ARCHIVES
OF BIOCHEMISTRY
Vol. 299, No. 1, November
AND
BIOPHYSICS
15, pp. lOO-104,1992
Reactions of Thrombin-Serpin with Thrombospondin’
Complexes
Andrew C. Chang2 and Thomas C. Detwiler3 Department
of Biochemistry,
State University
of New York Health Science Center at Brooklyn, Brooklyn, New York 11203
Received May 18, 1992, and in revised form July 21, 1992
Activated platelets release proteins that form stable complexes with thrombin (J. J. Miller, P. C. Browne, and T. C. Detwiler, Biochem. Biophys. Res. Commun. 151,915,1988). A working model for the reaction (P. C. Browne, J. J. Miller, and T. C. Detwiler, Arch. Biochem. Biophys. 265, 534-538, 1988) includes a dissociable complex of thrombin with released platelet protease nexin, leading to formation of a nondissociable thrombin-nexin complex that then becomes disulfide linked to thrombospondin. This disulfide-linked complex is converted back to the thrombinnexin complex by reduction of disulfide bonds. Results that allow elaboration on this model are presented. After longer periods of incubation or after incubation with higher concentrations of thrombin, the amount of thrombin complexed with thrombospondin exceeded the amount of thrombinnexin complex recovered after reduction of disulfide bonds. When the reaction mixture included inhibitors of formation of the thrombin-nexin complex, a slow formation of the thrombin-thrombospondin complex was observed. It was concluded that there is a nexin-independent as well as the faster nexin-dependent disulfide linkage of thrombin to thrombospondin. Addition of thrombin-antithrombin III complexes to the supernatant solution of activated platelets also led to complexes with thrombospondin, demonstrating that serpins other than platelet protease nexin facilitate incorporation of thrombin into complexes with thrombospondin. By heparin allinity chromatography, it was shown that thrombin-nexin complexes dissociably associate with thrombospondin prior to formation of disulfide-linked complexes. These observations are incorporated into a more detailed model of the reaction. o lssz Academic PWS, I~C.
1 This work was supported by Grant HL37250 from the National Institutes of Health, United States Department of Health and Human Services. ’ Current address: Bldg. 4, Room 413, NIAID, Bethesda, MD 20892. 3 To whom correspondence should be addressed.
Platelets are activated by several agonists, of which thrombin is the most potent (1). Activated platelets release a variety of proteins (2,3), several of which interact with thrombin as substrates, inhibitors, or other binding proteins (3-6). This paper concerns two thrombin-binding proteins released from activated platelets. One is platelet protease nexin (PNiJ4 (6), a serpin that is released from activated platelets (7). It forms SDS-stable complexes with, and inhibits, thrombin and other proteases (8, 9). The other is thrombospondin (Tsp), a 450-kDa cell-adhesion, or matrix, protein that undergoes thiol-disulfide exchange with thrombin-PN, complexes (5, 7, 8), leading to a large trimolecular complex. This obviously is a complicated reaction. From several studies (5, 7-lo), we have proposed the working model (8) T+PN,+ SC?
T.*.PN,* EA T -PN,
s/-\s
T-PN,
Tsp 7-, Tsp
I
111
A
S-S
SH
SH Thrombin (T) is shown with a disulfide bond (ll), and Tsp is shown with a free thiol (12). An initial reversible reaction of thrombin with PN, has been inferred (8). It is followed by formation of a nondissociable thrombinPN, complex (approximately 77 kDa) that presumably 4 Abbreviations used: APMSF and APMS-thrombin, p-amidinophenylmethylsulfonyl fluoride and its thrombin derivative; ATIII, Antithrombin III; Hepes, 4-(2.hydroxyethyl)-l-piperazineethanesulfonic acid, PAGE, polyacrylamide gel electrophoresis; PN,, platelet-derived protease nexin; PDI, protein disulfide isomerase; SDS, sodium dodecyl sulfate; Tsp, thrombospondin.
100 All
Copyright 0 1992 rights of reproduction
0003.9861/92 $5.00 by Academic Press, Inc. in any form reserved.
