monitoring for neuroprotection : new technologies for the new millennium

101

Insight from Studies withRecombinant Fibrinogens

SUSAN T. LORD

a

AND OLEG V. GORKUN

b

a

Departments of Pathology and Laboratory Medicine, and Chemistry, Curriculum in Genetics and Molecular Biology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

b

Department of Pathology and Laboratory Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

A

BSTRACT

: Using a two-step cloning strategy, we have synthesized more than20 variant human fibrinogens for biochemical studies. In preliminary experi-ments we showed that normal fibrinogen produced in CHO cells serves as anaccurate model for plasma fibrinogen. We focus here on those variants whosecharacterization has provided insight into the mechanism of thrombin-catalyzed polymerization. Analysis of N-terminal variants showed that throm-bin specificity dictates the ordered release of fibrinopeptides. Nevertheless,analysis of C-terminal variants indicated that fibrinopeptide B (FpB) release isdependent on polymerization. Changes in the

a

polymerization site and thehigh-affinity calcium-binding site were associated with a complete loss of poly-merization. These experiments showed that alterations in the calcium-bindingsite influenced function of the

a

site; in contrast, alterations in the

a

site did notalter calcium binding. Analysis of variants in the N-terminus of the B

�

chainprovided the first direct evidence that this region impacts predominantly onlateral aggregation, as has long been presumed. These experiments also sug-gested that lateral aggregation facilitated by this region proceeds without therelease of FpB. From these studies we learned that individual sites withinfibrinogen do not function in isolation. We conclude that thrombin-catalyzedpolymerization is mediated by a continuum of concerted interactions.

K

EYWORDS

: Recombinant; Fibrinogen; Fibrin; Polymerization; Lateral aggre-gation; Fibrinopeptide; Clot.

INTRODUCTION

We described the successful synthesis of fully assembled, functional recombinanthuman fibrinogen in cultured Chinese Hamster Ovary cells in 1993.

1

Using this sys-tem, we have been able to produce milligrams of recombinant fibrinogen, sufficientfor efficient purification and subsequent biochemical analysis. By incorporating atwo-step cloning strategy, this system permits the expedient synthesis of new variantfibrinogens.

Address for correspondence: Susan T. Lord, Ph.D., Department of Pathology and Labora-tory Medicine, University of North Carolina, CB#7525, Brinkhous-Bullitt 603, Chapel Hill,NC 27599-7525, USA. Voice: 919-966-3548; fax: 919-966-6718.

[email protected]

102 ANNALS NEW YORK ACADEMY OF SCIENCES

An important preliminary experiment for analysis of recombinant fibrinogenvariants was the comparison of normal recombinant fibrinogen to normal plasmafibrinogen.

2

Plasma fibrinogen is a heterogeneous mixture of molecules due to vari-ations in mRNA processing, posttranslational modification and proteolysis in the cir-culation.

3

Because the significance of this heterogeneity is not known, it wasnecessary to compare the recombinant and plasma proteins for each biochemicaltrait under study. In multiple experiments, we have found that in essence these mol-ecules are the same. Transmission electron microscopy studies showed trinodularstructures for both fibrinogens, and scanning electron microscopy studies showedthat the structures of the two fibrin clots were indistinguishable. We found thatthrombin-catalyzed fibrinopeptide release, FXIIIa-catalyzed cross-link formation,and fibrin monomer polymerization were remarkably similar for both proteins. Wesaw small differences in thrombin-catalyzed polymerization, suggesting that theordered assembly of protofibrils and fibers was not identical. We speculate that thesesubtle differences reflect the heterogeneity of plasma fibrinogen. We conclude thatthe normal recombinant fibrinogen produced in CHO cells serves as an accuratemodel for plasma fibrinogen, and thus is an appropriate control for studies withrecombinant variant fibrinogens.

We have subsequently synthesized and characterized more than 20 variant fibrin-ogens. We designed our variants to test implications based on published data. Ourresults have demonstrated that the availability of planned variant fibrinogens pro-vides a novel approach to examine current ideas relevant to the many functions offibrinogen. A list of the variants that have been described, including those firstdescribed at this conference, is presented in T

ABLE

1. In this communication wefocus on variants made at sites that have provided insight into the details of throm-bin-catalyzed polymerization: sites that contribute to thrombin specificity andsequential fibrinopeptide release; sites that contribute to protofibril formation andhigh-affinity calcium binding; a site that contributes to D:D interactions; and a sitethat contributes to lateral aggregation.

T

ABLE

1. Described variant fibrinogens

N

OTE

: In addition to the changes in single chains, we have also examined one variant withchanges in two chains, A

α

D97E,D597E,

γ∆

408-411.

27

a

Described in this paper.

b

Synthesized in BHK cells as described by Farrell

et al.

