Bestrophin-1 influences transepithelial electrical properties and Ca2+ signaling in human retinal pigment epithelium.

PURPOSE: Mutations in BEST1, encoding Bestrophin-1 (Best1), cause Best vitelliform macular dystrophy (BVMD) and other inherited retinal degenerative diseases. Best1 is an integral membrane protein localized to the basolateral plasma membrane of the retinal pigment epithelium (RPE). Data from numerous in vitro and in vivo models have demonstrated that Best1 regulates intracellular Ca2+ levels. Although it is known from in vitro and crystal structure data that Best1 is also a calcium-activated anion channel, evidence for Best1 functioning as a channel in human RPE is lacking. To assess Best1-associated channel activity in the RPE, we examined the transepithelial electrical properties of fetal human RPE (fhRPE) cells, which express endogenous Best1.
METHODS: Using adenovirus-mediated gene transfer, we overexpressed Best1 and the BVMD mutant Best1W93C in fhRPE cells and assessed resting transepithelial potential (TEP), transepithelial resistance, short circuit current (Isc), and intracellular Ca2+ levels. Cl- currents were directly measured in transfected HEK293 cells using whole-cell patch clamp.
RESULTS: Best1W93C showed ablated Cl- currents and, when co-expressed, suppressed the channel activity of Best1 in HEK293 cells. In fhRPE, overexpression of Best1 increased TEP and Isc, while Best1W93C diminished TEP and Isc. Substitution of Cl- in the bath media resulted in a significant reduction of Isc in monolayers overexpressing Best1, but no significant Isc change in monolayers expressing Best1W93C. We removed Ca2+ as a limit on transepithelial electrical properties by treating cells with ionomycin, and found that changes in Isc and TEP for monolayers expressing Best1 were absent in monolayers expressing Best1W93C. Similarly, inhibition of calcium-activated anion channels with niflumic acid reduced both Isc and TEP of control and Best1 monolayers, but did not notably affect Best1W93C monolayers. Stimulation with extracellular ATP induced an increase in TEP in control monolayers that was greater than that observed in those expressing Best1(W93C). Examination of [Ca2+]i following ATP stimulation demonstrated that the expression of Best1W93C impaired intracellular Ca2+ signaling.
CONCLUSIONS: These data indicate that Best1 activity strongly influences electrophysiology and Ca2+ signaling in RPE cells, and that a common BVMD mutation disrupts both of these parameters. Our findings support the hypothesis that Best1 functions as an anion channel in human RPE.

Introduction

The bestrophinopathies are a collection of five diseases causally associated with mutations in the gene BEST1 (BEST1 database). The most common bestrophinopathy is Best vitelliform macular dystrophy (BVMD, OMIM 153700), a retinal disease characterized by an autosomal dominant mode of inheritance, an accumulation of lipofuscin in the retinal pigment epithelium (RPE), the formation of macular lesions, and vision loss [1-4]. BEST1 encodes the protein bestrophin-1 (Best1), a homo-pentameric anion channel [5,6] and integral membrane protein that is localized to the basolateral plasma membrane of the RPE [7,8]. Best1 has also been reported in the cytosolic compartment adjacent to the basolateral plasma membrane [9-12], where it may regulate Ca2+ stores by conducting anions as a counter to Ca2+ [9-11].

Prior to the identification of the BEST1 gene, the only fully penetrant symptom of BVMD was considered a reduced electrooculogram (EOG) light peak (LP) with a normal clinical electroretinogram [1-3,13]. The LP is recorded as a change in the corneo-retinal standing potential of the eye. This change in trans-tissue potential is the result of a depolarization of the basolateral plasma membrane of the RPE, and correlates with a change in the transepithelial electrical potential (TEP) of the RPE [14-16]. The change in TEP that is recorded is the LP and is believed to be generated by a Ca2+-dependent Cl- conductance across the basolateral plasma membrane of the RPE [14-16], where Best1 is localized [7].

Whole-cell patch clamp analysis of Best1 and other bestrophins in heterologous systems demonstrates that they function as Ca2+-activated anion channels (CAAC), and that disease-causing mutations in BEST1 impair anion channel activity [17-19]. This led to the hypothesis that the diminished EOG characteristic of BVMD was due to a loss of Best1 CAAC activity. However, our prior studies using a whole-cell patch clamp on RPE from Best1 knock-in and knockout mice failed to find any effect of either the absence of Best1 or the Best1 mutation W93C on Ca2+-activated Cl- conductances in RPE cells isolated from those mice [20,21]. Furthermore, we have shown that the LP is not generated by Best1, but is regulated by it [20]. Although it is known from in vitro data [17-19], crystal structure data [5,6], and in vivo neuronal data [22,23] that Best1 is an anion channel, Best1 anion channel activity in the RPE of any species has yet to be documented.

