Von Willebrand Factor and the Endothelium

DENISA D. WAGNER, Ph.D., ROBERTA BONFANTI, M.D., Center ofHemostasis and Thrombosis Research, Division ofHematology-Oncology, New England Medical Center, Boston, Massachusetts

Endothelial cells are the principal source of plasma and basement membrane von Willebrand factor (vWF). To arrive at its biologically active multimeric form, vWF undergoes a series of intracellular processing steps. The protein is synthesized as a large precursor pro-vWF, which dimerizes in the endoplasmic reticulum through disulfide bonds located in the carboxyl-terminal portion of the subunit. Only dimers are transported to the Golgi apparatus. Expression of truncated pro-vWF subunits, which lack the last 20 kd, abolishes the requirement for dimerization and thereby allows the monomeric protein to be secreted. Another requirement for intracellular transport from the endoplasmic reticulum is N-linked glycosylation. Inhibition of N-linked glycosylation prevents exit of both the wild-type and the truncated vWF from the endoplasmic reticulum. In the acidic environment of the trans-Golgi and post-Golgi compartments, pro-vWF dimers multimerize by a second set of interchain disulfide bonds. The presence of the vWF propolypeptide and acidic pH conditions are necessary for the multimerization process. The largest vWF multimers are stored in endothelial cell-specific organelles called Weibel-Palade bodies. At the site of vascular injury and inflammation, physiologic secretagogues such as thrombin, fibrin, and histamine may cause release of these large, biologically potent vWF multimers from the Weibel-Palade bodies into the surrounding blood and subendothelium.

Vascular endothelium is the primary source of plasma von Willebrand factor (vWF). The only other type of cell that synthesizes vWF is the This work was supported in part by Grant HL 41002 R01 from the National Institutes of Health, Public Health Service. Individual reprints of this article are not available. The entire Bowie Symposium on von Willebrand's Disease will be available for purchase as a bound booklet from the Proceedings Circulation Office in September. Mayo Clin Proc 66:621-627, 1991

megakaryocyte. The endothelium also deposits vWF into the basement membrane of blood vessels. This matrix form of vWF is important for initial attachment of platelets during vascular injury and also for adhesion of the endothelial cells to the vessel wall. 1 vWF is an unusual adhesive molecule in that it is heterogeneous in size because of its polymeric nature. The vWF polymers, composed of identical subunits held together by disulfide bonds, range in size from 0.5 to 20 million d. The functional significance of




this heterogeneity is unknown. Although the largest multimers are the most effective in hemostasis, all sizes of vWF carry factor VIn and protect it against proteolysis. 1 Predominantly small multimers are secreted constitutively (without stimulation) byendothelial cells in culture. In contrast, only the largest multimers are rapidly released on stimulation through what is called the regulated pathway of secretion.v' This pool of vWF is stored in endothelial cell-specific granules known as WeibelPalade bodies," which are elongated organelles enclosed by a unit membrane. The tubular structures found within these organelles are also observed in the a granules of platelets" and most likely represent highly organized vWF polymers. Besides vWF and its propolypeptide, the only other protein identified in WeibelPalade bodies is PADGEM (GMP-140, CD62), a transmembrane receptor for neutrophils and monocytes.v" Preliminary evidence from our laboratory indicates that the vWF propolypeptide is involved in directing vWF molecules into this storage compartment. Whether PADGEM has its own targeting signal or whether its storage depends on vWF has not been determined. Several physiologic secretagogues for endothelial cells have been identified, including thrombin," fibrin,"? and histamine.l! All these secretagogues disrupt the junctions between the endothelial cells and thereby allow the released vWF to diffuse in all directions. In contrast, when secretagogues that do not disturb the monolayer, such as the tumor promoter phorbol myristate acetate or a calcium ionophore, are used, Weibel-Palade bodies release their contents predominantly basolaterally. Microtubules may provide the mechanical means by which this polarized transport of Weibel-Palade bodies to the cell surface can occur. 12 The large vWF multimers released from Weibel-Palade bodies bind avidly to extracellular matrices. The binding of vWF to the matrix or to any other solid substratum renders the protein competent for interaction with glycoprotein Ib on resting platelets. To arrive at its final multimeric form, vWF undergoes several processingsteps, some ofwhich