32

A

α

chain B

β

chain

γ

chain

F8Y

8

R14H

a

D318A

a

G14V

9

A

′β

9

D320A

a

D97E

b

27

A68T

28

D318A,D320A

a

D574E

b

27

P70S

29

∆

319,320

18

A

α

251

30

L72S

29

Y363A

14

B

β

H16A,R17A,P18A

a

D364A

14

D364H

14

∆

408–411

31

103LORD & GORKUN: RECOMBINANT FIBRINOGENS

In most instances we chose to synthesize variants with the intention to interruptone specific interaction, and thereby one specific function. Instead we found thatthese variants influence multiple activities. From these studies we learned somethingthat might be considered intuitively obvious from the beginning: because they arejoined in a single molecule, individual sites do not function in isolation. Thus, thesteps in thrombin-catalyzed polymerization are a continuum and only a very few, ifany, individual steps can be examined alone.

RESULTS

Thrombin Specificity and Sequential Fibrinopeptide Release

Thrombin catalyses the hydrolysis of four peptide bonds in fibrinogen releasingtwo fibrinopeptides A (FpA) and two fibrinopeptides B (FpB) from the N-termini ofthe A

α

and B

β

chains, respectively. The release of FpA follows simple first orderkinetics.

4

The release of FpB is more complex, but can be described as a first orderprocess under the assumption that FpB release occurs subsequent to FpA release.

4

The efficient release of FpB has been linked to fibrin polymerization, in part becausethe release of FpB is slower in the presence of the peptide GPRP, which inhibitspolymerization.

5

To examine the basis for the sequential release of fibrinopeptides,and the correlation of sequential release with polymerization,

6

we synthesized fourvariants. To reverse the ordered release, such the FpB release precedes FpA release,we synthesized A

α

F8Y fibrinogen where Phe 8 in the A

α

chain is replaced with Tyr.To eliminate the ordered release, such that all four peptides are released concomi-tantly, we synthesized A

′β

fibrinogen, where we replaced FpB with an FpA-like sub-strate, FpA

′

where Gly 14 in FpA is replaced by Val; we also synthesized therequired control A

′α

fibrinogen, with FpA

′

replacing FpA. To exaggerate the orderedrelease, such that FpB release is substantially delayed, we synthesized B

β

R14Hfibrinogen where Arg14 in the B

β

chain is replaced with His.Based on prior results,

7

we anticipated that the A

α

F8Y substitution would impairthrombin-catalyzed hydrolysis of the A

α

Arg16-Gly17 bond. Indeed, kinetic analy-sis demonstrated that A

α

F8Y1-16 was a very poor thrombin substrate, with a spec-ificity constant 280-fold lower than the normal fibrinopeptide.

8

These experimentsconfirmed that Phe8 plays a critical role in thrombin-catalyzed FpA release. Therelease of FpB was also impaired, but less so than FpA, such that FpB release pre-ceded FpA release. Thrombin-catalyzed polymerization of A

α

F8Y fibrinogen, mon-itored by the change in turbidity at 350 nm, was markedly delayed relative to normal.These data suggest that exposure of the N-terminus of the

α

chain is required for nor-mal polymerization. Thus, FpA release occurs first, and then polymerization begins.

Also based on prior results,

7

we anticipated that FpA

′

release from the N-terminiof the

β

chains of A

′β

fibrinogen would occur concomitantly with FpA release fromthe

α

chains. In control experiments with Gly14 replaced with Val in the A

α

chain,we found that this variant fibrinopeptide was a good thrombin substrate, with releaseof FpA

′

essentially identical to FpA.

9

Kinetic analysis of thrombin-catalyzed fibrin-opeptide release from A

′β

fibrinogen showed that the release of both peptidesfollowed simple first order kinetics, and that the specificity constants for bothsubstrates were similar. We concluded that the substrate specificity directs the


ordered release of FpA prior to FpB. Thrombin-catalyzed polymerization of A

′β

fibrinogen was the same as normal fibrinogen, and the two clot structures, examinedby scanning electron microscopy, were indistinguishable. We concluded that earlyexposure of the N-terminus of the

β

chain does not affect polymerization. We alsoexamined the release of fibrinopeptides in the presence of 1 mM GPRP. Consistentwith previous reports,

5

we found that the rate of FpB release from normal fibrinogenwas threefold slower in the presence of GPRP. In contrast, the rate of FpA

′

releasefrom A

′β

fibrinogen was unaffected. This remarkable result indicates that it is thestructure of the fibrinopeptide

per se

that links the influence of GPRP to both arrest-ed polymerization and delayed FpB release. Further experiments are clearly neededto appreciate the link between FpB release and polymerization.

Based on the analysis of dysfibrinogen Petrosky, A

α

R16H fibrinogen,

10

we antic-ipated that the thrombin-catalyzed release of the variant fibrinopeptide fromB

β

R14H fibrinogen would be dramatically slower than the release of FpB fromnormal fibrinogen. Indeed, the rate was reduced 300-fold. Thrombin-catalyzedpolymerization of B

β

R14H fibrinogen was also significantly impaired, with a lowerfinal turbidity than normal, as expected when lateral aggregation is impaired. Thisresult suggests that the timely release of FpB is critical for normal lateral aggrega-tion. Further experiments, however, did not support this suggestion. We examinedBatroxobin-catalyzed fibrinopeptide release and fibrin polymerization. Neither the

