Apical phosphatidylserine externalization in auditory hair cells

Abstract

In hair cells of the inner ear, phosphatidylserine (PS), detected with fluorescent annexin V labeling, was rapidly exposed on the external leaflet of apical plasma membranes upon dissection of the organ of Corti. PS externalization was unchanged by caspase inhibition, suggesting that externalization did not portend apoptosis or necrosis. Consistent with that conclusion, mitochondrial membrane potential and hair-cell nuclear structure remained normal during externalization. PS externalization was triggered by forskolin, which raises cAMP, and blocked by inhibitors of adenylyl cyclase. Blocking Na⁺ influx by inhibiting the mechanoelectrical transduction channels and P2X ATP channels also inhibited external PS externalization. Diminished PS externalization was also seen in cells exposed to LY 294002, which blocks membrane recycling in hair cells by inhibiting phosphatidylinositol 3-kinase. These results indicate that PS exposure on the external leaflet, presumably requiring vesicular transport, results from elevation of intracellular cAMP, which can be triggered by Na⁺ entry into hair cells.

• cAMP
• hair cells
• annexin V
• membrane cycling

Introduction

Hair cells of the inner ear, the sensory cells responsible for hearing and balance, are epithelial cells with specialized microvilli, called stereocilia, located on their apical surfaces. Sound moves stereocilia and gates an ionic current that affects the hair cell membrane potential. Hair cells have two major domains: an apical domain, bathed in high K⁺, low Na⁺, and low Ca^2 + endolymph, and a basolateral domain, bathed by plasma-like perilymph (Figure 1). Apical tight junctions not only precludes endolymph-perilymph exchange, but prevent proteins and lipids from readily diffusing between domains [1]

Figure 1.  Organization of the apical pole of an auditory hair cell. Stereocilia are located at the top of a hair cell and insert into the cuticular plate, a meshwork of actin filaments located between the stereocilia and the soma. Tight junctions connecting hair cells and supporting cells divide hair cells into apical and basolateral domains. This Figure is reproduced in colour in Molecular Membrane Biology online.

Biological membranes are highly asymmetric, with phospholipids non-randomly distributed between the two leaflets [2]. The outer leaflet primarily contains phosphatidylcholine and sphingomyelin, while the inner leaflet contains the aminophospholipids, including phosphatidylethanolamine and phosphatidylserine (PS). Phosphatidic acid and the phosphatidylinositols are also concentrated on the cytoplasmic side of the membrane. Mechanisms for maintaining transmembrane lipid distribution are only partially understood. An ATP-dependent aminophospholipid translocase (flippase) moves PS to the inner (cytoplasmic) leaflet of the plasma membrane. Another ATP-dependent translocase (floppase) moves lipids in the opposite direction. In addition, a nonspecific scramblase allows all types of phospholipids, regardless of headgroup, to rapidly equilibrate across the bilayer [3]. Exposure of PS on the external cell surface occurs in both normal and pathologic conditions [4], [5]. For example, PS exposure on platelet surfaces is fundamental to blood coagulation [5]. PS surface exposure in a hepatocytic cell line has been implicated in biliary lipid secretion [6]. Regional PS exposure in the acrosomal region of the spermatozoa is critical for binding to the zona pellucida, oocyte penetration, and signaling [7]. A subset of T lymphocytes with low levels of the tyrosine phosphatase CD45 externalize PS in response to P2X7 activation and other signals [8]; such cells are functionally normal. In a final example, massive degranulation of mast cells in response to IgE stimulation leads to reversible PS externalization [9]. Of course, PS exposure that follows a rapid rise in cytosolic Ca^2 + concentration and ATP depletion is also well known to precede apoptosis and necrosis [10], [11].

PS is partially exposed to the extracellular environment when bullfrog hair cells are isolated [12]. We used annexin V to examine PS exposure in hair cells of the mammalian cochlea. Annexin V is a ∼35 kD, Ca^2 +-dependent phospholipid-binding protein that has a high affinity for PS; it is often used as a probe for apoptosis [13]. Here we show that non-apoptotic PS exposure occurs in the apical plasma membrane of hair cells, and, like in other cell types [9], this PS externalization occurs by a mechanism related to vesicular trafficking.

