Differential Kinetics of Cell Surface Loss of von Willebrand Factor and Its Propolypeptide after Secretion from Weibel-Palade Bodies in Living Human Endothelial Cells*

The time course for cell surface loss of von Willebrand factor (VWF) and the propolypeptide of VWF (proregion) following exocytosis of individual Weibel-Palade bodies (WPBs) from single human endothelial cells was analyzed. Chimeras of enhanced green fluorescent protein (EGFP) and full-length pre-pro-VWF (VWF-EGFP) or the VWF propolypeptide (proregion-EGFP) were made and expressed in human umbilical vein endothelial cells. Expression of VWF-EGFP or proregion-EGFP resulted in fluorescent rod-shaped organelles that recruited the WPB membrane markers P-selectin and CD63. The WPB secretagogue histamine evoked exocytosis of these fluorescent WPBs and extracellular release of VWF-EGFP or proregion-EGFP. Secreted VWF-EGFP formed distinctive extracellular patches of fluorescence that were labeled with an extracellular antibody to VWF. The half-time for dispersal of VWF-EGFP from extracellular patches was 323.5 ± 146.2 s (±S.D., n = 20 WPBs). In contrast, secreted proregion-EGFP did not form extracellular patches but dispersed rapidly from its site of release. The half-time for dispersal of proregion-EGFP following WPB exocytosis was 2.98 ± 1.88 s (±S.D., n = 32 WPBs). The slow rate of loss of VWF-EGFP is consistent with the adhesive nature of this protein for the endothelial membrane. The much faster rate of loss of proregion-EGFP indicates that this protein does not interact strongly with extracellular VWF or the endothelial membrane and consequently may not play an adhesive role at the endothelial cell surface.

Weibel-Palade bodies (WPBs) 1 are large rod-shaped secretory granules found in endothelial cells (1)(2)(3)(4). Regulated exocytosis of WPBs delivers the adhesive protein von Willebrand factor (VWF) and the leukocyte adhesion molecule P-selectin to the endothelial cell surface, where they play important roles in vascular hemostasis and inflammation (1,5,6). Correct targeting of VWF into WPBs requires the propolypeptide of VWF (proregion) (7,8). The cleaved proregion is co-packaged with VWF in WPB in a 1:1 stoichiometry and released along with VWF during WPB exocytosis (9). The extracellular function of proregion is not clear. Previous biochemical studies have shown that although proregion does associate transiently with VWF inside the secretory pathway (10), it does not associate with extracellular VWF, matrix components, or the endothelial cell surface, indicating that it is unlikely to play an important role in endothelial-platelet adhesion (9). A putative role in regulating collagen-induced platelet-platelet aggregation has been suggested (11,12), and it may also act as a ligand for the VLA-4 (Very Late Antigen-4) integrin present on monocytes (13). VLA-4 plays an important role in the recruitment of monocytes into the blood vessel wall and is implicated in the vascular accumulation of monocytes that marks the early development of atherosclerotic lesions (14). These data indicate that WPB-derived VWF and proregion perform distinct extracellular functions, VWF on the endothelial surface and proregion at sites distant from the endothelial surface. Little is known about rates of dispersal of these proteins from the endothelial cell surface following WPB exocytosis, a process that will influence the effectiveness of these molecules at their sites of action. A recent study using viral infection of human umbilical vein endothelial cells (HUVECs) with a fusion protein of VWF and green fluorescent protein (GFP), in which the GFP replaced the A2 domain of VWF, has shown that WPBs can be fluorescently labeled and that stimulation results in the formation of extracellular patches of VWF-GFP that remained visible on the cell surface for up to 20 min (15). A detailed analysis of the rate of loss of surface-associated VWF was not made. In this study, we constructed chimeras of VWF and its proregion, fused to enhanced green fluorescent protein (EGFP) (16). We show that both constructs are correctly targeted to WPBs when expressed in HUVECs and that the resulting fluorescent WPBs undergo exocytosis in response to the WPB secretagogue histamine. Using these constructs, we have determined the time course of VWF-EGFP and proregion-EGFP loss from individual WPBs following exocytosis.

