V-ATPase V0a1 promotes Weibel–Palade body biogenesis through the regulation of membrane fission

  1. Yasuo Yamazaki  Is a corresponding author
  2. Yuka Eura
  3. Koichi Kokame  Is a corresponding author
  1. Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, Japan

Abstract

Membrane fission, the division of a membrane-bound structure into two discrete compartments, is essential for diverse cellular events, such as endocytosis and vesicle/granule biogenesis; however, the process remains unclear. The hemostatic protein von Willebrand factor is produced in vascular endothelial cells and packaged into specialized secretory granules, Weibel–Palade bodies (WPBs) at the trans-Golgi network (TGN). Here, we reported that V0a1, a V-ATPase component, is required for the membrane fission of WPBs. We identified two V0a isoforms in distinct populations of WPBs in cultured endothelial cells, V0a1 and V0a2, on mature and nascent WPBs, respectively. Although WPB buds were formed, WPBs could not separate from the TGN in the absence of V0a1. Screening using dominant–negative forms of known membrane fission regulators revealed protein kinase D (PKD) as an essential factor in biogenesis of WPBs. Further, we showed that the induction of wild-type PKDs in V0a1-depleted cells does not support the segregation of WPBs from the TGN; suggesting a primary role of V0a1 in the membrane fission of WPBs. The identification of V0a1 as a new membrane fission regulator should facilitate the understanding of molecular events that enable membrane fission.

Editor's evaluation

This study is of particular interest to endothelial cell biologists and cell biologists working on intracellular transport. The experiments provide new insights into the contribution of a proton pump and protein kinase D in the formation of Weibel-Palade bodies that contain von Willebrand factor in vascular endothelial cells.

https://doi.org/10.7554/eLife.71526.sa0

Introduction

Approximately one-third of newly synthesized proteins, including secretory proteins, transition from the endoplasmic reticulum to the Golgi apparatus, thereby allowing versatile protein modifications, such as disulfide bond formation (Chen et al., 2005; Farquhar and Palade, 1981; Klumperman, 2011). Subsequently, the proteins reach the membrane compartment, called the trans-Golgi network (TGN) (Klumperman, 2011). At the TGN, the proteins are sorted and packaged as cargos into distinct transport carriers that are targeted to individual subcellular destinations, such as the plasma membrane, endosomes, and/or secretory granules in specialized cells (De Matteis and Luini, 2008; Kienzle and von Blume, 2014; Pakdel and von Blume, 2018; Anitei and Hoflack, 2011). Biogenesis of such membrane-bound structures is initiated by the deformation of the TGN membrane, which functions as a donor membrane, to form a bud where cargo proteins are packaged. The formation of nascent buds is sometimes associated with the recruitment of the coat proteins, clathrin, and coat protein I (Kümmel and Reinisch, 2011; Antonny, 2006). Lastly, the buds separate from the TGN membrane for transport. The process of division of one membrane into two is termed membrane fission, which is expected to occur in leak-free luminal contents (Campelo and Malhotra, 2012; Renard et al., 2018). To achieve such a high-fidelity system, cells must employ elaborate machinery; however, the mechanism of membrane fission remains elusive. One plausible reason is the difficulty in segregating the membrane fission process from the other steps of biogenesis in cells due to the small membrane-bound structures, such as secretory vesicles that are 50–100 nm in diameter, formed at the TGN (Campelo and Malhotra, 2012; Renard et al., 2018).

The von Willebrand factor (VWF) is an extra-large multimeric glycoprotein involved in primary hemostasis by adhering platelets to the sites of vascular injury (Sadler, 2005; Sadler, 1998; Wagner, 1990). The multimers found in plasma vary in size; the largest multimer has a molecular mass of 20 MDa (Fowler et al., 1985; Zheng, 2015). Larger multimers recruit platelets more efficiently; thus, the multimerization of VWF is crucial for hemostasis (Federici et al., 1989). VWF is synthesized in vascular endothelial cells as a single polypeptide with a molecular mass of 350 kDa in the endoplasmic reticulum and immediately forms a disulfide bond-dependent dimer. The dimers are then transported through the Golgi apparatus, where they convert dimers to multimers via disulfide bond formation. The multimers are stored in large secretory granules called Weibel–Palade bodies (WPBs) (McCormack et al., 2017; Metcalf et al., 2008; Mourik and Eikenboom, 2017; Valentijn et al., 2011; Weibel, 2012). WPBs are endothelial cell-specific, elongated, rod-shaped secretory organelles varying in size from 0.5 to 5 μm. VWF multimers in WPBs are exocytosed in response to several stimulants, such as histamine. In electron tomography, VWF multimers in the WPB lumen are found as striated, paracrystalline matrices called VWF tubules (Berriman et al., 2009; Valentijn et al., 2008; Zenner et al., 2007). As VWF tubules are aligned in parallel along the longitudinal axes of WPBs, it is thought that the formation of VWF tubules drives the formation of the unique, elongated shape of the organelles. In vitro reconstitution studies suggest that VWF tubule formation is a highly coordinated process of the alignment of multiple VWF dimers into a right-handed coil and the conversion of the aligned dimers into large multimers (McCormack et al., 2017; Huang et al., 2008; Mayadas and Wagner, 1989; Zhou et al., 2011). This tubulation allows approximately 50-fold compaction of the VWF in WPBs. The tubulation process requires acidic pH and Ca2+ in vitro (Huang et al., 2008; Zhou et al., 2011; Gerke, 2011; Springer, 2011) thus, VWF is expected to be exposed in such a subcellular environment during maturation, presumably at the TGN and/or in the newly forming WPB buds from the TGN (Gerke, 2011; Springer, 2011; Wagner et al., 1986). The biological significance of the acidic milieu in mature WPBs is also evident. Forced neutralization of the WPB lumen by a chemical treatment causes a loss of visible VWF tubules, resulting in rounding of WPBs (Michaux et al., 2006). Neutralization does not reduce agonist-evoked VWF exocytosis; instead, it markedly reduces platelet recruitment because the multimers are unable to disperse efficiently upon exposure to an extracellular neutral environment. Thus, this evidence indicates the biological importance of the subcellular acidic milieu in primary hemostasis mediated by VWF; nevertheless, the mechanism by which endothelial cells achieve such an environment has not yet been addressed.

Vacuolar H+-ATPase (V-ATPase), an ATP-driven proton pump, is a major regulator of the subcellular acidic pH in cells (Collins and Forgac, 2020; Maxson and Grinstein, 2014; Nakanishi-Matsui et al., 2010; Vasanthakumar and Rubinstein, 2020). It is a multi-subunit complex composed of two large domains, V0 and V1. V0 is a membrane-embedded domain that translocates protons across the membrane, whereas the V1 domain hydrolyzes ATP. In mammalian cells, several subunits exist in multiple isoforms encoded by distinct genes, suggesting a large structural diversity of the V-ATPase complex in cells. This study focused on identifying the localization of V0a isoforms (‘subunit a’ in the V0 domain) in WPBs. We further analyzed the separation of newly forming WPB buds from the TGN, and investigated the role of V0a1 in PKD-dependent biogenesis of WPBs.

Results

Human umbilical vein endothelial cell culture

Although VWF is the most commonly used vascular endothelial cell marker in vivo (Wagner, 1990), not all cultured endothelial cells express VWF, particularly when cells are maintained in rather sparse conditions. To use cultured endothelial cells in physiologically relevant conditions, we optimized our culture and gene transfer conditions. Primary cultured human umbilical vein endothelial cells (HUVECs) form a stable monolayer with a characteristic cobblestone-like appearance when cells are maintained on fibronectin for 5–6 days. Under these conditions, essentially all cells expressed VWF at a comparable level, and the cellular junction proteins were located along the cell boundary, as expected in differentiated endothelial cells (Figure 1—figure supplement 1A). Although our tested lipofection- and electroporation-based transfection considerably disturbed the cellular junction, recombinant lentivirus-mediated expression of EGFP showed virtually no effect (Figure 1—figure supplement 1B). Therefore, we used a recombinant lentivirus system throughout this study.

Cellular V-ATPase activity promotes VWF multimer dispersion

The acidic milieu of the WPB lumen is crucial for the hemostatic function of VWF (Huang et al., 2008; Zhou et al., 2011; Gerke, 2011; Springer, 2011; Michaux et al., 2006). V-ATPase is a major regulator of the subcellular acidic pH in cells (Collins and Forgac, 2020; Maxson and Grinstein, 2014; Nakanishi-Matsui et al., 2010; Vasanthakumar and Rubinstein, 2020). To address whether V-ATPase is a determinant of the acidic milieu of WPBs, we used bafilomycin A1 and concanamycin A, which are well-characterized specific V-ATPase inhibitors (Collins and Forgac, 2020). Four hours of exposure to the inhibitors caused rounding of WPBs without affecting the secreted amount of VWF (Figure 1—figure supplement 2A,B), as previously observed in chemically neutralized cells (Michaux et al., 2006), suggesting the possible role of V-ATPase in WPBs. Note that prolonged exposure to bafilomycin A1 (16 h) is reported to reduce the basal/constitutive secretion of VWF (Torisu et al., 2013). Upon exocytosis, VWF tubules in the acidic pH inside WPBs are instantly exposed to the extracellular neutral pH. The rapid change in pH causes the unfurling of VWF tubules, resulting in an efficient dispersion of VWF multimers on the cellular surface (as observed in the histamine-exposed cells in Figure 1—figure supplement 2C; Michaux et al., 2006). While pretreatment with the V-ATPase inhibitors did not alter the regulated exocytosis induced by histamine (Figure 1—figure supplement 2D), it impaired the dispersion of VWF multimers, as observed in the cells exposed to NH4Cl, which neutralizes subcellular acidic compartments. The strings formed by tangled VWF multimers upon histamine exposure of the V-ATPase inhibitor- or NH4Cl-pretreated cells were rather thin and fragile compared to the prominent, thick structures observed in the non-pretreated cells (Figure 1—figure supplement 2C). These observations suggest the contribution of V-ATPase activity in the acidic milieu inside WPBs. Indeed, we confirmed the WPB localization of V0c and V1A, two common subunits of the mammalian V-ATPase complex, in HUVECs (Figure 1—figure supplement 3A,B).

The V0a isoforms are found on the distinct populations of WPBs

To gain further insight into the role of V-ATPase, we used confocal microscopy to observe and screen the subcellular localization of multiple V-ATPase subunits by expressing them as EGFP-fusion proteins in HUVECs. As a result, we found that two out of four isoforms of V0a (‘subunit a’ in the V0 domain) were localized in distinct populations of WPBs. As observed in the EGFP-tagged V0a1 (Figure 1—figure supplement 4A), the endogenous V0a1 (isoform 1 of the V0a subunit) was mainly found in WPBs (Figure 1A). The signal was evident on the WPBs observed in the cell periphery, whereas virtually no signal was observed on the WPBs in the perinucleus (Figure 1A). WPBs are formed at the TGN and transported to the cell periphery upon maturation, suggesting V0a1 localization in mature WPBs. Rab27a, a small GTPase, is a well-characterized molecule recruited to WPBs during maturation (Bierings et al., 2012; Nightingale et al., 2009; Rojo Pulido et al., 2011). Post-budded WPBs from the TGN move along microtubules to the cell periphery, where they are anchored to actin fibers. Rab27a is required for anchoring WPBs to actin. We found well-matched colocalization of V0a1 and EGFP-Rab27a (Figure 1B). Thus, V0a1 appears to be localized in mature WPBs in HUVECs.

Figure 1 with 6 supplements see all
V0a1 is localized on peripheral mature Weibel–Palade bodies (WPBs) in human umbilical vein endothelial cells (HUVECs).

(A) Endogenous V0a1 protein was detected on the peripheral WPBs (closed arrowheads); however, it was not detected on the perinuclear WPBs (open arrowheads). (B) V0a1 was well-colocalized with EGFP-Rab27a, a mature WPB marker. Scale bars, 20 μm (wide field) and 5 μm (expanded). Asterisks, nucleus.

