Search

The trypsin from Fusarium oxysporum is equally homologous to trypsins from Streptomyces griseus, Streptomyces erythraeus and to bovine trypsin. A DFP (diisopropylfluorophosphate) inhibited form of the enzyme has been crystallized from 1.4 M Na2SO4, buffered with citrate at pH 5.0-5.5. The crystals belong to space group P2(1) with cell parameters a = 33.43 A, b = 67.65 A, c = 39.85 A and beta = 107.6 degrees. There is one protein molecule in the asymmetric unit. X-ray diffraction data to a resolution of 1.8 A were collected on film using synchrotron radiation. The structure was solved by molecular replacement using models of bovine and S. griseus trypsins and refined to an R-factor of 0.141. The overall fold is similar to other trypsins, with some insertions and deletions. There is no evidence of the divalent cation binding sites seen in other trypsins. The covalently bound inhibitor molecule is clearly visible.

The toxin vipoxin, which is a complex between a basic toxic phospholipase A2 and an acidic non-toxic protein inhibitor, is found in the venom of the Bulgarian viper (Vipera ammodytes ammodytes), the most toxic snake in Europe. The two polypeptide chains each consist of 122 residues and are highly homologous (62%). The vipoxin complex is the first reported example of a high degree of structural homology between an enzyme and its natural inhibitor. The present crystals diffract in the X-ray beam to 1.8 A resolution. The space group is P2(1)2(1)2(1). The cell dimensions are a = 45.80 A, b = 55.36 A and c = 107.69 A. Native data to a resolution of 2.8 A have been recorded.

Neurotoxin-1 from Naja naja oxiana venom (NTX-1) has been crystallized by vapor diffusion in sitting drops. The crystals have cell dimensions of a = 25.2 A, b = 75.6 A, c = 35.9 A, and are in space group P2(1)2(1)2(1). Three-dimensional data to 1.9 A have been recorded by a Syntex P2(1) automatic diffractometer. The atomic structure of the toxin has been determined by molecular replacement using the alpha-cobratoxin (alpha-CTX) as the search model. The position of 534 non-hydrogen protein atoms have been determined. The model contains 65 water molecules. Refinement has led to an R-factor of 19.3% at 1.9 A resolution. The secondary and tertiary structures of NTX-1 have been analyzed and a comparison with structure of the alpha-CTX has been made.

The inhibition of serine proteinases by both synthetic and natural inhibitors has been widely studied. Eglin c is a small thermostable protein isolated from the leech, Hirudo medicinalis. Eglin c is a potent serine proteinase inhibitor. The three-dimensional structure of native eglin and of its complexes with a number of proteinases are known. We here describe the crystal structure of hydrolysed eglin not bound to a proteinase. The body of the eglin has a conformation remarkably similar to that in the known complexes with proteinases. However, the peptide chain has been cut at the 'scissile' bond between residues 45 and 46, presumed to result from the presence of subtilisin DY in the crystallisation sample. The residues usually making up the inhibiting loop of eglin take up a quite different conformation in the nicked inhibitor leading to stabilising contacts between neighbouring molecules in the crystal. The structure was solved by molecular replacement techniques and refined to a final R-factor of 14.5%.

A novel actinomycin (Act SG3) from a strain of Streptomyces galbus var. C-72, as well as actinomycin D (Act D) were found to act as competitive inhibitors of serine proteinases from microorganisms. The inhibitory properties of Act SG3 and Act D are compared with these of other peptide antibiotics, namely bacitracin A (Bac A) and gramicidin S (Gr S). The last compound has only a weak inhibitory effect. The following order of affinity for the four peptide antibiotics towards subtilisin DY and proteinase K was observed: Bac A > Act D > Act SG3 = Gr S. The affinity towards thermitase changes as follows: Act SG3 = Act D > Bac A > Gr S.

