Interaction of Bestrophin-1 and Ca2+ Channel β-Subunits: Identification of New Binding Domains on the Bestrophin-1 C-Terminus

Bestrophin-1 modulates currents through voltage-dependent L-type Ca2+ channels by physically interacting with the β-subunits of Ca2+ channels. The main function of β-subunits is to regulate the number of pore-forming CaV-subunits in the cell membrane and modulate Ca2+ channel currents. To understand the influence of full-length bestrophin-1 on β-subunit function, we studied binding and localization of bestrophin-1 and Ca2+ channel subunits, together with modulation of CaV1.3 Ca2+ channels currents. In heterologeous expression, bestrophin-1 showed co-immunoprecipitation with either, β3-, or β4-subunits. We identified a new highly conserved cluster of proline-rich motifs on the bestrophin-1 C-terminus between amino acid position 468 and 486, which enables possible binding to SH3-domains of β-subunits. A bestrophin-1 that lacks these proline-rich motifs (ΔCT-PxxP bestrophin-1) showed reduced efficiency to co-immunoprecipitate with β3 and β4-subunits. In the presence of ΔCT-PxxP bestrophin-1, β4-subunits and CaV1.3 subunits partly lost membrane localization. Currents from CaV1.3 subunits were modified in the presence of β4-subunit and wild-type bestrophin-1: accelerated time-dependent activation and reduced current density. With ΔCTPxxP bestrophin-1, currents showed the same time-dependent activation as with wild-type bestrophin-1, but the current density was further reduced due to decreased number of Ca2+ channels proteins in the cell membrane. In summary, we described new proline-rich motifs on bestrophin-1 C-terminus, which help to maintain the ability of β-subunits to regulate surface expression of pore-forming CaV Ca2+-channel subunits.

Voltage-dependent Ca 2+ channels are composed of poreforming Ca V -subunits (a1-subunits) which determine the basic Ca 2+ properties and of the auxiliary b, a2dand sometimes the csubunits [6,7]. The Ca 2+ channel b-subunits have complex functions [8,9]: they modulate the electrophysiological properties of the pore-forming Ca V -subunits, interact with kinases and are required for the transport of Ca V -subunits to the cell membrane. It is most likely that described effects of bestrophin-1 on L-type channel activity are due to modulation of b-subunit function. Using heterologous co-expression of L-type Ca 2+ channels and Cterminus fragments of bestrophin-1, proline-rich motifs between the amino acid positions 330 and 346 were identified to enable interaction with b-subunits of voltage-dependent Ca 2+ channels via SH3 domains [4,5] . The physical interaction of full-length bestrophin-1 with Ca 2+ channel b-subunits was confirmed by Reichhart et al. [5].
Mutations in the gene for bestrophin-1, BEST1, lead to different types of retinal or macular degenerations [1,10]. The most common phenotype is Best's vitelliforme macular dystrophy [11,12,13]. Symptoms include macular degeneration, fast accumulation of lipofuscin and a reduction of the so called ''light-peak'' in the patient's electro-oculogram [14]. However, little is known about how mutations in BEST1 lead to disease. It is possible that BEST1's role as an anion channel and/or Ca 2+ channel regulator could explain the typical changes seen in the electro-oculogram. Several lines of evidence show that the light-peak in the electrooculogram is generated by activation of Cl 2 conductance and is dependent on the presence of b4and Ca V 1.3 subunits [3,15,16,17] in the retinal pigment epithelium (RPE) which closely interacts with the photoreceptors in the retina [18]. Bestrophin-1 is known to be basolaterally located in the RPE and can function as both Cl 2 channel and Ca 2+ channel regulator. The combined function as a Ca 2+ -dependent anion channel and Ca 2+ channel regulator would provide an efficient feedback loop to control Ca 2+dependent Cl 2 transport, for example, by the RPE [1,18]. There are several studies which investigated eyes from Best patients [19,20,21,22,23,24]. So far, only one pathologic effect of a mutant bestrophin-1 has been found. This mutant form does not show uniform basolateral localization [20]. Thus, the proper trafficking of bestrophin-1 and probably its interaction partners seem to be important for understanding bestrophin-1 function in disease. In this regard, determining the influence of bestrophin-1 on bsubunits derived regulation of pore-forming Ca V 1.3-subunits in the plasma membrane would allow for better understanding of Best's disease.