THROMBIN-SERPIN-THROMBOSPONDIN
involves some type of acyl bond between thrombin and PN, (13). Thrombin in the 77-kDa complex shows no catalytic activity (8, 9). This complex then reacts with Tsp by thiol-disulfide exchange involving a thiol of Tsp and a disulfide bond of thrombin (5, 10, 12). This leads to an intermolecular disulfide bond between Tsp and thrombin (complex >450 kDa), and it results in a free thiol on thrombin (5). When the progress of the reaction is followed by SDS-PAGE analysis of samples with labeled thrombin, quick formation of a 77-kDa complex is followed by its slower disappearance and concomitant formation of the >450-kDa complex (8). After reduction of disulfide bonds, all of the complexed thrombin is recovered as the 77-kDa thrombin-PN, complex (8). This model has proved adequate for design of experiments and interpretation of experimental data, but it clearly is not a satisfactory description of the reaction. In this paper we report additional observations that allow us to elaborate on the model and more fully explain the reaction. The new model includes serpins other than PN,, a serpin-independent reaction in addition to the PN,dependent one, a dissociable complex that precedes the disulfide-linked complex with Tsp, and an enzyme-catalyzed disulfide bond formation. MATERIALS
AND
METHODS
Purified human oc-thrombin was a gift from John W. Fenton II, Wadsworth Center for Laboratories and Research, School of Public Health Sciences, New York Department of Health (Albany, NY). The labeling of thrombin with radioiodine and its inhibition with APMSF have been described (7). Hirudin was from Sigma Chemical Co. (St. Louis, MO) and A23187 was from Calbiochem (San Diego, CA). Human AT III was from Sigma. Human platelets were prepared from blood collected from healthy volunteers into 0.14 vol of anticoagulant (85 mM sodium citrate, 65 mM citric acid, 2% glucose). The red cells and white cells were removed by centrifugation at 300g for 20 min at 4’C. The platelets were then collected by centrifugation at 1OOOgfor 20 min; they were washed twice with 150 mM NaCl, 10 mM Hepes (pH 7.4), 1 mM EDTA. They were finally suspended in the same buffer at l-3 X 109/ml. To prepare the supernatant solution of activated platelets, the suspension was warmed to 37°C and made 2 mM with CaCl* before addition of 0.01 ~01500 pM A23187 in dimethyl sulfoxide. After 2 min with gentle shaking, 5 mM EDTA was added and the suspension was cooled on ice. The suspension was centrifuged at 12,000g for 15 min at 4°C to obtain the supernatant with secreted proteins. The formation of protein complexes was analyzed by SDS-PAGE/ autoradiography by methods that have been described (7). The supernatant solution of activated platelets was incubated with iz51-labeled thrombin at 37°C. Aliquots were analyzed by SDS-PAGE/autoradiography with either nonreducing (for Tsp-thrombin-PN, complexes) or reducing gels (for thrombin-PN, complexes). The amount of each complex was quantified by measurement of radioactivity in slices cut from the gels.
RESULTS
It was shown previously Stoichiometry of the reaction. that labeled thrombin in a >450-kDa complex (thrombinPN,-Tsp) could be recovered nearly entirely as a 77-kDa
101
COMPLEXES
complex (thrombin-PN,) after reduction of disulfide bonds (8). This led to the conclusion of a stoichiometric reaction in which one thrombin and one PN, formed a complex that was then incorporated into a larger complex with Tsp by intermolecular thiol-disulfide exchange. When we extended the period of incubation beyond the 60 min used previously or when we used thrombin concentrations above the 10 nM used previously, we found that the amount of thrombin-PN,-Tsp complex substantially exceeded the levels of thrombin-PN, recovered after reduction of disulfide bonds (Fig. 1). These results argue against the stoichiometry of our model for the reaction. A possible explanation is turnover of the thrombin-PN, complex, such that one molecule of PN, sequentially mediates the formation of multiple thrombinTsp complexes. We were unable, however, to detect any turnover or dissociation of thrombin-PN, complexes. For example, in an experiment in which 5 nM labeled thrombin was incubated with the supernatant solution for 60 min and then incubated an additional 60 min with no addition or with 500 nM unlabeled thrombin, the amount of labeled thrombin-PN, complex after reduction was constant and the same under both conditions (data not shown). An alternative explanation for the lack of stoichiometry is that PN, facilitates and accelerates the incorporation of thrombin into complexes with Tsp but is not essential for formation of those complexes. This was suggested by experiments with inhibitors that block formation of the thrombin-PN, complex (Fig. 2). When reaction with PN, was blocked, there was still a slow formation of the large complex of thrombin with Tsp. In fact, the rate of this formation was approximately equal to the rate of formation above the thrombin-PN, complex in Fig. 1A. While APMS-thrombin can form the nondissociable complex with PN,, , hirudin prevents this complex, so the data of Fig. 2 indicate that neither the SDS-stable nor
Time (min)
Thrombin (nM)
FIG. 1. The amount of >450-kDa complex can exceed the amount of 77.kDa complex recovered after reduction of disulfide bonds. Labeled thrombin was incubated with the supernatant solution of activated platelets. The >450-kDa complex (Tsp-thrombin-PN,) was measured on nonreducing gels, and the 77.kDa complex (thrombin-PN,), which dissociates from the >450-kDa complex on reduction of disulfide bonds (8), was measured on reducing gels. (Left) The periods of incubation were varied with 10 nM thrombin. (Right) Incubation was for 60 min with the concentration of thrombin varied.