Circular dichroic spectroscopy has been used to study the effect of pH, guanidinium hydrochloride concentration and temperature on the conformation of the fungal subtilisin-like proteinase K and the bacterial DY. The ellipticity of the bands in the far ultraviolet region remains almost unchanged in the pH range 3.0-11.0 (PMS-proteinase K) and 5.0-10.0 (PMS-subtilisin DY). The same ranges of pH stability were determined from the pH dependence of the near ultraviolet dichroic spectra. Hence the changes in the tertiary and secondary structure occur in parallel. Proteinase K is considerably more stable at acidic and somewhat more stable at alkaline pH than subtilisin DY. At neutral pH proteinase K is more resistant to denaturation by guanidinium hydrochloride than is subtilisin DY. The midpoints of the denaturation curves were 6.2 M and 3.2 M guanidinium, respectively. The thermal unfolding of proteinase K occurred at a higher temperature than for subtilisin DY, the transition midpoints being 65 degrees and 48 degrees, respectively. Thus proteinase K is overall a much more robust molecule than subtilisin DY, showing greater resistance to all three forms of denaturation. The differences in the stability of the two proteinases can be partly explained by differences in their calcium binding sites.

The X-ray crystal structure of the molecular complex of penicillin G with a deacylation-defective mutant of the RTEM-1 beta-lactamase from Escherichia coli shows how these antibiotics are recognized and destroyed. Penicillin G is covalently bound to Ser 70 0 gamma as an acyl-enzyme intermediate. The deduced catalytic mechanism uses Ser 70 0 gamma as the attacking nucleophile during acylation. Lys 73 N zeta acts as a general base in abstracting a proton from Ser 70 and transferring it to the thiazolidine ring nitrogen atom via Ser 130 0 gamma. Deacylation is accomplished by nucleophilic attack on the penicilloyl carbonyl carbon by a water molecule assisted by the general base, Glu 166.

The crystal structure of the antigen-binding fragment of a monoclonal antibody (8F5) that neutralizes human rhinovirus serotype 2 has been determined by X-ray diffraction studies. Antibody 8F5, obtained by immunization with native HRV2 virions, cross-reacts with peptides of the viral capsid protein VP2, which contribute to the neutralizing immunogenic site B in this serotype. The structure was solved by the molecular replacement method and has been refined to an R-factor of 18.9% at 2.8 A resolution. The elbow angle, relating the variable and constant modules of the molecule is 127 degrees, representing the smallest elbow angle observed so far in an Fab fragment. Furthermore, the charged residues of the epitope can be well accommodated in the antigen-binding site. This is the first crystal structure reported for an antibody directed against an icosahedral virus.

The three-dimensional structure of one of the three lipoamide dehydrogenases occurring in Pseudomonas putida, LipDH Val, has been determined at 2.45 A resolution. The orthorhombic crystals, grown in the presence of 20 mM NAD+, contain 458 residues per asymmetric unit. A crystallographic 2-fold axis generates the dimer which is observed in solution. The final crystallographic R-factor is 21.8% for 18,216 unique reflections and a model consisting of 3,452 protein atoms, 189 solvent molecules and 44 NAD+ atoms, while the overall B-factor is unusually high: 47 A2. The structure of LipDH Val reveals the conformation of the C-terminal residues which fold "back" into the putative lipoamide binding region. The C-terminus has been proven to be important for activity by site-directed mutagenesis. However, the distance of the C-terminus to the catalytically essential residues is surprisingly large, over 6 A, and the precise role of the C-terminus still needs to be elucidated. In this crystal form LipDH Val contains one NAD+ molecule per subunit. Its adenine-ribose moiety occupies an analogous position as in the structure of glutathione reductase. However, the nicotinamide-ribose moiety is far removed from its expected position near the isoalloxazine ring and points into solution. Comparison of LipDH Val with Azotobacter vinelandii lipoamide dehydrogenase yields an rms difference of 1.6 A for 440 well defined C alpha atoms per subunit. Comparing LipDH Val with glutathione reductase shows large differences in the tertiary and quaternary structure of the two enzymes. For instance, the two subunits in the dimer are shifted by 6 A with respect to each other. So, LipDH Val confirms the surprising differences in molecular architecture between glutathione reductase and lipoamide dehydrogenase, which were already observed in Azotobacter vinelandii LipDH. This is the more remarkable since the active sites are located at the subunit interface and are virtually identical in all three enzymes.

The major protein component in seeds is storage protein. These have no known enzymatic activity and act to provide amino acids as a source of metabolites in the developing seedling. We report here the first three dimensional crystal structure of a seed storage globulin at high resolution. The molecule of the 2S globulin, narbonin, from Vicia narbonensis L., consists of an eight-stranded parallel alpha/beta barrel structure similar to that observed in triose phosphate isomerase (TIM). Narbonin is the first protein with this topology possessing no known enzymatic activity. Because of the lack of sequence information most of the primary structure was determined directly from the electron density.

