Efficiency of von Willebrand factor-mediated targeting of interleukin-8 into Weibel–Palade bodies



    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    Search for more papers by this author

    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    2. Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht
    Search for more papers by this author
  • A. KRAGT,

    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    Search for more papers by this author

    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    2. Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht
    Search for more papers by this author

    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    Search for more papers by this author

    1. Department of Plasma Proteins, Sanquin Research and Landsteiner Laboratory AMC, University of Amsterdam, Amsterdam
    2. Department of Vascular Medicine, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands
    Search for more papers by this author

Jan A. van Mourik, Department of Plasma Proteins, Sanquin Research, Plesmanlaan 125, 1066 CX Amsterdam, The Netherlands.
Tel.: +31 20 5123120; fax: +31 20 5123680; e-mail: j.vanmourik@sanquin.nl


Summary. Background: After de novo synthesis in endothelial cells, the chemokine interleukin-8 (IL-8) is targeted to endothelial cell-specific storage vesicles, the Weibel–Palade bodies (WPBs), where it colocalizes with von Willebrand factor (VWF). Objective: In this study we investigated a putative regulator function for VWF in the recruitment of IL-8 to WPBs. Methods: We performed a quantitative analysis of the entry of IL-8 into the storage system of the endothelium using pulse-chase analysis and subcellular fractionation studies. Results: Using pulse-chase analysis of IL-1β-stimulated human umbilical vein endothelial cells, we found that a small part of de novo synthesized IL-8 was retained in endothelial cells after 4 h. In density gradients of endothelial cell homogenates nearly equimolar amounts of VWF and IL-8 were present in subcellular fractions that contained WPBs. Furthermore, we found that IL-8 binds to immobilized VWF under the slightly acidic conditions thought to prevail in the lumen of the late secretory pathway. Conclusions: These observations indicate that the sorting efficiency of IL-8 into the regulated secretory pathway of the endothelium is tightly controlled by the entry of VWF into WPBs.


Endothelial cells contain typical, elongated vesicles, the so-called Weibel–Palade bodies (WPBs) [1], which serve as a storage compartment for von Willebrand factor (VWF) [2], a large, multimeric, adhesive glycoprotein involved in platelet plug formation after vascular damage. Following stimulation of endothelial cells with agonists that raise intracellular Ca2+ or cyclic AMP (cAMP) levels, such as thrombin or epinephrine, WPBs are transported to and fuse with the plasma membrane, thereby delivering their contents onto the cellular surface or into the circulation and subendothelial tissue. Besides VWF, the cargo of WPBs may consist of a wide variety of membrane and secretory proteins, including P-selectin, angiopoietin-2, endothelin-1, osteoprotegerin-1 (OPG) and the chemokines interleukin-8 (IL-8) and eotaxin-3 (reviewed in [3]). Resting endothelial cells do not synthesize IL-8 or eotaxin-3 in significant amounts. De novo synthesis of these and other chemokines requires exposure of endothelial cells to cytokines such as IL-1β or IL-4. Subsequently, IL-8 and eotaxin-3 are stored in WPBs. Thus, endothelial cells are able to adapt to dynamic changes in the microenvironment of the vasculature by modifying the composition of their WPBs (reviewed in [3]).

Previously, we and others have shown that VWF can direct the storage of IL-8 and P-selectin in endothelial and non-endothelial cell types [4–6], and accumulating evidence indicates that a variety of other molecules are costored with VWF in WPBs. However, little is known about the nature and efficiency of the sorting process of IL-8. What portion of de novo synthesized protein destined for regulated secretion is targeted to WPBs? We have initiated experiments aimed at studying the efficiency of the sorting of IL-8 to WPBs after induction of its synthesis in primary endothelial cells. Using density gradient centrifugation of homogenates of endothelial cells, we made the novel observation that there is a linear relationship between the concentrations of IL-8 and VWF in WPB-containing subcellular fractions. We show that, irrespective of the amounts of IL-8 and VWF synthesized, these proteins are targeted to WPBs in virtually equimolar amounts. A similar linear relationship is found between the amounts of VWF and IL-8 that are released from the WPBs by thrombin-stimulated human umbilical vein endothelial cells (HUVEC). We also demonstrate that IL-8 is able to bind to immobilized VWF under the slightly acidic conditions thought to exist in the lumen of the trans-Golgi network (TGN). These observations suggest a mechanism in which IL-8 interacts with VWF in the TGN prior to vesicle formation in such a manner that the amount of VWF that enters the forming WPB governs the sorting efficiency of the cotargeted IL-8.