FIGURE 1. Representative polymerization curves. The change in turbidity with timeof normal (solid lines) and BβR14H (dashed lines) fibrinogen with 0.2 mg/mL fibrinogen,0.1 U/mL enzyme in 20 mM HEPES, pH 7.4, and ionic strength held constant at 0.15 withNaCl. Thin lines represent polymerization in the absence of added calcium and thick linesrepresent polymerization with 10 mM CaCl2. Panel A, thrombin-catalyzed polymerization.Panel B, batroxobin-catalyzed polymerization. From Reference 11.


normal nor the variant FpB was released by Batroxobin. However, the polymeriza-tion curves for these fibrinogens were different. Moreover, as is shown in F

IGURE

1,

11

the magnitude of this difference varied with the calcium concentration, such that atmM calcium the variant and normal curves were quite similar. We concluded that theN-terminus of the variant B

β

chain participates in polymerization in a calcium-dependent association, probably during lateral aggregation. This result suggests thatthe N-terminus of the B

β

chain normally participates in lateral aggregation, and thatthe normal association is calcium-dependent.

Protofibril Formation and the Influence of Calcium

The formation of half-staggered, two-stranded protofibrils is mediated through

A

:

a

interactions, between the

A

site in the central domain exposed by FpA releaseand

a

site in the C-terminal domain of the

γ

chain. Recent high-resolution structuraldata of the C-terminal domain with bound GPRP, an

A

site analog, have identifiedspecific

A

:

a

interactions, including the side chain of Asp364.

12,13

The structural dataalso identified the residues that compose the high-affinity calcium binding site,including the side chains of Asp318 and Asp320. These two binding sites are juxta-posed in the C-terminal domain. We have synthesized three variants with substitu-tions in the

a

site, Y363A fibrinogen where Tyr363 is replaced with Ala, D364A andD364H fibrinogens where Asp364 is replaced with either Ala or His; and four vari-ants with substitutions in the calcium binding site,

∆

319,320 fibrinogen whereAsn319 and Asp 320 have been deleted; D318A fibrinogen where Asp 318 isreplaced with Ala; D320A fibrinogen where Asp 320 is replaced with Ala; andD318A,D320A fibrinogen where both Asp318 and Asp320 are replaced with Ala.

We focused our

a

site studies,

14

which were initiated prior to the availability ofthe high resolution structure, on Tyr

γ

363, identified by photoaffinity labeling offibrinogen with a GPRP analog,

15

and Asp

γ

364, identified in the dysfibrinogen Mat-sumoto I.

16

As is shown in F

IGURE

2, thrombin-catalyzed polymerization of variantswith substitutions at either site was significantly impaired, but polymerization wasevident for both Y363A and D364A. In contrast, with D364H there was no changein turbidity under these conditions. These results indicated that the

A

:

a

interactionswere compromised by the two Ala substitutions and eliminated by the His substitu-tion. Recently, we have used plasmin protection assays to independently assess func-tion of the

a

site in the

γ

364 variants. As shown by Lounes

et al.

(these proceedings),plasmin digestion of either D364A or D364H was not changed by the presence of1 mM GPRP; in contrast, as previously reported,

17 plasmin cleavage of normalfibrinogen was limited by the presence of GPRP. This result confirmed that the a sitewas defective in both variants. The polymerization and plasmin protection experi-ments together suggest that both Ala and His substitutions cripple the A:a interac-tion, but the D364H substitution has more profound consequences than the D364Asubstitution. We conclude that multiple interactions, including protofibril formation,were impacted by the histidine substitution. In contrast, whereas normal A:a interac-tions may be lost with the alanine substitution, remaining interactions were able tosupport polymer formation. Further experiments are needed to characterize the dif-ferences between these variants.

We initiated studies of the calcium binding domain by synthesizing a variant anal-ogous to the dysfibrinogen Vlissingen/Frankfurt IV.18 Using turbidity at 350 nm,


FIGURE 2. Thrombin-catalyzed fibrinogen polymerization. The change in turbidityover time with normal recombinant fibrinogen (N) and three γ chain variant fibrinogens,Y363A, D364A, and D364H. Thrombin (Panels A and B, 0.05 unit/ml; Panel C, 0.5unit/ml) was added at 0 min to fibrinogen at 0.09 mg/ml (Panels A and C) or to fibrinogenat 0.45 mg/ml (Panel B). From Reference 14.


we have never detected polymerization with this variant, even when the thrombinconcentration was increased to 10 U/ml, the fibrinogen concentration increased to1.3 mg/ml, the calcium concentration increased to 10 mM, the temperature loweredto 14°C, or the experiment monitored for 14 hours. Using dynamic light scattering,we confirmed that this variant fibrin does not polymerize. The experiments shown inFIGURE 3 were performed under conditions where within a few seconds normalfibrinogen would show a six- to 10-fold increase in mass and a 10-fold drop in nor-malized diffusion coefficient, consistent with the formation of protofibrils. Withγ∆319,320 fibrinogen the mass increased about twofold and the diffusion coefficientdropped about 1.5-fold after four hours consistent with the formation of small aggre-gates. The formation of small aggregates was also seen by transmission electronmicroscopy. We concluded that this variant could not form any ordered polymerstructure. In additional experiments, we were unable to detect either ADP-inducedplatelet aggregation or FXIIIa-catalyzed fibrinogen cross-linking with the γ∆319,320 variant. These results suggest that this two-residue deletion affected the overallstructure of the C-terminal domain of the γ chain. Therefore, it is not appropriate tolink these multiple changes in function specifically with the alteration at the high-affinity calcium binding site.