Lactotransferrin is an iron-binding protein. It has been purified from buffalo colostrum. The purified lactotransferrin has been crystallized in 10% ethanol solution. The crystals are orthorhombic and the space group is P2(1)2(1)2(1) with unit cell dimensions a = 161.70 A, b = 155.75 A, c = 113.48 A. The asymmetric unit contains three molecules of the protein with a solvent content of about 59%. The crystals were stable in the X-ray beam and diffract beyond 3.5 A resolution. The native data have been collected and the structure determination is in progress.

The fluorescence properties of proteinase K are described and related to the X-ray model refined at 1.48 A resolution. Upon excitation of proteinase K at 295 nm the fluorescence is determined by the two tryptophan residues, Trp-8 and Trp-212. The tryptophans are partly buried just below the surface of the molecule. Neither Trp is in a highly hydrophobic environment, suggesting that this cannot be the explanation for the fluorescence at 330 nm: formation of exiplexes with adjacent peptide bonds would seem to be the more likely cause. Trp-8 is located in a 'cavity', close to an internal cluster of water molecules. The contribution of Trp-8 to the total indole emission is 60% and that of Trp-212 is 40%. The tryptophan fluorescence quantum yield is constant in the pH range 3-9. The fluorescence spectrum resulting from the simultaneous excitation of the tyrosyl and tryptophyl residues at 280 nm is dominated by the indole fluorophores: 61% of the light absorbed by the tyrosyl side chains is transferred to the two indole rings. Iodide and caesium are not efficient quenchers of the proteinase K tryptophan fluorescence, which is explained by restricted access of the ions to the somewhat buried Trp side chains and by electrostatic repulsion of caesium ions. Acrylamide quenching proceeds via both a dynamic and a static process and the data show homogeneity of the indole fluorescence arising from fluorophores in similar environments. The activation energy for the thermal deactivation of the excited tryptophans is 54 kJ mol-1. This value is substantially higher than those found for other proteinases from microorganisms and arises from the thermostability of proteinase K. Photooxidation of proteinase K in the presence of proflavine follows the kinetics of a first order reaction. The two tryptophans differ in their photoreactivity, Trp-212 being considerably more reactive.

Savinase (EC3.4.21.14) is secreted by the alkalophilic bacterium Bacillus lentus and is a representative of that subgroup of subtilisin enzymes with maximum stability in the pH range 7 to 10 and high activity in the range 8 to 12. It is therefore of major industrial importance for use in detergents. The crystal structure of the native form of Savinase has been refined using X-ray diffraction data to 1.4 A resolution. The starting model was that of subtilisin Carlsberg. A comparison to the structures of the closely related subtilisins Carlsberg and BPN' and to the more distant thermitase and proteinase K is presented. The structure of Savinase is very similar to those of homologous Bacillus subtilisins. There are two calcium ions in the structure, equivalent to the strong and the weak calcium-binding sites in subtilisin Carlsberg and subtilisin BPN', well known for their stabilizing effect on the subtilisins. The structure of Savinase shows novel features that can be related to its stability and activity. The relatively high number of salt bridges in Savinase is likely to contribute to its high thermal stability. The non-conservative substitutions and deletions in the hydrophobic binding pocket S1 result in the most significant structural differences from the other subtilisins. The different composition of the S1 binding loop as well as the more hydrophobic character of the substrate-binding region probably contribute to the alkaline activity profile of the enzyme. The model of Savinase contains 1880 protein atoms, 159 water molecules and two calcium ions. The crystallographic R-factor [formula; see text].

Bacitracins are a group of widely used peptide antibiotics. There has been interest in determining the three-dimensional structure of the bacitracins. However, solution studies indicate significant flexibility in their structure and to date native bacitracins have resisted attempts at crystallisation despite considerable efforts over a number of years by several groups. Here we report the first three-dimensional X-ray structure of a bacitracin, complexed to a subtilisin proteinase. X-Ray diffraction data were collected using synchrotron radiation in combination with the Image Plate Scanner system. The complex structure including two enzymes, two bacitracins, 220 water molecules and two Ca2+ ions was refined by restrained least-squares to a crystallographic R factor ( = sigma [[Fo-Fc]]/sigma [Fo]]) of 16.3% at 2.0 A.

