Deficiency in the Wiskott-Aldrich protein induces premature proplatelet formation and platelet production in the bone marrow compartment

Megakaryocytes (MKs) shed platelets by forming cytoplasmic extensions called proplatelets.¹  At the end of maturation, MKs migrate toward marrow sinusoids and either entirely transmigrate before forming platelets²  or protrude their proplatelets through the vascular endothelium.³  Ectopic localization of this process within the bone marrow environment would severely compromise efficient platelet production because platelets have low capacities for chemotaxis and transmigration, and therefore need to be directly released into the bloodstream. For this reason, efficient platelet production requires a synchronous process of cell migration and proplatelet formation. ECM⁴  and stromal cells⁵  appear to play an important role in the inhibition of platelet production by MKs within the marrow microenvironment, and the regulated loss of this specific interaction may precede MK migration from the bone marrow, and promote the onset of proplatelet formation.

The Wiskott-Aldrich syndrome protein (WASp), which is exclusively expressed in hematopoietic cells, has emerged as a key signaling molecule⁶  involved in cellular processes such as (1) reorganization of the actin cytoskeleton in response to specific signals,⁷  (2) formation of podosomes, specialized actin-rich structures identified in motile primary cells such as macrophages, dendritic cells (DCs), and osteoclasts,⁸  and (3) migration and cell trafficking of lymphocytes, DCs, and granulocytes.⁹  Accordingly, abnormalities of immune-cell adhesion and migration observed in WAS, the X-linked primary immunodeficiency caused by mutations in the gene encoding WASp, have been partly ascribed to the lack of podosomes and concomitant cell migration defects in human and mouse models. In contrast, the pathophysiology of microthrombocytopenia, which is characteristic of WAS and its attenuated forms, remains unclear, although it is considered that increased platelet peripheral destruction is the most important mechanism.⁶  Indeed, splenectomy of WAS patients usually results in substantial (though sometimes incomplete) correction of thrombocytopenia, but persistence of small platelet size.¹⁰  Isotope studies of platelet survival and release have also demonstrated that a component of the thrombocytopenia is related to inefficient platelet production.¹¹  This is unlikely to be due to a direct role of WASp in the process of pseudopod extension, which leads to proplatelet formation because platelet production by human WAS MKs is normal in liquid culture in vitro.¹²  In this system, platelet shedding occurs directly in the absence of cell-cell and cell-ECM interactions, and it is therefore limited in its ability to evaluate the important influences of MK interaction with the microenvironment and MK endothelial transmigration. WASp, an important regulator of the actin cytoskeleton,⁷  may play an important role in these last 2 mechanisms. Mice generated by gene targeting have provided a powerful model to study the role of WASp in multiple cell lineages of the hematopoietic system. However, until now study of the mechanism of thrombocytopenia in WAS has been limited because, unlike human patients, mice have only a moderate decrease in their platelet count and no abnormalities in their platelet size.¹³ 

In spite of these species differences, we report here that proplatelet formation in WASp-deficient mice may occur ectopically within the bone marrow. We also show that WASp-deficient MKs have a defect in the negative regulation of proplatelet formation mediated by α2β1, one of the main receptors of fibrillar collagen I (CI), which is an important component of the bone marrow microenvironment.¹⁴  In addition, in the presence of CI, WASp-deficient MKs show a profound defect in chemotactic migration, which is consistent with marked cytoskeletal abnormalities, such as the lack of actin-rich podosome structures that were selectively induced by CI in wild-type MKs.

Taken together, we suggest that WASp is involved in the modulation of MK actin reorganization, and that this participates in the prevention of premature transmigration of immature MKs in response to chemotactic factors and early platelet biogenesis in the bone marrow microenvironment. These events may provide a novel explanation for the defective platelet production in WASp-deficient mice and human patients.

C57Bl/6 and wild-type 129Sv/Ev mice were purchased from Iffa Credo (L'arbresle, France). WASp-deficient mice were provided by Drs Scott B. Snapper (Boston, MA) and Adrian Thrasher (London, United Kingdom). Mice were housed in animal facilities at the Institut Gustave Roussy (Villejuif, France) under specific pathogen-free conditions.