102
CHANG
0
0
20
40
AND
60
DETWILER
bin-PN,-Tsp complexes. Tsp and PN, were eluted with 500 and 700 mM NaCl (Fig. 4A). When PN, that first had been separated from Tsp was complexed with thrombin, the complex also eluted at 700 mM NaCl (Fig. 4B), showing that the heparin-binding site of PN, was not masked by thrombin. When thrombin was incubated with the supernatant solution of activated platelets (a solution that contains PN, and Tsp) for 1 min, two-thirds of the thrombin-PN, complex was eluted at 500 mM NaCl (i.e., with Tsp; Fig. 4C). Disulfide linkage of thrombin-PN, with Tsp was prevented by using a medium with Ca2+, a very short incubation, and a temperature of 4°C for the
Time (min) FIG. 2. Inhibitors of the formation of the thrombin-PN, complex did not completely block formation of the >450-kDa complex. Labeled thrombin (8 nM) was incubated with the supernatant solution of activated platelets at 37°C. Aliquots were removed and analyzed by nonreducing SDS-PAGE/autoradiography as for Fig. 1. Incubations were with control thrombin (0), with APMS-thrombin (a), or with thrombin with a threefold excess of hirudin (A).
NON-REDUCING
Th-AT Ill
the dissociable complex of thrombin with PN, is an essential intermediate for formation of a thrombin-Tsp complex. Complex formation is facilitated by another serpin. Do other serpins facilitate incorporation of thrombin into complexes with Tsp? Figure 3 shows that formation of large labeled complexes was similar after the addition of labeled thrombin or labeled thrombin-antithrombin III (ATIII) complexes to the supernatant solution of activated platelets. Note specifically in each case the appearance of similar multimers of the large labeled complexes and the inhibition of formation of the large complexes by Ca2+. Since a serpin that could be obtained pure underwent the same reaction, we were able to test the formation of the large complex with three pure proteins: Tsp, thrombin, and ATIII. The amount of >450kDa complex resulting from incubation of 400 nM Tsp with 5 nM thrombin-AT111 was less than 5% the amount when the same thrombin-AT111 was incubated with the supernatant solution of activated platelets, which contained a comparable amount of Tsp. Similarly, the formation of multimers of Tsp (10) occurred to a very limited extent in the absence of the supernatant solution (data not shown). We conclude that another factor in the supernatant solution is required. It is likely that the factor is the enzyme protein disulfide isomerase (PDI), which was observed in the supernatant solution of activated platelets (14). Evidence for a nondissociable complex of Tsp with thrombin-PN,. Tsp and PN, bind to heparin, and they can be resolved with a heparin affinity column (7). We used such a column to analyze thrombin-PN, and throm-
REDUCING
15 60
Time (mid
0
EDTA
+++ttt-+
Ca2+
SUPERNATANT + THROMBIN
0 I_-
15 60
60
60
t,
-
SUPERNATANT + THROMBIN-AT III
; z’ z iz
FIG. 3. Thrombin-Tsp complexes formed with ATIII. Labeled thrombin (100 nM) was incubated with AT111 (20 U/ml) and heparin (20 U/ml) for 10 min at 22°C in 150 mM NaCI, 10 mM Hepes (pH 7.4), 1 mM EDTA. An aliquot was transferred to a supernatant solution containing either 5 mM EDTA or 3 mM CaClz to give 5 nM thrombin, at least half of which was in a thrombin-AT111 complex. Thrombin (5 nM) without AT111 was added to control supernatant solutions. The solutions were incubated at 37°C. Aliquots were analyzed by SDS-PAGE/autoradiography. Photographs of the autoradiograms are shown.
THROMBINPSERPIN-THROMBOSPONDIN
400
600
800
1000
[NaCI] (mM) FIG. 4. Thrombin-PN, binds dissociably to Tsp. Various protein mixtures were chromatographed on a l-ml heparin-agarose column (Pierce Chemical Co., Rockford, IL) at 4°C. Elution was with step gradients of NaCl in 10 mM Hepes (pH 7.6), 2 mM CaCl*. Eluted Tsp was monitored by Coomassie blue-stained SDS-PAGE and quantified by densitometry. Eluted PN, was measured by its ability to form a labeled complex was complex with ‘2SI-labeled thrombin; the thrombin-PN, quantified by SDS-PAGE/autoradiography and counting of gel slices corresponding to the 77-kDa-labeled band. The elution of labeled complexes of thrombin-PN, was similarly quantified by SDS-PAGE/autoradiography and counting of gel slices. (A) The supernatant solution was resolved into Tsp (a) and PN, (0). (B) The 700 mM fraction from (A) was dialyzed against 10 mM Hepes (pH 7.4) and 2 mM CaCl* for 3 h at 4°C before reaction with 5 nM labeled thrombin for 5 min at 37°C. It was then rechromatographed on the same column. The thrombinPN, complex (0) was measured by SDS-PAGE/autoradiography followed by counting of the appropriate gel slice. (C) The supernatant solution of activated platelets was incubated for 1 min with 5 nM lz51labeled thrombin prior to chromatography on heparin-agarose. The eluted thrombin-PN, complex (A) was measured by SDS-PAGE/autoradiography and counts of the appropriate gel slice.
