Correlation of polarized absorption spectroscopic and X-ray diffraction studies of crystalline cytosolic aspartate aminotransferase of pig hearts

From protein structure to function via single crystal optical spectroscopy

The more than 100,000 protein structures determined by X-ray crystallography provide a wealth of information for the characterization of biological processes at the molecular level. However, several crystallographic “artifacts,” including conformational selection, crystallization conditions and radiation damages, may affect the quality and the interpretation of the electron density maps, thus limiting the relevance of structure determinations. Moreover, for most of these structures, no functional data have been obtained in the crystalline state, thus posing serious questions on their validity in infereing protein mechanisms. In order to solve these issues, spectroscopic methods have been applied for the determination of equilibrium and kinetic properties of proteins in the crystalline state. These methods are UV-vis spectrophotometry, spectrofluorimetry, IR, EPR, Raman, and resonance Raman spectroscopy. Some of these approaches have been implemented with on-line instruments at X-ray synchrotron beamlines. Here, we provide an overview of investigations predominantly carried out in our laboratory by single crystal polarized absorption UV-vis microspectrophotometry, the most applied technique for the functional characterization of proteins in the crystalline state. Studies on hemoglobins, pyridoxal 5′-phosphate dependent enzymes and green fluorescent protein in the crystalline state have addressed key biological issues, leading to either straightforward structure-function correlations or limitations to structure-based mechanisms.

Polarized absorption spectra of green fluorescent protein single crystals: transition dipole moment directions

Polarized absorption spectra of orthorhombic crystals of wild-type green fluorescent protein (GFP) were measured between 350 and 520 nm to obtain information on the directions of the electronic transition dipole moments ((-->)m) of the chromophore relative to the molecular axes. The transition dipole moment orientation is a basic spectroscopic parameter of relevance to biologists when interpreting Förster-type fluorescence resonance energy transfer data and for comparing absorbance and fluorescence spectra of GFP with quantum chemical calculations. Maximal extinction was obtained throughout the spectrum when the polarization direction of the electric vector of incident light was parallel to the c-axis of the crystal. The transition dipole moments were assumed to be parallel to the plane of the chromophore. With this assumption and the measured dichroic ratios in the crystals, the transition dipole moments associated with the neutral (lambda(max) = 398 nm) and anionic (lambda(max) = 478 nm) forms of the chromophore were found to subtend angles of approximately 26 degrees and 13 degrees (counterclockwise), respectively, with the vector that joins the phenolic and imidazolinone oxygen atoms of the chromophore.

Structure of mammalian ornithine decarboxylase at 1.6 A resolution: stereochemical implications of PLP-dependent amino acid decarboxylases

BACKGROUND

Pyridoxal-5'-phosphate (PLP) dependent enzymes catalyze a broad range of reactions, resulting in bond cleavage at C alpha, C beta, or C gamma carbons of D and L amino acid substrates. Ornithine decarboxylase (ODC) is a PLP-dependent enzyme that controls a critical step in the biosynthesis of polyamines, small organic polycations whose controlled levels are essential for proper growth. ODC inhibition has applications for the treatment of certain cancers and parasitic ailments such as African sleeping sickness.

RESULTS

The structure of truncated mouse ODC (mODC') was determined by multiple isomorphous replacement methods and refined to 1.6 A resolution. This is the first structure of a Group IV decarboxylase. The monomer contains two domains: an alpha/beta barrel that binds the cofactor, and a second domain consisting mostly of beta structure. Only the dimer is catalytically active, as the active sites are constructed of residues from both monomers. The interactions stabilizing the dimer shed light on its regulation by antizyme. The overall structure and the environment of the cofactor are compared with those of alanine racemase.

CONCLUSIONS

The analysis of the mODC' structure and its comparison with alanine racemase, together with modeling studies of the external aldimine intermediate, provide insight into the stereochemical characteristics of PLP-dependent decarboxylation. The structure comparison reveals stereochemical differences with other PLP-dependent enzymes and the bacterial ODC. These characteristics may be exploited in the design of new inhibitors specific for eukaryotic and bacterial ODCs, and provide the basis for a detailed understanding of the mechanism by which these enzymes regulate reaction specificity.

