Brian Storrie, Department of Physiology and Biophysics, University of Arkansas for Medical Sciences, 4301 West Markham Street, Little Rock, AR 72205, USA. Tel.: +1 501 526 7418; fax: +1 501 686 8167; e-mail: firstname.lastname@example.org
Summary. Background: von Willebrand factor (VWF) and fibrinogen are major storage proteins of platelet α-granules. VWF is synthesized by the megakaryocyte, the cell from which platelets bud, and fibrinogen is delivered to α-granules by endocytosis.Aim: Considering biosynthetic origins, VWF and fibrinogen might be differentially packaged within platelets. We applied immunofluorescence microscopy to provide whole platelet, global information on the distributions of VWF and fibrinogen.Results: The distribution of VWF and fibrinogen were characterized in both the resting state and handling activated human platelets. Full cell volume image stacks were collected by spinning-disk confocal microscopy, corrected for a small pixel shift between green and red channels, deconvolved, and visualized in a three-dimensional space. In sum, we found that there was little overlap in the distribution of VWF and fibrinogen in resting state platelets. In an important control, the distributions of green and red secondary antibodies overlapped completely when different color secondary antibodies directed against the same first antibody were used. Moreover, the same result was observed using different first antibodies and switching second antibody color to switch the color of VWF and fibrinogen staining. No accumulation of fibrinogen in late endosomes or lysosomes was detected by co-staining with LAMP2, a late endosome/lysosome membrane protein. Significantly, we found that in handling activated platelets there was differential retention of fibrinogen-positive granules relative to VWF positive granules.Conclusion: Our results indicate that VWF and fibrinogen are differentially packaged in human platelets. Moreover, the results suggest that differential packaging could support differential release of α-granule proteins.
Platelets are anucleate cell fragments with limited biosynthetic function that are critical for normal hemostasis and blood coagulation [1,2]. Upon activation at sites of vascular injury, platelets release secretory granule proteins to initiate the formation of a platelet aggregate that aids in wound healing. Three types of secretory organelles are present in platelets: α-granules, dense granules and lysosomes. These organelles have characteristic molecular composition, ultrastructure and sensitivity to stimuli [3–5]. α-Granules are the major secretory organelle in platelets and, based on electron microscopy of thin sections, there are 50–80 spherical to ellipsoid α-granules per cell with a diameter of 200–500 nm [4,6,7]. These granules contain coagulation factors (e.g. fibrinogen), pro-angiogenic agents and some adhesion molecules [e.g. von Willebrand factor (VWF)], P-selectin, fibronectin and growth factors (e.g. platelet-derived growth factor). Some of these molecules such as VWF and the granule membrane protein, P-selectin, which is also found in association with dense granules, are synthesized primarily by the platelet precursor cell, the megakaryocyte. On the other hand, fibrinogen is the type example of a second class of α-granule proteins that are synthesized by other cells and internalized by receptor-mediated endocytosis [8,9]. Using electron microscopy, α-granule polypeptides are differentially distributed in distinct regions within the granule. By immunogold labeling of thawed cryosections, VWF, for example, is found concentrated in the eccentric rims of α-granules [10,11]. As immunogold labeling cryosections are ∼50 nm thick , only a very small portion of an individual α-granule or more importantly the 1 μm or greater thickness platelet is sampled. Hence, the existing electron microscopy provides little data on the more global distribution of α-granule proteins with respect to one another.
Using immunofluorescence, a lower resolution and more global technique, α-granule proteins show a distinct punctate intracellular staining pattern suggestive of their granular localization, but often fail to show more than a general co-localization in wide-field micrographs . Using cell fractionation, various α-granule proteins tend to co-distribute [14,15]. In response to stimuli such as thrombin, α-granule proteins, for example, P-selectin and PF4, are secreted with a similar dose response . However, the kinetics of release can be significantly different with fibrinogen being released faster than VWF in response to collagen [15,16]. In patients with inherited gray platelet syndrome, platelets are deficient in α-granules and multiple α-granule proteins including fibrinogen and VWF [17,18]. In sum, these data are commonly interpreted to suggest that there is a single α-granule population within platelets that consists of individual granules in which different α-granule proteins are individually segregated . Less commonly, an alternative explanation is suggested, namely, that there is one, or more, subsets of α-granules . In a second cell type, endothelial cells, the Weibel–Palade body, an analogous organelle to the α-granule of platelets , is heterogenesis in protein composition and displays differential exocytosis of P-selectin and VWF .
