Essential role of zyxin in platelet biogenesis and glycoprotein Ib-IX surface expression

Platelets are generated from the cytoplasm of megakaryocytes (MKs) via actin cytoskeleton reorganization. Zyxin is a focal adhesion protein and wildly expressed in eukaryotes to regulate actin remodeling. Zyxin is upregulated during megakaryocytic differentiation; however, the role of zyxin in thrombopoiesis is unknown. Here we show that zyxin ablation results in profound macrothrombocytopenia. Platelet lifespan and thrombopoietin level were comparable between wild-type and zyxin-deficient mice, but MK maturation, demarcation membrane system formation, and proplatelet generation were obviously impaired in the absence of zyxin. Differential proteomic analysis of proteins associated with macrothrombocytopenia revealed that glycoprotein (GP) Ib-IX was significantly reduced in zyxin-deficient platelets. Moreover, GPIb-IX surface level was decreased in zyxin-deficient MKs. Knockdown of zyxin in a human megakaryocytic cell line resulted in GPIbα degradation by lysosomes leading to the reduction of GPIb-IX surface level. We further found that zyxin was colocalized with vasodilator-stimulated phosphoprotein (VASP), and loss of zyxin caused diffuse distribution of VASP and actin cytoskeleton disorganization in both platelets and MKs. Reconstitution of zyxin with VASP binding site in zyxin-deficient hematopoietic progenitor cell-derived MKs restored GPIb-IX surface expression and proplatelet generation. Taken together, our findings identify zyxin as a regulator of platelet biogenesis and GPIb-IX surface expression through VASP-mediated cytoskeleton reorganization, suggesting possible pathogenesis of macrothrombocytopenia.


INTRODUCTION
Platelets, the central regulator of thrombosis and hemostasis, are generated from the cytoplasm of megakaryocytes (MKs). In bone marrow (BM), hematopoietic stem cells differentiate into MKs in response to thrombopoietin (TPO) [1]. Immature MKs undergo endomitosis to increase their size and ploidy and develop a demarcation membrane system (DMS) that forms the plasma membrane of future platelets [2]. In the meantime, MK progenitors migrate from BM osteoblastic niche to the vascular niche [3,4], in which proplatelets are formed and released from the matured MKs into the bloodstream. Finally, platelets are shed from proplatelets in the bloodstream.
The actin cytoskeleton plays an important role in platelet biogenesis [5]. Actin filaments localizing around the DMS network generate mechanical forces to initiate and mediate DMS formation [6]. Actin cytoskeleton reorganization guides the bending and bifurcating of the proplatelet shafts to increase the number of proplatelet ends [7]. Therefore, human mutations and genetargeted mice deficient in the components of the actin cytoskeleton, such as non-muscle myosin heavy chain IIA (NMMHC-IIA), α-actinin, filamin A, tropomyosin 4, diaphanous-related formin 1 (DIAPH1), and tropomodulin 3, have shown thrombocytopenia due to abnormal platelet generation [8][9][10][11][12][13]. However, the molecular mechanisms underlying the complex process still remain elusive.
Zyxin is a focal adhesion protein wildly expressed in tissues [14]. Zyxin regulates actin cytoskeleton remodeling by providing docking sites for actin regulatory proteins [15][16][17][18]. The C-terminal of zyxin contains three LIM domains, which are essential for its localization to focal adhesions [18]. The N-terminus of zyxin contains the binding sites for actin filament crosslinker α-actinin [19], and displays the four proline-rich ActA repeats that can interact with the actin assembly modulator enabled (Ena)/vasodilator-stimulated phosphoprotein (VASP) proteins [20]. Mutations in α-actinin cause macrothrombocytopenia by inducing multiple defects of proplatelet formation [9]. We recently demonstrated that zyxin interacts with NMMHC-IIA [21], whose coding gene MYH9 has the most macrothrombocytopeniacausing mutations [8]. Zyxin has been shown to express in platelets [14] and was found to be upregulated during megakaryocytic differentiation [22], whereas it is unclear whether zyxin plays a role in platelet biogenesis.

RESULTS
Zyx −/− mice display macrothrombocytopenia Zyxin has been demonstrated to express in both human and murine platelets [14,23]. In order to investigate the role of zyxin in platelet biogenesis, we first performed a hematologic analysis in Zyx −/− mice. Interestingly, we found that Zyx −/− mice displayed severe macrothrombocytopenia, with a 65% reduction of platelet count and a 68% increase of platelet volume compared with wildtype (WT) mice (Fig. 1A, B). The Zyx −/− platelets presented normal shape, but obviously enlarged size (Fig. 1C, D). In contrast, the number and morphology of leukocytes and red blood cells were not altered in Zyx −/mice (Table S1).
To investigate whether the thrombocytopenia is due to the enhanced platelet turnover, platelet lifespan was measured in Zyx −/− mice. We found that platelet lifespan was not obviously altered in Zyx −/− mice (t 1/2 = 49 h for WT vs t 1/2 = 51 h for Zyx −/− mice) (Fig. 1E). In addition, the percentage of reticulated platelets was not altered, while the reticulated platelet number was markedly reduced in Zyx −/− mice (Fig. S1). Since TPO is essential for MK proliferation and maturation, serum TPO level was examined. There were comparable levels of serum TPO between Zyx −/− and WT mice (Fig. 1F). These data show that the macrothrombocytopenia in Zyx −/− mice is not caused by accelerated platelet clearance or impaired TPO generation.