The aim of our study is to first investigate the interaction of fulllength bestrophin-1 with b-subunits and secondly, the influence of bestrophin-1 on the ability of b-subunits to regulate the surface expression of Ca V -subunits. In order to test this hypothesis we performed immunoprecipitation experiments with heterologously expressed bestrophin-1, b-subunits and a1-subunit Ca V 1.3 corresponding to the Ca 2+ channel expressed in the RPE [3,25]. These interactions and the influence on membrane localization of Ca V 1.3-subunits were verified by correlation with the subcelluar localization using confocal microscopy. The functional effects were studied by patch-clamp analysis of Ca 2+ channel currents from heterologously expressed Ca V 1.3-subunits and b4-subunits.

DNA manipulations
Deletion of the C-terminal proline rich (CT-PxxP) region between 462-575aa (113 amino acids) was introduced into human bestrophin-1 by PCR using the following primers: 59-ATC-GCTCGAGCCACCATGACCATCACTTACACA and 59-CG-ATGGATCCATGGCAGACTTGAAGGCGTC. Resulting PCR product was digested using XhoI and BamHI restriction enzymes, and subsequently inserted into pCDNA3-bestrophin-1 plasmid digested with XhoI/BamHI. All the constructs were verified by DNA sequencing.

Immunoprecipitation
Subconfluent (70-80%) culture of CHO, HEK-293 or COS-7 cells were transiently transfected with combinations of plasmids encoding a human-bestrophin-1 (N-terminal EGFP-tagged or untagged), b3 (untagged), b4-His 6 and Ca V 1.3 (N-terminal eGFPtagged or untagged) subunits of voltage-dependent calcium channels using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturer's protocol. The pcDNA3.1 and pEGFP plasmids were used as negative control. After a 24-h incubation, cells were washed with 1xPBS and then lysed in culture dish with shaking for 15 min at 4uC with ice-cold lysis buffer (150 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet-P40, 0.5% natrium deoxycholate, 1 tablet Complete Mini protein inhibitor mixture/10 ml (Roche Applied Science), and 0.7 mg/ml pepstatin). Cell lysate was scraped and transferred to a new tube and lysed for additional 15 min at 4uC with rocking. The lysates were clarified by centrifugation at 13,0006 g for 10 min at 4uC.

Western Blot Analysis
Western Blot analysis were performed as previously described in detail [3]. Lysates of membrane proteins were prepared by three freezing and thawing steps (liquid N 2 ; 42u) and two centrifugation steps at 500 and 43.000 g, the pellet was suspended in lysis buffer and subjected to SDS-PAGE (7.5-10% gel) The proteins were blotted to nitrocellulose filter membranes (Polyscreen; NEN Life Science Products, Boston, MA). The blots were blocked in 5% non-fat dry milk and 5% bovine serum albumin. Primary antibodies were diluted as follow: anti-human-bestrophin, anti-b3, and anti-GFP (1:5000), anti His6 (1:2500), anti Cav1.3, and anti beta-actin (1:1000). After incubation with primary antibodies, blots were visualized with a peroxidase-conjugated secondary antibody and a chemiluminiscence kit according to manufacturers instructions. Chemi-luminescence detection (of bound secondary antibodies) was carried out using Immobilon Western HRP substrate detection kit (Millipore). The images were digitalized using an image analyzer (Chemimager, Biozym).

Immunohistochemistry and confocal microscopy
For immunofluorescence experiments, transiently transfected CHO and ARPE-19 cells were grown on a sterile glass cover slip. 24 hours after the transfection cells were washed with 16 PBS, and fixed for 10 min at room temperature with 4% (w/v) paraformaldehyde. After three additional washing steps with 16 PBS, cells were permeabilized with blocking/permeabilization solution [10% (v/v) normal goat serum, 0.5% (v/v) Triton X-100 in 16 PBS) for 30 min. Cells were then labeled for 1 hour with antibestrophin-1 antibody, and anti-b3 antibody diluted 1:500, and goat anti-Ca v 1.3 antibody diluted 1:100 in 2% normal goat serum, 0.1% Triton X-100 in 16 PBS. After three additional washing steps cells were incubated for 1 hour with appropriate secondary antibodies diluted 1:500 (conjugated with Alexa 488, Alexa 546, and Alexa 633, (Invitrogen). Cells were mounted in confocal matrix (Micro Tech Lab, Graz, Austria) and then examined using confocal microscope LSM510 (Carl Zeiss, Göttingen, Germany).

