Complex Regulation of Voltage-dependent Activation and Inactivation Properties of Retinal Voltage-gated Cav1.4 L-type Ca2+ Channels by Ca2+-binding Protein 4 (CaBP4)*

Background: Cav1.4 provides sustained influx of Ca2+ into photoreceptor terminals, which is required for continuous release of glutamate and normal vision. Results: We characterize the effects of CaBP4 on Cav1.4. Conclusion: CaBP4 increases Cav1.4 channel availability. CaBP4 mutations lead to reduced Cav1.4 availability and to retinal disease. Significance: CaBP4 tailors voltage gating of Cav1.4 for its function in photoreceptor terminals. Cav1.4 L-type Ca2+ channels are crucial for synaptic transmission in retinal photoreceptors and bipolar neurons. Recent studies suggest that the activity of this channel is regulated by the Ca2+-binding protein 4 (CaBP4). In the present study, we explored this issue by examining functional effects of CaBP4 on heterologously expressed Cav1.4. We show that CaBP4 dramatically increases Cav1.4 channel availability. This effect crucially depends on the presence of the C-terminal ICDI (inhibitor of Ca2+-dependent inactivation) domain of Cav1.4 and is absent in a Cav1.4 mutant lacking the ICDI. Using FRET experiments, we demonstrate that CaBP4 interacts with the IQ motif of Cav1.4 and that it interferes with the binding of the ICDI domain. Based on these findings, we suggest that CaBP4 increases Cav1.4 channel availability by relieving the inhibitory effects of the ICDI domain on voltage-dependent Cav1.4 channel gating. We also functionally characterized two CaBP4 mutants that are associated with a congenital variant of human night blindness and other closely related nonstationary retinal diseases. Although both mutants interact with Cav1.4 channels, the functional effects of CaBP4 mutants are only partially preserved, leading to a reduction of Cav1.4 channel availability and loss of function. In conclusion, our study sheds new light on the functional interaction between CaBP4 and Cav1.4. Moreover, it provides insights into the mechanism by which CaBP4 mutants lead to loss of Cav1.4 function and to retinal disease.


Cav1.4 L-type Ca 2؉ channels are crucial for synaptic transmission in retinal photoreceptors and bipolar neurons. Recent studies suggest that the activity of this channel is regulated by the Ca 2؉ -binding protein 4 (CaBP4). In the present study, we explored this issue by examining functional effects of CaBP4 on heterologously expressed Cav1.4. We show that CaBP4 dramatically increases Cav1.4 channel availability. This effect crucially depends on the presence of the C-terminal ICDI (inhibitor of Ca 2؉ -dependent inactivation) domain of Cav1.4 and is absent in a Cav1.4 mutant lacking the ICDI.
Using FRET experiments, we demonstrate that CaBP4 interacts with the IQ motif of Cav1. 4 and that it interferes with the binding of the ICDI domain. Based on these findings, we suggest that CaBP4 increases Cav1.4 channel availability by relieving the inhibitory effects of the ICDI domain on voltage-dependent Cav1.4 channel gating. We also functionally characterized two CaBP4 mutants that are associated with a congenital variant of human night blindness and other closely related nonstationary retinal diseases. Although both mutants interact with Cav1.4 channels, the functional effects of CaBP4 mutants are only partially preserved, leading to a reduction of Cav1.4 channel availability and loss of function. In conclusion, our study sheds new light on the functional interaction between CaBP4 and Cav1.4. Moreover, it provides insights into the mechanism by which CaBP4 mutants lead to loss of Cav1.4 function and to retinal disease.
Ribbon synapses of retinal photoreceptors and bipolar cells encode graded and sustained changes in membrane potential that are generated during photoreception and maintained over long periods of time throughout the duration of a light stimulus. These graded signals are coupled with high fidelity to tonic glutamate release via permanent Ca 2ϩ influx through presynaptic Cav1.4 L-type Ca 2ϩ channels. Cav1.4 channels are tailored for their function as long-lasting presynaptic Ca 2ϩ entry pathways by a lack of Ca 2ϩ -dependent inactivation (CDI) 3 and very slow voltage-dependent inactivation (1)(2)(3)(4)(5)(6). In Cav1.4 channels, CDI is specifically abolished by a recently discovered inhibitory domain (ICDI: inhibitor of CDI) in the C-terminal tail of the Cav1.4 channel, which binds to upstream C-terminal regulatory motifs (4,5).
