Plasticity in Membrane Cholesterol Contributes toward Electrical Maturation of Hearing*

Advances in refining the “fluid mosaic” model of the plasma membrane have revealed that it is wrought with an ordered lipid composition that undergoes remarkable plasticity during cell development. Despite the evidence that specific signaling proteins and ion channels gravitate toward these lipid microdomains, identification of their functional impact remains a formidable challenge. We report that in contrast to matured auditory hair cells, depletion of membrane cholesterol in developing hair cells produced marked potentiation of voltage-gated K+ currents (IKv). The enhanced magnitude of IKv in developing hair cells was in keeping with the reduced cholesterol-rich microdomains in matured hair cells. Remarkably, potentiation of the cholesterol-sensitive current was sufficient to abolish spontaneous activity, a functional blueprint of developing and regenerating hair cells. Collectively, these findings provide evidence that developmental plasticity of lipid microdomains and the ensuing changes in K+ currents are important determinants of one of the hallmarks in the maturation of hearing.

Advances in refining the "fluid mosaic" model of the plasma membrane have revealed that it is wrought with an ordered lipid composition that undergoes remarkable plasticity during cell development. Despite the evidence that specific signaling proteins and ion channels gravitate toward these lipid microdomains, identification of their functional impact remains a formidable challenge. We report that in contrast to matured auditory hair cells, depletion of membrane cholesterol in developing hair cells produced marked potentiation of voltagegated K ؉ currents (I Kv ). The enhanced magnitude of I Kv in developing hair cells was in keeping with the reduced cholesterol-rich microdomains in matured hair cells. Remarkably, potentiation of the cholesterol-sensitive current was sufficient to abolish spontaneous activity, a functional blueprint of developing and regenerating hair cells. Collectively, these findings provide evidence that developmental plasticity of lipid microdomains and the ensuing changes in K ؉ currents are important determinants of one of the hallmarks in the maturation of hearing. The matured auditory system is equipped with salient features that are essential for establishing the temporal fidelity required for speech and pitch perception (1). To achieve the post-hearing phenotype, pre-hearing hair cells undergo a precise and conserved transition from spontaneous action potentials (SAPs) 2 to graded generator potentials (2,3). The importance of SAPs in the developing auditory system for synaptic refinement is reflected in the recapitulation of the preserved process in regenerating hair cells (4). Moreover, the electrical transition to graded receptor potential, which occurs at embryonic day (E) 18 in the chick and postnatal day (P) 12 in the mouse, coincides perfectly well with increased hearing sensitivity (3,(5)(6)(7)(8). Previous studies have shown that several contributory conductances emerge and disappear in developing hair cells to culminate in the cessation of SAP. The functional expression of low voltage-activated K ϩ currents (I K,n ) derived from K v 7.4 channels and Ca 2ϩ -activated K ϩ currents at later stages of development clamps the membrane potential close to the reversal potential of K ϩ (2,3,9,10). Additionally, a reduction in the Ca 2ϩ current density and the late-stage disappearance of low voltage-activated Ca 2ϩ current (4) both contribute toward the transition from SAPs to graded receptor potentials, yet the underlying mechanisms for the latestage functional appearance of K ϩ currents are unknown, despite evidence to demonstrate that K ϩ channel genes and products are expressed several days prior to hair cell maturation (11,12).
A revised version of the structure of the plasma membrane consists of ordered microdomains rich in cholesterol, sphingolipids, and saturated phospholipids (13,14), which may be organized by scaffolding proteins (15,16). These cholesterol/ sphingolipid-rich domains are thought to confer spatial segregation of signaling pathways in cells (13). Indeed, several ion channels gravitate toward these cholesterol-rich "icebergs" in the sea of lipids, which regulate channel protein functions or their cell-surface expression (18,19). Moreover, the composition of the lipid microdomains undergoes drastic and consistent plasticity during neurodevelopment, serving as stagespecific markers for certain lineages of cells (20 -22). The plasma membrane of hair cells has been shown to contain lipid microdomains (23). Furthermore, hair cell membranebound motor proteins such as prestin and Myo1c have biochemical properties that are consistent with their association with lipid rafts (24 -28), raising the possibility that lipid microdomains and their potential developmental plasticity play important roles in hair cell functions.