Mayo Clin Proc, June 1991, Vol 66

may be unique to vWF. Similar to other proteins, vWF processing follows a defined sequence of events, and the exit of the protein from the different cellular compartments seems to be tightly controlled. vWF processing can be divided into early (in the endoplasmic reticulum, where dimerization occurs) and late (in the Golgi complex and post-Golgi compartments, where the dimers multimerize and the prosequence is removed by proteolytic cleavage) steps. At any specific time, about halfofthe total cellular vWF is found in the endoplasmic reticulum of endothelial cells.!" In this compartment, N-linked high mannose oligosaccharides are added to the nascent pro-vWF chain, and the monomeric provWF subunits are cross-linked by an unknown number of disulfide bonds present at the carboxyl-terminal end of the subunit. 14 The resulting pro-vWF dimers are then transported from the endoplasmic reticulum to the Golgi apparatus. vWF exits from umbilical vein endothelial cells approximately 2 hours after the onset of synthesis of vWF. Pro-vWF monomers are retained in the endoplasmic reticulum and are not secreted by endothelial cells, megakaryocytes, or cell lines transfected with complementary DNA encoding for pro-vWF.1 Treatment of human endothelial cells with the antibiotic tunicamycin results in the failure of pro-vWF dimerization and subsequent retention of the monomers in the endoplasmic reticulum of human umbilical vein endothelial cells!" (Fig. I). Is this retention due solely to the lack of dimerization, or is the addition of carbohydrate also important for intracellular transport? Can truncated vWF that is unable to form carboxylterminal interchain disulfide bonds be secreted? Recently, experiments performed by us and by Voorberg and associates!" provided some answers to these questions. Pro-vWF consists of four repeated domains, denoted DI-D2-D'-D3AI-A2-A3-D4-BI-B2-B3-CI-C2, in whichDI and D2 represent the amino-terminal prosequence. Voorberg and colleagues have used site-directed mutagenesis to introduce a stop codon beyond domain D3. Consequently, transfection of COS cells with this construct resulted in synthesis of vWF subunits, consisting of only domains DI-

Mayo Clin Proc, June 1991, Vol 66



Fig. 1. Effect of tunicamycin on pro-von Willebrand factor (vWF) cleavage and secretion by human umbilical vein endothelial cells. Cells were incubated overnight in the presence (+) or absence (-) of 1 ug/ml of tunicamycin and then metabolically labeled with 35S-methionine for 24 hours in the same medium. vWF was purified from both the culture medium and the cell lysate and analyzed reduced on a 5% polyacrylamide gel, an autoradiograph of which is shown. Tunicamycintreated cells contained the precursor subunit (provWf> , but no cleaved subunit (mvWf> was detected. Secretion ofvWF was completely inhibited in tunicamycintreated cells, in that the medium lacked the precursor and cleaved subunits that were secreted in control cultures. (From Wagner and assoeiates.P By permission of the Rockefeller University Press.)

D2-D'-D3 and lacking the majority ofthe mature C2 domain, was also efficiently secreted from subunit, that were unable to dimerize in the en- the endoplasmic reticulum. doplasmic reticulum. These monomeric subunits, Figure 2 shows the vWF fragment purified however, were secreted normally from the endo- from the culture medium of metabolically laplasmic reticulum.!" In our studies, during beled RIN5F cells and analyzed reduced and preparation of stable cell lines that express pro- nonreduced on a 5% polyacrylamide gel. The vWF, we isolated an RIN5F cell (rat insulinoma subunit composition, being either the mature line) clone that produced a mutant pro-vWF vWF or the pro-vWF fragment (labeled mf and protein. The protein migrated slightly faster prof, respectively), was determined for the five than did the wild-type protein on reduced poly- bands denoted on the nonreduced gel by two-diacrylamide gels (Fig. 2), an indication that the mensional gel electrophoresis. The nonreduced subunits were approximately 20 kd shorter than gel strip was incubated in the presence of a the wild-type protein. The mutant protein was reducing agent before it was placed on top of a not recognized by the monoclonal antibody 2.29 second polyacrylamide gel into which the vari(provided by Dr. Zaverio Ruggeri), which binds ous vWF species migrated under reducing condito an epitope located at the carboxyl-terminal tions (not shown). The autoradiograph of the end ofvWF (the site ofinterchain disulfide bonds nonreduced gel shown in Figure 2 was intentioninvolved in dimerization). This truncated pro- ally overexposed to detect all the vWF fragments tein, which most likely lacks a portion of the last and their combinations secreted by this cell line.