Zyxin deficiency impairs MK maturation, DMS development, and proplatelet generation
To investigate the mechanism for zyxin deficiency-mediated macrothrombocytopenia, we first characterized the megakaryopoiesis in Zyx −/− mice. We found that, compared with WT mice, Zyx −/− mice showed a 50% increase in MK number of the BM sections (16 MKs per visual field for WT vs 24 for Zyx −/− mice, Fig. 2A). Spleen is also an important hematopoietic organ in mice. We found that there was an 83% increase in splenic MK number (6 MKs per visual field for WT vs 11 for Zyx −/− mice, Fig. 2A). Therefore, the thrombocytopenia in Zyx −/− mice does not result from a lack of MKs. We next examined the effect of zyxin deletion on MK differentiation in vitro by culturing mouse fetal liver hematopoietic progenitor cells (FL HPCs). The percentages of CD41 positive (CD41 + ) (Fig. 2B) and CD61 positive (CD61 + ) (Fig. S2) cells were comparable between WT and Zyx −/− FL HPC cultures. We further cultured bone marrow hematopoietic progenitor cells (BM HPCs) from Zyx −/− and WT mice. There was also not difference in the percentages of CD41 + and CD61 + cells between the two cultures (Fig. S3). These data suggest that zyxin is not required for MK differentiation.
We next investigated the effect of zyxin deficiency on MK maturation and proplatelet production. Giemsa-stained BM smears showed more Stage I MKs (immature MKs) and fewer Stage II and III MKs (mature MKs) in Zyx −/− mice (Fig. 2C). The ultrastructure of Stage III MKs from BM aspirates was further analyzed by transmission electron microscopy (TEM). WT MKs showed well-organized DMS, but Zyx −/− MKs showed disorganized DMS and a lot of vacuoles (Fig. 2D), suggesting the defective development of DMS in Zyx −/− MKs.
Finally, platelet generation was examined in Zyx −/− FL HPCderived MKs. Culture-derived MKs started to release CD41positive (CD41 + ) platelet-sized particles from day 5 (Fig. 2E). On day 5 and day 6, there were 13.6% and 11.8% CD41 + plateletsized particles from WT MKs, while only 2.6% and 2.9% from Zyx −/− MKs (Fig. 2E). Furthermore, Fig. 2F showed that proplatelet protrusions from Zyx −/− MKs were obviously reduced compared with that from WT MKs. Taken together, these data show that zyxin deletion impairs MK maturation, DMS development, and proplatelet generation. and Zyx −/− (n = 17) mice. Means are indicated by horizontal lines. ***P < 0.001 compared with WT mice by Mann-Whitney test in (A) and unpaired Student's t-test with Welch correction in (B). C Representative images of Wright-Giemsa stained blood smears from WT and Zyx −/− mice (original magnification × 1,000). Scale bar: 10 μm. Platelets were indicated with arrows. D Representative transmission electron microscopy (TEM) images of WT and Zyx −/− platelets (original magnification ×13,500). Scale bar: 1 μm. Images were obtained from five mice in each genotype in (C) and (D). E WT and Zyx −/− mice were intravenously injected with NHS-biotin, and peripheral blood was taken at the indicated time points after injection. The percentage of biotinylated platelets was determined by flow cytometry; n = 10 mice per genotype. Data are expressed as means ± SD. F Serum TPO levels in WT and Zyx −/− mice; n = 7 mice per genotype. Means are indicated by horizontal lines.
We further confirmed these findings by detecting platelet function. Consistently, botrocetin-induced platelet aggregation, which is mediated by the interaction of GPIbα and von Willebrand factor, was obviously reduced in Zyx −/− platelets (Fig. S5A). Furthermore, Zyx −/− platelets tend to disaggregate in response to 0.05 U/mL thrombin, consistent with the requirement of GPIbα in low dose thrombin-induced platelet activation [29]. However, zyxin ablation did not affect platelet aggregation induced by relatively high doses of thrombin (0.1 U/mL and 0.2 U/mL) (Fig.  S5B). In addition, there was no significant difference in ADP-or U46619-induced platelet aggregation between WT and Zyx −/− platelets (Fig. S5C, D). These data suggest that zyxin deficiency impairs only GPIb-IX-dependent platelet function.
We also examined the expression levels of GPIb-IX/GPIbα in MKs. The mean fluorescence intensities (MFI) of GPIbα, GPIbβ, and GPIX were significantly decreased in Zyx −/− MKs in the immunostained BM sections (Fig. 3D). The GPIbα−positive (GPIbα + ) cells were markedly reduced in Zyx −/− FL HPC-derived MKs (Fig. 3E). Moreover, the GPIbα + platelet-sized particles were also reduced (Fig. 3F). These data indicate that zyxin deficiency results in the reduction of GPIb-IX in both platelets and MKs.
Zyxin knockdown results in lysosome-mediated GPIbα degradation in human megakaryocytic Dami cells GPIb-IX complex is assembled in the endoplasmic reticulum, transported into Golgi, and expressed on the plasma membrane [30]. The presence of all three subunits and proper transportation are required for the efficient surface expression of the whole complex [31]. To explore the reason for the defective GPIb-IX expression in Zyx −/− platelets, we first examined whether zyxin deletion reduced transcription of these three genes. The quantitative real-time PCR (qRT-PCR) results showed that the mRNA levels of the three subunits did not decrease in Zyx −/− platelets (Fig. 4A), excluding the possibility of defective gene transcription. Further analysis of GPIbα by Western blot showed additional anti-GPIbα antibody-positive bands (60 kDa and 25 kDa) in Zyx −/− platelets (Fig. 4B), suggesting that GPIb-IX complex may be degraded in Zyx −/− platelets.
We next used a human megakaryocytic cell line, Dami cells, which express the GPIb-IX complex on the cell surface [32], to explore the role of zyxin in GPIb-IX expression. The small interfering RNAs (siRNAs) targeting zyxin were transfected into Dami cells to knockdown zyxin. We first verified that the zyxin siRNAs significantly reduced the mRNA level of zyxin but not that of GPIbα, GPIbβ, or GPIX (Fig. S6). Zyxin siRNAs reduced zyxin expression, and in the meantime, the levels of GPIbα and GPIX were obviously reduced (Fig. 4C). Moreover, GPIbα and GPIX surface levels were significantly decreased (Fig. 4D). These data, consistent with the data from Zyx −/− platelets and MKs, reveal that knockdown of zyxin reduces GPIb-IX level in human megakaryocytic Dami cells.
GPIbα was reported to be degraded through a lysosomedependent manner [30]. We treated Dami cells with lysosomal inhibitors (leupeptin + NH 4 Cl) and a proteasome inhibitor (MG-132) after zyxin siRNA transfection. As shown in Fig. 4E and F, inhibition of lysosomes but not proteasomes restored GPIbα surface expression as well as the total level of full-length GPIbα. We further used another lysosomal inhibitor (bafilomycin A1) to verify this. Bafilomycin A1 indeed rescued zyxin knockdowninduced GPIbα reduction (Fig. S7). These data demonstrate that zyxin deficiency results in lysosome-mediated GPIbα degradation in the human megakaryocytic cell line.

