Two families of synthetic peptides that enhance fibrin turbidity and delay fibrinolysis by different mechanisms.

When fibrin clots are formed in vitro in the presence of certain positively charged peptides, the turbidity is enhanced and fibrinolysis is delayed. Here we show that these two phenomena are not always linked and that different families of peptides bring about the delay of lysis in different ways. In the case of intrinsically adhesive peptides corresponding to certain regions of the fibrinogen gammaC and betaC domains, even though these peptides bind to fibrin(ogen) and enhance turbidity, the delay in lysis is mainly due to direct inhibition of plasminogen activation. In contrast, for certain pentapeptides patterned on fibrin B knobs, the delay in lysis is a consequence of how fibrin units assemble. On their own, these B knob surrogates can induce the gelation of fibrinogen molecules. The likely cause of enhanced clot turbidity and delay in fibrinolysis was revealed by a crystal structure of the D-dimer from human fibrinogen cocrystallized with GHRPYam, the packing of which showed the direct involvement of the ligand tyrosines in antiparallel betaC-betaC interactions.

Crystal structure of human fibrinogen.

A crystal structure of human fibrinogen has been determined at approximately 3.3 A resolution. The protein was purified from human blood plasma, first by a cold ethanol precipitation procedure and then by stepwise chromatography on DEAE-cellulose. A product was obtained that was homogeneous on SDS-polyacrylamide gels. Nonetheless, when individual crystals used for X-ray diffraction were examined by SDS gel electrophoresis after data collection, two species of alpha chain were present, indicating that some proteolysis had occurred during the course of operations. Amino-terminal sequencing on post-X-ray crystals showed mostly intact native alpha- and gamma-chain sequences (the native beta chain is blocked). The overall structure differs from that of a native fibrinogen from chicken blood and those reported for a partially proteolyzed bovine fibrinogen in the nature of twist in the coiled-coil regions, likely due to weak forces imparted by unique crystal packing. As such, the structure adds to the inventory of possible conformations that may occur in solution. Other features include a novel interface with an antiparallel arrangement of beta chains and a unique tangential association of coiled coils from neighboring molecules. The carbohydrate groups attached to beta chains are unusually prominent, the full sweep of 11 sugar residues being positioned. As was the case for native chicken fibrinogen, no resolvable electron density could be associated with alphaC domains.

Probing the beta-chain hole of fibrinogen with synthetic peptides that differ at their amino termini.

In a recent report, we showed that alanine can replace glycine at the amino terminus of synthetic B-knobs that bind to human fibrin(ogen). We now report a survey of 13 synthetic peptides with the general sequence XHRPYam, all tested with regard to their ability to delay fibrinolysis in an in vitro system activated by t-PA, the results being used as measures of binding affinity to the betaC hole. Unexpectedly, some large and bulky amino acids, including methionine and arginine, are effective binders. Amino acids that branch at the beta carbon (valine, isoleucine, and threonine) do not bind effectively. Crystal structures were determined for two of the peptides (GHRPYam and MHRPYam) complexed with fibrin fragment D-dimer; the modeling of various other side chains showed clashing in the cases of beta-carbon substituents. The two crystal structures also showed that the enhanced binding observed with pentapeptides with carboxyl-terminal tyrosine, compared with that of their tetrapeptide equivalents, is attributable to an interaction between the tyrosine side chain and a guanidino group of a nearby arginine (beta406). The equivalent position in gamma-chains of human fibrin(ogen) is occupied by a lysine (gamma338), but in chicken and lamprey fibrin(ogen), it is an arginine, just as occurs in beta chains. Accordingly, the peptides GPRPam and GPRPYam, which are surrogate A-knobs, were tested for their influence on fibrin polymerization with fibrinogen from lamprey and humans. In lampreys, GPRPYam is a significantly better inhibitor, but in humans, it is less effective than GPRPam, indicating that in the lamprey system the same tyrosine-arginine interaction can also occur in the gamma-chain setting.

Differences in binding specificity for the homologous gamma- and beta-chain "holes" on fibrinogen: exclusive binding of Ala-His-Arg-Pro-amide by the beta-chain hole.

The beta-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 beta-chain hole and, unlike GHRPam, not at all to the homologous gamma-chain hole. The difference can be attributed to the methyl group of the alanine residue clashing with a critical carboxyl group in the gammaC hole but being accommodated in the roomier betaC hole where the equivalent carboxyl is situated more flexibly.