Materials and methods

Reagents and antibodies

Culture media, trypsin, penicillin and streptomycin were from Invitrogen (Breda, The Netherlands). IL-1β, IL-8, and polyclonal antihuman IL-8 were from Strathmann Biotech (Hannover, Germany). Monoclonal anti-IL-8 was from Sigma-Aldrich Chemie (Steinheim, Germany). Alexa 488- and Alexa 633-conjugated antibodies were from Molecular Probes (Breda, The Netherlands). Protein A-Sepharose, CNBr-activated Sepharose 4, Percoll and Pro-mix L[35S] were from Amersham Biosciences (Buckinghamshire, UK).

Cell culture

Endothelial cells were isolated from umbilical veins and cultured essentially as described previously [7]. All experiments were performed with early passage HUVEC (passage 3–5).

Pulse-chase analysis

HUVEC were seeded on 10 cm culture dishes (Nunc, Rochester, NY, USA), grown to confluency and incubated overnight with 10 ng mL−1 IL-1β. Cells were pre-incubated for 60 min in culture medium lacking l-methionine and l-cysteine supplemented with IL-1β. Cells were labeled for 30 min with 0.9 mCi Pro-mix L[35S], followed by chases for the indicated time periods in normal culture medium without IL-1β. After radiolabeling, IL-8 was immunoprecipitated from medium and cell lysates using preformed anti-IL-8–Protein A-Sepharose complexes.

Subcellular fractionation

HUVEC were grown in 175 cm2 flasks until they reached confluency, and were subsequently incubated for 24 h with 10 ng mL−1 IL-1β. Subcellular fractionation using Percoll density gradient centrifugation was performed essentially as described previously [8]. Density gradients were established by centrifugation in a Beckmann Optima™ LX-100 XP ultracentrifuge (Beckmann Instruments, Palo Alto, CA, USA) equipped with the Ti50.2 fixed angle rotor for 30 min at 100 000 × g. Fractions (1.6 mL) were taken from bottom-up and the density was determined by weighing or by optical refraction.


HUVEC were grown on 6-well plates. Upon reaching confluency they were incubated for 48 h, either with or without 10 ng mL−1 IL-1β. After pre-incubation for 6 h in serum-free medium (20% serum replaced by 1% HSA) without IL-1β, cells were treated for 15 min with either 2U mL−1 thrombin or serum-free medium. Levels of VWF, VWF propeptide (VWFpp) and IL-8 released into the medium were determined by ELISA.


VWF antigen, VWFpp antigen and proVWF antigen were measured by ELISA as described previously [9,10]. IL-8 antigen was determined by ELISA (Sanquin, Amsterdam, The Netherlands).


Endothelial cells were grown on gelatin-coated coverslips. After 24 h of treatment either with or without 10 ng mL−1 IL-1β, cells were fixed with 3.7% formaldehyde for 15 min. VWF was visualized using CLB-RAg20 [11] and Alexa 633-conjugated antimouse immunoglobulin G (IgG)2b. VWFpp was visualized using CLB-Pro17 [9] and IL-8 was visualized using monoclonal anti-IL-8. For both, Alexa 488-conjugated antimouse IgG1 was used as secondary antibody. Cells were embedded in Vectashield mounting medium (Vector Laboratories, Burlington, CA, USA) and analyzed by confocal microscopy using a Zeiss LSM510 (Carl Zeiss, Sliedrecht, the Netherlands). For both conditions, 10–15 cells were randomly selected and images were generated by making optical sections (Z-stacks with 0.36-μm intervals). Maximal projections were analyzed using Image Pro Plus 6.0 (Media Cybernetics, Breda, the Netherlands) to quantify the number of IL-8-, VWFpp- and VWF- positive vesicles in single cells.

