Fibrinogen Sialic Acid Residues Are Low Affinity Calcium-binding ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 25, Issue of September 5, pp. 15104-15108,1989 Printed in U.S.A.

Fibrinogen Sialic Acid Residues Are Low Affinity Calcium-binding Sites That Influence FibrinAssembly* (Received for publication, May 2, 1989)

Chi V. DangSgV, Chung K. Shin$, William R. Bell$$ Chandrasekar Nagaswarnill, and John W. Weisell1 From the $Hematology Division, Departmentof Medicine, The Johns Hopkins UniversitySchool of Medicine, Baltimore, Maryland 21205 and the I(Department of Anatomy, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104

Calcium ions occupy low ( n = 10; Kd = 1 mM) and high ( n = 3; K d 1 pM) affinity sites on fibrinogen and facilitate fibrin monomer polymerization. We have previously localized two of the three high affinity Ca2+ sites to y3 11-7336.However, optimal enhancement of fibrin monomer polymerization occurs only at physiological millimolar Ca2+concentrations which are two orders of magnitudehigher than the concentration required for occupancy ofthe high affinity Ca2+-binding sites. In this study, we show that removal of fibrinogen sialic acid residues results in lossof low affinityCaz+binding sites. Clotting of asialofibrinogen appears to be Ca2+-independent and results in fiber bundles thicker in diameter than normal fibrin bundles as determined by turbidometryand scanning and transmission electron microscopy. By using a Ca2+-sensitive electrode, free sialic acid is shown to bind Ca2+(Kd= 1 mM). These observations suggest that the high affinity fibrinogen D-domain Ca2+-bindingsites may play a role in the tertiary structure of the D-domain, whereas, sialic acid residues are low affinity sites whose occupancy by Ca2+at physiological calcium concentration facilitates fibrin polymerization.

Fibrinogen is a 340-kDa plasma glycoprotein which plays a pivotal role in blood clot formation (1, 2). It consists of three pairs of nonidentical polypeptide chains: A a (Mr67,000), BP (Mr56,000), and y (Mr47,000). The carbohydrate occurs as a complex biantennary oligosaccharide with terminal sialic acid residues N-linkedand to (3). The biochemical events leading to theconversion of fibrinogen to cross-linked fibrin gel can be described by four general steps: 1) release of fibrinopeptides and formation of fibrin monomers by thrombin proteolytic cleavage of fibrinogen; 2) formation of doublestranded protofibrils by linear association of fibrin monomers; 3) formation of fibrin fibers of varying diameters by the association of protofibrils; and 4) covalent cross-linking of polymerized fibrin by the transglutaminase Factor XIII. * This work was supported in part by National Institutes of Health (NIH) GrantHL30954 (to J. W. W.), by NIH Grant HL36260 and a grant from the American Heart Association (Maryland chapter) (to W. R. B.), and by an NIH biomedical research support grant (to C. V. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § W. M. Keck Foundation clinician scientist and recipient of H. M. and L. Stratton Foundation Award. To whom correspondence should be addressed: 109 Hunterian, The JohnsHopkins University School of Medicine, Baltimore, MD 21205. 11 Hubert E. and Anne E. Rogers Scholar.

Although it has been known for several decades that calcium facilitates fibrin monomer polymerization (4), the molecular mechanism by which Ca2+accelerates fibrin polymerization is not understood. Calcium does not appear to affect the ratesof release of fibrinopeptides A or B by thrombin ( 5 ) . Fibrinogen contains three high affinity (& = 1 p M ) calciumbinding sites and 8-10 low affinity (& 2 1 mM) sites (6-8). Although the properties of the high affinity sites have received considerable scrutiny, to the point that the peptide loci of the binding sites have been identified (7, 8), the loci of the low affinity sites remain unclear. Equally unclear are the functions of these low affinity sites and thebiochemical basis for the calcium-dependent rate of fibrin polymerization. It has been observed that an abnormally high number of sialic acid residues on fibrinogen from patients with hepatic disorders decreases the rate of fibrin polymerization ( 9 - l l ) , whereas removal of sialic acid by neuraminidase results in asialofibrin which polymerizes rapidly (9). The enhancement of fibrin polymerization is not due to enhanced fibrinopeptide release (9) nor exposure of new terminal carbohydrate moieties because total removal of carbohydrates from fibrinogen also enhances fibrin polymerization (12). From these observations, it appears that sialic acid residues may contribute to electrostatic repulsion between fibrin molecules. In this report, we show that removal of fibrinogen sialic acid residues results in the loss of low affinity Ca2+-binding sites. Moreover, polymerization of asialofibrin appears to be independent of Ca2+concentration and results in fibrinfibers greater in diameter than normal fibrin. Also, free sialic acid is shown to bind Ca2+directly. These data support the hypothesis that sialic residues constitute most of the fibrinogen low affinity Ca2+-bindingsites. Occupancy of these sites at physiological free Ca2+ concentration(0.8-1.5 m M ) decreases intermolecular repulsion and facilitates fibrin polymerization through specific protein-protein interactions. Thus, these unoccupied binding sites provide the biochemical basis for delayed polymerization of acquired dysfibrinogens with increased sialic acid content. MATERIALSANDMETHODS

