The Ca²⁺-activated Cl⁻ channel Ano1 controls microvilli length and membrane surface area in the oocyte

Oogenesis encompasses the growth phase of vertebrate oocytes during which time they accumulate macromolecular complexes essential for supporting early embryogenesis (Machaca, 2009). This is followed by a maturation phase before oocytes acquire the competency to be fertilized. Oocyte maturation encompasses the reductionist meiotic division in addition to cytoplasmic remodeling that prepares the mature egg for fertilization and the egg-to-embryo transition. In the frog oocyte, maturation is initiated in response to steroid hormones that induce a commitment to progressing through meiosis, until the process is arrested again at metaphase of meiosis II up to fertilization (Machaca, 2009; Nader et al., 2013). An important component of oocyte maturation is the remodeling of the Ca²⁺ signaling machinery to support a specialized Ca²⁺ transient at fertilization (Machaca, 2009; Nader et al., 2013). The Ca²⁺ signal at fertilization possesses specialized spatial and temporal dynamics that encode cellular processes, such as the block to polyspermy and the completion of meiosis, in a sequential fashion to mediate the egg-to-embryo transition. Ca²⁺ is in fact the universal signal for egg activation at fertilization in all sexually reproducing species investigated to date, although the fertilization-specific Ca²⁺ transient takes on different spatio-temporal dynamics in a species-specific fashion (Stricker, 1999).

Frogs have evolved two complimentary, but mechanistically distinct, Ca²⁺-dependent mechanisms to efficiently block polyspermy given their large egg size. The fast block to polyspermy occurs almost immediately after fertilization, and results from the stimulation of Ca²⁺-activated Cl⁻ channels (CaCCs) leading to membrane depolarization, which precludes further sperm–egg fusion (Cross and Elinson, 1980; Jaffe et al., 1983). On a slower time scale Ca²⁺ induces the fusion of cortical granules, which release enzymes that modify the egg's extracellular matrix and mediate the more sustained slow-block to polyspermy (Grey et al., 1974; Wolf, 1974).

CaCCs were first described in the Xenopus oocyte as regulators of the resting membrane potential, in addition to their role in the block to polyspermy (Duran and Hartzell, 2011; Machaca et al., 2001). CaCCs represent the primary current in the oocyte and their biophysical properties have been well characterized (Hartzell et al., 2005; Kuruma and Hartzell, 2000). CaCCs are broadly distributed among different species from invertebrates to mammals, and across diverse tissues in vertebrates (Kidd and Thorn, 2000; Robertson and Martin, 1996). Functionally CaCCs regulate multiple physiological processes, including epithelial secretion, vision, olfaction, vascular tone and neuronal excitability (Hartzell et al., 2005; Machaca et al., 2001). Our understanding of CaCCs was hampered for many years by controversy regarding their molecular identity; however, in 2008, the Xenopus oocyte CaCC was identified as TMEM16A using an expression cloning approach (Schroeder et al., 2008), and knockdown of endogenous Xenopus TMEM16A was shown to reduce native Ca²⁺-activated Cl⁻ current (Yang et al., 2008). Given that TMEM16A is an anion channel with eight transmembrane domains, Yang et al. renamed the protein anoctamin 1, or Ano1 (for anion+octa) (Yang et al., 2008). Mammalian genomes encode 10 anoctamin homologs. Ano1 is widely expressed in various epithelia where it is important for Cl⁻ secretion, as illustrated by secretion defects in Ano1 knockout mice (Ousingsawat et al., 2009). In addition to its role as a secretory channel in epithelia, Ano1 has also been shown to play important roles in gut smooth muscle, airway smooth muscle, nociception, the biogenesis of the primary cilium, and interestingly, Ano1 expression has been shown to correlate with cancer, providing a good biomarker for some cancers (Duran and Hartzell, 2011; Ruppersburg and Hartzell, 2014).