Mayo CIin Proc, June 1991, Vol 66

R Fig. 2. Von Willebrand factor (vWF) species secreted by RIN5F insulinoma cells expressing truncated pro-vWF without the carboxyl-terminal region ofthe molecule. The RIN5F cells were metabolically labeled with 35S-cysteine for 30 minutes with 1 mCi/ml and then chased for 5 hours in unlabeled medium from which the mature vWF and pro-vWF fragments (labeled mf and prof, respectively) were isolated by using polyclonal antibodies to vWF attached to protein A-Sepharose beads. Samples were analyzed both reduced (R) and nonreduced (NR) on 5% polyacrylamide gels, the autoradiograph of which is shown. The reduced gel shows that the truncated protein is efficiently cleaved to the mature form. Some uncleaved pro-vWF subunits are also secreted. Bars indicate the position of migration of the wild-type pro-vWF and mature vWF subunits on the same gel. The nonreduced gel shows that both cleaved and uncleaved monomeric fragments and three species of dimers that contain combinations of uncleaved and cleaved subunits were secreted.

On a brief exposure, the only prominent band corresponded to the dimer of the mature fragment subunits (mf-mf), a finding that indicates that most ofthe expressed protein formed dimers and was proteolytically processed. Because the monomeric pro-vWF fragments were efficiently secreted from the endoplasmic reticulum, apparently the carboxyl-terminal region of pro-vWF is responsible for retention of the monomeric protein in the endoplasmic reticulum. This retention may be accomplished by binding ofvWF to the heat shock protein BiPY

Rothmari'" proposed that BiP may recognize exposed regions on misfolded or not yet completely folded or oligomerized proteins. Dimerization at the carboxyl-terminal end may mask a potential BiP binding sequence and therefore preventvWFfrom binding to BiP. Interestingly, both types of the truncated protein, one lacking the carboxyl-terminal 20 kd (Fig. 2) and the other containing only the D' and D3 domains, 16 were able to make the amino-terminal interchain disulfide bonds, which resulted in the formation of dimeric molecules. In contrast to

Mayo CUn Proc, June 1991, Vol 66


endothelial cells, where the intracellular pool of dimers is sensitive to endoglycosidase H (that is, contains high mannose-type carbohydrate), the dimeric vWF fragments in the RIN5F cells were completely resistant to this enzyme (not shown). This result confirms that the dimers in the RIN5F cells were formed late in processing, only after the carbohydrate was fully processed. In addition, the formation ofthe dimeric molecules could be inhibited by the weak base ammonium chloride (not shown), further evidence that dimerization occurs in the acidic trans-Golgi or post-Golgi compartment. Collectively, these observations demonstrate that the formation of the amino-terminal interchain disulfide bonds leading to vWF multimerization is entirely independent of prior formation of carboxyl-terminal interchain disulfide bonds that leads to dimerization in the endoplasmic reticulum. Although the D1-D2- D'-D3 truncated protein expressed in COS cells by Voorberg and cclleagues'" clearly contains the information necessary for disulfide bonding through the D3 domain, additional interchain disulfide bonds may exist in the Al domain of the wild-type vWF molecule." To examine the importance of N-linked glycosylation on exit of vWF from the endoplasmic reticulum, we subjected the RIN5F cells that produce the large pro-vWF fragment to the drug tunicamycin (Fig. 3). The pattern observed was similar to that seen with wild-type vWF treated with tunicamycin (compare with Figure 1)namely, only pro-vWF subunits were detected in the treated cells, and the vWF fragment was not secreted (an indication of retention of the protein in the endoplasmic reticulum). Therefore, N-linked carbohydrates apparently playa role in intracellular transport of vWF that is independent of their role in dimerization. The mechanism of formation of interdimer disulfide bonds is yet unknown. Several conditions were identified as necessary for the multimerization process. First, the presence of the propolypeptide is an absolute requirement. This conclusion was obtained from both expression studies with use of vWF complementary DNA with deleted propolypeptide'v-" and in vitro multimerization studies with mature and pro-


