Probing the substrate specificity of the intracellular brain platelet-activating factor acetylhydrolase

Platelet-activating factor acetylhydrolases (PAF-AHs) are uniquePLA2s which hydrolyze the sn-2 ester linkage in PAF-like phospholipidswith a marked preference for very short acyl chains, typicallyacetyl. The recent solution of the crystal structure of the₁ catalytic subunit of isoform Ib of bovine brain intracellularPAF-AH at 1.7 Å resolution paved the way for a detailedexamination of the molecular basis of substrate specificityin this enzyme. The crystal structure suggests that the sidechains of Thr103, Leu48 and Leu194 are involved in substraterecognition. Three single site mutants (L48A, T103S and L194A)were overexpressed and their structures were solved to 2.3 Åresolution or better by X-ray diffraction methods. Enzyme kineticsshowed that, compared with wild-type protein, all three mutantshave higher relative activity against phospholipids with sn-2acyl chains longer than an acetyl. However, for each of themutants we observed an unexpected and substantial reductionin the V_max of the reaction. These results are consistent withthe model in which residues Leu48, Thr103 and Leu194 indeedcontribute to substrate specificity and in addition suggestthat the integrity of the specificity pocket is critical forthe expression of full catalytic function, thus conferring veryhigh substrate selectivity on the enzyme.     

Comparison of backbone structures of glucose isomerase from Streptomyces and Arthrobacter

The C backbones of the glucose isomerase molecules of Streptomycesrubiginosus and Arthrobacter have been determined by X-ray crystallographyand compared. Each molecule is a tetramer of eight-stranded/ß barrels, and the mode of association of the tetramersis identical in each case. The Arthrobacter electron densityshows four additional amino acids at the carboxyl terminus.There is also an insertion of six amino acids at position 277,and two individual insertions at about positions 348 and 357(numbering according to the Streptomyces structure). There isa close structural homology throughout the whole molecule, whichis most accurate up to position 325. The r.m.s. displacementfor 315 homologous C positions up to this position is 0.92 Å.     

Three-dimensional structures of chimeric enzymes between Bacillus subtilis and Thermus thermophilus 3-isopropylmalate dehydrogenases

The 3-D structures of two chimeric enzymes (4M6T and 2T2M6T)between the Bacillus subtilis and Thermus thermophilus 3-isopropylmalatedehydrogenases were analysed by X-ray diffraction in order toinvestigate their different thermostabilities. The structureof 2T2M6T was determined by the difference Fourier method andthat of 4M6T by rigid body refinement, as based on the structureof the T. thermophilus enzyme. These structures were refinedstereochemically to an R-factor of 0.193 at 2.5 Å resolutionfor 4M6T and to an R-factor of 0.195 at 2.2 Å resolutionfor 2T2M6T. The 3-D structures of 4M6T and 2T2M6T were veryclose to the structure of the T.thermophilus enzyme, conspicuousdifferences being at the molecular surface, In particular, 2T2M6Thaving a larger reduction in thermostability was more closelyrelated to the T.thermophilus enzyme. However, their correlationsbetween C-atom displacements and the root squares of the temperaturefactors were significantly different from each other.     

The 2 A crystal structure of subtilisin E with PMSF inhibitor

Using enzyme prepared by the DNA recombination technique, subtilisinE from Bacillus subtilis was crystallizedin space group P2₁2₁2₁with two molecules in an asymmetric unit. The crystal structureof PMSF-inhibited subtilisin E was solved by molecular replacementfollowed by refinement with the X-PLOR program. This resultedin the 2.0 Å structure of subtilisin E with an R-factorof 0.191 for 8–2 Å data and r.m.s. deviations fromideal values of 0.021 Å and 2.294° for bond lengthsand bond angles respectively. The PMSF group covalently boundto Ser221 appeared very clearly in the electron density map.Except for the active site disturbed by PMSF binding, the structuralfeatures of subtilisin E are almost the same as in other subtilisins.The calcium-binding sites are different in detail in the twoindependent molecules of subtilisin E. Based on the structure,the remarkably enhanced heat stability of mutant N118S of subtilisinE is discussed. It is very likely that there is an additionalwater molecule in the mutant structure, which is hydrogen bondedto side chains of Serll8 and its neighbouring residues Lys27and Asp 120.     