We report for the first time that depletion of cholesterol produces a differential modulation of voltage-gated K ϩ currents (I Kv ) in developing hair cells compared with matured cells. Significantly, the data are in keeping with the developmental plasticity of cholesterol in plasma membranes of developing and matured hair cells. The ensuing impact of cholesterol depletion is a profound attenuation of SAPs in developing hair cells. We propose that alteration of the lipid composition of membrane lipids is fundamental to the transition from SAPs to graded receptor potentials, an important feature in the maturation of hearing.

EXPERIMENTAL PROCEDURES
Isolation of the Chicken Basilar Papilla-This investigation was performed in accordance with the guidelines of the Animal Care and Use Committee of the University of California Davis. Basilar papillae were isolated as described previously (4). All experiments were performed within 5-45 min of isolation. This study included chickens at different stages of embryonic development, ranging from E6 and E21, as well as post-hatched chickens. Fertilized eggs were incubated at 37°C in a Marsh automatic incubator (Lyon Electric). Before experiments, chicken embryos were killed and staged according to the number of somites present. Basilar papillae were isolated as described previously (4). The preparations were dissected in oxygenated chicken saline containing 155 mM NaCl, 6 mM KCl, 4 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, and 3 mM glucose (pH 7.4). The tegmentum vasculosum and the tectorial membrane were removed without any prior enzymatic treatment using a fine minutia needle. Chicken basilar papillae were stored in a 37°C incubator in minimal essential medium (Invitrogen) before recordings from hair cell in situ. All experiments were performed at room temperature (21-23°C) within 5-45 min of isolation. All of the reagents were obtained from Sigma unless specified otherwise.
Electrophysiology-K ϩ currents were recorded in a wholecell voltage-clamp configuration using 3-5-megohm pipettes. Currents were amplified with an Axopatch 200B amplifier and filtered at 2 kHz through a low-pass Bessel filter, and data were digitized at 5-20 kHz using an analog-to-digital converter. This study included cells (n ϭ 537) of healthy appearance whose leak current was not greater than 30 pA at a Ϫ80-mV holding potential throughout the recording period (ϳ40 min). The sampling frequency was determined by the protocols used. No on-line leak current subtraction was made, and as such, only recordings with a holding current less than 20 pA were accepted for analyses. The liquid junction potentials were measured and corrected. The capacitative transients were used to estimate the capacitance of the cell as an indirect measure of the cell size. Capacitative decay was fitted with a single exponential curve to determine the membrane time constant. Series resistance was estimated from the membrane time constant, given its capacitance. The series resistances were within the 5-20-megohm range. After 60 -90% compensation, the mean residual uncompensated resistance was 6.3 Ϯ 0.5 megohms (n ϭ 60). The seal resistance was typically 5-20 gigohms. Action potentials were amplified (100ϫ), filtered (band pass of 2-10 kHz), and digitized at 5-500 kHz as described (4). The stock solutions of channel blockers used were made in either double-distilled H 2 O or dimethyl sulfoxide and stored at Ϫ20°C. The final concentration of dimethyl sulfoxide in the recording bath solution was ϳ0.001%.
Data Analysis-The number of cells (n) is given with each data set. Data were analyzed using pClamp8 (Axon Instruments), Origin7.0 (MicroCal Software, Northampton, MA), and Excel 2000 (Microsoft, Redmond, WA). Voltage dependence of activation was examined from currents elicited by step depolarizations to potentials between Ϫ70 and 50 mV at different developmental stages (E10 -P2), and then normalized curves were fitted with the Boltzmann distribution. Pooled data are presented as the mean Ϯ S.D. Significant differences between groups were tested using Student's t test, with p Ͻ 0.05 or 0.01 indicating a statistically significant difference. Data were analyzed using pClamp8 and Origin7.0.