Recombinant VWF

Recombinant human VWF was expressed in HEK293 cells that were transfected with the pcDNA3.1+ wtVWF plasmid containing full-length human VWF cDNA. VWF was purified by immunoaffinity chromatography using CLB-RAg20 coupled to CNBr-activated Sepharose 4. Protein concentration was determined by Bradford assay. VWF concentration was determined by ELISA. Purity of the VWF was assessed by sodium dodecylsulfate–polyacrylamide gel electrophoresis (SDS–PAGE) under reduced conditions and silver staining, while the multimeric composition was analyzed using agarose gel electrophoresis and Western blotting with Dako-HRP (Dako A/S, Glostrup, Denmark).

IL-8 binding to immobilized VWF

Purified recombinant VWF (0.5 μg well−1) was immobilized to microtiter wells (Maxisorp®, Nunc). Non-bound VWF was removed by washing four times with a washing buffer containing 50 mm MES, 150 mm NaCl, 5 mm CaCl2 and 0.1% Tween-20 (pH 5.5). The remaining binding sites were blocked with binding buffer (washing buffer supplemented with 10 mg mL−1 bovine serum albumin). Wells were incubated with IL-8 (0–0.5 μm) for 2 h at 37 °C in binding buffer. In case of coincubation with soluble VWF, 0.15 μm IL-8 was supplemented with 0.01–0.15 μm VWF. Non-bound IL-8 was washed away with washing buffer. Bound IL-8 was detected using biotin-labeled anti-IL-8 followed by streptavidin-polyHRP and was corrected for non-specific binding, determined by incubation of IL-8 in empty wells. Surface plasmon resonance (SPR) studies were performed on a Biacore3000 (Biacore AB, Uppsala, Sweden). Twenty-two fmol mm−2 purified VWF was immobilized on a CM5 sensor chip by amine-coupling, as described by the manufacturer. As a control we used an uncoupled flow cell without VWF, which was subsequently blocked with ethanolamide. Binding to the control flow cell was typically less than 5% of binding to a flow cell to which VWF was coupled. Binding of recombinant IL-8 to immobilized VWF was analyzed in 150 mm NaCl, 10 mm CaCl2, 0.005% Tween-20, 5% glycerol and 50 mm HEPES (pH 7.4 and 6.7) or 50 mm MES (pH 6.2 and 5.5) for 4 min at 25 °C with a flow rate of 20 μL min−1. Dissociation was initiated upon replacement of IL-8 solution with buffer. For regeneration of the surface, each incubation was followed with three incubations using the same buffer supplemented with 1 m NaCl. For calculation of affinity constants (KD), responses at 240 s of association (Req) were plotted against protein concentration. If maximum binding was not reached during the first 240 s of association, we estimated maximum binding by fitting the association curves to an exponential association function. Non-linear regression was performed on the resulting binding isotherms assuming a one-site binding hyperbola: Req = Req max* (IL-8)/[KD+(IL-8)].

Statistical analysis

Non-linear and linear regression analyses using the Deming method and Student’s or paired t-tests were performed with graphpad prism version 4.03 (GraphPad Software, San Diego, CA, USA).