vWF dimers used as starting material. 22 Recently, Voorberg and colleagues'" demonstrated that the short D' domain at the amino-terminal end of the mature subunit is also necessary for vWF multimerization, even though no interchain disulfide bonds h.ave been localized to this domain. Second, an acidic pH environment is needed. This requirement was demonstrated in two ways. Human umbilical vein endothelial cells were grown in the presence of a weak base (ammonium chloride), which increases the pH of acidic cellular compartments, such as the transGolgi network and secretory granules. The presence of the weak base inhibits the multimerization ofdimers." Similarly, in vitro multimerization ofpro-vWF dimers occurs only under acidic pH conditions with an optimal pH between 5.4 and 5.8. 2 2 Third, a requirement for divalent cations has been observed. In vitro multimerization was inhibited with ethylenediaminetetraacetic acid, and the inhibition could be overcome by the addition of excess Ca2 + (Mayadas TN, Wagner DD: Unpublished data). Storage granules are known to contain high Ca 2 + concentrations, which may be one of the reasons why vWF multimerization is efficient in the storage granules. vWF multimerization does not seem to require cellular enzymes, as demonstrated by the fact that it can proceed in vitro in the absence of any cellular components. Mayadas and Wagner'" proposed that the propolypeptide may facilitate disulfide interchange involving free sulfhydryl groups on the mature subunit. This hypothesis, which currently has no experimental backing, stems from several observations. First, spontaneous formation of disulfide bonds from free sulfhydryls is promoted by basic pH conditions and is unlikely to occur at acidic pH conditions. Next, cotransfection of vWF propolypeptide coding complementary DNA with mature vWF coding complementary DNA promoted multimerization of the mature subunit.>' Finally, the vicinal disulfide cysteine-glycine x cysteine, which forms the catalytic site of the enzyme protein disulfide isomerase found only in the endoplasmic reticulum, and that of thioredoxins, which are small oxidation-reduction proteins first discovered in



Mayo Clin Proc, June 1991, Vol 66


Fig. 3. Effect oftunicamycin on pro-von Willebrand factor (vWF) fragment cleavage and secretion by RIN5F insulinoma cells expressing the truncated vWF protein. Cells were pretreated overnight with (+) or without (-) 2 ug/ml of tunicamycin and then metabolically labeled for 24 hours with 35S-cysteine (30 !lCi/ml) in the presence or absence ofthe drug. The vWF fragments were purified from both the lysed cells and the culture medium by using vWF polyclonal antibody-coated protein A-Sepharose and analyzed reduced on 5% polyacrylamide gel, the autoradiograph of which is shown. Tunicamycin-treated cells contained the precursor subunit (prof) but no cleaved subunit (mf). Secretion ofvWF was completely inhibited in tunicamycin-treated cells, an indication that the unglycosylated protein was retained in the endoplasmic reticulum.

bacteria." are also present once in each D1 and D2 domain of the propolypeptide.v' Another proposed function for the propolypeptide may be in physically aligning the dimeric molecules and thereby facilitating the formation of disulfide bonds between the subunits of neighboring dimers. The propolypeptide must dimerize by noncovalent interactions at some point during vWF biosynthesis, inasmuch as the circulating free plasma propolypeptide is dimeric.P Thus for the moment, we can report the following conclusion:

vWF's way ofmultimerizing Is unique and quite enterprising. It requires the pro To facilitate the show Under conditions we're closely scrutinizing.


We thank Tanya N. Mayadas for critical reading of the manuscript and for writing the limerick onvWF.


Wagner DD: Cell biology of von Willebrand factor. Annu Rev Cell BioI 6:217-246,1990 2. Loesberg C, Gonsalves MD, Zandbergen J, Willems C, VanAken WG, Stel HV, Van MourikJA, De Groot PG: The effect of calcium on the secretion of factor VIIIrelated antigen by cultured human endothelial cells. Biochim Biophys Acta 763:160-168,1983 3. Sporn LA, Marder VJ, Wagner DD: Inducible secretion of large, biologically potent von Willebrand factor multimers. Cell 46:185-190, 1986 4. Wagner DD, Olmsted JB, Marder VJ: Immunolocalization of von Willebrand protein in Weibel-Palade bodies of human endothelial cells. J Cell Biol 95:355360, 1982

Mayo Clin Proc, June 1991, Vol 66


6. 7.



10. 11.



14. 15.


