Pseudomonas aeruginosa exotoxin A interaction with eucaryotic elongation factor 2. Role of the His426 residue.

Pseudomonas aeruginosa exotoxin A (ETA) catalyzes the transfer of the ADP-ribose moiety of NAD+ onto eucaryotic elongation factor 2 (EF-2). To study the ETA site of interaction with EF-2, an immobilized EF-2 binding assay was developed. This assay demonstrates that ETA, in the presence of NAD+, binds to immobilized EF-2. Additionally, diphtheria toxin was also found to bind to the immobilized EF-2 in the presence of NAD+. Comparative analysis was performed with a mutated form of ETA (CRM 66) in which a histidine residue at position 426 has been replaced with a tyrosine residue. This immunologically cross-reactive, ADP-ribosyl transferase-deficient toxin does not bind to immobilized EF-2, thus explaining its lack of ADPRT activity. ETA bound to immobilized EF-2 cannot bind the monoclonal antibody TC-1 which specifically recognizes the ETA epitope containing His426. Immunoprecipitation of native ETA by mAb TC-1 is only achieved by incubating ETA in the presence of NAD+. Diethyl pyrocarbonate modification of the His426 residue blocks ETA binding to EF-2 and prevents the binding of the TC-1 antibody. Analogs of NAD+ containing a reduced nicotinamide ring or modified adenine moieties cannot substitute for NAD+ in the immobilized binding assay. Collectively, these data support our proposal that the site of ETA interaction with EF-2 includes His426 and that a molecule of NAD+ is required for stable interaction.     

An enzyme-linked immunosorbent assay for the association of the catalytic domain of diphthamide-specific ribosyltransferases to eukaryotic elongation factor-2.

The X-ray structure of the catalytic domain of Pseudomonas aeruginosa exotoxin A (PE24) has recently been solved to high resolution, facilitating studies on the interaction of PE24 with its target substrate, eukaryotic elongation factor-2 (eEF-2). PE24 exhibits mono-ADP-ribosyltransferase (ADPRT) activity in a mechanism that has been proposed to feature a nucleophilic attack by the diphthamide residue (nucleophile) of eEF-2 on the C-1 of the nicotinamide ribose of NAD(+). The interaction of wheat germ eEF-2 with PE24 was studied by employing an enzyme-linked immunosorbent assay (ELISA), devised to assess protein-protein interactions. It was shown that the proteins associate with each other only in the presence of the enzyme's nucleotide substrate, NAD(+), and exhibit a dose-dependent association that is saturable. The apparent dissociation constant (K(d)) for this protein-protein interaction is 50 nM and is salt-dependent. The association is maximal at low ionic strength and is progressively weaker at higher salt concentrations, which corroborates previous findings on the salt dependence of ADPRT activity for this toxin. This finding suggests that the sensitivity of ADPRT activity toward high salt resides in the interaction between the catalytic domain of the toxin and eEF-2. A major product of the glycohydrolase activity of PE24, nicotinamide, inhibits the binding between PE24 and eEF-2 with an ID(50) of 20 microM. The naturally occurring, noncatalytic mutant of PE24, H426Y, did not bind eEF-2 in the ELISA, verifying that His 426 is located at the center of the eEF-2 binding site within ETA.     

Biochemical analysis of CRM 66. A nonfunctional Pseudomonas aeruginosa exotoxin A

Pseudomonas aeruginosa exotoxin A (ETA) is an ADP-ribosyltransferase which inactivates protein synthesis by covalently attaching the ADP-ribose portion of NAD+ onto eucaryotic elongation factor 2 (EF-2). A direct biochemical comparison has been made between ETA and a nonenzymatically active mutant toxin (CRM 66) using highly purified preparations of each protein. The loss of ADP-ribosyltransferase activity and subsequent cytotoxicity have been correlated with the presence of a tyrosine residue in place of a histidine at position 426 in CRM 66. In the native conformation, CRM 66 demonstrated a limited ability (by a factor or at least 100,000) to modify EF-2 covalently and lacked in vitro and in vivo cytotoxicity, yet CRM 66 appeared to be normal with respect to NAD+ binding. Upon activation with urea and dithiothreitol, CRM 66 lost ADP-ribosyltransferase activity entirely yet CRM 66 retained the ability to bind NAD+. Replacement of Tyr-426 with histidine in CRM 66 completely restored cytotoxicity and ADP-ribosyltransferase activity. These results support previous findings from this laboratory (Wozniak, D. J., Hsu, L.-Y., and Galloway, D. R. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 8880-8884) which suggest that the His-426 residue of ETA is not involved in NAD+ binding but appears to be associated with the interaction between ETA and EF-2.     