Preparation of Asialofibrinogen-Asialofibrinogen (AF)’ was prepared by incubating purified human fibrinogen (grade L; Kabi-Vitrum, Stockholm, Sweden) with neuraminidase(0.02units/mg fibrinogen) from Clostridiumperfringens (Sigma) for 3 h at 37 “cin 10 mM bis-Tris, pH 6.3,and 0.15 M NaCl (9,10).Untreated normal fibrino-

’

The abbreviations used are: AF, asialofibrinogen; bis-Tris, 2[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)-propane-1,3-diol; F, untreated normal fibrinogen; SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis; AFD, Factor XIII-deficient asialofibrinogen; FD, Factor XIII-deficient fibrinogen.

15104

FibrinogenCa2+-binding Sialic Sites Acid gen (F) was incubated under the same conditions in the absence of neuraminidase. Complete desialylation of AF was verified by hydrolyzing the protein a t 80 "C for 1h in 0.1 N H2S04, and thesialic acid content was measured as described (13). Both AF and F were dialyzed first at 4°C against 100 volumes of 25 mM Tris-HC1, pH 7.26,lO mM EDTA, and 0.15 M NaCl and then dialyzed exhaustively (2000 volumes) against 25 mM Tris-HC1, pH 7.26, 150 mM NaC1. Fibrinogen concentration was determined spec= 1.5 (14). trophotometrically using ziEm Preparation of Factor XZZZ-deficient Fibrinogen-Factor XIII-deficient fibrinogen was prepared by linear gradient DEAE-cellulose chromatography using 0.005 M HaPo,, 0.039 M Tris,pH 8.6, as starting buffer and 0.5 M H3P04, 0.5 M Tris, pH 4.1, as final buffer (15). Purified fibrinogen was first dialyzed extensively (1000 volumes) against the starting buffer before being loaded onto the DEAEcellulose column. Eluted fractionswere collected and dialyzed against 10 mM bis-Tris, pH 6.3,0.15 M NaCl in preparation for desialylation. Factor XIII-deficient AF was prepared by desialylation as described (10). The presence and absence of Factor XI11 in protein solutions were determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to examine the extent of y-chain crosslinking after clottingwith purified a-thrombin (1unit/mg fibrinogen; a gift from Dr. J. Fenton, New York State Department of Health, Albany, NY). Detection of High Affinity Ca2' Sites-AF and F were electrophoresed using SDS-PAGE in nonreducing conditions and then transferred onto nitrocellulose sheets using a constant current of 400 mA for 2 h at 4 "C in 25 mM Tris, 0.2 M glycine, pH 8.8. The sheet was washed three times in 60 mM KC1, 5 mMMgC12, 10 mM imidazole, pH 6.8, for 20 min each and incubated in the same buffer containing 1 mCi/liter 45Ca2+(20 mCi/mg; Du Pont-New England Nuclear) for 10 min at 20°C (16). The membrane was then rinsed twice with deionized water for 5 min each, air-dried, and autoradiographed with Kodak XAR-5 x-ray film for 24 h. Fibrinogen Clotting Assay-Clotting assays were performed with fibrinogen solutions (1 mg/ml) in 10 mM Tris-HC1, pH 7.4,0.15 M NaCl as described (7). To 1 ml of fibrinogen solution, 1 unit of athrombinin10 pl was added to initiate the reaction which was continuously monitored by absorbance a t 350 nm using a PerkinElmer model X 4B spectrophotometer. The water-jacketed cuvette chamber was thermostated to 20 "C. To determine relative fibrin fiber radii, 1 mg/ml fibrinogen solutions in 10 mM Tris-HC1, pH 7.4, 0.15 M NaCl were clotted with 1 unit/ml thrombin for 1 h a t 20 "C. Turbidities ( 7 ) of fibrin gels were determined by absorbances a t various wavelengths between 350 and 650 nmat 20 "C. Relative fibrin fiber radii were determined according to Equation 1 (18), 7