V0a2 showed the opposite localization to V0a1; V0a2-EGFP was localized only on the perinuclear WPBs but not on the peripheral WPBs (Figure 2A). Accordingly, endogenous V0a2 protein was detected in the buds emerging from the TGN as well as on the TGN (Figure 1—figure supplement 4B). Super resolution microscopy revealed that V0a2-EGFP is located on the perinuclear WPB buds; however, they are not often found in post-budded WPBs (Figure 2B). The signal intensity of VWF found in the V0a2-positive WPBs was rather faint compared to the strong signal observed in the post-budded WPBs (Figure 2B). Live cell imaging confirmed that V0a2-EGFP was considerably localized in the newly forming bud and disappeared as soon as the bud segregated from the Golgi apparatus (Figure 2C and Video 1). The V0a2 buds found around the TGN were completely absent from EGFP-Rab27a, a mature WPB marker (Figure 2D). These observations indicated the limited localization of V0a2 in nascent WPBs. Newly forming WPB buds in the TGN are clathrin-coated Zenner et al., 2007. Depletion of the Golgi-associated clathrin adaptor protein-1 (AP-1) results in an almost total loss of visible WPBs Lui-Roberts et al., 2005. Thus, clathrin appears to be an essential factor for biogenesis of WPBs, presumably required for a very early step, such as the bud-forming process at the TGN. Corroborating the findings of earlier studies, nascent WPBs found in the perinucleus were partially decorated with the components of AP-1 complex (Figure 1—figure supplement 5A,B); however, this was not as evident as observed in V0a2-EGFP (Figure 2A–B). As the two other V0a isoforms did not show evident WPB localization as observed in V0a1 and V0a2 (Figure 1—figure supplement 6A,B), we concluded that V0a1 and V0a2 are the predominant V0a isoforms located on WPBs in HUVECs.

Figure 2 with 1 supplement see all
V0a2 is localized on the perinuclear newly forming Weibel–Palade bodies (WPBs).

(A) V0a2-EGFP was found on perinuclear WPBs (arrowheads); however, it was not found in the peripheral WPBs. N-terminal EGFP-tagged V0a2 (EGFP-V0a2) exhibited the same subcellular localization. Scale bars: 20 μm (wide field) and 5 μm (expanded). (B) Super resolution image around trans-Golgi network (TGN). V0a2-EGFP was observed on buds originating from the TGN (closed arrowheads). The von Willebrand factor signal observed in the V0a2 buds was faint compared to the strong signal in the post-budded WPBs (open arrowheads). Scale bar, 5 μm. (C) V0a2-EGFP (green) was localized on the newly forming buds in the Golgi apparatus (arrowhead) and disappeared as soon as the bud separates from the Golgi apparatus (mKate2-Giantin, red). Images were captured from the Video 1. Scale bar, 5 μm. (D) V0a2 buds were absent in Rab27a, a mature WPB marker. Scale bars: 20 μm (wide field) and 5 μm (expanded). Asterisks, nucleus.

Video 1
V0a2-EGFP is found on bud of newly forming Weibel–Palade bodies.

V0a2-EGFP (green) disappeared as soon as it segregates from the Golgi apparatus (red, mKate2-Giantin).

V0a1 and V0a2 are dispensable for VWF multimerization

As bafilomycin A1 and concanamycin A bind to V0c, a common subunit of the V-ATPase complex (Bowman et al., 2006; Huss et al., 2002; Wang et al., 2021), they are expected to block ‘pan’ V-ATPase independent of the variations of subunits/isoforms. To test whether cellular V-ATPase activity is required for the maturation of VWF, we examined the effects of V-ATPase inhibitors on VWF multimerization. VWF is secreted at a low level, even in the absence of exogenous agonist stimulation. The secretion occurs from the WPBs and the TGN (referred to as ‘basal’ and ‘constitutive’ secretions, respectively). Although the secretion is at a low level, it is thought that the continuous supply of VWF multimers via these pathways is crucial for the maintenance of plasma VWF levels for the immediate hemostatic response (Giblin et al., 2008). While V-ATPase blockage did not change the amount of basally/constitutively secreted VWF (Figure 1—figure supplement 2B), the multimer analysis revealed that it caused a selective induction of dimer secretion (Figure 2—figure supplement 1A), suggesting a requirement for cellular V-ATPase activity in multimerization. As the dimer secretion is blocked by brefeldin A, an ER-to-Golgi transport inhibitor (Figure 2—figure supplement 1B), it is likely that the dimer secretion caused by V-ATPase blockage occurs from the TGN via the constitutive secretion pathway. The V-ATPase inhibitor exposure also caused a rapid disappearance of the nascent WPB buds (Figure 2—figure supplement 1C, compared with Figure 2B). The above observations suggested that VWF multimerization occurring at the TGN and/or in nascent WPB buds require V-ATPase activity. Next, we tested whether the V0a isoforms found in WPBs contribute to multimerization. Although the depletion of either V0a1 or V0a2 caused an increase in the secreted amount of VWF multimers, unexpectedly, no detectable alteration was observed in the multimer status even when both isoforms were depleted (Figure 3—figure supplement 1A,B). V-ATPase contributes to the formation of intracellular acidic microenvironment in various cell types (Collins and Forgac, 2020; Maxson and Grinstein, 2014; Nakanishi-Matsui et al., 2010; Vasanthakumar and Rubinstein, 2020). We introduced ratiometric pHluorin coupled with beta-1,4-galactosyltransferase 1 (GalT-pHluorin) (Llopis et al., 1998; Cole et al., 1996) to see if the V0a isoforms are involved in the luminal pH regulation of the TGN where WPBs are formed. GalT-pHluorin is able to measure pH in a range of 5.5–7.5 (Figure 3—figure supplement 2A). Bafilomycin A1 and concanamycin A exposure caused almost complete neutralization of the TGN, whereas the depletion of each V0a isoform resulted in a modest increase of the pH (Figure 3—figure supplement 2A). These observations suggest that the luminal acidic environment of the TGN is largely maintained by V-ATPase activity in HUVECs, but the contribution of V0a1 and V0a2 is limited or masked by functional redundancies. Consistent with the nascent WPB/TGN localization, V0a2 depletion resulted in a slight but homogenous increase of the luminal TGN pH (Figure 3—figure supplement 2B). In contrast, although no statistical difference was observed when the whole TGN pH was compared (Figure 3—figure supplement 2A), V0a1 depletion caused relatively large pH increase only in the some TGN subdomains (arrowheads in Figure 3—figure supplement 2B); suggesting an involvement of V0a1 in the pH regulation of the TGN, although the main cellular pool is found on peripheral WPBs. Taken together, despite the considerable localization on WPBs as well as the partial contribution on the acidic microenvironment of the TGN, V0a1 and V0a2 appear to be dispensable for multimer formation; however, the roles of these V0a isoforms remain elusive.

The V0a isoforms are required for the biogenesis of WPBs

An increase in basal/constitutive VWF secretion is often observed in cells showing an alteration in biogenesis of WPBs (Nightingale et al., 2009; Lui-Roberts et al., 2005). To evaluate the importance of the V0a isoforms in the biogenesis of WPBs, we examined WPBs in V0a-depleted cells (Figure 3—figure supplement 1A). The WPBs were visualized using an anti-VWF antibody. Depletion of V0a2, which is found in the nascent WPBs, resulted in the formation of irregular, twisted WPBs compared to those found in the control (Figure 3A). To clarify the morphological change of WPBs, we determined the longest Feret’s diameter (Fmax) of individual WPBs and the aspect ratio (Fmax/Fmin) calculated from the longest and shortest Feret’s diameters. Both values are expected to be smaller if WPBs are twisted or in irregular shape compared with relatively straight structure observed in the control. Accordingly, Feret’s diameters of WPBs, particularly Fmax/Fmin in the V0a2-depleted cells were obviously smaller than those in the control (Figure 3—figure supplement 3). To confirm the shRNA-mediated phenotype, we introduced HA-tagged shRNA-resistant V0a2 in shRNA-expressing cells. The phenotype was specifically rescued in the resistant V0a2 expressing cells; however, this was not observed in the non-transfected cells (Figure 3B). The number of WPBs formed in the V0a2-depleted cells was comparable to that in the control (Figure 4B), suggesting that V0a2 is not essential for the separation of WPBs from the TGN at the site of biogenesis. WPBs are storage organelles; that is, once formed, they remain in cells until they are exocytosed or turned over. Efficient reduction of target proteins by shRNAs requires time. Thus, it is likely that immunofluorescence of shRNA-transfected cells using anti-VWF antibody detects not only the WPBs that are formed in the absence of the target protein, but also the pre-formed WPBs before depletion. To further confirm the significance of V0a2 in biogenesis of WPBs, HUVECs were transduced with recombinant lentiviral particles expressing HA-tagged VWF after confirmed the depletion of endogenous V0a2 protein (Figure 3—figure supplement 1C). VWFD1-A1-HA was found in the irregularly shaped WPBs located away from the TGN (Figure 3C). The WPBs were decorated with EGFP-Rab27a (Figure 3D). These observations indicate that the membrane-bound structure of WPBs can still be formed and transported to the cell periphery, even in the absence of V0a2, although it appears to be important for the formation of properly shaped WPBs.

Figure 3 with 3 supplements see all
Depletion of V0a2 causes the formation of irregular, twisted Weibel–Palade bodies (WPBs).

(A) Irregular WPBs were generated in the V0a2-depleted cells. WPBs were visualized using immunofluorescence with anti-von Willebrand factor (VWF) antibody. Scale bars, 20 μm (wide field) and 5 μm (expanded). (B) The irregular WPBs were rescued by the induction of shRNA-resistant V0a2 (V0a2-HA). Scale bar, 20 μm. (C, D) WPBs separated from the trans-Golgi network and maturated even in the absence of V0a2. VWFD1-A1-HA (C) and EGFP-Rab27a (D) were introduced in the V0a2-depleted human umbilical vein endothelial cells. Due to limitations in the size for packaging into a lentivirus particle, we used VWFD1-A1 instead of full-length VWF. VWFD1-A1 is shown to function in the same manner as full-length VWF in cells (Michaux et al., 2006). Scale bars, 20 μm (wide field) and 5 μm (expanded). Asterisks, nucleus.

Figure 4 with 1 supplement see all
Weibel–Palade bodies (WPBs) cannot be separated from the trans-Golgi network in the absence of V0a1.

(A) V0a1 depletion caused the perinuclear accumulation of WPBs (arrowheads, compared to that in the shControl in Figure 3A). Scale bars, 20 μm (wide field) and 5 μm (expanded). (B) The depletion of V0a1 resulted in the significant decrease in the number of WPBs (counts per cell, average± SD). We confirmed that there are no obvious difference in cell size between the V0a isoform depleted cells and the control. n = 30 (shControl), n = 29 (shV0a1), and n = 31 (shV0a2). Eight independent fields were collected from each imaging sample. The images were analyzed using Fiji. *p < 0.05 to shControl. (C) The phenotype mediated by V0a1 depletion was rescued by the induction of a shRNA-resistant V0a1 (HA-V0a1). Scale bar, 20 μm. (D, E) WPB contents were trapped in the perinuclear WPBs that accumulated in the absence of V0a1. VWFD1-A1-HA (D) and P-selectin-EGFP (E) were transfected in V0a1-depleted cells. Both proteins were observed in the accumulated WPBs (arrowheads, compared to that in the shControl in Figure 3C for VWFD1-A1-HA). (F) EGFP-Rab27a was absent in perinuclear WPBs that accumulated in V0a1-depleted cells (arrowheads). Scale bars, 20 μm (wide field) and 5 μm (expanded). Asterisks, nucleus. Source files of all graph data shown in this figure are available in Figure 4—source data 1.