We synthesized two single residue substitutions in the calcium binding site,D318A and D320A, because we thought the single changes would not impact the

FIGURE 3. Dynamic light scattering. Fractional change in translational diffusioncoefficient over time. Thrombin (0.3 U/ml) was added to purified γ∆319,320 fibrinogen(0.2 mg/ml) and monitored for 4 h. D/D0, the normalized diffusion coefficient, is shown onone ordinate-axis and size distribution (units are monomers of fibrin) are shown on theopposite ordinate-axis. Three reactions are shown. From Reference 18.


overall structure of the C-terminal domain of the γ chain. We also synthesized thevariant where both residues are changed, D318A,D320A. In preliminary studieswith these variants (Lounes et al., this volume), we found that both D318A andD320A did polymerize, as detected by changes in turbidity, although in both casespolymerization was impaired. No change in turbidity was seen with theD318A,D320A variant. We also examined plasmin digests with these variants in thepresence of 1 mM GPRP, 5 mM calcium, and 1 mM EDTA. The results from theseplasmin protection studies indicate that the function of the a site depends uponthe structure and function of the high-affinity calcium binding site. This result con-trasts with the inverse interdependence, that is, the structure and function of the high-affinity calcium binding site is independent from the a site. We (Lounes et al., thisvolume) and others19 have shown in plasmin protection assays with variants at the asite, that GPRP protection was lost and calcium protection was normal.

Experiments are in progress to determine whether the D318A and D320A substi-tutions alter other functions associated with C-terminal domain of the γ chain.

The Connection between FpB Release and Polymerization

We synthesized three variants at residue γ308—N308A, N308I, and N308K—because several dysfibrinogens have been identified at this site. The crystallographicstructures show that this residue likely participates in D:D interactions and therebyinfluences polymerization.13 In preliminary experiments, we have found that poly-merization was impaired with all three variants, but the severity of the defect variedwith the specific substitution. Moreover, the differences were markedly dependenton the concentration of calcium. As is shown in FIGURE 4A and B in the absence ofadded calcium the lag time was longer and the rise in turbidity was slower for allthree variants relative to normal. In contrast, in the presence of 1 mM calcium, onlythe N308K variant remained abnormal; both the lag time and the rise in turbiditywere normal for N308A and N308I. These data indicate that the charged substitutiondisrupts the alignment of monomers in protofibrils, irrespective of calcium concen-tration. In contrast, the uncharged substitutions apparently influence calcium bind-ing and thereby influence protofibril formation. It is notable that our findings areconsistent with data from analysis of the dysfibrinogen N308I, or Baltimore III. Thisdysfibrinogen was identified as a heterozygous mutation in an asymptomatic womanwho had a prolonged thrombin time that was corrected by excess calcium.20 Thus,the connection between the substitution and the calcium dependent polymerizationwas detected in both plasma fibrinogen, which likely contained some normal γchains, and purified recombinant fibrinogen.

Furthermore, we found a similar association between calcium concentration andthe kinetics of thrombin-catalyzed FpB release. As is shown in FIGURE 4C, in theabsence of added calcium the rate of FpB release was reduced for all three variants.In the presence of 1 mM calcium the rate of FpB release from N308K remainedabnormal, whereas FpB release from N308A and N308I was normal. The coincidentcalcium dependence of the turbidity data and the kinetics of FpB release demonstratethe interdependence of these two events. We conclude that defective alignment of theD:D interface is associated with impaired release of FpB, because the addition ofcalcium not only normalized polymerization, but also normalized FpB release.Alternatively, it is reasonable to conclude that the Ala and Ile substitutions impaired


FIGURE 4. Calcium dependence of polymerization and FpB release. The lag time(Panel A) and Vmax (Panel B) determined from turbidity curves2 with normal (N), γN308A(A), γN308I (I), and γN308K (K) fibrinogens at 0.4 mg/ml fibrinogen, 0.05 U/ml thrombinin 20 mM HEPES, pH 7.4, in the absence of added calcium (0) or with 1 mM CaCl2 (1) and0.12 M NaCl. The specificity constants (Panel C) determined2 from progress curves offibrinopeptide release with normal (N), γN308A (A), γN308I (I), and γN308K (K) fibrino-gens at 0.1 mg/ml fibrinogen and 0.005 U/ml thrombin, with and without added calcium asfor Panel A.


calcium binding, such that in the absence of added calcium the D:D interactions andthe release of FpB were independently impaired.

