, and the assignment of interproton distances generated by the NMR experiment is not always feasible; at present the technique is restricted to small proteins of less than approximately 15-20 kDa.The practical limitations encountered in high-resolution structural studies of proteins stimulate continual progress in f This is National Research Council of Canada Publication No. NRCC 34263.

Fourier transform infrared (FTIR) spectroscopy is an established tool for the structural characterization of proteins. However, many potential pitfalls exist for the unwary investigator. In this review we critically assess the application of FTIR spectroscopy to the determination of protein structure by (1) outlining the principles underlying protein secondary structure determination by FTIR spectroscopy, (2) highlighting the situations in which FTIR spectroscopy should be considered the technique of choice, (3) discussing the manner in which experiments should be conducted to derive as much physiologically relevant information as possible, and (4) outlining current methods for the determination of secondary structure from infrared spectra of proteins.

The general theory of Fourier self-deconvolution, i.e., spectral deconvolution using Fourier transforms and the intrinsic lineshape, is developed. The method provides a way of computationally resolving overlapped lines that can not be instrumentally resolved due to their intrinsic linewidth. Examples of the application of the technique to synthetic and experimental infrared spectra are presented, and potential applications are discussed. It is shown that lines in spectra having moderate signal/noise ratios (∼1000) can readily be reduced in width by a factor of 3. The method is applicable to a variety of spectroscopic techniques.

Infrared spectra of human central nervous system tissue and human breast carcinoma are presented. The spectra are discussed in terms of the composition of the tissues. It is shown that differences between spectra of white and grey matter can be rationalised on the basis of differences in lipid content. Spectra of the choroid plexus and arachnoid villus of the meninges show a series of absorptions not observed in other CNS tissue. These absorptions are discussed in terms of the connective tissue content of the samples. We demonstrate that the presence of collagen results in the appearance of a series of characteristic absorptions which may be mis-assigned as DNA phosphate absorptions. The implications of the presence of collagen in tissues for the diagnosis of disease states by IR spectroscopic methods, with particular reference to cancer, is discussed.