The intracellular and plasma membrane pools of phosphatidylinositol-4-monophosphate control megakaryocyte maturation and proplatelet formation

Background Megakaryocytes (MKs) develop from hematopoietic stem cells after stimulation by the cytokine thrombopoietin. During megakaryopoiesis, MKs enlarge, undergo the process of endomitosis, and develop intracellular membranes (demarcation membrane system, DMS). During DMS formation, there is active transport of proteins, lipids, and membranes from the Golgi apparatus to the DMS. The most important phosphoinositide that controls anterograde transport from the Golgi apparatus to the plasma membrane (PM) is phosphatidylinositol-4-monophosphate (PI4P), whose levels are controlled by suppressor of actin mutations 1–like protein (Sac1) phosphatase at the Golgi and endoplasmic reticulum. Objectives Here we investigated the role of Sac1 and PI4P in megakaryopoiesis. Methods We analyzed Sac1 and PI4P localization in primary MKs derived from fetal liver or bone marrow and in the DAMI cell line by immunofluorescence. The intracellular and PM pools of PI4P in primary MKs were modulated by expression of Sac1 constructs from retroviral vector and inhibition of PI4 kinase IIIα, respectively. Results We showed that in primary mouse MKs, PI4P is mostly found in the Golgi apparatus and the PM in immature MKs, while in mature MKs, it is found in the cell periphery and at the PM. The exogenous expression of wild-type but not C389S mutant (catalytically dead) Sac1 results in the perinuclear retention of the Golgi apparatus resembling immature MKs, with decreased ability to form proplatelets. The pharmacologic inhibition of PI4P production specifically at the PM also resulted in a significant decrease in MKs that form proplatelets. Conclusion These results indicate that both intracellular and PM pools of PI4P mediate MK maturation and proplatelet formation.


| I N T R O D U C T I O N
Megakaryocytes (MKs) are the largest (50-100 μm in diameter) and one of the rarest (0.01%) cells in the bone marrow (BM) [1]. They differentiate from hematopoietic stem cells in response to thrombopoietin, a major cytokine that is produced by the liver and binds to its receptor c-Mpl, promoting the development and growth of MKs [1,2].
During megakaryopoiesis, MKs undergo the process of endomitosis, increase in size, pack proteins into αand dense granules, and develop a system of membranes called the demarcation membrane system (DMS) that is a source of membranes for future platelets (PLTs) [3].
The DMS is composed of numerous cisternae and tubules that are continuous with the PM, and it requires high levels of proteins and lipids for its formation [1]. The formation of DMS starts in the perinuclear region near the Golgi apparatus [4]. It has been shown that vesicles from the trans-Golgi network (TGN) are localized close to the DMS and fuse with the DMS, while the blockage of Golgi trafficking increases the number of immature MKs [4]. In the later stages of DMS formation, the endoplasmic reticulum (ER) makes membrane contact sites with the DMS [4]. This suggests that the transport vesicles deliver the necessary proteins, lipids, and membranes for the formation of DMS. One of the most abundant phosphoinositides (PIs) at the Golgi apparatus and the ER is phosphatidylinositol-4-monophosphate (PI4P) [5]. It can be produced by 4 different phosphatidylinositol-4 kinases (PI4Ks) divided into 2 subfamilies: type II kinases (PI4KIIα and PI4KIIβ) and type III kinases (PI4KIIIα and PI4KIIIβ) [6]. By binding effector proteins, PI4P regulates anterograde vesicular transport from the TGN to the plasma membrane (PM) [7][8][9]. Although PI4P is mostly present at the Golgi in animal cells, it can be found at the PM as well [5].