Best1 also serves as a regulator of intracellular Ca2+ levels. We and others have shown that, in vitro, Best1 regulates the activation/inactivation kinetics of voltage-dependent Ca2+ channels (VDCCs) [20,24,25], that it physically interacts with VDCC subunits [25-27], and that Best1 mutants alter the functional interaction of Best1 and VDCCs [24,28]. In Best1-deficient mice, we have shown that stimulation of the RPE with ATP, a candidate light peak substance [29], results in increased [Ca2+]i in comparison with wild-type (WT) mice. Best1-deficient mice also exhibit a more robust LP luminance response than WT mice [20]. Conversely, mice harboring the BVMD-associated W93C mutation in Best1 exhibit a LP luminance response reminiscent of VDCC-deficient mice [21]. Interestingly, the LP is diminished by inhibition of VDCCs [30], and the LP luminance response is desensitized and diminished in mice lacking either the Cav1.3 [30] or B4 [20] subunits of VDCCs. Best1 has also been reported to regulate Ca2+ stores in RPE [9-11]. Thus, defective Ca2+ signaling and/or Ca2+ store release may underlie the LP defect in BVMD. This could occur either indirectly via its channel activity or directly via its regulation of VDCCs. Although the latter function remains to be validated in vivo, RPE cells from Best1W93C knock-in mice exhibit no detectable Ca2+ release following ATP stimulation [21].

Based on these observations, we sought to examine the effects of Best1 on transepithelial electrical properties and intracellular Ca2+ signaling in human RPE. To accomplish this, we studied Best1 and the BVMD-associated mutant Best1W93C using cultured fetal human RPE (fhRPE) monolayers. Best1W93C was chosen because we have previously established using mouse and rat models that this mutant diminishes the LP response [21,31]. We have also shown that Best1W93C disrupts the functional interaction of Best1 with VDCCs in a heterologous system [24] and physically interacts with WT Best1 [32], consistent with the dominant nature of BVMD. W93C is also one of the most frequently described mutations associated with BVMD. In contrast to other RPE culture models (e.g., ARPE-19, D407, RPE-J), fhRPE expresses endogenous human Best1 (hBest1) [32,33], mimics many of the ion transport properties of RPE in the eye [33], and generates a high enough transepithelial resistance (TER) to permit studies of transepithelial ion flux [33,34]. This is the first manuscript to investigate the effects of Best1 on transepithelial electrophysiology in human RPE. Our studies lead us to conclude that Best1 activity regulates both transepithelial electrical properties and Ca2+ signaling of RPE, and that disruption of both functions of Best1 may contribute to the pathogenesis of BVMD.

Methods

Cell culture, adenovirus-mediated gene transfer, and transfection

Cultures of fhRPE, established as described by Hu and Bok [33], were maintained as before [32]. For experiments, fhRPE cells were plated on poly-L-lysine–coated glass coverslips, 96-well plates, or laminin-coated 0.4 µm pore size Millicell-HA filters (12-mm; Millipore, Bedford, MA) at 2.5×105 cells/cm2 in Chee’s essential medium, containing 1.8 mM Ca2+, 1% bovine retinal extract, and 1% FBS. Monolayers were differentiated for a period of > 8 weeks before use in any experiments. Expression of hBest1 or hBest1W93C was accomplished using replication defective adenovirus vectors at an MOI of 3, as described elsewhere [7,32]. Controls were performed using adenovirus-mediated gene transfer with an empty “null” viral vector.

Immunofluorescence

Immunofluorescence staining of hBest1 in fhRPE was performed as described previously [32]. In brief, fhRPE grown on Millicell HA filters were stained for hBest1 using the previously described antibody E6–6 [7]. Nuclei were stained using 4′,6-diamidino-2-phenylindole (DAPI; Life Technologies, Carlsbad, CA) for positional referencing. Images in the X-Y and X-Z planes were obtained using a Leica SP5 confocal microscope with a 40X oil immersion objective. Images were procured and processed using Leica’s LAS AF Lite software.

Whole-cell patch clamp

WT or mutant hBest1 fused at the C-terminus to enhanced yellow fluorescent protein (eYFP, Invitrogen, Grand Island, NY) or enhanced cyan fluorescent protein (eCFP, Invitrogen, Grand Island, NY) were transfected into HEK293 cells as described previously [32,35]. Single cells identified by CFP or YFP fluorescence were used for whole-cell patch clamp experiments within 48 h. Transfected HEK293 cells were recorded using a conventional whole-cell patch-clamp technique with an EPC-10 amplifier and Patchmaster software (HEKA, Bellmore, NY). Fire-polished borosilicate glass patch pipettes were 3–5 MΩ. Experiments were conducted at RT (20–24 °C). Since the liquid junction potentials were small (<2 mV), no correction was made. The high Ca2+ intracellular solution contained (mM): 146 CsCl, 2 MgCl2, 5 Ca2+-EGTA (free Ca2+ ∼20 μM), 10 HEPES, 10 sucrose, pH 7.3, adjusted with NMDG. The standard extracellular solution contained (mM): 140 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 15 glucose, 10 HEPES, pH 7.4 with NaOH. This combination of intracellular and extracellular solutions set Erev for Cl- currents to zero, while cation currents carried by Na+ or Cs+ had very positive or negative Erev, respectively. Osmolarity was adjusted with sucrose to 303 mOsm for all solutions.