Zyxin deficiency enhances microtubular ring in platelets
Microtubules play critical roles in the extension of proplatelets [5,7]. The proplatelets from a mouse model of BSS contained twice as many tubulin fibers in the marginal microtubular ring [33]. Thus, we also examined the microtubular system in Zyx −/− platelets. First, we constructed a mouse model of BSS by deletion of GPIbα (Fig. S8). The GPIbα-deficient (Gp1ba -/-) mice presented the typical macrothrombocytopenia phenotype (Table S2). The same as the finding in Gp1ba −/− platelets, a brighter and thicker marginal band was found in Zyx −/− platelets compared with that in WT platelets (Fig. 5A). These data suggest that the phenomenon of Zyx −/− platelets is similar to that of the platelets from a mouse model of BSS [32].

Zyxin deficiency alters actin cytoskeleton organization
It has been implied that actin cytoskeleton organization is essential for appropriate insertion or removal of membrane proteins [34]. And alteration of actin cytoskeleton reorganization impairs proplatelet formation leading to macrothrombocytopenia [5]. Zyxin regulates actin assembly [16][17][18]. Therefore, zyxin ablation may alter actin cytoskeleton organization, contributing to defective GPIb-IX complex expression and proplatelet generation. To test this hypothesis, F-actin organization was visualized by phalloidin staining in platelets spreading on the fibrinogen surface. The stress fibers were bundled along the sides of the WT platelets; in contrast, the stress fibers were disseminated in the whole cytoplasm of Zyx −/− platelets (Fig. 5B). We further detected actin cytoskeleton in MKs from WT and Zyx −/− mice. In MKs spreading on immobilized collagen, only 41% of WT MKs displayed an organized F-actin network along collagen fibers, Fig. 2 Normal MK differentiation but impaired maturation, DMS development, and proplatelet formation in Zyx −/− MKs. A MK counts in mouse BM and spleen. Left panels, representative confocal images of MKs in femoral BM and spleen sections immunostained with anti-CD41 antibody. MKs (CD41) are in green, and nuclei are in blue. Original magnification × 200 for BM sections and × 100 for spleen sections. Scale bar: 100 μm. Right panel, the quantification of MK numbers per visual field; n = 15 visual fields from five mice per genotype. Data are expressed as means ± SD. B Percentage of CD41 + cells differentiated from WT and Zyx −/− FL HPCs on day 4 was analyzed by flow cytometry. Data are from five independent experiments. C Upper panels, representative microscopy images of MKs in BM smears stained with Wright-Giemsa stain from WT and Zyx −/− mice (original magnification × 400). Scale bar: 20 μm. Lower panel, the quantification of different stages of MKs using BM smears from WT (n = 4) and Zyx −/− mice (n = 7). No less than 100 MKs were analyzed for each mouse. D Representative TEM images of MKs isolated from WT and Zyx −/− mouse BM (original magnification, ×3,400 for left panels, and ×13,500 for right panels). Images were obtained from five mice in each genotype. Scale bar: 1 μm. E Percentage of CD41 + platelet-sized particles released from culture-derived MKs was analyzed by flow cytometry. Data are from four independent experiments. F Representative microscopy images of proplatelet formation from culture-derived MKs (E) on day 4 (original magnification ×200). Scale bar, 20 μm. Proplatelet protrusions were indicated with arrows. Means are indicated by horizontal lines in (B), (C), and (E). **P < 0.01, ***P < 0.001, compared with WT mice by unpaired Student's t-test in (A) and (B), and by two-way ANOVA followed by Bonferroni's post hoc test in (C) and (E). NS, not significant.
whereas 75% of Zyx −/− MKs presented the phenomena (Fig. 5C, D). Furthermore, for MKs spreading on immobilized fibrinogen, there was 61% of WT MKs displaying the lamellipodia, whereas this was only seen in about 26% of Zyx −/− MKs (Fig. 5E, F). The spread areas were not affected by zyxin deficiency on both collagen and fibrinogen surfaces (Fig. S9). These data indicate that actin cytoskeleton organization is altered in Zyx −/− platelets and MKs.