The phenotype induced by the depletion of V0a1, which is found in the peripheral mature WPBs, is more striking. The number of WPBs in the V0a1-depleted cells was substantially decreased compared to that in the control, and the formed WPBs were largely accumulated around the TGN (Figure 4A–B). Large decrease in Feret’s diameter in the V0a1-depleted cells (Figure 3—figure supplement 3) seemed to represent the appearance of rounded WPBs observed in cell periphery (Figure 4A), although the WPBs accumulated around the TGN still retain the characteristic elongated shape. As the rounded WPBs were not often observed when WPBs were visualized by the introduction of VWFD1-A1-EGFP and P-selectin-EGFP in the V0a1-depleted cells (Figure 4D and E), we conclude that they are WPBs formed before the V0a1 deletion was achieved. The perinuclear accumulation of WPBs was rescued by the induction of shRNA-resistant V0a1 (Figure 4C). The phenotype was further confirmed by the induction of tagged VWFD1-A1 and P-selectin in V0a1-depleted cells (Figure 4D–E). P-selectin is a leukocyte-recruiting protein that is packaged into WPBs at the TGN (Harrison-Lavoie et al., 2006; Larsen et al., 1989). Both proteins were trapped in the perinuclear buds originating from the TGN in the absence of V0a1 (Figure 4D–E). The accumulated WPB buds were not EGFP-Rab27a-positive (Figure 4F). Live cell imaging showed that the accumulated WPBs in the V0a1-depleted cells were rather static compared to those in the control (Videos 2 and 3). These observations suggest that newly forming WPBs cannot be separated from the TGN in the absence of V0a1. Post-budded WPBs from the TGN are transported along the microtubule network to the cell periphery, where they are tethered to actin fibers (Bierings et al., 2012; Nightingale et al., 2009; Rojo Pulido et al., 2011). We believe that disruption of cytoskeletal architecture is unlikely because we were unable to observe any apparent alteration of microtubule and actin networks in the V0a1-depleted cells (Figure 4—figure supplement 1A). Accordingly, the induction of a dominant-negative form of either dynactin 2 (Rondaij et al., 2006) or myosin Va (Rojo Pulido et al., 2011), which disturbs the microtubule or actin network-mediating transport of WPBs, did not mimic the phenotype upon V0a1-depletion (Figure 4—figure supplement 1B,C).

Video 2
Weibel–Palade bodies (WPBs) in the control shRNA introduced cells.

WPBs were marked with P-selectin-EGFP (green), and the Golgi apparatus was identified using mKate2-Giantin (red). shControl, Video 2; shV0a1, Video 3.

Video 3
Accumulated Weibel–Palade bodies (WPBs) at the Golgi apparatus in the V0a1-depleted cells are static compared to those in the control.

Protein kinase D, a membrane fission regulator is required for the biogenesis of WPBs

At the TGN, newly produced proteins that transition through the Golgi apparatus are sorted and packaged into distinct membrane-bound structures (De Matteis and Luini, 2008; Kienzle and von Blume, 2014; Pakdel and von Blume, 2018; Anitei and Hoflack, 2011). The formation is initiated by the deformation of the TGN membrane to form a bud where the proteins are packaged. The formed bud then segregates from the TGN membrane by membrane fission for further transport (Campelo and Malhotra, 2012; Renard et al., 2018). In the absence of V0a1, WPBs seemed unable to separate from the TGN, although the buds were formed (Figure 4). We postulated that V0a1 is required for membrane fission of WPBs at the TGN. Thus, we screened known TGN membrane fission regulators, such as dynamins (Ferguson and De Camilli, 2012; Hinshaw, 2000), CtBP1/BARS (Bonazzi et al., 2005; Pagliuso et al., 2016; Valente et al., 2012; Weigert et al., 1999), and serine/threonine-protein kinases PKDs (Bard and Malhotra, 2006; Bossard et al., 2007; Liljedahl et al., 2001; Malhotra and Campelo, 2011; Van Lint et al., 2002), to determine whether any of these contribute to the biogenesis of WPBs. Although the dominant-negative forms of various splicing variants of dynamin 2, dynamin 3, and CtBP1/BARS are virtually ineffective in biogenesis of WPBs (Figure 5—figure supplement 1, Figure 5—figure supplement 2), we found that the induction of the kinase-dead mutants of the PKD family members (PKD1K612N, PKD2K580N, and PKD3K605N) in HUVECs significantly impaired the formation of WPBs (Figure 5A, Figure 5—figure supplement 3A,B,C). The expression of kinase-dead PKD mutants leads to the formation of tubules originating from the TGN because of the blockage of PKD-mediated membrane fission. In such cells, cargo proteins that should be transported in a PKD-dependent manner are trapped in the tubules (Liljedahl et al., 2001). Endogenous VWF protein was found in the tubules formed by all three PKD mutants (Figure 5A, Figure 5—figure supplement 3A,B), indicating that the membrane fission of WPBs at the TGN requires PKDs. We further investigated whether V0a1 had any effect on the subcellular localization of wild-type PKDs (PKDsWT). EGFP-PKD1WT was found mainly in the cytosol and certain parts of the TGN in the control cells, as reported earlier (Figure 5B; Bard and Malhotra, 2006; Malhotra and Campelo, 2011; Van Lint et al., 2002). EGFP-PKD2WT and EGFP-PKD3WT showed comparable localization to EGFP-PKD1WT (Figure 5—figure supplement 3D,E). In contrast, EGFP-PKDsWT substantially accumulated on the WPBs, which were retained around the TGN upon V0a1 depletion (Figure 5B–E, Figure 5—figure supplement 3D,E). The induction of PKDsWT appears to be not enough to induce the separation of WPBs from the TGN in the V0a1-depleted cells. A similar accumulation of endogenous PKD2 protein was observed in the V0a1-depleted cells (Figure 5—figure supplement 4). Thus, the above observations indicate that the segregation of newly forming WPB buds from the TGN is mediated by PKD-dependent membrane fission, and suggest that either V0a1 is required for this process or that V0a1 and PKD coordinate with each other in this process.

Figure 5 with 4 supplements see all
Weibel–Palade bodies (WPB) biogenesis requires protein kinase D1 (PKD1).

(A) The von Willebrand factor (VWF) was trapped in the tubules (arrowheads) formed by the induction of a dominant-negative form of PKD1 (EGFP-PKD1K612N) because of the blockage of PKD1-mediating membrane fission at the TGN (arrowheads). (B) EGFP-PKD1WT was accumulated on the perinuclear WPBs that accumulated in the absence of V0a1. The channel overlap of EGFP-PKD1WT and VWF are shown as ‘colocalization’ on the right. Scale bars, 5 μm. C–E. The localization of EGFP-PKDs on WPBs was significantly increased in the V0a1-depleted cells. The colocalization of VWF and EGFP-PKD1WT (C), EGFP-PKD2WT (Figure 5—figure supplement 3D), and EGFP-PKD3WT (Figure 5—figure supplement 3E) is shown as Pearson’s correlation coefficient (PCC, average± SD). n = 14 (VWF and EGFP-PKD1WT), n = 12 (VWF and EGFP-PKD2WT), and n = 13 (VWF and EGFP-PKD3WT). *p < 0.05 to shControl. Asterisks, nucleus. Source file of the graph data shown in C–E is available in Figure 5—source data 1.

Diacylglycerol and phosphatidylinositol-4-phosphate are present on nascent WPBs

To further elucidate the underlying mechanisms of V0a1-mediated regulation of membrane fission, we investigated how PKDs could be recruited to newly forming WPBs. Subcellular localization of PKDs is regulated by diacylglycerol (DAG), a membrane lipid (Baron and Malhotra, 2002; Maeda et al., 2001). We used the C1 domain from protein kinase Cε (EGFP-PKCε-C1) as a high-affinity DAG reporter (Lehel et al., 1995; Stahelin et al., 2005). A large DAG pool was detected on the TGN, as expected from an earlier study (Figure 6A; Lehel et al., 1995). The reporter was also found in the perinuclear WPB buds in the control group (Figure 6A). The signal was more evident in the accumulated WPB buds formed upon V0a1-depletion (Figure 6A). DAG localization on perinuclear WPBs was further confirmed using two additional DAG reporters, including the C1a domain from PKD1 (Figure 6B, Figure 6—figure supplement 1A; Baron and Malhotra, 2002; Maeda et al., 2001; Tsai et al., 2014). These observations showed that DAG was present on the membrane of newly forming WPBs, and PKDs were recruited through the interaction with DAG. We further investigated whether the PKDs on the WPB buds were enzymatically active. PI4-KIIIβ, a type III phosphatidylinositol-4-kinase localized at the TGN, has been identified as a PKD substrate (Hausser et al., 2005). Its phosphorylation on the activation loop causes the activation of the enzyme, resulting in the conversion of phosphatidylinositol to phosphatidylinositol-4-phosphate (PI4P). We introduced the PH domain from the oxysterol-binding protein (EGFP-PH-OSBP) as a PI4P reporter in HUVECs (Balla et al., 2005). As observed in the DAG reporters, the signal was detected on the accumulated WPBs in the V0a1-depleted cells and on the perinuclear WPBs in the control to a lesser extent (Figure 6C). Two additional PI4P reporters, PH-FAPP1-EGFP (Balla et al., 2005) and EGFP-P4M-SidM (Hammond et al., 2014), also showed similar results (Figure 6—figure supplement 1B,C), indicating a presence of PI4P on perinuclear newly forming WPBs. To see if the PI4P on the forming WPBs is provided through the activation of PKDs, we performed western blotting of HUVEC lysate by using an anti-phospho-PKD1 antibody that detects the phosphorylation on the activation loop (pSer744/748). V0a1-depletion mediated by two independent shRNAs caused an increase of total PKD1 protein compared with the control, whereas we were unable to see any detectable changes in the level of phosphorylated PKD1 (Figure 6—figure supplement 2). As the substantial amount of PKDs were observed on the accumulated WPBs (Figure 5B–E, Figure 5—figure supplement 3D,E), an increase of PKD1 protein might be due to a difference in the protein stability dependent on the subcellular localization. Alternatively, it may be due to an upregulation of PKD1 expression by a cell autonomous compensation or unknown mechanism mediated by the V0a1-depletion. Because of the low endogenous expression of PKDs, it is hard to test whether the PKDs observed on the accumulated WPBs in the V0a1-depleted cells are active/phosphorylated or not. Therefore, although total amount of cellular active PKD1 is unchanged independent from the absence or presence of V0a1, we were unable to clarify the requirement of V0a1 on the PKDs activation/phosphorylation.

Figure 6 with 2 supplements see all
Diacylglycerol (DAG) and phosphatidylinositol-4-phosphate (PI4P) are present on newly forming Weibel–Palade bodies (WPBs).

(A, B) DAG was present on newly forming perinuclear WPBs. Two specific DAG reporters, EGFP-PKCε-C1 (A) and EGFP-PKD1-C1a (B), were introduced into human umbilical vein endothelial cells. EGFP-PKCε-C1 (A) is found in the perinuclear WPBs of the control (arrowheads). The signal is more evident in the accumulated WPBs in V0a1-depleted cells (arrowheads). EGFP-PKD1-C1a (B) is found in the accumulated WPBs in V0a1-depleted cells (arrowheads). Since PKD1-C1a is not a high-affinity DAG reporter as PKCε-C1, it may explain why the signal is not evident in the control. The colocalization of von Willebrand factor (VWF) and EGFP-PKCε-C1 (A) and EGFP-PKD1-C1a (B) is shown as Pearson’s correlation coefficient (PCC, average± SD) on the right. n = 19 (VWF and EGFP-PKCε-C1), and n = 14 (VWF and EGFP-PKD1-C1a). *p < 0.05 to shControl. (C) PI4P was present in perinuclear WPBs. EGFP-PH-OSBP, a PI4P reporter, was found in few perinuclear WPBs in the control (arrowheads). The signal is more evident in V0a1-depleted cells (arrowheads). The colocalization of VWF and EGFP-PH-OSBP is shown as Pearson’s correlation coefficient (PCC, average ± SD) on the right. n = 18 (VWF and EGFP-PH-OSBP). *p < 0.05 to shControl. Scale bars, 5 μm. Asterisks, nucleus. Source file of the graph data shown in 6 A–C is available in Figure 6—source data 1.