Pulse-chase analysis of storage and secretion of IL-8 in endothelial cells

Initially, pulse-chase experiments were performed in order to explore and compare the cellular retention of IL-8 and VWF. Unlike endothelial cells of different origin, such as human intestinal microvascular endothelial cells, resting HUVEC do not express IL-8 unless activated to upregulate IL-8 expression by proinflammatory cytokines such as IL-1β [12]. HUVEC were incubated for 18 h with IL-1β in order to induce IL-8 synthesis. These cells were subsequently pulse-labeled and at various chase times labeled IL-8 from medium and cell lysates was immunoprecipitated and analyzed by SDS–PAGE and autoradiography. During the first hour of chase IL-8 was almost completely secreted (Fig. 1). After a chase period of 4 h only a small amount of IL-8 remained associated with cells. Apparently the relative amount of IL-8 that enters the storage pathway of IL-1β-stimulated HUVEC is low; virtually all de novo produced IL-8 leaves the cell within 40 min following its synthesis.

Figure 1.

 Release and retention of interleukin-8 (IL-8) in endothelial cells.
Endothelial cells, incubated overnight with 10 ng mL−1 IL-1β in order to upregulate IL-8 gene expression, were metabolically labeled for 30 min, followed by chases in unlabeled medium for the indicated time periods. IL-8 was immunoprecipitated from respectively conditioned media and cell lysates 1, 40, 80, 120 and 240 min after addition of unlabeled medium. Immunoprecipitates were subjected to SDS–PAGE under reducing conditions, after which autoradiography was performed.

IL-8 and VWF cosediment in comparable amounts in WPB-containing subcellular fractions

Next, we determined the amounts of VWF and IL-8 that are stored in WPBs after upregulation of IL-8 synthesis by incubation with IL-1β. Homogenates of IL-1β-stimulated HUVEC were subjected to density gradient centrifugation and the concentrations of IL-8, VWF, VWFpp and proVWF were determined in the various fractions by ELISA. Figure 2A shows the analysis of a density gradient, which is representative of seven independent experiments (Supplementary Table 1). IL-8 and VWF were found in three subcellular fractions: a dense fraction with a density ranging from 1.10–1.16 g mL−1, containing WPBs [8]; a buoyant fraction with a density ranging from 1.06–1.09 g mL−1, containing subcellular organelles of the secretory pathway [endoplasmic reticulum (ER), Golgi, TGN] and vesicles containing proteins that constitutively leave the cell; and a top fraction, containing material from cell organelles that lyzed during the cell disruption procedure. The dense fractions contain primarily mature VWF and approximately equimolar amounts of VWFpp (Supplementary Table 1). This is consistent with the observation that VWF and VWFpp are secreted in a 1:1 molar ratio after stimulation with agonists [9,13,14]. The VWF that was found in the buoyant fractions was largely in its unprocessed, proVWF form (Fig. 2A). In the experiment depicted in Fig. 2, about 1.8% of the cellular IL-8 was found in the dense granules. The IL-8 concentration in these fractions was in the same order of magnitude as the concentration of VWF and VWFpp (0.2–1.0 pmol fraction−1, Supplementary Table 1). In this experiment the total amounts of IL-8, VWF and VWFpp recovered after gradient centrifugation were 87%, 95% and 91% respectively.

Figure 2.

 von Willebrand factor (VWF), VWF propeptide (VWFpp), proVWF and interleukin-8 (IL-8) distribution in subcellular fractions. (A) Postnuclear supernatants of IL-1β-stimulated cells were fractionated by Percoll density gradient centrifugation. Shown is a representative experiment (indicated with ? in Supplementary Table 1). (○), VWF; (bsl00066), VWFpp; (□), proVWF; (◆), IL-8; (line), density. (B, C) Linear regression analysis of the total molar amount of VWF vs. the total molar amount of IL-8 (B) and VWFpp (C) in the dense fraction. The equations for the plotted lines are given in the graphs (five to seven experiments, Supplementary Table 1).