Here, we describe a newly identified function of Ano1 that is independent from its channel activity. We show that Ano1 regulates the length and diameter of microvilli in Xenopus oocytes through its interaction with ERM proteins and the actin cytoskeleton. The number of microvilli in the oocyte has been estimated to be ≈40×10⁶, and they expand the surface of the oocyte three- to fourfold (Kado et al., 1981). The structure of a microvillus has been extensively described in the intestinal brush border (Brown and McKnight, 2010; Crawley et al., 2014). Microvilli are shaped around parallel bundles of actin fibers that stabilize the thin, highly curved microvillus membrane against significant physical forces including surface tension, bending stiffness and membrane–cytoskeleton adhesion energy (Crawley et al., 2014; Nambiar et al., 2010). Single actin filaments are insufficient to overcome these forces, generating the need for actin bundles that are cross-linked and stabilized by a family of actin-binding proteins, including villin, fimbrin and espin (Crawley et al., 2014). In turn, actin bundles are physically linked to integral plasma membrane proteins and membrane lipids through two primary classes of proteins, the unconventional myosins and ERM proteins. The ERM (ezrin, radixin and moesin) family of proteins share 70-80% identity among the different members, and are highly conserved throughout the animal kingdom (Arpin et al., 2011). The ERM proteins bind membrane lipids such as PIP2 and also connect the cytoskeleton to a wide variety of transmembrane proteins such as ions channels (e.g. CFTR) (Guerra et al., 2005), G-protein coupled receptors (e.g. κ-opioid receptor and β2-adrenergic receptor) (Heydorn et al., 2004; Li et al., 2002), as well as adhesion proteins (e.g. ICAM1, CD44) (Bretscher et al., 2002). The ERM proteins share a common structure; the N-terminal (N_ter) domain [termed FERM (4.1 protein and ERM) or N-ERMAD (ERM associated domain)] is composed of three lobes that mediate ERM binding to the plasma membrane. The C-terminal, or C-ERMAD, domain contains the binding site for actin (see Fig. 3A). At rest the protein is folded and the C-ERMAD and FERM domains are in contact, resulting in auto-inhibition of the ERM crosslinking function. ERM proteins need to unfold to unmask the functional sites allowing binding to their specific partners, a process promoted by phosphorylation of a threonine residue in the C-ERMAD domain (Fehon et al., 2010). The interaction of the ERM protein with its target at the plasma membrane involves either direct binding to the cytoplasmic portion of membrane proteins, or requires linker proteins such as EBP50 (ERM-binding phosphoprotein 50) (also known as NHERF; Na-H⁺ exchanger regulatory factor) or E3KARP (NHE type 3 kinase regulatory protein) that contain a PDZ domain (Fehon et al., 2010).

Data presented herein reveal a newly discovered function of Ano1 that is independent of channel activity in modulating the structure of microvilli. Ano1 supports microvilli elongation through its interaction with ERM in the oocyte. This results in an expansion of membrane surface area as a function of Ano1 expression. Hence, Ano1 plays a crucial role in regulating membrane surface area in the oocyte by physically linking to the cytoskeleton through ERM proteins. Oocyte maturation is associated with the loss of microvilli resulting from endocytosis, a process that is crucial for the biogenesis of the first polarized epithelium leading to blastocoel formation.

The effect of expression of Xenopus laevis Ano1 (10 ng RNA) on CaCC currents in stage VI oocytes was evaluated using a voltage jump from the resting membrane potential of −30 mV to +40 mV (Fig. 1A). There was a significant increase in the resting current recorded at +40 mV in cells expressing Ano1 (Fig. 1B), that was associated with a decrease in membrane resistance (Fig. 1C) and a small but significant depolarization (Fig. 1D). At the resting membrane potential [i.e. −35.8±0.8 mV (mean±s.e.m.), n=42] the driving force for chloride ions is minimal, with the reversal potential for Cl⁻ (E_Cl) in our recording conditions being between −25 mV and −30 mV (Barish, 1983; Costa et al., 1989). Furthermore, the current through CaCC in the Xenopus oocyte displays a strong outward rectification at depolarizing potential and low Ca²⁺ concentration. Taken together, this explains the moderate effect of Ano1 expression on the membrane potential.

To increase intracellular Ca²⁺ and stimulate Ano1, we mobilized intracellular Ca²⁺ stores using ionomycin, and recorded Cl⁻ currents at the peak of the Ca²⁺ release phase (Courjaret and Machaca, 2014). As illustrated in Fig. 1A and E, the current recorded at +40 mV was greatly increased in response to intracellular Ca²⁺ release in Ano1-expressing cells. This indicates that the over-expressed Ano1 increases the Ca²⁺-activated chloride currents in the oocyte with similar properties to the endogenous CaCC, as previously reported in axolotl oocytes (Schroeder et al., 2008). Strikingly, however, these expected changes in membrane conductance properties following Ano1 expression were associated with a significant increase in the membrane capacitance (from 230.6±4.4 nF to 311.8±9.2 nF; Fig. 1F). This increase in membrane surface area is specific to Ano1, as overexpression of another ion channel, Orai1, does not affect membrane capacitance (Fig. 1G).