Antibodies and Reagents
Rabbit polyclonal anti-human VWF was purchased from Dako Ltd. (Cambridgeshire, UK). Mouse monoclonal anti-P-selectin (clone AK-6) was from Serotec (Oxon, UK). The anti-CD63 monoclonal antibody (clone H5C6) developed by J. T. August and J. E. K. Hildreth was obtained from The Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa. Secondary antibodies coupled to fluorophores were purchased from Jackson Immunoresearch. All other reagents were purchased from Sigma-Aldrich, unless stated otherwise.

Tissue Culture and Transfection
Primary HUVECs were purchased from TCS Cellworks and grown as previously described (17) or in the TCS Cellworks HUVEC culture medium. HUVECs were transfected using 5-10 g of expression vector DNA using the Nucleofection device and buffers according to the manufacturer's instructions (Amaxa GMBH). Cells were used between 1 and 3 days after Nucleofection.

Expression Vectors
pVWF-EGFP-A 229-bp PCR fragment flanked by EcoRI and AgeI restriction sites was generated from the carboxyl terminus of human VWF using primers 5Ј-ctgaattcgaggtggatatccactac-3Ј and 5Ј-ctgaccggtcccttgctgcacttcctggggg-3Ј and the vector pMT2-ADA-VWF (18) (American Type Culture Collection) as template. This fragment included a unique EcoRV site 204 bp from the VWF stop codon. The reverse primer was designed to mutate the VWF Stop codon (tga) into a glycine (gga). The PCR product was digested with EcoRI and AgeI and cloned into the EcoRI/AgeI sites of pEGFP-N1 (Clontech), giving rise to plasmid pVWFcterm-N1, which encodes a fusion of the VWF carboxyl terminus inframe with the EGFP cDNA. This cloning strategy gave rise to a joining sequence of four extraneous amino acids (Pro-Val-Ala-Thr) between the carboxyl-terminal glycine of VWF (mutated from Stop) and the aminoterminal methionine of EGFP. The entire 8442-bp human VWF coding sequence minus the 204-bp 3Ј of the unique EcoRV site (and with some additional 5Ј-UTR sequence) was cut out of pMT2-ADA-VWF using EcoRI and EcoRV and cloned into EcoRI/EcoRV-digested pVWF-cterm-N1, producing pVWF-FL-N1. A 350-bp EcoRI/XbaI fragment encoding the amino-terminal ATG of human VWF without any additional 5Ј-UTR was generated from pMT2-ADA-VWF using PCR with the primers 5Ј-gagaattcatgattcctgccagatttgc-3Ј and 5Ј-caccggacagcttgtagtacc-3Ј. This fragment was ligated into EcoRI/XbaI-digested pVWF-FL-N1 to give pVWF-EGFP. All PCR products were verified by sequencing. It should be noted that VWF is actually synthesized as a precursor protein (pre-pro-VWF) that has to undergo proteolytic processing in the secretory pathway to give rise to the mature (circulating) VWF (1). After post-translational modifications, the predicted product of the pre-pro-VWF-EGFP construct is EGFP-fused to the carboxyl terminus of mature VWF. For simplification, we refer to this construct as pVWF-EGFP, and we refer to the protein to which it gives rise as VWF-EGFP (rather than pre-pro-VWF-EGFP).
pProregion-EGFP-The proregion of VWF refers to the 741-amino acid protein that is generated from the amino terminus of pre-pro-VWF by its cleavage at a tetrabasic, furin-like cleavage site (19,20). To manufacture a proregion-EGFP fusion protein, we took advantage of the presence of a unique HindIII restriction site just 50 bp 5Ј of the proregion-VWF cleavage site. The HindIII site in the multiple cloning site of pEGFP-N2 (Clontech) was destroyed by restriction digest and subsequent filling-in, giving rise to pEGFP-N2 (minus HindIII). A 2.8-kb EcoRI/BamHI fragment encoding the entire proregion of VWF plus some additional carboxyl-terminal amino acids was cut out of pVWF-EGFP (above) and ligated into the EcoRI/BamHI sites of pEGFP-N2(minus HindIII), giving rise to pVWF-nterm-N2(minus). This was digested with HindIII/BamHI, and a 57-bp linker encoding a synthetic proregion carboxyl terminus in-frame with EGFP and with the tetrabasic cleavage site mutated (from Arg-Ser-Lys-Arg to Arg-Ser-Lys-Gly) was ligated in to generate pProregion-EGFP. The linker was generated with overlapping oligonucleotides (5Ј-cggaagcttgctgcctgacgctgtcctcagcagtcccctg-3Ј and 5Ј-gtggatccctttgctgcgatgagacaggggactgctgag-3Ј) that were annealed, filled in, and digested with HindIII/BamHI. The linker and the regions surrounding the restriction sites were verified by sequencing.