Disulfide stabilization of antibody Fv: computer predictions and experimental evaluation

Using molecular modeling technology we have recently identifiedpositions in conserved framework regions of Fvs which can beused to stabilize antibody Fvs by an interchain disulfide bondengineered in between the structurally conserved framework positionsof the variable domains of heavy (V_H) and light (V_L) immunoglobulinchains (disulfide-stabilized Fv; dsFv). The computer model indicatedthe existence of other potential sites in the framework regionsthat might be suitable for disulfide bond formation betweenV_H and V_L. The possibility of obtaining dsFvs using these positionsis evaluated here experimentally by constructing dsFv immunotoxinsin which the Fv moiety is fused to a truncated form of Pseudomonasexotoxin. We analyzed the extent of dsFv formation and the activityof the resulting dsFv immunotoxins, and compared various dsFvmolecules with the scFv immunotoxin. Our results demonstratethat position H44-L105 is the only one which gives high productionyields of active dsFv. All other positions gave either low yieldsand activity or completely failed to produce active dsFv. Withone exception, the formation and activities of the dsFvs correspondedto the C-C distance between the VH and VL positions, with anoptimal distance of 5.7 Å producing the best dsFv. Distancesof 6.0–6.9 Å resulted in a' low yield of proteinthat was still capable of binding antigen, whereas distances>7.0 Å resulted in molecules in which dsFv formationwas not obtained.     

The structure-function relationship of the lipases from Pseudomonas aeruginosa and Bacillus subtilis

Within the BRIDGE T-project on lipases we investigate the structure-functionrelationships of the lipases from Bacillus subtilis and Pseudomonasaeruginosa. Construction of an overproducing Bacillus. strainallowed the purification of > 100 mg lipase from 30 l culturesupernatant. After testing a large variety of crystallizationconditions, the Bacillus lipase gave crystals of reasonablequality in PEG-4000 (38-45%), Na₂SO₄ and octyl-ß-glucosideat 22°C, pH 9.0. A 2.5 Å; dataset has been obtainedwhich is complete from 15 to 2.5 A resolution. P.aeruginosawild-type strain PAC1R was fermented using conditions of maximumlipase production. More than 90% of the lipase was cell boundand could be solubilized by treatment of the cells with TritonX-100. This permitted the purification of 50 mg lipase. So far,no crystals of sufficient quality were obtained. Comparisonof the model we built for the Pseudomonas lipase, on the basisof sequences and structures of various hydrolases which werefound to possess a common folding pattern (/ß hydrolasefold), with the X-ray structure of the P.glumae lipase revealedthat it is possible to correctly build the structure of thecore of a protein even in the absence of obvious sequence homologywith a protein of known 3-D structure.     

Crystallization and structure solution at 4 A resolution of the recombinant synthase domain of N(5'-phosphoribosyl)anthranilate isomerase:indole-3-glycerol-phosphate synthase from Escherichia coli complexed to a substrate analogue

The recombinant synthase domain of the bifunctional enzyme N-(5'-phosphoribosyl)anthranilateisomerase:indole-3-glycerol-phosphate synthase from Escherichiacoli has been crystallized, and the structure has been solvedat 4 Å resolution. Two closely related crystal forms grownfrom ammonium sulphate diffract to 2 Å resolution. Oneform (space group R32, a = 163 Å, = 29.5°) containsthe unliganded synthase domain; the second crystal form (spacegroup P6₃22, a = 144 Å, c = 158 Å) is co-crystallizedwith the substrate analogue N-(5'-phosphoribit-1-yl)anthranilate.The structure of the synthase–inhibitor complex has beensolved by the molecular replacement method. This achievementrepresents the first successful use of a (ß)_g-barrelmonomer as a trial model. The recombinant synthase domain associatesas a trimer in the crystal, the molecules being related by apseudo-crystallographic triad. The interface contacts betweenthe three domains are mediated by those residues that are alsoinvolved in the domain interface of the bifunctional enzyme.This system provides a model for an interface which is usedin both intermolecular and intramolecular domain contacts.     