Epithelial electrophysiology

All experiments were performed in a water jacketed Ussing chamber similar to that described by Oakley et al. [36] and Miller and Steinberg [37]. The chamber was modified to hold Millicell HA cell culture inserts and to contain larger bath volumes (8.5 ml apical, 8.0 ml basal). All experiments began by perfusing both sides of the chamber at a rate of 7 ml/min with Ringers solution containing in (mM): 113.4 NaCl, 5 KCl, 5.6 glucose, 0.8 MgCl2, 2 glutathione, 1.8 CaCl2, and 26.2 NaHCO3, pH 7.4, until the TEP stabilized. All solutions were constantly gassed with 95% O2, 5% CO2. For the Cl- ion-substitution experiments, 90% of the normal Cl- content in both the apical and basal baths was replaced by gluconate [in (mM): 108.9 D-gluconic acid (Na+ salt), 4.5 NaCl, 5 KCl, 5.6 glucose, 0.8 MgCl2, 2 glutathione, 1.8 CaCl2, and 26.2 NaHCO3, pH 7.4]. For experiments examining Ca2+ effects, monolayers were stimulated with 500 nM ionomycin and 100 µM niflumic acid (NFA). Ionomycin was added to both the apical and basal baths, while NFA was added only to the basal bath.

Voltage across fhRPE monolayers or the TEP was recorded using Ag/AgCl pellet electrodes bridged with agar 0.5 M KCl bridges in each bath. Signals were passed through a low-pass Bessel filter, amplified using a DP-304 Differential Amplifier (Warner Instruments, Inc., Hamden, CT), digitized, and recorded and analyzed on a Dell PC using LabScribe 1.821 software (iWorx, Dover, NH). The TER of the monolayers was determined by passing bipolar 10 µA pulses (DS8000 Digital Stimulator, World Precision Instruments, Inc., Sarasota, FL) across the monolayer every 30 s between Ag-AgCl pellets placed in each bath. TER was measured from the current-induced voltage changes across the monolayer. Short circuit current (Isc) was calculated as Isc=(TEP/TER).

Ca2+ measurements

fhRPE were grown on glass coverslips or in 96-well cell culture plates. Overexpression of hBest1 or hBest1 mutants was achieved by adenovirus transduction as described above and in previous work [32], with a null vector serving as a control.

For Ca2+ concentration measurements, cells on coverslips were loaded with the Ca2+ indicator dye FURA-2 AM 48 h following infection. Coverslips were then secured in a perfusion chamber on the stage of a Nikon TE200 inverted microscope with a xenon light source. The chamber was continuously perfused with CO2/HCO3- containing Ringers solution (see above) warmed to 37 °C by an in-line heater. FURA-2 AM fluorescence was excited sequentially with 340 nm and 380 nm light and acquired with a 515 / 70 nm band pass filter and 40X air or oil objective using a Cascade 512b cooled CCD camera (Photometrics, Tucson, AZ). Data were collected every 5 s and analyzed using MetaFluor Software (Molecular Devices, Sunnyvale, CA) and Excel 2004. [Ca2+]i was determined as follows: The ratios of the intensity of emissions at 340 nm / 380 nm was determined for each individual cell in a field. Absolute values of [Ca2+]i were estimated as before [20,21] according to Grynkiewicz et al. [38], using a bath solution with 10 µM ionomycin to saturate Fura-2 with Ca2+ and Ca2+-free bath solution with 10 µM ionomycin to deplete Fura-2 from Ca2+. Resting [Ca2+]i was determined from the average [Ca2+]i of all cells in a field recorded for 3 min before addition of a stimulus.

To evaluate changes in [Ca2+]i in response to varying doses of ATP, cells grown on 96-well plates were loaded with a proprietary Ca2+ sensitive fluorescent indicator using the Calcium 4 Assay Kit (Molecular Devices, Sunnydale, CA) according to the manufacturer’s instructions. The cells were then placed in a Flexstation 3 (Molecular Devices) and imaged using 485 nm excitation/525±10 nm emissions. A 90 s baseline was established, after which ATP was added to the solutions and the responses recorded every 3 s for an additional 5 min. Data were acquired using SOFTmax Pro v5.2 software (Molecular Devices). Changes in cell Ca2+ were evaluated by normalizing fluorescence intensity levels to that measured immediately preceding the addition of ATP. Following the experiment, hBest1 expression was verified by immunoprecipitation with anti-Best1 pAb-125, and western blotting with anti-Best1 maB E6–6, as previously described [7].

Statistics

Statistics were done using ANOVA (ANOVA) and t tests. Significance was defined as p<0.05.