B� Chain N-Terminal Residues and Lateral Aggregation

The release of FpB and exposure of the B site is thought to mediate lateral aggre-gation of protofibrils. High-resolution structures of fragment D and crosslinked frag-ment D crystallized in the presence of the peptide GHRP, which represents the B site,showed that GHRP binds to the C-terminal domain of the β-chain, in a pocket whosegeometry is analogous to the γ-chain GPRP binding pocket.21 Thus, the structuraldata show that the B:b interaction can occur. Nevertheless, biochemical data to sup-port a model where this B:b interaction mediates lateral aggregation are limited.Studies with fibrinogen lacking Bβ chain residues 1–42 and synthetic peptidesderived from the N-terminus of the Bβ chain support the conclusion that a polymer-ization center is located within this extended region.22,23 Both of these studies sug-gested, however, that residues in N-terminus of the β chain are involved with bothlateral aggregation and protofibril formation. In addition, photooxidation of BβH16specifically impaired protofibril formation.24 Finally, although GHRP binds tofibrinogen, it does not prevent clot formation.25 Together these studies demonstratethat residues in the B site influence polymerization, but the specific association ofB:b interactions with lateral aggregation remains ambiguous.

We synthesized BβH16A,R17A,P18A fibrinogen to study the role of the B site.Thrombin catalyzed polymerization of BβH16A,R17A,P18A showed no change inturbidity under conditions where normal fibrinogen reached a maximum turbiditywithin 10 min. For example, with 0.1 mg/ml fibrinogen, 0.1 U/ml thrombin inHEPES/NaCl with 0.1 mM CaCl2, no change in turbidity was seen with BβHRP16-18AAA fibrinogen, even when monitored overnight. Under these conditions wefound that thrombin catalyzed release of FpA was similar to normal, while releaseof FpB was delayed about 20-fold relative to normal. We also monitored polymer-ization at 20 mM CaCl2 and 110 mM NaCl. As is shown in FIGURE 5, polymerizationof BβH16A,R17A,P18A was apparent, although still remarkably impaired relativeto normal (FIG. 5A). We also examined Ancrod-catalyzed polymerization (FIG. 5B)and fibrin monomer polymerization (data not shown) with this variant. In both exper-iments polymerization was markedly impaired relative to normal, even at 20 mMCaCl2. We noted that Ancrod catalyzed polymerization of BβH16A,R17A,P18A wassimilar to thrombin-catalyzed polymerization of BβH16A,R17A,P18A (compareFIG. 5A and 5B), suggesting that polymerization proceeded by the same path irre-spective of which enzyme was used. We conclude that the N-terminal region of theBβ-chain is important for polymerization, with or without FpB release. Furthermore,because high calcium concentrations were essential for any polymerization, boundcalcium must affect a structure or structures that facilitate polymerization. That is,bound calcium either changed a facilitating structure in the distal domain, so it couldinteract with the altered central domain, or changed the structure of the variant siteitself, or both.

To determine whether protofibril formation or lateral aggregation or both wereimpaired, we monitored polymerization by static light scattering. In contrast to theturbidity experiments, we saw changes in light scattering even in the absence of cal-cium. We performed several experiments at varying the calcium concentrations, plot-ted the normalized intensity of scattered light versus time, and determined the lag


time and average rate constants from exponential fits of the data. We found that therate of polymerization was increased and the lag time decreased in the presence ofcalcium (1–5 mM) relative to no added calcium. We also performed dynamic lightscattering experiments and these likewise showed a calcium dependent increase inthe rate of protofibril formation. Aside from the rate effects seen on adding calcium,protofibril formation was normal such that normal sized aggregates were formed.Because the turbidity experiments with no added calcium showed no polymerization,we concluded that lateral aggregation of the BβH16A,R17A,P18A was defective.Thus, these experiments provide direct evidence that the N-terminal region of the Bβchain participates in lateral aggregation. In the presence of calcium we saw a slowrise in turbidity, demonstrating that elevated calcium facilitated polymerization ofthis variant, although lateral aggregation likely remained compromised. Additionalexperiments, such as transmission electron microscopy, are needed to confirm thatthe light scattering data is consistent with normal protofibril formation. Neverthe-less, these preliminary data clearly demonstrate that changes in the hypothesized B

FIGURE 5. Average polymerization curves. The change in turbidity with time ofnormal (solid lines, n = 3) and BβH16A,R17A,P18A (dashed lines, n = 3) fibrinogen with0.1 mg/mL fibrinogen and 0.1 U/mL thrombin (Panel A) or 0.1 U/ml Ancrod (Panel B) in20 mM HEPES, pH 7.4, 0.11 M NaCl, 20 mM CaCl2.


site markedly altered lateral aggregation, in a calcium dependent manner, with littleor no change in protofibril formation. These experiments are the first to link the Bsite specifically to lateral aggregation.

DISCUSSION

The variants described here were designed to examine the three steps in thrombincatalyzed polymerization: fibrinopeptide release, protofibril formation, and lateralaggregation. Analysis of each of these variants has proven to be informative. More-over, when considered together, these analyses provide novel insight into the mech-anism of thrombin catalyzed polymerization. Furthermore, by considering thesebiochemical data alongside the high-resolution structural data, it has been possibleto achieve a more detailed description of the interactions that support the conversionof fibrinogen to a fibrin clot.