CI (Horm) was purchased from Nycomed (Munich, Germany). Poly-l-lysine (PLL), fibronectin, antivinculin mAb, and tetramethyl rhodamine isothiocyanate (TRITC)-conjugated phalloidin were from Sigma (St Quentin, France). Growth factor-reduced matrigel was obtained from BD Biosciences (San Jose, CA). Bovine fibrinogen was from Pentex (Kankakee, IL). Rabbit anti-von Willebrand factor (VWF) and anti-WASp (H-250) antibodies were obtained from Dakocytomation (Trappes, France) and Santa Cruz Biotechnology (Perray en Yvelines, France), respectively. Secondary antibodies were Alexa 488 goat anti-rabbit and Alexa 647 anti-mouse IgG (Molecular Probes, Eugene, OR). Convulxin (CVX) and GFOGER peptide were purified as described.^15,16 

Bone marrow (BM) nucleated cells from 129Sv, WASp-deficient mice, or C57Bl/6 mice were obtained from femurs and tibiae of 6- to 8-week-old adult mice. The lineage-negative (lin^-) fraction was purified from BM nucleated cells and grown for 72 hours in serum-free medium containing 10 ng/mL murine recombinant thrombopoietin (TPO; R&D Systems, Oxon, United Kingdom) as previously described.¹⁷ 

Standard. Epon-embedded thick sections (1 μm) of wild-type and WASp-deficient mouse bone marrow were stained with toluidine blue and examined by optical microscopy.

Immunohistochemistry. Immunohistochemistry was performed on paraffin-embedded 4-μm tissue sections from the WASp and control femora previously fixed in 10% neutral-buffered formalin and embedded in paraffin. Standard indirect immunoperoxidase procedures were used. The primary antibody was a polyclonal rabbit antibody factor VIII anti-humanrelated antigen (dilution, 1:2; DAKO, Carpinteria, CA); an avidinbiotinylated peroxidase system was applied for the immunohistochemical stains. Hematoxylin was the nuclear counterstain. Negative controls were performed by replacing the primary antibody with buffer.

Samples were fixed in 1.5% glutaraldehyde for 1 hour and washed 3 times in 0.1 M phosphate buffer, pH 7.4. For morphologic examination, samples were postfixed in 1% osmic acid, dehydrated in ethanol, and embedded in Epon by standard methods. Samples were counterstained and were observed on a Philips CM 10 electron microscope (Eindhoven, the Netherlands).

Cell migration was quantified through 12-μm-pore filters (Transwell, 24-well cell clusters; VWR, Strasbourg, France). Serum-free medium (100 μL) containing 2 × 10⁵  cells was placed in the upper chamber of the transwell precoated by matrigel or CI. In the lower chamber, 600 μL serum-free medium with (30, 100, or 300 ng/mL) or without SDF-1α (Abcys, Paris, France) was added. After 4 hours at 37°C in 5% CO₂, the cells from the lower chambers and the initial population were recovered in phosphate-buffered saline (PBS) containing 2% BSA, and then incubated with a purified rat anti-mouse CD41 antibody (MWReg30) for 20 minutes at 4°C. Negative controls were performed using purified rat IgG1 antibody. The cells were washed and labeled with an allophycocyanin-conjugated goat antirat antibody for 20 minutes at 4°C, washed, recovered in equal volumes of PBS containing 7-AAD, and counted by flow cytometry. The percentage of migration was calculated according to the following formula: number of CD41⁺ cells in the lower chamber divided by the number of CD41⁺ cells in the initial population. The 7-AAD DNA dye was used to exclude dead cells. All antibodies were purchased from BD Biosciences, and flow cytometry analyses were performed on a FACSort instrument (BD Biosciences).