column, each a condition unfavhrable for disulfide linkage. We conclude that a noncovalent association of thrombinPN, with Tsp occurs much more quickly than does the formation of the SDS-stable, disulfide-linked complex. DISCUSSION
Our model for the reaction of thrombin with Tsp has been modified (reaction [2]) in three ways to account for the observations described here: T+S= Li
T...S” s-s
T--S-TSP...T--S~T~P /\ I /\ s-s SH S-S
7
T-S I /\ S-S SH
Tsp
/‘\ s-s
kTsp I /‘\ S-S SH
El
COMPLEXES
103
First, the demonstration that AT111 can substitute for PN, shows that the reaction is more general than shown in Eq. [ 11. While it is not yet clear how many serpins can serve this role, we have replaced PN, with S, for serpin, in Eq. [ 21. It should be noted that this is not any complex with thrombin, because the thrombin-hirudin complex was not incorporated. Second, it is necessary to account for the observations that (i) the amount of thrombin-Tsp complex can exceed the amount of thrombin-PN, complex and (ii) the thrombin-Tsp complex forms when formation of the thrombin-PN, complex is blocked. That is, while PN, accelerates the incorporation of thrombin into a complex with Tsp so much that the thrombin-PN, complex is a normal intermediate in this process, it is not an essential intermediate; the reaction will proceed, albeit slowly, in the absence of a serpin. Equation [2] shows this serpin-independent pathway in addition to the serpindependent pathway. It is important to note that the slower pathway would occur to an appreciable extent only if there were no serpin or if the concentration of thrombin greatly exceeded the concentration of serpin (or if an inhibitor blocked formation of the thrombin-PN, complex, as in Fig. 2). Third, the observation of dissociable binding of thrombin-PN, complexes to Tsp prior to formation of the covalent complex is an important clue to why the reaction is so much faster in the presence of a serpin. It is likely that Tsp has a binding site for serpin-protease complexes, and that it is this initial binding that leads to thiol-disulfide exchange. Since the heparin affinity column separates Tsp from PN, much better than from thrombin-PN, complexes, it is likely that the binding site recognizes the complexes rather than the serpin itself, and the affinity of thrombin-PN, for Tsp must be greater than that for heparin. A dissociable complex preceding the covalent complex has been added in reaction [a]. Finally, from other studies we can explain why the reaction does not occur with purified proteins. It has been shown (14) that activated platelets release a protein disulfide isomerase that catalyzes the thiol-disulfide exchange. This enzyme has been added in the last step of the serpindependent reaction in reaction [a]. It is not clear whether PDI catalyzes the serpin-independent formation of thrombin-Tsp complexes, but it is reasonable that it does. REFERENCES 1. Berndt, M. C., and Phillips, D. R. (1981) in Platelets in Biology and Pathology-2 (Gordon, J. L., Ed.), pp. 43-76, Elsevier/North-Holland, Amsterdam. 2. Niewiarowski, S., and Paul, D. (1981) in Platelets in Biology and Pathology2 (Gordon, J. L., Ed.), pp. 91-106, Elsevier/North-Holland, Amsterdam. 3. Walsh, P. N., and Schmaier, A. H. (1987) in Hemostasis and Thrombosis (Colman, R. W., et al., Eds.), pp. 689-709, Lippincott, New York. 4. Danishefsky, K. ,J., and Detwiler, T. C. (1984) Biochim. Biophys. Acta 801, 48-57.
104 5. Danishefsky, Biochemistry
CHANG K. J., Alexander, 23,4984-4990.
6. Gronke, R. S., Bergman, Chem. 262,3030-3036.
R. J., and Detwiler,
AND
T. C. (1984)
B. L., and Baker, J. B. (1987) J. Biol.
7. Miller, J. J., Browne, P. B., and Detwiler, Biophys. Res. Commun. 151,9-15.
T. C. (1988) Biochem.
8. Browne, P. C., Miller, J. J., and Detwiler, T. C. (1988) Arch. Biochem. Biophys. 265,534-538. 9. Miller, J. J., and Detwiler, 276,364-368.
T. C. (1990) Arch. Biochem. Biophys.
DETWILER 10. Turk, J. L., and Detwiler, 245,446-454.
T. C. (1986) Arch. Biochem. Biophys.
11. Elion, J., Downing, M. R., Butkowski, R. J., and Mann, K. G. (1977) in Chemistry and Biology of Thrombin (Lundblad et al., Eds.), pp. 97-111, Ann Arbor Science, Ann Arbor, MI. 12. Speziale, M. V., and Detwiler, 17859-17867.