Mapping of the epitope on lipoprotein lipase recognized by a monoclonal antibody (5D2) which inhibits lipase activity.

A monoclonal antibody, 5D2, which inhibits human lipoprotein lipase (hLPL) activity has been widely used for assessment of LPL immunoreactive mass in the clinical evaluation of patients [1] and for analysis of structure-function relationships of LPL [2,3]. We have mapped the epitope on LPL, recognized by the 5D2 antibody, within residues 396-405. Ala400 is the critical amino acid residue conferring epitope specificity. This knowledge confirms that the C-terminal domain of LPL plays a critical role in LPL activity and also provides important information for studies exploring the structure-function relationship of LPL using this antibody.     

Mapping of binding site III in the leptin receptor and modeling of a hexameric leptin.leptin receptor complex

The leptin.leptin receptor (LR) system shows strong similarities to the long chain cytokine interleukin-6 (IL-6) and granulocyte colony-stimulating factor (G-CSF) cytokine.cytokine receptor systems. The IL-6 family cytokines interact with their receptors through three different binding sites (I-III). We demonstrated previously that leptin has similar binding sites I-III and mapped the interactions between binding site II and cytokine receptor homology domain II (CRH2) (Peelman, F., Van Beneden, K., Zabeau, L., Iserentant, H., Ulrichts, P., Defeau, D., Verhee, A., Catteeuw, D., Elewaut, D., and Tavernier, J. (2004) J. Biol. Chem. 279, 41038-41046). In this study, we built homology models for the CRH1 and Ig-like domains of the LR. The Ig-like domain shows a large conserved surface patch in the beta-sheet formed by beta-strands 3, 6, and 7. Mutations in this patch almost completely abolished the leptin-induced STAT3-dependent reporter activity. We propose that a conserved cluster of residues Leu370, Ala407, Tyr409, His417, and His418 forms the center of binding site III of the LR. We built a hexameric leptin.LR complex model based on the hexameric IL-6 complex. In this model, a conserved hydrophobic protuberance of Val36, Thr37, Phe41, and Phe43 in the A-B loop of leptin fits perfectly in the CRH2 domain, corresponding to the IL-6 alpha-receptor, and forms the center of binding site I. The 2:4 hexameric leptin.LR complex offers a rational explanation for mutagenesis studies and residue conservation.     

Mapping the collagen-binding site in the von Willebrand factor-A3 domain

The multimeric glycoprotein von Willebrand factor (VWF) mediates platelet adhesion to collagen at sites of vascular damage. The binding site for collagen types I and III is located in the VWF-A3 domain. Recently, we showed that His(1023), located near the edge between the "front" and "bottom" faces of A3, is critical for collagen binding (Romijn, R. A., Bouma, B., Wuyster, W., Gros, P., Kroon, J., Sixma, J. J., and Huizinga, E. G. (2001) J. Biol. Chem. 276, 9985-9991). To map the binding site in detail, we introduced 22 point mutations in the front and bottom faces of A3. The mutants were expressed as multimeric VWF, and binding to collagen type III was evaluated in a solid-state binding assay and by surface plasmon resonance. Mutation of residues Asp(979), Ser(1020), and His(1023) nearly abolished collagen binding, whereas mutation of residues Ile(975), Thr(977), Val(997), and Glu(1001) reduced binding affinity about 10-fold. Together, these residues define a flat and rather hydrophobic collagen-binding site located at the front face of the A3 domain. The collagen-binding site of VWF-A3 is distinctly different from that of the homologous integrin alpha(2) I domain, which has a hydrophilic binding site located at the top face of the domain. Based on the surface characteristics of the collagen-binding site of A3, we propose that it interacts with collagen sequences containing positively charged and hydrophobic residues. Docking of a collagen triple helix on the binding site suggests a range of possible engagements and predicts that at most eight consecutive residues in a collagen triple helix interact with A3.     

Both the dimerization and immunochemical properties of E-cadherin EC1 domain depend on Trp(156) residue

Using site-directed mutagenesis, we show in this paper that the adhesive interface detected in cadherin crystals is unlikely to mediate adhesive interaction between myc- and flag-tagged E-cadherin molecules in human A-431 cells. We also found that a critical residue within this interface, His(233), is part of the epitope for mAb SHE78-7. This epitope was accessible to the antibody in the adhesive E-cadherin dimers, which is consistent with uninvolvement of the site containing His(233) in cell-cell adhesion. However, both the adhesive dimerization and the integrity of the SHE78-7 epitope depended on the same intramolecular interaction between Trp(156) and its hydrophobic pocket. Our data suggest that this interaction may have an important regulatory function in controlling the surface topology of the NH(2)-terminal domain of E-cadherin.     