There was a considerable variation in terms of VWF and IL-8 production among the different HUVEC cell cultures following IL-1β stimulation. Not only did the total amount of VWF and IL-8 that was found in the total lysate fluctuate, but also the portion of VWF and IL-8 found in the dense fractions (Supplementary Table 1). The variation in the amount of VWF found in the dense fractions has been observed before [8]. It is reasonable to assume that this can be attributed to differences in VWF expression in various HUVEC cultures. In addition, individual HUVEC cultures responded differently to IL-1β stimulation. The total amount of IL-8 found in lysates of IL-1β-stimulated cells varied more than thirtyfold (Supplementary Table 1). However, this was not reflected by the amount of IL-8 found in the dense fractions. In fact, there was a significant linear correlation (= 0.0035) between the molar amounts of IL-8 and VWF that were found in the dense fractions. The WPB fraction, on average, contained 0.7 moles IL-8 per mole VWF monomer (Fig. 2B). Similarly, there was a significant linear correlation (= 0.0009) between the molar amounts of VWF recovered in the WPB fractions and the molar amounts of propeptide found in these fractions (1.2 mole propeptide per mole VWF, Fig. 2C). The difference between the molar ratios of VWFpp:VWF and IL-8:VWF was not significant (= 0.2450 by paired t-test).

IL-8 and VWF are released in similar amounts by IL-1β-treated HUVEC upon thrombin stimulation

As there is a linear relationship between the amounts of VWF, VWFpp and IL-8 stored in WPBs of IL-1β-treated endothelial cells, it is to be expected that this is also reflected in the quantities of these proteins secreted when cells are stimulated to release the contents of their WPBs. To test this, IL-1β-treated and untreated HUVEC were stimulated for 15 min with 2U mL−1 thrombin (Fig. 3A). Indeed, a linear relationship was found between the amounts of VWF and IL-8 apparently released by IL-1β-treated endothelial cells in response to thrombin stimulation. On average, per mole VWF 0.6 moles IL-8 were recovered (P=0.0104; Fig. 3B). This was consistent with the observed molar ratio revealed by the density gradient fractionation studies. No detectable IL-8 secretion was found when endothelial cells were not subjected to IL-1β treatment. Treatment with IL-1β is known to decrease VWF expression [15], which was apparent from the reduction in VWF and VWFpp levels released from endothelial cells under these conditions (Fig. 3A). As expected [9,14] for VWF and VWFpp, significant linear relationships were also found in resting HUVEC (= 0.0039) as well as after IL-1β treatment (= 0.0028). The molar ratios of VWF:VWFpp were 1:1.3 (no IL-1β) and 1:1.6 (+IL-1β) respectively (Fig. 3B–C). The difference between the molar ratios of VWFpp:VWF and IL-8:VWF released was not significant (= 0.1505 by paired t-test).

Figure 3.

 Regulated release of von Willebrand factor (VWF), VWF propeptide (VWFpp) and interleukin-8 (IL-8) from thrombin-stimulated endothelial cells. (A) Human umbilical vein endothelial cells were treated for 48 h with or without 10 ng mL−1 IL-1β, as described in Materials and methods. After 6 h of pre-incubation in serum-free medium without IL-1β, cells were stimulated for 15 min with serum-free medium containing 2 U mL−1 thrombin (black bars) or medium alone (white bars). The concentrations of VWF, VWFpp and IL-8 secreted in the medium were measured by ELISA. Error bars represent SEM. *P<0.05, **P<0.005, ***P<0.0001 by Student’s t-test. (B, C) Linear regression analysis of the amounts of VWF vs. the amounts of VWFpp (□) and IL-8 (bsl00066) released after thrombin stimulation (corrected for constitutive release) in resting (B) and IL-1β-stimulated (C) endothelial cells. The equations for the plotted lines are given in the graphs. Data represent the mean of at least seven experiments.