Suppressor of actin mutations 1-like protein (Sac1) is a SAC1
domain-containing phosphatase [10] that mainly dephosphorylates PI4P at the Golgi apparatus and ER [5]. It is a transmembrane protein with a CX5R(T/S) catalytic motif [11] that localizes to both ER and Golgi, which is growth factor-dependent [12]. When the cells are stimulated with growth factors, Sac1 is mostly found at the ER, resulting in the accumulation of PI4P at the Golgi apparatus and stimulation of the anterograde transport from the Golgi to the PM. In growth factor-deprived conditions, Sac1 mostly shifts from the ER to the Golgi apparatus, where it dephosphorylates PI4P, resulting in a decrease in anterograde trafficking from the Golgi apparatus to the PM. Data from the human and mouse PLT proteome revealed that Sac1 is highly expressed in human (5800 copies) and mouse (10,500 copies) PLTs [13], suggesting high copy numbers in MKs as well.
Considering the proximity of the Golgi apparatus to maturing DMS, the contribution of its transport vesicles in supplying growing membranes, and the PI4P that regulates anterograde Golgi trafficking, we studied the role of PI4P and Sac1 in primary mouse MKs.
Here we show that in the early stages of primary mouse MK maturation, PI4P is enriched at the Golgi apparatus, while in the later stages, it is found to be enriched at the PM. The exogenous expression of wild-type Sac1 results in smaller MKs that produce fewer proplatelets. Since the inhibition of PI4P production at the PM by PI4KIIIα also results in a decrease in proplatelet formation, we suggest that both the intracellular and PM pool of PI4P mediate MK maturation and proplatelet formation.
• We studied PI4P in relation to Golgi apparatus in primary mouse megakaryocytes (MKs).
• In immature MKs, PI4P is enriched at the Golgi, but in mature MKs, it is enriched at the plasma membrane.
• Disrupting PI4P localization or its levels leads to smaller MKs that produce fewer proplatelets.

| Mouse BM and fetal liver isolation and MK production
Mouse BM from 8-to 12-week-old mice and fetal liver (FL) from 13.5 days postcoitum old embryos were used. Cell aggregates from the BM were dissociated by mechanical aspiration 10 times through a 21G × 1" needle, 10 times through a 23G × 1" needle, and 2 times through a

| Mouse PLT isolation
Mouse PLTs were isolated from mouse peripheral blood. Peripheral blood was drawn in 1/10 of the Aster-Jandl anticoagulant (85 mM Na citrate, 69 mM citric acid, 20 g/L glucose, pH 4.6). Blood was centrifuged for 8 minutes at 100 g. Platelet rich plasma was diluted 1:5 in washing buffer (140 mM NaCl, 5 mM KCl, 12 mM trisodium citrate, 10 mM glucose, 12.5 mM sucrose; pH = 6) and centrifuged for 6 minutes at 100 g. Platelet rich plasma supernatant was taken and centrifuged for 5 minutes at 1200 g. The pellet was resuspended in 1 mL of washing buffer and again centrifuged for 5 minutes at 1200 g. The pellet was resuspended in 200 μL of resuspension buffer (10 mM HEPES, 140 mM NaCl, 3 mM KCl, 0.5 mM MgCl 2 , 0.5 mM sodium hydrogen carbonate, 10 mM glucose; pH = 7.4). PLTs were then left to rest for 30 minutes at 37 • C before lysing for western blot assay.

| Constructs, retroviral production, and MK transduction
Human Sac1 (shares a 95.06% homology with the mouse protein) from pEGFP-Sac1 (wild-type [WT]) and C389S-Sac1 (mutated) from pEGFP-C389S-Sac1 were amplified by polymerase chain reaction with primers Immunostaining of intracellular pools of PI4P, as well as Golgi markers through MK development, was performed as previously described [14]. Briefly, all steps were performed at room temperature.

| Statistical analysis
All experiments were performed at least in triplicate. Data are represented as mean ± SEM. Data were analyzed by t-test or ANOVA using Prism software (GraphPad). Differences were considered significant when P values were < .05 (*P < .05; ** P < .01; ***P < .001; ****P < .0001; nsP > .5). produce immature or mature MKs, respectively (described in detail in reference [17]). Briefly, immature MKs are significantly smaller than mature MKs and they have a small nuclear/cytoplasmic ratio, while mature MKs enlarge 2-to 4-fold, have a large nuclear/cytoplasmic ratio, and increase the expression of GPIbβ. In BM-derived MKs, we could observe 2 Sac1 isoforms, a highly expressed 67 kDa isoform and an isoform of 55.6 kDa, with lower expression ( Figure 1A). Further quantification of the highly expressed 67 kDa isoform showed that there was no change in expression levels during MK maturation and compared to mouse PLTs ( Figure 1B, HEK293T as control cells). In immature MKs, Sac1 staining was predominantly vesicular, while in mature MKs, it was more dispersed (Figure 1C, D). In both immature and mature BM-derived MKs, Sac1 was following the shape and occasionally colocalizing with the ER marker KDEL ( Figure 1D) and in the vicinity of the Golgi apparatus ( Figure 1C). We observed the same phenotype in FL-derived MKs (Figure 2A, B). We confirmed the specificity of the used antibodies by staining for irrelevant IgG rabbit with anti-rabbit Alexa Fluor 488 secondary antibody ( Supplementary   Fig. S1A), irrelevant IgG mouse with anti-mouse Alexa Fluor 555 secondary antibody ( Supplementary Fig. S1B), and secondary antibodies only (Supplementary Fig. S1C).