Discussion

Membrane fusion and fission are fundamental subcellular events essential for life. In contrast to membrane fusion, which is mainly driven by SNARE proteins and cofactors (Südhof and Rothman, 2009), the process of membrane fission is less understood (Campelo and Malhotra, 2012; Renard et al., 2018). Our study reveals that the V-ATPase subunit V0a1 promotes the biogenesis of WPBs through the regulation of membrane fission (Figure 7). V-ATPase has been suggested to regulate membrane fusion by generating electrochemical gradients or as a structural component of the fusion machinery (Merz, 2015), whereas any involvement in membrane fission has not been described previously. Although we were unable to determine whether the requirement of V0a1 is associated with the proton pump function because of the pleiotropic effects of V-ATPase in biogenesis or multimerization of WPBs, we believe that the identification of V0a1 as a new membrane fission regulator should facilitate the elucidation of the molecular events that allow membrane fission. V0a1 is found on the peripheral mature WPBs in endothelial cells, whereas V0a2, the other V0a isoform, is only observed in nascent WPBs. Consistent with our findings, these V0a isoforms were detected in WPB proteomes (Holthenrich et al., 2019; van Breevoort et al., 2012). V0a2 appears to be required for the formation of proper, elongated shaped WPBs, whereas it is not essential for bud formation or membrane fission of WPBs despite the specific localization on newly forming WPBs. Nevertheless, the requirement of V0a1, which is mainly found in peripheral WPBs, for membrane fission that occurs at the perinuclear TGN is mystifying. V0a1 is also found on small intracellular vesicles in addition to WPBs; perhaps, these are acidic subcellular compartments such as endosomes and lysosomes. Endosomes are observed to form close contacts with WPBs in electron tomographic analysis (Berriman et al., 2009). The endosomal/lysosomal tetraspanin CD63/LAMP3 is delivered to WPBs during maturation (Harrison-Lavoie et al., 2006). Thus, it is likely that the small vesicle-mediated transfer of V0a1 could drive the membrane fission of nascent WPBs. Relatively large pH increase was observed in some TGN subdomains in the V0a1-depleted cells (Figure 3—figure supplement 2B) possibly suggests that the biogenesis of WPBs occurs at particular subdomains of the TGN. Further work is required to elucidate the role of V0a1 in membrane fission.

The biogenesis of Weibel–Palade bodies (WPBs) requires vacuolar H+-ATPase V0a1 and PKD for the membrane fission.

V0a1 on small vesicles such as endosomes could drive the membrane fission (see Discussion). In the absence of V0a1, nascent WPBs could not separate from the trans-Golgi network (right). PKD activity was also required for the biogenesis of WPBs. Since the induction of wild-type PKDs did not support the segregation of WPBs from the TGN in the absence of V0a1, V0a1 appeared to be a primary factor in the membrane fission of WPBs.

We also found that DAG is present on the membrane of newly forming WPBs; however, it was not found in the post-budded WPBs. DAG accumulation on nascent WPBs became more evident in the absence of V0a1. Thus, PKDs appear to be recruited to the WPB through the interaction of the DAG-binding domain. Insertion of the N-terminal hydrophobic stretch of PKDs into the outer leaflet of the TGN is suggested to assist membrane fission by generating membrane curvature (Campelo and Malhotra, 2012; Malhotra and Campelo, 2011). We believe that the hydrophobic stretch-mediating mechanism is unlikely in the biogenesis of WPBs because we observed essentially equivalent effects on all PKD members, including PKD3, which lacks the hydrophobic stretch in the structure (Campelo and Malhotra, 2012). Instead, our observations suggest functional redundancy of PKDs in biogenesis of WPBs. In fact, a single knockdown of either PKD1 or PKD2 does not show any significant effect on WPB formation as well as the secretion in earlier reports (Ketteler et al., 2017; Hao et al., 2009). DAG is a conical-shaped lipid that is thought to be relatively concentrated in the TGN buds to increase membrane curvature (Bankaitis et al., 2012). A theoretical model suggests that the presence of DAG on buds can stabilize the structure and prevent membrane fission (Shemesh et al., 2003). The disappearance of DAG in the post-budded WPBs may suggest a conversion of DAG into other lipids that destabilize the membrane structure to enable membrane fission. We also observed PI4P, a well-characterized membrane signaling lipid, on perinuclear nascent WPBs similar to DAG. PI4P and phosphatidic acid, a lipid generated from DAG through single-step metabolism, are shown to work as pH biosensors in cells; the pH-dependent interaction with particular proteins regulates cellular processes such as membrane biogenesis and intracellular trafficking (Shin et al., 2020; Young et al., 2010). TGN pH measurements suggest that V0a1 appears to regulate luminal pH in particular TGN subdomains in HUVECs (Figure 3—figure supplement 2B). Thus, the possible contribution of these lipids and cellular pH, as well as related lipid metabolism enzymes and lipid transporting proteins in WPB biogenesis should be investigated in the future.

In conclusion, we have shown that two V0a isoforms of V-ATPase promote the biogenesis of WPBs in endothelial cells. In particular, V0a1 is essential for the membrane fission in nascent WPBs at the TGN. We also found that PKD activity is required for WPB biogenesis. As the induction of wild-type PKDs in V0a1-depleted cells did not support WPB segregation from the TGN, V0a1 appears to be a primary factor in the membrane fission of WPBs, although the molecular link between V0a1 and PKDs is still unclear. Qualitative or quantitative deficiency of plasma VWF multimers causes a bleeding disorder called von Willebrand disease (Sadler, 2005; Sadler, 1998; Sharma and Haberichter, 2019). A V0a2 mutation has recently been reported to cause bleeding diathesis in humans (Karacan et al., 2019). Plasma VWF activity is significantly impaired in patients without any substantial loss of plasma VWF antigen level, thus exhibiting a typical sign of von Willebrand disease, although no associating mutation was found in the VWF gene. We believe that our findings will facilitate further understanding of the pathology associated with VWF, such as von Willebrand disease, as well as the mechanism of membrane fission.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (Homo sapiens)ATP6V0A1NAQ93050 (VPP1_HUMAN)V-type proton ATPase 116 kDa subunit a1
Gene (Homo sapiens)ATP6V0A2NAQ9Y487 (VPP2_HUMAN)V-type proton ATPase 116 kDa subunit a2
Cell line (Homo sapiens)Lenti-X 293T Cell LineClontechCat#632180
Cell line (Homo sapiens)HUVECLonzaCat#C2519A
Antibodyanti-VE-Cadherin (mouse monoclonal)R&D SystemsCat#123413RRID:AB_2260374(1:1,000)
Antibodyanti-ZO-1 (rabbit monoclonal)CSTClone:D6L1ERRID:AB_2798287(1:50)
Antibodyanti-VWF (rabbit polyclonal)Dako/Agilent TechnologiesCat#P0226RRID:AB_579516(1:1,000)
Antibodyanti-VWF (sheep polyclonal)AbcamCat#ab11713RRID:AB_298501(1:500)
Antibodyanti-V0c (rabbit polyclonal)Thermo-Fisher ScientificCat#PA5-66746RRID:AB_2662731(1:25)
Antibodyanti-V1A (rabbit polyclonal)Thermo-Fisher ScientificCat#PA5-65137RRID:AB_2662833(1:125)
Antibodyanti-V0a1 (rabbit polyclonal)Novus BiologicalsCat#NBP1-89342RRID:AB_11015740(1:200)
Antibodyanti-HA (mouse monoclonal)CovanceClone:16B12RRID:AB_10064068(1:1,000)
Antibodyanti-TGN46 (sheep polyclonal)Bio-RadCat#AHP500GTRRID:AB_2203291(1:500)
Antibodyanti-V0a2 (rabbit polyclonal)Atlas AntibodiesCat#HPA044279RRID:AB_10961274(1:50)
Antibodyanti-V0a2 (rabbit polyclonal)My BioSourceCat#MBS8527160(1:100)
Antibodyanti-α-tubulin (mouse monoclonal)CSTClone:DM1ARRID:AB_1904178(1:2,000)
Antibodyanti-PKD2 (rabbit monoclonal)AbcamClone:EP1495YRRID:AB_922233(1:200)
Antibodyanti-PKD2 (rabbit polyclonal)GeneTexCat#C1C3RRID:AB_1951437(1:200)
Antibodyanti-phospho-PKD/PKCμ(Ser744/448) (rabbit polyclonal)CSTCat#2054RRID:AB_2172539(1:1,000)
Antibodyanti-PKD/PKCμ (rabbit monoclonal)CSTClone:D4J1NRRID:AB_2800149(1:1,000)
AntibodyAlexa 488-, 555-, 647 secondariesThermo Fisher Scientific(1:400)
AntibodyCy3-secondaryJackson Immunoresearch(1:400)
Recombinant DNA reagentpLVSIN-CMV Pur VectorClontechCat#6183
Recombinant DNA reagentMission shRNA plasmid non-targeting controlSigma-AldrichSHC016
Recombinant DNA reagentMission shRNA plasmid against V0a1Sigma-AldrichTRCN0000380234Mainly used throughout the study.
Recombinant DNA reagentMission shRNA plasmid against V0a1Sigma-AldrichTRCN0000333635Used in Figure 6—figure supplement 2B.
Recombinant DNA reagentMission shRNA plasmid against V0a2Sigma-AldrichTRCN0000043494
Chemical compound, drugBafilomycin A1Cayman ChemicalsCat#11038100 nM
Chemical compound, drugiFluor555-conjugated phalloidinCayman ChemicalsCat#20552(1:400)
Chemical compound, drugConcanamycin AAdipoGen Life SciencesCat#BVT-0237100 nM
Chemical compound, drugBrefeldin AThermo Fisher ScientificCat#B7450100 ng/ml
Chemical compound, drugHistamineSigma-AldrichCat#H7250100 μM
Chemical compound, drugTransIT Lenti transfection reagentClontechCat#MIR6600
Chemical compound, drugLentiviral High Titer Packaging MixClontechCat#6194

Reagents

Bafilomycin A1 and iFluor555-conjugated phalloidin were purchased from Cayman Chemicals (MI, USA). Concanamycin A was purchased from AdipoGen Life Sciences (CA, USA). Brefeldin A was purchased from Thermo Fisher Scientific (MA, USA). Histamine was procured from Sigma-Aldrich (MO, USA). The following primary antibodies were used: mouse monoclonal anti-VE-Cadherin (#123413, 1:1,000, R&D Systems, MN, USA), rabbit monoclonal anti-ZO-1 (D6L1E, 1:50, CST, MA, USA), rabbit polyclonal anti-VWF (P0226, 1:1,000, Dako/Agilent Technologies, CA, USA), sheep polyclonal anti-VWF (ab11713, 1:500, Abcam, MA, USA), rabbit polyclonal anti-V0c (PA5-66746, 1:25, Thermo-Fisher Scientific), rabbit polyclonal anti-V1A (PA5-65137, 1:125, Thermo-Fisher Scientific), rabbit polyclonal anti-V0a1 (NBP1-89342, 1:200, Novus Biologicals, CO, USA), mouse anti-HA (16B12, 1:1000, Covance, MI, USA), sheep anti-TGN46 (AHP500GT, 1:500, Bio-Rad, CA, USA), rabbit polyclonal anti-V0a2 (HPA044279, 1:50, Atlas Antibodies, Bromma, Sweden), rabbit polyclonal anti-V0a2 (MBS8527160, 1:100, My BioSource, CA, USA), mouse monoclonal anti-α-tubulin (DM1A, 1:2000, CST), rabbit monoclonal anti-PKD2 (EP1495Y, 1:200, Abcam), rabbit polyclonal anti-PKD2 (C1C3, 1:200, GeneTex, CA, USA), rabbit polyclonal anti-phospho-PKD/PKCμ(Ser744/448) (#2054, 1:1000, CST), and rabbit monoclonal anti-PKD/PKCμ (D4J1N, 1:1,000, CST). The secondary antibodies used were Alexa 488, Alexa 555, and Alexa 647 (1:400, Thermo Fisher Scientific), and Cy3 (1:400, Jackson Immunoresearch, PA, USA).