The theoretical surface area of a smooth oocyte of 1.2∼1.3 mm in diameter would result in a cell capacitance of 45–55 nF (with a conversion factor of 1 µF/cm²; Hille, 1992). This value is close to the measured capacitance of Xenopus eggs where the microvilli have been lost during oocyte maturation. When the stage VI oocyte matures, a significant portion of the membrane area in microvilli is internalized, leading to a loss of microvilli and flattening of the cell membrane (Kado et al., 1981). Whereas the apparent diameter of the cell does not change during oocyte maturation, the cell capacitance drops from 224.0±4.5 nF (n=51) to 74.3±2.8 nF (n=67) (see also Kado et al., 1981; Machaca and Haun, 2000; Peres and Bernardini, 1985). This indicates that the microvilli surface accounts for the majority of the cell capacitance.

We imaged oocyte microvilli and evaluated their dimensions under different conditions. We first measured the size and shape of microvilli on the animal pole of naïve cells using wheat germ agglutinin to stain the plasma membrane of live oocytes (El-Jouni et al., 2008). Cells were cooled to 4°C during the staining process to limit vesicular trafficking (Fig. 2A). This was then compared with membrane structure in cells expressing mCherry-tagged Ano1 by imaging the red fluorescence of mCherry. The first striking observation was the presence of very long microvilli and, on some cells, of large vesicular-shaped structures consisting of broad membrane extensions of shorter length, similar to the length of microvilli in control cells (Fig. 2A, arrowheads). These structures are likely fused microvilli. Both features were absent in naïve cells (Fig. 2A). We first quantified the maximum microvillus length in naïve cells, Ano1-expressing cells, and cells expressing the Ca²⁺ channel Orai1–GFP. The length of the 10 longer microvilli was measured in a single orthogonal section for each cell and averaged. There was a significant increase in the maximum microvillus length when Ano1–mCherry was expressed but not with Orai1–GFP (Fig. 2B). Additionally, the maximum microvillus diameter was also increased in Ano1-expressing, but not Orai1-expressing, cells (Fig. 2C). To obtain an estimate of the average increase in the length of the microvilli resulting from Ano1 expression, we analyzed confocal z-stacks by plotting the intensity of the fluorescent signal across orthogonal sections from the membrane intracellular boundary to the tip of the microvilli (Fig. 2D–F). The width of the curves at half-maximum intensity provides a good estimate for the average microvillus length, which was significantly increased in Ano1-expressing cells (Fig. 2F). In addition, we used a mesh reconstruction algorithm of the oocyte surface (Fig. 2G) that allows the quantification of the actual membrane surface. In naïve cells the estimated actual membrane surface was 2880±260 µm² (n=17) for a 400 µm² section, indicating that the convolutions increase the membrane surface by ∼7.2× at the animal pole of naïve cells. This value is larger than the theoretical value of an ∼fivefold increase for the whole cell but this probably results from the uneven distribution of the microvilli and their higher density at the animal pole, and that the measure of the surface is highly sensitive to the mesh parameters. In cells expressing Ano1 the increase factor reached ∼10.3× (Fig. 2H). The mesh analysis suggests that the increased length of microvilli in Ano1-expressing cells is associated with a decrease in their density. To obtain an estimate of microvilli density in control versus Ano1-expressing cells, we used the average microvilli length and maximal diameter measured as discussed above to calculate microvilli density for a theoretical oocyte of 1.25 mm in diameter (surface area of 4.9 mm², equivalent to 49 nF at 1 µF/cm²). The measured surface area of a control oocyte is ∼23 mm² (230 nF), yielding a total microvilli surface area of 18.1 mm². In control cells the average microvillus length is 1.74 μm with a maximum diameter of 0.42 μm. If we model the microvillus as a cone, this results in an individual microvillus surface area of 1.29 µm², and a total number of ∼14×10⁶ (low estimate) with a density of 2.9×10⁶ villi/mm². Ano1-expressing cells have a surface area of 31.2 mm² (312 nF) with an average microvillus length of 2.87 µm and a maximum diameter of 0.91 µm. This gives a microvillus surface area of 4.8 µm² and a total of 5.5×10⁶ microvilli with a density of 1.1×10⁶ villi/mm². This represents a two- to threefold reduction in microvilli density following Ano1 expression. These calculations are consistent with the imaging data (Fig. 2A), the mesh analysis (Fig. 2G) and the dome-like structures, which could result from several microvilli fusing together (Fig. 2A, arrowheads).

To assess whether Ano1 Cl⁻ channel function is required for its role in modulating microvilli length, we blocked channel function using the Ano1 inhibitor MONNA (Oh et al., 2013) during the expression of Ano1. MONNA (1 µM) effectively blocked the resting CaCC current in both naïve oocyte (Fig. 3A), and in oocytes expressing Ano1 (Fig. 3B). However, despite the fact that MONNA blocked the CaCC current it had no effect on the increased capacitance observed following Ano1 expression (Fig. 3C). This is also visible in the confocal images recorded on cells overexpressing Ano1 where no change in the microvilli density or structure is visible (Fig. 3D). Collectively, these results support the conclusion that Ano1 increases the membrane capacitance by increasing the length and diameter of microvilli while decreasing their overall density and that this effect does not require the activation of the channel.