| PI4P changes its localization from the Golgi apparatus toward the plasma membrane during MK maturation
In other cells, PI4P is mostly localized on the Golgi apparatus but it can also be found at the PM [5]. we tested the specificity of the anti-PI4P antibody in PLTs [15] and in

MKs by staining for an irrelevant IgM mouse with anti-mouse Alexa
Fluor 568 antibody ( Supplementary Fig. S2A, B).
The change in PI4P localization could be due to the changes in the  Fig. S3D).

| The expression of wild-type Sac1 negatively impacts megakaryocyte maturation
To determine the role of the intracellular pool of PI4P that is controlled by Sac1, we expressed WT and catalytically inactive C389S  Fig. S4A).
The EGFP-MSCV bend was located at 28 kDa, while the Sac1 constructs coupled with EGFP were located at 95 kDa, which corresponds to the sum of the sizes of EGFP (28 kDa) and Sac1 (67 kDa).
To test the function of the cloned constructs before transducing primary MKs, we transfected DAMI cells. PI4P is mostly localized at the Golgi in untransfected DAMI cells ( Supplementary Fig. S5, top panel). Interestingly, upon expression of WT Sac1, the levels of PI4P do not change ( Supplementary Fig. S5B), but its localization changes.
From its compact localization at the Golgi, PI4P disperses throughout the cell (Supplementary Fig. S5A) in approximately 80% of the cells ( Supplementary Fig. S5C). In contrast, the expression of mutant Sac1 significantly increases PI4P levels ( Supplementary Fig. S5B) with no change in its localization ( Supplementary Fig. S5A, C). Since we could not observe a decrease in PI4P with the expression of WT-Sac1, we additionally tested if the antibody for PI4P is specific by expressing the P4M-SidM probe [19], which was shown to specifically bind to PI4P. As shown in Supplementary Figure 6 Fig. S4B). The transduction rate between constructs was comparable, as shown by epifluorescent imaging ( Supplementary   Fig. S4C, D). We observed that WT Sac1 localizes mostly at the center of the cell in the perinuclear region, while mutant Sac1 is mostly Interestingly, expression of WT-Sac1, but not of mutant, changes the morphology of the Golgi apparatus, keeping it at the perinuclear region (further analyzed in Figure 5E, F). Quantification of the colocalization coefficient shows that endogenous Sac1 localizes more at the ER than at the Golgi apparatus in mature FL-derived MKs, although with no significant difference between the two ( Figure 4C). However, the WT and mutant Sac1 localize significantly more at the Golgi apparatus than at endogenous Sac1 ( Figure 4C). The colocalization of WT and mutant Sac1 was also higher with the ER marker than endogenous Sac1, and this was significant in MKs expressing mutant Sac1 ( Figure 4C).
Interestingly, there was no significant difference in PI4P fluores-  Fig. S4E). This is also consistent with the hypothesis that they are smaller and less mature, as shown previously [17].