IL-8 and VWF are heterogeneously distributed in the WPBs of IL-1β-treated HUVEC

The observation that about 0.7 mole IL-8 stored per mole VWF monomeric subunit is recovered in the WPB-containing fractions raises the question of whether IL-8 and VWF are equally distributed over all WPBs or, alternatively, that subpopulations of WPBs exist that contain different amounts of IL-8 and VWF. Previously we have shown that the latter possibility is probably the case [6]. In order to address this issue in a quantitative manner, we immunolocalized IL-8 and VWF in IL-1β-treated endothelial cells and quantified the number of WPBs that contained both VWF and IL-8 or VWF alone (Fig. 4). IL-8 colocalizes with about 70% of the VWF-containing vesicles. In addition, IL-8 is found in structures typical of the late secretory pathway (Golgi, TGN), as well as in small spherical organelles that do not contain VWF (Fig. 4A, upper panels). The latter are presumably vesicles of the constitutive secretory pathway, as inhibition of de novo protein synthesis by prolonged incubation with cycloheximide results in the selective disappearance of these vesicles (not shown). Virtually all WPBs of resting endothelial cells contain VWF as well as VWFpp (Fig. 4A, lower panels), a phenomenon predicted by the non-covalent association of VWF and VWFpp after the endoproteolytic processing events preceding sorting and storage of these proteins in WPBs [16]. Similarly, VWFpp shows virtually complete colocalization with all vesicles staining for VWF in IL-1β-stimulated endothelial cells (not shown).

Figure 4.

 Interleukin-1β (IL-1β)-stimulated endothelial cells contain a heterogeneous population of Weibel–Palade bodies (WPBs). Human umbilical vein endothelial cells were treated for 24 h with (upper panels) or without (lower panels) 10 ng mL−1 IL-1β. The intracellular localization of IL-8 or VWF propeptide (VWFpp) was compared with that of VWF. Shown are maximal projections; scale bar represents 10 μm. Detailed insets are shown. WPBs exclusively containing VWF are indicated by arrowheads, and small spherical vesicles that contain only IL-8 are indicated by arrows. The Golgi apparatus is indicated with *. (B) Quantification of relative numbers of WPBs that contain VWF and IL-8 or VWF and VWFpp respectively. Mean values are indicated, error bars represent SEM.

IL-8 displays specific and saturable binding to VWF

The simplest explanation for our observations that VWF and IL-8 are present in nearly equimolar concentrations in the WPBs is that VWF controls IL-8 targeting to storage organelles by stoichiometric interaction. In order to confirm potential intracellular interactions of IL-8 and VWF, we used an ELISA-based approach to detect binding of IL-8 to immobilized VWF in the presence of 5 mm Ca2+ at pH 5.5. This is a condition thought to prevail in the in statu nascendi secretory granule of (neuro)endocrine cells [17] and also at the interface of the trans-most Golgi cisternae and the forming trans-Golgi vesicles [18], the site where entry of both proteins could take place. We demonstrated that under these conditions IL-8 binds to VWF in a specific and saturable manner (Fig. 5A). Soluble VWF competed with immobilized VWF for IL-8 binding in a dose-dependent manner (Fig. 5B), demonstrating the specificity of the interaction. The late secretory pathway constitutes a gradually acidifying environment, with the overall pH of the respective subcellular compartments dropping from 7.2–7.4 (ER) to 6.2–6.4 (Golgi and TGN) and 5.2–5.5 inside the nascent vesicle [19]. Complementary to the approach described above, we explored the pH dependence of VWF–IL-8 interactions by SPR analysis to determine at which stage IL-8 and VWF are able to associate. VWF was immobilized to a CM5 sensor chip and was perfused with 200 nm IL-8 at various pH values (Fig. 5C). While only minimal interaction of IL-8 with VWF was observed under neutral conditions, binding significantly increased at pH 6.2 and (less pronouncedly) at pH 5.5.

Figure 5.