Cav1.4 channels are multiprotein complexes. The core Ca 2ϩ channel complex is composed of the principal pore-forming ␣ 1 subunit and the auxiliary ␤ 2 and ␣ 2 ␦ subunits. Mutations in the gene coding for the ␣ 1 subunit of Cav1.4 channels have been identified in patients suffering from congenital stationary night blindness type 2 (CSNB2) (7). This retinal disease is characterized by loss of neurotransmission from rods to second order bipolar cells, which is attributable to a loss of Cav1.4. These findings are corroborated by the analysis of a genetic mouse model deficient for the Cav1.4 ␣ 1 subunit (8,9). In addition, in the mouse, deletion of another component of the Cav1.4 channel complex, the auxiliary ␤ 2a subunit, also leads to CSNB2-like phenotype (10).
Recently, it has been shown that mutations in the gene coding for CaBP4 are also associated with CSNB2 and other closely related nonstationary retinal diseases in humans (11)(12)(13)(14). In line with these findings, deletion of CaBP4 in the mouse leads to a CSNB2-like phenotype (15). CaBP4 is a neuronal Ca 2ϩ sensor protein with structural homology to calmodulin (CaM). Like CaM, CaBP4 contains two globular domains, the N-and the C-lobe, each containing a pair of EF-hand motifs connected by a central helix (16 -18). The second EF-hand in the N-lobe of CaBP4 is nonfunctional. It has been shown that CaBP4 is part of the Cav1.4 channel complex in the retina (15). In addition, there is evidence that CaBP4 interacts with the C terminus of Cav1.4. However, the exact binding motif in Cav1.4 is unknown. CaBP4 has been implicated in the modulation of voltage-dependent Cav1.4 activation (15). Beyond these initial observations, the physiological effects of CaBP4 on Cav1.4 channels remains incompletely understood. The aim of this study was to explore the details of the role of CaBP4 in the regulation of Cav1.4 channel function. We provide evidence that CaBP4 has differential effects on different aspects of Cav1.4 channel function such as voltage-dependent activation and inactivation as well as CDI. We also characterized the functional consequences of two mutations in CaBP4 that are associated with human retinal disease.

EXPERIMENTAL PROCEDURES
Constructs for Electrophysiology-For expression of murine Cav1.4 ␣ subunit wild type channel (1) (accession number AJ579852) and Cav1.4⌬ICDI (6), stable HEK293 cell lines were generated using the Flp-In TM system (Invitrogen) according to manufacturer's protocol. Wild type CaBP4 (accession number NM144532, CaBP4 of Mus musculus, corresponds to NM145200, CaBP4 of Homo sapiens) was cloned into pIRES2-EGFP (Clontech) expression vector by using the restriction sites: BglII and SalI. In addition to murine wild type CaBP4, two mutant variants of CaBP4 that are associated with autosomal recessive forms of human congenital retinal disease CSNB2 were cloned (see Fig. 5A). In CaBP4-R216X, a single nucleotide exchange (C646T) replaces an arginine residue at position 216 by a premature stop codon (13). This mutation leads to a truncated CaBP4 protein lacking the C-lobe containing EF-hands 3 and 4. In the second mutation (CaBP4-E267fs), a deletion of two nucleotides causes a frameshift at the last residue of EFhand 4, which elongates the protein by 91 novel amino acid residues (12). The mutations were introduced into mouse CaBP4 sequence to keep in line with the findings in genetic mouse models. This approach is valid as mouse Cav1.4 and CaBP4 proteins are more than 94 and 83% homologous to their respective human counterparts. Secondly, mutations in the genes coding for Cav1.4 or CaBP4 both lead to a similar phenotype in the mouse and human. Thirdly, functional properties of Cav1.4 as well as regulation by CaM and ICDI domain are also similar in mouse and human. According to the numbering of the mouse Cav1.4 protein sequence, human CaBP4-R216X and human CaBP4-E267fs correspond to murine CaBP4-R212X and CaBP4-E263fs, respectively. Throughout this study, we adhere to human terminology.