Zyxin deficiency results in VASP mislocalization
We next explored the mechanism for zyxin in regulating actin cytoskeleton organization. α-Actinin and VASP are two important zyxin binding partners [19,20]. We recently identified NMMHC-IIA as a zyxin binding protein [21]. The three proteins all exist in platelets and play important roles in actin cytoskeleton reorganization. Zyxin deletion did not affect the expression of the three proteins (Fig. 3A, Fig. S10A). In the absence of zyxin, the distributions of α-actinin and NMMHC-IIA were not obviously altered in platelets spreading on the fibrinogen surface (Fig. S10B). Moreover, zyxin deficiency did not affect the phosphorylation of NMMHC-IIA at Ser1916 and Ser1943, which is related to the activity of NMMIIA [35], in the platelets either under basal condition or stimulated with various agonists (Fig. S11). In contrast, we found that zyxin colocalized with VASP ( Fig. 6A) in WT platelets, and the distribution of VASP was obviously disrupted in Zyx −/− platelets compared with that in WT platelets (Fig. 6B), suggesting the essential role for zyxin in VASP localization. Next, we verified this finding in MKs. Consistent with the finding in platelets, VASP was localized at focal adhesions (Fig. 6C), which is the same as zyxin in WT MKs (Fig. S12). Without zyxin, VASP was distributed all over the cells in Zyx −/− MKs (Fig. 6C). These data suggest that zyxin is essential for subcellular VASP localization in platelets and MKs.
Next, we explored the mechanism for zyxin in regulating GPIb-IX surface expression. As known, VASP is an actin regulatory protein. Filamin A that crosslinks actin filaments associates with the cytoplasmic domain of GPIbα anchoring the GPIb-IX complex to the membrane skeleton for surface expression. Therefore, we hypothesized that there might be an association between VASP and filamin A or GPIbα. Indeed, we find that VASP could be coimmunoprecipitated with both filamin A and GPIbα (Fig. 6D). These findings suggest the possibility that zyxin, through VASP, regulates GPIb-IX surface expression.
The interaction of zyxin with VASP is required for GPIbα surface expression and platelet production We further investigated whether zyxin regulates GPIb-IX expression and platelet production through the interaction with VASP. The binding sites for α-actinin and NMMHC-IIA are located in the N-terminal 42 amino acids of zyxin [19,21], and the binding site for VASP is in the four proline-rich ActA repeats in zyxin (Fig. 7A) [20]. In order to distinguish which site is essential for GPIb-IX expression and platelet production, we constructed zyxin mutants lacking binding sites for α-actinin and NMMHC-IIA (Zyx 43-564 ), and VASP (Zyx 4F > A ) (Fig. 7A). Zyx -/− FL HPCs were infected with adenoviruses expressing WT zyxin and zyxin mutants, and similar expression levels of these genes were verified with Western blot (Fig. 7B). The expressions of WT zyxin (Zyx 1-564 ) and zyxin mutants (Zyx  and Zyx 4F > A ) in Zyx −/− FL HPCs did not affect the percentage of CD41 + cells compared with that of vector control (Fig. 7C), suggesting that these genes did not affect the MK Fig. 4 Knockdown of zyxin results in lysosome-mediated GPIbα degradation. A mRNA levels of GPIbα, GPIbβ, and GPIX in WT and Zyx −/− platelets. mRNA expression was analyzed by qRT-PCR and determined by a ratio relative to the control GAPDH. The data were expressed as the ratio relative to WT; n = 5 mice per genotype. B Western blot analysis of GPIbα protein in WT and Zyx −/− platelets. Protein concentration was adjusted to the same level between WT and Zyx −/− platelet lysates. Blots are representative of five independent experiments. C, D Dami cells were transfected with siRNAs targeting zyxin (siZYX-1, and -2) or negative control siRNA (control). The expression of GPIbα and GPIX in Dami cells was analyzed by Western blot; the blots are representative of five independent experiments (C). The surface level of GPIbα and GPIX was analyzed by flow cytometry (D). E, F Dami cells were treated with or without 10 μg/mL leupeptin plus 10 mM NH 4 Cl (Leu + NH 4 Cl) and MG-132 (100 nM) for 12 h after zyxin siRNA (siZYX-1) transfection. The total expression of GPIbα was analyzed with Western blot; the blots are representative of five independent experiments (E). The surface level of GPIbα was analyzed by flow cytometry (F). Data are from five independent experiments in (C-F). Means are indicated by horizontal lines in (A) and (C-F). *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA followed by Dunnett's post hoc test in (C-F). NS, not significant.
differentiation. However, WT zyxin and Zyx 43-564 significantly increased the percentage of GPIbα + cells; in contrast, Zyx 4F > A did not have the effect (Fig. 7D), indicating that WT and Zyx 43-564 but not Zyx 4F > A could rescue GPIbα surface expression in Zyx −/− FL HPC-derived MKs. Consistently, WT zyxin and Zyx 43-564 but not Zyx 4F > A enhanced CD41 + platelet-sized particles and GPIbα + platelet-sized particles from Zyx −/− FL HPC-derived MKs (Fig. 7E, F). Moreover, Fig. 7G showed that the expression of WT zyxin and Zyx 43-564 but not Zyx 4F > A promoted proplatelet formation. These data suggest that the interaction of zyxin with VASP is essential for GPIbα surface expression and platelet production.