In the present study, we re-examined the distribution of the major α-granule proteins, VWF and fibrinogen, using current state-of-the-art confocal fluorescence microscopy approaches. Full cell volume image stacks were collected, corrected for a slight pixel shift between green and red light, deconvolved, and visualized in a three-dimensional space. We found that the distribution of VWF and fibrinogen was decidedly heterogeneous in resting state platelets. In addition, fibrinogen showed little overlap in distribution with late endosomes or lysosomes indicating little fibrinogen was endosomal. Moreover, we found that in handling activated platelets there was differential loss of fibrinogen-positive granules relative to VWF positive granules. Our results indicate that VWF and fibrinogen are differentially packaged in platelets and are consistent with the possibility that they could be differentially released in response to physiological stimuli.
Purification of platelets
For preparation of resting state platelets, human blood (approximately 10 mL) was drawn into citrated tubes and mixed immediately with an equal amount of fixative. Four different conditions were used: (i) mix with an equal volume of ice-cold 4% formaldehyde prepared in phosphate-buffered saline (PBS) and fix for 2 h on ice, for example, formaldehyde, room temperature (RT) −>4 °C; (ii) mix with an equal volume of RT 4% formaldehyde prepared in PBS and fix for 2 h at RT, for example, formaldehyde, RT; (iii) mix with an equal volume of 4% paraformaldehyde prepared in PBS and fix for 2 h at RT, for example, paraformaldehyde, RT; or (iv) mix with an equal volume of 37 °C 4% formaldehyde prepared in PBS and fix for 2 h at 37 °C, for example, formaldehyde, 37 °C. In most cases, formaldehyde, RT −>4 °C fixation was used. Subsequent platelet purification steps were at 3–4 °C. Platelet-rich plasma supernatant was then obtained by centrifugation (Beckman TJ-6R centrifuge, TH-4 swinging bucket rotor; Beckman Coulter, Fullerton, CA, USA) at 300× g (high brake setting) for 15 min. The supernatant was centrifuged at 1300× g (low brake setting) for 12 min to pellet the platelets. The pellet was suspended in Tyrode’s buffer, pH 6.5, and platelets were sequentially pelleted twice at 1300× g (low brake setting) for 12 min in Tyrode’s buffer, pH 6.5. The resulting platelet pellet was suspended in RT Tyrode’s buffer, pH 6.5, supplemented with 0.1% bovine serum albumin, 1 mm CaCl2, 1 mm MgCl2 and prostaglandin E1 (0.35 μL mL−1). An equal volume of 4% paraformaldehyde in PBS was added to the final platelet suspension and isolated platelets were fixed for an additional 30 min at RT. For preparation of handling activated platelets, fixation was delayed until after the centrifugation steps. Platelet count was measured with a Hemavet HV 950FS (Drew Scientific Inc., Oxford, CT, USA) and was about 360 000–450 000 in the final suspension. Collection of normal blood was after consent and institutional review board approval in accordance with the Declaration of Helsinki.
Fibrinogen was labeled with either mouse monoclonal antibody (mAb) (clone 2C2-G7; BD Biosciences, San Jose, CA, USA) directed against human fibrinogen at 1:100 dilution or rabbit polyclonal antibodies (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) directed against human fibrinogen-α chain at 1:100 dilution. VWF was labeled with rabbit polyclonal antibodies (Dako A/S, Glostrup, Denmark) directed against human VWF at 1:200 dilution or mouse mAb (clone 2 F2-A9, BD Biosciences) directed against human VWF at 1:100 dilution. P-selectin was labeled with goat polyclonal antibodies (Santa Cruz Biotechnologies) directed against human P-selectin (at 1:100 dilution). LAMP-2 was labeled with mouse mAb (clone H4B4, Developmental Studies Hybridoma Bank, Iowa City, IA, USA) directed against human LAMP-2 (1:100 dilution). Fluorescein isothiocyanate (FITC)-conjugated (1:100 dilution) or Cy3-conjugated (1:1000 dilution) donkey anti-rabbit, donkey anti-mouse and donkey anti-goat immunoglobulin G were used as secondary antibodies. LAMP-2 secondary antibody was labeled with Alexa Fluor 488-conjugated goat anti-mouse immunoglobulin G (1:100 dilution).
One hundred microliters of fixed platelet suspension at an approximate platelet count of 400 000 μL−1 was incubated on the surface of 11-mm glass coverslips for 90 min at 37 °C to allow the platelets to adhere to the coverslip surface. High humidity was maintained during the incubation by placing the coverslips inside covered tissue culture dishes that floated in a water bath. For subsequent cell handling steps, coverslips were placed cell side down on top of reagent drops. Next, 50 mm ammonium chloride was used to quench the fixative and saponin-gelatin in PBS was used to permeabilize the platelets. Alternatively, methanol (−20 °C, 4 min), acetone (−20 °C, 4 min) or 0.5% Triton X-100 in PBS (RT, 4 min) was used for permeabilization. Use of methanol or acetone for permeabilization was particularly important for staining with the VWF mAb. Platelets were then sequentially stained with the respective antibodies (20 min each for first and second antibodies) diluted in saponin-gelatin-PBS. Platelets were washed three times with saponin-gelatin in PBS after each staining step. Coverslips were mounted in Mowiol (Calbiochem, San Diego, CA, USA).