CI (50 μg/mL), CVX (15 μg/mL), fibronectin (20 μg/mL), or fibrinogen (20 μg/mL) was incubated on coverslips overnight at 4°C. GFOGER peptide or PLL (20 μg/mL) was incubated for 1 hour at room temperature (RT). Cells were seeded on precoated coverslips at 37°C for the indicated times and processed for labeling as described.¹⁷  Briefly, cells were fixed in 2% paraformaldehyde for 10 minutes, washed with PBS, and permeabilized using Triton X-100 (0.2%) for 3 minutes. They were then incubated sequentially with antivinculin (1/400), rabbit anti-VWF (1/1000), or anti-WASp (1/400) antibodies. After washing, coverslips were incubated with TRITC-phalloidin (1/1000) and Alexa 647 goat anti-mouse or 488 anti-rabbit IgG (1/200), washed, and mounted with Vectashield antifading mounting medium (Vector, Burlingame, CA).

Cells were examined under an epifluorescence microscope (Nikon Eclipse 600; Nikon, Tokyo, Japan) equipped with a 63 ×/1.4 NA objective lens and a Zeiss laser scanning microscope (LSM 510; Zeiss, Jena, Germany) equipped with a Planapo oil-immersion lens (63 × magnification). Images were processed using CLSM5 Zeiss Browse Image software.

At least 300 MKs showing VWF⁺ staining were counted in each experiment in 3 independent experiments to determine the percentage of MKs showing VWF⁺ proplatelets (defined as cells exhibiting one or more cytoplasmic processes with areas of constriction) or actin-rich podosome structures and/or stress fibers.

The results are presented as mean plus or minus SD. The data were analyzed using the 2-tailed Student t test.

Microthrombocytopenia is a consistent feature of WAS and its mild form X-linked thrombocytopenia (XLT).⁶  However, in previous studies, we did not detect any obvious cytologic alterations in primary human WASp-deficient MKs grown in vitro.¹²  To investigate whether WASp might be involved in MK function under more physiologic conditions, we analyzed bone marrow femurs of WASp-deficient mice and their normal counterpart 129Sv mice by conventional histology, immunohistochemistry, and electron microscopy.

In WASp-deficient bone marrow, MKs were more numerous compared with the normal mouse marrow as observed by conventional histology (data not shown).

Epon-embedded sections of wild-type bone marrow MKs (toluidine blue staining) showed centrally located nuclei (Figure 1Ai arrows). In contrast, the nuclei were eccentrically located at the cell periphery of MKs in WASp-deficient bone marrows (Figure 1Aii arrows).

Immunohistochemistry confirmed the increased number and abnormal structure of WASp-deficient MKs (Figure 1Bi,ii). In addition, short cytoplasmic extensions resembling proplatelets were found in the vicinity of some mature MKs, and numerous deposits of specific immunostaining were scattered in the hematopoietic space of the bone marrow, suggesting that local platelet production had occurred (Figure 1Biv,vi). In contrast, control bone marrow did not display any background staining (Figure 1Biii,v).

Electron microscopy demonstrated that mature WASp-deficient MKs exhibited delineated platelet territories within their cytoplasm more often than in control mice (Table 1; Figure 2A). In addition, numerous MKs in the process of shedding and release of platelets were observed in WASp-deficient bone marrows (Table 1; Figure 2B), and this aspect could not be seen in wild-type marrows. Proplatelets and platelets were also observed in WASp-deficient bone marrow either in clusters or scattered in the hematopoietic space (Table 1; Figure 2C), and these images were virtually absent from our control mice. Naked MK nuclei were also found in the WASp-deficient bone marrow space (Table 1; Figure 2D), again indicating that cytoplasmic shedding had occurred in situ. Naked nuclei were not observed in EM control marrow sections, suggesting that these are rare events in 129Sv mice marrow in contrast to Wasp-deficient mice marrow.

Recently, we showed that proplatelet formation of human MKs is inhibited by CI, the main component of ECM in the bone marrow.⁴  We investigated whether adhesion to CI would affect proplatelet formation in wild-type and WASp-deficient MKs derived from lineage-depleted (lin^-) bone marrow. Cells harvested from 129Sv and WASp-deficient mice were cultured in the presence of TPO for 3 days, and plated for 2 hours on CI. Cells were then labeled by TRITC-phalloidin to visualize F-actin and VWF to confirm MK identity and discriminate between proplatelets and pseudopodal projections that could be due to cell spreading. First, we confirmed that as in WAS patients, WASp-deficient murine MKs formed proplatelets in liquid culture with the same efficiency as wild-type MKs. In addition, wild-type or WASp-deficient murine MKs seeded on CI produced only few MKs bearing proplatelets.