Catalytic competence of O-acetylserine sulfhydrylase in the crystal probed by polarized absorption microspectrophotometry

The reactions of the pyridoxal 5'-phosphate-dependent enzyme O-acetylserine sulfhydrylase with the substrate O-acetyl-L-serine and substrate analogs have been investigated in the crystalline state by single-crystal polarized absorption microspectrophotometry. This approach has allowed us to examine the catalytic competence of the enzyme in different crystalline states, one of which was used to determine the three-dimensional structure; experimental conditions were defined for the accumulation of catalytic intermediates in the crystal suitable for crystallographic analyses.O-Acetyl-L-serine reacts with the enzyme in one of the crystal forms leading via a beta-elimination reaction to the accumulation of the alpha-aminoacrylate Schiff base, absorbing maximally at 320 and 470 nm, as in solution. The dissociation constant for the alpha-aminoacrylate Schiff base is in the millimolar range, 500-fold higher than in solution, suggesting that crystal lattice interactions may oppose functionally relevant conformational changes. The dissociation constant exhibits a bell-shaped dependence on pH centered at pH 7. At this pH the alpha-aminoacrylate species slowly decays with time (30% decrease in 24 hours). The alpha-aminoacrylate intermediate readily reacts with sodium azide, an analog of sulfide, the natural nucleophilic agent, to give a new amino acid and the native enzyme, indicating that the crystalline enzyme catalyzes the overall beta-replacement reaction as in solution. In other crystal forms, including that used for the X-ray investigation, O-acetyl-L-serine either has an even higher dissociation constant or causes crystal damage upon binding. When the crystalline enzyme reacts with either L-cysteine or L-serine, the external aldimine intermediate is formed. The dissociation constants for both substrate analogs are closer to those observed in solution and are modulated by pH as in solution. Findings demonstrate that O-acetylserine sulfhydrylase is catalytically competent in the crystal although some regions of the molecule, likely involved in an open-closed transition induced by O-acetyl-L-serine binding, may have a limited flexibility. The accumulation in the crystal of both the external aldimine and the alpha-aminoacrylate intermediate makes feasible their structural determination and, therefore, the elucidation of the catalytic pathway at the molecular level.

Aspartate aminotransferase complexed with erythro-beta-hydroxyaspartate: crystallographic and spectroscopic identification of the carbinolamine intermediate

The crystal structure of mitochondrial aspartate aminotransferase (mAAT) of chicken complexed with erythro-beta-hydroxyaspartate has been determined at 2.4 A resolution. Pregrown crystals of mAAT complexed with the inhibitor maleate (closed enzyme conformation, orthorhombic space group C222(1)) were soaked in solutions of erythro-beta-hydroxyaspartate. The ligand exchange was monitored by microspectrophotometry. The active site turned out to be predominantly occupied by the carbinolamine intermediate. The carbinolamine is a true intermediate of the catalytic cycle forming the last covalently bound enzyme:substrate complex before release of the keto acid product. Occupancies of approximately 80% for the carbinolamine and of approximately 20% for the quinonoid intermediate were obtained. Two hydrogen bonds were identified that are potentially relevant for the accumulation of the carbinolamine intermediate: one to the hydroxyl group of Tyr 70* and the other to the epsilon-NH2 group of Lys 258.

The reaction catalyzed by Escherichia coli aspartate aminotransferase has multiple partially rate-determining steps, while that catalyzed by the Y225F mutant is dominated by ketimine hydrolysis

