Wolfgang Steigemann

Structure of the complex formed by bovine trypsin and bovine pancreatic trypsin inhibitor: II. Crystallographic refinement at 1.9 Å resolution☆

Abstract The crystal structure of the complex formed by bovine trypsin and bovine pancreatic trypsin inhibitor has been refined with data to 1.9 A resolution, using a procedure described by Deisenhofer & Steigemann (1974) in their refinement of the crystal structure of the free inhibitor. This procedure involves cycles consisting of phase calculation using the current atomic model, Fourier synthesis using these phases and the observed structure factor amplitudes and Diamonds real-space refinement (Diamond, 1971,1974). At various stages, difference Fourier syntheses are calculated to detect and correct gross errors in the model and to localize solvent molecules. The refinement progressed smoothly, starting with the model obtained from the isomorphous Fourier map at 2.6 A resolution. The R -factor is 0.23 for 20,500 significantly measured reflections to 1.9 A resolution, using an over-all temperature factor of 20 A 2 . The estimated standard deviation of atomic positions is 0.09 A. An objective assessment of the upper limit of the error in the atomic coordinates of the final model is possible by comparing the inhibitor component in the model of the complex with the refined structure of the free inhibitor (Deisenhofer & Steigemann, 1974). The mean deviation of main-chain atoms of the two molecular models in internal segments is 0.25 A, of main-chain dihedral angles 5.1 ° and side-chain dihedral angles 6.5 °. A comparison of the trypsin component with α-chymotrypsin (Birktoft & Blow, 1972) showed a mean deviation of main-chain atoms of 0.75 A. The structures are closely similar and the various deletions and insertions cause local structural differences only.

Structure of erythrocruorin in different ligand states refined at 1.4 A resolution.

Abstract The crystal structure of erythrocruorin has been refined by constrained crystallographic refinement at 1·4 A resolution in the following ligand states: aquomet (Fe 3+ , high spin), cyanomet (Fe 3+ , low spin), deoxy (Fe 2+ , high spin) and carbonmonoxy (Fe 2+ , low spin). The final R -value at this resolution is better than 0·19 for each of these models. The positional errors of the co-ordinates are less than 0·1 A. The root-mean-square differences between the deoxygenated and the ligated erythrocruorin are about 0·1 A, being largest for cyanomet-erythrocruorin. The changes in tertiary structures propagate from the location of primary events and often fade out at the molecular surface. Helix E passing the distal side of the haem group is affected most by the direct contact with the ligand bound to the haem iron. Steric hindrance by the distal residue IleE11 forces the cyanide and carbonmonoxide ligands to bind at an angle to the haem axis. The strain at the ligand is partially relieved by movement of the haem deeper into the haem pocket and rearrangement of neighbouring residues. The differences in iron location with respect to the mean haem plane are spin-dependent but unexpectedly small (the largest value is 0·15 A between deoxy and carbonmonoxy-erythrocruorin). Spin state changes seem to have little influence on the porphyrin stereochemistry; it is determined primarily by the chemical properties of the ligand and its interaction with the haem and the globin. These non-covalent interactions are largely responsible for the initiation of the structural changes on ligand binding.

Two cis-prolines in the Bence-Jones protein Rei and the cis-pro-bend

The stabilities of the cisand transforms of the X-Pro peptide group differ only slightly by -2.0 to 2.0 kcal/mol-’ for linear polypeptides [ 11. In small cyclic peptides, the ring closure enforces the formation of cis-peptide groups [2,3]. It is surprising, therefore, that there are only few reports of X-Pro cis-peptide groups found in globular protein molecules. Furthermore, most of these are questioned by the authors due to the uncertainty in the interpretation of a Fourier map calculated with phases obtained from isomorphous replacement [4,5a,Sb,6-81. During the course of the constrained crystallographic refinement of the crystal structure of the Bence-Jones Protein Rei [9a,9b] we found evidence for two X-Pro cis-peptide groups and were able to confirm this by difference Fourier methods.