The n values reported reflect the number of cells.
Histochemical Analysis-To visualize cholesterol in the plasma membrane of hair cells, we used filipin staining (30). Filipin binds to cholesterol with high affinity and has natural fluorescence under UV excitation. Filipin (125 g/ml) staining was performed in a light-protected room for 2 h at room temperature. After incubation with primary reagents, the preparation was washed with 0.5 M Tris containing no serum and counterstained with NeuroTrace TM 500/525 green fluorescent Nissl stain (N-21480, Molecular Probes, Eugene, OR) according to the manufacturer's instructions. The preparation was visualized using a confocal microscope (Zeiss LSM 510).

Methyl-␤-cyclodextrin Increases I Kv Only in Immature Auditory Hair
Cells-To examine the roles of membrane lipids in the functional development of ionic conductances in hair cells, we determined the sensitivity of voltage-gated K ϩ currents (I Kv ) to membrane cholesterol by application of the cholesterol-depleting compound M␤CD. M␤CD removes cholesterol from the plasma membrane and disrupts the function of lipid rafts in eukaryotic cell membranes (29). Whole-cell currents were elicited with ϳ250-ms depolarizing voltage steps in 10-mV increments from a holding potential of Ϫ80 mV. To suppress Ca 2ϩ -activated I K , we omitted Ca 2ϩ and included 100 nM iberiotoxin in the bath solution. Fig. 1A shows traces of I Kv elicited from an E12 midsection basilar papilla hair cell before and after application of M␤CD. M␤CD significantly enhanced I Kv . Shown in Fig. 1B is the current-voltage relationship obtained from data from midsection basilar papilla hair cells (n ϭ 16). The current modulated by M␤CD or the cholesterol depletion-sensitive current is denoted as the difference current. The voltage dependence of current activation was not altered significantly, as shown in Fig. 1E. The Boltzmann fits are plotted with solid lines. Half-activation voltages were Ϫ31.3 Ϯ 3.6 mV for the control and Ϫ35.1 Ϯ 5.8 mV after application of M␤CD (p ϭ 0.3, n ϭ 9). The maximum slope factors for the activation curves were 18.3 Ϯ 1.8 mV and 16.8 Ϯ 3.9 mV for control currents and after application of M␤CD (p ϭ 0.1, n ϭ 9), respectively. In stark contrast to immature hair cells, the sensitivity of I Kv to M␤CD was lost in matured hair cells (Fig. 1C). Application of M␤CD in P2 midsection hair cells left I Kv unaltered, as illustrated in Fig. 1 (C  and D). As shown in Fig. 1F, the M␤CD-sensitive current density declined by ϳ4-fold from E12 to E18. M␤CD-sensitive current density was 18.6 Ϯ 2.1 pA/picofarads at E12 and 4.6 Ϯ 1.3 pA/picofarads at E18. After E18, the M␤CD-sensitive current disappeared. Of note, the density of I Kv increased by ϳ3-fold from E12 to P2.
Methyl-␤-cyclodextrin Exerts Its Effects via Membrane Cholesterol Depletion-To determine the specificity of the M␤CD on I Kv , M␤CD was saturated with cholesterol at a ratio of 8:2 (ratio shown to abolish M␤CD lipid-buffering capacity (29,31)). Application of a cholesterol/M␤CD mixture did not potentiate I Kv in immature hair cells, as illustrated in Fig. 2. The magnitude of I Kv was unaltered after exposure to a cholesterol-M␤CD 8:2 saturated complex (Fig. 2, A-D). Together, these experiments support the notion that M␤CD exerts its effects on I Kv by cholesterol depletion of the plasma membrane of developing hair cells.