Results

hBest1W93C impairs anion channel activity

To date, all disease-causing mutants of hBest1 tested have exhibited diminished anion channel activity [17,18]. We confirmed this for hBest1W93C by performing a whole-cell patch clamp of HEK293 cells heterologously expressing hBest1 fused with YFP or CFP (Best1-YFP, Best1-CFP), Best1W93C-YFP, or YFP alone. We have previously shown that the addition of a CFP or YFP tag to the C-terminus of hBest1 does not mitigate its ability to function as an anion channel [35]. As shown in Figure 1A, Best1-YFP and Best1-CFP produce significant Cl- currents, which are nearly absent from control cells. Cells transfected with Best1-YFP and Best1-CFP produced currents with amplitudes between those of Best1-YFP and Best1-CFP (Figure 1A). In contrast, Best1W93C-YFP produced little current in comparison to Best1-YFP (Figure 1B). In cells expressing both Best1W93C-YFP and Best1-CFP, Cl- currents were substantially diminished relative to cells expressing Best1-YFP and Best1-CFP (Figure 1C). This is the first time that differently tagged forms of fluorescent Best1 have been co-expressed together for whole-cell patch-clamp analysis, and the first time that such an analysis was done where WT and mutant Best1 could both be confirmed to be co-expressed. This is particularly significant, as it was not uncommon to observe cells in co-transfected populations only expressing one of the two Best1 variants (e.g., Best1W93C-YFP only or Best1-CFP only).

Figure 1

Whole-cell chloride currents mediated by Best1-CFP, Best1-YFP, and Best1W93C-YFP. A: Whole-cell patch-clamp recordings were performed on HEK293 cells transfected with Best1-CFP, Best1-YFP, or Best1-CFP and Best1-YFP. Cells expressing fluorescently tagged Best1 exhibited robust chloride currents compared to untransfected cells recorded from the same plate. B: In contrast, Best1W93C exhibited a significant loss of anion channel activity compared to Best1-YFP (B), and dominantly inhibited the anion channel activity of wild-type Best1-YFP when co-expressed (C).

Effects of hBest1 and hBest1W93C on transepithelial electrical properties

As we have done in previous work [32], primary cultures of fhRPE were grown on permeable supports. These cells produce monolayer cultures that normally express hBest1 (Figure 2A,B) and can generate a TER of >400 Ω x cm2 (Figure 3B) [33,34] when measured in Ringer’s solution. hBest1 and hBest1W93C were overexpressed in the monolayers using adenovirus mediated gene transfer. Controls were performed using adenovirus-mediated gene transfer with an empty “null” viral vector. We have previously demonstrated that adenovirus-mediated gene transfer causes overexpression of these proteins in fhRPE cells relative to endogenous levels of Best1 [32]. Localization of overexpressed hBest1 and hBest1W93C was previously observed to be to the basolateral plasma membrane in fhRPE cells [32] and the RPE of rat eyes [31]. We confirm this here by showing immunofluorescence and confocal microscopy of endogenous Best1 (Figure 2A,B), overexpressed hBest1 (Figure 2C,D), and overexpressed hBest1W93C (Figure 2E,F).

Figure 2

Localization of endogenous hBest1, overexpressed hBest1, and overexpressed hBest1W93C in fhRPE cells. Monolayers of fhRPE were stained for hBest1 expression (green) and examined by confocal microscopy in the X-Y (A, C, E) and X-Z (B, D, F) planes. Endogenous hBest1 (A, B) was observed to be localized to the basolateral plasma membrane, as was hBest1 in monolayers overexpressing hBest1 (C, D) or hBest1W93C (E, F). Nuclei (red) were stained for positional referencing. Scale bars: 20 µM.

Figure 3

Baseline characteristics of hBest1 and hBest1W93C in fhRPE and the contribution of Cl- to short circuit current (Isc). The transepithelial electrical properties of monolayers of fhRPE overexpressing hBest1 or hBest1W93C (W93C) were compared with control fhRPE cells. A: Transepithelial potential (TEP), (B) transepithelial resistance (TER), and (C) short-circuit current (Isc) are shown. The contribution of Cl- to differences in Isc was determined by substitution of Cl- with gluconate in the bath media. Representative voltage recordings are shown in D, and Isc is shown in E. Vertical bars in the recordings in D represent the voltage change in response to a 10 µA bipolar current pulse used to determine TER at 30 s intervals. For cells overexpressing hBest1, a significant (p<0.03) reduction in Isc occurred in response to Cl- substitution, the effects of which were reversible. Bars in E indicate the Isc recorded in Ringer’s (Cl- Ringers), during ion substitution (gluconate), and following return to Ringer’s (recovery). Bars in A-C and E represent mean ± SD with n=23 control, n=16 hBest1, and n=11 W93C for A-C. For panel E, n=20 control, n=14 hBest1, and n=9 W93C. *p<0.04 versus control.