We previously demonstrated that CI mediates a very strong negative signal to proplatelet formation, and that a more discriminatory effect can be obtained by studying the response to ligation of the 2 main collagen receptors GPVI and α2β1 integrin.⁴  Wild-type and WASp-deficient MKs were thus plated on CVX¹⁴  or GFOGER peptide,¹⁵  2 high-affinity substrates for GPVI and α2β1 integrin, respectively. Adhesion to CVX mediated a very similar inhibition of proplatelet formation in wild-type and WASp-deficient MKs (Figure 3). Remarkably, we found that the proportion of WASp-deficient MKs forming VWF⁺ proplatelets on GFOGER peptide-coated surfaces increased significantly compared with wild-type-derived MKs (Figure 3A), and reached an even higher level than that observed in liquid culture. In addition, WASp-deficient MKs underwent striking morphologic alterations leading to the elaboration of long, branched proplatelets with diffuse actin staining, similar to those observed in liquid culture of wild-type MKs (Figure 3B). These findings indicate that inhibition of proplatelet formation in response to α2 integrin ligation is dependent on the presence of WASp.

WASp plays a crucial role in chemotactic migration and trafficking of several hematopoietic cell lineages.⁹  It has also been suggested that immunodeficiency in WAS might at least in part be related to migration defects of immune cells.¹⁸  To assess how migration of WASp-deficient MKs might be modulated by the bone marrow microenvironment including ECM and SDF-1 (the central chemokine in cytokine-independent regulation of thrombopoiesis),¹⁹  we tested the ability of wild-type and WASp-deficient MKs to migrate in vitro through a gradient of SDF-1 using ECM-coated transwell chambers. In the absence of SDF-1, only a few cells were able to migrate (Figure 4). When transwells were not coated, the migration efficiency of wild-type and WASp-deficient MKs was similar in the presence of SDF-1. In matrigel-coated chambers, addition of SDF-1 induced a higher migration of wild-type CD41⁺ MKs compared with that observed in CI-coated chambers (37.7% ± 3.8% and 10.3% ± 0.98%, respectively), at suboptimal doses of SDF-1 (30 ng/mL). Similar rates of cell migration were observed at 100 and 300 ng/mL. The migration efficiency of WASp-deficient MKs was significantly decreased on both matrigel-coated (Figure 4A) and CI-coated (Figure 4B) transwells. Thus, in the presence of CI or matrigel, WASp-deficient MKs have a marked defect in migration. Of importance, migration of WASp-deficient MKs was barely detectable on CI, suggesting a selective role of WASp on CI signaling. In combination with a loss of α2β1 integrin-mediated suppression of platelet production, these findings may explain the occurrence of ectopic and premature platelet shedding within the bone marrow microenvironment.

We hypothesized that alterations of the actin cytoskeleton may underlie the observed differences in the migration behavior of WASp-deficient MKs. Thus, we investigated cytoskeletal actin reorganization induced by adhesion to ECM components in wild-type and WASp-deficient MKs. Cells harvested from wildtype and WASp-deficient mice were cultured in the presence of TPO for 3 days and plated on CI, fibronectin, fibrinogen, matrigel, or PLL for 30 minutes, 1 hour, or 2 hours. Unexpectedly, on CI-coated slides, actin dotlike structures reminiscent of podosomes were noticeable at 30 minutes (Figure 5Ai) and were prominent in 59% ± 10% of MKs within 2 hours (Figure 5Aii,iv,D). The MK identity of cells was verified by staining for VWF (blue pseudocolor). Confocal sections revealed punctate actin structures that were restricted to the ventral MK surface (Figure 5Aii: section = 0.70 μm) and were not observed in upper sections (Figure 5Aiii: section = 1.40 μm). Punctate actin structures at the cell margin were surrounded by vinculin rings (green), which are characteristic of podosome organization (Figure 5Aiv). Of interest, podosomes were not observed when cells were seeded on PLL, matrigel, fibrinogen, or fibronectin-coated slides (data not shown). Thus, podosomes in murine MKs were not spontaneously assembled as in monocytic-derived cells and were specifically induced by CI, likely through its main receptors GPVI and/or α2β1 integrin.