Quantitative c-olocalization analysis
For the quantitative co-localization analysis, ARPE-19 cells grown on glass cover slips were either double or triple transfected with various bestrophin constructs and labeled with corresponding primary antibody. After subsequent incubation with secondary antibodies conjugated with Alexa 488, 546, and 633 diluted 1:500 (Invitrogen, Germany), cover slips were examined using confocal microscope LSM 510 (Zeiss, Göttingen, Germany). Confocal microscopy has advantage over the standard fluorescence microscopy, because it generates thin optical sections and thus allows quantification of the co-localization of antigens. Triple fluorescence for green, red and infrared channels was obtained using excitation of an argon-helium-neon laser at wave lengths of 488, 546, and 633 nm. Emission of the different fluorophores was detected using appropriate filter sets and multi channel acquisition. Triple stained images were obtained by sequential scanning for each channel to eliminate the crosstalk of chromophores and to ensure reliable quantification of co-localization. Images were recorded at intensity levels below saturation, estimated by intensity analysis module. Confocal images were quantitatively analyzed using an ImageJ software package. Pearson's correlation coefficient (PCC) was employed to evaluate co-localization according to Abramoff [26]. PCC is one of the standard techniques applied in pattern recognition for matching one image to another in order to describe the correlation of the intensity distributions between channels. It takes into consideration only for the similarity of shapes between two images, and does not depend upon image pixel intensity values. Values of PCC are defined from -1 to 1 where -1 indicates no overlap and 1 is a complete co-localization. For surface expression analysis, confocal image files were loaded into ImageJ (version.1.45b), and were submitted to edge detection process using built in edge detection algorithm (363 Sobel edge filter). In the next step, singe cells were selected and cell surface was labelled using freehand tool. Intracellular regions were additionally selected, and all selected regions were saved as a region of the interest (ROI). Total number of pixels were counted using analyze particles command for each channel separately. The number of pixels from the whole cell was subtracted from the intracellular regions, thus giving the proportion of the pixel localized to the membrane. Furthermore, membrane pixel values were divided with intracellular pixel values, giving relative surface expression.

Patch-Clamp recordings of Ca V 1.3 channel currents -Microinjection
For patch-clamp experiments and confocal microscopy CHO cells were microinjected with plasmids coding Ca 2+ channel subunits and different bestrophin-1 depending on the experiment as indicated in the results. Microinjections were performed under an inverted microscope (Carl Zeiss) equipped with a micromanipulator InjectMan NI2 (Eppendorf) by using an automated FemtoJet (Eppendorf) and Femtotip (Eppendorf) glass microcapillaries. CHO cells were plated on 12-mm glass plates and microinjected with plasmid DNA (50 ng/ml for each construct). Immediately after microinjection, cells were incubated overnight at 30uC, and next day cells were transferred into 37uC cell culture incubator. Patch-clamp analysis: Membrane currents were measured in the whole-cell configuration of the patch-clamp technique. While recording, transfected cells were superfused in a bath solution containing (mM): choline chloride 150, BaCl 2 10, MgCl 2 1, HEPES 10; pH 7.4 adjusted with CsOH; 333mOsm). To elicite voltage-dependent currents, cells were stimulated from a holding potential of 270 mV by stepwise depolarization. Gating currents were measured according to Fan et al. [27]. To measure gating currents, the Ca V 1.3 pore was blocked using 10 mM Co 2+ , which was used to replace Ba 2+ . Co 2+ is known to block L-type channels as it is not able to permeate through the pore [27]. Before measuring the gating current amplitude the Ca V 1.3 currents were measured currents using Ba 2+ as charge carrier. Only cells which showed a strong and robust Ba 2+ current were used for gating current analysis. For proper measurement of gating currents, special care was taken to compensate for residual capacitative currents. The perfusion chamber was assessed by fluorescent microscope. Transfected cells were selected by their GFP fluorescence. For whole-cell recording, patch-pipettes of 3-5 MV were made from borosilicate tubes using a DMZ-Universal Puller (Zeitz, Augsburg Germany). Pipettes were filled with a pipettesolution containing (mM): CsCl 135, MgCl 2 1, CsEGTA 10; pH 7.4 adjusted with CsOH; 283 mOsm). Membrane currents were recorded using an EPC-10 computer-controlled patch-clamp amplifier in conjunction with the TIDA software for data acquisition and analysis. The mean membrane capacitance was 22.0461.3 pF (n = 25). The access resistance was compensated for to values lower than 10 MV. Analysis of voltage-dependent activation was done by plotting steady-state currents against membrane potentials of electrical stimulation. Individual cell plots were fitted using the Boltzmann equation. Gating current amplitudes were plotted against the voltages of the electrical stimulation. For comparison gating current density was calculated at +20 mV.

Calculations and statistical analysis
Experiments were repeated at least three times. The Western blots shown in the figures show a representative experiment. Mean values were given as mean +/2 SEM; n refers to the number of experiments. Statistical difference was tested by ANOVA; statistic significant difference was considered at p values smaller than 0.05.