1.85 A structure of anti-fluorescein 4-4-20 Fab

The crystal complex of fluorescein bound to the high-affinityanti-fluorescein 4-4-20 Fab {K_a = 10¹⁰ M^–1 at 2°C)has been determined at 1.85 Å. Isomorphous crystals oftwo isoelectric forms (p1 = 7.5 and 7.9) of the antifluorescein4-4-20 Fab, an IgG2A [Gibson et al (1988)Proteins: Struct. FunctGenet., 3, 155–160], have been grown. Both complexes crystallizewith one molecule in the asymmetric unit in space group P1,with a = 42.75 Å, b =43.87 Å, c = 58.17 Å, = 95.15° , ß = 86.85° and = 98.01°.The final structure has an R value of 0.188 at 1.85 Åresolution. Interactions between bound fluorescein, the complementarity-determiningregions (CDRs) of the Fab and the active-site mutants of the4-4-20 single-chain Fv will be discussed. Differences were foundbetween the structure reported here and the previously reported2.7 Å 4-4-20 Fab structure [Herron et al. (1989) Proteins:Struct. Fund., 5, 271–280]. Our structure determinationwas based on 26 328 unique reflections — four times theamount of data used in the previous report. Differences in thetwo structures could be explained by differences in interpretingthe electron density maps at the various resolutions. The r.m.s.deviations between the variable and constant domains of thetwo structures were 0.77 and 1.54 Å, respectively. Fourregions of the light chain and four regions of the heavy chainhad r.m.s. backbone deviations of >4 Å. The most significantof these was the conformation of the light chain CDR 1.     

Modelling the polypeptide backbone with 'spare parts' from known protein structures

An automatic procedure for building a protein polyalanine backbonefrom C positions and ‘spare parts’ retrieved froma data base of 66 high-resolution protein structures is described.Protein backbones are constructed from over-lapping fragmentsof variable length, which allows the backbone of regular secondarystructure elements to be built in one block. The procedure isshown to yield backbones which compare very favourably withthose from highly refined X-ray structures (r.m.s. deviationbetween generated and crystal structures <lÅ). Themethod is furthermore quite insensitive to experimental errorsin C positions as well as to the size of the data base, andis seen to yield valuable insight into the relationships betweensequence and 3-D structure: one example on triose phosphateisomerase, a ß-barrel protein, shows that ßloops can be considered as structurally more uncommon than ßloops. The ‘spare parts’ approach is also foundto be useful for general-purpose modelling of local structuralchanges produced by insertion or deletion of residues. It should,however, be used with caution. Crude selection criteria basedsolely on fragment length and geometric fit to the loop baseregions yield realistic backbones in about two-thirds of thetest cases (r.m.s. deviations from refined crystal structure{small tilde}lÅ). In the remaining cases, sequence information,in particular the presence of glycine residues which tend toadopt more unusual backbone conformations, must be consideredto obtain comparable results.     

Crystal structure of human serum albumin at 2.5 A resolution

A new triclinic crystal form of human serum albumin (HSA), derivedeither from pool plasma (pHSA) or from a Pichia pastoris expressionsystem (rHSA), was obtained from polyethylene glycol 4000 solution.Three-dimensional structures of pHSA and rHSA were determinedat 2.5 Å resolution from the new triclinic crystal formby molecular replacement, using atomic coordinates derived froma multiple isomorphous replacement work with a known tetragonalcrystal form. The structures of pHSA and rHSA are virtuallyidentical, with an r.m.s. deviation of 0.24 Å for allC atoms. The two HSA molecules involved in the asymmetric unitare related by a strict local twofold symmetry such that theC atoms of the two molecules can be superimposed with an r.m.s.deviation of 0.28 Å in pHSA. Cys34 is the only cysteinewith a free sulfhydryl group which does not participate in adisulfide linkage with any external ligand. Domains II and IIIboth have a pocket formed mostly of hydrophobic and positivelycharged residues and in which a very wide range of compoundsmay be accommodated. Three tentative binding sites for long-chainfatty acids, each with different surroundings, are located atthe surface of each domain.     

Modular mutagenesis of a TEM-barrel enzyme: the crystal structure of a chimeric E.coli TIM having the eighth {beta}{alpha}-unit replaced by the equivalent unit of chicken TIM

Abstract The crystal structure of a hybrid Escherichia coli triosephosphateisomerase (TIM) has been determined at 2.8 Å resolution.The hybrid TIM (ETIM8CHI) was constructed by replacing the eighthß-unit of E.coli TIM with the equivalent unit of chickenTIM. This replacement involves 10 sequence changes. One of thechanges concerns the mutation of a buried alanine (Ala232 instrand 8) into a phenylalanine. The ETIM8CHI structure showsthat the A232F sequence change can be incorporated by a side-chainrotation of Phe224 (in helix 7). No cavities or strained dihedralsare observed in ETIM8CHI in the region near position 232, whichis in agreement with the observation that ETIM8CHI and E.coliTIM have similar stabilities. The largest CA (C-alpha atom)movements, 3 Å, are seen for the C-terminal end of helix8 (associated with the outward rotation of Phe224) and for theresidues in the loop after helix 1 (associated with sequencechanges in helix 8). From the structure it is not clear whythe k_cat of ETIM8CHI is 10 times lower than in wild type E.coliTIM     