| PI4P is necessary for proplatelet formation
We performed a proplatelet formation assay to determine the ability of Sac1-expressing MKs to produce proplatelets. MKs expressing WT Sac1 produced significantly fewer proplatelets than MKs expressing only EGFP or mutant Sac1 with no apparent change in proplatelet morphology ( Figure 6A, B). These data suggest that the intracellular pool of PI4P controlled by the Sac1 phosphatase is important for proplatelet formation and that the perturbations in PI4P and/or Golgi localization do not allow full MK maturation and proplatelet formation.
Since PI4P localizes more on the PM with MK maturation, we wanted to see if perturbation of the PM pool of PI4P would change the production of proplatelets. PI4KIIIα was shown to be the major source of PI4P at the PM in other cells [20] and it is highly expressed in both human (protein copy number: 1800) [13] and mouse (protein copy number: 1568) [21] PLTs. Here, we show that PI4KIIIα is highly expressed in mature BM-and FL-derived MKs, as well as mouse PLTs and HEK293T cells (Supplementary Fig. S7). Next, we pharmacologically inhibited PI4KIIIα with GSK-A1 (100 nM) and examined PI4P intensity and the ability of FL-derived MKs to produce proplatelets. and ER in other cells [5], as well as late endosomes and lysosomes [19].  [23,27]. In our experiments, Sac1 is stably expressed at all stages of MK development, and although we did not observe different Sac1 localization with maturation, its appearance changes: in young MKs, it is observed in a punctate pattern that follows the ER, while in mature MKs, it is more dispersed (Figure 1C, D). Differential Sac1 activity could lead to intracellular PI4P hydrolysis; however, accumulation of PI4P at the PM could be due to PI4KIIIα production. The exogenous expression of WT-Sac1 in MKs results in a retention of the Golgi apparatus perinuclearly and its colocalization with TGN46, which subsequently leads to increased intracellular localization of PI4P ( Figure 5). These results indicate that a balanced amount of functional Sac1 needs to be present in MKs. There is a clear correlation between exogenous Sac1 expression and PI4P/Golgi localization, leading to the formation of smaller MKs that produce significantly fewer proplatelets, while catalytically dead Sac1 had no effect. WT, but not catalytically dead Sac1, was shown to interact with COPI [28], and therefore, excess of WT-Sac1 molecules in MKs could bind additional COPI molecules and favor retrograde transport from the Golgi to the ER. In addition, an acute PI4P depletion was shown to diminish anterograde trafficking from the Golgi to PM and late endosomes [29]. Both of these scenarios could reduce the dispersion of the Golgi and consequently maturation of WT-Sac1-expressing MK.
The changes in Golgi morphology have been investigated in mice [4,18], rats [24,25], or guinea pigs [26] but not in human MKs. It would be interesting to see if the same applies to human MKs. MKs are not the only cells in which Golgi fragmentation has been observed. Small Golgi fragments have been found in neuronal dendrites that are separated from the perinuclear Golgi and serve as secretory organelles [30]. These neuronal Golgi fragments are called Golgi outposts (GOs) [31]. It has been shown that GOs are important for dendrite growth and branching because ablation of GOs reduces dendritic extension and retraction [32] and they can also interact with cytoskeletal remodeling and motor proteins [30]. Since GOs in neurons are  [20]. PI4KIIIα is highly expressed in MKs, and its inhibition leads to a significant decrease in PI4P at the PM and a decrease in the formation of proplatelets, suggesting its role in the formation of PI4P at the PM. However, this does not exclude a role of other PI4 kinases in the process of MK maturation. While type III PI4Ks are cytosolic and shuffle between membranes of different cellular compartments or transiently associate with the PM [5], type II PI4Ks are mostly tethered at the Golgi and endosomal membranes [34]. Proteome data revealed that mouse PLTs express all 4 isoforms of PI4K; however, the most abundant isoform is PI4KIIIα [21] and only this isoform was found in human PLT proteome [13]. Our preliminary data indicate the expression of other PI4K isoforms also in mouse MKs, and thus could contribute to PI4P production and its different localization. Further studies are underway to decipher their role.
Several PI4P effector proteins have been shown to directly bind to PI4P in mammalian cells [35]. For some of the PI4P effector proteins, it has been shown to be highly expressed in mouse PLTs such as GGA1 (copy number: 1677), GGA3 (copy number: 342), and GOLPH3 (copy number: 1943) [21]. While GGAs are protein adaptors that localize at the TGN and mediate the transport from the Golgi to endosomes [36], GOLPH3 also localizes to the TGN via binding to PI4P but it is required for the exit of Golgi vesicles and their anterograde trafficking by inducing curvature of the Golgi membranes [37].
It is possible that perturbations in the PI4P levels or localization (eg, when WT-Sac1 is expressed) disable the Golgi recruitment of GOLPH3 to the TGN, which results in a defect of the Golgi anterograde trafficking. What are the exact PI4P effectors that mediate MK maturation remain to be elucidated.
In conclusion, we demonstrate that both the intracellular pool of PI4P that is controlled by the Sac1 phosphatase and the PM pool of PI4P that is controlled by the PI4KIIIα kinase are necessary for MK maturation and proplatelet formation. These results suggest that tight BURA ET AL.

RELATIONSHIP DISCLOSURE
There are no competing interests to disclose.