Materials and methods

Tissue preparation

Guinea pigs (both female and male albino guinea pigs, 250–350 g) were anesthetized with an intramuscular injection of 1 ml/kg of 60 mg/ml ketamine, 2.4 mg/ml xylazine, and 1.2 mg/ml acepromazine in H₂O, then killed by decapitation. Cochleae were rapidly removed from the bulla, opened, and dissected in a Petri dish filled with a standard artificial perilymph composed of (in mM): NaCl 125, KCl 3.5, glucose 5, HEPES 10, CaCl₂ 1.3, and MgCl₂ 1.5. NaH₂PO₄ 0.51. The osmolarity of the solution was adjusted to 310 mOsm with NaCl and the pH was adjusted to 7.4 with NaOH. All experiments were performed at room temperature. A segment of the organ of Corti from the third turn of the cochlea was carefully removed and used as the experimental tissue. The organ of Corti was then transferred to a perfusion bath that was coated with silicon rubber (Sylgard 184, Dow Corning). The tissue was pinned to the silicone rubber layer in the bottom of a chamber (4 ml) that was equipped with a perfusion system. The procedures of this study were approved by the Animal Institutional Care and Use Committee of the Oregon Health & Sciences University.

Isolation of hair cells

The isolation procedure was essentially the same as described in our earlier study [14]. The anesthetized animal was decapitated, and the bulla and the otic capsule were opened. The apical and third turns of the cochlea were removed and transferred into a drop of perilymph solution containing collagenase (1 mg/ml Sigma Type IV) in a Petri dish for 10 min. After enzyme treatment, the tissues were washed three times in fresh perilymph. Hair cells were manually dissociated with a fine needle (WPI, Inc) and gently transferred into small Petri dishes for experiments. After dissociation, these cells are relatively depolarized (−13 to −15 mV [15]). We have not measured transduction currents from these cells, although in the experience of others with this preparation (e.g., ref. [16]), such currents will be at best extremely small.

Annexin V staining

In order to detect exposure of PS on the external layer of the plasma membrane, tissues and isolated hair cells were labeled with either annexin V-Alexa Fluor 488 or annexin V-Alexa Fluor 568 (Molecular Probes, Eugene, Oregon USA), by incubating tissues with 1 µg/ml fluorescent annexin V in the medium for 10 min. Tissues were washed and mounted. Labeling was visualized by fluorescence microscopy using 488 nm or 568 excitation and at 519 nm or 618 emission filters using a Nikon E 600 fluorescence microscope equipped with a CCD camera controlled by Metamorph software or with a Bio-Rad MRC 1024 confocal laser microscope system using a 40X, NA = 1.0 objective lens. To co-localize PS and the whole external leaflet of plasma membranes, isolated hair cells were placed in artificial perilymph solution containing 1 µg/ml annexin V and 10 µM N-(4-sulfobutyl)-4-(4-(4-(dipentylamino)phenyl)butadienyl)pyridinium (RH421; Molecular Probes, Eugene Oregon USA) for 10 min, then were washed with dye free solution. To detect mitochondrial activity, tissues and isolated hair cells were placed in an artificial perilymph solution containing annexin V and 100 nM of the membrane potential dye tetramethyl rhodamine methyl ester (TMRM⁺; Molecular Probes, Eugene, Oregon USA) for 10 min, and then were washed with dye-free solution.

Co-localization of F-actin and PS externalization

To detect whether PS externalization is correlated with cytoskeleton disassembly, tissues were labeled with fluorescent annexin V, fixed with 4% paraformaldehyde for 2 h, washed with phosphate-buffered saline (PBS) for 30 min, and permeabilized in 0.5% Triton-X 100 for 1 h. Tissues were incubated with either rhodamine-phalloidin or FITC-phalloidin (Molecular Probes, Eugene, Oregon USA) at 4 units/ml for 1 h, then rinsed with PBS for 10 min. Imaging was performed using confocal microscopy.