T. C. (1990) J. Biol. Chem. 265,
13. Carrell, R., and Travis, J. (1985) Trends Biochem. Ski. 10, 20-24. 14. Chen, K., Lin, Y., and Detwiler,
T. C. (1992) Blood 79, 2226-2228.
OF BIOCHEMISTRY
Vol. 299, No. 1, November
AND
BIOPHYSICS
15, pp. lOO-104,1992
Reactions of Thrombin-Serpin with Thrombospondin’
Complexes
Andrew C. Chang2 and Thomas C. Detwiler3 Department
of Biochemistry,
State University
of New York Health Science Center at Brooklyn, Brooklyn, New York 11203
Received May 18, 1992, and in revised form July 21, 1992
Activated platelets release proteins that form stable complexes with thrombin (J. J. Miller, P. C. Browne, and T. C. Detwiler, Biochem. Biophys. Res. Commun. 151,915,1988). A working model for the reaction (P. C. Browne, J. J. Miller, and T. C. Detwiler, Arch. Biochem. Biophys. 265, 534-538, 1988) includes a dissociable complex of thrombin with released platelet protease nexin, leading to formation of a nondissociable thrombin-nexin complex that then becomes disulfide linked to thrombospondin. This disulfide-linked complex is converted back to the thrombinnexin complex by reduction of disulfide bonds. Results that allow elaboration on this model are presented. After longer periods of incubation or after incubation with higher concentrations of thrombin, the amount of thrombin complexed with thrombospondin exceeded the amount of thrombinnexin complex recovered after reduction of disulfide bonds. When the reaction mixture included inhibitors of formation of the thrombin-nexin complex, a slow formation of the thrombin-thrombospondin complex was observed. It was concluded that there is a nexin-independent as well as the faster nexin-dependent disulfide linkage of thrombin to thrombospondin. Addition of thrombin-antithrombin III complexes to the supernatant solution of activated platelets also led to complexes with thrombospondin, demonstrating that serpins other than platelet protease nexin facilitate incorporation of thrombin into complexes with thrombospondin. By heparin allinity chromatography, it was shown that thrombin-nexin complexes dissociably associate with thrombospondin prior to formation of disulfide-linked complexes. These observations are incorporated into a more detailed model of the reaction. o lssz Academic PWS, I~C.
1 This work was supported by Grant HL37250 from the National Institutes of Health, United States Department of Health and Human Services. ’ Current address: Bldg. 4, Room 413, NIAID, Bethesda, MD 20892. 3 To whom correspondence should be addressed.
Platelets are activated by several agonists, of which thrombin is the most potent (1). Activated platelets release a variety of proteins (2,3), several of which interact with thrombin as substrates, inhibitors, or other binding proteins (3-6). This paper concerns two thrombin-binding proteins released from activated platelets. One is platelet protease nexin (PNiJ4 (6), a serpin that is released from activated platelets (7). It forms SDS-stable complexes with, and inhibits, thrombin and other proteases (8, 9). The other is thrombospondin (Tsp), a 450-kDa cell-adhesion, or matrix, protein that undergoes thiol-disulfide exchange with thrombin-PN, complexes (5, 7, 8), leading to a large trimolecular complex. This obviously is a complicated reaction. From several studies (5, 7-lo), we have proposed the working model (8) T+PN,+ SC?
T.*.PN,* EA T -PN,
s/-\s
T-PN,
Tsp 7-, Tsp
I
111
A
S-S
SH
SH Thrombin (T) is shown with a disulfide bond (ll), and Tsp is shown with a free thiol (12). An initial reversible reaction of thrombin with PN, has been inferred (8). It is followed by formation of a nondissociable thrombinPN, complex (approximately 77 kDa) that presumably 4 Abbreviations used: APMSF and APMS-thrombin, p-amidinophenylmethylsulfonyl fluoride and its thrombin derivative; ATIII, Antithrombin III; Hepes, 4-(2.hydroxyethyl)-l-piperazineethanesulfonic acid, PAGE, polyacrylamide gel electrophoresis; PN,, platelet-derived protease nexin; PDI, protein disulfide isomerase; SDS, sodium dodecyl sulfate; Tsp, thrombospondin.
100 All
Copyright 0 1992 rights of reproduction
0003.9861/92 $5.00 by Academic Press, Inc. in any form reserved.