Molecular modeling of coiled-coil {alpha}-tropomyosin: analysis of staggered and in register helix--helix interactions

In register and staggered models of tropomyosin coiled-coilwere built from X-ray C coordinates and refined via moleculardynamics. The two models show similar structural features withthe X-ray structure of GCN4 leucine zipper. Empirical energeticmethods used to compare the in register and staggered modelsindicate that both are equally probable. The two models havesimilar profiles of solvation free energy of folding for residuesat positions a and d of the repeating heptad, indicating thatresidues at these positions are as well buried in an in registerstructure as in a staggered one. Neither the in register northe 14 residues staggered structure can be ruled out based onhydrophobic or e–g' (g–e') electrostatic interactionswhich are not able to distinguish between the two models andare therefore not selective. However, the eg–b'c' electrostaticinteractions, although smaller in magnitude, are in favor ofthe in register model. Furthermore, analysis of hydrophobicand electrostatic interactions along the tropomyosin sequenceshows that bulky residues in positions a and d prevent the formationof inter-chain salt bridges.     

Site-directed mutagenesis and X-ray crystallography of two phospholipase A2 mutants: Y52F and Y73F

Tyr52 and Tyr73 are conserved amino acid residues throughoutall vertebrate phospholipases A₂. They are part of an extendedhydrogen bonding system that links the N-terminal -NH⁺₃ -groupto the catalytic residues His48 and Asp99. These tyrosines werereplaced by phenylalanines in a porcine pancreatic phospholipaseA₂ mutant, in which residues 62–66 had been deleted (62–66PLA₂).The mutations did not affect the catalytic properties of theenzyme, nor the folding kinetics. The stability against denaturatlonby guanidine hydrochloride was decreased, however. To analysehow the enzyme compensates for the loss of the tyrosine hydroxylgroup, the X-ray structures of the Y52F and AY73F mutants weredetermined. After crystallographic refinement the final crystallographicR-factors were 18.1% for the %Y52F mutant (data between 7 and2.3 Å resolution) and 19.1% for the Y73F mutant (databetween 7 and 2.4 Å resolution). No conformational changesoccurred in the mutants compared with the 62–66PLA₂, butan empty cavity formed at the site of the hydroxyl group ofthe former tyrosine. In both mutants the Asp99 side chain losesone of its hydrogen bonds and this might explain the observeddestabilization.     

Modeling the uncleaved serpin antichymotrypsin and its chymotrypsin complex

In order to provide a structural reference for protein engineeringexperiments involving the serpin ₁-antichymotrypsin (ACT) andits complexes with chymotrypsin and DNA, a homology model ofACT has been constructed based on the 3-D structure of the relatedprotein ovalbumin [29% identical and 44% similar; see Stein,P., Leslie, A., Finch, J., Turnell, W., McLaughlin, P. and Carrell,R. (1990) Nature, 347, 99–102]. After mapping the aminoacid sequence of ACT onto the peptide backbone of ovalbumin,the resulting model was subjected to simulated annealing andenergy minimization. Overall, the final ACT model is structurallysimilar to ovalbumin, although the 2.4 Å root mean squaredeviation of corresponding C atoms reflects the presence ofregions exhibiting notable structural differences. The hydrogenbond stereochemistry of the ACT model is consistent with patternsfound in high resolution protein structures and 92% of its backboneatoms have acceptable conformations when evaluated in a Ramachandrananalysis. Significantly, the homology model serves as a structuralreference for protein engineering experiments aimed at redesigningthe functional properties of ACT, particularly with regard toits protease-bound conformation. Additionally, the homologymodel may be useful as a probe for solving the crystal structuresof certain ACT variants (e.g. Thr345 Arg) by molecular replacementmethods. Ultimately, the homology approach may be applied towardthe construction of other serpin models starting with an experimentallydetermined structure of uncleaved ACT as a template.     