Results
In order to determine the possible physical interaction between Ca 2+ channel subunits and bestrophin-1, immunoprecipitation experiments were performed using heterologously expressed proteins. All proteins were determined to have no endogeneous expression in CHO cells ( Figure S1A). We first tested our experimental settings using known physiological interaction between pore-forming Ca V 1.3 subunits and accessory b-subunits [6,7] (Figure 1). CHO cells were co-transfected with Ca V 1.3 subunits and b3-subunits. Immunoprecipitation of b3-subunits showed the presence of Ca V 1.3 subunits ( Figure 1A, left panel) and vice versa ( Figure 1A, right panels). Using immunocytochemistry, we studied possible interaction of these proteins in intact cells ( Figure 1B). ARPE-19 cells were chosen for co-localization experiments because bestrophin-1 is normally expressed in the RPE and protein trafficking differs in the RPE of other cell types [18]. ARPE-19 cells were transfected with Ca V 1.3 subunits and b3-subunits and their subcellular localization was investigated by confocal microscopy. Both Ca V 1.3 and b3-subunits were found in the cell membrane. To quantify the co-staining and co-localization of the two proteins in the cell membrane, Pearson's correlation coefficient was calculated. The resulting coefficient of 85.068% (n = 3; Table 1) indicates a good co-localization. In order to further analyze membrane localization, we analyzed fluorescence profiles across the cells whilst avoiding the nuclear region. The confocal images were used to quantify the plasmalemmal localization by pixel analysis using edge detection ( Figure S2A-C). This revealed a relative surface expression of 3.460.13 for Ca V 1.3 and 2.960.6 for b3-subunits (n = 3; n.s.) ( Figure 1C).
In order to detect the interaction of bestrophin-1 and bsubunits, co-localization studies were performed in intact cells ( Figure 2) in the same way as described for interaction of Ca V 1.3 and b-subunits. ARPE-19 cells were co-transfected with b3subunits and bestrophin-1. In these cells, the majority of the two proteins showed a good co-localization (Table 1) and was diffuse across the cytosol (Figure 2A). Measurement of the relative surface expression by pixel analysis revealed values of 0.7860.06 for b3subunits and 0.87760.15 for bestrophin-1 ( Figure 2E; n = 3; n.s.) indicating a cytoplasmic localization for b3-subunits and bestrophin-1 when expressed together. When ARPE-19 cells were cotransfected with bestrophin-1, b3-subunits and Ca V 1.3 subunits, all three proteins were located in the cell membrane ( Figure 2B, Table 1). This differs significantly compared to the bestrophin-1 and b3-subunit localization shown before. The localization of the three proteins is shown through pixel analysis which revealed a good surface expression with values of 3.8060.26 for Ca V 1.3, 3.0660.16 for b3-subunits and 3.5160.25 for bestrophin-1 (all n = 3; Figure 2E). The same could be observed in cells cotransfected with Ca V 1.3, b4-subunits and bestrophin-1 ( Figure 2C). The relative surface expression were 3.6560.34 for Ca V 1.3, 2.9960.10 for b4-subunits and 3.3460.25 for bestrophin-1 (all n = 3; Figure 2E). For comparison and validation of the fluorescence ratios, the ratio for the purinergic receptor P2Y 2 -His 6 , a typical membrane protein, was calculated ( Figure 2D). This protein showed a relative surface expression of 5.1960.24 ( Figure 2E).
