Theoretical investigations of the hydrolysis pathway of verdoheme to biliverdin

Conversion of iron(II) verdoheme to iron biliverdin in the presence of OH(-) was investigated using B3LYP method. Both 3-21G and 6-31G* basis sets were employed for geometry optimization calculation as well as energy stabilization estimation. Calculation at 6-31G* level was found necessary for a correct spin state estimation of the iron complexes. Two possible pathways for the conversion of iron verdoheme to iron biliverdin were considered. In one path the iron was six-coordinate while in the other it was considered to be five-coordinate. In the six-coordinated pathway, the ground state of bis imidazole iron verdoheme is singlet while that for open chain iron biliverdin it is triplet state with 4.86 kcal/mol more stable than the singlet state. The potential energy surface suggests that a spin inversion take place during the course of reaction after TS. The ring opening process in the six-coordinated pathway is in overall -2.26 kcal/mol exothermic with a kinetic barrier of 9.76 kcal/mol. In the five-coordinated pathway the reactant and product are in the ground triplet state. In this path, hydroxyl ion attacks the iron center to produce a complex, which is only 1.59 kcal/mol more stable than when OH(-) directly attacks the macrocycle. The activation barrier for the conversion of iron hydroxy species to the iron biliverdin complex by a rebound mechanism is estimated to be 32.68 kcal/mol. Large barrier for rebound mechanism, small barrier of 4.18 kcal/mol for ring opening process of the hydroxylated macrocycle, and relatively same stabilities for complexes resulted by the attack of nucleophile to the iron and macrocycle indicate that five-coordinated pathway with direct attack of nucleophile to the 5-oxo position of macrocycle might be the path for the conversion of verdoheme to biliverdin.     

Theoretical investigation of the ring opening process of verdoheme to biliverdin in the presence of dioxygen

The conversion of ferrous verdoheme to ferric biliverdin in the presence of O₂ was investigated using the B3LYP method. Both 6-31G and 6-31G (d) basis sets were employed for geometry optimization calculation as well as energy stabilization estimation. Three possible pathways for the conversion of iron verdoheme to iron biliverdin were considered. In the first route oxygen and reducing electron were employed. In this path formation of ferrous verdoheme-O₂ complex was followed by the addition of one electron to the ferrous-oxycomplex to produce ferric peroxide intermediate. The ferric peroxide intermediate experienced an intramolecular nucleophilic attack to the most positive position at 5-oxo carbons on the ring to form a closed ring biliverdin. Subsequently the ring opening process took place and the iron (III) biliverdin complex was formed. Closed ring iron biliverdin intermediate and open ring iron biliverdin formed as a product of verdoheme cleavage were respectively 13.20 and 32.70 kcal mol⁻¹ more stable than ferric peroxide intermediate. Barrier energy for conversion of ferric peroxide to closed ring Fe (III) biliverdin and from the latter to Fe (III) biliverdin were respectively 8.67 and 3.35 kcal mol⁻¹. In this path spin ground states are doublet except for iron (III) biliverdin in which spin state is quartet. In the second path a ferrous-O₂ complex was formed and, without going to a one electron reduction process, nucleophilic attack of iron superoxide complex took place followed by the formation of iron (III) biliverdin. This path is thermodynamically and kinetically less favorable than the first one. In addition, iron hydro peroxy complex or direct attack of O₂ to macrocycle to form an isoporphyrin type intermediate have shown energy surfaces less favorable than aforementioned routes.     

Stereoselectivity of each of the three steps of the heme oxygenase reaction: hemin to meso-hydroxyhemin,meso-hydroxyhemin to verdoheme,and verdoheme to biliverdin

Heme oxygenase catalyzes the regiospecific oxidation of hemin to biliverdin IXalpha with concomitant liberation of CO and iron by three sequential monooxygenase reactions. The alpha-regioselectivity of heme oxygenase has been thought to result from the regioselective oxygenation of the heme alpha-meso position at the first step, which leads to the reaction pathway via meso-hydroxyheme IXalpha and verdoheme IXalpha intermediates. However, recent reports concerning heme oxygenase forming biliverdin isomers other than biliverdin IXalpha raise a question whether heme oxygenase can degrade meso-hydroxyhemin and isomers other than the alpha-isomers. In this paper, we investigated the stereoselectivity of each of the two reaction steps from meso-hydroxyhemin to verdoheme and verdoheme to biliverdin by using a truncated form of rat heme oxygenase-1 and the chemically synthesized four isomers of meso-hydroxyhemin and verdoheme. Heme oxygenase-1 converted all four isomers of meso-hydroxyhemin to the corresponding isomers of verdoheme. In contrast, only verdoheme IXalpha was converted to the corresponding biliverdin IXalpha. We conclude that the third step, but not the second, is stereoselective for the alpha-isomer substrate. The present findings on regioselectivities of the second and the third steps have been discussed on the basis of the oxygen activation mechanisms of these steps.     