Cramer EM, Meyer D, Ie Menn R, Breton-Gorius J: Eccentric localization of von Willebrand factor in an internal structure of platelet a-granule resembling thatofWeibel-Paladebodies. Blood 66:710-713,1985 Bonfanti R, Furie BC, Furie B, Wagner DD: PADGEM (GMP140) is a component of Weibel-Palade bodies of human endothelial cells. Blood 73:1109-1112, 1989 Hattori R, Hamilton KK, Fugate RD, McEver RP, Sims PJ: Stimulated secretion of endothelial von Willebrand factor is accompanied by rapid redistribution to the cell surface of the intracellular granule membrane protein GMP-140. J BioI Chern 264:77687771, 1989 Larsen E, Celi A, Gilbert GE, Furie BC, Erban JK, Bonfanti R, Wagner DD, Furie B: PADGEM protein: a receptor that mediates the interaction of activated plateletswithneutrophilsandmonocytes. Cell 59:305312, 1989 LevineJD,HarlanJM,HarkerLA,JosephML, Counts RB: Thrombin-mediated release of factor VIII antigen from human umbilical vein endothelial cells in culture. Blood 60:531-534, 1982 Ribes JA, Francis CW, Wagner DD: Fibrin induces release of von Willebrand factor from endothelial cells. J Clin Invest 79:117-123, 1987 Hamilton KK, Sims PJ: Changes in cytosolic Ca 2 + associated with von Willebrand factor release in human endothelial cells exposed to histamine: study of microcarrier cell monolayers using the fluorescent probe indo-I. J Clin Invest 79:600-608, 1987 Sporn LA, Marder VJ, Wagner DD: Differing polarity of the constitutive and regulated secretory pathways for von Willebrand factor in endothelial cells. J Cell BioI 108:1283-1289, 1989 Wagner DD, Marder VJ: Biosynthesis of von Willebrand protein by human endothelial cells: processing steps and their intracellular localization. J Cell BioI 99:2123-2130, 1984 Titani K, Marti T, Takio K, Walsh KA: Primary structure of human von Willebrand factor. Hematology 9:99-116, 1989 Wagner DD, Mayadas T, Marder VJ: Initial glycosylation and acidic pH in the Golgi apparatus are



18. 19.








required for multimerization of von Willebrand factor. J Cell BioI 102:1320-1324, 1986 Voorberg J, Fontijn R, van Mourik JA, Pannekoek H: Domains involved in multimer assembly of von Willebrand factor (vWF): multimerization is independent of dimerization. EMBO J 9:797-803, 1990 Dorner AJ, Bole DG, Kaufman RJ: The relationship of N-linked glycosylation and heavy chain-binding protein association with the secretion of glycoproteins. J Cell BioI 105:2665-2674, 1987 Rothman JE: Polypeptide chain binding proteins: catalysts of protein folding and related processes in cells. Cell 59:591-601, 1989 MohriH, Fujimura Y, ShimaM, Yoshioka A, Houghten RA, Ruggeri ZM, Zimmerman TS: Structure of the von Willebrand factor domain interacting with glycoprotein lb. J BioI Chern 263:17901-17904,1988 Verweij CL, Hart M, Pannekoek H: Expression of variant von Willebrand factor (vWF) eDNA inheterologous cells: requirement ofthe pro-polypeptide in vWF multimer formation. EMBO J 6:2885-2890, 1987 Wise RJ, Pittman DD, Handin RI, Kaufman RJ, Orkin SH: The propeptide of von Willebrand factor independently mediates the assembly of von Willebrand multimers. Cell 52:229-236, 1988 Mayadas TN, Wagner DD: In vitro multimerization of von Willebrand factor is triggered by low pH: importance of the propolypeptide and free sulfhydryls. J BioI Chern 264:13497-13503,1989 EdmanJC, EllisL, BlacherRW, RothRA, RutterWJ: Sequence of protein disulphide isomerase and implications of its relationship to thioredoxin. Nature 317:267-270,1985 Bonthron D, Orr EC, Mitsock LM, GinsburgD, Handin RI, Orkin SH: Nucleotide sequence of pre-pro-von Willebrand factor eDNA. Nucleic Acids Res 14:71257127, 1986 Wagner DD, Fay PJ, Sporn LA, Sinha S, Lawrence SO, Marder VJ: Divergent fates of von Willebrand factor and its propolypeptide (von Willebrand antigen II) after secretion from endothelial cells. Proc Nat! Acad Sci USA 84:1955-1959,1987

von Willebrand factor and the endothelium.

Endothelial cells are the principal source of plasma and basement membrane von Willebrand factor (vWF). To arrive at its biologically active multimeri...
1MB Sizes 0 Downloads 0 Views