= [ (88/l5).lr3n(dn/dc)'Crcl/NX3

*SF.

niques for the preparation of clots for transmission electron microscopy by the method of negative contrast have been described previously (17). Clots were dispersed with a Pasteur pipette, applied to 300-mesh carbon-coated Formvar grids and negatively contrasted with 1%uranyl acetate. Specimens were examined in a Philips 400 electron microscope equipped with a low dose kit; microscope magnifications were calibrated with negatively contrasted tropomyosin paracrystals. Specimens for scanning electron microscopywere prepared by variations of methods developed in the laboratory of John W. Weisel and described previously (12). Clots were formed in specially designed Plexiglass microdialysis cells, washed to remove excess salt, and fixed with 2% glutaraldehyde. After osmium treatment and dehydration, the clots were critical-point dried, mounted, and sputter-coated with gold palladium. Specimens were examined in an AMRAY 1400 scanning electron microscope equipped with a turbomolecular pump.

RESULTS

Calcium-independent Clotting of Asialofibrinogen-We examined the calcium-dependent clotting rate of AF and untreated F that were depleted of calcium by EDTA treatment (see"Materials and Methods"). The untreated F displays virtually no polymerization inthe presence of 20 mM EDTA (Fig. lA),whereas AF displays significant polymerization even in the presence of 20 mM EDTA (Fig. 1B).Moreover, F shows a monotonic increase in polymerization rate with increasing Ca2+concentration (Fig. lA). In contrast, AF displays essentially calcium concentration-independent polymerization between no added (-0.1 PM; Ref. 7) and 1 mM Ca2* (Fig. 1B). W i t h a supraphysiological calcium concentration of 10 mM Ca2+,both F and AF display additionalenhancement of polymerization. The times to reach half-maximal turbidity for F and AF at various Ca2+ concentrations are shown in Table I. Analysis of fibrin subunits bySDS-PAGE indicated that significant Factor XIII-mediated cross-linking of the ychain occurred within 5 m i n at 10 mM Ca2+(Fig. 3). In t h e

@

@

2

a

E

b

(1)

where n is the solution refractive index, dn/dc is the refractive index increment, X is the wavelength, C is the concentration of fibrinogen in g/ml, N is Avogadro's number, and p is the mass/length ratio. The mass/length ratio is related to fiber radius, r, and fiber density, 6 (18), as follows: p =

15105

(2)

0

1

2 time [min]

3

4

0

1

2

3

4

time [mln]

FIG. 1. A, calcium-dependent clotting of human fibrinogen determined by turbidometry (absorbance a t 350 nm). Clotting of Ca2+depleted fibrinogen (1 mg/ml) (see "Materials and Methods") was initiated with thrombin (1 unit/ml) at 20 "C and at various Ca'+ or EDTA concentrations: 20 mM EDTA (O), no added Ca2+(v),O.l mM CaZ+(A), 1 mM Ca2+(0),or 10 mM Ca2+(V). The inset represents a Coomassie Blue-stained SDS-polyacrylamide gel of untreated fibrinogen subunits ( l a n e a ) and asialofibrinogen subunits(lane b). B, clotting of asialofibrinogen a t various EDTA or CaZ+concentrations as described in A.