Perez-Cornejo et al. comprehensively defined the Ano1 interactome, and identified the ERM protein family as robust Ano1-binding partners, thus providing a molecular link between the plasma membrane and the cytoskeleton (Perez-Cornejo et al., 2012). They also found that the interaction between Ano1 and moesin influences the presence of both proteins at the plasma membrane. As discussed in the Introduction the microvillus membrane associates with the actin cytoskeleton, allowing it to overcome the physical forces against its bending to fit the structure of the microvillus. The observation that Ano1 expression increases microvilli length suggests that Ano1 links the plasma membrane to the cytoskeleton through ERMs, thus supporting the increased microvillus length and area. To test whether ERM proteins provide the molecular link between Ano1 and the cytoskeleton within the microvilli, we expressed different human moesin constructs and tested the effect on membrane surface area. We chose moesin as the ERM protein as it is known to be expressed in Xenopus oocytes (Thorn et al., 1999). Human and Xenopus moesin share 88% overall identity, reaching 97% in the functional FERM domain and 100% in the F-actin binding site (Fig. 4A; Fig. S1). Cells were injected with cRNAs coding for three different proteins; full-length moesin (FL-moesin), a truncated form containing the FERM domain (the first 381 residues; N_ter-moesin), and a protein encoding the F-actin-binding domain (382 to C-terminal end; C_ter-moesin) (Fig. 4A) (Amieva et al., 1999). Oocytes tolerated the expression of FL-moesin well; however, prolonged overexpression of N_ter-moesin lead to depigmentation of the animal pole and cell death at longer time points (Fig. 4B). C_ter-moesin was poorly expressed (Fig. 6). Microvilli were visualized using wheat germ agglutinin (WGA) staining and orthogonal reconstruction of confocal z-stacks. There was no obvious effect of FL-moesin and C_ter-moesin expression on the plasma membrane (Fig. 4B). For cells expressing N_ter-moesin, membrane structure depended on the extent of the discoloration of the animal pole of the oocyte. In non-discolored cells, some microvilli were longer, although patches of the membrane lacked microvilli (Fig. S2). When the cells were discolored the membrane became more chaotic, showing large round-shaped structures stained by WGA and no visible microvilli (Fig. S2). Therefore, for the rest of the experiments with N_ter-moesin, analysis was limited to cells with no discoloration on the animal pole to avoid non-specific toxicity effects. The cell capacitance was significantly increased following expression of FL-moesin (Fig. 4C). In contrast, expression of the N-terminal truncation resulted in a reduction of membrane capacitance (Fig. 4C). The increase in membrane area following FL-moesin expression is likely a result of its crosslinking function between the cytoskeleton and plasma membrane. The decrease in membrane capacitance correlated with the levels of expression of N_ter-moesin in a dose-dependent fashion (Fig. 4D).

To better control the expression levels of the different moesin truncations, we further normalized for the cRNA copy number injected per oocyte, based on the size of the different constructs (5 ng full length, 4 ng N_ter-moesin and 3 ng C_ter-moesin). The results were comparable with those obtained on injection of 10 ng of cRNA; a slight increase in the capacitance for the expression of the FL-moesin and a reduction for the N_ter-moesin (Fig. S3A). There was no significant change in the membrane potential or in the membrane resistance (Fig. S3B,C). Importantly, expression of the different moesin constructs did not significantly alter the levels of the CaCC current at rest or after mobilization of store Ca²⁺ stores with ionomycin (Fig. S3D,E). This indicates that expression of FL-moesin, N_ter-moesin or C_ter-moesin does not significantly alter the membrane residence of Ano1, and that the observed decrease in membrane capacitance in N_ter-moesin-expressing cells results from a disruption of the molecular link between the cytoskeleton and the plasma membrane. The absence of any phenotype with C_ter-moesin expression likely results from the poor expression of this construct as shown below.