With AαF8Y and A′β fibrinogens, we have seen that the substrate specificity ofthrombin dictates the order of fibrinopeptide release, suggesting that enzyme alonecontrols the timing of FpB release. Yet, analyses of the γ308 variants suggest thatpolymerization, specifically proper alignment of the D:D interface in protofibrils,controls the timing of FpB release. Perhaps the enzyme specificity ensures that FpArelease precedes FpB release, whereas protofibril formation controls the length ofthe interval between FpA and FpB release. Thus, when we examined fibrinopeptiderelease from A′β fibrinogen, where all four fibrinopeptides are equivalent, the sub-strate specificity preempted the influence of polymerization and the addition ofGPRP was inconsequential. Alternatively, perhaps the link between polymerizationof the γ308 variants and FpB release is coincidental. Perhaps calcium controls boththe order and timing of fibrinopeptide release. Thus, if we altered a calcium bindingsite when we replaced FpB with FpA′, and this is the same calcium binding site thataffected polymerization of N308A and N308I, then the timing of FpB release wouldbe changed in both variants. Clearly more experiments are needed to resolve the sig-nificance of the timing of FpB release.

We have seen that lateral aggregation occurs without the release of FpB. Ofcourse, this has been known for some time, since Batroxobin- or Ancrod-catalyzedpolymerization, which removes only FpA, yields essentially normal fibers, albeitwith somewhat slower kinetics. Our studies with N-terminal Bβ chain variants(BβR14H and BβH16A,R17A,P18A) suggest that irrespective of the release of FpB,the N-terminus of Bβ chain participates in lateral aggregation. With both variants, theturbidity studies with of desA fibrin were abnormal. With both variants the magnitudeof the differences in polymerization, catalyzed by either thrombin or the thrombin-like protease, varied with calcium concentration. We conclude that the N-terminus ofthe variant Bβ chain participates in polymerization in a calcium dependent associa-tion. Because the BβH16A,R17A,P18A light scattering data demonstrated apparent-ly normal protofibril formation, we surmise that protofibril formation is normal forboth N-terminal Bβ chain variants. Thus, with both variants, lateral aggregation isdefective. Therefore, we conclude that the N-terminus of the Bβ chain normally par-ticipates in lateral aggregation, and that the normal association is calcium dependent.Previous studies have also linked this N-terminal domain with calcium dependent


polymerization. Polymerization of the dysfibrinogen New York I, lacking Bβ chainresidues 9–72, occurred only in the presence of calcium.26 Reptilase-catalyzed poly-merization of desBβ1–42 fibrinogen increased significantly in the presence of 10 mMcalcium.23 Taken all together, these experiments suggest a specific role for calciumin the lateral aggregation of the N-terminal Bβ chain variants. These analyses are notsufficient to understand the link between calcium and lateral aggregation, but they doconnect calcium with the N-terminal domain of the Bβ chain. We suggest that in thesevariants, the high-affinity calcium binding site in the E domain has been changed toa low-affinity site. Thus, higher calcium concentrations were required for lateralaggregation of these altered fibrinogens.

Our experiments have also demonstrated that measurements of turbidity at 350nm do correlate with polymerization, such that a change in the lag time or the rateof turbidity rise does demonstrate a change in polymerization. As expected, however,abnormal turbidity analyses do not distinguish between abnormal protofibril forma-tion and abnormal lateral aggregation. In particular, we demonstrated that a com-plete loss of turbidity can be associated with either a loss of protofibril formation, asshown with γ∆319,320, or normal protofibril formation and a loss of lateral aggre-gation, as shown with BβH16A,R17A,P18A in the absence of added calcium.

FUTURE DIRECTIONS

The results described here suggest two ambiguous aspects of thrombin catalyzedpolymerization that can be addressed by further analysis of variant fibrinogens. First,what is the mechanism by which bound calcium so profoundly influences thrombincatalyzed polymerization? Do both high-affinity and low-affinity sites regulate poly-merization? Second, where is the b site and does it have a role in lateral aggregation?Does the role of the N-terminus of the Bβ chain depend on the release of FpB, suchthat B:b interactions mediate lateral aggregation? Based on the recently publishedcrystallographic structure determined in the presence of GPRP and GHRP, we planto synthesize new variants to facilitate our understanding of these two aspects.

Our findings suggest that both strong and weak calcium binding sites have aninstrumental role in polymerization. We plan further analysis of existing variantswith substitutions in the strong, γ chain calcium binding sites; these experiments willlikely demonstrate whether this site participates directly in polymerization, or par-ticipates only by influencing the overall conformation of the D domain. We also planto measure calcium binding to A′β, BβR14H, and BβH16A,R17A,P18A fibrinogens.In addition, we plan to synthesize variants with substitutions in the three weak cal-cium binding sites that were recently identified in the structural data.21 One of thesesites, which is altered in the presence of GHRP, lies in the C-terminal module of theBβ chain. The second is found in the γ chain when GHRP is bound to the a site, andthe third is found in the C-terminus of the β chain at a position homologous to thehigh-affinity site in the γ chain. Biochemical analyses of these variants will addresswhether calcium bound at specific sites is critical to thrombin catalyzed polymeriza-tion, in particular, to lateral aggregation.