Fluorescent imaging of intracellular Na⁺

Intracellular Na⁺ concentrations were estimated by incubation with Sodium Green (10 µM) (Molecular Probes, Eugene, Oregon USA) for 15 min, followed by a wash with fresh standard perilymph solution. Sodium Green fluorescence signals were imaged with 488 nm excitation and 519 nm emission filters using confocal microscopy.

Cell viability measurements

Tissues were dissected and placed in a flow bath. After 10 min incubation, propidium iodide (PI, 1 µg/ml) was added to the solution for 10 min, following washout with fresh perilymph solution. Images were acquired with 568 nm excitation and 618 nm emission filters.

FM 4-64 fluorescence

To monitor the effect of LY 294002 on cell endocytosis and exocytosis, 1 µM FM 4-64 (Molecular Probes, Eugene, Oregon USA) was added into the artificial perilymph solution. Cells were observed on a Nikon microscope with a 40X, NA = 0.8 water-immersion lens. FM 4-64 was excited with band pass filters (peak 568 nm) and collected with an emission filter (peak 635 nm). The mean fluorescence intensity of FM 4-64 in the control and LY 294002 treated tissues, was determined by the software in a window approximately covering the entire inner hair cells in the four control tissue images (20 cells were selected in each sample picture and one sample picture from each animal) and four LY 294002 treated tissue images. A mean background was determined in a small window located away from fluorescence tissue, and was subtracted from the mean fluorescence intensity value. The data were averaged within the total four control images and the four LY 294002 treated tissue images.

Scala media perfusion

The pipette (5–7 mm tip diameter) was held in a hydraulic microdrive/micromanipulator and advanced into scala media by penetration through a slit in the round window and through the basilar membrane. The 3rd turn lateral wall of the cochlea was opened to allow the endolymph to escape. Perfusion used either artificial endolymph, composed of (in mM) KCl 126, NaCl 1, KHCO₃ 25, MgCl₂ 0.025 CaCl₂ 0.025, K₂HPO₄ 1.4, mannitol 25, or artificial perilymph, composed of (in mM): NaCl 125, KCl 3.5, glucose 5, HEPES 10, CaCl₂ 1.3, and MgCl₂ 1.5. NaH₂PO₄ 0.51; each solution contained annexin V-Alexa Fluor 568 (1 µg/ml). Perfusion was accomplished using a precision syringe pump at the rate of 200 nl/min for 40 min. After perfusion, animals were sacrificed and the organ of Corti was isolated; fluorescence signals from tissues were observed as above under a Bio-Rad MRC 1024 confocal laser microscope system using 40X and 60X objective lenses.

Reagent applications

To modulate PS externalization in hair cells of isolated organ of Corti, reagents were applied in the artificial perilymph solution. Unless indicated, the reagent remained in artificial perilymph throughout the entire experiment (Table I).

Table I.  Reagents employed.

Agent	Concentration	Product number and source
Z-VAD-FMK (pan-caspase inhibitor)	50 µM	260-020-M001, Alexis
Amiloride hydrochloride (MET channel inhibitor)	500 µM	A4562, Sigma
N-methyl-D-glucamine (NMDG; to replace sodium)	125 mM	M2004, Sigma
Pyridoxal-phosphate-6-azophenyl-2′,4′-disulfonate (PPADS; inhibitor of P2X2 channel)	100 µM	P178, Sigma
6-acetyl-7-deacetylforskolin (a cAMP agonist)	10 µM	A-9420, Sigma
SQ 22536 (adenylyl cyclase inhibitor)	500 µM	S153, Sigma
N-methyl-N′-nitro-N-nitrosoguanidine (MNNG)	500 µM	15427LO, Sigma
2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY 294002, PtdIns 3-kinase inhibitor)	10 µM	270-038-M001, Alexis

Statistical analysis

Data are presented as mean±SD; values for n refer to either the numbers of animals or the numbers of hair cells. Differences between data sets were assessed with the Student's t-test. Differences were considered significant at p<0.05.