To identify the mechanism of interaction between bestrophin-1 and b-subunits of Ca 2+ channels, bestrophin-1 sequences were analyzed for interaction domains ( Figure 3A). We searched for proline-rich (PxxP) motifs which could bind to the SH3-domain of the b-subunits. Together with the already known cluster of PxxP motifs between amino acid position 330 and 346 on bestrophin-1 C-terminus [4] we detected a cluster of 4 proline-rich motifs, which are highly conserved among many species, between the amino acid positions 468 and 486. To explore the role of the newly detected cluster, we generated a deletion mutant of bestropin-1 lacking the proline-rich motifs between amino acid positions 462 and 575 (named DCT-PxxP). Using this mutant, immunoprecipitation experiments were performed to analyze binding between several b-subunits and mutant bestrophin-1. For this purpose, HEK-293 cells were transfected with wild-type or with mutant bestrophin-1 together with b3or b4-subunits. Wild-type bestrophin-1 could be co-immmunoprecipitated with either b3or b4subunits ( Figures 3B and 3C). Similar results were obtained using CHO or COS-7 ( Figure S1B). Western blot analysis of the precipitates using antibodies directed against bestrophin-1 showed that DCT-PxxP could be precipitated with the same efficiency as the wild-type bestrophin-1 ( Figure S3A). These precipitates were further analyzed for the presence of either b3-subunits or b4subunits showing co-precipitation. The DCT-PxxP bestrophin-1 also showed co-precipitation with b3-subunits or b4-subunits. Based on the summarized amount of protein in the lysate, precipitation fraction and non-bound fraction, we calculated the relative co-precipitation efficiency of the DCT-PxxP mutant and b3-subunits ( Figures 3D and 3E). b3-subunit showed a significant decrease in co-precipitation efficiency from 1263% with wild-type bestrophin-1 to 461% (n = 5; p = 0.035 unpaired t-test) with the DCT-PxxP mutant of bestrophin-1. The b4-subunits also showed a significant decrease in co-precipitation efficiency from 26.666% with wild-type bestrophin-1 to 460.1% with DCTPxxP bestrophin-1 (n = 3; p = 0.02 unpaired t-test, Figure S3B To study the functional implications of the b-subunit and bestrophin-1 interaction, patch-clamp analysis of Ca 2+ channel currents of heterologously expressed Ca V 1.3/b4 channels was performed ( Figure 5). We chose b4-subunits because they are known to be expressed in the RPE. To obtain stable membrane localization and reliable membrane current measurements, a2d1subunits were additionally expressed. CHO cells were microinjected with all required plasmids. Cells which expressed Ca V 1.3, b4-subunit and a2d1 subunits showed Ba 2+ currents associated with L-type channels as previously reported ( Figure 5A; see Table 2) [5,28,29,30]. To analyze whether the change in Ba 2+ current density is due to modulation of the Ca V 1.3 pore or due to less Ca V 1.3 protein in the cell membrane, gating currents of Ca V 1.3 channels were measured. When ionic currents were blocked after removal of Ba 2+ and addition of Co 2+ , depolarization led to voltage dependent outward currents which activated at membrane potentials more positive than 232.4562.0 mV (n = 5; Figure 5B/C). These currents represent the gating currents of the Ca V 1.3 subunit [27]. The maximal ionic Ba 2+ current density was reduced from 15.6362.19 pApF 21 (n = 10) to 8.9561.40 pApF 21 (n = 11; p = 0.0169 unpaired t-test) by the additional presence of wild-type bestrophin-1 ( Figure 5D). Without bestrophin-1 the gating current density of 1.3360.122 pApF 21 (n = 5; Figure 5E) was not significantly different compared to the gating current density in the presence of bestrophin-1 (1.2460.26 pApF 21 ; n = 4; p = 0.75 unpaired t-test). However, in the presence of the DCTPxxP bestrophin-1 mutant the current density was with values of 5.0360.44 pApF 21 (n = 8) further decreased compared to that in the presence wild-type bestrophin-1 ( Figure 5D; p = 0.034 unpaired t-test). In the presence of DCT-PxxP, the gating currents were reduced to a density of 0.1360.033 pApF 21 (n = 4; p = 0.0001 unpaired t-test; Figure 5E) indicating less Ca V 1.3 protein in the cell membrane. The voltage-dependent activation of the currents was analyzed by fitting the normalized current/voltage relation by the Boltzmann function. Neither the voltage of half maximal activation nor the slope is significantly changed in the presence of bestrophin-1 ( Figure 5F/G). In the presence of wild-type bestrophin-1, we found an acceleration of the time-dependent activation ( Figure 5H) with an activation time constant at +20 mV of 1.2660.15 ms (n = 11) versus 2.306 0.19 ms (n = 10; p = 0.0005 unpaired t-test) under control conditions without bestrophin-1 ( Figure 5I). Using the DCT-PXXP form of the bestrophin-1 the Ca V 1.3/b4 currents showed acceleration of time-dependent activation (activation time constant of 1.1860.18 ms; n = 8) comparable to that in the presence of wild-type bestrophin-1 ( Figure 5H/I). No differences of voltagedependence of currents in the presence of the DCT-PxxP mutant compared to that in the presence of wild-type bestrophin-1 were observed ( Figure 5F/G). CHO cells were analyzed by confocal microscopy after patch-clamp analysis ( Figure 6). Ca V 1.3 and b4subunits were localized in the cell membrane which was not changed by the presence of wild-type bestrophin-1. Under these conditions all three proteins were found in the cell membrane ( Figure 6A, Table 1) with relative surface expression values of 3.1960.13 for bestrophin-1, 3.0260.07 for b4-subunits and 3.6760.122 for Ca V 1.3 (all n = 3; Figure 6D). However, in the presence of DCT-PxxP mutant, a larger proportion of bestrophin-1, Ca V 1.3 and b4-subunit were found in the cytosol ( Figure 6B Figure 6D). The surface expression values measured in the presence of wild-type bestrophin-1 were significantly different to those measured in the presence of DCTPxxP bestrophin-1 (p = 0.02-0.002). As a control comparison, the effect of DCTPxxP mutant on non-interacting protein P2Y 2 receptor [31] was investigated by the same means ( Figure 6C). Here the P2Y 2 receptor was found in the cell membrane (surface expression value 5.1760.42; n = 3) whereas the DCTPxxP bestrophin-1 was found in the cytoplasm (surface expression value 1.2960.03; n = 3) indicating an independent trafficking of the two proteins (p = 0.0008).