We therefore sought to determine the respective roles of CI receptors in podosome assembly. Unlike primary human MKs,⁴  adhesion to CVX induced significant spreading, podosome formation, and stress fiber assembly (Figure 5Bi,ii,D). Adhesion to GFOGER peptide, the specific substrate of α2β1 integrin, induced podosome assembly to the same extent as CVX (Figure 5Biii,iv,D) independently of GPVI receptor. Furthermore, we found that podosome formation was not restricted to the mouse 129Sv background and was similarly induced by CI in C57Bl/6-derived MKs (Figure 5Ci,ii,iv,D). As noted in 129Sv-derived MKs, actin structures were discernible by phase contrast as dark dots at the leading edges and the tips of lamellae (Figure 5Ci inset), and their spatial distribution and average diameter (0.48 ± 0.06 μm, Figure 5Cii) resembled the phenotype noted previously for migrating macrophages and DCs. Adhesion to CVX (Figure 5Ciii,D) or GFOGER peptide (Figure 5Civ,D) induced also similar morphologic alterations to that observed in 129Sv-derived MKs. As illustrated in high magnification views (Figure 5Ciii-iv bottom panels), vinculin rings were evident mainly at the cell periphery.

WASp has been shown to be critical for podosome formation in other hematopoietic-cell types in both human and murine models and has been found to localize to the core of podosomes in DCs, macrophages, and transformed fibroblasts.⁸  Costaining of actin and WASp showed colocalization of WASp and actin in the core of podosomes, mainly at the cell periphery of wild-type 129Sv MKs adherent to CI, CVX, or GFOGER peptide (Figure 6Ai, ii, and iii, respectively). To further investigate the role of WASp in podosome assembly, we analyzed actin cytoskeletal reorganization of WASp-deficient MKs. WASp-deficient MKs adherent to CI, CVX, or GFOGER peptide were completely devoid of podosomes (Figure 6Bi, ii, and iii, respectively, and Figure 6C), and exhibited diffuse vinculin staining (data not shown). The inability of MKs to form podosomes was coupled with other morphologic defects. The number of wild-type MKs seeded on CVX and exhibiting stress fibers (61% ± 2.8%) was significantly decreased compared with WASp-deficient MKs (26% ± 9.2%). Altogether, these data suggest that WASp deficiency leads to a defective MK actin cytoarchitecture. Failure to assemble podosomes may compromise normal adhesive interactions with ECM components and chemokine-induced MK migration.

Microthrombocytopenia is an invariable characteristic of WASp-deficient disease in humans. The mechanisms of thrombocytopenia are not fully understood, but are thought predominantly to be due to accelerated platelet destruction in the spleen associated with an intrinsic defect of platelet production.⁶  WASp-deficient mice have only a moderate thrombocytopenia, which is not associated with reduced platelet size.¹³  Although splenectomy has not been performed in these animals to evaluate the role of platelet destruction in the mechanism of thrombocytopenia, it is possible that a defect in platelet biogenesis may be an important mechanism. Our data support this hypothesis by showing that WASp deficiency affects platelet production by disturbing the negative regulation of proplatelet formation mediated by CI through α2β1, and MK migration induced by SDF-1 in the presence of CI.

Previously, we did not observe significant cytologic differences at the ultrastructural level between normal and WASp-deficient human MKs grown in vitro. In addition, proplatelet formation occurs normally in liquid culture in the presence of TPO.¹²  As shown in this study, WASp-deficient murine MKs behaved similarly in vitro. However, in situ within the WASp-deficient mouse bone marrow, MKs were more numerous than in the controls, and were morphologically abnormal. This immediately suggested that the contribution of the bone marrow microenvironment is crucial for WASp-mediated effects on MKs. Moreover, proplatelet formation, true platelets, and naked MK nuclei were ectopically localized in the bone marrow of WASp-deficient mice. The increase in naked nuclei in the marrow could be explained both by the premature platelet production in the marrow and the defect in phagocytosis function of Wasp-deficient macrophages. Thus, WASp may be involved in the 2 main mechanisms, which suppress premature occurrence of proplatelet formation in the bone marrow microenvironment prior to platelet release into the blood stream (ie, MK migration and exit from the bone marrow² ) and inhibition of proplatelet formation by the bone marrow environment (ECM⁴  and stromal cells⁵ ).