THROMBIN-SERPIN-THROMBOSPONDIN
involves some type of acyl bond between thrombin and PN, (13). Thrombin in the 77-kDa complex shows no catalytic activity (8, 9). This complex then reacts with Tsp by thiol-disulfide exchange involving a thiol of Tsp and a disulfide bond of thrombin (5, 10, 12). This leads to an intermolecular disulfide bond between Tsp and thrombin (complex >450 kDa), and it results in a free thiol on thrombin (5). When the progress of the reaction is followed by SDS-PAGE analysis of samples with labeled thrombin, quick formation of a 77-kDa complex is followed by its slower disappearance and concomitant formation of the >450-kDa complex (8). After reduction of disulfide bonds, all of the complexed thrombin is recovered as the 77-kDa thrombin-PN, complex (8). This model has proved adequate for design of experiments and interpretation of experimental data, but it clearly is not a satisfactory description of the reaction. In this paper we report additional observations that allow us to elaborate on the model and more fully explain the reaction. The new model includes serpins other than PN,, a serpin-independent reaction in addition to the PN,dependent one, a dissociable complex that precedes the disulfide-linked complex with Tsp, and an enzyme-catalyzed disulfide bond formation. MATERIALS
AND
METHODS
Purified human oc-thrombin was a gift from John W. Fenton II, Wadsworth Center for Laboratories and Research, School of Public Health Sciences, New York Department of Health (Albany, NY). The labeling of thrombin with radioiodine and its inhibition with APMSF have been described (7). Hirudin was from Sigma Chemical Co. (St. Louis, MO) and A23187 was from Calbiochem (San Diego, CA). Human AT III was from Sigma. Human platelets were prepared from blood collected from healthy volunteers into 0.14 vol of anticoagulant (85 mM sodium citrate, 65 mM citric acid, 2% glucose). The red cells and white cells were removed by centrifugation at 300g for 20 min at 4’C. The platelets were then collected by centrifugation at 1OOOgfor 20 min; they were washed twice with 150 mM NaCl, 10 mM Hepes (pH 7.4), 1 mM EDTA. They were finally suspended in the same buffer at l-3 X 109/ml. To prepare the supernatant solution of activated platelets, the suspension was warmed to 37°C and made 2 mM with CaCl* before addition of 0.01 ~01500 pM A23187 in dimethyl sulfoxide. After 2 min with gentle shaking, 5 mM EDTA was added and the suspension was cooled on ice. The suspension was centrifuged at 12,000g for 15 min at 4°C to obtain the supernatant with secreted proteins. The formation of protein complexes was analyzed by SDS-PAGE/ autoradiography by methods that have been described (7). The supernatant solution of activated platelets was incubated with iz51-labeled thrombin at 37°C. Aliquots were analyzed by SDS-PAGE/autoradiography with either nonreducing (for Tsp-thrombin-PN, complexes) or reducing gels (for thrombin-PN, complexes). The amount of each complex was quantified by measurement of radioactivity in slices cut from the gels.
RESULTS
It was shown previously Stoichiometry of the reaction. that labeled thrombin in a >450-kDa complex (thrombinPN,-Tsp) could be recovered nearly entirely as a 77-kDa
101
COMPLEXES
complex (thrombin-PN,) after reduction of disulfide bonds (8). This led to the conclusion of a stoichiometric reaction in which one thrombin and one PN, formed a complex that was then incorporated into a larger complex with Tsp by intermolecular thiol-disulfide exchange. When we extended the period of incubation beyond the 60 min used previously or when we used thrombin concentrations above the 10 nM used previously, we found that the amount of thrombin-PN,-Tsp complex substantially exceeded the levels of thrombin-PN, recovered after reduction of disulfide bonds (Fig. 1). These results argue against the stoichiometry of our model for the reaction. A possible explanation is turnover of the thrombin-PN, complex, such that one molecule of PN, sequentially mediates the formation of multiple thrombinTsp complexes. We were unable, however, to detect any turnover or dissociation of thrombin-PN, complexes. For example, in an experiment in which 5 nM labeled thrombin was incubated with the supernatant solution for 60 min and then incubated an additional 60 min with no addition or with 500 nM unlabeled thrombin, the amount of labeled thrombin-PN, complex after reduction was constant and the same under both conditions (data not shown). An alternative explanation for the lack of stoichiometry is that PN, facilitates and accelerates the incorporation of thrombin into complexes with Tsp but is not essential for formation of those complexes. This was suggested by experiments with inhibitors that block formation of the thrombin-PN, complex (Fig. 2). When reaction with PN, was blocked, there was still a slow formation of the large complex of thrombin with Tsp. In fact, the rate of this formation was approximately equal to the rate of formation above the thrombin-PN, complex in Fig. 1A. While APMS-thrombin can form the nondissociable complex with PN,, , hirudin prevents this complex, so the data of Fig. 2 indicate that neither the SDS-stable nor
Time (min)
Thrombin (nM)
FIG. 1. The amount of >450-kDa complex can exceed the amount of 77.kDa complex recovered after reduction of disulfide bonds. Labeled thrombin was incubated with the supernatant solution of activated platelets. The >450-kDa complex (Tsp-thrombin-PN,) was measured on nonreducing gels, and the 77.kDa complex (thrombin-PN,), which dissociates from the >450-kDa complex on reduction of disulfide bonds (8), was measured on reducing gels. (Left) The periods of incubation were varied with 10 nM thrombin. (Right) Incubation was for 60 min with the concentration of thrombin varied.