Reactions of benzocyclobutene chromium complexes with carbon, nitrogen and oxygen nucleophiles: nucleophilic addition and unexpected proximal ring opening reactions

Tricarbonyl(η⁶-1-oxobenzocyclobutene)chromium(0) (1) can be transformed to tricarbonyl(η⁶-1-endo-hydroxybenzocyclobutene) chromium(0) derivatives with substituents R (R=CH₃, CH=CH₂, (CH₂)₄CH=CH₂, (CH₂)₄OSi(Me)₂tBu) at Cl on the exo face of the complex. The relative configuration is proven by an X-ray crystal structure analysis of the trimethylsilyl ether 8 (C₁₆H₁₈CrO₄Si: a=8.693(1), b=9.490(1), c=11.063(1) Å, =97.51(1), β=110.32(1), γ=95.38(1)°, triclinic, space group P (No.2), R=0.037, R_w=0.052 for 4609 observed reflections. Attempts directed at an intramolecular cycloaddition of the ortho-quinodimethane complex derived from 17 by anion promoted ring opening unexpectedly resulted in the formation of 12 as the product of an opening of the proximal bond of the anellated ring located between the hydroxy group and the coordinated aromatic ring in 16. The fact that the intermolecular cycloaddition reaction for 16 is possible in the presence of a dienophile is taken as evidence for an equilibrium between the alcoholate 17 and the two ring opened products 16 and 18. The proximal ring opening of 6 is not observed when the free organic ligand 21 is used as the educt. Ketone complexes 1 and 25 undergo proximal ring opening reaction when treated with alcoholate or primary amines.     

Separation and identification of the regioisomers of verdoheme by reversed-phase ion-pair high-performance liquid chromatography, and characterization of their complexes with heme oxygenase

We report an HPLC method for separating the four regioisomers of verdoheme formed in the coupled oxidation of hemin with oxygen and ascorbate in aqueous pyridine. The reversed-phase ion-pair system uses hexafluoroacetone and pyridine as ion-pair agents. The regiochemistry of the separated isomers was established both by HPLC of the corresponding biliverdin IX derivatives and by 1H NMR of each isomer. Optical spectra of the pyridine verdohemochrome isomers were similar to each other, but showed differences in the absorption maxima in the red region, which appear at 680, 663, 648 and 660 nm for the alpha, beta, gamma, and delta-isomers, respectively. Each of the four isomers was incorporated anaerobically into heme oxygenase-1, yielding the corresponding verdoheme-enzyme complex. The ferrous forms had absorption maxima at 690, 667, 655, and 663 nm, and their CO-bound forms had maxima at 638, 624, 616, and 626 nm for alpha, beta, gamma, and delta-isomer, respectively. Addition of ferricyanide to the alpha-verdoheme-heme oxygenase complex brought about a ferric low-spin heme-like signal, which is identical with the ferric alpha-verdoheme complexed with the heme oxygenase that was observed in the heme oxygenase reaction.     

Mild, rapid and selective alcoholysis of terpyridine-appended amide substrates by Cu-catalysis: protonation state and reactivity of the complex

An amide bond of a terpyridine-appended substrate, 6-(L-phenylalanylamino)-2,2′:6′,2″-terpyridine (2a), was cleaved to yield phenylalanine methyl ester quantitatively in the presence of catalytic amounts of Cu²⁺ not only in methanol but also in aqueous methanol at 30 °C. The reaction proceeds via formation of an N₃O (three terpyridine nitrogens and one carbonyl oxygen) type 1:1 metal complex 2a-Cu²⁺. From spectral titration, the structure of the 2a-Cu²⁺ complex was confirmed to have three different protonation states, i.e., A (non-deprotonated amide with α-ammonium), D1 (deprotonated-amide with α-ammonium) and D2 (deprotonated-amide with α-amino) states. Among them, the complex in the D2 state was exclusively responsible for the observed mild, rapid and selective alcoholysis, showing the first-order rate constant of 6 × 10⁻³ s⁻¹ or half-decay time of 2 min in methanol at 30 °C. Acidity of the amide proton was found to be higher than that of the ammonium proton in the complex, allowing formation of the highly reactive amide-deprotonated D2 state in methanol even without addition of external bases. Factors contributing to the high reactivity of the complex were discussed.     

