The three-dimensional structures of two cytochromes c' have been determined in order to analyse the common features of proteins of this family and their relationship with other four-helix bundle structures. The structure of cytochrome c' from Alcaligenes sp was determined by molecular replacement supplemented with the iron anomalous scattering and the use of a single isomorphous heavy-atom derivative, and was refined using synchrotron data to 1.8 A resolution. The final model, comprising 956 protein atoms (one monomer) and 89 water molecules, has a final R value of 0.188 for all data in the range 20.0-1.8 A resolution (14 673 reflections). The structure of the cytochrome c' from Alcaligenes denitrificans is isomorphous and essentially identical (r.m.s. deviation for all atoms 0.36 A). Although its amino-acid sequence has not been determined chemically, only four differences from that of Alcaligenes sp cytochrome c' were identified by the X-ray analysis. The final model for Alcaligenes denitrificans cytochrome c', comprising 953 protein atoms and 75 water molecules, gave a final R factor of 0.167 for all data in the range 20.0-2.15 A (8220 reflections). The cytochrome c' monomer forms a classic four-helix bundle, determined by the packing of hydrophobic side chains around the enclosed haem group. There are very few cross-linking hydrogen bonds between the helices, the principal side-chain hydrogen bonding involving one of the haem propionates and a conserved Arg residue. The cytochrome c' dimer is created by a crystallographic twofold axis. Monomer-monomer contacts primarily involve the two A helices, with size complementarity of side chains in a central solvent-excluded portion of the interface and hydrogen bonding at the periphery. Both species have a pyroglutamic acid N-terminal residue. The haem iron is five-coordinate, 0.32 A out of the haem plane towards the fifth ligand, His120. The unusual magnetic properties of the Fe atom may be linked to a conserved basic residue, Arg124, adjacent to His120.

The crystal structure of an orthorhombic form of human apolactoferrin (ApoLf) has been determined from 2.8 ~ diffractometer data by molecular replacement methods. A variety of search models derived from the diferric lactoferrin structure (Fe2Lf) were used to obtain a consistent solution to the rotation function. An R-factor search gave the correct translational solution and the model was refined by rigid-body least-squares refinement (program CORELS). Only three of the four domains were located correctly by this procedure, however; the fourth was finally placed correctly by rotating it manually onto three strands of electron density which were recognized as part of its central/3-sheet. The final model, refined by restrained least-squares methods to an R factor of 0.214 for data in the resolution range 10.0 to 2.8 A~, shows a large domain movement in the N-terminal half of the molecule (a 54 rotation of domain N2) and smaller domain movements elsewhere, when compared with Fe=Lf. A feature of the crystal structure is that although the ApoLf and Fe2Lf unit cells appear very similar, their crystal packing and molecular structures are quite different.

The structure of apo-azurin from Alcaligenes denitrill'cans has been determined at high resolution by X-ray crystallography. Two separate structure analyses have been carried out, (i) on crystals obtained from solutions of apo-azurin and (ii) on crystals obtained by removal of copper from previousl],' formed crystals of holo-azurin. Data to 1.8A resolution were collected from the apo-azurin crystals, by Weissenberg photography (with image plates) using synchrotron radiation and by diffractometry, and the structure was refined by restrained least-squares methods to a final R value of 0.160 for all data in the range 10.0-1.8 A,. The final model of 1954 protein atoms, 246 water molecules (66 halfweighted), four SO4-ions, and two low-occupancy (0.13 and 0.15) Cu atoms has r.m.s, deviations of 0.012, 0.045 and 0.013A, from standard bond lengths, angle distances and planar groups. For copper-removed azurin, data to 2.2 A, were collected by diffractometry and the structure refined by restrained least squares to a final R value of 0.158 for all data in the range 10.0-2.2 A,. The final model of 1954 protein atoms, 264 water molecules, two SO4: ions, two low occupancy (0.18 and 0.22) metal atoms and one unidentified atom (modelled as S) has r.m.s. deviations of 0.013, 0.047 and 0.012 A, from standard bond lengths, angle distances and planar groups. The two structures are essentially identical to each other and show no significant differences from the oxidized and reduced holo-azurin structures. The ligand side chains move slightly closer together following the removal of copper, with the radius of the cavity between the three strongly binding ligands, His 46, His 117 and Cys 112, shrinking from 1.31 A, in reduced azurin to 1.24 A, in oxidized azurin and 1.16 A, in apo-azurin. There is a suggestion of increased flexibility in one of the copperbinding loops but the structure supports the view that the copper site found in holo-azurin is a stable structure, defined by the constraints of the polypeptide structure even in the absence of a bound metal ion.

Crystal structures of two isomorphous solvates of the macrolide antibiotic borrelidin [ borrelidin 3-methylbutan-1-ol (A) and borrelidin (S)-2-methylbutan-1-ol (B)] have been determined from X-ray diffractometer data recorded at 294 and 152 K respectively. Crystals are monoclinic, space group P21, with a 13.501(2), b 12.000(2), c 11.863(2) � , β 110.76(1)� [(A); T = 294 K], and a 13.565(2), b 11.646(2), c 11 .376(3) �, β 109.76(1)� [(B); T = 152 K], (Z = 2). The structures were solved by direct-methods procedures and refined by full-matrix least-squares analysis to conventional R-factors of 0.061 [(A), 2676 reflections] and 0.048 [(B), 2851 reflections]. The absolute configuration of borrelidin follows from the known chirality of the (S)-2-methylbutan-1-ol solvent in (B), and is 3S,4S,6S,8R,1OS,11R,17S,18R,22R. The unusual structural and stereochemical features of borrelidin are discussed in relation to other macrolide antibiotics.

(3aRS,4RS,6RS,8aSR)-5-Methyleneoctahydro-4H-3a,6-methanoazulene-4-carboxylic acid (7) has been synthesized and a detailed comparison made between its molecular structure and that of gibberellic acid (4). Indane-5-carboxylic acid (8) was converted by an alternating sequence of reduction and isomerization into the hydrindene acid (11). The diazomethyl ketone derived from (11) was transformed to cyclopropyl ketone (12) which was reduced by Li/NH3 with stereochemical inversion at the β-carbon to give a 1:3 mixture of ketones (13) and (14) respectively. Ketone (14) was then converted into the title acid (7) by a standard procedure. ��� Molecules of (7) crystallize in space group P2/c with a 7.782(1), b 11.055(1), c 14.559(3) Ǻ, β 116.05(1)°. The structure was solved by direct methods and refined by full-matrix least squares to a final R value of 0.053.