The mechanism of transamination catalyzed by Escherichia coli wild-type aspartate aminotransferase (AATase) and the mutant AAtase in which Tyr-225 is converted to Phe (Y225F) was investigated. The absorbance spectrum of wild-type AATase in the presence of excess L-Asp and oxalacetate is dominated by species absorbing near 330 nm. The primary C alpha 2H-Asp kinetic isotope effects (KIEs) on reactions catalyzed by wild-type AAtase at pH 8.9 and 7.5 on kcat/KMAsp are approximately 2, and the KIEs on kcat are 1.9 (pH 8.9) and 1.4 (pH 7.5). The C alpha 2H-Asp KIEs on reactions catalyzed by Y225F are near unity at both pH values. The solvent deuterium KIEs (SKIEs) on kcat for reactions with L-Asp catalyzed by wild-type AATase and Y225F at their pH/pD maxima approximately 2, and the SKIE on kcat/kMAsp is increased from 1.3 to 2.3 by the mutation. The C4' (S)-2H-pyridoxamine 5'-phosphate KIE values on reactions of alpha-ketoacids with both enzymes are near unity. The viscosity effects on kcat/KMAsp and kcat for wild-type AAtase at pH 9 are 0.10 and 0.31, respectively, indicating that the reaction is partially diffusion limited. The viscosity effects on kcat/KMAsp and kcat for Y225F are reduced to -0.02 and 0.06, respectively, indicating that the mutant catalyzed reaction is almost fully chemistry-limited. A free-energy profile for the L-Asp-to-oxalacetate half-reaction was constructed for wild-type AAtase. C alpha H abstraction, ketimine hydrolysis, and oxalacetate dissociation are partially rate-determining. Ketimine hydrolysis is the sole rate-determining step for the corresponding Y225F- catalyzed reaction.

Transaldimination induces coenzyme reorientation in pig kidney dopa decarboxylase

Dopa decarboxylase from pig kidney is a pyridoxal-5'-phosphate (PLP) dependent enzyme that displays positive circular dichroism (CD) relative to the coenzyme absorption bands at 335 and 420 nm, which are characteristic of an asymmetrically bound coenzyme. It has been previously noted that the presence of various substrates and substrate analogs gives rise to similar absorbance changes, independently of whether or not the enzyme-ligand interaction is accompanied by the conversion of the internal aldimine to an external aldimine. The effects of various ligands on the CD spectral properties of the enzyme bound PLP are presented herein. It was observed that changes in the optical activity are seen only in the presence of ligands capable of Schiff base formation. These results imply that a reorientation of the pyridoxal phosphate ring occurs upon formation of an external aldimine. Moreover, external aldimines formed with catecholic and indolic compounds are characterized by quite dissimilar optical activities, suggesting that with respect to vicinal residues, different coenzyme microenvironments exist in these complexes.

Photophysical study of the Schiff bases of 5'-deoxypyridoxal and n-hexylamine in cationic micelles

The absorption and fluorescence spectra of the Schiff bases formed between 5'-deoxypyridoxal and n-hexylamine in aqueous media containing different concentrations of the cationic surfactant hexadecyltrimethylammonium bromide were recorded at 25 degrees C. The quantum yields of fluorescence of the different zwitterionic and enol forms of the chemical species of the Schiff bases occurring in media of pH 4.5-8.5 were determined. Also, the fluorescence quenching resulting from the presence of the surfactant and that of iodide ion were analyzed. From the results obtained it follows that the zwitterionic forms do not interact with the cationic surfactant, whereas the enol forms do interact with it.

Crystal structures of true enzymatic reaction intermediates: aspartate and glutamate ketimines in aspartate aminotransferase

The crystal structures of the stable, closed complexes of chicken mitochondrial aspartate aminotransferase with the natural substrates L-aspartate and L-glutamate have been solved and refined at 2.4- and 2.3-A resolution, respectively. In both cases, clear electron density at the substrate-coenzyme binding site unequivocally indicates the presence of a covalent intermediate. The crystallographically identical environments of the two subunits of the alpha 2 dimer allow a simple, direct correlation of the coenzyme absorption spectra of the crystalline enzyme with the diffraction results. Deconvolution of the spectra of the crystalline complexes using lognormal curves indicates that the ketimine intermediates constitute 76% and 83% of the total enzyme populations with L-aspartate and L-glutamate, respectively. The electron density maps accommodate the ketimine structures best in agreement with the independent spectral data. Crystalline enzyme has a much higher affinity for keto acid substrates compared to enzyme in solution. The increased affinity is interpreted in terms of a perturbation of the open/closed conformational equilibrium by the crystal lattice, with the closed form having greater affinity for substrate. The crystal lattice contacts provide energy required for domain closure normally supplied by the excess binding energy of the substrate. In solution, enzyme saturated with amino/keto acid substrate pairs has a greater total fraction of intermediates in the aldehyde oxidation state compared to crystalline enzyme. Assuming the only difference between the solution and crystalline enzymes is in conformational freedom, this difference suggests that one or more substantially populated, aldehydic intermediates in solution exist in the open conformation. Quantitative analyses of the spectra indicate that the value of the equilibrium constant for the open-closed conformational transition of the liganded, aldehydic enzyme in solution is near 1. The C4' pro-S proton in the ketimine models is oriented nearly perpendicularly to the plane of the pyridine ring, suggesting that the enzyme facilitates its removal by maximizing sigma-pi orbital overlap. The absence of a localized water molecule near Lys258 dictates that ketimine hydrolysis occurs via a transiently bound water molecule or from an alternative, possibly more open, structure in which water is appropriately bound. A prominent mechanistic role for flexibility of the Lys258 side chain is suggested by the absence of hydrogen bonds to the amino group in the aspartate structure and the relatively high temperature factors for these atoms in both structures.