The 3.0 A crystal structure of xylose isomerase from Streptomyces olivochromogenes

The crystal structure of xylose isomerase [E.C. 5.3.1.5 [EC] ] fromStreptomyces olivochromogenes has been determined to 3.0 Åresolution. The crystals belong to space group P22₁2₁ with unitcell parameters a = 98.7, b = 93.9, c = 87.7. The asymmetricunit contains half of a tetrameric molecule of 222 symmetry.The two-fold axis relating the two molecules in the asymmetricunit is close to where a crystallographic two-fold would beif the space group were 1222. This causes the diffraction patternto have strong 1222 pseudo-symmetry, so all data were collectedin this pseudo-space group. Since the sequence of this enzymehas not been reported, a polyalanine backbone has been fittedto the electron density. Xylose isomerase has two domains: theN-terminal domain is an eight-stranded /ß barrel of299 residues. The C-terminal domain is a large loop of 50 residueswhich is involved in inter-molecular contacts. Comparison ofxylose isomerase with the archetypical /ß barrel protein,triose phosphate isomerase, reveals that the proteins overlapbest when the third (ß) strand of xylose isomeraseis superimposed on the first (ß) strand of triosephosphate isomerase. This same overlap has also been found betweenthe muconate lactonising enzyme and triose phosphate isomerase[Goldman et al. (1987) J. Mol. Biol., in press].     

Interactions of (Ala*Ala*Lys*Pro)n and (Lys*Lys*Ser*Pro)n with DNA. Proposed coiled-coil structure of AlgR3 and AlgP from Pseudomonas aeruginosa

The proteins, AlgR3 and AlgP, are involved in the regulationof alginate synthesis in Pseudomonas. They contain multiplerepeats of Ala^*Ala^*Lys^*Pro as do several other proteins thatresemble histones. The interactions of synthesis oligopeptidescomposed of repeated Ala^*Ala^*Lys^*Pro or Lys^*Lys^*Ser^*Pro unitswith DNA were studied by fluorescence of the Fmoc (9-fluorenylmethyloxycarbonyl)group attached to the N-termini of the peptides. DNA quenchingof the Fmoc fluorescence of the peptides was used to estimatethe apparent association constants for the interaction of Fmoc(AAKP)_nOH(n = 2, 4, 8, 18, 32) and of Fmoc(KKSP)_nOH (n = 2, 4, 8, 16,20, 32) with DNA. The Fmoc(AAKP)_nOH peptides bind to DNA onlyat low ionic strength; the Fmoc(KKSP)_n OH peptides interactwith DNA at both low (0.05 M KCl) and high (0.2 M KCl) saltAt low ionic strength an increase in the number of the repeatunits causes an increase in the apparent association constantup to {small tilde}2 x 10⁶ M^–1 for both types of peptidesat N 24. The insertion of an AAKTA unit into the middle ofthe Fmoc(AAKP)₈OH peptide increases its affinity to DNA. Wepropose a model of (AAKP)_n and of its interaction with DNA.The repeat unit consists of a single turn of -helix followedby a bend necessitated by Pro. The resultant coiled-coil formsa right-handed superhelix with 10 AAKPs per repeat distanceof {small tilde}33 Å. With only slight modification ofthe canonical parameters of this model the AAKP super helixfits into the major groove of B-form DNA with one AAKP tetramerper base pair repeat of 3.4 Å. The -amine nitrogen ofLys can form a polar hydrogen bond with a phosphate oxygen atomof the DNA backbone. A better fit is obtained when the modelis modified to accommodate [(AAKP)₅AAKTA]_n as actually observedin AlgR3. We suggest that this coiled-coil represents a generalmotif for other protein–DNA interactions.     

An 8-fold {beta}{alpha} barrel protein with redundant folding possibilities

Protein sequences containing redundant segments of secondarystructure at both termini have the choice a priori of foldinginto several possible circularly permuted variants of the wild-typetertiary structure. To test this hypothesis the gene of phosphoribosylanthranilate isomerase from yeast, which is a single-domain8-fold ß barrel protein, was modified to produce a10-fold ß homologue in Escherichia coli. It containeda duplicate of the two C-terminal ß units of supersecondarystructure fused to its N-terminus. Most of the protein was recoveredfrom the insoluble fraction of disrupted cells by dissolutionin guanidinium chloride solutions and refolding. Pristine proteinwas purified from the soluble fraction. The purified (ß)₁₀proteins were enzymically almost fully active. Absorbance, fluorescenceand circular dichroism spectra as well as the reversible unfoldingbehaviour of both proteins were also very similar to the propertiesof the original (ß)₈ protein. Digestion with endopeptidasesconverted both the pristine and the refolded (ß)₁₀variant to the same large fragment that had the N-terminal sequenceand mol. wt of the wild-type ß)₈ protein. The datasuggest that the folding of the (ß)₁₀ variant is controlledthermodynamically both in vivo and in vitro.     