Calcium Titration-Solutions of AF and F were prepared as described above and diluted to a final concentration of 1.1% (w/v). The reference electrode for the calcium-sensitive electrode was made by filling a 200-p1 micropipette tube with 1.5% (w/v) agar (FMC BioProducts, Rockland, ME) dissolved in 3 M KCl.Voltage readings were taken on a Corning pH meter 140 set on the mVmode. The calcium-sensitive electrode was obtained from World Precision Instruments, Inc. (New Haven, CT). In the presence of calcium-sensitive and reference electrodes, 10TABLE I pl aliquots of 0.01 M CaC12dissolved in 25 mM Tris, 0.15 M NaCl, pH Time to reach half-maximal ( t d turbidity of clots forming from 7.26, were added to 2 ml of protein solution in a 5-ml beaker. The untreated or Factor XIZI-free normal and asialofibrinogen beaker was gently swirled manually, and the solution was allowed to t, was the time inmin to reach half-maximal turbidity determined reach steady state (-1 min) before readings were taken. Electromotive forces were corrected for volume changes. For titration of free sialic by absorbance a t 350 nm (Figs. 1 and 2). acid (Sigma), 1%(w/v) solutions of N-acetylneuraminic acid were [Ca2+] added FAFD FD AF used. All steps were performed at room temperature. mM Electron Microscopy-To facilitate direct comparison of results, 01.5 3.2 1.2 samples for electron microscopywere identical to those used for 2.4 0.1 1.6 1.1 1.5 0.8 biochemical experiments. Additionally, the results obtained by ex1 1.4 0.7 1.1 1.2 amination of specimens by electron microscopy were first interpreted 10 1.3 0.6 1.1 0.7 without knowledge of the results of biochemical experiments. Tech-

15106

Sites

ea2+-binding Acid Fibrinogen Sialic

absence of thrombin, 10 mM Ca2+does not cause nonspecific fibrinogen aggregation (data not shown). As Factor XI11 might enhance the turbidity of fibrin solutions, we examined Factor XIII-deficient AF (AFD) and F (FD) for calcium-dependent clotting. FD displays a monotonic increase in polymerization rate with increasing calcium concentrations (Fig. 2 A ) . AFD also displays an increase in polymerization at 10 mM Ca2+ (Fig. 2 B ) as compared to AF (Fig. 1B).Polymerization of AFD appears to be calcium concentration-independent between no added and 1 mM Ca2+. The times to reach half-maximal turbidity of FD and AFD are shown in Table I. It is noteworthy that AFD displays slightly less maximum turbidity as compared to AF at various Ca2+ concentrations. Analysis of fibrin subunits on SDS-PAGE confirmed the relative paucity of y-chain cross-linking in AFD as compared to AF (Fig. 3). Hence, it appears that Factor XI11 is not necessary for the enhancement of turbidity at 10 mM Ca2+. Asialofibrinogen displays higher maximal turbidity than F at different calcium concentrations, suggesting that AF forms thicker fibrin fibers than F (Fig. 1).To determine fibrin fiber thickness, we took two different approaches. The relative fibrin fiber radii were determined by turbidities measured in the range from 350 to 650 nm at various calcium concentrations (18).F displays increasing fiber bundle thickness with

increasing calcium concentrations, whereas AF displays much thicker fiber bundles than F virtually independentof calcium concentration (Table 11). Note thatin theseexperiments absorbances were measured 1h after clotting was initiated by thrombin, as compared to shorter time points in the polymerization assays( sites in asialofibrinogen and the direct interaction of free E sialic acid with calcium strongly support the localization of 0 some of the low affinity fibrinogen calcium-binding sites to 0 3 sialic acid residues. 100 j The hypothesis that fibrinogen sialic acid residues are low 4.0 3.8 3.4 3.6 3.2 affinity Ca2+-bindingsites makes several specific predictions that will each be discussed below. First, removal of sialic acid -I og [ ~ a ” ] residues should enhance fibrin polymerization because calFIG. 6. Binding of Ca2’ to free sialic acid determined by a cium is not required to neutralize sialic acid moieties. Second, Ca2+-sensitiveelectrode. Electromotive forces of distilled deion- asialofibrinogen should clot independently of Ca2+concentraized water (pH 6.9) at various calcium concentrations are shown in the presence of 1% (w/v)glycerol (0) or sialic acid (0).Inset, tion. Third, asialofibrinogen should bind less calcium than Scatchard analysis of sialic acid calcium binding. E,bound Ca2+;B / untreated fibrinogen. Fourth, sialic acid is capable of binding calcium directly. F, bound/free Ca2+. Removal of sialic acid residues (9-11) or complete removal fibers weremeasuredfrom the micrographs of specimens of carbohydrates (12) from fibrinogen resulted in significant observed under each of the conditions listed in Tables I and enhancement of the rate and extent of lateral aggregation. 11. Fiber bundles formed from AF under all conditions or from Our results confirm that asialofibrinogen clots more rapidly F in the presence of all levels of Ca2+were about 70% larger and displays higher maximal turbidity than normal fibrinogen than those formed from F without Ca2+.Quantitative differ- (9-11). The clotting behaviors of asialofibrinogen and deglyences among the clots formed under these various conditions cosylated fibrinogen are similar; however, it appears that the are more accurately calculated from the turbidity measure- magnitude of the observed effects are larger upon deglycosylation (12). Furthermore, although the effects of removal of ments as described above (Table 11). sialic acid are similar to those of adding calcium to normal Loss of Low Affinity Calcium Binding in AsialofibrinogenTo determine calcium binding in fibrinogen, we used a Ca2+- fibrinogen, the effects of deglycosylation are considerably sensitive electrode which determines free Ca2+concentrations greater in magnitude. Although we do not understand the in the physiological range. Titration of AF and F with Ca2+ differences in behaviors between deglycosylated and desialyclearly showslesscalcium binding by asialofibrinogen as lated fibrinogens, these observations support the hypothesis evidenced by the presence of more free Ca2+ (Fig. 5A) at that sialic acid residues might be neutralized by Ca2+ to various amounts of added Ca2+when compared to untreated facilitate clotting of fibrinogen. Consistent with the hypothesis, asialofibrinogen clots infibrinogen (Fig. 5A). Scatchard analysis was performed (Fig. 5B) to estimate Ca2+-bindingaffinities of AF and F. Based dependently of calcium concentration up to 1 mM. However,