Theoretical study of the structure of adenosine deaminase complexes with adenosine analogues: I. Aza-, deaza- and isomeric azadeazaanalogues of adenosine

The conformational models of the active site of adenosine deaminase (ADA) and its complexes in the basic state with adenosine and 13 isosteric analogues of the aza, deaza, and azadeaza series were constructed. The optimization of the conformational energy of the active site and the nucleoside bound with it in the complex was achieved in the force field of the whole enzyme (the 1ADD structure was used) within the molecular mechanics model using the AMBER 99 potentials. The stable conformational states of each of the complexes, as well as the optimal conformation of the ADA in the absence of ligand, were determined. It was proved that the conformational state that is close to the structure of the ADA complex with 1-deazaadenosine (1ADD) known from the X-ray study corresponds to one of the local minima of the potential surface. Another, a significantly deeper minimum was determined; it differs from the first minimum by the mutual orientation of side chains of amino acid residues. A similar conformational state is optimal for the ADA active site in the absence of the bound ligand. A qualitative correlation exists between the values of potential energies of the complexes in this conformation and the enzymatic activity of ADA toward the corresponding nucleosides. The dynamics of conformational conversions of the active site after the binding of substrate or its analogues, as well as the possibility of the estimation of the inhibitory properties of nucleosides on the basis of calculations, are discussed.     

Evolution of erect helical colony form in the Bryozoa: phylogenetic, functional, and ecological factors

Erect helical colony forms have evolved at least six separate times within the Bryozoa, but only among those in which branches are composed of a single layer of feeding zooids. The four best known genera with helical colony forms evolved independently, and each occupied different benthic marine environments, achieved different growth habits, and utilized different aspects of an array of functional potentials resulting from the radially symmetrical colonies. Examination of the distribution of these four genera ( Archimedes , Bugula , Crisidmonea , and Retiflustra ) within a theoretical morphospace of hypothetical helical colony form reveals that each occupies its own characteristic region of morphospace, broadly overlapping in some dimensions but separated in others. The genera differ markedly in the branching densities within their filtration-sheet whorls, reflecting their phylogenetic legacies rather than constructional or functional constraints associated with helical growth. In contrast, all tend toward helices in which the radiating whorls of the unilaminate branches are held at an average of 50–60° to the central axis of the colony, and this may reflect a common functional optimum associated with the cilia-driven, auto-generated currents within the helix. The region of morphospace characterized by high values of surface area – i.e. helical geometries with branches orientated at very low angles to the central axis, and with very closely spaced whorls along the axis – is entirely empty of bryozoans, and these geometries apparently represent functionally unrealistic colony forms.  © 2003 The Linnean Society of London, Biological Journal of the Linnean Society , 2003, 80 , 235–260.     

Chemistry and reactivity of dinuclear iron oxamate complexes: alkane oxidation with hydrogen peroxide catalysed by an oxo-bridged diiron(III) complex with amide and carboxylate ligation

A new dinuclear iron(III) complex with the tetradentate ligand N,N^′-o-phenylenebis(oxamate) (opba) has been synthesised, and structurally, magnetically and electrochemically characterised. It possesses an unprecedented triply bridged Fe₂(μ-O)(μ-RCO₂?H₂O?O₂CR)₂ core, whereby two N-amides from the opba ligand complete the square-pyramidal coordination sphere of the O-carboxylate rich iron site (Fe-N=2.053 Å and Fe-O=2.015 Å). The antiferromagnetic exchange interaction between the two high-spin Fe^III ions through the oxo bridge (J=−190 cm⁻¹; H=−JS₁·S₂) is weaker than that found in related μ-oxo singly bridged diiron(III) complexes. The lessened antiferromagnetic coupling correlates to the remarkably long Fe-O distance (R=1.808 Å), which is associated to the relatively bent Fe-O-Fe angle (α=131.8°) in the Fe₂O unit. It experiences an irreversible one-electron oxidation process in acetonitrile at 25 °C (E=0.63 V versus SCE) to give the putative mixed-valent diiron(III,IV) species which is unstable under the experimental conditions. The reactivity of this unique oxo-bridged diiron(III) oxamate complex toward hydrogen peroxide activation and hydrocarbon oxidation in the presence of dioxygen has been examined. It catalyses the oxidation of alkanes like cyclohexane and adamantane to the corresponding alcohols and ketones by H₂O₂ and O₂ in acetonitrile at room temperature with limited catalytic activities (total yields of ca. 5% after 24 h) and modest selectivities (A/K=0.9, KIE=2.4, tert/sec=3.0). Overall, these results are more in line with a mechanism involving alkyl radicals as transient intermediates, although they do not exclude the possibility that a metal-based active oxidant is also involved in this C-H bond oxidation chemistry.     