We next co-expressed Ano1–mCherry with FL-moesin and its fragments to evaluate their co-localization and influence on Ano1 expression and function. We did not observe any synergistic effect of co-expression of Ano1 with FL-moesin or C_ter-moesin, with both resulting in the same level of enhancement of membrane capacitance as observed with expression of Ano1 alone (Fig. 5B). In contrast, co-expression of N_ter-moesin with Ano1 reverses the enhancement of membrane surface area observed following Ano1 expression (Fig. 5A,B). We interpret this result to indicate that N_ter-moesin saturates the new Ano1 ERM binding sites generated following Ano1 overexpression, and as such inhibits microvilli elongation and the associated increase in membrane surface area. However, another potential explanation is that moesin is somewhat involved in the trafficking of Ano1 to the plasma membrane and with the co-expression of the dominant-negative N_ter-moesin, trafficking is inhibited leading to a block of the increase in surface area. However, this is not the case, as revealed by measurements of the amplitude (Fig. 5C) and density (Fig. 5D) of the CaCC current (measured at the peak of the ionomycin-induced Ca²⁺ release event). Co-expression of N_ter-moesin did not reduce the amplitude (Fig. 5C) or density (Fig. 5D) of the current, showing that overexpressed Ano1 reaches the plasma membrane to the same extent whether expressed alone or co-expressed with N_ter-moesin.

We further investigated the interaction between Ano1 and moesin in vivo using co-localization and immunoprecipitation. The different moesin constructs used above were tagged with GFP and co-expressed with Ano1–mCherry to evaluate their respective subcellular localization. When N_ter-moesin–GFP was expressed the discoloration of the cell and reduction of the microvilli could be observed only after substantially longer periods of expression (3 to 5 days as opposed to 24 to 48 h in the case of untagged N_ter-moesin), although the fluorescence was clearly observed earlier. We therefore used cells with clear expression of N_ter-moesin but without any discoloration for our analysis. In the case of the C_ter-moesin–GFP, the fluorescence of the GFP could be detected on some cells but was very faint or absent in most of them (Fig. 6A). The co-localization of the Ano1 and moesin was then quantified using orthogonal projections from z-stacks (Fig. 6A). Two values were used as a reference; for maximum co-localization we measured the values obtained from an experiment where Ano1–mCherry-expressing cells were stained with WGA. For minimum co-localization (exclusion) we used the Pearson's coefficient obtained from two proteins that are not located in the same cellular compartment, the Ca²⁺ sensor STIM1 (located in the endoplasmic reticulum) and the plasma membrane Ca²⁺ channel Orai1 (Fig. 6B, S/O) (Courjaret and Machaca, 2014). There was a high degree of co-localization between Ano1 and either FL-moesin or N_ter-moesin (Fig. 6B). For the C_ter-moesin the co-localization analysis was limited only to selected cells that expressed a significant GFP signal and showed reduced co-localization (Fig. 6B); however, this could be caused by poor expression of C_ter-moesin in the oocyte. Nonetheless, these results indicate that moesin is located at the plasma membrane and is bound through its N-terminal domain to its target. Next, to confirm that one of the targets of moesin in the plasma membrane is Ano1, we tested the ability of both proteins to co-immunoprecipitate. The experiments were performed on cells expressing the GFP-tagged moesin constructs. As illustrated in Fig. 6C, immunoprecipitation of FL-moesin–GFP or N_ter-moesin–GFP pulls down Ano1, revealing a single immunoreactive band close to the theoretical value of 113 kDa for Ano1. These results show that Ano1 interacts with moesin in vivo through its N-terminal domain leading to co-localization of both proteins to microvilli.

To further support the finding that moesin interacts with Ano1, we next performed FRAP experiments in order to measure Ano1 diffusion in the plane of the membrane (see Materials and Methods). We used cells expressing Ano1–mCherry with or without untagged moesin proteins and bleached a selected membrane area while monitoring the rate of fluorescence recovery over time (Fig. 7A,B). Proteins responsible for the recovery of fluorescence after photobleaching can originate from the cytoplasm (fast diffusion) and from the plasma membrane (slower) (Goehring et al., 2010). In this case, the bleaching was performed using a line scan over the z-axis and consequently, the intracellular pool of proteins under the membrane is bleached as well. The existence of a substantive intracellular Ano1 pool ready to fuse with the plasma membrane is also unlikely as, even under conditions of Ano1 overexpression, no detectable vesicular Ano1-positive fluorescence is visible (Fig. 6A). Consequently, most of the fluorescence recovery after photobleaching is likely to occur from lateral diffusion in the plane of the plasma membrane. There was a significant increase in the rate of recovery in cells expressing the N_ter-moesin but no major change for cell expressing the full-length or the C-terminal fragment (Fig. 7B,C). This indicates that the linkage of Ano1 to the cytoskeleton through ERM reduces its mobility in the membrane and is disrupted when the N_ter-moesin competes for the Ano1 binding site.