 Interaction of interleukin-8 (IL-8) and von Willebrand factor (VWF) in vitro. (A) Immobilized VWF (□) was incubated with various concentrations of IL-8 (0–0.5 μm). In a competition experiment 0.15 μm IL-8 was coincubated with 0.15 μm soluble VWF (bsl00066, arrow). Data represent mean values (±SEM) of five experiments. Data were fitted in a non-linear regression curve assuming a one-site binding model. (B) 0.15 μm IL-8 was coincubated with 0–0.15 μm soluble recombinant VWF. Bound IL-8 was expressed as the percentage of IL-8 bound in the absence of soluble VWF. Data represent mean values (±SEM) of four experiments. **P=0.0051, ***P=0.0008 by paired t-test. (C) VWF immobilized onto a CM5 sensor chip was incubated with 200 nm IL-8 in neutral (pH 7.4, curve I) or acidic (pH 6.2, curve II; pH 5.5, curve III) conditions. The amount of associated IL-8 is expressed as response units and is corrected for bulk refractive index changes and aspecific binding to an uncoupled control channel. Data shown are representative binding curves of three to four independent experiments.

For further analysis of the equilibrium binding isotherms under these conditions, we plotted the maximum binding response for a range of IL-8 concentrations as a function of those concentrations. We estimated an apparent KD by fitting these data to a hyperbola using non-linear regression. This revealed that VWF and IL-8 bind with low affinity at pH 7.4 (KD = 2.7 μm), and with increased affinity under more acidic conditions (KD = 0.6 μm and 0.2 μm at pH 5.5 and pH 6.2 respectively).


In order to obtain comprehensive information about the significance of the targeting of IL-8 to the regulated secretory pathway we used subcellular density centrifugation as a technique to assess IL-8 sorting on a quantitative basis. We have made the interesting observation that the amount of de novo synthesized IL-8 that is sorted to WPBs correlates in a linear fashion with the amount of VWF associated with these organelles (Fig. 2B, Supplementary Table 1). On a molar basis IL-8 and VWF are stored in WPBs, isolated by density centrifugation, in nearly equal amounts. Also, the amounts of VWFpp and VWF recovered in the dense fractions were virtually equimolar (Fig. 2C). As it has been demonstrated previously by different techniques that VWFpp is a typical WPB resident that, together with mature VWF, is stored in WPBs in equimolar amounts [9,14,16], this observation underscores the validity of the use of density gradient centrifugation to assess the molecular composition of WPBs in a quantitative manner. The observation that IL-8, VWF and VWFpp are secreted in similar amounts (Fig. 3, also [9]) is in concert with the 1:1 stoichiometry of these molecules found in WPBs (Supplementary Table 1). Differences in the molar amounts of IL-8 and VWF recovered in the medium might be due to differential binding of the proteins to the cell surface after release and/or, as discussed below, heterogeneity of the WPB population.

The amount of IL-8 synthesized after stimulation with IL-1β, as well as the amount of VWF synthesized under these conditions, varies considerably between individual HUVEC isolates (Supplementary Table 1). Nevertheless, we found that irrespective of the amounts of IL-8 (and VWF) synthesized, the molar concentration of IL-8 and VWF in WPBs was in the same order of magnitude (Fig. 2B, Supplementary Table 1). Apparently, the VWF-mediated targeting of IL-8 to WPBs is at a 1:1 stoichiometric level. This suggests specific interactions between these secretory molecules at the molecular level (see below). Depending on the amount of IL-8 synthesized, about 2–15% of the cellular IL-8 was associated with WPBs. Most of the IL-8 (and VWF) recovered was found in the buoyant fractions. Taking into account that the bulk of IL-8 is rapidly released after its synthesis (Fig. 1), this indicates that only a minor part of newly synthesized IL-8 is targeted to WPBs. In this respect, the sorting efficiency of IL-8 in endothelial cells may seem inadequate. On the other hand, if one takes into account that VWF facilitates the entry of IL-8 into WPBs on a virtually equimolar basis (that is, 1 mole IL-8 per mole 220 kD VWF monomer), the observed targeting efficiency could be considered as significant and efficient.