Biomimetic metal-radical reactivity: aerial oxidation of alcohols, amines, aminophenols and catechols catalyzed by transition metal complexes

The contributions of the authors to the research program 'Radicals in Enzymatic Catalysis' over the last ca. 5 years are summarized. Significant efforts were directed towards the design and testing of phenol-containing ligands for synthesizing radical-containing transition metal complexes as potential candidates for catalysis of organic substrates like alcohols, amines, aminophenols and catechols. Functional models for different copper oxidases, such as galactose oxidase, amine oxidases, phenoxazinone synthase and catechol oxidase, are reported. The copper complexes synthesized can mimic the function of the metalloenzymes galactose oxidase and amine oxidases by catalyzing the aerial oxidation of alcohols and amines. Even methanol could be oxidized, albeit with a low conversion, by a biradical-copper(II) compound. The presence of a primary kinetic isotope effect, similar to that for galactose oxidase, provides compelling evidence that H-atom abstraction from the alpha-C-atom of the substrates is the rate-limiting step. Although catechol oxidase and phenoxazinone synthase contain copper, manganese(IV) complexes containing radicals have been found to be useful to study synthetic systems and to understand the naturally occurring processes. An 'on-off' mechanism of the radicals without redox participation from the metal centers seems to be operative in the catalysis involving such metal-radical complexes.     

Stabilization of microtubules by inorganic phosphate and its structural analogues, the fluoride complexes of aluminum and beryllium

In order to elucidate how the elementary reactions of GTP cleavage and subsequent inorganic phosphate (Pi) release, which accompany microtubule assembly, regulate microtubule dynamics, the effect of Pi and of its structural analogues AlF4- and BeF3- on the stability of GDP-microtubules has been investigated. Inorganic phosphate binds to microtubules with a low affinity (KD = 25 mM) and slows down the rate of GDP-subunit dissociation by about 2 orders of magnitude. AlF4- and BeF3- exhibit phosphate-like effects with 1000-fold higher affinity. Evidence has been obtained for direct binding of BeF3- to microtubules with a stoichiometry of 1 mol of BeF3- per mole of GDP-subunit and an equilibrium dissociation constant of 12-15 microM. AlF4- and Pi compete for this site. Phosphate analogues abolish oscillatory polymerization kinetics and slow down microtubule turnover at steady state. In view of these results, we propose that Pi and its structural analogues bind to the site of the gamma-phosphate of GTP in the E site and reconstitute a GDP-Pi-microtubule, from which tubulin subunits dissociate very slowly. We therefore understand that, following GTP cleavage on microtubules, Pi release in the medium is accompanied by a structural change resulting in a large destabilization of the polymer. A cap of slowly dissociating GDP-Pi-subunits prevents depolymerization of the microtubule GDP-core at steady state. The similarity with the actin system [Carlier, M.-F., & Pantaloni, D. (1988) J. Biol. Chem. 263, 817-825] is underlined.     