cDNA and plasmids

Request a detailed protocol

The cDNA used in this study were amplified from HUVEC cDNA, unless otherwise stated. AP1M1 and AP1G1 were provided by RIKEN BRC through the National Bio-Resource Project of MEXT, Japan. P4M-SidM was obtained from Dr. Tamas Balla (National Institute of Health, USA). For the expression of tagged proteins, we introduced EGFP, mKate2, or HA cDNA into the multiple cloning site of pLVSIN lentiviral vector (Clontech, CA, USA), such that the inserted protein was tagged at either the N- or C-terminus with a flexible linker (GGGGS ×2). The pLVSIN constructs used in this study were as follows: pLVSIN-EGFP, pLVSIN-N-EGFP-V0a1, pLVSIN-V0a1-C-EGFP, pLVSIN-N-HA-V0a1, pLVSIN-N-EGFP-V0a2, pLVSIN-V0a2-C-EGFP, pLVSIN-V0a2shRNAresist-C-3× HA pLVSIN-N-EGFP-V0a3, pLVSIN-V0a3-C-EGFP, pLVSIN-N-EGFP-V0a4, pLVSIN-V0a4-C-EGFP, pLVSIN-N-EGFP-Rab27a, pLVSIN-AP1M1-C-EGFP, pLVSIN-AP1G1-C-EGFP, pLVSIN-VWFD1-A1(Met1-Pro1480)-C-3× HA, pLVSIN-P-selectin-C-EGFP, pLVSIN-GalT(Met1-Gln61)-pHluorin, pLVSIN-Dynamin 3–1 K44A-C-EGFP, pLVSIN-Dynamin 3–2 K44A-C-EGFP, pLVSIN-Dynamin 2–2 K44A-C-EGFP, pLVSIN-Dynamin 2–3 K44A-C-EGFP, pLVSIN-Dynamin 2–4 K44A-C-EGFP, pLVSIN-Dynamin 2–5 K44A-C-EGFP, pLVSIN-CtBP1/BARSWT-C-EGFP, pLVSIN-CtBP1/BARSS147A-C-EGFP, pLVSIN-N-EGFP-PKD1WT, pLVSIN-N-EGFP-PKD1K612N, pLVSIN-N-EGFP-PKD2WT, pLVSIN-N-EGFP-PKD2K580N, pLVSIN-N-EGFP-PKD3WT, pLVSIN-N-EGFP-PKD3K605N, pLVSIN-N-EGFP-PKCε-C1, pLVSIN-N-EGFP-PKD1-C1a, pLVSIN-PKCγ-C1A-C1A-C-EGFP, pLVSIN-N-EGFP-PH-OSBP, pLVSIN-PH-FAPP1-C-EGFP, pLVSIN-N-EGFP-P4M-SidM(Thr546-Lys647), pLVSIN-dynactin 2-C-mKate2, pLVSIN-N-mKate2-myosin Va tail, and pLVSIN-N-mKate2-Giantin(Glu3131-Leu3259). Mission shRNA plasmids against V0a1 (TRCN0000380234, targeting the 3′-UTR region, and TRCN0000333635, targeting the coding region around Tyr620-Met627) and V0a2 (TRCN0000043494, targeting the coding region around Phe37 -Val55) and non-targeting control (SHC016) were purchased from Sigma-Aldrich. For the construction of HA-tagged shRNA-resistant V0a2, four nucleotides located in the middle of the targeting region were substituted with synonymous codons.

Cell culture

Request a detailed protocol

Primary HUVECs (C2519A, Lonza, Basel, Switzerland) were cultured at 37 °C in EGM-2 supplemented with 2% fetal bovine serum (Lonza). All plates and coverslips used for HUVECs were freshly coated with fibronectin from human plasma (20 μg/mL; Fujifilm Wako, Osaka, Japan). Lenti-X293T cells (Clontech) were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Fujifilm Wako) supplemented with 10% EqualFETAL (Atlas Biologicals, CO, USA) and 1 μg/mL penicillin and streptomycin (Fujifilm Wako) at 37 °C.

Lentivirus particle preparation

Request a detailed protocol

Lenti-X293T cells (Clontech) were plated at a density of 6 × 106 cells in 90 cm2 plates coated with soluble type I-C collagen (Kurabo, Osaka, Japan). After overnight culture, the pLVSIN lentiviral vector was transfected together with lentivirus packaging mix using TransIT Lenti transfection reagents according to the manufacturer’s protocol (Clontech). Conditioned media were collected three times for 96 hr as recombinant lentivirus particles. The virus particles were then precipitated with PEG6000, resuspended in an appropriate amount of phosphate-buffered saline (PBS)/DMEM, and stored at −80 °C until use.

Immunofluorescence and image analysis

Request a detailed protocol

HUVECs were seeded at 50,000 cells/well on fibronectin-coated coverslips in 24-well plates. After 16–18 hr of culture, the cells were transduced with recombinant lentivirus particles overnight. The medium was replaced daily. For the expression of tagged proteins on the shRNA background, cells were exposed to the virus particles that expressed the tagged protein after 48 hr of shRNA transfection. On day 6, the cells were fixed with 4% paraformaldehyde for 25 min at 37 °C. After rinsing with PBS, cells were quenched with 50 mM NH4Cl/PBS for 5 min, permeabilized, and blocked in PBS containing 0.3% Triton X-100% and 5% normal serum from the respective species for 30 min. The cells were incubated with primary antibodies in blocking buffer overnight at 4 °C. Primary antibodies were detected using the appropriate fluorescent secondary antibodies. Coverslips were mounted using Fluoromount-G (SouthernBiotech, AL, USA) and imaged using LSM880 (Plan Apochromat ×63 objective, NA 1.4, ZEISS, Oberkochen, Germany), FV3000 (UPSAPO60XO × 60 60 objective, NA 1.35, Olympus, Tokyo, Japan), and FV1000 (UPlanSApo ×60 objective, NA 1.35, Olympus) microscopes. For the super-resolution analysis (Figure 2B and Figure 2—figure supplement 1C), images were acquired with LSM880 with AiryScan (estimated lateral resolution: 180 nm). EGFP was excited with 488 nm line from an argon laser and collected with MBS 488/561/633 (acquisition band: 500–540 nm). Cy3 was excited with 561 nm line from a DPSS laser and collected with MBS 488/561/633 (acquisition band: 590–630 nm). AlexaFluor647 was excited with 633 nm line from a HeNe laser and collected with MBS 488/561/633 (acquisition band: 645–755 nm). The acquired images were processed with Zen Black software by performing filtering, deconvolution and pixel reassignment to enhance spatial resolution. Images were analyzed using Fiji to count the number and size (Feret’s diameter) of WPBs and co-localization analysis Colocalization Finder for the generation of channel overlap (Figure 5, Figure 5—figure supplements 3 and 4); https://imagejdocu.tudor.lu/plugin/analysis/colocalizationfinder/start, and EzColocalization for the calculation of Pearson’s correlation coefficient (Figures 5 and 6; https://github.com/DrHanLim/EzColocalization/releases/tag/1.1.4 Sheng, 2020) was performed. For the ratiometric pH measurement, GalT-pHluorin was consecutively excited at 405 and 488 nm and the emitted fluorescence was acquired through a spectral slit of 510–550 nm. Mean emission intensities obtained from the identical ROI were used to calculate 405/488 ratio. pH calibration standard was obtained by using monensin-containing calibration buffers adjusted to the indicated pH (Grillo-Hill et al., 2014; Ma et al., 2017). The ratiometric images were generated by the image calculator function on Fiji and the pH calibration data was applied to convert the ratio to pH.

VWF multimer analysis

Request a detailed protocol

HUVECs were seeded and transduced using the same protocol used for immunofluorescence assay, except for the use of coverslips. On the day of analysis, cells were rinsed twice with endothelial basal medium (EBM)–2 without growth supplements or fetal bovine serum. After 2 hr of resting in EBM-2 containing 10 mM HEPES pH 7.4, conditioned media was collected for 1 hr to obtain basally/constitutively secreted VWF. The collected media was mixed with sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis sample buffer without reducing reagent and then loaded onto a ‘SDS–agarose’ gel. After separation, proteins on the gel were capillary blotted onto PVDF membranes and labeled with rabbit anti-VWF antibody (Dako/Agilent Technologies), followed by IRDye 800CW-conjugated anti-rabbit IgG secondary antibody (LI-COR Biosciences, Lincoln, NE, USA). The signal was visualized using the Odyssey CLx Imaging System (LI-COR Biosciences). Normal human plasma (Siemens Healthineers, Erlangen, Germany) was used as a VWF multimer control for each gel. Although we attempted to perform multimer analysis of the conditioned media collected on day six from shRNA-transfected cells, similar to the immunofluorescence (see above), we found that a considerable number of cells were detached from plates during 2 hr of rest. We believe that the depletion of the V0a isoform disturbs the adherence of HUVECs; thus, we used conditioned media collected from day 5 culture for the analysis (Figure 3—figure supplement 1B). We performed CBB staining of cell lysate to confirm equal loading of the samples in each experiment.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting file; Source Data files have been provided for all blots and graphs shown in the manuscript.

References

Decision letter

  1. Suzanne R Pfeffer
    Senior and Reviewing Editor; Stanford University School of Medicine, United States
  2. Felix Campelo
    Reviewer; The Barcelona Institute of Science and Technology, Spain
  3. Vivek Malhotra
    Reviewer; The Barcelona Institute of Science and Technology, Spain

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "V-ATPase V0a1 promotes Weibel-Palade body biogenesis through protein kinase D-dependent membrane fission" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Suzanne Pfeffer as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Felix Campelo (Reviewer #1); Vivek Malhotra (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. We include the full reviews to help guide your next steps in revising this manuscript.

Essential revisions:

The reviewer’s request:

1. Quantitative assessment of changes in WPB size and shape in V0a2/V0a1 depleted cells.

2. Measure lumenal Golgi/TGN pH in WT and a1/a2 KD cells.

3. The claim that PKD is active in the shV0a1 cells is not supported by the data. Please measure PKD activity using more adequate assays (phospho-specific proteins etc) in the absence of V0a1. it seems that PKD in V0a1-depleted cells is predominantly at the Golgi. Therefore, detection of PKD activity by WB using the commercially available activation-loop Ab or an IVK should reflect Golgi activity. Alternatively, you could also use a genetically encoded reporter to measure PKD activity.

4. Provide a more thorough discussion including how pH and lipid metabolism (DAG, PI4P) might be related.

Reviewer #1:

In this work, the authors used human umbilical vein endothelial cells (HUVEC cells) to investigate the biogenesis of Weibel-Palade bodies (WPBs). Since it is well known that an acidic pH in this compartment is crucial for the hemostatic function of its most important component – the protein von Willebrand factor (VWF) -, the authors investigated the role of a particular set of V-ATPase subunits in WPB biogenesis.

They reported that both the endogenous and the exogenously-expressed GFP-tagged V-ATPase subunits V0a1 and V0a2 localize to WPBs. Particularly, V0a1 appears to be found in peripheral (probably mature) WPBs, whereas V0a2 localized to perinuclear regions containing VWF (TGN/nascent WPBs). The authors then found that these proteins are required for the proper morphology of WPBs: when V0a2 is absent, WPBs formed at normal rates but present twisted shapes; whereas when V0a1 is absent, less WPBs are found in cells, and those that form are located at close apposition (maybe still connected to) of the TGN. The authors therefore suggested that V0a1 is a component required for the fission from the TGN of nascent WPBs.

Next, they investigated known TGN fission factors, and found the involvement of protein kinase D (PKD) in WPB formation. Specifically, when expressing kinase-dead (KD) mutants of PKD – which form long tubes connected to the TGN and enriched with cargo proteins (because of defective carrier fission) -, they found that VWF was included in those PKD-KD tubes, suggesting that VWF is a PKD-dependent cargo protein. It is not clear how PKD is required for WPB biogenesis, and what, if any, is the connection between V0a1 and PKD in this process. Interestingly, there's reports in the literature (e.g. doi:10.1152/ajpcell.00504.2008) that show that in the absence of PKD1, endothelial cells can still secrete normal levels of VWF. Finally, the authors suggested that PKDs present in nascent WPBs are enzymatically active because they reported that DAG and PI4P sensors are present on those membranes. However, normal levels of PI4P do not necessarily need to report an enzymatically active PKD, since, for instance, active PKD has been shown to reduce PI4P levels at TGN membranes (by phosphorylation of its substrate OSBP). Hence, it is not clear how V0a1 promotes WPB biogenesis in a PKD-dependent manner.