The chemokine SDF-1 has been shown to participate in the regulation of platelet formation, especially in the absence of TPO. SDF-1 appears to play an important role in allowing the interaction of MKs with endothelial cells that regulate MK terminal differentiation.¹⁹  In addition, SDF-1 controls the migration of hematopoietic cells and their retention in the bone marrow.²⁰  It has been previously shown that WASp-deficient T cells and CD34⁺ cells have an altered migration in response to SDF-1.^21,22  In contrast, human WAS MKs derived in vitro from CD34⁺ cells have a normal migratory response to SDF-1.¹²  We found a similar result for WASp-deficient murine MKs in the absence of ECM. In contrast, when transwells were coated with a complex ECM (matrigel) or CI to mimic a more physiologic substratum, murine WASp-deficient MKs exhibited a marked defect in migration. The mechanism for defective migration under these conditions is not fully resolved, but these findings are reminiscent of abnormalities reported in other cell lineages including DCs.²³  As for DCs, we have demonstrated that primary wild-type murine MKs are able to assemble actin-rich classic podosome structures, which, interestingly, were selectively induced by adhesion to CI, and by engagement of either GPVI or α2β1 receptors. Ventral localization of podosomes⁸  and the condensation of vinculin²⁴  surrounding the actin core suggest that podosomes may play a role in the dynamics of cell adhesion. In contrast to DCs, podosome assembly was temporally restricted to mature MKs (days 3 and 4 of culture), which may tie in with previous observations that differentiation of primary murine²⁵  and human²⁶  MKs is accompanied by an increase of GPVI expression and functional activation. In human platelets, WASp is selectively tyrosine-phosphorylated by CI,²⁷  likely through GPVI receptor.²⁸  The detailed role of α2β1 integrin and GPVI in WASp activation and recruitment to podosomes remains to be investigated in murine MKs. However, as shown for other WASp-deficient cell types,⁸  we have demonstrated here that WASp-deficient MKs were completely devoid of podosome structures, further indicating that WASp is a nonredundant regulator of podosome assembly in hematopoietic cells.

In accordance with our previous findings in primary human MKs,⁴  we have found that CI inhibits proplatelet formation in murine MKs. Surprisingly, we also found that WASp participates in suppression of proplatelet formation in MKs adhering to CI through a mechanism involving engagement of α2β1 integrin. Microtubule-based forces represent the main motor for proplatelet elongation, whereas a dynamic F-actin network is involved in the amplification and bifurcation of proplatelet ends to increase the yield of released platelets.²⁹  WASp-induced podosome assembly may partly account for the negative regulation of proplatelet formation. In accordance with their ventral surface localization, WASp-induced podosomes provide localized adhesive contacts, which may generate mechanical constraints to the extension of proplatelets.

This study supports the notion that WASp participates as an intrinsic negative effector of megakaryocytopoiesis in vivo and suppresses the occurrence of premature platelet shedding within the bone marrow compartment. Whether these findings are applicable to human disease requires further investigation, particularly as there are presently several strands of evidence suggesting that human⁴  and murine megakaryopoiesis differ, such as in the reorganization of the actin cytoskeleton in response to CI, as shown in this study. These studies will be facilitated by the development of reagents that enable the selective down-regulation of WASp expression in normal hematopoietic cells, as recently demonstrated for human DCs.²⁹  However, this study paves the way toward a better comprehension of the mechanisms of microthrombocytopenia in WAS patients and defines a novel role for WASp in suppression of premature platelet shedding.

We are grateful to Dr Stephan Linder (Institut fur Prophylaxe und Epidemiologie der Kreislaufkrankheiten, Munich, Germany) for discussions about this work.

We would like to thank P. Ardouin and A. Rouches (Institut Gustave Roussy) for animal facilities.