Imidazole,imidazolate, and hydroxide complexes of (protoporphyrin IX)iron(III) and its dimethyl ester as model systems for ferric hemoproteins: Electron paramagnetic resonance and electronic spectral study

The EPR and electronic spectral changes upon titration of systems consisting of (protoporphyrin IX)iron(III) chloride (Fe(PPIX)Cl) or its dimethyl ester (Fe-(PPIXDME)Cl) and imidazole derivatives with tetrabutylammonium hydroxide solution have been measured at 77 and 298 °K in various solvents. The EPR and electronic spectra of the melt of Fe(PPIXDME)Cl in imidazole derivatives have been also measured. The imidazole derivatives studied here were imidazole and 4-methyl-, 4-phenyl-, 2-methyl-, 2,4-dimethyl-, 1-methyl-, and 1-acetylimidazole. The spectral changes upon addition of hydroxide were markedly different between the systems containing NH imidazoles (BH), with a dissociable proton, and those containing NR imidazoles (BR), without it. In the former systems, five spectral species were successively formed at 77 °K and were assigned to following complexes: [Fe(P)(BH)₂]⁺, Fe(P)(BH)(B), [Fe(P)(B)₂]^?, Fe(P)(BH)(OH), and [Fe(P)(B)(OH)]^?, where P is PPIX or PPIXDME. In the latter systems, initial complex, [Fe(P)(BR)₂]⁺, was found to be changed to final complex, Fe(P)(BR)(OH), through an intermediate at 77 °K. At 298 °K, both systems were found to react with hydroxide to finally form Fe(P)(OH). The crystal field parameters were evaluated using the EPR g values in low-spin complexes studied here and in hemoproteins. The five regions corresponding to five low-spin complexes could be distinguished in crystal field diagrams.     

Interaction of amide inhibitors with the active site of carbonic anhydrase: metal-induced deprotonation of the bound amide group is indicated by slow binding kinetics, by visible spectra of complexes with cobalt enzyme, and by pH effects on binding affinity

Most carbonic anhydrase (CA) inhibitors bind at the active site metal and either are anions or are capable of deprotonation to yield anions. Much less is known about the interaction of CA with inhibitors that have hitherto been considered to bind as neutral species. We report a study of the reversible amide inhibition of Co(II)-substituted CA by iodoacetamide and ethyl carbamate (urethane), as well as the ambivalent oxamate, the monoamide of oxalate. Visible cobalt spectral changes indicate coordination of all these inhibitors to the metal. The pH dependence of the affinity of carbonic anhydrase isozyme I (CA I) for ethyl carbamate and iodoacetamide is formally consistent with their binding either as anionic species to the acid form of the enzyme or as neutral species to the basic form of the enzyme. The former view is in better accord with the spectral data. Most strikingly, reversible binding of iodoacetamide and ethyl carbamate leads to uniquely slow kinetics of ligand association and dissociation that could be followed by simple mixing. The slow association kinetics suggest the involvement of energetically unfavorable deprotonation of the amide group preceding final coordination. The complex pH profile for inhibition of CA I by the ambivalent oxamate is consistent with coordination through the carboxylate group at low pH and through the deprotonated amide group at high pH. The visible spectrum of the complex of Co(II)CA I with oxamate shows a parallel dependence on pH, reflecting this dual coordination mode. Similarly, oxamate dissociation kinetics were biphasic and could be correlated with the pH-dependent spectral changes.(ABSTRACT TRUNCATED AT 250 WORDS)     

Monoclonal antibody against chicken type IX collagen: preparation, characterization, and recognition of the intact form of type IX collagen secreted by chondrocytes

A series of monoclonal antibodies was prepared against the pepsin-resistant fragment of type IX collagen designated HMW. One of these antibodies (called 2C2) was selected for further analysis. Antibody 2C2 showed no cross-reactivity with other collagen types by inhibition enzyme-linked immunosorbent assays. It recognized an epitope present in native HMW, but failed to recognize any of the three chains of HMW fractionated after denaturation followed by reduction and alkylation of interchain disulfide bridges. Electron microscopic observations after rotary shadowing showed that the location of the epitope for antibody 2C2 was close to the carboxy-terminus of HMW. Immunofluorescent staining of sections of embryonic and adult cartilage with antibody 2C2 after removal of proteoglycans by testicular hyaluronidase digestion showed that type IX collagen is distributed throughout the cartilage matrix, and is not present in other connective tissues or skeletal muscle. The intact type IX collagen molecule, which was secreted by a suspension culture of freshly isolated embryonic chick chondrocytes, was recognized by rotary shadowing in the presence of antibody 2C2 after first precipitating the procollagens from the culture medium with ammonium sulfate (30%). Two different collagenous molecules were present in the precipitate: a longer molecule of type II procollagen (average length, 335 nm) with both amino- and carboxy-propeptides still remaining uncleaved, and a shorter molecule (average length, 190 nm) which was identified as type IX collagen. Antibody 2C2 consistently bound to the shorter molecules at a site located 136 nm from a distinctive knob at one end of the molecule, and did not bind to any specific site on the type II procollagen molecules. The structure of the intact type IX collagen molecule with the location of both collagenous and noncollagenous domains was as predicted after converting the nucleotide sequence of a cDNA clone encoding for one of the chains of type IX collagen to an amino acid sequence (Ninomiya, Y., and B. R. Olsen, 1984, Proc. Natl. Acad. Sci. USA, 81:3014-3018).     