High-level expression, purification, kinetic characterization and crystallization of protein farnesyltransferase -subunit C-terminal mutants

Protein farnesyltransferase (FPT) is a 97 000 Da heterodimericenzyme that catalyzes post-translational farnesylation of manycellular regulatory proteins including p21 Ras. To facilitatethe construction of site-directed mutants, a novel translationallycoupled, two-cistron Escherichia coli expression system forrat FPT has been developed. This expression system enabled yieldsof >5 mg of purified protein per liter of E.coli cultureto be obtained. The E.coli-derived FPT demonstrated an activitycomparable to that of protein isolated from other sources. Thereported expression system was used to construct three ß-subunitC-terminal truncation mutants, 5, 10 and 14, which were designedto eliminate a lattice interaction between the ß-subunitC-terminus of one molecule and the active site of a symmetry-relatedmolecule. Steady-state kinetic analyses of these mutants showedthat deletion up to 14 residues at the C-terminus did not reducethe value of k_cat; however, K_m values for both peptide and FPPincreased 2–3-fold. A new crystalline form of FPT was obtainedfor the 10 C-terminal mutant grown in the presence of the substrateanalogs acetyl-Cys-Val-Ile-Met-COOH peptide and -hydroxyfarnesylphosphonicacid. The crystals diffract to beyond 2.0 Å resolution.The refined structure clearly shows that both substrate analogsadopt extended conformations within the FPT active site cavity.     

Functional roles and subsite locations of Leu177, Trp178 and Asn182 of Aspergillus awamori glucoamylase determined by site-directed mutagenesis

Fungal glucoamylases contain four conserved regions. One regionfrom the Aspergillus niger enzyme contains three key carboxylicacid residues, the general acid catalytic group, Glu179, alongwith Asp176 and Glu180. Three site-directed mutations, Leu177– His, Trp178 – Arg and Asn182 – Ala, wereconstructed near these acidic groups to reveal the functionof other conserved residues in this region. Leu177 and Trp178are strictly conserved among fungal glucoamylases, while anamide, predominantly Asn, always occurs at position 182. Substitutionsof Leu177 or Trp178 cause significant decreases in k_cat withthe substrates tested. Similar increases in activation energiesobtained with Leu177 – His with both -(1,4)- and -(1,6)-linkedsubstrates indicate Leu177 is located in subsite 1. K_M valuesobtained with the Trp178 – Arg mutation increase for an-(1,6)-linked substrate, but not for -(1,4)-linked substrates.Calculated differences in activation energy between substratesindicate Trp178 interacts specifically with subsite 2. The Asn182 Ala mutation did not change k_cat or K_M values, indicating thatAsn182 is not crucial for activity. These results support amechanism for glucoamylase catalytic activity consisting ofa fast substrate binding step followed by a conformational changeat subsite 1 to stabilize the transition state complex.     

The structure of a designed peptidomimetic inhibitor complex of {alpha}-thrombin

Thrombin displays remarkable specificity, effecting the removalof fibrinopeptides A and B of fibrinogen through the selectivecleavage of two Arg–Gly bonds between the 181 Arg/Lys–Xaabonds in fibrinogen. Significant advances have been made inrecent years towards understanding the origin of the specificityof cleavage of the Argl6–Gly17 bond of the A-chain ofhuman fibrinogen. We have previously proposed a model for thebound structure of fibrinopeptide A_7–16 (FPA), based uponNMR data, computer-assisted molecular modeling and the synthesisand study of peptidomimetic substrates and inhibitors of thrombin.We now report the structure of the ternary complex of an FPAmimetic (FPAM), hirugen and thrombin at 2.5 Å resolution(R-factor = 0.138) and specificity data for the inhibition ofthrombin and related trypsin-like proteinases by FPAM. The crystallographicstructures of FPA and its chloromethyl ketone derivative boundto thrombin were determined. Although there are differencesbetween these structures in the above modeled FPA structureand that of the crystal structure of FPAM bound to thrombin,the , angles in the critical region of P1–P2–P3in all of the structures are similar to those of bovine pancreatictrypsin inhibitor (BPTI) in the BPTI–trypsin complex andD–Phe–Pro–Arg (PPACK) in the PPACK–thrombinstructure. A comparison between these and an NMR-derived structureis carried out and discussed.