Overexpression of hBest1 or hBest1W93C did not affect TER (Figure 3B) when compared to control monolayers transduced with a null adenovirus vector. Monolayers overexpressing hBest1 exhibited a significant (p<0.05) increase in TEP (Figure 3A) and the calculated short circuit current (Isc, Figure 3C), while monolayers expressing hBest1W93C exhibited a significantly diminished TEP (Figure 3A) and Isc (Figure 3C).

To test whether the changes in Isc in cells overexpressing hBest1 and hBest1W93C were due in part to an altered Cl- conductance, 90% of the Cl- in the apical and basal baths was substituted with gluconate. The addition of hBest1 anion channels to the basolateral membrane would theoretically increase electrogenic Cl- flux and therefore account for the observed increase in Isc (Figure 3C). Removing bath Cl- would inhibit Cl- uptake by the apical Na+-K+-Cl- co-transporter and therefore serve as an indirect test of the basolateral Cl- flux contributing to this observed change in Isc. Substitution resulted in observable voltage changes (Figure 3D) in all monolayers. However, statistically significant changes in Isc (Figure 3E) were observed only in monolayers overexpressing hBest1, which showed a significant reduction in Isc following substitution with gluconate. This change in Isc was reversible, as it returned to its previous value after restoring Cl- to the bath media (Figure 3E).

Effects of hBest1 and hBest1W93C on transepithelial electrical properties following application of ionomycin

We have previously shown that, following ATP stimulation, Ca2+ signaling is suppressed in RPE from mice heterozygous or homozygous for the W93C mutation [21]. To assess whether the diminished TEP (Figure 3A) and Isc (Figure 3C) observed in monolayers overexpressing hBest1W93C was due to altered Ca2+ signaling, we eliminated Ca2+ influx as a limit by treating monolayer cultures with the Ca2+ ionophore ionomycin (Figure 4). This would be predicted to have the effect of overcoming any suppressive effects hBest1 could exert on [Ca2+]i and, as such, any observed changes would reflect its activity as an anion channel.

Figure 4

Effect of hBest1 and hBest1W93C expression on Ca2+-stimulated changes in transepithelial electrical properties. A: fhRPE cells were treated with 500 nM ionomycin, which caused a biphasic increase in TEP. Each phase of the response was measured at the indicated point. When the TEP reached a plateau, 100 µM NFA was added to the bathing solution to inhibit calcium-activated anion channel activity. Representative recordings of responses are shown in (B), demonstrating that overexpression of either hBest1 or hBest1W93C resulted in a suppression of the overall response, and that the response to the addition of NFA was similar in control and monolayers overexpressing hBest1. Isc (C) was determined from TEP (D) and TER (E) for P0, P1, P2, and NFA. Responses were elevated at P0 for hBest1-overexpressing monolayers compared to controls and monolayers overexpressing hBest1W93C. However, at P1, P2, and NFA, Isc was not significantly different between controls and monolayers overexpressing hBest1. Neither ionomycin nor NFA had a significant effect on the transepithelial electrical properties of fhRPE monolayers expressing hBest1W93C. Compared to control and hBest1-overexpressing monolayers, TEP (D) and Isc (C) were significantly reduced for hBest1W93C monolayers at P0, P1, P2, and NFA. For both control and hBest1-overexpressing monolayers, the differences in TEP (D) and Isc (C) between different time points were significant (p<0.05). Data are mean ± SD, n=9 for control, n=5 for hBest1, and n=6 for W93C.

In control monolayers, ionomycin treatment resulted in a biphasic increase in TEP above the resting potential (indicated as P0; Figure 4A). The two peaks in TEP stimulated by ionomycin were designated P1 and P2, respectively (Figure 4A). After P2 reached a plateau, the monolayers were treated with NFA, a broad inhibitor of hBest1 and other Ca2+-dependent ion channels. At any point in the time course, there was no significant change in Isc (Figure 4C), TEP (Figure 4D), or TER (Figure 4E) for monolayers expressing hBest1W93C. Compared to control and overexpressing hBest1 monolayers, however, monolayers expressing hBest1W93C exhibited significantly lower (p<0.05) TEP and Isc at the P0, P1, P2, and NFA time points (Figure 4C,D). The overall response (Figure 4B) of monolayers overexpressing hBest1 appeared to be similar to controls’. Although the Isc and TEP during the P0 phase were higher in monolayers overexpressing hBest1 compared to controls’, the maximum Isc, TEP, and TER at P1 and P2 (Figure 4C,D,E) were nearly identical in controls and monolayers overexpressing hBest1. Moreover, the effect of NFA on these properties in monolayers overexpressing hBest1 was indistinguishable from controls’ (Figure 4C,D). While NFA reduced TEP and Isc of control and hBest1-overexpressing monolayers (Figure 4C,D), NFA failed to impact the TEP or Isc of hBest1W93C-expressing monolayers (Figure 4C,D). For control and hBest1-overexpressing monolayers, the differences in TEP and Isc at each time point were significant.