15108

Fibrinogen Sialic Acid Ca2+-binding Sites

at 10 mM Ca”, there is a significant increase in turbidity of the clot. In the absence of thrombin, 10 mM Ca2+ does not cause nonspecific aggregation of fibrinogen; however, in its presence,significant y-chain cross-linking was observed. These results are consistent with the activation of Factor XI11 a t high calcium concentrations (19). T o determine a possible contribution of contaminating Factor XI11 to the increased turbidity a t 10 mM Ca”, fibrinogenwassubjected to ionexchange chromatographyto remove Factor XIII. AF and FactorXIII-depleted AF both display an enhancement of turbidity at 10 mM Ca2+, suggestingthat FactorXI11 does not contribute to increased turbidity by cross-linking reactions. Because there is a detectable increase in polymerization rate at 10 mM Ca2+ with Factor XIII-depleted fibrinogen, we can not rule out thepresence of other low affinity calcium-binding sites. We measured the binding of calcium to asialofibrinogen and untreated fibrinogen by using acalcium ion-sensitive electrode and observed a significant decrease in low affinity calcium binding with asialofibrinogen. This observation suggests that sialic acid residues are required for low affinity calcium binding. Using a protein blotting techniqueto determine high affinity calcium binding, treatment of fibrinogen with neuraminidasedoes not appear toaffect the high affinity Ca2+-binding sites as compared to untreatedfibrinogen. We demonstrated further that free sialic acid in solution binds calciumwith an affinity in the order of Kd c- 1 mM. Our results are consistent withprevious reports that sialic acid is a calcium-binding carbohydrate (20-22). Hence, these results support the hypothesis that sialic acid residuesare fibrinogen low affinity calcium-binding sites. Sialic acidcontent in normal fibrinogen has been estimated to be 6 mol of sialic acid/mol of fibrinogen, but there arefour biantennary carbohydrate moieties capable of holding 8 sialic acid residues/fibrinogen molecule (3, 23). If each sialic acid residue is capable of binding one calcium ion (18-20), then the average normal fibrinogen (6 mol of sialic acid/mol of fibrinogen) should have six low affinity Ca2+-binding sites. However, the number oflow affinity calcium-binding sites has been determined tobe as high as 15 for bovine fibrinogen and 5-8 for rat fibrinogen (24, 25). Because the sialic acid residues account for only sixlow affinity sites and the polymerization rate of asialofibrinogen is further increased at 10 mM Ca2+,we cannot rule out theexistence of other non-sialic acid low affinity calcium-binding sites in fibrinogen. In fact, a number of new calcium-binding sites have been observed upon fibrin monomer polymerization(26,27). These new sites might be secondary to exposureof cryptic Ca2+-binding sites upon polymerization. The subsaturationsialic acid content of normal fibrinogen (6 versus an expected 8 mol of sialic acid/mol of fibrinogen) is consistent with the occurrence of “hypersialylated” fibrinogen (8-10 mol of sialic acid/mol of fibrinogen) in patients