Wide-field and confocal microscopy
A Zeiss Axiovert 200 M microscope (Carl Zeiss MicroImaging Inc., Thornwood, NY, USA) with 100×, 1.4 numerical aperture and 63×, 1.4 numerical aperture objectives was used to take single plane wide-field images. Spinning-disk confocal image stacks were taken with a BD Biosciences CARVII accessory fitted to the Zeiss microscope as described earlier . IPLab 3.9.5 software (BD Biosciences) was used to capture the images.
Image processing and analysis
Small pixel shifts were corrected between the green and red channels based on the observed shift between the two channels when fibrinogen was labeled with green and red secondary antibodies directed against the same first antibody. Following correction, there was full overlap of green and red channels when secondary antibodies directed against the same first antibody were used to stain platelets (Supplementary Fig. S1). The raw confocal stack images were deconvolved with iterations limited to ten (huygens essential software, version 2.10; Scientific Volume Imaging, Hilversum, the Netherlands). Maximum intensity projections (MIP) were normally carried out with iplab software. Simulated three-dimensional fluorescence projections of the deconvolved images were generated using the huygens essential software. Typically, simulated fluorescence is shown with the illumination light coming from the NW and the images are tilted 45° to the vertical. Fig. 2E with the XZ projection was generated with the simulated projection image tilted at 90° to the vertical. Surface rendered projections were generated with huygens essential by adjusting the thresholds to match the corresponding MIP images. Quantification of granule number was performed using single channel images of the surface projections and counting the protuberances of either color channel. The average number of granules was calculated over a number of individual platelets (n = 38 for resting state and n = 40 for handling activated platelets). The quantification was performed for platelets co-labeled with fibrinogen and VWF.
Widefield microscopy demonstrates limited co-localization of fibrinogen and VWF in resting state platelets
Freshly fixed human platelets [formaldehyde, RT −>4 °C] were permeabilized and stained using commercial antibodies against fibrinogen and VWF. Using differential interference contrast (DIC) microscopy, the coverslip-attached platelets were round without any pseudopodial extensions (Fig. 1A), indicating that the isolated platelets were maintained in a resting state. Using dual-label wide-field fluorescence microscopy, much of the VWF (green) and fibrinogen (red) fluorescence appeared distinct from each other with very little yellow color indicative of co-localization observed (Fig. 1B). Images shown were corrected for small pixel shifts between green and red channels. Note that as expected for wide field microscopy, the fluorescence images are somewhat blurry. These results suggest that there may be a differential packaging of VWF and fibrinogen in platelets.
The differential distribution of VWF and fibrinogen becomes more obvious with confocal microscopy
To investigate further the apparent limited co-localization of VWF and fibrinogen, we used spinning-disk confocal microscopy. This is a high signal to noise technique compared with laser scanning confocal microscopy . Before analysis, all image stacks were corrected for small pixel shifts between green and red channels (Supplementary Fig. S1). As shown in Fig. 2A, individual confocal images spaced 125 nm apart in the Z plane included almost all the image information for a single cell in eight planes, 1 μm. Most individual structures whether positive for VWF (green) or fibrinogen (red) spanned a few image planes consistent with the 0.5 μm Z-dimension resolution of the microscope. In comparison to the previous wide field imaging, the confocal images were sharper and again there was very limited co-localization of VWF and fibrinogen as indicated by a yellow color. With deconvolution to correct for the point spread function of the objective, the images became even sharper and the lack of co-localization was very apparent (Fig. 2B). Gray-scale images of the individual confocal optical sections for each channel are shown in Supplementary Fig. S2; differences in intensity are much more apparent in gray-scale images. As seen in Fig. 2C, MIP of deconvolved image stacks provided a crisp single image characteristic of the full-image stack. In the second approach, three-dimensional simulated fluorescence projections (SFPs) of the image stacks were computed at a 45 ° tilt. As seen in Fig. 2D, this approach had the advantage of showing where an individual stained granule was located in a three-dimensional space. The individual VWF and fibrinogen channel SFPs are shown in Supplementary Fig. S2. In the third approach, the cell was rotated 90 ° to provide an XZ projection (Fig. 2E). This approach provided good Z-resolution but discarded XY information. Qualitatively, dissimilarity in distributions was seen in each of the hundreds of platelets examined. In sum, the differential distribution of VWF and fibrinogen was clearest in the deconvolved confocal image stacks, particularly when these were displayed as SFPs of the individual channels and the merged image (Supplementary Fig. S2). The image stacks were fully corrected for any green vs. red channel shift and the distributions of green and red secondary antibodies directed against the same first antibody displayed a full overlap in distribution when presented as described in Methods (Supplementary Fig. S1).