Results

Apical PS externalization in hair cells

When we labeled external plasma membranes of isolated hair cells with RH 421 (Figure 2A) and fluorescent annexin V, we found that PS externalization – detected by annexin V – only occurred in apical membranes (Figure 2B). In a whole-mount preparation of the organ of Corti labeled with fluorescent annexin V, PS externalization was seen in apical domains of both inner hair cells (IHCs) and outer hair cells (OHCs) (Figure 2D). The intensity of fluorescence annexin V labeling of IHCs was substantially higher than of OHCs (Figure 2E).

Figure 2.  Confocal images of hair cells labeled with annexin V-Alexa fluor 488 and RH 421 in regular perilymph solution. (A–C) RH 421 labels the entire cell membrane, while annexin V labeling is restricted to the apical domain, including the membranes of the stereocilia and in the cuticular plate region. (D) annexin V labeling of a whole-mount organ of Corti. Annexin V labeled PS labels IHCs more intensely than OHCs. (E) quantification of annexin V labeling of 30 IHC and 114 OHCs. Labeling was significantly different (p<0.05). Bar: 10 µm.

When tissues were fixed and permeabilized prior to labeling, annexin V fluorescence was observed throughout cell membranes in all cells of the organ of Corti (Figure 3). The strongest fluorescence signals were still in the apical regions of IHCs and OHCs, however, particularly in IHCs (Figure 3). This result suggests either than the concentration of PS is higher in apical membranes or that PS was more readily extracted from basal membranes during permeabilization.

Figure 3.  Annexin V in fixed and permeabilized organ of Corti. (A) annexin V fluorescence was strong in OHC stereocilia (SC/OHC, stereocilia of outer hair cell). (B) Annexin V fluorescence in IHC stereocilia and cuticular plate region (SC/IHC, stereocilia of inner hair cell). (C) Annexin V labeling was present throughout the IHC cell membrane but was particularly strong at the apical pole; there was less annexin labeling in the basal poles (AP/IHC, apical pole of inner hair cell; BP/IHC, basal pole of inner hair cell).

Apical PS externalization of PS is not a sign of necrosis or apoptosis

PS externalization occurs rapidly in many cell types exposed to high calcium or ATP. Under such conditions, a cell may commit to cell death by either apoptotic or necrotic processes [10], [11]. To test if PS externalization foreshadows apoptosis or necrosis, we labeled cells with fluorescent annexin V and the DNA-binding dye propidium iodide (PI) (Figure 4), which detects the integrity of cell plasma membranes [17]. We never observed PI nuclear labeling of hair cells during the 0.5–3 h incubation in artificial perilymph, despite high levels of PS externalization (Figure 4D). By contrast, when we exposed the organ of Corti or isolated OHCs with N-methyl-N′-nitro-N-nitrosoguanidine (MNNG), a DNA-alkylating agent that potently activates cell apoptosis for 30 min [18], we observed apoptotic cells with PI staining in their nuclei and PS externalization throughout the plasma membrane. Figure 4E shows an example of a control isolated OHC where annexin fluorescence was only seen in the apical membrane. In contrast, Figure 4F shows an isolated OHC exposed to MNNG for 30 min; annexin V fluorescence was seen throughout the membrane and PI labeled nucleus.

Figure 4.  PS exposure is not linked to apoptosis or necrosis. (A–D) Organ of Corti bathed for 30 min in perilymph solution. (A) DIC image. (B) annexin V labeling. (C) annexin V, focusing on the IHC layer. (D) PI staining. No PI staining is observed in the level of the nucleus of either IHCs or OHCs. (E) PS externalization in the apical cellular membrane in an isolated control OHC. Representative of results from ten animals (∼50 imaged cells). (F) After treatment with MNNG, PS externalization was apparent in all membranes (arrows), including the basolateral membrane. In addition, the nucleus was labeled with PI. Two animals were used (3 cells imaged). (G) No significant difference in PS externalization between cells treated with an apoptotic inhibitor (z-VAD-fmk) or with a control solution for a 30-min incubation period (p>0.05; n=3 animals for each condition).

To further test if fluorescent annexin V labeling is an early apoptotic event, isolated tissues were exposed to artificial perilymph contained z-VAD-fmk (50 µM), a broad-spectrum caspase inhibitor [19], [20]. In many cell types, PS externalization is prevented upon inhibition of caspases [21], [22]. In contrast, we found no significant difference in fluorescence between the cells treated with z-VAD-fmk and untreated cells (Figure 4G).