Structural model of the collagen-like region of C1q comprising the kink region and the fibre-like packing of the six triple helices☆

A detailed three-dimensional model of the collagenous part of C1q was derived by model building and computer-aided energy refinement calculations. The proposed structure is based on the collagen-like (-Gly-Xaa-Yaa-) repeating sequence of 78 to 81 residues in the N-terminal regions of the constituent A, B and C chains, on the mode of disulphide linkage between the 18 chains of C1q, and on its electron microscopically derived gross structure. It is demonstrated that the interruptions of the repeating sequence about half-way along the length of the collagenous regions (Gly36-Ile37-Arg38-Thr39 in the A chain and Ala36-Ile37-Hy138 in the C chain) do not lead to a disruption of the triple helical conformation but rather to a bend of about 60 degrees in an otherwise continuous triple helix. These features are consistent with a flexibility comparable with that of regular triple helices and with the observed low proteolytic susceptibility of the kink region. The azimuthal orientation of the kink is defined approximately by ArgA38 being located in the cap of the knee. Because of this extra residue between two glycine residues, a bad contact that would arise between the methyl group of AlaC36 and the peptide carbonyl of IleA37 in a straight triple helix is relaxed. The model features also a cluster of hydrophobic contacts between large hydrophobic side-chains in the interaction edges between the six collagen triple helices aligned with their about 10 nm long N-terminal regions in the fibril-like endpiece of C1q. The azimuthal orientations of the triple helices were derived by energy calculations of side-chain interactions previously applied to fibre-forming collagens. Independently, the same orientations and interaction edges were derived from the azimuthal orientation of the kink and the electron microscopically observed orientations of the triple helical arms that emerge from the endpiece, and which carry the C-terminal globular binding domains. The structural model has a number of implications for the assembly of the first component of complement from C1q and the zymogen complex C1r2C1s2 and possible mechanisms of its activation.

The structure of oxy-erythrocruorin at 1.4 Å resolution

Abstract Stable oxy-erythrocruorin crystals were prepared from the met form and diffraction intensities recorded by a novel procedure applying a hydrogen sulphide/oxygen gas mixture in a flow cell. The crystal structure was analysed by difference Fourier methods and partly refined at 1.4 A resolution. The dioxygen was found to be bound end-on and hydrogen-bonded to a water molecule. The iron, which is 0.17 A out of the haem plane in the deoxy form (Steigemann & Weber, manuscript in preparation), changes its position only very slightly and moves to the proximal side. Further conformational changes are minimal and far smaller than observed for the carbonmonoxy and cyano derivatives (Steigemann & Weber, manuscript in preparation).

Experience with various techniques for the refinement of protein structures

Publisher Summary This chapter reviews the experiences with the refinement techniques Real Space Refinement (RLSP), COnstrained-REstrained Least-Squares (CORELS) and EREF. From Real Space Refinement experience, several advantages and disadvantages, which are inherent to the method, have been recognized. The principal advantage for improvement of the initial model by fitting to an MIR map lies in the fact that an optimum interpretation of such a map can be obtained before the observed phases are replaced by calculated ones. A new version of CORELS became available, in which the definition of rigid groups especially, and the treatment of different space groups, have been simplified significantly. The main advantages of the Jack–Levitt method became apparent during the refinement of the Protein structures: (1) refinement at low resolution is possible, (2) geometric restraints can be relaxed temporarily, (3) treatment of branched chains is easy, (4) the method requires about 30% less computing time than Diamonds I real space refinement, and (5) distorted geometry can be repaired.

Crystallographic evidence for structurally similar domains in the human κ-type Bence-Jones protein Rei

Abstract Patterson search calculations demonstrated the presence of a local pseudo 2-fold axis in the crystalline state of the monomeric Bence-Jones protein Rei. These local diads are also observed in a preliminary Fourier synthesis at 4 A resolution. They appear to relate the variable and constant halves of neighbouring molecules. The presence of local diads is evidence that the tertiary structures of the constant and variable halves of the Bence-Jones protein are similar.