Goodpasture syndrome. Localization of the epitope for the autoantibodies to the carboxyl-terminal region of the alpha 3(IV) chain of basement membrane collagen.

The autoantibodies of patients with Goodpasture syndrome are primarily targeted to the noncollagenous (NC1) domain of the alpha 3(IV) chain of basement membrane collagen (Saus, J., Wieslander, J., Langeveld, J. P. M., Quinones, S., and Hudson, B. G. (1988) J. Biol. Chem. 263, 13374-13380). In the present study, the location of the Goodpasture epitope in human alpha 3NC1 was determined, and its structure was partially characterized. This was achieved by identification of regions of alpha 3NC1 which are candidates for the epitope and which are structurally unique among the five known homologous NC1 domains (alpha 1-alpha 5); amino acids that are critical for Goodpasture antibody binding, by selective chemical modifications; and regions that are critical for Goodpasture antibody binding, by synthesis of 12 alpha 3NC1 peptides and measurement of their antibody binding capacity. The carboxyl-terminal region, residues 198-233, was identified as the most likely region for the epitope. By experiment, lysine and cysteine were identified as critical amino acids for antibody binding. Three synthetic peptides were found to inhibit Goodpasture antibody binding to alpha 3NC1 markedly: a 36-mer (residues 198-233), a 12-mer (residues 222-233), and a 5-mer (residues 229-233). Together, these results strongly indicate that the Goodpasture epitope is localized to the carboxyl-terminal region of alpha 3NC1, encompassing residues 198-233 as the primary antibody interaction site and that its structure is discontinuous. These findings provide a conceptual framework for future studies to elucidate a more complete epitope structure by sequential replacement of residues encompassing the epitope using cDNA expression products and peptides synthesized chemically.     

Antithrombin III Basel. Identification of a Pro-Leu substitution in a hereditary abnormal antithrombin with impaired heparin cofactor activity

Antithrombin III Basel is a hereditary abnormal antithrombin with normal progressive inhibition activity (normal reactive site) and reduced heparin cofactor activity (impaired heparin binding site). Structures of antithrombin III Basel and normal antithrombin III isolated from the same patient were compared by peptide mapping using the dimethylaminoazobenzene isothiocyanate precolumn derivatization technique. Of the approximately 50 tryptic peptides of normal and abnormal antithrombin III, one peptide comprising residues 40-46 had a different retention time in reversed-phase high performance liquid chromatography. The amino acid sequence of the peptide from antithrombin III Basel had a single substitution of Pro (normal) by Leu (abnormal) at position 41. This substitution is close to an Arg (residue 47) and a Trp (residue 49) which have previously been shown to be critical for heparin binding by antithrombin III. Although additional amino acid substitutions in antithrombin III Basel cannot be ruled out, this Pro-Leu replacement could cause a conformational change by increasing both the helical structure and the hydrophobicity around residue 41. These data suggest that: (i) the heparin binding site of antithrombin III encompasses the region containing residues 41, 47, and 49; and (ii) the impaired heparin cofactor activity of antithrombin III Basel is likely due to a conformational change of the heparin binding site induced by the Pro-Leu substitution at position 41.     

Functional analysis of exotoxin A-related protein of Pseudomonas aeruginosa lacking residues 225-412

The crystal structure of the exotoxin A (ETA) of Pseudomonas aeruginosa showed that this protein is folded into three distinct domains. Domain I (Ia and Ib), the amino-terminal domain, is the receptor-binding domain of ETA and domain III, the carboxy-terminal domain, is responsible for the ADP-ribosyl transferase activity of the toxin. To elucidate the function(s) of domains 1b and II in the intoxication process and to define the region of the domain III necessary for ADP-ribosylating activity, a defined deletion in the structural gene of P. aeruginosa ETA encompassing residues 225-412 was constructed and an ETA-related product DeID, (from which all of domains II and Ib were deleted) was expressed. The ETA-related protein did not penetrate sensitive cells, but retained the same specific activity to ADP-ribosylate elongation factor-2 as wild-type toxin. This suggests that domain II is necessary to allow toxin internalization by sensitive cells and that the absence of domain Ib does not interfere with enzymic activity. The domain strictly involved in ADP-ribosylation activity encompasses residues 412-613.     