PS externalization in the normal hair cells

In many types of cells, external exposure of PS correlates with cell damage. In this study, we found that the regional PS externalization occurred in apparently normal hair cells, which lacked cytoplasmic granulations, displaced nuclei, or swelling, and exhibited no organellar Brownian motion under DIC microscopy [23]. Figure 5A–5F shows that PS was externalized only in the apical domain of an OHC (Figure 5A) and IHCs (Figure 5D). PS externalization was correlated with high TMRM⁺ fluorescence signals where mitochondria were located (Figure 5B, 5E). The strong TMRM⁺ signals, combined with normal cellular morphology, indicated that hair cells are not damaged, despite PS externalization in the apical domain. By contrast, in swollen cells with organelles showing Brownian motion and their nuclei displaced, PS exposure was not regional but instead occurred in the entire cell membrane as shown in Figure 5G. The obviously damaged OHC of Figure 5G, without a hair bundle, had a significantly reduced mitochondrial membrane potential (no TMRM⁺ red signal).

Figure 5.  Mitochondrial membrane potential is apparently normal in cells showing apical PS externalization. A,D, PS apical externalization in an isolated OHC (A) and a group of IHCs (D) (annexin V-Alexa fluor 488, green). (B, E) normal mitochondria labeling pattern (TMRM⁺, red). (C, F) Merged images. (G) PS externalization over the entire cellular plasma membrane (annexin V-FITC, arrow) is accompanied with a loss of mitochondrial membrane potential (TMRM⁺, no red signal) in this damaged cell. Data are representative of ∼10 cells from two animals. Bar: 10 µm.

PS regional externalization is stimulated by intracellular Na⁺

When we, in vivo, perfused the cochlea's scala media compartment, which normally contains endolymph, with artificial endolymph containing fluorescent annexin V, there was very little labeling on apical membrane domains of hair cells (Figure 6B). By contrast, after endolymphatic perfusion of annexin V-containing artificial perilymph (which includes Na⁺), surface exposure of PS was considerably higher (Figure 6E). These results suggest that apical exposure of the organ of Corti to high Na⁺ induced PS external exposure.

Figure 6.  PS externalization after scala media perfusion in vivo. (A–C) scala media perfusion with artificial endolymph. (A) Phalloidin labeling of IHCs to detect F-actin, indicating normal stereocilia morphology. (B) Very weak annexin V-labeling. (C) Merge. In four perfusion experiments with artificial endolymph, we saw weak labeling of hair cells in two experiments and no labeling in two others. (D–F) Scala media perfusion with artificial perilymph. (D) Phalloidin labeling, showing normal stereocilia and cuticular plates. Arrows indicate IHC stereocilia (SC/IHC) and the cuticular plate (CP/IHC). (E) Annexin V labeling is increased in animals perfused with artificial perilymph. (F) Merge. Strong PS labeling in the cochlea's apical turn was observed in two of three perfusion experiments. (G) Experimental design of scala media (SM) perfusion; perfusion with artificial endolymph maintains the normal high K⁺ concentration at the apical surface of hair cells. (H) Perfusion with artificial perilymph produces a high Na⁺ concentration at the apical surface of hair cells. The natural perilymph in the scala vestibuli (SV) and scala tympani (ST) is not affected by the SM perfusion. Bar: 5 µm.

To determine the role of Na⁺ in PS externalization, we combined fluorescent annexin V labeling with detection of Na⁺ with Sodium Green dye. When we incubated tissue with an artificial perilymph solution at 60 min, PS labeling was high on the apical hair cell surface (Figure 7A). Apical blebs were also present, as previously described [24]. At the same time, the intracellular sodium fluorescent signal was high (Figure 7B). When tissues were incubated with NMDG-Cl solution, where Na⁺ was replaced by NMDG⁺, PS externalization was reduced substantially over the 60 min incubation period (Figure 7C). The intracellular Na⁺ signal remained low (Figure 7D). After the NMDG-Cl solution was replaced with standard perilymph solution and maintained for 10 min, annexin V labeling (Figure 7E) and Sodium Green fluorescence (Figure 7F) both rapidly increased. In another experiment, we bathed tissues in standard perilymph solution that contained amiloride (to block the mechanoelectrical transduction channel) and PPADS (to block P2X2 channels); this treatment prevents Na⁺ loading of IHCs (Figure 7H). Annexin V labeling was greatly decreased after incubation for 60 min with amiloride and PPADS (Figure 7G). After 10 min washout, Na⁺ fluorescent signals rapidly increased (Figure 7J) and annexin V label appeared on the surface of the majority of IHCs (Figure 7I). Thus, PS externalization was highly correlated with the degree of intracellular Na⁺ loading.