To confirm the role of ERM–Ano1 interactions in mediating the increase in membrane surface area following Ano1 expression, we sought to express different fragments of Ano1 that would act as binding sites for ERM. Given that the Ano1 binding site for ERM proteins is not known, we expressed the large cytoplasmic N-terminal fragment of Ano1 (amino acids 1–322) and a longer fragment that ends just before the presumed third transmembrane domain of Ano1 (Fig. S4). Expression of the 1–322 fragment of Ano1 effectively reduced membrane capacitance (Fig. S4C), without affecting CaCC current (Fig. S4B). This supports the role of ERM–Ano1 interactions in modulating membrane surface area.

To increase the exchange capacity of a cell while limiting volume variations (i.e. increase the surface/volume ratio) epithelial cells, oocytes, lymphocytes, hepatocytes and placental cells grow thin extensions termed microvilli. They extend the membrane surface to facilitate exchanges (absorption and secretion), and depending on the cell type, can also serve mechanical functions, including physical support and mechanotransduction. Densely packed microvilli form a brush border extending the surface area of epithelial cells such as the ones lining the small intestine and the proximal tubule of the kidney by ∼tenfold (Crawley et al., 2014; Helander and Fändriks, 2014). Xenopus oocytes are covered with microvilli that vary in shape and density during the growth and development of the gamete. This is particularly obvious following oocyte maturation in preparation for fertilization. During that process the number of microvilli is drastically and rapidly reduced (2000/s), inducing a large drop in the membrane capacitance in a few hours, without any change in the apparent diameter of the oocyte (Kado et al., 1981; Machaca and Haun, 2000). This rapid internalization of membranes forming the microvilli generates an intracellular vesicular pool that is essential for the rapid cellularization during early embryogenesis (Müller and Hausen, 1995). Furthermore, this membrane internalization forms the initial step in the biogenesis of the first polarized epithelium in the embryo that supports vectorial transport of Na⁺ and water into the blastocoel (Muller, 2001). This is because endocytosis during oocyte maturation is associated with a selective internalization of membrane proteins resulting in two populations of membranes; the egg plasma membrane, which represents the future apical membrane of the epithelium, and the internalized vesicular pool, which represents the future basolateral portion, with distinct protein content. For example, Orai1 and PMCA are internalized during oocyte maturation but not Ano1 (El-Jouni et al., 2008; Yu et al., 2009; Yu et al., 2010).

Here we show that Ano1 regulates microvilli length and, thereby, membrane surface area, independently of its channel function. Ano1 expression increases the oocyte surface area and microvilli length and diameter (Figs 1 and 2). The Ano1-dependent effect on microvilli is mediated through ERM proteins, as co-expression of the dominant-negative N-terminal domain of moesin with Ano1 abolishes the Ano1-dependent increase in microvilli length and surface area (Fig. 5). Furthermore, full-length moesin and the N-terminal domain co-localize and co-immunoprecipitate with Ano1 (Fig. 6). In addition, co-expression of Ano1 with N_ter-moesin increases Ano1 lateral mobility in the plane of the membrane, supporting the conclusion that moesin links Ano1 to the cytoskeleton, and as such anchors it and limits its lateral diffusion (Fig. 7). Collectively, our results support a model in which Ano1 traffics to the microvillus, where it binds ERM proteins and cross-links to the cytoskeleton. This stabilizes the microvillus membrane, revealing an important role of Ano1 in microvilli biogenesis in the oocyte.

Not surprisingly, expression of N_ter-moesin alone decreases membrane surface area (Fig. 4). This likely partly results from inhibition of Ano1–ERM interaction, but also from the inhibition of other interactions between ERM and plasma membrane proteins and lipids. This is highlighted by the discoloration observed in the oocyte at high levels or after prolonged expression of N_ter-moesin (Fig. 4). Interestingly, however, co-expression of Ano1 with full- length moesin does not show a synergistic effect over Ano1 expression alone. This could be caused by a physical limit on the length of the microvilli, independent of the amount of available scaffolding proteins. The effect of Ano1 on microvilli can be readily dissociated from its channel function. Blocking the activation of Ano1 using MONNA does not affect the ability of overexpressed Ano1 to increase microvilli length and membrane capacitance. Conversely, when the Ano1–ERM link is perturbed, either by expressing the dominant-negative N_ter-moesin mutant (Fig. 5) or the Ano1 N-terminal cytosolic fragment (Fig. S4), the global chloride current is not affected, yet the increased capacitance resulting from Ano1 expression is reversed. This shows that Ano1 regulates microvilli length through a physical/scaffolding interaction and not through its Cl⁻ property of conducting channel ions.