A 1H NMR study of a ternary peptide complex that mimics the interaction between troponin C and troponin I.

The troponin I peptide N alpha-acetyl TnI (104-115) amide (TnIp) represents the minimum sequence necessary for inhibition of actomyosin ATPase activity of skeletal muscle (Talbot, J.A. & Hodges, R.S. 1981, J. Biol. Chem. 256, 2798-3802; Van Eyk, J.E. & Hodges, R.S., 1988, J. Biol. Chem. 263, 1726-1732; Van Eyk, J.E., Kay, C.M., & Hodges, R.S., 1991, Biochemistry 30, 9974-9981). In this study, we have used 1H NMR spectroscopy to compare the binding of this inhibitory TnI peptide to a synthetic peptide heterodimer representing site III and site IV of the C-terminal domain of troponin C (TnC) and to calcium-saturated skeletal TnC. The residues whose 1H NMR chemical shifts are perturbed upon TnIp binding are the same in both the site III/site IV heterodimer and TnC. These residues include F102, I104, F112, I113, I121, I149, D150, F151, and F154, which are all found in the C-terminal domain hydrophobic pocket and antiparallel beta-sheet region of the synthetic site III/site IV heterodimer and of TnC. Further, the affinity of TnIp binding to the heterodimer (Kd = 192 +/- 37 microM) was found to be similar to TnIp binding to TnC (48 +/- 18 microM [Campbell, A.P., Cachia, P.J., & Sykes, B.D., 1991, Biochem. Cell Biol. 69, 674-681]). The results indicate that binding of the inhibitory region of TnI is primarily to the C-terminal domain of TnC. The results also indicate how well the synthetic peptide heterodimer mimics the C-terminal domain of TnC in structure and functional interactions.     

Mutational analysis of Cvab, an ABC transporter involved in the secretion of active colicin V

CvaB is the central membrane transporter of the colicin V secretion system that belongs to an ATP-binding cassette superfamily. Previous data showed that the N-terminal and C-terminal domains of CvaB are essential for the function of CvaB. N-terminal domain of CvaB possesses Ca(2+)-dependent cysteine proteolytic activity, and two critical residues, Cys32 and His105, have been identified. In this study, we also identify Asp121 as being the third residue of the putative catalytic triad within the active site of the enzyme. The Asp121 mutants lose both their colicin V secretion activity and N-terminal proteolytic activity. The adjacent residue Pro122 also appears to play a critical role in the colicin V secretion. However, the reversal of the two residues D121P - P122D results in loss of activity. Based on molecular modeling and protein sequence alignment, several residues adjacent to the critical residues, Cys32 and His105, were also examined and characterized. Site-directed mutagenesis of Trp101, Asp102, Val108, Leu76, Gly77, and Gln26 indicate that the neighboring residues around the catalytic triad affect colicin V secretion. Several mutated CvaB proteins with defective secretion were also tested, including Asp121 and Pro122, and were found to be structurally stable. These results indicate that the residues surrounding the identified catalytic triad are functionally involved in the secretion of biologically active colicin V.     

Discrimination between rat thiostatin (T-kininogen) and one of its cystatin-like inhibitory fragments by a monoclonal antibody, and localization of the epitope.

A monoclonal antibody (mAb D3) raised against rat thiostatin (T-kininogen) strongly interacted with a fragment, identified as cystatin-like domain 3, which inhibits cysteine proteinases but did not recognize intact, native thiostatin. The antigen-antibody reaction requires cleavage of the single peptide chain of thiostatin in its inter-domain 2-3 region. This mAb can also differentiate between the two molecular varieties of thiostatin, reacting only with immobilized domain 3 from T1 thiostatin, which differs from the T2 variety by only 10 out of 125 residues. mAb D3 did not react with an N-terminally truncated domain 3 of T1 thiostatin prepared by submaxillary gland kallikrein k10 proteolysis. This suggests that the epitope, or an essential part of it, is located on a stretch of 12 residues at the N-terminal of the T1 thiostatin domain 3. This sequence in T1 thiostatin differs from that in T2 thiostatin by four amino acids, two of which are arginyl residues in T1. Chemical modification of these residues located at positions 246 and 250 decreased the reactivity of T1 domain 3 towards the antibody, suggesting that at least one of them is a critical residue of the epitope. Arginine 246 is part of a small disulfide loop between cysteines 245 and 248 which is also necessary for antibody recognition. This antibody does not change the inhibitory properties of purified domain 3 towards papain or rat liver cathepsin L, indicating that the N-terminal part of domain 3 is not involved in inhibition. mAb D3 was used to demonstrate the presence of inhibitory thiostatin fragments in ascites fluid but not in plasma from normal or turpentine-injected rats.     