Next, we verified that the differential distributions were not a consequence of fixation conditions, the antibodies used, or the permeabilization conditions. A range of different formaldehyde fixation conditions all yielded a similar differential distribution of fibrinogen and VWF (Methods, Supplementary Fig. S3). In Fig. 2 and Supplementary Figs S1–3, fibrinogen was visualized using a mAb and VWF was visualized using polyclonal antibodies and red and green secondary antibodies. Similar experimental outcomes were observed when the colors of the secondary antibodies were switched and when fibrinogen was visualized with polyclonal antibodies and VWF with a mAb (data not shown). We found similar staining if Triton X-100, methanol or acetone were used instead of saponin (data not shown).
Fibrinogen and LAMP-2, a late endosomal (multivesicular body)/lysosomal membrane marker, fail to co-localize
Fibrinogen is delivered to platelet granules after its endocytosis and trafficking through late endosomes/multivesicular bodies . To test whether fibrinogen might be retained in late endosomes (multivesicular bodies), we compared the distribution of fibrinogen with LAMP-2, a late endosomal (multivesicular body)/lysosomal membrane marker. In deconvolved confocal image stacks, LAMP-2 was localized in structures that were typically restricted to a small, central region of the platelet while fibrinogen showed little co-localization with LAMP2 and instead was located in structures that were distributed throughout the platelet (Fig. 3). These results indicate little fibrinogen in resting state platelets is retained within a late endosomal (multivesicular body) population.
A simple retention assay provides a second line of evidence for differential packaging of the major α-granule proteins VWF and fibrinogen
The above results suggest strongly that VWF and fibrinogen are differentially packaged within human platelets. If so, a retention assay might well provide a second line of evidence for differential packaging. For example, differential retention might well be observed upon weak activation. We took as a simple assay the frequency of VWF- and fibrinogen-positive structures in resting state and handling activated platelets. Platelets have long been known to display traits of weak activation with handling/cold treatment. As shown in Fig. 4, when fixation was delayed until after platelet purification, the isolated platelets displayed traits indicative of their activation. Using DIC microscopy, the attached platelets appeared to be more spread and frequently had pseudopodial extensions (Fig. 4A, arrows). Activation was further confirmed by the surface localization of P-selectin. Having validated that handling activated platelets display traits characteristic of activation, we asked if there might be differential retention of fibrinogen or VWF after such weak activation. Qualitatively, MIP images and three-dimensional SFPs of handling activated platelets appeared to contain few fibrinogen positive (red channel) compared with VWF positive (green channel) structures (Fig. 4B,C). We proceeded to quantify these with the assumption that separated structures were too close together to be resolved fully by confocal microscopy and hence that individual fluorescence thickenings (protuberances) corresponded to individual α-granules. We scored granule number in this manner for surface rendered three-dimensional projections of resting and handling activated platelets stained for fibrinogen and VWF. Fig. 4D shows an example of surface-rendered granules in a handling activated platelet. As shown in Fig. 5, approximately equal numbers of fibrinogen and VWF positive granules were quantified in resting platelets. In contrast, in handling activated platelets, there was an appreciable decrease in fibrinogen positive granules with only a small decrease in VWF positive granules. Quantitatively, there was little overlap in the distribution of fibrinogen and VWF positive granules in either case. In conclusion, this approach provides a second line of evidence for the differential packaging of VWF and fibrinogen in human platelets.
We characterized the global distribution of fibrinogen and VWF, two major stored adhesive proteins in platelets, relative to each other and LAMP2 (late endosomes/lysosomes). Fibrinogen is endocytosed into platelets while VWF is synthesized by megakaryocytes, the cells from which platelets are derived. We found at the limits of resolution of confocal light microscopy, ∼0.2 μm XY and ∼0.5 μm Z, that there was limited co-localization between fibrinogen and VWF, a clear indication of differential packaging. Co-localization comparisons with LAMP2 indicated that there was little retention of fibrinogen in late endosomes (multivesicular bodies). Moreover, we found in a second assay for differential packaging, a retention assay after weak activation, that fibrinogen-positive structures were preferentially reduced in number relative to VWF-positive structures indicating again the differential packaging of these major granule proteins. The differential packaging of these major platelet granule proteins could well have important implications for platelet function.