Cholesterol Depletion Affects the TEA-sensitive I Kv -delayed Rectifier-M␤CD-sensitive currents showed slow kinetics of activation and a lack of current inactivation (Fig. 1), raising the possibility that the delayed rectifier I Kv in developing hair cells is the targeted current. We tested the effect of TEA on M␤CD-sensitive current (8) . Fig. 3A (panels a-c) illustrates the effects of 5 mM TEA on the M␤CD-sensitive I Kv recorded in E12 midsection basilar papilla hair cells. Note the increase in current amplitude after application of M␤CD (Fig. 3, A and  B) and subsequent current sensitivity to 5 mM TEA (Fig. 3A,  panel c). The reverse experiment in which developing hair cells were pre-exposed to TEA showed no noticeable increase in current after exposure to M␤CD (Fig. 3, C, panels a-c, and  D). Thus, the data suggested that the TEA-sensitive I Kv is the M␤CD-sensitive conductance. Because TEA-sensitive currents are present in adult hair cells (3,8), it may be inferred that interaction of the channel producing the delay rectifier current with cholesterol is functionally relevant only in developing hair cells. Next, we cross-checked whether M␤CD had any effect on inward Ca 2ϩ currents in developing versus matured hair cells. As shown in supplemental Fig. S1, Ca 2ϩ currents in hair cells were reduced after cholesterol depletion. However, the effects of M␤CD were not statistically significant (supplemental Fig. S1).
Changes in Membrane Cholesterol during Hair Cell Development-To determine whether hair cell membrane cholesterol undergoes changes during development, we used filipin staining to evaluate cholesterol contents. Filipin labeling was markedly greater in E14 compared with P2 hair cells (Fig. 4). As demonstrated in Fig. 4, M␤CD was effective in  12). E, the Boltzmann fits were derived from the tail currents and plotted with solid lines from data obtained from hair cells at the midsection of E12 basilar papilla. Half-activation voltages were Ϫ31.3 Ϯ 3.6 mV and Ϫ35.1 Ϯ 5.8 mV (p ϭ 0.3, n ϭ 9) for control currents and after application of M␤CD, respectively. The maximum slope factors for the activation curves were 18.3 Ϯ 1.8 mV and 16.8 Ϯ 3.9 mV (p ϭ 0.1, n ϭ 9) for control currents and after application of M␤CD, respectively. F, summary data of the mean current density measured at 0-mV step potential from data collected from hair cells at the midsection of the basilar papilla (pA/picofarads (pF)) at E12, E16, E18, and P2. The sensitivity to M␤CD decreased as hair cells became more mature. E12, n ϭ 16; E16, n ϭ 14; E18, n ϭ 13; and P2, n ϭ 12. *, p Ͻ 0.05. depleting cholesterol in hair cells. If indeed these findings have any physiological relevance, we would predict that the magnitude of the TEA-sensitive I Kv would increase as part of the normal developmental plasticity of hair cell membrane cholesterol. The data shown in Fig. 1F are in keeping with the correlation between developmental changes in cholesterol and the magnitude of I Kv , solidifying the physiological relevance of the present findings.

Depletion of Membrane Cholesterol Has Profound Functional Consequences on Spontaneous Electrical Activity of
Hair Cells-To further investigate the functional ramifications of changes in membrane cholesterol in hair cell development, we examined the effects of M␤CD on the spontaneous electrical activity. Fig. 5 illustrates membrane SAP from immature midsection basilar papilla hair cells at E12 before (Fig.  5A) and after (Fig. 5B) exposure to M␤CD. Membrane depletion of cholesterol resulted in loss of spontaneous activity. The summary data on the effects of M␤CD on the resting membrane potential are shown in Fig. 5C. Alterations of M␤CD concentrations produced concentration-dependent changes in the spike frequency (Fig. 5D) and the resting membrane potential of hair cells, which is reminiscent of the effects of TEA on spike activity during development (3,8). In addition, the spike width was reduced markedly (Fig. 5E).