CoMFA and CoMSIA investigations of dopamine D3 receptor ligands leading to the prediction, synthesis, and evaluation of rigidized FAUC 365 analogues

Taking advantage of our in-house experimental data on dopamine D3 receptor modulators, we have successfully established highly significant CoMFA and CoMSIA models (q(cv)2 = 0.82/0.76). These models were carefully investigated to assure their stability and predictivity (r(pred)2 = 0.65/0.61) and subsequently applied to guide experimental investigations on the synthesis and receptor binding of three conformationally restricted D3 ligands. Besides the high D3 affinity, the test compound 45, incorporating a trans-1,4-cyclohexylene partial structure, exhibited improved (approximately 3200-fold) selectivity over the D4 subtype.     

Selective recognition of metal complexes by macrocyclic ethers: further observations on the macrocycle size dependence and the first-sphere ligand composition dependence of recognition thermodynamics

Additional studies of solution phase recognition of Ru(NH₃)_x(pyridine)_y²⁺, Ru(NH₃)_x(2,2′-bipyridine)_y²⁺ and Ru(NH₃)₄(1,10-phenantroline)²⁺ species by dibenzo crown ethers are reported. The factors most closely examined were crown size, ammine ligand content and ancillary ligand composition. The overall study confirms that recognition or association derives primarily from H-bond formation (ammine hydrogen/ether oxygen. Evidently opposing these interactions, however, are crown conformational rearrangements. Consequently, straight-forward correlatins between association strength and potential number of H-bond interactions are found only in selected cases. Based on comparisons of association constants for (bis) pyridine, bipyridine and phenanthroline ligand-containing species with dibenzo crowns, evidence is also found for favorable polypyridine/benzene interactions. NMR (NOE) measurements indicate that the preferred association geometrics in solution are those that make each of the benzenes of the crown coplanar (or nearly coplanar) with the ligated polypyridine.     

Biomimetic hydrolytic activation by Fe(III) aggregates: structures,reactivity and properties of novel oxo-bridged iron complexes

The tetranuclear aggregate (enH(2))[Fe(4)(mu(3)-O)(heidi)(4)(mu-O,O'-O(2)CNHC(2)H(4)NH(3))] x 4H(2)O contains a novel bidentate zwitterionic carbamic acid ligand. Magnetic studies indicate that the unsymmetrical Fe(4) core is ferrimagnetic with an S=4 ground state. Similar ligands have been obtained on rectangular tetranuclear aggregates [M(4)(mu-O)(mu-OH)(hpdta)(2)(mu-X)(2)](n-) (M[double bond]Fe, Al, Ga). The carbamic acid ligands are considered to result from the hydrolytic activation (fixation) of atmospheric CO(2) by the aggregate precursor to give a carbonato intermediate, which then reacts with the organic diamine used as base in the synthesis. Similar aggregates with acetate ligands result from hydrolytic activation of the DMA used as cosolvent. Closely related mechanisms for these two activation processes are proposed, which are also related to the accepted mechanisms for carbonic anhydrase and urease.     

Structural milestones in the reaction pathway of an amide hydrolase: substrate,acyl, and product complexes of cephalothin with AmpC beta-lactamase

Beta-lactamases hydrolyze beta-lactam antibiotics and are the leading cause of bacterial resistance to these drugs. Although beta-lactamases have been extensively studied, structures of the substrate-enzyme and product-enzyme complexes have proven elusive. Here, the structure of a mutant AmpC in complex with the beta-lactam cephalothin in its substrate and product forms was determined by X-ray crystallography to 1.53 A resolution. The acyl-enzyme intermediate between AmpC and cephalothin was determined to 2.06 A resolution. The ligand undergoes a dramatic conformational change as the reaction progresses, with the characteristic six-membered dihydrothiazine ring of cephalothin rotating by 109 degrees. These structures correspond to all three intermediates along the reaction path and provide insight into substrate recognition, catalysis, and product expulsion.