Immunocytochemistry
Nucleofected HUVECs grown on gelatin-coated glass coverslips were processed for immunofluorescence as described previously (21).

Fluorescence Imaging of WPB Exocytosis in Living Cells
Nucleofected HUVECs were grown on poly-D-lysine-coated glassbottomed chambers (MatTek Corp.). Time-lapse images of EGFP fluorescence were recorded in release medium (medium 199 supplemented with 10 mM HEPES, pH 7.2, and 0.2% bovine serum albumin), at room temperature or at 37°C as indicated, on a Deltavision Imaging system (Applied Precision Inc., Seattle, WA) housed within a temperaturecontrolled incubator. Images were obtained using either an Olympus U-Plan-Apo ϫ100 objective 1.35 numerical aperture or PLANAPO ϫ60 1.4 numerical aperture objective in conjunction with a Prinston Instru-ments Micromax air-cooled interlined charge-coupled device camera. 490 Ϯ 20 nm excitation and 528 Ϯ 38 nm emission filters were used in conjunction with a 4-band pass standard Deltavision dichroic (Chroma; C13022-84100). Excitation light was attenuated with a 50% neutral density filter to reduce photo-bleaching of EGFP. For experiments using proregion-EGFP, images were acquired at ϳ3.1 frames/s with continuous illumination. For experiments using VWF-EGFP, images were acquired for ϳ90 s at ϳ1.3 frames/s with continuous illumination (during the period in which WPB exocytosis was evoked), followed by intermittent illumination (300-ms exposure once every 5 s) for a period of Ͼ30 min. Under these conditions, imaging for 30 -35 min resulted in ϳ10% loss of EGFP fluorescence due to bleaching. Secretion was evoked by application of 100 M histamine. In some experiments, a rabbit polyclonal antibody to VWF (Dako Ltd.) conjugated to Alexa-546 using an anti-rabbit IgG Zeon Alexa Fluor 546 kit (Molecular Probes) according to the manufacturer's instructions was added along with histamine at a final bath dilution of 1:1000. In these experiments, EGFP and Alexa-546 fluorescence were acquired sequentially in time-lapse using 555 Ϯ 38 nm excitation and 617 Ϯ 73 nm emission filters for Alexa-546. The time course for changes in total WPB fluorescence following exocytosis was determined from time-lapse movies using the image analysis software Image J (rsb.info.nih.gov/ij/). Briefly, a region of interest was selected to include a single fluorescent WPB, and the total fluorescence within the region of interest was determined frame by frame through the image sequence. The fluorescence signal was corrected for background fluorescence by frame-by-frame subtraction of fluorescence from an adjacent region of interest lacking fluorescent WPBs. Data were normalized to the peak fluorescence after WPB exocytosis, and the time taken for fluorescence intensity to decline by 50% was used to compare the times for dispersal of proregion or VWF-EGFP from the release sites. (Fig. 1, aϪc) or proregion-EGFP (Fig. 1, d-f) in HUVECs resulted in the appearance of fluorescent, rod-shaped organelles that contained VWF. The VWF-EGFP-or proregion-EGFP-positive organelles recruited the known WPB membrane proteins P-selectin (22) and CD63 (23), although, as previously reported, the majority of CD63 is found in late endosomes/ lysosomes (17, 23, 24) (see Supplemental Figs. 1 and 2).