Crystals of purified 5 S rRNA from Thermus flavus have been obtained. The crystals diffract up to 8 A resolution, using synchrotron radiation, and have the monoclinic space-group C2. The unit cell has the dimensions a = 190 A, b = 110 A, c = 138 A and beta = 117 degrees. The cell volume suggests the presence of four 5 S rRNA molecules per asymmetric unit.

We compare the three-dimensional structures of thermitase and of proteinase K determined by X-ray crystallography to a resolution of 1.4 and 1.48 A respectively. Both enzymes are relatively stable towards heat and denaturating agents and are representative of a subgroup of subtilisins characterized by a free SH group close to the active site histidine. Even though they have low sequence homology, the overall tertiary structures are highly conserved. The high resolution structures are compared in terms of the overall fold of the molecules, the active sites, the calcium binding sites, disulphide bridge positions, the positions of the charged residues and the solvent structure. Most subtilisins such as thermitase are of prokaryotic origin and proteinase K is up to now the only known eukaryotic structure.

Thermitase, the thermostable alkaline protease from Thermoactinomyces vulgaris, has been crystallised in a 1:1 complex with eglin, the inhibitor from the medical leech. Two large crystals were grown, with cell dimensions of a = 49.3 A, b = 67.3 A, c = 90.5 A and space group P2(1)2(1)2(1). The crystals are relatively tightly packed with Vm = 2.1 A3/Da. Three-dimensional data to 1.9 A have been recorded from one of these crystals. The orientation and position of the complex in the unit cell have been established using the subtilisin Carlsberg-eglin structure as a model. The structure of the complex is being refined by restrained least-squares. The present crystallographic R factor (= sigma parallel Fo - Fc parallel/sigma/Fo parallel) is 26% at 2.5 A resolution.

In the crystal structure of the polyiodide complex (p-nitrophenyl-alpha-maltohexaose(2)) . Ba(I(3))(2) . 22H(2)O, the maltohexaose units form an antiparallel, left-handed double helix with O-2 ... O-3 and O-6 ... O-6 hydrogen bonding and a central cavity that encloses two triiodide units. This structure contrasts with the parallel, left-handed double helix with no central cavity proposed for the A-and B-starch helix and the left-handed single helix in V-amylose and may be relevant for the stabilization of glycogen Structure.

The crystal and molecular structure of proteinase K was determined by X-ray diffraction data to 0.15-nm resolution. The enzyme belongs to the subtilisin family with an active-site catalytic triad Asp39--His69--Ser224 but is a representative of a subgroup with a free Cys73 close to and 'below' the active His69. Besides this Cys72, proteinase K has two disulfide bonds, Cys34--Cys123 and Cys178--Cys249, which contribute to the stability of the tertiary structure consisting of an extended central parallel beta-sheet decorated by six alpha-helices, three short antiparallel beta-sheets, 18 beta-turns and involving several internal, structurally important water molecules. Proteinase K exhibits two Ca2+-binding sites, one very strong and the other weak, which were the sites of the heavy atoms (Pb2+, Sm3+) used to solve the crystal structure. The weak binding site is liganded to the N and C termini, Thr16 and Asp260, and is only incompletely coordinated by oxygen ligands. The strong binding site is coordinated in the form of a pentagonal bipyramid with the side chain carboxylate of Asp200 and the C = O of Pro175 as apex, and C = O of Val177 and four water molecules in the equatorial plane. Upon removal of this Ca2+, proteinase K loses activity which is interpreted in terms of a local structural deformation involving the substrate-recognition site (Ser132--Gly136), probably associated with a cis----trans isomerization of cis Pro171. Several water molecules are located in the active site. One, W335, is positioned in the 'oxyanion hole' and is displaced by the C = O of the scissile peptide bond of the substrate, as indicated by crystallographic studies with peptide chloromethane inhibitors. Based on these experiments, a reaction mechanism is proposed where the peptide substrate forms a three-stranded antiparallel pleated sheet with the recognition site of proteinase K consisting of Ser132--Leu133--Gly134 on one side and Gly100--Ser101 on the other, followed by expulsion of the oxyanion hole water W335 and hydrolytic cleavage by the Asp39--His69--Serr224 triad. These latter residues display low thermal motion corresponding to well-defined geometry and are hardly accessible to solvent molecules, whereas the recognition-site amino acids are more flexible and partially exposed to solvent.