Effects of hBest1 and hBest1W93C on transepithelial electrical properties following stimulation with ATP

We next examined the effects of hBest1 and hBest1W93C overexpression on the responses of fhRPE cells to a physiologic stimulus that would raise [Ca2+]i, extracellular ATP. ATP has been demonstrated to increase intracellular Ca2+ in the RPE [29], a response that is impaired in Best1W93C knock-in mice [21]. Furthermore, ATP is a candidate for the LP substance [16] and is known to activate RPE Cl- conductances [29]. Figure 5 shows that stimulation of control fhRPE monolayers with 50 µM ATP results in a rapid increase in TEP (Figure 5A,B) and a modest but slower decrease in TER (Figure 5C). Isc in control monolayers was increased, reaching a plateau within 6 min of stimulus (Figure 5D). In contrast, cells overexpressing hBest1 exhibited a rapid decrease in TEP (Figure 5A,B) and TER (Figure 5C), as well as a smaller increase in Isc compared to controls (Figure 5D). The increase in Isc (Figure 5D) in cells overexpressing hBest1W93C stimulated with extracellular ATP was smaller and slower than both control and hBest1-overexpressing monolayers. This was due primarily to a smaller, slower increase in TEP (Figure 5A,B) in cells overexpressing hBest1W93C compared to controls, although the change in TER (Figure 5C) was comparable. The very large, very rapid decrease in TER (Figure 5C) in monolayers overexpressing hBest1 suggested that the tight junctions of the fhRPE had been compromised. As a result, although we show TEP (Figure 5A,B) and calculated Isc (Figure 5D) for these monolayers, these data cannot be interpreted as representing events occurring at the plasma membrane.

Figure 5

Effect of ATP on transepithelial electrical properties of fhRPE monolayers. Representative recordings from monolayers expressing endogenous hBest1 (control) or overexpressing hBest1 or hBest1W93C (W93C) in response to a 50 µM ATP stimulus are shown in A. TEP (B) increased in controls and cells expressing hBest1W93C in response to ATP, but diminished significantly in cells overexpressing hBest1. Vertical bars in the recordings in A correspond to the voltage deflection induced by a 10 µA bipolar current pulse used to determine TER (C), which diminished precipitously in monolayers overexpressing hBest1. Isc (D) was calculated from TEP and TER. Data are representative of ≥4 experiments.

Effects of hBest1 and hBest1W93C on ATP stimulated Ca2+ release

The dramatic hBest1- and hBest1W93C-dependent differences observed in fhRPE monolayers in response to ATP motivated us to follow the ATP-induced changes in [Ca2+]i resulting from hBest1 or hBest1W93C overexpression using the indicator dye fura-2. Mean [Ca2+]i in fhRPE cells differed significantly between controls and cells overexpressing hBest1 or hBest1W93C, suggesting that hBest1 can affect resting [Ca2+]i in human RPE (Figure 6). Compared to controls, mean resting [Ca2+]i was substantially reduced in monolayers overexpressing hBest1 (Figure 6B). Prior to ATP stimulation, resting [Ca2+]i was elevated beyond that of controls in monolayers overexpressing hBest1W93C (Figure 6B). Measured resting [Ca2+]i was 83±3 nM for controls, 66±11 nM for hBest1, and 90±5 nM for hBest1W93C (average ± SE, n=4 experiments; Figure 6B). Following ATP stimulation, [Ca2+]i was observed to increase in controls, hBest1-overexpressing cells, and cells expressing hBest1W93C (Figure 6A). However, the increase in mean [Ca2+]i in cells overexpressing hBest1 was not sustained and [Ca2+]i returned rapidly to near resting levels. Cells expressing hBest1W93C exhibited an initial fast rise in [Ca2+]i similar to that observed in hBest1-overexpressing cells, but did not continue to increase and instead remained steady at a level approximately half of that observed in control cells (Figure 6A).

Figure 6

Effect of hBest1 and hBest1W93C on ATP-induced release of Ca2+ stores. A: The concentration of intracellular Ca2+ was measured using the indicator dye fura-2 in fhRPE following stimulation with 100 µM ATP. [Ca2+]i was compared between control monolayers and monolayers overexpressing hBest1 or hBest1W93C after ATP-stimulation, as well as after subsequent washout of ATP. B: Prior to ATP stimulation, mean baselines [Ca2+]i between control and overexpressing monolayers are shown. Differences between monolayers were significant (p<0.005 versus controls, n=4). Data are mean ± SEM.