102
CHANG
0
0
20
40
AND
60
DETWILER
bin-PN,-Tsp complexes. Tsp and PN, were eluted with 500 and 700 mM NaCl (Fig. 4A). When PN, that first had been separated from Tsp was complexed with thrombin, the complex also eluted at 700 mM NaCl (Fig. 4B), showing that the heparin-binding site of PN, was not masked by thrombin. When thrombin was incubated with the supernatant solution of activated platelets (a solution that contains PN, and Tsp) for 1 min, two-thirds of the thrombin-PN, complex was eluted at 500 mM NaCl (i.e., with Tsp; Fig. 4C). Disulfide linkage of thrombin-PN, with Tsp was prevented by using a medium with Ca2+, a very short incubation, and a temperature of 4°C for the
Time (min) FIG. 2. Inhibitors of the formation of the thrombin-PN, complex did not completely block formation of the >450-kDa complex. Labeled thrombin (8 nM) was incubated with the supernatant solution of activated platelets at 37°C. Aliquots were removed and analyzed by nonreducing SDS-PAGE/autoradiography as for Fig. 1. Incubations were with control thrombin (0), with APMS-thrombin (a), or with thrombin with a threefold excess of hirudin (A).
NON-REDUCING
Th-AT Ill
the dissociable complex of thrombin with PN, is an essential intermediate for formation of a thrombin-Tsp complex. Complex formation is facilitated by another serpin. Do other serpins facilitate incorporation of thrombin into complexes with Tsp? Figure 3 shows that formation of large labeled complexes was similar after the addition of labeled thrombin or labeled thrombin-antithrombin III (ATIII) complexes to the supernatant solution of activated platelets. Note specifically in each case the appearance of similar multimers of the large labeled complexes and the inhibition of formation of the large complexes by Ca2+. Since a serpin that could be obtained pure underwent the same reaction, we were able to test the formation of the large complex with three pure proteins: Tsp, thrombin, and ATIII. The amount of >450kDa complex resulting from incubation of 400 nM Tsp with 5 nM thrombin-AT111 was less than 5% the amount when the same thrombin-AT111 was incubated with the supernatant solution of activated platelets, which contained a comparable amount of Tsp. Similarly, the formation of multimers of Tsp (10) occurred to a very limited extent in the absence of the supernatant solution (data not shown). We conclude that another factor in the supernatant solution is required. It is likely that the factor is the enzyme protein disulfide isomerase (PDI), which was observed in the supernatant solution of activated platelets (14). Evidence for a nondissociable complex of Tsp with thrombin-PN,. Tsp and PN, bind to heparin, and they can be resolved with a heparin affinity column (7). We used such a column to analyze thrombin-PN, and throm-
REDUCING
15 60
Time (mid
0
EDTA
+++ttt-+
Ca2+
SUPERNATANT + THROMBIN
0 I_-
15 60
60
60
t,
-
SUPERNATANT + THROMBIN-AT III
; z’ z iz
FIG. 3. Thrombin-Tsp complexes formed with ATIII. Labeled thrombin (100 nM) was incubated with AT111 (20 U/ml) and heparin (20 U/ml) for 10 min at 22°C in 150 mM NaCI, 10 mM Hepes (pH 7.4), 1 mM EDTA. An aliquot was transferred to a supernatant solution containing either 5 mM EDTA or 3 mM CaClz to give 5 nM thrombin, at least half of which was in a thrombin-AT111 complex. Thrombin (5 nM) without AT111 was added to control supernatant solutions. The solutions were incubated at 37°C. Aliquots were analyzed by SDS-PAGE/autoradiography. Photographs of the autoradiograms are shown.
THROMBINPSERPIN-THROMBOSPONDIN
400
600
800
1000
[NaCI] (mM) FIG. 4. Thrombin-PN, binds dissociably to Tsp. Various protein mixtures were chromatographed on a l-ml heparin-agarose column (Pierce Chemical Co., Rockford, IL) at 4°C. Elution was with step gradients of NaCl in 10 mM Hepes (pH 7.6), 2 mM CaCl*. Eluted Tsp was monitored by Coomassie blue-stained SDS-PAGE and quantified by densitometry. Eluted PN, was measured by its ability to form a labeled complex was complex with ‘2SI-labeled thrombin; the thrombin-PN, quantified by SDS-PAGE/autoradiography and counting of gel slices corresponding to the 77-kDa-labeled band. The elution of labeled complexes of thrombin-PN, was similarly quantified by SDS-PAGE/autoradiography and counting of gel slices. (A) The supernatant solution was resolved into Tsp (a) and PN, (0). (B) The 700 mM fraction from (A) was dialyzed against 10 mM Hepes (pH 7.4) and 2 mM CaCl* for 3 h at 4°C before reaction with 5 nM labeled thrombin for 5 min at 37°C. It was then rechromatographed on the same column. The thrombinPN, complex (0) was measured by SDS-PAGE/autoradiography followed by counting of the appropriate gel slice. (C) The supernatant solution of activated platelets was incubated for 1 min with 5 nM lz51labeled thrombin prior to chromatography on heparin-agarose. The eluted thrombin-PN, complex (A) was measured by SDS-PAGE/autoradiography and counts of the appropriate gel slice.