The crystal structure of the complex of thermitase with eglin-c in crystal form II, obtained in the presence of 5 mM-CaCl2, has been determined at 1.98 A resolution. The structure was solved by a molecular replacement method, then molecular dynamics crystallographic refinement was started using the thermitase-eglin-c structure as determined for crystal form I. A ten degrees rigid body misplacement of the core of eglin-c was corrected by the molecular dynamics crystallographic refinement without any need for manual rebuilding on a graphics system. A final crystallographic R-factor of 16.5% was obtained for crystal form II. The comparison of the complexes of thermitase with eglin-c in the two crystal forms shows that the eglin-c cores are differently oriented with respect to the protease. The inhibiting loop of eglin binds in a similar way to thermitase as to subtilisin Carlsberg. A tryptophanyl residue at the S4 site explains the preference of thermitase for aromatic residues of the substrate at the P4 site. The difference in the P1 binding pocket, asparagine in thermitase instead of glycine in subtilisin Carlsberg, does not change the binding of eglin-c. The preference for an arginyl residue at the P1 site of thermitase can be explained by the hydrogen bonding with Asn170 in thermitase. Three ion-binding sites of thermitase have been identified. The strong and weak calcium-binding sites resemble the equivalent sites of subtilisin Carlsberg and subtilisin BPN', though there are important amino acid differences at the calcium-binding sites. The medium-strength calcium-binding site of thermitase is observed in the subtilisin family for the first time. The calcium is bound to residues from the loop 57 to 66. A difference in chelation is observed at this site between the two crystal forms of thermitase, which differ in calcium concentration. Additional electron density is observed near Asp60 in crystal form II, which has more calcium bound than form I. This density is possibly due to a water molecule ligating the calcium ion or the result of Asp60 assuming two significantly different conformations.

Many attempts have been made to elucidate the three-dimensional structure from elongation factor Tu, but so far the only crystals suitable for X-ray crystallography contained a partially degraded protein. Here, we report the crystallization of a fully active, intact EF-Tu from thermus aquaticus. The crystals belong to hexagonal space group P6(3)(22) and diffract up to 2.6 A. The cell dimensions are a = b = 178 A, c = 238 A and 6 molecules are contained per asymmetric unit.

Proteinase K (EC 3.4.21.14) from the fungus Tritirachium album Limber is the most active known serine endopeptidase. The sequence of its 275-residue long polypeptide chain and its three-dimensional folding show a high degree of homology with the bacterial subtilisin proteases. Using difference Fourier methods, the binding mode of the synthetic carbobenzoxy-Ala-Ala-chloromethyl ketone inhibitor to the active site of proteinase K was determined. In several cycles of restrained least-squares, the enzyme-inhibitor complex was refined to a current R = 22% for 9400 X-ray diffraction data between 2.2 and 5.0 A resolution. The inhibitor is attached to proteinase K by two covalent bonds: one between the methylene carbon of the inhibitor and N epsilon 2 of the catalytic His 68, the other between the ketone carbon atom of the inhibitor and O gamma of the catalytic Ser 221. In addition, two hydrogen bonds donated by the peptide NH of Ser 221 and by the side chain NH2 of Asn 160 hold the hemiketal O- in the oxyanion hole. The peptide inhibitor is further hydrogen bonded to the proteinase polypeptide chain in a three-stranded antiparallel pleated sheet.

One of the three wheat germ inhibitors of proteinase K is bifunctional and inhibits simultaneously proteinase K (or subtilisin but not enzymes of the trypsin family) and insect alpha-amylase. The molecular mass of this inhibitor called PKI-3 is 21 kDa, and the binding constant for proteinase K is 0.8 nM at pH 8.2, 25 degrees C, in 1:1 molar ratio. PKI-3 was crystallized by microdialysis against 10-12% polyethylene glycol 6000, 50 mM NaH2PO4, pH 6.7. The crystals have monoclinic space group P2(1) with a = 42.5, b = 65.3, c = 31.5 A, beta = 110 degrees, and diffract beyond 2.0 A resolution. The complex proteinase K X PKI-3 was crystallized by equilibrium vapor diffusion under the same conditions. The crystals are needle-shaped and still too small for X-ray analysis. Gel electrophoresis established the composition of the crystals.