DISCUSSION
In this study, we report for the first time that zyxin is essential for platelet biogenesis and GPIb-IX surface expression. Our data show that: (1) Zyx −/− mice display macrothrombocytopenia, (2) zyxin is essential for DMS formation and proplatelet production, (3) zyxin ablation results in GPIbα degradation leading to reduction of GPIb-IX surface expression and GPIb-IX-dependent platelet function, (4) deletion of zyxin alters actin cytoskeleton organization, and (5) the interaction of zyxin with VASP is required for proplatelet production and GPIb-IX surface expression.
Defective proplatelet formation has been reported in more than ten forms of inherited thrombocytopenia [36]. Most of these diseases are macrothrombocytopenia caused by mutations in genes encoding for components of the actomyosin cytoskeleton or microtubular system [36]. Reorganization of the actin cytoskeleton is crucial for DMS formation or extension and release of proplatelets [6,7]. Therefore, defect in genes associated with the actomyosin cytoskeleton, such as MYH9 and ACTN1, results in alteration of proplatelet formation [8][9][10][11]. Microtubules composed of αand β-tubulin are the driving force of proplatelet elongation [5,7]. Therefore, defects in TUBB1 induce abnormalities of microtubule function, affecting proplatelet formation [26]. In line with the reported macrothrombocytopenia, we found that the absence of zyxin resulted in actin cytoskeleton disorganization and enhanced tubulin filaments in the marginal band. Thus, our data support that zyxin deficiency alters the actin cytoskeleton and microtubular system resulting in defective proplatelet formation and macrothrombocytopenia.
Zyxin is essential for actin filament remodeling. Our data showed that zyxin deficiency affects F-actin organization. Without zyxin, VASP is disseminated in the whole platelets. VASP is capable of bundling actin filaments and inducing actin polymerization [37,38]. The dislocated VASP may induce actin polymerization in altered places, leading to altered organization of actin skeleton in Zyx −/− platelets and MKs. Therefore, it is reasonable to find that the actin filaments were disorganized in Zyx −/− platelets and MKs, resulting in defective proplatelet formation. Consistently, reconstitution of zyxin lacking VASP binding site in Zyx −/− FL HPCs did not restore platelet production.
Interestingly, we found that zyxin deficiency diminished GPIb-IX surface expression both in Zyx −/− platelets and MKs. These findings suggest that the molecular mechanism of zyxin deficiency-induced macrothrombocytopenia may be similar to that of BSS. GPIb-IX complex is assembled in the endoplasmic reticulum and traffics to the plasma membrane for surface expression [30]. Filamin A that binds to actin associates with the cytoplasmic domain of GPIbα anchoring the GPIb-IX complex for surface expression. One of the roles of actin is for membrane trafficking [39]. Therefore, actin organization might be critical for GPIb-IX trafficking and surface expression. In the current study, the data showed that VSAP localized with zyxin, and the interaction of zyxin with VASP is required for GPIb-IX surface expression. Thus, it is most likely that zyxin through VASP-mediated actin reorganization regulates GPIb-IX trafficking and surface expression. According to this theory, loss of zyxin results in actin cytoskeleton disorganization, which impairs GPIb-IX trafficking to the membrane and incurs GPIbα degradation. In support of this, GPIbα degradation bands were observed in Zyx −/− platelets, and knockdown of zyxin in Dami cells results in lysosome-mediated degradation of GPIbα. GPIb-IX complex, the receptor for the von Willebrand factor, plays an important role in thrombosis and hemostasis. This is the first time we have disclosed that zyxin regulates platelet function through regulating the GPIb-IX surface expression.