Quantitative description of the absorption spectra of the coenzyme in glycogen phosphorylases based on log-normal distribution curves

The absorption spectra of the coenzyme [pyridoxal 5'-phosphate (PLP)] in glycogen phosphorylase a (GPha), glycogen phosphorylase b (GPhb) and of the latter bound to various effectors and substrates were analysed on the basis of log-normal distribution curves. The results obtained showed that the ionization state of the PLP and GPha environment differs from that of GPhb. This divergence was interpreted in terms of tautomeric equilibria between some forms of the Schiff base of PLP and enzymic Lys-679. The ionic forms are slightly more predominant in GPha than they are in GPhb, so ionic and/or hydrogen-bonding interactions between the aromatic ring of PLP and GPha must be stronger than with GPhb. This confirms the purely structural role of the aromatic ring of the coenzyme. Binding of GPhb to AMP and Mg2+ results in the coenzyme adopting a similar state as in GPha. On the other hand, binding to IMP gives rise to no detectable changes in the tautomeric equilibrium of the coenzyme.

Domain closure in mitochondrial aspartate aminotransferase

The subunits of the dimeric enzyme aspartate aminotransferase have two domains: one large and one small. The active site lies in a cavity that is close to both the subunit interface and the interface between the two domains. On binding the substrate the domains close together. This closure completely buries the substrate in the active site and moves two arginine side-chains so they form salt bridges with carboxylate groups of the substrate. The salt bridges hold the substrate close to the pyridoxal 5'-phosphate cofactor and in the right position and orientation for the catalysis of the transamination reaction. We describe here the structural changes that produce the domain movements and the closure of the active site. Structural changes occur at the interface between the domains and within the small domain itself. On closure, the core of the small domain rotates by 13 degrees relative to the large domain. Two other regions of the small domain, which form part of the active site, move somewhat differently. A loop, residues 39 to 49, above the active site moves about 1 A less than the core of the small domain. A helix within the small domain forms the "door" of the active site. It moves with the core of the small domain and, in addition, shifts by 1.2 A, rotates by 10 degrees, and switches its first turn from the alpha to the 3(10) conformation. This results in the helix closing the active site. The domain movements are produced by a co-ordinated series of small changes. Within one subunit the polypeptide chain passes twice between the large and small domains. One link involves a peptide in an extended conformation. The second link is in the middle of a long helix that spans both domains. At the interface this helix is kinked and, on closure, the angle of the kink changes to accommodate the movement of the small domain. The interface between the domains is formed by 15 residues in the large domain packing against 12 residues in the small domain and the manner in which these residues pack is essentially the same in the open and closed structures. Domain movements involve changes in the main-chain and side-chain torsion angles in the residues on both sides of the interface. Most of these changes are small; only a few side-chains switch to new conformations.(ABSTRACT TRUNCATED AT 400 WORDS)

Spectroscopic study of the Schiff bases of dodecylamine with pyridoxal 5'-phosphate and 5'-deoxypyridoxal. A model for the Schiff bases of pyridoxal 5'-phosphate in biological systems

We recorded the absorption spectra of the Schiff bases of pyridoxal 5'-phosphate (PLP) and 5'-deoxypyridoxal (DPL) with dodecylamine (DOD) at different pH values. By applying deconvolution techniques to the spectra and analysing their different components we found that the above-mentioned Schiff bases in aqueous solutions of pH 7 adopted a conformation in which the pyridine ring is embedded in a very hydrophobic medium from which water is virtually completely excluded. This conformation in the same as that adopted by PLP when it acts as coenzyme for some enzymes such as glycogen phosphorylase. The experimental results obtained also show such a conformation to be highly favoured but sensitive to the protonation of the pyridine nitrogen, which makes the aromatic ring more readily accessible to the solvent.