with liver disorders (9-11). These hypersialylated fibrinogens uniquely display delayed fibrin polymerization which is corrected by removal of excess sialic acid (9-11). Our results provide a mechanistic basis for this abnormality. That is, a t physiological Ca2+concentration, thelow affinity interaction of sialic acid with calcium is inadequate to neutralize additional charge repulsion between hypersialylated fibrin molecules resulting in delayed fibrin polymerization. Acknowledgments-We thank Dr. Eduardo Marban for the use of the Ca2+electrode and Dr. Mack Mitchell for use of the spectrophotometer.

REFERENCES 1. Doolittle, R. F. (1984) Annu. Reu. Biochem. 53, 195-229 2. Weisel, J.W., Stauffacher, C. V., Bullett, E., and Cohen, C. (1985) Science 2 3 0 , 1388-1391 3. Townsend, R. R., Hilliker, E., Li, Y.-T., Laine, R. A., Bell, W. R., and Lee, Y.C. (1982) J. Biol. Chem. 2 5 7 , 9704-9710 4. Ratnoff, 0. D., and Potts, A. M. (1954) J. Clin. Znuest. 3 3 , 206210 5. Boyer, M. H., Shainoff, J. R., and Ratnoff, 0.D. (1972) Blood 39,382-387 6. Nieuwenhuizen, W., and Haverkate, F. (1983) Ann. N. Y. Acod. Sci. 408,92-96 7. Dang, C. V., Ebert, R. F., and Bell, W. R. (1985) J. Biol. Chem. 260,9713-9719 8. Viradi, A., and Scheraga, H.A. (1986) Biochemistry 25,519-528 9. Martinez, J., Keane, D. M., Gilman, P. B., and Palascak, J. E. (1983) Ann. N.Y. Acad. Sci. 408, 388-396 10. Martinez, J., MacDonald, K. A., and Palascak, J. E. (1983) Blood 6 1 , 1196-1202 11. Gralnick, H.R., Givelber, H., and Abram, E. (1978) N. Engl. J . Med. 299,221-226 12. Langer, B. G., Weisel, J. W., Dinauer, P. A., Nagaswami, C., and Bell, W. R. (1988) J. Biol. Chem. 2 6 3 , 15056-15063 13. Warren, L. (1959) J . Biol. Chem. 2 3 4 , 1971-1975 14. Doolittle, R. F. (1975) in Plasma Protein (Putnam, F. W., ed) Vol. 2, pp. 109-161, Academic Press, New York 15. Finlayson, J. S., and Mosesson, M. W. (1963) Biochemistry 2 , 42-46 16. Maruyama, K., Mikawa, T., and Ebashi, S. (1984) J. Biochem. (Tokyo)95,511-519 17. Weisel, J.W. (1987) J. Ultrastruct. Mol. S t r u t . Res. 9 6 , 176188 18. Carr, M. E., and Gabriel, D. A. (1980) Macromolecules 13,14731471 19. Curtis, C. G., Janus, T. J., Credo, R. B., and Lorand, L. (1983) Ann. N.Y. Acad. Sci. 408, 567-576 20. Jaques, L. W., Brown, E. B., Barrett, J. M., Brey, W. S., Jr., and Weltner, W., Jr. (1977) J . Biol. Chem. 252, 4533-4538 21. Behr, J. P., and Lehn, J. M. (1972) FEBS Lett. 22,178-180 22. Czamiecki, M. F., and Thornton, E. R. (1977) Biochem. Biophys. Res. Commun. 74,553-558 23. Townsend, R. R., Heller, D. N., Fenselau, C. C., and Lee, Y. C. (1984) Biochemistry 23,6389-6392 24. Marguerie, G., Chagniel, G., and Suscillon, M. (1977) Biochem. Biophys. Acta 490,94-103 25. VanRuijven-Vermeer, I. A. M., Neuwenhuizen, W., and Nooijen, W. J. (1978) FEBS Lett. 93, 177-180 26. Mihalyi, E. (1988) Biochemistry 27,967-976 27. Mihalyi, E. (1988) Biochemistry 27, 976-982