Our immunolocalizations are valid light microscopic representations of the distribution of granule proteins in resting and handling activated platelets. Similar results were observed with all fixation conditions tested including RT goes to chilled, constant RT and constant 37 °C fixation. The confocal image stacks were collected through the entire platelet depth using the best high-resolution objectives. Spinning-disk confocal microscopy was used to eliminate bleaching during image collection and to maximize light collection to the camera. The technique gives a much higher signal-to-noise ratio than laser scanning confocal microscopy. Green and red channels were corrected for a minor pixel shift between wavelengths. The resulting image stacks were deconvolved to remove blur. The use of SFPs and surface rendering allowed visualization of structure (granule) staining through the entire platelet in a single image. The same differential distributions were observed for fibrinogen and VWF whether visualized with (i) monoclonal or polyclonal antibodies or vice versa, (ii) red or green second antibodies or vice versa, and (iii) different permeabilization methods.
Electron microscopy indicates that α-granule proteins may be packaged into discrete zones within a 50-nm-thick α-granule section . We observed near total separation in the distribution of the major α-granule proteins, fibrinogen and VWF, within the full volume of individual platelets. In our quantification, we equated thickenings (protuberances) in the fluorescence distributions to individual granules. Making this assumption, granule size would be a few tenths of a micron and the scored total number of granules per resting platelet was about 33, approximately half the number estimated from thin section electron microscopy . Comparisons with the distribution of LAMP2 as a late endosomal (multivesicular body)/lysosome marker indicated that fibrinogen was not retained within late endosomes (multivesicular bodies) to lysosomes. Preliminary studies on GPIIbIIIa distribution indicate that open canicular system (OCS) invaginations of the cell surface are few in number compared with granule numbers (our unpubl. obs.) indicating that little fibrinogen is present in the OCS. Our results have parallels in previous observations. Early wide field microscopy from the mid-1980s points to fibrinogen and VWF, then called factor VIII-related antigen, not exhibiting tight co-localization . There are reports that fibrinogen release from platelets is more rapid than that of VWF [15,16]. Moreover, the analogous organelle in endothelial cells, Weibel–Palade bodies have been reported recently to exhibit a similar heterogeneity in composition and exocytosis .
In sum, our results indicate extensive heterogeneity in α-granule protein distributions. Whether the underlying source of this heterogeneity at the ultrastructural level is as a result of differential packaging within individual α-granules or heterogeneity in the granule population will require further study. A clear future challenge will be to relate individual global protein distributions to α-granules as classically defined by electron microscopy. New techniques in light microscopy such as stimulated emission depletion fluorescence microscopy that are capable of 50 nm resolution  and electron microscopy tomography with thick sections  hold great promise in answering this question definitely. These techniques combine high resolution with the ability to globally image distributions. The relationship of the observed heterogeneity in mature platelets to organelle and platelet biogenesis in the megakaryocyte is for now an open question. However, based on the present data, some mechanistic speculations can be made. Firstly, the difference in fibrinogen and VWF distribution is unlikely to be because of their differing biosynthetic origin, endocytosis vs. megakaryocytic. These data provide little to no evidence for fibrinogen being sequestered to a late endosomal (multivesicular body) compartment or other non-granule structures. Secondly, multiple adaptor proteins for protein targeting to different destinations are known. In endothelial cells, VWF, for example, is packaged into Weibel–Palade bodies at the Golgi apparatus and is adaptor protein independent while the transport of other proteins into the elongate, cigar-shaped Weibel–Palade body is AP-3 adaptor complex dependent  indicating multiple routes of delivery to the Weibel–Palade body. Thirdly, likely, functionally important clusters of α-granule proteins are found in either discrete zones within large individual α-granules or in distinct α-granule subpopulations. The limited overlap in protein distributions observed in the present work is suggestive of multiple subpopulations of α-granules. Our observations are the novel outcome of the application of state-of-the-art techniques and definitive within the limits of the approaches taken. In conclusion, we suggest that differential packaging may be a hallmark feature of platelets that could have important implications for their biogenesis and function in hemostasis. We note also in closing that our results provide a possible underlying subcellular basis for the recently reported differential release of endostatin vs. vascular endothelial growth factor by platelet granules in response to proteinase-activated receptor (PAR)1 vs. PAR4 agonists [25, see also for review, 26].
We express our appreciation to Jerry Ware, UAMS, for continuous comments on the work. This work was supported by a grant from the Arkansas Biosciences Institute.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.