Discussion
In four independent studies bestrophin-1 appears to function as a regulator of voltage-dependent L-type Ca 2+ channels [2,3,4,5]. We have newly discovered an additional cluster of highly conserved proline-rich motifs on the C-terminus of bestrophin-1 and show that this cluster is required for bestrophin-1 -dependent modulation of b-subunit function.
In order to study direct interaction of bestrophin-1 with Ca 2+ channel subunits, co-immunoprecipitation and co-localization experiments of heterologously expressed bestrophin-1 and different Ca 2+ channel subunits were performed. In our system, coprecipitation of Ca V 1.3 subunits with its physiological interaction partner b3-subunits could be observed [6,7]. Co-precipitation was independent of the expression system. Co-localization detection and co-precipitation were dependent on certain amino acid motifs on the C-terminus of bestrophin-1. Thus our experimental system allowed detecting physiological interaction between Ca 2+ channel subunits and regulatory proteins.
Heterologously expressed bestrophin-1 showed co-precipitation with b3or b4-subunits but not with Ca V 1.3 subunits. In the presence of b-subunits precipitation of Ca V 1.3 subunits resulted in indirect co-precipitation of bestrophin-1. Thus Ca V 1.3/bsubunits can form complexes with bestrophin-1 via binding of bestrophin-1 with b-subunits. Confocal microscopy of cells transfected with bestrophin-1 and b3-subunits showed a colocalization of the two proteins which was however more uniformly distributed in the cytoplasm. When the cells were transfected with Ca V 1.3, b3-subunit and bestrophin-1 or Ca V 1.3, b4-subunit and bestrophin-1, all three proteins were found to be localized in the cell membrane. This indicates close and direct interaction of bestrophin-1 with Ca 2+ channel b-subunits. However, the methods used here could only indicate direct interaction. A stronger proof of this interaction would require experiments showing detection of FRET (fluorescence resonance energy transfer) which is beyond the scope of this study. The presence of wild-type bestrophin-1 had two effects on the Ca V 1.3/b4 currents: an acceleration of the time-dependent activation and a reduction of ionic current density. The acceleration of the time-dependent activation has also been previously reported for b2-subunit modulation of Ca V 1.2 currents in heterologous expression and endogenously expressed L-type channels in a RPE cell line [2,3]. The reduction in the maximal activity was reported for b1-, b2and b4-subunit/bestrophin-1 interaction in the modulation of rat Ca V 1.3 currents [4] and for human Ca V 1.3/b4-subunit currents [5]. Since the gating currents were not different in the absence or presence of bestrophin-1, the reduction of the ionic current density was most likely not due to a reduced number of Ca V 1.3 subunits in the cell membrane. Thus, wild-type bestrophin-1 influences the ability of b-subunits to modulate the pore-function of Ca V 1.3 subunits. This differs from observations made by Yu et al. [4] who used only the C-terminus of bestrophin-1 and not full length bestrophin-1 for gating current analysis.
The binding of b-subunits and bestrophin-1 could depend on the interaction between SH3 domains of b-subunits [6,7,32] with proline-rich motifs, PxxP, present on the C-terminus of bestrophin-1. One cluster with two PxxP motifs is between the amino acid positions 330 and 346 and has been reported to be responsible for bestrophin-1/b-subunit interaction [4]. We found another cluster located between the amino acid positions 468-486 containing four PxxP motifs. To study its functional role, we made a deletion mutant lacking the PxxP motifs between amino acid positions 468-486 (named DCTPxxP). This mutant showed a reduced efficiency to co-precipitate with b-subunits by 70-80% depending on the isoform of b-subunit. However, the weak coprecipitation of DCTPxxP bestrophin-1 with b-subunits might result from the PxxP motifs between amino acid positions 330-346 which are still present. Furthermore, when studying indirect coprecipiation of Ca V 1.3/b4-subunit complex with bestrophin-1, we found no difference between wild-type bestrophin-1 and DCTPxxP mutant bestrophin-1. This can be explained by the occlusion of the SH3 domains in the free b-subunit crystal structure. [33,34,35]. It is hypothesized that the SH3 becomes accessible when the b-subunits bind to the Ca V -subunits [36]. Thus, bestrophin-1 can probably bind to b-subunits with higher efficiency when b-subunits are part of the Ca V 1.3/b-subunit complex.