Figure 7.  Na⁺ entry is correlated with PS externalization in vitro. (A) Annexin V labeling is high in artificial perilymph solution (Na⁺ 125 mM). (B) Corresponding intracellular Na⁺ fluorescent signal of Sodium Green detection is high. C, annexin V labeling is reduced in NMDG-Cl (Na⁺ free) solution. (D) Corresponding Na⁺ levels are low with Sodium Green detection. (E) Annexin V labeling increases after replacing NMDG-Cl with regular artificial perilymph. F, corresponding increase in Sodium Green signal. (G) Blocking mechanotransduction and P2X2 channels with amiloride and PPADS reduces annexin V labeling. (H) Corresponding reduced Sodium Green signal. (I) Annexin V labeling increases after channel blockers were washed out. (J) Increase in Sodium Green signal. Note correlation of high annexin V labeling with high Sodium Green signal (arrows). Three animals were used for each condition (41 inner hair cells for NMDG treatment; 46 cells for P2X2–amiloride treatment). Bars: 10 µm.

Calcium levels are negatively correlated with PS expression

PS scrambling can be triggered by a rapid rise in cytosolic Ca^2 + concentration and ATP depletion [10], [11]. By contrast, using fluorescent annexin V and fluo-3 AM, a calcium indicator, PS externalization was negatively correlated with high intracellular Ca^2 + levels: cells with high fluo-3 fluorescence had little or no annexin V label (Figure 8).

Figure 8.  Cells with high levels of Ca^2 + have low PS externalization in vitro. (A) Annexin V labeling. PS externalization and blebs occur variably from cell to cell and result from an imbalance of endo- and exocytosis [24]. (B) Fluo-3 AM labeling. Note that high fluorescence signals indicating elevated Ca^2 + were associated with cells with no apparent external PS (e.g., cell 3) and vice versa (cell 4).

PS externalization depends on cAMP

We have shown that bleb formation on apical membranes of IHCs depends on levels of intracellular cAMP [24]. To test whether PS externalization follows adenylyl cyclase activation, we compared externalization with cyclase inhibitors and activators. When tissues were bathed for 40 min in perilymph solution, we observed that the Na⁺ fluorescent signal gradually increased (Figure 9A–9C) and correspondingly, PS was exposed on the surface of hair cells (Figure 9D–9F). However, when tissues were bathed in perilymph solution containing 500 µM SQ 22536, an adenylyl cyclase inhibitor, although intracellular Na⁺ levels gradually increased (Figure 9G–9I); annexin V labeling was barely detectable (Figure 9J–9L).

Figure 9.  PS externalization is inhibited by block of adenylyl cyclase with SQ 22536 in vitro. (A, B, C) Sodium Green fluorescence increased when tissue was bathed in artificial perilymph solution. (D, E, F) Annexin V labeling gradually increased. (G, H, I) Sodium Green fluorescence increased during SQ 22536 incubation. (J, K, L) Annexin V labeling did not change during SQ 22536 incubation. Similar results were seen in experiments from three animals. Bar: 50 µm.

No apical annexin V labeling was seen in organ of Corti cells bathed in NMDG-Cl solution for up to 60 min (Figure 10D). By contrast, bath application of 20 µM forskolin, an adenylyl cyclase activator, produced a gradual increase in annexin V labeling in every OHC and IHC, but not in supporting cells (Figure 10C, 10E). Surprisingly, the orientation of the hair bundles of some OHCs in the first and second rows appeared to rotate during the treatment (Figure 10C, 10E). Cell rotation was never seen when tissue were incubated in artificial perilymph (not visible in Figure 10A due to labeling intensities). The mechanism of the cell rotation is the subject of a separate report.