VWF-EGFP and Proregion-EGFP Are Correctly Targeted to WPBs When Expressed in HUVECs-Expression of VWF-EGFP
Histamine Evokes Exocytosis of VWF-EGFP-and Proregion-EGFP-containing WPBs-Stimulation of VWF-EGFP-expressing HUVECs with histamine (100 M, room temperature) caused the disappearance of fluorescent WPBs and the formation of distinctive patches of VWF-EGFP (Movie 1). Fig. 2A and Movie 2 confirm that the collapse of individual VWF-EGFPcontaining WPBs represents true exocytosis. In this experi- ment, an extracellular Alexa-546-conjugated anti-VWF antibody was applied at the same time as histamine to detect newly secreted VWF. The rapid binding (within a few seconds of WPB collapse) and co-localization of the anti-VWF antibody with patches of fluorescent VWF-EGFP confirm the extracellular nature of the newly formed patches of fluorescent VWF-EGFP. Exposure of non-stimulated cells to the fluorescent VWF antibody alone did not evoke exocytosis or formation of extracellular patches of VWF (data not shown). Histamine also evoked exocytosis of proregion-EGFP-containing WPBs (Movie 3). Fig.  2B and Movie 4 show the exocytosis of a single proregion-EGFP-containing WPB evoked by histamine and in the presence of an extracellular Alexa-546-conjugated anti-VWF antibody. In this case, proregion-EGFP was seen to rapidly disperse from the release site and did not form extracellular patches. The fluorescent VWF antibody detected extracellular VWF (co-packaged in the WPBs with the proregion-EGFP) at the release site within a few seconds of WPB exocytosis.
VWF-EGFP but Not Proregion-EGFP Is Lost Slowly from Release Site after WPB Exocytosis-Experiments were carried out at 37°C. When imaged with subsecond time resolution, an abrupt increase in the fluorescence of intra-WPB-EGFP is seen prior to or coinciding with release of contents (e.g. Figs. 2B and  3). This increase in fluorescence is due to the collapse of the WPB intra-granule pH to that of the extracellular medium (pH 7.4) following WPB fusion to the plasma membrane, and it is described in more detail elsewhere (25). WPB exocytosis can occur on both the upper and lower surfaces of the HUVEC cell membrane, and the time course data analyzed here represent a mixture of both cases. Fig. 3 shows the total fluorescence of a WPB containing proregion-EGFP (Fig. 3A) or VWF-EGFP (Fig.  3B) during exocytosis. The data are normalized to the peak fluorescence seen at the point of fusion. Exocytosis of proregion-EGFP-containing WPBs (Fig. 3A) resulted in an abrupt increase in fluorescence (mean, 2.7 Ϯ 0.7-fold, n ϭ 32 WPBs) followed by a rapid decline to background levels. No persistent extracellular patches of fluorescence could be observed. The time taken for fluorescence to decline to 50% of the initial peak increase was 2.98 Ϯ 1.88 s (ϮS.D., n ϭ 32 WPBs). Exocytosis of VWF-EGFP-containing WPBs also resulted in an abrupt in-crease in fluorescence (mean, 2.6 Ϯ 0.6-fold, n ϭ 20 WPBs) (Fig.  3B). Following the peak of the fluorescence increase, the fluorescence of newly formed extracellular patches of VWF-EGFP declined slowly toward background levels. The decline in fluorescence of extracellular patches of VWF-EGFP was interpreted as the loss of VWF-EGFP from the cell surface. The time taken for fluorescence to decline to 50% of the initial peak increase was 323.5 Ϯ 146.2 s (ϮS.D., n ϭ 20 WPBs). Extracellular fluorescent patches of VWF-EGFP at sites of WPB exocytosis could be observed 30 -40 min after exocytosis. The dramatic difference in the rate of dispersal of proregion-EGFP compared with VWF-EGFP is highlighted in Fig. 3C. For easier comparison, the normalized fluorescence data from Fig. 3A (black trace) and Fig. 3B (gray trace) were aligned at the point of WPB fusion and plotted on an expanded time scale to show the period immediately following fusion (Fig. 3C, iii). Images corresponding to the time points indicated by open circles in the traces in Fig. 3C, iii are shown in Fig. 3C, i (proregion-EGFP) and ii (VWF-EGFP).