A monoclonal antibody recognizes an epitope in the first extracellular loop of the plasma membrane Ca2+ pump.

A monoclonal antibody against the human erythrocyte Ca2+ pump (1E4) reacted with the enzyme in intact erythrocytes. Using trypsinized preparations of the pump the antibody only reacted with the N-terminal fragments of 33.5 and 35 kDa. The fragments span from the N terminus (35 kDa) or from residue 19 (33.5 kDa) to residue 314 of the hPMCA4 isoform of the pump. Exhaustive degradation with a number of agents produced smaller peptides which reacted with the antibody. Sequencing analysis on two chymotryptic fragments of 8.8 and 13.5 kDa identified the epitope in an approximately 80-residue domain beginning with Gly-81. Two peptides corresponding to the putative extramembrane portions of this region of the pump were synthesized. The antibody reacted with one of them, spanning residues Phe-121 to Gly-152 and containing the first putative external loop of the pump. Peptides corresponding to overlapping portions of this peptide were synthesized, leading to the location of the epitope in a 13-residue sequence (Glu-130 to Glu-142) in the first predicted extracellular loop (Verma, A. K., Filoteo, A. G., Stanford, D. R., Wieben, E. D., Strehler, E. E., Fischer, R., Heim, R., Vogel, G., Mathews, S., Strehler-Page, M-A., James, P., Vorherr, T., Krebs, J., Penniston, J. T., and Carafoli, E. (1988) J. Biol. Chem. 263, 14152-14159).     

Chimeric nectin1-poliovirus receptor molecules identify a nectin1 region functional in herpes simplex virus entry

Human nectin1 (hNectin1), an adhesion molecule belonging to the nectin family of the immunoglobulin superfamily, mediates entry of herpes simplex virus (HSV) into cells. The hNectin1 domain that mediates virus entry into cells and also binds glycoprotein D (gD) has been localized to the first N-terminal V-type domain. The poliovirus receptor (PVR) is a structural homolog to nectins, but it cannot function as an HSV entry receptor. hNectin1-PVR chimeras were constructed to functionally locate the site on hNectin1 involved in HSV entry (HSV entry site). The epitope recognized by monoclonal antibody (MAb) R1.302, which is able to block HSV entry, was also located. The chimeric receptors were designed to preserve the overall structure of the V domain. The HSV entry activity mapped entirely to the hNectin1 portion located between residues 64 and 94 (64-94), likely to encode the C, C', and C" beta-strands and intervening loops. In turn, this site consisted of two portions: one with low-level basal activity for HSV entry (77-94), and one immediately upstream (residues 64 to 76) which greatly enhanced the HSV entry activity of the downstream region. The gD-binding site mapped substantially to the same site, whereas the MAb R1.302 epitope also required a further downstream portion (95-102). The involvement of the 64-76 portion is at difference with previous indirect mapping results that were based on competitive binding studies (C. Krummenacher et al., J. Virol. 74:10863-10872, 2000). The A, A', B, D, E, F, and G beta-strands and intervening loops did not appear to play any role in HSV entry. According to the predicted three-dimensional structure of PVR, the C C' C" site is located peripherally in the V domain and very likely represents an accessible portion at the cell surface.     

Exotoxin A of Pseudomonas aeruginosa: substitution of glutamic acid 553 with aspartic acid drastically reduces toxicity and enzymatic activity.

Glutamic acid 553 of Pseudomonas aeruginosa exotoxin A (ETA) has been identified by photoaffinity labeling as a residue within the NAD binding site (S.F. Carroll and R.J. Collier, J. Biol. Chem. 262:8707-8711, 1987). To explore the function of Glu-553 we used oligonucleotide-directed mutagenesis to replace this residue with Asp in cloned ETA and expressed the mutant gene in Escherichia coli K-12. ADP-ribosylation activity of Asp-553 ETA in cell extracts was about 1,800-fold lower and toxicity for mouse L-M929 fibroblasts was at least 10,000-fold lower than that of the wild-type toxin. Extracts containing Asp-553 ETA inhibited the cytotoxicity of authentic ETA on L-M929 fibroblasts, suggesting that the mutant toxin competes for ETA receptors. The results indicate that Glu-553 is crucial for ADP-ribosylation activity and, consequently, cytotoxicity of ETA. Substitution or deletion of this residue may be a route to new ETA vaccines.     