The structure of the serine endopeptidase proteinase K (279 amino acid residues; 28,790 daltons) has been refined by restrained least-squares methods to a conventional R value of 16.7% employing synchrotron film data of 30,812 reflections greater than 3 sigma in the 5.0 to 1.5 A resolution range. During refinement, the molecular structure was restrained to known stereochemistry, with root-mean-square (r.m.s.) deviation of 0.015 A from ideal bond lengths. The average atomic temperature factor, B, is 11.1 A2 for all atoms. The final model comprises 2020 protein atoms and 174 solvent molecules (which were given unit occupancies). Four corrections to the amino acid sequence were made, which were confirmed later by sequence analysis of the proteinase K gene: a deletion of one glycine in position 80; a change of sequence in position 207-208 and insertions of the dipeptide 210-211 and of residue 270. The r.m.s. deviation in the alpha-C atomic positions between the final refined model and the initial model built on the basis of a 3.3 A mini-map is 1.72 A for 227 out of 266 residues, which were originally traced in the mini-map without sequence information. The positions of the remaining 39 residues deviate by more than 8 A from the original ones and are located in regions where extensive revision of the structural model was necessary.

Cyclodextrins (CD's) have proved useful as model systems for the study of hydrogen bonding. They are torus-shaped molecules composed of six(alpha), seven(beta) or eight(gamma) (1----4) linked glucoses. Because of their particular geometry, they are able to act as a "host" to form inclusion complexes with "guest" molecules very much like enzymes. Cyclodextrins have been shown to exert catalytic activity on suitable included-substrate molecules; they catalyze the hydrolysis of phenylacetates, of organic pyrophosphates and of penicillin derivatives. They also accelerate aromatic chlorinations and diazo coupling by means of their primary and/or secondary hydroxyl groups, so that the rates of hydrolysis are enhanced by up to a factor of 400. In order to understand the hydrogen bonding in these enzyme models, neutron diffraction data were collected to unambiguously determine the hydrogen atom positions, which could not be done from the x-ray diffraction data. alpha-CD has been shown to have two different structures with well-defined hydrogen bonds, one "tense" and the other "relaxed". An "induced-fit"-like mechanism for alpha-CD complex formation has been proposed. Circular hydrogen bond networks have also been found for alpha-CD due to the energetically favored cooperative effect. beta-CD with a disordered water structure possesses an unusual flip-flop hydrogen bonding system of the type O-H...H-O representing an equilibrium between two states: O-H...O in equilibrium O...H-O. gamma-CD with a disordered water structure similar to beta-CD also possesses the flip-flop hydrogen bond. This study demonstrates that hydrogen bonds are operative in disordered systems and display dynamics even in the solid state.

The crystal structure of lipoamide dehydrogenase from Azotobacter vinelandii has been determined by a combination of molecular replacement and isomorphous replacement techniques yielding eventually a good-quality 2.8 A electron density map. Initially, the structure determination was attempted by molecular replacement procedures alone using a model of human glutathione reductase, which has 26% sequence identity with this bacterial dehydrogenase. The rotation function yielded the correct orientation of the model structure both when the glutathione reductase dimer and monomer were used as starting model. The translation function could not be solved, however. Consequently, data for two heavy-atom derivatives were collected using the Hamburg synchotron facilities. The derivatives had several sites in common, which was presumably a major reason why the electron density map obtained by isomorphous information alone was of poor quality. Application of solvent flattening procedures cleaned up the map considerably, however, showing clearly the outline of the lipoamide dehydrogenase dimer, which has a molecular weight of 100,000. Application of the "phased translation function", which combines the phase information of both isomorphous and molecular replacement, led to an unambiguous determination of the position of the model structure in the lipoamide dehydrogenase unit cell. The non-crystallographic 2-fold axis of the dimer was optimized by several cycles of constrained-restrained least-squares refinement and subsequently used for phase improvement by 2-fold density averaging. After ten cycles at 3.5 A, the resolution was gradually extended to 2.8 A in another 140 cycles. The 2.8 A electron density distribution obtained in this manner was of much improved quality and allowed building of an atomic model of A. vinelandii lipoamide dehydrogenase. It appears that in the orthorhombic crystals used each dimer is involved in contacts with eight surrounding dimers, leaving unexplained why the crystals are rather fragile. Contacts between subunits within one dimer, which are quite extensive, can be divided into two regions separated by a cavity. In one of the contact regions, the level of sequence identity with glutathione reductase is very low but it is quite high in the other. The folding of the polypeptide chain in each subunit is quite similar to that of glutathione reductase, as is the extended conformation of the co-enzyme FAD.(ABSTRACT TRUNCATED AT 400 WORDS)