The functional effect of PxxP motifs deletion between the amino acid positions 468-486 was studied by patch-clamp analysis of currents through human Ca V 1.3 subunit/b4-subunits expressed together with DCTPxxP-bestrophin-1. The presence of the DCTPxxP mutant only further decreased the ionic current density. Analysis of the subcellular localization revealed that these cells have a larger proportion of DCTPxxP bestrophin-1 and Ca V 1.3 subunits in the cytoplasm. Furthermore, in the presence of DCTPxxP bestrophin-1, the gating current density was strongly reduced compared to that of wild-type bestrophin-1.
Both observations indicate a reduced number of pore-forming Ca V 1.3 subunits in the cell membrane in the presence of DCTPxxP bestrophin-1. Since the DCTPxxP mutant bestrophin-1 still binds to the Ca V 1.3/b4-subunit complex, the reduced number of Ca V 1.3 subunits is due to an influence of the DCTPxxP bestrophin-1 on the ability of b-subunits to regulate Ca V -subunit surface expression. Thus, when the PxxP motifs between amino acid positions 468-486 on the bestrophin-1 Cterminus are lacking b-subunits show reduced ability to enhance surface expression of Ca V 1.3 subunits. To show that the effect of the DCTPxxP bestrophin-1 specifically influences trafficking Ca 2+ channel proteins, we expressed the DCTPxxP bestrophin-1 together with the P2Y 2 receptor which is known not to interact with bestrophin-1 [31]. Here, the P2Y receptor trafficked into the cell membrane whereas DCTPxxP bestrophin-1 stayed in the cytoplasm. Thus, the reduced amount of Ca 2+ channel protein in the cell membrane by the presence by DCTPxxP bestrophin-1 is not due to unspecific protein aggregation in the cytosol. In addition, it should be noted that the ionic current density was reduced by 40% but the gating current density was reduced the cell shape. 2E: Relative surface expression quantified by edge detection analysis (data are mean 6 SEM; n = 3). (* = p,0.05 for bestrophin-1; # = p,0.05 for b3-subunits, unpaired t-test) Scale bar represents 10 mm. doi:10.1371/journal.pone.0019364.g002 Figure 3. Detection of interaction sites between b-subunits and bestrophin-1. 3A: Bestrophin-1 construct used in this study and alignment of amino acid sequences of the C-terminus of the bestrophin-1 from different species (Boxes: transmembrane domains). Two among vertebrate species highly conserved clusters of proline-rich motifs (PxxP) could be detected. In the DCTPxxP mutant form, PxxP motifs between amino acid 468 to 486 were removed with unchanged recognition sites for the anti bestrophin-1 antibody. 3B: HEK-293 cells were transfected with b3-subunits together with bestrophin-1 or DCTPxxP constructs. Proteins were precipitated using anti-bestrophin-1 antibody and blots were visualized for anti-b3subunit to show co-immunoprecipitation. 3C: HEK-293 cells were transfected with His-tagged b4-subunits together with bestrophin-1 wild type or DCTPxxP constructs. Proteins were precipitated using anti-His antibody and the blots were visualized with anti-bestrophin-1 antibody to show coimmunoprecipitation. 3D: Relative co-immunoprecipitation of b3-subunits with either wild-type or DCTPxxP bestrophin-1: efficiency was measured by densitometry (n = 5). 3E: Relative co-immunoprecipitation of b4-subunits with either wild-type or tenfold. Thus, the DCTPxxP mutant lost its ability to decrease the single channel conductance compared to wild-type bestrophin-1. In summary, the loss of the PxxP cluster between the amino acid positions 468-486 did not alter the binding of bestrophin-1 to the Ca V 1.3/b-subunit complex but reduced the ability of b-subunits to guide pore-forming Ca V -subunits to the cell membrane and possibly influences the gating behavior of the Ca V 1.3 subunit.
b-subunits modulate electrophysiological properties of the pore-forming Ca V -subunits, help to transport Ca V -subunits into the cell membrane and interact with protein kinases for further modulation of Ca 2+ channel activity [8,9]. The absence of the PxxP motifs between amino acid positions 468-486 of bestrophin-1 impaired the ability of b-subunits to regulate the surface expression of pore-forming Ca V -subunits and possibly modulate single channel properties of the Ca V 1.3 channel. The absence of the PxxP motifs between the amino acid positions 330-346 reduced binding to b-subunits but did not affect the trafficking of Ca V 1.3 subunits to the cell membrane [5]. Thus, the PxxP motifs between amino acid positions 330-346 are more responsible for the binding of b-subunits and bestrophin-1 and the PxxP motifs between amino acid positions 468-486 help maintain the bsubunit regulatory properties over the surface expression of Ca Vsubunits.
