Differences in Binding Specificity for the Homologous -and â-Chain Holes on Fibrinogen : Exclusive Binding of Ala-His-Arg-Pro-amide by the â-Chain Hole

Theâ-chain amino-terminal sequences of all known mammalian fibrins begin with the sequence Gly-His-Arg-Pro(GHRP-), but the homologous sequence in chicken fibrin begins with the sequence Ala-His-Arg-Pro(AHRP-). Nonetheless, chicken fibrinogen binds the synthetic peptide GHRPam, and a previously reported crystal structure has revealed that the binding is in exact conformance with that observed for the human GHRPam -fragment D complex. We now report that human fibrinogen, which is known not to bind APRP, binds the synthetic peptide AHRPam. Moreover, a crystal structure of AHRPam complexed with fragment D from human fibrinogen shows that AHRPam binds exclusively to the â-c ain hole and, unlike GHRPam, not at all to the homologous γ-chain hole. The difference can be attributed to the methyl group of the alanine residue clashing with a critical carboxyl group in the γC hole but being accommodated in the roomier âC hole where the equivalent carboxyl is situated more flexibly. In vertebrates, fibrinogen is transformed into a fibrin clot by the thrombin-catalyzed removal of peptide material from the amino-terminal segments of the Randâ-chains ( 1). The release of these peptides, fibrinopeptides A and B, exposes two sets of “knobs”, A and B, which fit into complementary “holes” on neighboring molecules ( 2). Typically, A-knobs begin with the sequence Gly-Pro-Argand B-knobs with the sequence Gly-His-Arg-, although in some nonmammalian vertebrates differences have been observed. Synthetic peptides corresponding to A-knobs bind to fibrinogen and prevent polymerization ( 3, 4). Synthetic B-knobs bind to fibrinogen but, under most conditions, do not inhibit fibrin formation (3-5). Instead, the binding of synthetic B-knobs leads to more turbid clots ( 3, 6), the apparent result of enhanced lateral association of the polymerizing protofibrils. Although numerous factors can affect the turbidity of clots, which in the extreme are denoted as “fine” or “coarse” ( 7), synthetic B-knobs are unique in that the enhanced turbidity is accompanied by resistance to fibrinolysis and a delay in the fibrin-catalyzed activation of tPA1 (8). When both synthetic Aand B-knobs are present in a solution of fibrinogen or fragment D or D-dimer, invariably the two peptides each find their respective holes, their different structures being readily distinguished ( 9, 10). The architectures of the two homologous holes are remarkably similar, however, and it has not been immediately obvious how the sites discriminate between the two ligands. If only one of the synthetic knobs is present, it will bind to both kinds of hole ( 3, 4). An opportunity for investigating specific differences arose when it was found that the â-chain of chicken fibrin has an amino-terminal alanine instead of glycine ( 11). The finding was unexpected because it had been previously demonstrated that human fibrinogen does not bind the synthetic peptide APRP or other non-glycyl peptides ( 4), and it was presumed the requirement for amino-terminal glycine extended to the B-knob also. Fortuitously, chicken fibrinogen bound the previously synthesized peptide GHRPam, and a crystal structure at 2.7 Å resolution showed it bound to the â-chain holes in the same way as in fragments D and D-dimer from human fibrinogen ( 12). The existence of amino-terminal alanine in the chicken fibrin â-chain suggested that the âC hole was a more promiscuous binder than the γC hole. As such, we examined the influence of the synthetic peptide AHRPam on the † This work was supported in part by grants from the National Heart, Lung and Blood Institute (HL-81553) and the American Heart Association. ‡ The atomic coordinates and structure factors for FDAH have been deposited in the Protein Data Bank as entry 2H43. * To whom correspondence should be addressed. Telephone: (858) 534-4417. Fax: (858) 534-4985. E-mail: rdoolittle@ucsd.edu. § Present address: Baylor College of Medicine, Houston, TX 77030. 1 Abbreviations: AHRPam, Ala-His-Arg-Pro-amide; GHRPam, GlyHis-Arg-Pro-amide; dAPRPam, D-Ala-Pro-Arg-Pro-amide; AHRPY, Ala-His-Arg-Pro-Tyr-amide; GHRPYam, Gly-His-Arg-Pro-Tyr-amide; FDAH, fragment D complexed with AHRPam; FDGH, fragment D complexed with GHRPam; DDGH, D-dimer complexed with GHRPam; DDBO, D-dimer complexed with GPRPam and GHRPam; DDNL, D-dimer with no ligands; PEG, polyethylene glycol; MPD, methylpentanediol; tPA, tissue plasminogen activator. 