Our findings suggest that the N-terminus of the Bβ chain has a role in lateralaggregation, though it is not clear whether the release of FpB is critical. Thus, the


role of the B:b interactions remains ambiguous. The structural data obtained in thepresence of GHRP and GPRP identified residues within the C-terminal module ofthe Bβ chain that participate in GHRP binding. Thus, to locate the b site and deter-mine whether B:b interactions are important for polymerization, specifically for lat-eral aggregation, we plan to synthesize variants that will impair GHRP binding. Weplan to incorporate single residue substitutions in the b site that are analogous to asite substitutions that are known to impair GPRP binding. Based on our previouswork, we anticipate that studies with these new variants will not only address the twotargeted areas (calcium binding and B:b interactions), but will also contribute to ourunderstanding of the concerted interactions in the continuum of events that partici-pate in normal thrombin-catalyzed polymerization.

ACKNOWLEDGMENTS

We thank Li Fang Ping, who assisted in essentially every aspect of this work, andin particular for her help in generating all the cell lines and maintaining the culturesrequired for synthesis of the variant fibrinogens. Culture medium with normalrecombinant fibrinogen was purchased from the Cell Culture Center (Minneapolis,MN). We thank Kasim McLain for her assistance with the purification of multiplerecombinant fibrinogens. We thank John Weisel and Roy Hantgan, who collaboratedon the electron microscopy and light scattering studies, respectively, and moreover,who contributed to our appreciation of the complex processes of polymerization.Finally, we thank our laboratory colleagues—Cameron Binnie, Andrew Coates,Kelly Hogan, Karim Lounes, Jennifer Mullin, Nobuo Okumura, Michael Rooney,and Elizabeth Steele—whose experiments and ideas have provided the foundationand inspiration for this paper. This work was supported in part by NIH grantsHL31048 and HL52706.

REFERENCES

1. BINNIE, C.G., J.M. HETTASCH, E. STRICKLAND, et al. 1993. Characterization of purifiedrecombinant fibrinogen: partial phosphorylation of fibrinopeptide A. Biochemistry32: 107–113.

2. GORKUN, O.V., Y.I. VEKLICH, J.W. WEISEL, et al. 1997. The conversion of fibrinogento fibrin: recombinant fibrinogen typifies plasma fibrinogen. Blood 89: 4407–4414.

3. HENSCHEN, A.H. 1993. Human fibrinogen—structural variants and functional sites.Thromb. Hæmostasis 70: 42–47.

4. NG, A.S., S.D. LEWIS & J.A. SHAFER. 1993. Quantifying thrombin-catalyzed release offibrinopeptides from fibrinogen using high-performance liquid chromatography.Meth. Enzymol. 222: 341–358.

5. HIGGINS, D.L., S.D. LEWIS & J.A. SHAFER. 1983. Steady state kinetic parameters forthe thrombin-catalyzed conversion of human fibrinogen to fibrin. J. Biol. Chem. 258:9276–9282.

6. WEISEL, J.W., Y. VEKLICH & O. GORKUN. 1993. The sequence of cleavage of fibrin-opeptides from fibrinogen is important for protofibril formation and enhancement oflateral aggregation in fibrin clots. J. Mol. Biol. 232: 285–297.

7. LORD, S.T., P.A. BYRD, K.L. HEDE, et al. 1990. Analysis of fibrinogen Aα-fusion pro-teins. Mutants which inhibit thrombin equivalently are not equally good substrates.J. Biol. Chem. 265: 838–843.


8. ROONEY, M.M., J.L. MULLIN & S.T. LORD. 1998. Substitution of tyrosine for phenyla-lanine in fibrinopeptide A results in preferential thrombin cleavage of fibrinopeptideB from fibrinogen. Biochemistry 37: 13704–13709.

9. MULLIN, J.L., O.V. GORKUN, C.G. BINNIE, et al. 2000. Recombinant fibrinogen studiesreveal that thrombin specificity dictates order of fibrinopeptide release. J. Biol.Chem. 275: 25239–25246.

10. HIGGINS, D.L. & J.A. SHAFER. 1981. Fibrinogen Petoskey, a dysfibrinogenemia char-acterized by replacement of Arg-Aα 16 by a histidyl residue. Evidence for thrombin-catalyzed hydrolysis at a histidyl residue. J. Biol. Chem. 256: 12013–12017.

11. MULLIN, J.L. 2000. An Expanded and Refined Model of Fibrin Polymerization Basedon Studies with Variant Recombinant Fibrinogens. Ph.D. Thesis, Department ofChemistry, University of North Carolina, Chapel Hill.

12. PRATT, K.P., H.C.F. COTE, D.W. CHUNG, et al. 1997. The primary fibrin polymeriza-tion pocket 3-dimensional structure of a 30-Kda C-terminal γ chain fragment com-plexed with the peptide Gly-Pro-Arg-Pro. Proc. Natl. Acad. Sci. U.S.A. 94: 7176–7181.

13. SPRAGGON, G., S.J. EVERSE & R.F. DOOLITTLE. 1997. Crystal structures of fragment Dfrom human fibrinogen and its crosslinked counterpart from fibrin. Nature 389: 455–462.