The role of plasma membrane channels in shaping the morphology of microvilli is largely unknown, with the exception of channelrhodopsin-2, which has been shown to enlarge microvilli (Zimmermann et al., 2008). Expression of other plasma membrane channels, such as CFTR or the sodium channel ENaC, increases the cell capacitance, but this process is attributed to the increased fusion of transport vesicles at the plasma membrane (Awayda, 2000; Weber et al., 1999). Several integral membrane proteins are involved in shaping the microvillus structure through interactions with ERM proteins such as CD43, CD44, ICAM-2 (Bretscher et al., 2002; Yamane et al., 2011; Yonemura et al., 1999) or CD9 in the oocyte (Runge et al., 2007). The microvillus is a spatially defined compartment with some proteins, including ions channels, localizing to microvilli (or even in microvillus sub-regions), through binding to intermediate scaffolding proteins such as ERMs (Lange, 2011). The binding of membrane proteins to the cytoskeleton would therefore have the dual role of stabilizing the microvilli by reinforcing the scaffolding link between the plasma membrane and the actin backbone, and of precisely distributing channels to specific functional locations.

What is the function of the anchoring of Ano1 to the cytoskeleton? The distribution of Ano1 is not homogenous in the oocyte and it is known that CaCCs localize preferentially to the animal pole of the oocyte at the expense of the vegetal pole (Machaca and Hartzell, 1998). This might in fact explain the increased microvilli density at the animal compared with the vegetal pole of the oocyte (Bluemink et al., 1983). We recently showed that Ca²⁺ influx through store-operated Ca²⁺ entry (SOCE) specifically activates distally located Ano1 through Ca²⁺ transfer within the ER lumen and release through IP₃ receptors (Courjaret and Machaca, 2014). We refer to this as ‘mid-range Ca²⁺ signaling’ as it spatially bridges elementary and global Ca²⁺ signaling. SOCE ‘clusters’ do not co-localize with Ano1. Consequently, the clusters are unlikely to make it to the microvilli and constitute a specific compartment in terms of Cl⁻ and Ca²⁺ signaling. The anchoring of Ano1 to the cytoskeleton would therefore allow a very precise localization of the channel in the vicinity or away from specific Ca²⁺ release sites.

Another important parameter is the function of the microvilli. In the oocyte the microvilli contact the macrovilli arising from the follicle cells and protruding through the vitteline envelope (Browne and Werner, 1984; Dumont and Brummett, 1978). At the contact point, gap junctions allow the passage of numerous molecules, particularly IP₃ and Ca²⁺ produced by the stimulation of PLC in follicular cells, a complex coupling that varies depending on the subtypes of follicular cells involved and/or gap junctions (Arellano et al., 2012; Sandberg et al., 1992; Supplisson et al., 1991). Precise positioning of Ano1 within the microvilli by anchoring it to the cytoskeleton is likely to enhance the efficiency of a signal transferred from the follicular cells to the oocyte through gap junctions, avoiding cytoplasmic buffers.

In summary, we have newly identified a function of Ano1 that regulates microvilli length, diameter and density within the oocyte through linkage to ERM proteins and the actin cytoskeleton. This in turn modulates the surface area of the cell. Given the wide distribution of Ano1 in different epithelia, which are often enriched in microvilli, and the conservation of the role of ERMs and the actin cytoskeleton, these findings are likely to translate to most epithelia.

Stage VI oocytes were harvested from wild-type and lab-bred Xenopus laevis (Xenopus Express, France and Nasco, U.S.A) using previously described procedures (Machaca and Haun, 2000). Oocytes were kept in 0.5× L15 (Sigma) and used no earlier than 24 h after surgery. Animals were handled according to Weill Cornell Medical College IACUC approved procedures.

Human moesin (Accession number: NP_002435, a gift from Criss Hartzell, Emory University, Atlanta, GA) was sub-cloned from pcDNA3.1 through Xba1 and HindIII restriction sites into the pSGEM vector using the following set of primers: 5′-ACTGCTCTAGAATGCCCAAAACGATC-3′; 5′-TCGATAAGCTTTTACATAGACTCAAATTCG-3′. Human moesin N-terminus (1–381) was sub-cloned from pcDNA3.1 through Xba1 and HindIII restriction sites into the pSGEM vector using the following set of primers: 5′-ACTGCTCTAGAATGCCCAAAACGATC-3′; 5′-TCGATAAGCTTTTAACGCTTCCGTTCCTGC-3′. Human moesin C-terminus (382–577) was sub-cloned from pcDNA3.1 through Xba1 and HindIII restriction sites into a psGEM vector using the following primers: 5′-ACTGCTCTAGAATGGCCCAGAGCGAGG-3′; 5′-TCGATAAGCTTTTACATAGACTCAAATTCG-3′. To create moesin–GFP fusion constructs, human moesin, moesin (1–381) and moesin (382–577) were sub-cloned from pcDNA3.1 using SacII and XbaI in-frame into pSGEM-GFP. The primers used were: for pSGEM-Moesin-GFP, 5′-ACTGCCCGCGGATGCCCAAAACGATC-3′, 5′-TCGATTCTAGACATAGACTCAAATTCGTCAATGCGCTGC-3′; for pSGEM-Moesin (1-381)-GFP, 5′-ACTGCCCGCGGATGCCCAAAACGATC-3′, 5′-TCGATTCTAGAACGCTTCCGTTCCTGCTCAAGTTC-3′; for pSGEM-Moesin (382-577)-GFP, 5′-ACTGCCCGCGGATGGCCCAGAGCG-3′, 5′-TCGATTCTAGACATAGACTCAAATTCGTCAATGCGCTGC-3′