We next examined Ca2+ release in fhRPE cells grown in 96-well plates in response to varying doses of ATP. At a dose of 5 µM ATP, controls and cells overexpressing hBest1 exhibited a rapid initial increase in [Ca2+]i (Figure 7A). However, while [Ca2+]i levels remained high in control cells, they rapidly returned to near starting levels in cells overexpressing hBest1 (Figure 7A), and the entire response was significantly diminished in monolayers overexpressing hBest1W93C (Figure 7A). To examine this more closely, we generated dose response curves for the initial rise in [Ca2+]i as a function of ATP concentration. A definitive change in dose response, measured by the initial rise in Ca2+, was noted in cells overexpressing hBest1W93C compared to cells expressing only endogenous hBest1 or cells overexpressing hBest1 (Figure 7B). Analysis of the differences in these responses between groups by ANOVA found them significant (p<0.003) for all doses between 1 and 100 µM for hBest1W93C compared to controls.

Figure 7

Dose-responsive effect of ATP on [Ca2+]i in fhRPE cells. A: Represented by normalized change in percent fluorescence, the responses in [Ca2+] to 5 µM ATP are shown for control fhRPE and fhRPE overexpressing hBest1 or hBest1W93C B: Dose response curves for the initial rise in [Ca2+]i as a function of ATP concentration are shown for control monolayers and for monolayers overexpressing wild-type or mutant hBest1. ATP doses were between 1 and 100 µM. Differences in responses between each group were significant by ANOVA (p<0.003, n≥5). Data are mean ± SEM.

Discussion

Our data demonstrate that hBest1 influences both transepithelial electrical properties and Ca2+ signaling in RPE cells. This is the first study to investigate the effects of Best1 on transepithelial electrical properties in any system, and the first to investigate the effects of hBest1 on Ca2+ signaling in fhRPE. We infer a few likely explanations for the mechanisms underlying these changes in fhRPE. The first, which we favor, is that these data provide evidence for hBest1 functioning as a CAAC in human RPE. Recent crystal structure data for bacterial and chicken Best1 [5,6], as well as in vivo neuronal data [22,23], have made it unambiguously clear that Best1 functions as a homo-pentameric anion channel with an ion conductance pore at its center. Moreover, we and others have shown that endogenous hBest1 can be cell-surface biotinylated [7,39], that overexpressed hBest1 in native and heterologous systems can be cell-surface biotinylated [7,24,40-42], that hBest1 co-localizes with the plasma membrane marker MCT-1 when expressed in MDCK cells [43], and that anti-Best1 antibody-labeled gold particles are found at the basal membrane of the RPE in eyes from rhesus monkeys [8]. With this in mind, overexpression of hBest1 in fhRPE should cause an increase in the number of anion channels in the basolateral plasma membrane. This would be theorized to result in an increase in TEP due to a relative depolarization of the basolateral plasma membrane. Isc, which reflects all net ionic fluxes, would also be increased by Best1-enhanced electrogenic anion (e.g., Cl-) transport. This is very similar to our reported data in Figure 3, where monolayers overexpressing hBest1 showed significantly increased TEP (Figure 3A) and Isc (Figure 3C) compared to controls. A slight reduction in TER was observed, but it did not reach statistical significance (Figure 3B). We have previously shown that hBest1W93C physically interacts with WT hBest1 [32], and we show in Figure 1 that hBest1W93C-YFP dominantly suppresses the Cl- channel activity of hBest1-CFP (Figure 1C) in HEK293 cells. As such, overexpression of hBest1W93C should physically interact with endogenous hBest1 channels in fhRPE and impair their activity. This loss of basolateral plasma membrane CAACs would be predicted to decrease TEP and Isc. Our data show that overexpression of hBest1W93C significantly decreases TEP (Figure 3A) and Isc (Figure 3C) compared to controls. While TER remained unchanged (Figure 3B), these observations are analogous to the predicted changes resulting from overexpression or impairment of an anion channel. The lack of significant change in TER may be caused by the compensatory activity of other channels or other confounding factors.

This interpretation is further supported by the gluconate substitution experiment shown in Figure 3E. That no significant change in Isc was observed in monolayers overexpressing Best1W93C following Cl- substitution (Figure 3E) suggests that the W93C mutation abolishes any Best1-associated Cl- conductance. For hBest1-overexpressing but not control monolayers, a significant decrease in Isc following gluconate substitution was observed (Figure 3E), suggesting that basolateral Cl- transport contributed to the increased Isc caused by overexpression of hBest1 (Figure 3C). Data in Figure 4 also support this interpretation for two reasons: 1) NFA reduced TEP and Isc of control and hBest1 monolayers, but not hBest1W93C monolayers. This indicates that the W93C mutation abolishes a function of hBest1 that is sensitive to inhibition by NFA. For control and hBest1-overexpressing monolayers, the reduction in Isc (Figure 4C) and TEP (Figure 4D) in response to NFA was roughly equal to the difference in these parameters between the P1 and P2 time points. Thus, this difference between P2 and P1 may represent the component corresponding specifically to CAAC activation. 2) Despite removing Ca2+ influx and altered Ca2+ signaling as a variable via the application of ionomycin, TEP and Isc were still reduced in hBest1W93C monolayers. This would suggest that these parameters are being impaired, at least in part, by a malfunction of Best1W93C that is independent on Ca2+ signaling. Oddly, the application of NFA decreased TEP and Isc to a comparable extent in both control and hBest1-overexpressing monolayers. This may have been due to the activation of other channels by ionomycin (e.g., another CAAC or the maxi-K+ Ca2+-dependent K+ channel).