DISCUSSION
In this study, we show that plasma membrane cholesterol is an important determinant of the magnitude of I Kv in hair cells. This effect of membrane cholesterol is an important functional feature specific to developing but not matured hair  Fig. 1. Shown are the current traces and amplitude for control conditions (A) and after exposure to the cholesterol-M␤CD 8:2 saturated complex (B). C, the difference current traces reflect the loss of M␤CD effects seen in Fig. 1. D, summary data of the I-V relationship obtained from 12 hair cells at E12 from the midsection of the basilar papilla. The data suggest that M␤CD exerts its effects on I Kv by depletion of membrane cholesterol. cells. The evidence and implications of this work also raise the suggestion that functional maturation of hair cells entails plasticity of membrane lipid composition. Moreover, because the TEA-sensitive I Kv currents are also expressed in matured hair cells (8), our findings further raise the possibility that interaction between membrane proteins and cholesterol changes with hair cell development. These effects of membrane cholesterol on I Kv function could have significant consequences for the regulation of developing hair cell excitability in physiological conditions. Cholesterol and other lipid compounds such as sphingolipids assemble to form microdomains in the sea of plasma membrane lipids. These cholesterol-rich niches serve not only as platforms for the delivery of ion channel proteins but also as sites for the functional interaction of other membrane proteins (32). Indeed, several K v channels have been identified in these cholesterol-rich microdomains and found to produce functional alterations of channel properties (33,34). However, there is a paucity of data on endogenous K v channels in cholesterol-rich membrane niches and the functional impact of these lipid niches in the auditory system. The importance of the findings on the effects of cholesterol depletion on increased K v channel current magnitude in developing hair cells is underpinned by the pivotal role of I Kv in the development of hair cells (2,5).
The composition of membrane lipids greatly affects the biophysical and mechanical properties of the plasma membrane such as the fluidity and stiffness, which can modulate the gating of voltage-gated channels (35)(36)(37)(38). Additionally, cholesterol may exert its effect by directly binding to the membrane proteins, affecting protein conformation and dynamics (39,40). Single-channel studies have eliminated the likelihood that membrane lipids modulate single-channel unitary conductances, thereby potentiating the current (28,41). Instead, it appears that cholesterol affects the number of active channels in the plasma membrane. In this study, the relatively short delay between M␤CD application (ϳ2 min) and the effect on I Kv precludes the possibility for changes in channel synthesis or subunit assembly. However, there are several reports indicating that the membrane cholesterol modulates the equilibrium between active and silent forms of channels (28,31,41). Alternatively, it is conceivable that cholesterol modulates the surface distribution of the channels, which, as it turns out, has marked impact on channel functions (31,42). Furthermore, in addition to direct interactions with membrane cholesterol, the observed potentiation of I Kv in developing hair cells could ensue from indirect association of K v channels with other binding proteins that are lipid-sensitive (33,43,44). Moreover, cholesterol depletion eliminates SAPs in developing hair cells, and in keeping with the roles of K v channels, the resting membrane potential and action potential durations were altered accordingly.
Lipid microdomains represent important niches for compartmentalization of cell functions. Indeed, several essential features of hair cells are derived and made possible by exclu-  sive expression of specific proteins in subcellular compartments. Whereas most of the mechanosensory apparatus such as transduction channels and adaptation motor are expressed at or in close proximity to apical stereociliary bundle membranes (45,46), prestin, the motor protein of outer hair cells (27), voltage-gated ion channels, and synaptic protein machinery are compartmentalized at the basolateral membrane. Although until now studies on the role of membrane lipid compartmentalization in hair cells remained limited, in heterologous expression systems, prestin floats in low-density membrane fractions and associates with markers of lipid microdomains (24). Also, components of the hair cell transduction apparatus resist extraction by nonionic detergents (25), suggesting that parts of the transduction apparatus are embedded in lipid raft microdomains. Thus, our findings are consistent with the growing evidence that specialized lipid microdomains may play a critical role in hair cell functions. On a pragmatic level, recent reports have associated hyperlipidemia and dyslipidemia with hearing loss (47,48). The findings that outer hair cell function is tuned to membrane cholesterol content (49,17) further strengthens the growing evidence suggesting that membrane lipid microdomains may be important for hair cell physiology and pathophysiology.