In summary, this article presents new data showing the involvement of V0a1 in WPB biogenesis, and reported the presence of VWF proteins in PKD-KD tubules. These results open new questions in the field, especially on the mechanism by which these two proteins control WPB fission, and if there is some interplay between them in doing so.

Major comments:

As I stated above, I think the results are potentially interesting, but it is not clear to me what is the molecular mechanism by which V0a1 controls WPB biogenesis. The link with PKD is mild (as said earlier, there are reports indicating that PKD1 is dispensable for VWF secretion, which need to be cited and put in context of the present findings). Some ideas:

– V-ATPases regulate luminel pH. In the absence of V0a1 (or V0a2), is there any changes in WPBs pH? The authors reported that VWF still form multimers, but this only quite indirectly reports on the pH at the TGN/WPBs.

– About the involvement of PKDs: for the quantitation shown in Figure 5A (and related supps), it would help to count the number of WPBs in the cells expressing the PKD-KD mutants, similarly to what has been done in Figure 4B for the knockdowns of the V0as. And as said, put that in the context of the existing literature on the topic.

– Line 240-41: according to the data, both V0a1 knockdown and PKD-KD expression show a similar phenotype (fission defect). What is the connection (if any) between these two processes?

– Figure 6: PKD activity is assessed by the presence of PI4P on the TGN-associated VWF-positive buds. However, PKD is also phosphorylating OSBP, which reduces PI4P levels (actually, acute activation of PKD leads to less PI4P in the Golgi). There are other means to establish PKD activity (use of phospho-substrate antibodies, etc.)

– Lack of mechanisms or upstream/downstream organisation of the V0as/PKD in fission. Do V0a localize to PKD-KD tubes? Is V0a1 a substrate of PKD?

Reviewer #2:

This is a nice story that merits publication because it addresses a long standing issue on the export of bulky WPBs from the Golgi. The involvement of VO-ATPase subunits and PKD pathway makes this process amenable to molecular analysis. My gut feeling is that trafficking of WPBs is likely related to the export of other bulky cargoes like the collagens.

The migration or trafficking of WPB's remains a fascinating challenge. The data on different location of VOa1 and VOa2 are very interesting. The question I have is whether there are different pools of VO-ATPases? Can the authors isolate ( immuno precipitate) two different pools of VO-ATPase; one enriched in VOa1 and another in VOa2? Or did I get this wrong and these subunits are not in complex with other parts of VO-ATPase.

The data on the PKD-DAG are very interesting, but I am a bit confused about the involvement of clathrin. My understanding is that DAG-PKD pathway is independent of the clathrin mediated cargo export from the TGN. Is clathrin found on PKD-Kinase dead containing tubules that are enriched in WBPs?

Finally, is this a budding event or is it that PKD-DAG sequester a part of TGN enriched in WBPs , while the unoccupied part concentrates cargoes exported by the clathrin pathway? The authors should explain this. It would be terrific to test whether a cargo destined to the endosomes/lyososomes is segregated from WPBs in the TGN.

Reviewer #3:

The formation of Weibel-Palade bodies (WPBs), the secretory granules of endothelial cells, is driven by von Willebrand factor (VWF), their most abundant cargo protein. VWF is a large multidomain and multimeric protein that further multimerises in the Golgi complex after dimerisation in the ER, leading to the formation of VWF tubules that are packed into nascent WPBs in the TGN. The number and size of WBPs formed is ultimately controlled by the number of VWF quanta (VWF dimers fused together) formed in the Golgi complex. Importantly, the acidic luminal environment of the Golgi complex and transport vesicles is required for conformational rearrangements of VWF dimers, their tubulation and packing of multimers, respectively, and thus determines the WPB size. Since WPB size determines endothelial function, it is important to understand the mechanism underlying VWF tubulation and multimerisation. The multivacuolar (V)-ATPase-mediated proton pump consists of two large domains, V0 and V1, and continuously pumps H+ ions into the Golgi lumen, contributing to the acidic pH of the organelle. While the contribution of V-ATPase activity to membrane trafficking is known, the V0a isoform involved in the formation of WPBs at the TGN has not yet been identified.

In this paper, the authors have identified two V0 isoforms, V0a1 and V0a2, to be associated with mature WPBs and nascent WPBs, respectively. By using confocal microscopy and biochemical analysis, they show that the two isoforms do not contribute to multimerisation of VWF but are instead required for biogenesis of WPBs at the TGN. Furthermore, they propose that downstream of V0a1 protein kinase D (PKD) is required for WPB membrane fission at the TGN. The paper thus sheds light on a potential function of V-ATPase V0a1 in WPB formation and adds important information on isoform-specific functions of this member of the V-ATPase family. However, while the data showing the localization to and involvement of V0a1 in the biogenesis of WPBs are convincing, the experiments on the molecular mechanisms, especially the link to PKD-dependent membrane fission, fall short and are insufficient to support the conclusions.

Figure 1, Supplement 3F: V0a4-EGFP shows considerable co-localization with peripheral VWF-positive structures, however, the authors do not pursue this finding further claiming that V0a1 and V0a2 are the pre-dominant isoforms located on WPBs. The authors should have investigated a potential effect of V0a4 to rule out that this isoform is involved in WPB biogenesis/transport.

Figure 4: The finding that V0a1 is found on mature WPBs but its depletion affects WPB formation at the TGN is exciting. To make their data more solid the authors should have investigated if and how the pH in the Golgi/TGN changes upon depletion of V0a1. For example, they could have used a pH biosensor.

Figure 5: The authors find that PKD family is involved in the biogenesis of WPBs. While these data are convincing the link to V0a1 is insufficient. Upon V0a1 knockdown more PKD is found on the TGN, which might indicate that more DAG is present. This is an interesting finding, however, not further pursued by the authors. The authors use several PI4P biosensors to show that PKD is enzymatically active. It is clear that there is increased binding of these biosensors to the TGN when V0a1 is depleted, however, many pathways besides PKD contribute to PI4P production and therefore additional evidence should have been provided.

Additionally, PI4P has been shown to sense the cytosolic pH and this impacts cargo sorting and trafficking. Acidification of cytosolic pH displaces PH domains from PI4P (Shin et al., Dev Cell, 2020). Given these results, the authors should have investigated whether silencing of V0a1 could also affect intracellular pH and thus the binding of PH domains to PI4P.

General comment: The authors make several references to morphological changes in WPB size and shape upon perturbation (e.g. Figure 3C). However, it is difficult to see these changes when looking at the images. In particular, in Figure 3A, the difference in VWF-positive structures between shControl and shV0a2 is not clear. Given the findings that number and size of WPBs might be affected by V0a1 and V0a2 depletion but mulitmerization and secretion are not impacted or even increased (Figure 3 – supplement 1) the authors should have considered to investigate VWF string formation as this is dependent on size of WPBs and VWF tubulation.

Figure 1: The authors use Bafilomycin A1, a specific vacuolar H+ ATPase (V-ATPase) inhibitor, to investigate the contribution of V-ATPase to WPB formation. They show that BafA1 treatment does not affect the amount basal and regulated VWF secretion in general but disrupts VWF multimerization. While the data shown are convincing they are in conflict with data published by Torisu et al., (2013) who showed that inhibition of autophagy by BafA1 treatment reduces the secretion of VWF from endothelial cells. Some discussion would be needed to clarify the discrepancy.

Figure 5: The authors could use published PKD activity reporters, perform an in vitro kinase assay, or measure activation loop phosphorylation of PKD via Western Blot analysis to infer on PKD activity in V0a1 depleted cells. Also, some discussion on how V0a1 depletion might affect DAG levels at the TGN would be needed.

General:

The authors should assess changes in WPB size and morphology using quantitative morphometric analysis of the VWF-positive structures (see Lopes da Silva et al., JCellSci, 2016).

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "V-ATPase V0a1 promotes Weibel-Palade body biogenesis through protein kinase D-dependent membrane fission" for further consideration by eLife. Your revised article has been evaluated by Suzanne Pfeffer (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

– Provide some more complete quantitation of WPBs shape. For that they can measure the two Feret's diameters (short and long) for each granule, and from there the aspect ratio (in this case, higher aspect ratio means straighter WPB, lower aspect ratio means more twisted).

– Revise the text to be careful when claiming a specific role of PKD in this process.

– provide a rescue experiment on the effect of V0a1 depletion on PKD1 expression levels because the authors do not use a second, independent shRNA for V0a1 depletion.

Reviewer #1:

The authors have revised the manuscript and presented a detailed response to the reviewers' questions and comments. From my side, although I very much value the new data provided by the authors, I am still not fully convinced that all the conclusions that the authors claim in this manuscript are well sustained by the data. In particular, in terms of the statement that V0a1 promotes WPB formation through a PKD-dependent membrane fission. I am still quite positive on this manuscript, but I still have the following comments regarding the new data provided by the authors:

– Main point #1: While I do appreciate that the authors have presented quantitative information of WPB size in their revised manuscript, I'm not fully convinced that presenting a only the measurement of the large Feret's is the best way to quantitate shape of the WPB. E.g. in pg 9 they say that for a twisted WPB, one would expect a shorter Feret's diameter, which is true if the WPB have the same size, but it is not a direct measure of "twist" (different Feret's could be caused by other shape changes as compared to twisting). Also, the authors do not present the total distribution of Feret's but otherwise the rather arbitrary % of WPBs larger than 2um. I think that the phentype based on the few images provided is clear, but I think a better quantitative analysis could have been performed.

– Main point #3: Pg. 10 the authors say "In the absence of V0a1, WPBs 238 were unable to separate from the TGN, although the buds were formed (Figure 4).". I still am not convinced that this claim is sustained by the experimental data. Also, the new data on PKD1 overexpression is quite interesting. What about PKD2 and 3, have't the authors tested that? Is the high PKD1 expression an off.target effect (could the authors test it in the rescue experiments)? Also I'm puzzled on why the authors used an anti-PKD2 antibody in Figure 5supp4 and not an anti-PKD1, since they shown that PKD1 is upregulated in shV0a1 cells?

Reviewer #3:

The authors have put a lot of effort into revising the manuscript and have addressed the major criticisms. Particularly valuable in this regard are the experiments on intraluminal pH in the Golgi that support the hypothesis that WPB formation occurs at specific subdomains of the TGN. The experiments also revealed that PKD activity is not dependend on V0a1 and therefore the relationship between V0a1 and PKD activity remains unresolved. However, in my opinion, this does not prevent publication of the manuscript.

https://doi.org/10.7554/eLife.71526.sa1

Author response

Essential revisions:

The reviewer’s request:

1. Quantitative assessment of changes in WPB size and shape in V0a2/V0a1 depleted cells.

We thank the reviewers for this comment. To clarify the difference of WPB size and shape generated in the absence of V0a subunits, we have measured Feret’s diameter, the longest distance between any two points along individual WPB particles. Feret’s diameter is expected to be smaller if WPBs are twisted or in irregular shapes compared with relatively straight structures observed in the control. The data is shown as a percentage of WPBs of which Feret’s diameter is larger than 2 µm long (Figure 4B right). As expected from the confocal images, Feret’s diameter of WPBs in the V0a2-depleted cells was significantly smaller than those in the control. We also found that Feret’s diameter of WPBs in the V0a1-depleted cells were even smaller than those in the V0a2-depleted cells. We think that this represents the appearance of the rounded WPBs found in cell periphery in the V0a1-depleted cells (Figure 4A), although the WPBs accumulated around the TGN still retain the characteristic elongated shape. As the rounded WPBs were not that much observed when WPBs were visualized by the introduction of VWFD1-A1-EGFP and P-selectin-EGFP in the V0a1-depletion confirmed cells (Figure 4D,E), we concluded that they are WPBs formed before the V0a1 deletion is achieved. In regards to this, we have amend the text (Lines 195 to 199) and the corresponding figure legends (Lines 691 to 701).