A second possibility is that that these changes are caused by hBest1-induced regulation of intracellular Ca2+ signaling. This could occur via known interaction between Best1 and VDCCs [18], or by a recently proposed subpopulation of Best1 that resides in an internal compartment very near the basolateral plasma membrane [10]. For the latter, Best1 would function as an intracellular Cl- channel and conduct Cl- as a counterion for Ca2+, working to help accumulate and release Ca2+ from intracellular stores [9-11]. Overexpression of hBest1 caused a faster rise in [Ca2+]i and a faster release following ATP stimulation (Figure 7A), lending some credence to this theory. These effects on transepithelial electrical properties and Ca2+ could have also been mediated by functional and physical interaction with VDCCs. Such interaction has been observed independently by three different laboratories [24-28]. None of these possibilities preclude the other, however, and it is feasible that hBest1 anion channels in the plasma membrane, hBest1 anion channels in an internal compartment adjacent to the plasma membrane, and VDCCs all play a role in mediating the reported changes in transepithelial electrical properties (Figure 3, Figure 4, Figure 5) and intracellular Ca2+ signaling (Figure 6 and Figure 7).

In Figure 5, we show that stimulation with ATP differentially affects the transepithelial electrical properties of control, hBest1-overexpressing, and hBest1W93C-overexpressing monolayers. Since ATP stimulation of RPE increases [Ca2+]i, and Best1 is a CAAC, we would have expected ATP application to increase both TEP and Isc in fhRPE monolayers overexpressing hBest1. Instead, we observed a large decrease in both TEP (Figure 5B) and TER (Figure 5C) in these monolayers. The drop in TEP resulted from the drop in TER. This drop is an indicator of a tight junction breakdown. Since Best1 has been reported to regulate and be regulated by volume [44,45], one possibility for this drop in TER is that ATP stimulation following overexpression of hBest1 may have caused cell shrinkage. That this was only observed following ATP stimulation (Figure 5) and not after ionomycin (Figure 4) is interesting, and indicates that increasing [Ca2+]i via the ATP-specific pathway results in different effects than more broadly increasing Ca2+ via application with ionomycin.

Our Ca2+ data indicate that resting, baseline [Ca2+]i is significantly altered by the expression of hBest1 or hBest1W93C (Figure 6B). This indicates that, independent of a signal (e.g., ATP), hBest1 can still exert significant effects on Ca2+ homeostasis. We have previously observed an increase in the release of [Ca2+]i triggered by extracellular ATP in RPE explants from Best1 mice [20], and complete suppression of ATP-stimulated Ca2+ release in the RPE of Best1 knock-in mice [21]. It is important to note that the ATP-induced release of Ca2+ in human RPE, fetal or otherwise, differs from that observed in mouse RPE. Mice lack the fast component of Ca2+ release observed in Figure 6 and Figure 7, by Petersen and coworkers [29], and by Singh et al. [12], who used iPSC-derived hRPE cells. In the present study, we show that overexpression of hBest1 or hBest1W93C greatly diminished the sustained increase in cytosolic Ca2+ in response to ATP in fhRPE (Figure 6 and Figure 7). However, overexpression of hBest1 did result in a more rapid initial increase in [Ca2+]i (Figure 6 and Figure 7), and the overall response of monolayers expressing hBest1W93C were impaired compared to control and hBest1-overexpressing monolayers (Figure 6 and Figure 7). While our data for hBest1W93C are similar in mouse and human RPE in that hBest1W93C suppresses Ca2+ release following stimulation with ATP, they differ from that observed by Singh et al. [12] in human iPSC-derived RPE cells for the BVMD associated mutations A146K and N296H. These mutants caused an increase in the fast response more akin to that observed in our study for overexpression of WT hBest1 [12] (Figure 6 and Figure 7). One possible explanation for the difference is that different mutations can either constitutively activate the Ca2+ regulatory function of Best1, or alternatively inactivate it. Regardless, in both cases, intracellular Ca2+ in response to ATP stimulation is altered by mutations in hBest1, and it is likely that this effect varies with different mutations.

In summary, our data demonstrate that Best1 can influence transepithelial electrical properties and intracellular Ca2+ signaling in human RPE. They also provide support for Best1-associated anion channel activity in human RPE. The current data illustrate that mutations in BEST1 do this detrimentally, suggesting a potential mechanism for the primary defects observed in BVMD: a diminished EOG LP, fluid-filled retinal detachments, and lipofuscin accumulation [1-3,46]. Further studies are required to determine how impairing these functions individually or together contribute to the phenotypes of the different diseases caused by BEST1 mutations.