We selected a human megakaryocytic cell line, Dami cells, to demonstrate our findings. Consistent with the findings with Zyx −/− mouse platelets and MKs, knockdown of zyxin also reduced GPIb-IX surface level in Dami cells. These data provide important evidence that deletion of zyxin in human cells causes the same effects as it does in mice. We found that Zyx −/− platelets showed brighter and thicker marginal microtubular rings. This is consistent with the findings that microtubules were obviously increased in GPIbα and filamin A null platelets, which also showed defective GPIb-IX surface expression [33,40]. Furthermore, increased microtubules were observed in giant platelets from patients with May-Hegglin anomaly [41]. These findings suggest the possible relevance of GPIbα to the microtubular system. Future study is needed to disclose the relationship between them.
Therefore, these findings delineate the mechanism of zyxin in platelet biogenesis and GPIb-IX expression. In the absence of zyxin, (1) VASP was mislocalized, leading to disorganization of actin cytoskeleton, (2) GPIb-IX could not be properly transported to the plasma membrane for surface expression, (3) microtubule filaments were elevated in the marginal ring. These factors independently or mutually promoting impair proplatelet formation which results in macrothrombocytopenia.
The relevance of zyxin with diseases has been reported [42,43]. Here we show that the absence of zyxin incurs macrothrombocytopenia. Macrothrombocytopenia could cause severe or lifethreatening hemorrhage. However, due to the pathogenic mechanism is not totally understood, many patients with macrothrombocytopenia still have not been diagnosed or were misdiagnosed [36]. In this sense, the current study not only identifies the role of zyxin in the pathogenesis of macrothrombocytopenia, more importantly, but also suggests that defect of zyxin or other cytoskeleton proteins may have the potential to incur macrothrombocytopenia in humans as well. Thus, the finding expands our horizons in finding novel macrothrombocytopeniacausing genes.
In conclusion, our study demonstrates that zyxin ablation results in macrothrombocytopenia in mice. Zyxin, through VASPmediated cytoskeleton reorganization, regulates proplatelet formation and GPIb-IX surface expression. Our findings help to understand the mechanism of platelet biogenesis and suggest possible pathogenesis of macrothrombocytopenia.
The Gp1ba −/− mouse model was established by CRISPR/Cas9 genome editing technology on a C57BL/6 J background to induce two doublestranded breaks flanking exon 2 of Gp1ba. The optimized sgRNAs were constructed in the pT7-sgRNA plasmid backbone, and then were in vitro transcribed. The sequences of two independent guide RNAs targeting exon 2 of the Gp1ba gene were gRNA1 5'-TCTCACAGTTTACTTCCAGC −3' and gRNA2 5'-TATTGGGCACAGTGGGCATT −3'. The in vitro-transcribed Cas9 mRNA together with two sgRNAs were microinjected into the cytoplasm of C57BL/6 J zygote. In total, 145 healthy zygotes were transferred to pseudopregnant recipients and allowed to develop to term. Genotyping was performed by PCR amplification across the targeted Fig. 7 The interaction of zyxin with VASP is required for zyxin-mediated GPIbα expression and platelet production. A Cartoons for the constructs of zyxin mutants lacking α-actinin and NMMHC-IIA (Zyx 43-564 ), and VASP (Zyx 4F > A ) binding sites. NES, nuclear export signals. region (F: 5′-AGAAGCTCTGTTCCTCCAAAGGAC-3′, R: 5′-GGTAGTAGTGAC-CATGTAGCCTGAC −3′) to screen the born 15 pups. In total, ten founder mice were initially established and confirmed with the right PCR product (597 bp). The founder mice were further bred to WT C57BL/6 J mice to generate heterozygous Gp1ba-knockout mice and then were intercrossed to generate homozygous Gp1ba −/− mice.
Mice were 6-12 weeks old, and experiments included balanced groups of male and female mice. All animal experiments complied with the regulatory standards of and were approved by the Ethics Committee of the First Affiliated Hospital of Soochow University.