Linear dichroism studies of tryptophanase and its quasisubstrate complexes oriented in polyacrylamide gel

Tryptophanase from Escherichia coli was oriented in a compressed slab of polyacrylamide gel and its linear dichroism (LD) and absorption spectra have been measured. The free enzyme displays four LD bands at 305, 340, 425 and 490 nm. Two bands at 340 and 425 nm belong to the internal coenzyme-lysine aldimine. The 305-nm band apparently belongs to an aromatic amino acid residue. The 490-nm band disappears after treatment with NaBH4 or after incubation with L-alanine and subsequent dialysis. It is suggested that the 490-nm band belongs to a quinonoid enzyme subform. The reaction of tryptophanase with threo-3-phenyl-DL-serine, L-threonine and D-alanine leads to formation of an external aldimine with an intense absorption band at 420-425 nm. The values of reduced LD (delta A/A) in this band strongly differ from that in the 420-nm band of the free enzyme. The LD value of the complex with D-alanine is intermediate between those of the free enzyme and the complex with 3-phenylserine. In the presence of indole the complex with D-alanine displays the same LD as that observed with 3-phenylserine. The reaction of tryptophanase with L-alanine or oxindolyl-L-alanine leads to formation of a quinonoid intermediate with an absorption band near 500 nm. The LD value in this band is close to that of an external aldimine with L-threonine. It is concluded that reorientations of the coenzyme occur in the course of the tryptophanase reaction.

Aspartate aminotransferase: investigation of the active sites

An investigation of the crystal structure of cytosolic pig-heart aspartate aminotransferase (AAT, E.C.2.6.1.1) was carried out to determine the structural requirements for ligand recognition by the active site. Structural differences were observed between the two active sites of the AAT dimer. The natural ligand, L-aspartate, was docked into both active sites using various methods. However, due to structural differences, the ligand was able to form all the necessary interactions for initial binding in only one of the active sites. The program GRID (P.J. Goodford, J. Med. Chem. 1985, 28, 849-857) was used to predict favorable binding sites for the functional groups of the aspartate ligand. These binding sites corresponded to the position of the docked aspartate ligand, indicating that substrate recognition takes place before any major conformational changes occur within the enzyme.

Reactions of phosphonate analogs of pyridoxal phosphate with apo-aspartate aminotransferase

We have investigated reactions of the 5-phosphonoethyl and 5-phosphonoethenyl analogs of pyridoxal 5'-phosphate in the coenzyme site of cytosolic aspartate aminotransferase. Acid dissociation constants and equilibrium constants for hydration and for tautomerization have been evaluated for these compounds. In confirmation of previous results, both compounds are partially active. They bind to apoenzyme well and undergo conversion in the presence of glutamate to amine forms which show induced circular dichroism comparable to that of native enzyme. A normal "external" Schiff base is evidently formed with 2-methylaspartate, but the amounts of quinonoid intermediate formed with erythro-3-hydroxyaspartate are less than those formed with pyridoxal phosphate. The pKa of the imine group of the enzyme reconstituted with the phosphonoethyl analog is more than two units lower than that in the native enzyme. Binding of the dicarboxylates glutarate, 2-oxoglutarate, and succinate shifts the pKa upward. The absorption spectra of the resulting complexes indicate the existence of at least three low pH species. A shift of 2.3 to 2.9 ppm to a lower frequency was observed for the 31P NMR signal upon binding of these dicarboxylates or of 2-methylaspartate. Enzyme containing the analogs crystallizes. Polarized absorption spectra suggest that the coenzyme has an orientation similar to that of pyridoxal phosphate in the native enzyme.