DISCUSSION
Circulating VWF comprises predominantly dimers and low molecular weight multimers of VWF derived from constitutive secretion from vascular endothelial cells (see Ref. 1). These forms of soluble VWF do not readily adhere to the surface membrane of platelets or endothelial cells but can bind to exposed extracellular matrix components and promote platelet adhesion following vessel damage (26). VWF stored in WPBs consists of a highly multimeric form of VWF (27,28). Stimulation of cultured endothelial cells with thrombin or histamine results in WPB exocytosis and the appearance of complex patches of this form of VWF on the cell surface (15,27). The function of WPB-derived VWF on the endothelial surface is not clear. Basolateral secretion (toward the basement membrane) has been suggested to increase endothelial adhesion to the extracellular matrix components through interactions between VWF, endothelial integrin receptors, and basement membrane components (see Ref. 29). In vivo experiments have shown that apical secretion of VWF elicited by histamine supports platelet attachment and rolling on the endothelial cell membrane (6). The time course and extent of VWF-dependent platelet-endothelial surface interactions will depend in part on the kinetics of WPB exocytosis, VWF release, and the rate of its dispersal or loss from the endothelial cell surface. The time course for VWF-dependent platelet adhesion and rolling on the endothelial surface determined in vivo (6) includes a delay between endothelial stimulation and platelet adhesion of 15-20 s before an abrupt increase in the numbers of adherent and rolling platelets is seen, peaking ϳ30 -60 s after stimulation, and followed by a slow decline in the numbers of adherent and rolling platelets over 5-6 min to a lower but significantly elevated level (6). In previous studies, we have shown that WPB exocytosis in response to elevated [Ca 2ϩ ] i includes a delay of 10 -20 s (25,30), which may account in part for the delay in platelet adhesion observed following endothelial activation in vivo (6). In this study, we have shown that WPB exocytosis in living HUVECs leads to the formation of complex extracellular patches of VWF-GFP identical to those observed in histological studies following hormone stimulation (27) and in live HU-VECs (15). In the latter study (15), a time resolution of 60 s did not allow a detailed study of the formation and loss of VWF-EGFP from extracellular patches on the cell surface. Here we describe in more detail the slow, complex time course for loss of VWF-EGFP from extracellular patches with improved time resolution. There was some variability in the time course of VWF-EGFP loss from WPBs. This may reflect both a complex interplay of adhesive interactions between VWF-EGFP and components of endothelial surface and differences in the local environment in which the exocytotic events occurred (e.g. those that might exist on upper or lower surfaces of the cell). The overall slow time course for VWF-GFP dispersal seen here occurs over a similar time scale to the slow decrease in platelet adhesion and rolling observed in vivo (6), suggesting that the latter may be due in part to the slow loss of VWF from the endothelial surface membrane following WPB exocytosis. In contrast to VWF-EGFP, proregion-EGFP is rapidly lost from release sites (Ͼ2 orders of magnitude faster than VWF-EGFP), confirming that proregion does not form a strong association with either extracellular VWF or other endothelial surface components (9).