14. OKUMURA, N., O.V. GORKUN & S.T. LORD. 1997. Severely impaired polymerization ofrecombinant fibrinogen γ364 Asp—His, the substitution discovered in a heterozy-gous individual. J. Biol. Chem. 272: 29596–29601.

15. YAMAZUMI, K. & R.F. DOOLITTLE. 1992. Photoaffinity labeling of the primary fibrinpolymerization site: localization of the label to γ-chain Tyr-363. Proc. Natl. Acad.Sci. U.S.A. 89: 2893–2896.

16. OKUMURA, N., K. FURIHATA, F. TERASAWA, et al. 1996. Fibrinogen Matsumoto I: aγ364 Asp→His (GAT→CAT) substitution associated with defective fibrin polymer-ization. Thromb. Hæmostasis 75: 887–891.

17. YAMAZUMI, K. & R.F. DOOLITTLE. 1992. The synthetic peptide Gly-Pro-Arg-Pro-amide limits the plasmic digestion of fibrinogen in the same fashion as calcium ion.Prot. Sci. 1: 1719–1720.

18. HOGAN, K.A., O.V. GORKUN, K.C. LOUNES, et al. 2000. Recombinant fibrinogen Vliss-ingen/Frankfurt IV. The deletion of residues 319 and 320 from the γ chain of fibrin-ogen alters calcium binding, fibrin polymerization, cross-linking, and plateletaggregation. J. Biol. Chem. 275: 17778–17785.

19. COTE, H.C.F., K.P. PRATT, E.W. DAVIE, et al. 1997. The polymerization pocket “a”within the carboxyl-terminal region of the γ chain of human fibrinogen is adjacent tobut independent from the calcium-binding site. J. Biol. Chemistry 272: 23792–23798.

20. EBERT, R.F. & W.R. BELL. 1988. Fibrinogen Baltimore III: congenital dysfibrinogene-mia with a shortened gamma-subunit. Thromb. Res. 51: 251–258.

21. EVERSE, S.J., G. SPRAGGON, L. VEERAPANDIAN, et al. 1999. Conformational changes infragments D and double-D from human fibrin(ogen) upon binding the peptide ligandGly-His-Arg-Pro-amide. Biochemistry 38: 2941–2946.

22. PANDYA, B.V., J.L. GABRIEL, et al. 1991. Polymerization site in the β chain of fibrin:mapping of the Bβ 1–55 sequence. Biochemistry 30: 162–168.

23. SIEBENLIST, K.R., J.P. DIORIO, A.Z. BUDZYNSKI, et al. 1990. The polymerization andthrombin-binding properties of des-(Bβ 1–42)-fibrin. J. Biol. Chem. 265: 18650–18655.

24. HENSCHEN-EDMAN, A.H. 1997. Photo-oxidation of histidine as a probe for aminotermi-nal conformational changes during fibrinogen-fibrin conversion. Cell Mol. Life Sci.53: 29–33.

25. LAUDANO, A.P., B.A. COTTRELL, R.F. DOOLITTLE. 1983. Synthetic peptides modeledon fibrin polymerization sites. Ann. N.Y. Acad. Sci. 408: 315–329.

26. LIU, C.Y., J.A. KOEHN & F.J. MORGAN. 1985. Characterization of fibrinogen NewYork I. A dysfunctional fibrinogen with a deletion of Bβ (9–72) correspondingexactly to exon 2 of the gene. J. Biol. Chem. 260: 4390–4296.


27. ROONEY, M.M., D.H. FARRELL, B.M. VAN HEMEL, et al. 1998. The contribution of thethree hypothesized integrin-binding sites in fibrinogen to platelet-mediated clotretraction. Blood 92: 2374–2381.

28. MULLIN, J., O. GORKUN & S. LORD. 2000. Decreased lateral aggregation of a variantrecombinant fibrinogen provides insight into the polymerization mechanism. Bio-chemistry 39: 9843–9849.

29. LORD, S.T., E. STRICKLAND & E. JAYJOCK. 1996. Strategy for recombinant multichainprotein synthesis: fibrinogen Bβ-chain variants as thrombin substrates. Biochemistry35: 2342–2348.

30. GORKUN, O.V., A.H. HENSCHEN-EDMAN, L.F. PING, et al. 1998. Analysis of Aα251fibrinogen: the αC domain has a role in polymerization, albeit more subtle thananticipated from the analogous proteolytic fragment X. Biochemistry 37: 15434–15441.

31. ROONEY, M.M., L.V. PARISE & S.T. LORD. 1996. Dissecting clot retraction and plateletaggregation. Clot retraction does not require an intact fibrinogen γ chain C terminus.J. Biol. Chem. 271: 8553–8555.

32. FARRELL, D.H., P. THIAGARAJAN, D.W. CHUNG, et al. 1992. Role of fibrinogen α and γchain sites in platelet aggregation. Proc. Natl. Acad. Sci. U.S.A. 89: 10729–10732.

monitoring for neuroprotection : new technologies for the new millennium

Documents