Section 19.5.8. Structures determined by X-ray fibre diffraction

The structural details of the α-helix and β-sheet, the principal secondary-structure elements of proteins, have emerged from the analysis of synthetic polypeptides (Pauling & Corey, 1951, 1953). Analysis of noncrystalline fibre-diffraction patterns led to the triple-helical coiled-coil model of collagen (Ramachandran & Kartha, 1955; Rich & Crick, 1955). Recent studies on the organization of β-sheets in peptides of up to about 45 residues are providing an understanding of the molecular details of amyloid fibrils, related to Alzheimer's disease (Inouye et al., 1993; Malinchik et al., 1998).

The molecular structures of a series of DNA and RNA helices have been determined and refined using data from polycrystalline fibres (Arnott et al., 1969; Chandrasekaran & Arnott, 1989). These include the canonical A, B and C forms of DNA, corresponding, respectively, to 11-, 10- and 9.3-fold right-handed antiparallel Watson–Crick base-paired helices. Structural differences between the three have been attributed to changes in furanose puckerings and helical parameters: the A form has C3-endo, but B and C have C2-endo or analogous C3-exo puckers. All RNA duplexes are members of the A family. Later important structures included the sixfold single helix of poly (C) (Arnott et al., 1976), a compact eightfold double helix for poly d(AT) and poly d(IC) (Arnott et al., 1983), and the left-handed Z-DNA for poly d(GC) (Arnott et al., 1980). Difference Fourier syntheses were instrumental in locating a spine of water molecules in the minor groove and a series of sodium ions and water molecules that bridge the phosphate groups of adjacent DNA molecules in the tenfold helices of poly (dA)·poly (dT) (Chandrasekaran et al., 1995), poly (dA)·poly (dU) and poly d(AI)·poly d(CT) (Chandrasekaran et al., 1997). Data from noncrystalline fibres have been used to determine, among others, the structures of DNA·RNA hybrid duplexes (Arnott et al., 1986), a DNA triple-stranded helix (Chandrasekaran et al., 2000a) and two RNA triple-stranded helices (Chandrasekaran et al., 2000b,c). In each case mentioned, the best model was clearly preferred statistically (Hamilton, 1965) and had an R value between 0.2 and 0.3 to about 3 Å resolution.

Among the three-dimensional structures determined for industrially useful and biologically important polysaccharides are the gel-forming calcium i-carrageenan (Arnott, Scott et al., 1974), sodium pectate (Walkinshaw & Arnott, 1981), gellan (Chandrasekaran et al., 1988) and welan (Chandrasekaran, Radha & Lee, 1994), and a series of distinct helical forms of the glycosaminoglycan hyaluronan (Arnott & Mitra, 1984). The conformations of these molecules are delicately controlled by ions, such as sodium, potassium and calcium. The repeating units range from a simple monosaccharide to a branched pentasaccharide.

Specific interactions among the polysaccharides and their associated small molecules can be correlated with their observed properties. A number of neutral polysaccharides, such as cellulose, chitin and mannan, are twofold ribbon-like helices, which aggregate and are hence water insoluble. The A and B forms of amylose, the main constituents of starch granules, are sixfold left-handed parallel double helices. Derivatization of amylose leads to the formation of single helices (Chandrasekaran, 1997). The water-soluble galactomannan derives its high viscosity in aqueous solution from intermolecular side-chain interactions (Chandrasekaran et al., 1998).

The largest repeating units in structures determined by fibre diffraction are those of several members of the tobamovirus family, including tobacco mosaic virus (Namba et al., 1989), cucumber green mottle mosaic virus (Wang & Stubbs, 1994) and ribgrass mosaic virus (Wang et al., 1997). These viruses are rod-shaped, 3000 Å long and about 180 Å in diameter. Oriented sols yield exceptionally good diffraction patterns (Fig. 19.5.8.1). The asymmetric unit consists of a protein subunit of approximate molecular weight 18 000 Da and three nucleotides of RNA. The coat proteins are folded like globular proteins and are about 40% α-helical, with small regions of β-sheet. All of the amino acids, all three nucleotides, and in some cases water molecules and calcium ions, are seen in the electron-density maps. The TMV structure was determined by MDIR; the remaining structures were determined by molecular replacement from TMV or by a combination of molecular replacement and isomorphous replacement. All of the structures were refined by restrained least-squares or molecular-dynamics methods to R values of less than 0.10 at resolutions between 2.9 and 3.5 Å.

Figure 19.5.8.1| top | pdf | X-ray diffraction pattern from an oriented sol of the U2 strain of tobacco mosaic virus.

Several filamentous bacteriophage structures, including fd, Pf1 and related strains, have been determined and refined. Filamentous bacteriophages are flexible viruses, about 60 Å in diameter and 10 000 to 20 000 Å in length. Several thousand copies of a coat protein of about 50 residues wrap around a central single-stranded circular DNA. The DNA does not appear to be sufficiently ordered to appear in electron-density maps. The coat-protein molecules have an unusually simple structure, being almost entirely α-helical (Marvin et al., 1974). Model-building approaches have therefore been used, sometimes supplemented by isomorphous replacement (Bryan et al., 1983). Neutron scattering from bacteriophages with selectively deuterated amino-acid residues has also been used to assist model building (Nambudripad et al., 1991). Both restrained least-squares (Nambudripad et al., 1991) and molecular-dynamics (Gonzalez et al., 1995) refinement methods have been used. Although there is not complete agreement about the structure, the coat protein clearly forms two α-helical layers, possibly with a short intervening peptide loop (Nambudripad et al., 1991).

Low-resolution X-ray fibre-diffraction data have been successfully used to model the structural details of a number of complex assemblies. For example, the structure of the F-actin helix at 8 Å resolution has been described by combining the single-crystal structure of the G-actin monomer with fibre-diffraction data (Holmes et al., 1990). This structure, in turn, has been used to model the muscle thin filament, composed of F-actin monomers and tropomyosin, at about 25 Å resolution, both in the resting and activated states, and hence to understand the movement of tropomyosin in muscle function (Squire et al., 1993). The structure of the microtubule has been determined at 18 Å resolution using information from electron microscopy and fibre diffraction (Beese et al., 1987). A similar but more sophisticated approach was used for bacterial flagellar filaments at 9 Å resolution (Yamashita, Hasegawa et al., 1998); the diffraction patterns obtained from these filaments are of such high quality that prospects for a complete molecular structure are excellent.