13962 Biochemistry2006,45, 13962-13969 10.1021/bi061219e CCC: $33.50 © 2006 American Chemical Society Published on Web 11/02/2006 formation of human fibrin. In fact, the alanyl peptide enhances the turbidity of fibrin clots prepared from human fibrinogen in the very same way that GHRPam does. Accordingly, we prepared crystals of fragment D from human fibrinogen in the presence of the synthetic peptide AHRPam and determined a structure by X-ray crystallography. We have found that, unlike the synthetic peptide GHRPam, AHRPam binds exclusively to the â-chain hole. The presence or absence of ligand was verified not only by the presence or absence of electron density in suitably calculated omit maps but also by the conformations of key residues known to move as a result of ligand binding. We also conducted experiments on fibrin formation with the corresponding L-alanine analogue of the A-knob, APRPam, and confirmed that it does not have a significant effect on fibrin formation (human). A report ( 13) that D-alanine can replace glycine in these peptides (dAPRP) and can prevent fibrin formation could not be confirmed, no inhibition of thrombin-induced clotting or reassociation of fibrin monomers being observed with human fibrinogen in the presence of that synthetic peptide. The question of how theâ-chain hole is able to accommodate an alanyl peptide while the γ-chain hole cannot was addressed by comparing the new structure with numerous previously reported structures of fragment D and D-dimer complexed with various combinations of peptides. In the end, it was possible to pinpoint particular features that make the γ-chain hole more discriminating. MATERIALS AND METHODS The synthetic peptides Ala-His-Arg-Pro-amide, Ala-HisArg-Pro-Tyr-amide, Ala-Pro-Arg-Pro-amide, and D-Ala-ProArg-Pro-amide were purchased from Sigma-Genosys. The peptides Gly-His-Arg-Pro-amide and Gly-Pro-Arg-Pro-amide were those described in earlier publications from this laboratory ( 9, 12, 14-16) and had been synthesized by the BOC procedure ( 17). Fibrinogen was prepared from human blood plasma as described previously; protocols for the purification of fragments D and D-dimer have also been fully described in previous publications ( 9, 12, 14-16). Human thrombin was purchased from Enzyme Research Corp. Fibrin monomer reassociation assays ( 18) were conducted by diluting 50μL of NaBr-dispersed human fibrin with 1.0 mL of 0.08 M phosphate buffer (pH 6.8) containing various peptide additives, and following the resulting turbidity in a spectrophotometer at 350 nm ( λ). The details of this assay and others involving thrombin -fibrinogen and thrombin tPA-fibrinogen species have been described in a recent publication (8). Crystals were obtained by vapor diffusion from sitting drops at room temperature. Crystals of the fragment D-AHRPam complex were grown from drops containing equal volumes of a 10 mg/mL protein solution in 0.05 M Tris (pH 7.0), 5 mM CaCl 2, containing 8 mM AHRPam, and a well solution composed of 0.05 M Tris (pH 8.0) containing 15% PEG-3350, 10 mM CaCl 2, and 2 mM sodium azide. The final (pre-equilibration) concentration of calcium in the drop was 7.5 mM. Cryoprotection before crystals were frozen was achieved by gradually adding MPD to a final concentration of 15%. Preliminary screening of crystals was performed at the University of California at San Diego X-ray Crystallography Facility. Full sets of diffraction data were collected at beamline 5.0.1, the Advanced Light Source, Lawrence Berkeley National Laboratory (Berkeley, CA). Data were processed with HKL2000 (at the Advanced Light Source) and with Denzo and Scalepack ( 19). The structure was determined by molecular replacement with Amore ( 20). Refinement was accomplished with CNS ( 21), Rfree being used as a guide to rebuilding throughout; rebuilding depended heavily on O ( 22). Model Comparisons . Numerous comparisons with previously reported structures of fragments D and D-dimer were conducted in an effort to identify any features of the âand γ-chain holes that might be responsible for differences in discrimination. Structures were downloaded from the Protein Data Bank ( 23) or were the results of our own studies. These included crystallographic models with either GPRPam (or in one case GPRP) or GHRPam bound to one hole, the other hole, neither, or both. Special attention was paid to the conformational changes that occur when the various peptides are bound. Although the models varied with regard to their final resolution and had small discrepancies that were likely due to differences in data processing, model building, and protocols used for refinement, a consistent pattern of knob hole interactions and attendant conformational change was readily established. For the most part, comparisons were limited to the 90or 94-residue domains that contain the knob-binding pockets, and the synthetic knobs bound by them. Model comparisons were made after all structures were superimposed onto a common reference point with O ( 22). Various distance measuring programs contained in the original XPLOR package ( 24) or in CCP4 ( 20) were employed to determine all distances ofe4.0 Å between every non-hydrogen knob atom and the host protein. Some other distances between key atoms were determined directly from PDB coordinate files by use of a perl script in conjunction with Excel (we are grateful to J. Nand for writing this script). Illustrations were prepared with the aid of PyMol ( 25). Because the emphasis is entirely on structures derived from fragments D