Platelet and MK imaging
For blood smear, the murine whole blood from postorbital veins anticoagulated with 1/7 volume of ACD was made into monolayer cells on a glass slide. For BM smear, the femoral BM was harvested and smeared on a glass slide with fetal bovine serum added in advance. Blood and BM smears were then stained with Wright-Giemsa Stain solution and examined by optical microscopy.
For TEM analysis of platelets, washed mouse platelets were centrifuged at 600 × g for 2 min and then fixed in 2.5% glutaraldehyde at 4°C overnight. For TEM analysis of MKs, femora and tibias of mice were isolated and the BM was flushed out by 0.01 M phosphate-buffered saline (PBS) and then centrifuged at 300 × g for 2 min. The platelets or BM pellets were washed with PBS, postfixed in 1.0% osmium tetroxide for 1 h, gradually dehydrated using acetone, and then stained with saturated uranyl acetate. The samples were infiltrated, embedded with resin, and polymerized. Finally, ultrathin sections were observed with a transmission electron microscope (Philips CM 120, Eindhoven, The Netherlands).

Platelet aggregation
Platelets in PRP or washed platelets (3 × 10 8 /mL) were stimulated with different concentrations of different agonists. Platelet aggregation was recorded in a Chrono-Log lumi-aggregometer at a stirring speed of 1200 rpm at 37°C. Platelet aggregation was monitored continuously over 10 min.

Platelet lifespan analysis
WT and Zyx −/− mice were intravenously injected with 600 μg NHS-biotin in buffer containing 140 mM NaCl and 10% DMSO. Whole blood was collected by capillary tube from the retro-orbital venous plexus at various time points and mixed with ACD. Whole blood was stained by PEconjugated anti-mouse CD41 antibody and FITC-conjugated streptavidin at room temperature (RT) for 1 h. The percentage of biotinylated platelets was determined by FC 500 flow cytometer (Beckman Coulter, Miami, FL, USA).

Serum TPO measurement
Mouse blood was collected from postorbital veins and then incubated at 37°C for 1 h and centrifuged to obtain the serum. Mouse serum TPO levels were measured according to the manufacturer's protocol. Briefly, 50 μL serum was diluted with 50 μL assay buffer and added to the 96-well plate which has been coated with anti-mouse TPO antibody. After incubation, TPO was detected with biotinylated anti-TPO antibody and HRPconjugated streptavidin. The plate was read at 450 nm in a Variskan Flash spectral scanning multimode reader (Thermo Fisher Scientific).
For MK counting in mouse spleens, splenic cryosections were fixed in ice-cold acetone and then blocked in 5% BSA/PBS at RT for 1 h. APCconjugated anti-mouse CD41 (10 μg/mL) was incubated at RT for 2 h. Cell nuclei were stained with 5 μg/mL DAPI. Samples were visualized with a Leica TCS SP8 confocal microscope.