Figure 10.  PS exposure was stimulated directly by forskolin in vitro. (A–E) Organ of Corti was bathed in an NMDG-Cl solution to prevent Na⁺ entry. (A, B) Control conditions. No annexin V labeling was seen. (C, E) After introduction of 20 µM forskolin at time 0, annexin V labeling gradually increased (15 and 60 min). (D) In the absence of forskolin, annexin V labeling remained low at 60 min. (F) Annexin V labeling for control (solid line) and forskolin (dotted line). Results are representative of three experiments for each group. Bar: 50 µm.

To further test if PS exposure depends on vesicular trafficking, we treated tissues with LY-294002, a PtdIns 3-kinase inhibitor. LY-294002 not only prevents bleb formation initiated by Na⁺ loading [24], but also significantly decreased fluorescent annexin V labeling in hair bundles. Although it is also possible that LY 294002 may inhibit other pathways related to PS externalization, including tyrosine kinases downstream of PI3-K [25], [26], because LY-294002 inhibits uptake of FM 4-64 (Figure 11C–11D), our results strongly suggests that PS externalization depends on PtdIns 3-kinase controlled membrane trafficking (Figure 11A–11B). Under the conditions of our experiments, the principal pathway for FM 4-64 uptake is likely to be through apical endocytosis, not transduction channels [27].

Figure 11.  Inhibition of PtdIns 3-kinase with LY-294002 prevents PS externalization and slows endocytosis in vitro. (A) Annexin V labeling under control conditions. (B) Annexin V labeling after LY-294002 treatment. Annexin V labeling was inhibited. (C–D) Endocytosis measured with FM 4-64 uptake. (C) Control. (D) LY-294002. (E) Quantitation. LY-294002 significantly inhibited FM 4-64 uptake (p<0.05). Four experiments were carried out for each treatment.

Discussion

Apical membranes of outer and inner hair cells, including stereocilia, were strongly labeled with fluorescent annexin V, suggesting that these cells display PS on their outer leaflets in this region. Although PS is generally confined to the inner leaflet in mammalian cells, it is externalized to the outer leaflet during membrane fusion, aging, and apoptosis [4], [28]. In hair cells, PS surface exposure was not inhibited by caspase inhibitors, nor did the mitochondrial membrane potential or nuclear accessibility and morphology change during externalization. Thus, PS externalization in these cells is not a harbinger of apoptosis, but instead occurs when cells become Na⁺-loaded and cAMP levels increase. Our results are consistent with suggestions that membrane turnover in the apical domain of hair cells is extremely rapid [29].

Mechanism of PS externalization

Annexin V binding to PS requires Ca2⁺ concentrations of 20 µM or more [30]; indeed, we found no significant difference in annexin V fluorescence signals when tissues were bathed with 25 µM–1.5 mM extracellular CaCl₂. Rapid PS exposure and scrambling are generally observed after a rapid rise in cytosolic Ca^2 + concentration and ATP depletion [10], [11]; by contrast, we found that the highest annexin V fluorescence occurred in cells with the lowest intracellular Ca^2 + (Figure 8), opposite of what is expected for scramblases.

We suggest the following model. PS is thought to be more symmetrically distributed in intracellular vesicles, than in the plasma membrane [5]; following exocytosis, PS will be incorporated into both leaflets of the hair cell apical membrane. Under normal conditions, a flippase activity should rapidly move PS to the intracellular leaflet, so that at steady state, little PS will be exposed to the external environment. In the presence of elevated cAMP, however, either the exocytosis rate is increased or flippase is inhibited, or both, with the consequence that the flippase can no longer match the rate of exocytosis. A new steady state is achieved with a substantially greater PS concentration on the extracellular leaflet. In our experiments, Na⁺ entry triggers this process, elevating cAMP levels by an unknown mechanism. In turn, cAMP increases apical membrane exocytosis via a mechanism dependent on PtdIns 3-kinase.