We demonstrated that bestrophin-1 modulates human Ca V 1.3/ b4-subunits which are expressed in the RPE. Morbus Best patients show reduced light-peak in the electro-oculogram [13,37,38]. Mouse models which show the same phenotype as Best patients are the Ca V 1.3 knock-out mouse [17] and the lethargic mice which are a natural knock-out of the b4-subunit [16]. Thus Ca V 1.3/b4-subunits are of importance in the generation of the light-peak. The decreased light-peak in Best patients would probably result from altered modulation of L-type channels composed of Ca V 1.3/b4 subunits. Since L-type channels of the RPE regulate cell functions such as secretion or phagocytosis, the understanding of bestrophin-1 and Ca 2+ channel interaction would then help to understand mechanisms of retinal degeneration [39,40]. In summary, we were able to demonstrate the physical binding of full length bestrophin-1 with b3and b4-subunits of Ca 2+ channels. Additionally, we identified a new cluster of PxxP motifs on the C-terminus of bestrophin-1 which is needed for the interaction of bestrophin-1 with Ca 2+ channel b-subunits and probably other SH3 domain carrying proteins. . 5D: Comparison of the ionic current density of L-type currents from heterologously expressed Ca 2+ channel proteins and different bestrophins measured at +20 mV (note: the values are close to those we have published [5] but represent a different set of data). 5E: Comparison of the gating current density in the presence of 10 mM Co 2+ from heterologously expressed Ca 2+ channel proteins and different bestrophins measured at +20 mV. 5F: Normalized current/voltage plots of L-Type currents either in the absence, presence of bestrophin-1 or the DCTPxxP mutant bestrophin-1, fit by Boltzmann equation. 5G: Comparison of the voltages of half maximal activation obtained from Boltzmann fits of the curves in Fig. 5F. 5H: Whole cell Ba 2+ -currents measured from CHO cells expressing either Ca V 1.3, b4-, a2d1 subunits or Ca V 1.3, b4-, a2d1 subunits plus wt-bestrophin-1. Currents were elicited by a voltage-jump from -70 mV to +20 mV. The recording shows the currents normalized to their maximal amplitude for comparison. 5I: Comparison of the time-dependent activation of L-type currents from different heterologously expressed Ca 2+ channel proteins and bestrophins measured as time constant from single-exponential fit of the currents (note: the values are close to those we have published in [5] but represent a different set of data). (* = p,0.05; ** = p,0.01, unpaired t-test; n depicts the number of experiments) doi:10.1371/journal.pone.0019364.g005  Table 2. Summary of basic electrophysiological data measured at +20 mV (except voltage of maximal current amplitude) of Ca V 1.3 channel currents in the presence of different b-subunits and bestrophins.

Supporting Information
Figure S1 S1A: Control experiments: Proteins detected in transfected CHO cells are products of the plasmids used for transfection. Western blot of proteins isolated from CHO cells either transfected or not transfected by the corresponding plasmid.
Only when the cells were transfected by plasmids carrying bestrophin-1, b3-subunit, bestrophin-1-GFP or Ca V 1.3 subunits the Western blots revealed the presence of the corresponding proteins (bestropin-1; 68kDa, b3 subunit; 55kDa, bestrophin-1-GFP; 100kDa, and Cav1.3 subunit; 240 kDa). Anti b-actin antibody was used as a loading control. S1B: Physical interaction between bestrophin-1 and auxiliary b3-subunits of voltagedependent Ca 2+ channels was independent from the cell line which was used as transfection system: CHO, COS-7, and HEK-293 cells were co-transfected with bestrophin-1-GFP and b3subunits, and b3-subunits and GFP. Precipitates were obtained using anti GFP antibodies, and Western blots were stained using anti-b3 and anti-GFP antibodies. b3-subunits were detected only when cells were transfected with bestrophin-1-GFP fusion construct and b3-subunits but not when cells were transfected with b3-subunits and GFP vector.