column, each a condition unfavhrable for disulfide linkage. We conclude that a noncovalent association of thrombinPN, with Tsp occurs much more quickly than does the formation of the SDS-stable, disulfide-linked complex. DISCUSSION
Our model for the reaction of thrombin with Tsp has been modified (reaction [2]) in three ways to account for the observations described here: T+S= Li
T...S” s-s
T--S-TSP...T--S~T~P /\ I /\ s-s SH S-S
7
T-S I /\ S-S SH
Tsp
/‘\ s-s
kTsp I /‘\ S-S SH
El
COMPLEXES
103
First, the demonstration that AT111 can substitute for PN, shows that the reaction is more general than shown in Eq. [ 11. While it is not yet clear how many serpins can serve this role, we have replaced PN, with S, for serpin, in Eq. [ 21. It should be noted that this is not any complex with thrombin, because the thrombin-hirudin complex was not incorporated. Second, it is necessary to account for the observations that (i) the amount of thrombin-Tsp complex can exceed the amount of thrombin-PN, complex and (ii) the thrombin-Tsp complex forms when formation of the thrombin-PN, complex is blocked. That is, while PN, accelerates the incorporation of thrombin into a complex with Tsp so much that the thrombin-PN, complex is a normal intermediate in this process, it is not an essential intermediate; the reaction will proceed, albeit slowly, in the absence of a serpin. Equation [2] shows this serpin-independent pathway in addition to the serpindependent pathway. It is important to note that the slower pathway would occur to an appreciable extent only if there were no serpin or if the concentration of thrombin greatly exceeded the concentration of serpin (or if an inhibitor blocked formation of the thrombin-PN, complex, as in Fig. 2). Third, the observation of dissociable binding of thrombin-PN, complexes to Tsp prior to formation of the covalent complex is an important clue to why the reaction is so much faster in the presence of a serpin. It is likely that Tsp has a binding site for serpin-protease complexes, and that it is this initial binding that leads to thiol-disulfide exchange. Since the heparin affinity column separates Tsp from PN, much better than from thrombin-PN, complexes, it is likely that the binding site recognizes the complexes rather than the serpin itself, and the affinity of thrombin-PN, for Tsp must be greater than that for heparin. A dissociable complex preceding the covalent complex has been added in reaction [a]. Finally, from other studies we can explain why the reaction does not occur with purified proteins. It has been shown (14) that activated platelets release a protein disulfide isomerase that catalyzes the thiol-disulfide exchange. This enzyme has been added in the last step of the serpindependent reaction in reaction [a]. It is not clear whether PDI catalyzes the serpin-independent formation of thrombin-Tsp complexes, but it is reasonable that it does. REFERENCES 1. Berndt, M. C., and Phillips, D. R. (1981) in Platelets in Biology and Pathology-2 (Gordon, J. L., Ed.), pp. 43-76, Elsevier/North-Holland, Amsterdam. 2. Niewiarowski, S., and Paul, D. (1981) in Platelets in Biology and Pathology2 (Gordon, J. L., Ed.), pp. 91-106, Elsevier/North-Holland, Amsterdam. 3. Walsh, P. N., and Schmaier, A. H. (1987) in Hemostasis and Thrombosis (Colman, R. W., et al., Eds.), pp. 689-709, Lippincott, New York. 4. Danishefsky, K. ,J., and Detwiler, T. C. (1984) Biochim. Biophys. Acta 801, 48-57.
104 5. Danishefsky, Biochemistry
CHANG K. J., Alexander, 23,4984-4990.
6. Gronke, R. S., Bergman, Chem. 262,3030-3036.
R. J., and Detwiler,
AND
T. C. (1984)
B. L., and Baker, J. B. (1987) J. Biol.
7. Miller, J. J., Browne, P. B., and Detwiler, Biophys. Res. Commun. 151,9-15.
T. C. (1988) Biochem.
8. Browne, P. C., Miller, J. J., and Detwiler, T. C. (1988) Arch. Biochem. Biophys. 265,534-538. 9. Miller, J. J., and Detwiler, 276,364-368.
T. C. (1990) Arch. Biochem. Biophys.
DETWILER 10. Turk, J. L., and Detwiler, 245,446-454.
T. C. (1986) Arch. Biochem. Biophys.
11. Elion, J., Downing, M. R., Butkowski, R. J., and Mann, K. G. (1977) in Chemistry and Biology of Thrombin (Lundblad et al., Eds.), pp. 97-111, Ann Arbor Science, Ann Arbor, MI. 12. Speziale, M. V., and Detwiler, 17859-17867.
T. C. (1990) J. Biol. Chem. 265,
13. Carrell, R., and Travis, J. (1985) Trends Biochem. Ski. 10, 20-24. 14. Chen, K., Lin, Y., and Detwiler,
T. C. (1992) Blood 79, 2226-2228.