The ionic currents were recorded using standard two-electrode voltage-clamp recording technique. Recording electrodes were filled with 3 M KCl and coupled to a Geneclamp 500B controlled with pClamp 10.5 (Axon Instruments). CaCC currents were recorded using depolarizing pulses to +40 mV and a previously described ‘triple jump’ protocol (Courjaret and Machaca, 2014; Machaca and Hartzell, 1999). Membrane capacitance was monitored using the built-in routine from pClamp to measure membrane parameters. Cells were continuously superfused with saline during voltage-clamp and imaging experiments using a peristaltic pump. The standard extracellular saline contained (in mM) 96 NaCl, 2.5 KCl, 1.8 CaCl₂, 2 MgCl₂, 10 HEPES, pH 7.4.

Confocal imaging of live oocytes was performed using a LSM710 (Zeiss, Germany) fitted with a Plan Apo 63×/1.4 oil immersion objective or a TCS SP5 (Leica, Germany) fitted with a 63×/1.4-0.6 oil immersion objective. Confocal z-stacks were taken in 0.5 µm sections using a 1 Airy unit pinhole aperture. The images were recorded using Zen black 2012 (Zeiss) or LAS AF 2.4.1 (Leica) software. Micrographs of oocytes were taken using a Zeiss Lumar V12 Stereomicroscope controlled using Zen blue 2012 (Zeiss). Images were analyzed with ImageJ 1.46q (Schneider et al., 2012). Cell surface calculation was performed using the ‘iso-surface’ function of the Bone J plugin (Doube et al., 2010). A 400 µm² image plane was selected and a mesh created to fit the 3D reconstruction of the surface. A resampling factor of 2 was used and a detection threshold for surface change of 20 pixels. The results are expressed as fold increase of the theoretical surface area (measured surface/400).

For WGA staining, the cells were cooled down to 4°C for 15 min then stained for 30 min at 4°C in Ringer solution containing 5 µg ml⁻¹ of Alexa-Fluor-488−WGA conjugate (Cat no. W11261, Thermo Fischer Scientific). The cells were then washed in Ringer solution and kept on ice prior to imaging to minimize internalization of the membrane marker.

For FRAP experiments, orthogonal images across the membrane were obtained directly using x-z scans: a line scan is performed at different z positions and the z -positioning is achieved using a ‘Super-Z’ galvanometric stage (Leica) that allows for fast monitoring of the fluorescence across the plasma membrane. To bleach a defined region of the membrane, the zoom factor of the scanner was increased two times and the laser power (561 nm) set at 100% for three scans. Scanning speed was set a 400 Hz and image frequency to 0.1 Hz for both bleaching and recording.

Oocytes were lysed in lysis buffer containing Triton X-100 (10 mM Hepes, 150 mM NaCl, 1 mM EGTA, 0.1 mM MgCl2, 0.5% Triton X-100, pH 7.4). The immunoprecipitation was performed by magnetic associated cell sorting (MACS) using microbeads according to the manufacturer's instructions (Miltenyi Biotec, Germany). The lysates were incubated with micro beads conjugated to an anti-GFP monoclonal antibody (µMACS GFP Isolation Kit 130-091-125; Miltenyi Biotec) for 30 min on ice. The lysate–beads complexes were then sorted in the magnetic field of a µMACS (Miltenyi Biotec). After the washing steps, the proteins were eluted with the denaturing elution buffer provided in the kit. Western blotting was performed as previously described (Courjaret et al., 2013) and repeated three times on different samples. The anti-Ano1 antibody was used at a dilution of 1:1000 (SAB2102460, QC13356 Sigma).

Values are given as means±s.e.m. Statistical analysis was performed when required using either Student’s paired and unpaired two-tailed t-test or ANOVA and corrected for multiple comparison. P-values are indicated as follows: *P<0.05; **P<0.01; ***P<0.001 and ns, not significant. Statistics were obtained using Prism 6.05 (GraphPad Software, La Jolla, USA).

We are grateful to Criss Hartzell for sharing the human moesin clones.