2. Measure lumenal Golgi/TGN pH in WT and a1/a2 KD cells.

To respond to this comment, we have added new data introducing β-1,4-galactosyltransferase 1 (GalT), a well-characterized luminal TGN resident protein coupled with pHluorin, a pH-sensitive GFP derivative (GalT-pHluorin). This data has been included as Figure 3—figure supplement 2 in the revised manuscript. Bafilomycin A1 and concanamycin A exposure caused almost total neutralization of the luminal TGN pH, while the depletion of the V0a isoforms resulted in modest effects. These observations are consistent with our data showing that V-ATPase inhibition resulted in the blockage of VWF multimerization, but the depletion of V0a isoforms did not (Figure 2—figure supplement 1A, Figure 3—figure supplement 1B), because the multimerization requires acidic pH. The V0a2-depletion resulted in a slight but homogenous pH increase over the TGN (Figure 3—figure supplement 2B). In contrast, we have found relatively high pH increase in some, but not all of the TGN subdomains in the V0a1-deplted cells (Figure 3—figure supplement 2B), although no statistical difference was observed when the whole TGN pH was compared (Figure 3—figure supplement 2A). We think the impacts of V0a1-depletion in the pH regulation of particular TGN subdomains is interesting, because the main cellular pool was found in peripheral WPBs (Figure 1). WPBs were found to form a close contact with endosomes by electron tomography analyses. We think that this suggest a way how V0a1 is delivered to WPBs during the biogenesis, and the specific effects on some TGN subdomains suggest that WPB biogenesis occurs in particular subdomain of the TGN. In regards to the above results and discussion, we have amended the text in the results and Discussion sections (Lines 171 to 186, 320 to 322) as well as in the corresponding figure legend (Lines 844 to 853) and the Materials and methods section (Lines 446 to 452).

3. The claim that PKD is active in the shV0a1 cells is not supported by the data. Please measure PKD activity using more adequate assays (phospho-specific proteins etc) in the absence of V0a1. it seems that PKD in V0a1-depleted cells is predominantly at the Golgi. Therefore, detection of PKD activity by WB using the commercially available activation-loop Ab or an IVK should reflect Golgi activity. Alternatively, you could also use a genetically encoded reporter to measure PKD activity.

Thank you very much for the comments. As the reviewers pointed, there are several routes to supply PI4P to the Golgi/TGN other than the PKD-mediating pathway. So we agree that PKD activity should be evaluated by a more relevant assay in addition to the data using PI4P reporters (Figure 6C, Figure 6—figure supplement 1B,C). To see the PKD activity in the absence of V0a1, we performed western blotting using anti-phospho-PKD1(Ser744/748) and anti-total PKD1 antibodies (Cell Signaling Technology) as the reviewers suggested. These data have been included as Figure 6–supplement 2 in the revised manuscript. The depletion of V0a1 unexpectedly caused an increase of total PKD1 protein, whereas we were unable to see any detectable changes of the phospho-PKD1 level compared to the control. As the substantial amount of PKDs was observed on the nascent WPB in the absence of V0a1 compared with the cytosolic distribution in the control (Figure 5B-E, Figure 5—figure supplement 3D,E), an increase of PKD1 protein in the V0a1-depleted cells may be explained by a difference in the protein stability dependent on the subcellular localization. Alternatively, it might be an upregulation of PKD1 expression by a cell autonomous compensation or unknown mechanism mediated by the V0a1-depletion. Because of the low endogenous expression of PKDs, it is hard to test whether the PKDs observed on the accumulated WPBs in the V0a1-depleted cells are active/phosphorylated or not. Therefore, although total amount of cellular active PKD1 is unchanged independent from the absence or presence of V0a1, we were unable to clarify the requirement of V0a1 on the PKDs’ activation/phosphorylation. Regarding the above results and discussion, we have included appropriate text in the result section (Lines 275 to 297) and the corresponding figure legend (Lines 914 to 918).

4. Provide a more thorough discussion including how pH and lipid metabolism (DAG, PI4P) might be related.

As discussed above, V0a1 appears to be involved in the regulation of TGN pH. Although it is unclear yet if V0a1-mediating pH change triggers membrane fission of WPBs, we think that pH-dependent biogenesis of membrane-bound structures is feasible. We have shown that PI4P and DAG are observed in newly forming WPB buds; it is particularly prominent in the V0a1-depleted cells (Figure 6, Figure 6—figure supplement 1). As these lipids were not found in peripheral WPBs, they must be removed or converted into other lipids during the biogenesis, suggesting a dynamic lipid metabolism in WPB membrane. PI4P and phosphatidic acid, a lipid generated through single-step metabolism from DAG, are shown to work as pH biosensors in cells, and the pH-dependent interaction with particular proteins regulates cellular processes such as biogenesis of membrane-bound structures and intracellular trafficking (Shin et al., Dev Cell 2020, Young et al., Science 2010). Thus, we think that possible involvement of lipids and TGN luminal pH in the biogenesis of WPBs should be tested in the future. Regarding this, we have added a discussion (Lines 338 to 345).

Reviewer #1:Major comments:

As I stated above, I think the results are potentially interesting, but it is not clear to me what is the molecular mechanism by which V0a1 controls WPB biogenesis. The link with PKD is mild (as said earlier, there are reports indicating that PKD1 is dispensable for VWF secretion, which need to be cited and put in context of the present findings). Some ideas:

– V-ATPases regulate luminel pH. In the absence of V0a1 (or V0a2), is there any changes in WPBs pH? The authors reported that VWF still form multimers, but this only quite indirectly reports on the pH at the TGN/WPBs.

Thank you very much for your comments. We have measured intraluminal pH of the TGN where WPBs are generated. Please refer our response to Essential revision #2.

– About the involvement of PKDs: for the quantitation shown in Figure 5A (and related supps), it would help to count the number of WPBs in the cells expressing the PKD-KD mutants, similarly to what has been done in Figure 4B for the knockdowns of the V0as. And as said, put that in the context of the existing literature on the topic.

We have added the WPB counts of PKD-KD expressing cells as Figure 5—figure supplement 3C in the revised manuscript. Thank you very much for the pointing the paper that was missed in the original manuscript. It has been cited in the discussion in the revised manuscript (Line 333-334, Reference #74). As we observed essentially same effects in all PKDs, we consider that PKDs are redundant each other in the context of WPB biogenesis as described in the discussion (Lines 327 to 334).

– Line 240-41: according to the data, both V0a1 knockdown and PKD-KD expression show a similar phenotype (fission defect). What is the connection (if any) between these two processes?

Thank you very much for your comments. As the fission defect phenotype observed in the absence of V0a1 was not rescued even by the induction of wild-type PKDs (Figure 5B-E, Figure 5—figure supplement 3D,E), we believe that V0a1 must be a primary factor required for the fission, although the connection between V0a1 and PKDs is uncertain yet. We would like to address on this in our future study.

– Figure 6: PKD activity is assessed by the presence of PI4P on the TGN-associated VWF-positive buds. However, PKD is also phosphorylating OSBP, which reduces PI4P levels (actually, acute activation of PKD leads to less PI4P in the Golgi). There are other means to establish PKD activity (use of phospho-substrate antibodies, etc.)

Please refer our response to Essential Revision #3.

– Lack of mechanisms or upstream/downstream organisation of the V0as/PKD in fission. Do V0a localize to PKD-KD tubes? Is V0a1 a substrate of PKD?

We attempted to see if endogenous V0a1 protein is localized to PKD-KD tubes; however, we were unable to obtain adequate resolution data that allows us to draw a conclusion, because of the broad punctate distribution of endogenous V0a1 protein in cells.

Reviewer #2:This is a nice story that merits publication because it addresses a long standing issue on the export of bulky WPBs from the Golgi. The involvement of VO-ATPase subunits and PKD pathway makes this process amenable to molecular analysis. My gut feeling is that trafficking of WPBs is likely related to the export of other bulky cargoes like the collagens.

The migration or trafficking of WPB's remains a fascinating challenge. The data on different location of VOa1 and VOa2 are very interesting. The question I have is whether there are different pools of VO-ATPases? Can the authors isolate ( immuno precipitate) two different pools of VO-ATPase; one enriched in VOa1 and another in VOa2? Or did I get this wrong and these subunits are not in complex with other parts of VO-ATPase.

Thank you very much for your comments. V0 and V1 sectors of V-ATPase are shown to associate reversibly dependent on cellular condition such as nutrient status; suggesting that V0a isoforms found in cells are not always in "complete" V-ATPase complex. As shown in Figure 1—figure supplement 3A,B, we found that V1c and V1A, common subunits in the V0 and the V1 sectors, are predominantly observed in WPBs in HUVECs. Thus, it is likely that V0a isoforms found on WPBs are in complete V-ATPase complex. Thus, we believe that there are two distinct pools of V-ATPase complex in HUVECs; however, because of the lack of specific tools/antibodies, currently it is hard for us to test the hypothesis.

The data on the PKD-DAG are very interesting, but I am a bit confused about the involvement of clathrin. My understanding is that DAG-PKD pathway is independent of the clathrin mediated cargo export from the TGN. Is clathrin found on PKD-Kinase dead containing tubules that are enriched in WBPs?

As PKDs are shown to be involved in transport carrier biogenesis of various cargos, we believe that careful approaches are required to respond to the comments. We would like to address on this in our future study. Thank you very much for your constructive comments.

Finally, is this a budding event or is it that PKD-DAG sequester a part of TGN enriched in WBPs , while the unoccupied part concentrates cargoes exported by the clathrin pathway? The authors should explain this. It would be terrific to test whether a cargo destined to the endosomes/lyososomes is segregated from WPBs in the TGN.

Thank you very much for your comments. We also believe that cargos destinated to endosomes/lysosomes are segregated from WPBs in the TGN. In the revised manuscript, we have included a data showing that V0a1 appears to be involved in the pH regulation of some TGN subdomains. We think that it suggests WPB formation in particular subdomains of the TGN. We would like to address on this carefully together with #2-2 in our future study.

Reviewer #3:

Figure 1, Supplement 3F: V0a4-EGFP shows considerable co-localization with peripheral VWF-positive structures, however, the authors do not pursue this finding further claiming that V0a1 and V0a2 are the pre-dominant isoforms located on WPBs. The authors should have investigated a potential effect of V0a4 to rule out that this isoform is involved in WPB biogenesis/transport.

Thank you very much for your comments. As no suitable antibodies to V0a4 were available, we were unable to confirm whether endogenous V0a4 protein is localized on WPBs. Because of the specific phenotypes upon V0a1/V0a2-depeltion, we believe that V0a4 must have a distinct role other than the roles described in this study.

Figure 4: The finding that V0a1 is found on mature WPBs but its depletion affects WPB formation at the TGN is exciting. To make their data more solid the authors should have investigated if and how the pH in the Golgi/TGN changes upon depletion of V0a1. For example, they could have used a pH biosensor.

Please refer our response to Essential Revision #2.

Figure 5: The authors find that PKD family is involved in the biogenesis of WPBs. While these data are convincing the link to V0a1 is insufficient. Upon V0a1 knockdown more PKD is found on the TGN, which might indicate that more DAG is present. This is an interesting finding, however, not further pursued by the authors. The authors use several PI4P biosensors to show that PKD is enzymatically active. It is clear that there is increased binding of these biosensors to the TGN when V0a1 is depleted, however, many pathways besides PKD contribute to PI4P production and therefore additional evidence should have been provided.

Additionally, PI4P has been shown to sense the cytosolic pH and this impacts cargo sorting and trafficking. Acidification of cytosolic pH displaces PH domains from PI4P (Shin et al., Dev Cell, 2020). Given these results, the authors should have investigated whether silencing of V0a1 could also affect intracellular pH and thus the binding of PH domains to PI4P.

Please refer our responses to Essential Revision #2 and #3.

General comment: The authors make several references to morphological changes in WPB size and shape upon perturbation (e.g. Figure 3C). However, it is difficult to see these changes when looking at the images. In particular, in Figure 3A, the difference in VWF-positive structures between shControl and shV0a2 is not clear. Given the findings that number and size of WPBs might be affected by V0a1 and V0a2 depletion but mulitmerization and secretion are not impacted or even increased (Figure 3 – supplement 1) the authors should have considered to investigate VWF string formation as this is dependent on size of WPBs and VWF tubulation.