Regulation of vesicle transport and exocytosis by cAMP has been reported in many cell types, including neurons [31]. cAMP-dependent PKA phosphorylation can enhance exocytosis by acting at different stages of vesicular transport, e.g., on the rate of vesicle cycling [32]. For example, PKA can affect protein-protein interactions important for vesicle fusion, including those involving trans-soluble N-ethylmaleimide-sensitive fusion protein attachment protein (t-SNARE), vesicle-soluble N-ethylmaleimide-sensitive fusion protein attachment protein (v-SNARE), and vesicle associated membrane protein 2 (VAMP2). Likewise, phosphoinositides modulate membrane trafficking [33]; we found that inhibition of PtdIns 3-kinase decreased PS externalization, consistent with a requirement for vesicular trafficking for PS exposure.

In an alternative model, high levels of Na⁺ might lead to decreased ATP levels due to overactive Na⁺/K⁺-ATPase; aminophospholipid translocase depends on ATP, and the hair bundle may be particularly sensitive to perturbation of cellular energy homeostasis due to its spatial isolation from mitochondria. Under these conditions, the aminophospholipid translocase might be incapable of matching the rate of PS externalization produced by exocytosis. In either case, PS exposure would be a consequence of rapid apical membrane turnover, which is also indicated by the slow formation of hair cell apical membrane blebs under these conditions, resulting from a slight imbalance between exo- and endocytosis [24].

Physiological role for PS externalization

PS is likely present only at low levels under normal conditions in the extracellular leaflet of hair-cell apical membranes. Is outer-leaflet PS physiologically relevant or simply a pathological consequence of Na⁺ loading? An important issue is under what conditions is Na⁺ or cAMP elevated in hair cells.

cAMP may be important in regulating hair-cell function. The resting force felt by mechanotransduction channels is reduced by cAMP [34]; similarly, spontaneous oscillations in bullfrog hair cells are inhibited by cAMP analogs [35]. A reduced cytoplasmic leaflet concentration of PS might limit clustering of myosin-1c [36], which would reduce resting force produced by the motor. Elevated cAMP levels could be a mechanism used by the hair cell to reduce the sensitivity of mechanotransduction.

Our experiments highlight the rapid turnover of apical membrane in hair cells, which has been noted repeatedly [27], [29], [37]. Why do these sensory cells turn their membrane over so rapidly? Hair cells release neurotransmitter continuously, and previous experiments have suggested an intimate connection between apical and basolateral membrane endo- and exocytosis in hair cells [27], [29]. Thus although the membrane composition in the apical and basolateral domains is likely to be substantially different, hair cells seem to cycle their membrane compartments in a coupled manner at rates substantially higher than those of most cells. Consistent with a higher total membrane turnover in synaptically active cells, inner hair cells were labeled with annexin V substantially more strongly than outer hair cells.

In conclusion, our data show a high concentration of PS in the apical domain of hair cells and demonstrate that (i) apical PS of hair cells becomes externalized in vitro; (ii) apical PS externalization is independent of cellular apoptosis and necrosis; (iii) apical PS externalization is activated by Na⁺ loading and a consequent increase in intracellular cAMP, and (iv) apical PS externalization requires membrane trafficking. We suggest that PS exposure reflects the remarkably fast turnover of apical membranes of hair cells [27], [29], which is stimulated to yet greater levels when Na⁺ enters hair cells.

This paper was first published online on prEview on 27 September 2006.

The authors appreciate the contributions of Bechara Kachar, M.D (NIDCD, National Institutes of Health) on plasma membrane dynamics in sensory hair cells and valuable discussions on the possible mechanisms of phosphatidylserine externalization in the apical area of hair cells. The authors also thank Prof. Robert A. Schlegel (Pennsylvania State University) for his advice and the important question of calcium control of phosphatidylserine externalization. The authors also would like to thank the editor-in-chief and two anonymous reviewers for their scientific critical review of the work. This research was supported by NIH-NIDCD grants P30 DC005983 (A.L.N and P.G.G.), R01 DC000105 (A.L.N), R01 DC000141 (A.L.N), R01 DC002368 (P.G.G.), and R01 DC00702 (P.G.G).