Binding of synthetic B knobs to fibrinogen changes the character of fibrin and inhibits its ability to activate tissue plasminogen activator and its destruction by plasmin.

Synthetic peptides corresponding to the amino-terminal sequence of the beta chain of fibrin increase the turbidity of fibrin clots, whether they are generated by the direct interaction of thrombin and fibrinogen or by the reassociation of fibrin monomers. The turbidity of batroxobin-induced clots, which are characteristically "fine," is increased even more dramatically. Pentapeptides are more effective than tetrapeptides. Surprisingly, the same peptides also delay fibrinolysis, whether activated by exogenously added plasmin or from the fibrin-enhanced stimulation of tissue plasminogen activator (tPA) activation of plasminogen. The peptides have only a very slight effect on the plasmic hydrolysis of a chromogenic peptide, either by the direct addition of plasmin or by plasmin generated from plasminogen by tPA. The synthetic peptides mimicking the B knobs appear to exert their action in two ways. First, they render fibrin less vulnerable to attack by plasmin. Second, they delay the fibrin activation of tPA. The latter is attributed to their ability to prevent the binding of the authentic B knob, which itself is located at the end of a flexible 50-residue tether and which needs time to find its elusive "hole". We propose that, when after a while the tethered knob does become inserted, it locks the betaC domain in a conformation that allows access to tPA-plasminogen-binding sites, whereas the untethered synthetic knobs restrict the fibrin to a conformation in which those sites remain inaccessible. Thus, although the interaction involving the A knob and gammaC hole is the basis for the polymerization of fibrin, the comparable but delayed interaction involving the B knob and the betaC hole is ultimately directed at preparing the clot for its eventual destruction.

The crystal structure of fragment double-D from cross-linked lamprey fibrin reveals isopeptide linkages across an unexpected D-D interface.

The crystal structure of fragment double-D from factor XIII-cross-linked lamprey fibrin has been determined at 2.9 A resolution. The 180 kDa covalent dimer was cocrystallized with the peptide Gly-His-Arg-Pro-amide, which in many fibrinogens, but not that of lamprey, corresponds to the B-knob exposed by thrombin. The structure was determined by molecular replacement, a recently determined structure of lamprey fragment D being used as a search model. GHRPam was found in both the gamma- and beta-chain holes. Unlike the situation with fragment D, the crystal packing of the cross-linked double-D structure exhibits two different D-D interfaces, each gamma-chain facing gamma-chains on two other molecules. One of these (interface I) involves the asymmetric interface observed in all other D fragments and related structures. The other (interface II) encompasses a completely different set of residues. The two abutments differ in that interface I results in an "in line" arrangement of abutting molecules and the interface II in a "zigzag" arrangement. So far as can be determined (the electron density could only be traced on one side of the cross-links), it is the gamma-chains of the newly observed zigzag units (interface II) that are joined by the reciprocal epsilon-amino-gamma-glutamyl cross-links. Auspiciously, the same novel D-D interface was observed in two lower-resolution crystal structures of human double-D preparations that had been crystallized under unusual circumstances. These observations show that double-D structures are linked in a way that is sufficiently flexible to accommodate different D-D interfaces under different circumstances.

Crystal structure of fragment D from lamprey fibrinogen complexed with the peptide Gly-His-Arg-Pro-amide.

The crystal structure of fragment D from lamprey fibrinogen has been determined at 2.8 A resolution. The 89 kDa protein was cocrystallized with the peptide Gly-His-Arg-Pro-amide, which in many fibrinogens-but not lamprey-corresponds to the B knob exposed by thrombin. Because lamprey fragment D is more than 50% identical in sequence with human fragment D, the structure of which has been reported previously, it was possible to use the method of molecular replacement. The space group of the lamprey crystals is P1; there are four molecules in the unit cell. Although the fragments are packed head to head by the same D:D interface as is observed in other related preparations containing fragments D, the tails are uniquely joined by an unnatural association of the terminal sections of the residual coiled coils from adjacent molecules. Some features of the lamprey structure are clearer than have been observed in previous fragment D structures, including the beta-chain carbohydrate cluster, for one, and the important gamma-chain carboxyl-terminal segment, for another. The most significant differences between the lamprey and human structures occur in connecting loops at the entryways to the beta-chain and gamma-chain binding pockets.