Regarding the WPB size and shape in the absence of V0a2, please refer our response to Essential Revision #1. We did not perform any analysis of the regulated secretion of VWF including VWF string formation using shRNA-introduced cells, because we think that use of those cells makes data interpretation difficult. As discussed in the manuscript (Lines 204 to 208), WPBs are storage organelles; that is, once formed, they remain in cell until they are being exocytosed or turned over. Efficient reduction of target proteins by shRNAs requires time. Thus, it is likely that shRNA-introduced cells contain not only the WPBs that are formed in the absence of the target protein, but also the pre-formed WPBs before the depletion is achieved. Therefore, the regulated secretion from shRNA-introduced cells must include sum of VWF from heterogenous populations of WPBs. And also, there is a possibility that their response to exocytic stimulants might be different. From the above reasons, we do not think we can find anything relevant from the experiments using shRNA-introduced cells in terms of the regulated secretion.

Figure 1: The authors use Bafilomycin A1, a specific vacuolar H+ ATPase (V-ATPase) inhibitor, to investigate the contribution of V-ATPase to WPB formation. They show that BafA1 treatment does not affect the amount basal and regulated VWF secretion in general but disrupts VWF multimerization. While the data shown are convincing they are in conflict with data published by Torisu et al., (2013) who showed that inhibition of autophagy by BafA1 treatment reduces the secretion of VWF from endothelial cells. Some discussion would be needed to clarify the discrepancy.

Thank you very much for your comments. To see the effects of V-ATPase blockage, we used HUVECs exposed to the V-ATPase inhibitors for up to 4 hours, because we found that relatively short exposure was enough to see the effects and that the prolonged exposure (16 hours) induces a formation of large vesicles/vacuoles that can be observed even under phase-contrast microscopy. In the paper, they used HUVECs exposed to bafilomycin A1 for 16 hours. So we think that the discrepancy is due to the difference in the conditions. In regards to this, we have added a note in the legend of Figure 1—figure supplement 2B (Lines 774-775).

Figure 5: The authors could use published PKD activity reporters, perform an in vitro kinase assay, or measure activation loop phosphorylation of PKD via Western Blot analysis to infer on PKD activity in V0a1 depleted cells. Also, some discussion on how V0a1 depletion might affect DAG levels at the TGN would be needed.

Please refer our responses to Essential Revision #3 and #4.

General:

The authors should assess changes in WPB size and morphology using quantitative morphometric analysis of the VWF-positive structures (see Lopes da Silva et al., JCellSci, 2016).

Please refer our response to Essential Revision #1.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

– Provide some more complete quantitation of WPBs shape. For that they can measure the two Feret's diameters (short and long) for each granule, and from there the aspect ratio (in this case, higher aspect ratio means straighter WPB, lower aspect ratio means more twisted).

Thank you very much for the comments. We have measured the longest (Fmax) and shortest (Fmin) Feret’s diameters of individual WPB particles to calculate Fmax/Fmin ratio as suggested. Frequency distribution of the ratio (Fmax/Fmin) as well as the longest Feret’s diameter (Fmax) has been shown as histograms in Figure 3—figure supplement 3 to present total WPB distribution. Obvious decrease in Fmax/Fmin is observed in WPBs formed in the absence of V0a2; indicating the irregular, twisted shape as seen in the images (Figure 3). Regarding this, we have amended text in the result section (Lines 201-206) and the corresponding figure legend (Lines 880-889). And we have removed a figure panel in Figure 4B showing a comparison of WPB proportions larger than 2 µm and the corresponding text (Lines 706-714).

– Revise the text to be careful when claiming a specific role of PKD in this process.

Thank you very much for the comments. Because we observed that the induction of wild-type PKDs does not support the segregation of WPBs in the absence of V0a1 (Figure 5B, Figure 5—figure supplement 3D,E), we think it indicates a primary role of V0a1 in the membrane fission of WPBs, although the molecular link between V0a1 and PKD in WPB biogenesis is still unclear. To respond to this comment, we have carefully amended the text throughout the manuscript including the title (Lines 2-3, 29-32, 275-276, 316, 362-366, 728-729, and 765-771).

– provide a rescue experiment on the effect of V0a1 depletion on PKD1 expression levels because the authors do not use a second, independent shRNA for V0a1 depletion.

Thank you very much for the comments. To respond to this, we have included new data in Figure 6—figure supplement 2B showing that a second, independent shRNA against V0a1 also caused an increase of PKD1 protein as similarly observed in the other shRNA. We also performed a rescue experiment by the induction of HA-Va01 as suggested, however, we were unable to observe any rescue effects on the PKD1 protein level. Because we observed the obvious rescue effects in the fission defect phenotype (Figure 4C), we believe HA-V0a1 must be functional in this context. We assume that HA-V0a1 might not be functional enough regarding the regulatory function of PKD1 protein level or alternatively the overexpression may cause something unexpected. In conclusion, as two independent shRNAs targeting distinct region of V0a1 mRNA (UTR and Tyr620-Met627) resulted in the similar induction, it is likely that V0a1 depletion itself causes an induction of PKD1 protein in HUVECs.

Reviewer #1:

The authors have revised the manuscript and presented a detailed response to the reviewers' questions and comments. From my side, although I very much value the new data provided by the authors, I am still not fully convinced that all the conclusions that the authors claim in this manuscript are well sustained by the data. In particular, in terms of the statement that V0a1 promotes WPB formation through a PKD-dependent membrane fission. I am still quite positive on this manuscript, but I still have the following comments regarding the new data provided by the authors:

Thank you very much for your comments. Please refer our response to the comment above.

– Main point #1: While I do appreciate that the authors have presented quantitative information of WPB size in their revised manuscript, I'm not fully convinced that presenting a only the measurement of the large Feret's is the best way to quantitate shape of the WPB. E.g. in pg 9 they say that for a twisted WPB, one would expect a shorter Feret's diameter, which is true if the WPB have the same size, but it is not a direct measure of "twist" (different Feret's could be caused by other shape changes as compared to twisting). Also, the authors do not present the total distribution of Feret's but otherwise the rather arbitrary % of WPBs larger than 2um. I think that the phentype based on the few images provided is clear, but I think a better quantitative analysis could have been performed.

Thank you very much for your suggestion. Please refer our response to the comment above.

– Main point #3: Pg. 10 the authors say "In the absence of V0a1, WPBs 238 were unable to separate from the TGN, although the buds were formed (Figure 4).". I still am not convinced that this claim is sustained by the experimental data. Also, the new data on PKD1 overexpression is quite interesting. What about PKD2 and 3, have't the authors tested that? Is the high PKD1 expression an off.target effect (could the authors test it in the rescue experiments)? Also I'm puzzled on why the authors used an anti-PKD2 antibody in Figure 5supp4 and not an anti-PKD1, since they shown that PKD1 is upregulated in shV0a1 cells?

Thank you very much for your comments. Regarding the sentence "In the absence of V0a1, WPBs were unable to separate from the TGN, although the buds were formed (Figure 4)", we have amended to " In the absence of V0a1, WPBs seemed unable to separate from the TGN, although the buds were formed (Figure 4)". Regarding the concern of off target effect of shRNA, please refer our response to the comment #C. As no suitable immunofluorescence antibodies to PKD1 were available, we were unable to confirm where endogenous PKD1 protein is localized in the absence of V0a1. Because we think that our results suggest a functional redundancy of PKDs in this process, we looked at PKD2 instead.

https://doi.org/10.7554/eLife.71526.sa2

Article and author information

Author details

  1. Yasuo Yamazaki

    Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, Osaka, Japan
    Contribution
    Conceptualization, Data curation, Funding acquisition, Investigation, Project administration, Validation, Visualization, Writing - original draft
    For correspondence
    yasuo.yamazaki@ncvc.go.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5297-9837
  2. Yuka Eura

    Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, Osaka, Japan
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5784-6339
  3. Koichi Kokame

    Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, Osaka, Japan
    Contribution
    Data curation, Funding acquisition, Project administration, Validation, Writing - review and editing
    For correspondence
    kame@ncvc.go.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9654-6299

Funding

Japan Society for the Promotion of Science (17K07401)

  • Yasuo Yamazaki

Japan Society for the Promotion of Science (17K07358)

  • Koichi Kokame

Ministry of Health, Labour and Welfare (JPMH20FC1024)

  • Koichi Kokame

SENSHIN Medical Research Foundation

  • Yasuo Yamazaki

Japan Agency for Medical Research and Development (JP20ek0210154)

  • Koichi Kokame

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Authors are grateful to Hiroko Shirotani-Ikejima and Taiko Kunieda for their technical assistance.

Senior and Reviewing Editor

  1. Suzanne R Pfeffer, Stanford University School of Medicine, United States

Reviewers

  1. Felix Campelo, The Barcelona Institute of Science and Technology, Spain
  2. Vivek Malhotra, The Barcelona Institute of Science and Technology, Spain

Version history

  1. Received: June 22, 2021
  2. Accepted: December 13, 2021
  3. Accepted Manuscript published: December 14, 2021 (version 1)
  4. Version of Record published: December 30, 2021 (version 2)

Copyright

© 2021, Yamazaki et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 968
    Page views
  • 196
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Yasuo Yamazaki
  2. Yuka Eura
  3. Koichi Kokame
(2021)
V-ATPase V0a1 promotes Weibel–Palade body biogenesis through the regulation of membrane fission
eLife 10:e71526.
https://doi.org/10.7554/eLife.71526

Further reading

    1. Cell Biology
    2. Developmental Biology
    Simon Schneider, Andjela Kovacevic ... Hubert Schorle
    Research Article

    Cylicins are testis-specific proteins, which are exclusively expressed during spermiogenesis. In mice and humans, two Cylicins, the gonosomal X-linked Cylicin 1 (Cylc1/CYLC1) and the autosomal Cylicin 2 (Cylc2/CYLC2) genes, have been identified. Cylicins are cytoskeletal proteins with an overall positive charge due to lysine-rich repeats. While Cylicins have been localized in the acrosomal region of round spermatids, they resemble a major component of the calyx within the perinuclear theca at the posterior part of mature sperm nuclei. However, the role of Cylicins during spermiogenesis has not yet been investigated. Here, we applied CRISPR/Cas9-mediated gene editing in zygotes to establish Cylc1- and Cylc2-deficient mouse lines as a model to study the function of these proteins. Cylc1 deficiency resulted in male subfertility, whereas Cylc2-/-, Cylc1-/yCylc2+/-, and Cylc1-/yCylc2-/- males were infertile. Phenotypical characterization revealed that loss of Cylicins prevents proper calyx assembly during spermiogenesis. This results in decreased epididymal sperm counts, impaired shedding of excess cytoplasm, and severe structural malformations, ultimately resulting in impaired sperm motility. Furthermore, exome sequencing identified an infertile man with a hemizygous variant in CYLC1 and a heterozygous variant in CYLC2, displaying morphological abnormalities of the sperm including the absence of the acrosome. Thus, our study highlights the relevance and importance of Cylicins for spermiogenic remodeling and male fertility in human and mouse, and provides the basis for further studies on unraveling the complex molecular interactions between perinuclear theca proteins required during spermiogenesis.

    1. Cell Biology
    2. Structural Biology and Molecular Biophysics
    Bronwyn A Lucas, Benjamin A Himes, Nikolaus Grigorieff
    Research Advance

    Previously we showed that 2D template matching (2DTM) can be used to localize macromolecular complexes in images recorded by cryogenic electron microscopy (cryo-EM) with high precision, even in the presence of noise and cellular background (Lucas et al., 2021; Lucas et al., 2022). Here, we show that once localized, these particles may be averaged together to generate high-resolution 3D reconstructions. However, regions included in the template may suffer from template bias, leading to inflated resolution estimates and making the interpretation of high-resolution features unreliable. We evaluate conditions that minimize template bias while retaining the benefits of high-precision localization, and we show that molecular features not present in the template can be reconstructed at high resolution from targets found by 2DTM, extending prior work at low-resolution. Moreover, we present a quantitative metric for template bias to aid the interpretation of 3D reconstructions calculated with particles localized using high-resolution templates and fine angular sampling.