Identification of a ligand-binding site in an immunoglobulin fold domain of the Saccharomyces cerevisiae adhesion protein alpha-agglutinin.

The Saccharomyces cerevisiae adhesion protein alpha-agglutinin (Ag alpha 1p) is expressed by alpha cells and binds to the complementary a-agglutinin expressed by a cells. The N-terminal half of alpha-agglutinin is sufficient for ligand binding and has been proposed to contain an immunoglobulin (Ig) fold domain. Based on a structural homology model for this domain and a previously identified critical residue (His292), we made Ag alpha 1p mutations in three discontinuous patches of the domain that are predicted to be in close proximity to His292 in the model. Residues in each of the three patches were identified that are important for activity and therefore define a putative ligand binding site, whereas mutations in distant loops had no effect on activity. This putative binding site is on a different surface of the Ig fold than the defined binding sites of immunoglobulins and other members of the Ig superfamily. Comparison of protein interaction sites by structural and mutational analysis has indicated that the area of surface contact is larger than the functional binding site identified by mutagenesis. The putative alpha-agglutinin binding site is therefore likely to identify residues that contribute to the functional binding site within a larger area that contacts a-agglutinin.     

Cysteine residues in the active site of Corynebacterium sarcosine oxidase

Sarcosine oxidase from Corynebacterium sp. U-96 is inhibited by iodoacetamide (IAM) and the inhibition is prevented by the substrate analog, sodium acetate. To elucidate the mechanism of inhibition of the enzyme by IAM, we determined the amino acid sequences around the IAM-reactive cysteine residues, and the effects of the modification on the enzyme activity and the oxidation-reduction of the FAD moieties of the enzyme. The enzyme was specifically labeled with [14C]IAM, and the labeled subunit B was digested with trypsin and chymotrypsin. The HPLC profiles of the proteolytic digests showed mainly two radioactive peaks. The 14C-labeled peptides were purified, and their N-terminal sequences were determined to be Cys-Gly-Thr-Pro-Gly-Ala-Gly-Tyr (TC-1) and Ala-Gly-Ile-Ala-Cys-Xaa-Asp-Xaa-Val-Ala(-)- (TC-2). Peptide TC-2 contains a covalent FAD-binding sequence [Asx-His-Val-Ala; Shiga et al. (1983) Biochem. Int., 6, 737]. [14C]IAM-incorporation into the TC-1 sequence was strongly inhibited by sodium acetate. The N-terminal amino acid sequence of the CNBr fragment containing the TC-1 sequence (65 residues) was determined. According to the secondary structure predictions, Gly-Thr-Pro-Gly-Ala-Gly of the TC-1 sequence is located between the beta sheet and alpha helix of the sequence, indicating the presence of an AMP-binding site in the TC-1 region. The activity of the enzyme treated with IAM in the presence and absence of sodium acetate was not inhibited by sodium sulfite, which is known to react specifically with covalent FAD.(ABSTRACT TRUNCATED AT 250 WORDS)     

Identification of the collagen-binding site of the von Willebrand factor A3-domain

Von Willebrand factor (vWF) is a multimeric glycoprotein that mediates platelet adhesion and thrombus formation at sites of vascular injury. vWF functions as a molecular bridge between collagen and platelet receptor glycoprotein Ib. The major collagen-binding site of vWF is contained within the A3 domain, but its precise location is unknown. To localize the collagen-binding site, we determined the crystal structure of A3 in complex with an Fab fragment of antibody RU5 that inhibits collagen binding. The structure shows that RU5 recognizes a nonlinear epitope consisting of residues 962-966, 981-997, and 1022-1026. Alanine mutants were constructed of residues Arg(963), Glu(987), His(990), Arg(1016), and His(1023), located in or close to the epitope. Mutants were expressed as fully processed multimeric vWF. Mutation of His(1023) abolished collagen binding, whereas mutation of Arg(963) and Arg(1016) reduced collagen binding by 25-35%. These residues are part of loops alpha3beta4 and alpha1beta2 and alpha-helix 3, respectively, and lie near the bottom face of the domain. His(1023) and flanking residues display multiple conformations in available A3-crystal structures, suggesting that binding of A3 to collagen involves an induced-fit mechanism. The collagen-binding site of A3 is located distant from the top face of the domain where collagen-binding sites are found in homologous integrin I domains.