In vitro differentiation of MKs and platelet production analysis
Female WT and Zyx −/− mice were sacrificed on day 14.5 of pregnancy. Fetal livers were harvested and single-cell suspension was prepared as previously described [45]. For isolation of HPCs from mouse BM cells, the femora and tibia of 6-8 weeks old mice were isolated and the BM was flushed by EasySep buffer and homogenized as previously described [45]. FL and BM HPCs were isolated by a mouse HPC isolation kit based on the manufacturer's protocol. The isolated HPCs (3 × 10 5 /mL) were cultured in Stemspan TM SFEM in the presence of 1% penicillin, 1% streptomycin, and 20 ng/mL recombinant murine TPO at 37°C for 6 days.
The cell suspension was collected from day 3 to 6 and centrifuged at 300 × g for 5 min to separate culture-derived MKs (pellet) and plateletsized particles (supernatant). The cell pellet was washed with PBS and resuspended in cell staining buffer. Non-specific binding was blocked by 20 μg/mL 2.4G2 at RT for 10 min. Then cells were labeled with APCconjugated rat anti-mouse CD41 antibody (MWReg30, 2 μg/mL) and FITC-conjugated rat anti-mouse GPIbα antibody (Xia.G5, 1:10) at RT for 20 min. After being washed with PBS, cells were labeled with 5 μg/mL PI at RT for 10 min and then measured by flow cytometer. The percentage of CD41 + and GPIbα + on cells was analyzed in the PI − population. The supernatant was further centrifuged at 3,500 rpm for 2 min, and the pelleted platelet-sized particles were washed with CGS resuspended in cell staining buffer, and labeled with APC-conjugated rat anti-mouse CD41 antibody (MWReg30, 2 μg/mL) and FITC-conjugated rat anti-mouse GPIbα antibody (Xia.G5, 1:10) at RT for 20 min. After washed with CGS buffer, platelet-sized particles were stained with 5 μg/mL PI at RT for 10 min and then washed and measured by flow cytometer. The percentage of CD41 + and GPIbα + on platelet-sized particles was analyzed in the PI − population.

Proteomic sample preparation
Washed platelets (1 × 10 8 ) from WT and Zyx −/− mice were lysed with 100 μL lysis buffer (0.05 M NH 4 HCO 3 , 2% sodium deoxycholate, 0.025 M NaCl). After centrifuged at 14,000 × g for 10 min to remove cell debris, protein concentration was measured by BCA kit and adjusted to the same level by lysis buffer. DTT (2 mM) was added to the platelet lysates and incubated at 95°C for 3 min. Proteins were hydrolyzed by 2.5 μg trypsin (V528A, Promega, Madison, WI, USA) at 37°C for 4 h and further hydrolyzed by the addition of 100 μL lysis buffer and 2.5 μg trypsin at 37°C overnight. The reaction was stopped by 1% formic acid. After centrifuged at 14,000 × g at 4°C for 15 min, supernatant was collected and lyophilized. The differential proteomic analysis was conducted by Human Phenome Institute (Fudan University, Shanghai).

Western blot
For detection of mouse platelet proteins, washed WT mouse platelets (3 × 10 8 /mL) and Zyx −/− platelets (1.5 × 10 8 /mL) were lysed with an equal volume of 2 × lysis buffer (100 mM Tris, pH 7.4, 2% Triton X-100, 20 mM MgCl 2 , 300 mM NaCl) containing phenylmethylsulfonyl fluoride (1 mM) and protease inhibitor cocktail on ice for 30 min. For detection of Dami cell proteins, approximately 2 × 10 6 cells were lysed in 100 μL 1 × lysis buffer on ice for 30 min and the supernatant was separated from insoluble cell fractions by centrifugation at 12,000 × g for 10 min at 4°C. Protein concentration was measured by the BCA protein assay kit and adjusted to the same level by lysis buffer. Proteins were separated by SDS-PAGE and immunoblotted with specific antibodies.

Confocal microscopy analysis of platelets and MKs
For confocal analysis of resting platelets, washed mouse platelets (1 × 10 6 in 100 μL Modified Tyrode's buffer) were fixed with an equal volume of 4% PFA and spun to slides. For spread platelets, the slides were first coated with 30 μg/mL fibrinogen diluted in 0.1 M NaHCO 3 (pH8.3) at 4°C overnight and blocked with 5% BSA. Washed mouse platelets (1 × 10 7 /mL) were allowed to adhere and spread on the fibrinogen-coated surface in the presence of 0.1 U/mL thrombin for 2 h at 37°C. Platelets were fixed with 4% PFA. For spread MKs, the confocal dishes were first coated with 100 μg/mL fibrinogen or 50 μg/mL type I collagen at 4°C overnight and blocked with 5% BSA. MKs cultured from mouse FL HPCs were resuspended in DMEM medium and allowed to adhere and spread on a fibrinogenor collagen-coated surface for 3 h at 37°C with 5% CO 2 . MKs were fixed with 4% PFA. Platelets and MKs were then permeabilized with 0.1% Triton X-100 and blocked with 5% BSA at RT. Platelets and MKs were stained with specific primary antibodies (10 μg/mL) and Alexa Fluor 488-or 555-conjugated secondary antibodies (4 μg/mL). For F-actin staining, Alexa Fluor 488conjugated phalloidin (0.165 μM) was incubated for 15 min. Sufficient washing was performed after each step. Platelets and MKs were observed with a LEICA TCS SP8 confocal microscope with a 63x oil immersion lens.