As relatively low amounts of de novo synthesized IL-8 are sorted into WPBs, the physiological significance of our observations could be questioned. On the other hand, as the estimated IL-8 concentration in WPBs is in the μm range (calculated from the amount of IL-8 released upon stimulation and from the amount of IL-8 found in dense fractions), this is sufficient to elicit an immediate inflammatory response in the vascular microenvironment. Therefore, the view that IL-8 storage could be considered as a mechanism to allow recruitment of this potent chemotactic agent to sites of vascular perturbation on demand remains a valid hypothesis [12,20].

We and others [6,12,20] have observed that endothelial cells, after prolonged treatment with IL-1β, contain subpopulations of WPBs that differ with respect to the presence of IL-8. Recent reports have confirmed that distinct subpopulations of WPBs exist. For instance, P-selectin, a typical WPB resident, does not consistently colocalize with VWF [21]. Subpopulations based on CD63 content [22] or the presence of Rab27a [23] have also been reported. Under the experimental conditions studied here, we do not find IL-8-negative WPBs in close proximity to the TGN (Fig. 4). We therefore assume that IL-8-negative WPBs represent vesicles that are formed before or shortly after the onset of IL-1β-induced upregulation of IL-8 synthesis. Clearly, these vesicles constitute a subpopulation of WPBs. It would be of interest to know whether these senior vesicles are subject to differential regulation in terms of vesicle dynamics and/or secretion competence [21,24–26].

Because VWF expression is the driving force behind the biogenesis of its own storage vesicle [27], any cosegregated protein will, to some extent, be stored in proportion. However, our finding that not only a linear relationship exists between the molar amounts of VWF and IL-8 found in WPBs, but also that this ratio is virtually equimolar, suggests that each monomeric VWF subunit acts as a cargo-receptor, providing a single binding site for IL-8. The observation that IL-8 is able to bind to immobilized VWF under conditions that mimic the nascent granule in a specific and saturable manner (Fig. 5) is in support of this view. We speculate that after upregulation of IL-8 synthesis, VWF multimers are saturated by IL-8, which thereby ‘gets a ride’ into newly forming vesicles. Alternatively, IL-8 and VWF may both enter the forming secretory granule without prior association. This will only occur after the budding event due to selective coaggregation in the acidifying and condensing immature secretory granule. Subsequently, the molar excess of IL-8 will be removed from the condensing vesicle, a mechanism that is consistent with the sorting-by-retention hypothesis [28]. However, our finding that also under less acidic conditions (pH 6.2) both proteins can interact suggests that the interaction may occur earlier in the TGN. The relatively high concentrations of IL-8 and VWF in this compartment (see above) are probably sufficient to allow binding in a quantitative manner despite the apparent low affinity of this interaction.

Direct interaction of VWF with other WPB components seems to be a more general phenomenon in the sorting of secretory proteins to the regulated secretory pathway of the endothelium. A lumenal part of P-selectin interacts intracellularly with cotransfected VWF in HEK293 cells, where this interaction is sufficient to trigger recruitment of P-selectin to pseudo-WPBs [22,29]. Another recently discovered component of the WPB, OPG, also displays intracellular interaction with VWF, a feature that could play a role in its sorting to WPBs [30].

We conclude that by binding of IL-8 in the late secretory pathway, VWF actively recruits IL-8 to the newly forming WPBs. We postulate that this interaction determines the sorting efficiency of IL-8 into the WPBs. Similarly, the efficiency of the recruitment of other WPB residents could be governed by VWF. This would underscore the significance of VWF as a pleiotropic modulator of homeostasis.


B. Luken and C. van der Zwaan are acknowledged for help with purification of VWF. The authors wish to thank A. Meijer for helpful discussion. This study is supported by the Netherlands Heart Foundation (grant 2002.B187), and the Landsteiner Foundation for Blood Transfusion Research (grant LSBR 03.15).

Disclosure of Conflicts of Interest

The authors state that they have no conflict of interest.