Electrophysiology-HEK293 cells stably expressing Cav1.4 ␣ subunits (wild type and Cav1.4⌬ICDI) were transiently transfected with expression vectors encoding the calcium channel ␤ 2a (accession number: X64297) and ␣ 2 ␦ 1 (accession number: M21948) as described in Ref. 1. In some cases, wild type CaBP4 or CaBP4 variants were also transfected (1.5 g from each construct). I Ca and I Ba were measured by using the following solutions: pipette solution, 112 mM CsCl, 3 mM MgCl 2 , 3 mM Mg-ATP, 10 mM EGTA, 5 mM HEPES, adjusted to pH 7.4 with CsOH; standard bath solution, 102 mM NaCl, 10 mM BaCl 2 or 10 mM CaCl 2 , 5.4 mM CsCl 2 , 1 mM MgCl 2 , 20 mM tetraethylam-monium chloride, 5 mM HEPES, 10 mM glucose, adjusted to pH 7.4 with NaOH. For experiments carried out in 2 mM Ca 2ϩ in the extracellular solution, Ca 2ϩ was equimolarily replaced by Na ϩ . I Ca and I Ba were measured from the same cell. Bath solution was changed by a local solution exchanger. Currents were recorded at room temperature 1-3 days after transfection by using whole-cell patch clamp technique. Data were analyzed by using Origin 7.5 software (MicroCal).
The peak I/V relationship was measured by applying 350-ms voltage pulses to potentials between Ϫ80 and ϩ70 mV in 10-mV increments from a holding potential of Ϫ80 mV (see Fig. 2A). For determination of half-maximum activation voltage (V 0.5 ), the chord conductance (G) was calculated from the current voltage curves by dividing the peak current amplitude by its driving force at that respective potential G ϭ I/(V m Ϫ V rev ), where V rev is the interpolated reversal potential, V m is the membrane potential, and I is the peak current. The chord conductance was then fitted with a Boltzmann equation G ϭ G max / ͑1 ϩ e ͑V0.5 Ϫ Vm͒/kact ͒, where G max is the maximum conductance, V 0.5 is the half-maximum activation voltage, V m is the membrane potential, k act is the slope factor of the activation curve and e is the Euler's number.
Cav1.4 channel inactivation was quantified by calculating the fraction of peak Ba 2ϩ and Ca 2ϩ currents remaining after 350 ms of depolarization (R 350 ) as described in Ref. 4. R 350 is used to quantify CDI.
Half-maximum inactivation voltage (V 0.5,inact ) was determined using a pseudo-steady-state inactivation protocol (19 -21) rather than true steady-state inactivation for the following reasons. For the induction of complete inactivation, a prepulse length of more than 30 s and a recovery interval between individual trials of more than 10 -20 s are required. For the recovery from ultraslow inactivation, even more than 10 min have been reported (22). To efficiently record inactivation, we adjusted the protocol and decreased prepulse lengths and intertrial interval as follows. From a holding potential of Ϫ80 mV, a series of 2.5-s conditioning prepulses to various voltages between Ϫ100 mV and ϩ50 mV was used. The conditioning pulse was followed by a 10-ms return to the holding potential and a 150-ms test pulse to ϩ10 mV. The individual trials of the protocol were applied at a time interval of 10 s (see Fig. 2B).
The conditioning prepulse duration of 2.5 s was sufficient to induce considerable open state inactivation, which was comparable with the 5-s prepulse used in Ref. 1. The time interval between trials (10 s) was sufficient to recover the majority of inactivated channels. Ultraslow inactivation of Cav1.4, which needs more than 10 min to recover (22), was not addressed in the present study. Therefore, we cannot fully exclude the possibility that a small fraction of channels did not completely recover from ultraslow inactivation, leading to some cumulative inactivation that may sequentially add up in subsequent trials of the inactivation protocol.
The family of current traces obtained by applying the pseudosteady-state inactivation protocol was analyzed by normalizing tail currents immediately after the test pulse to ϩ10 mV to maximum current amplitude and plotted as a function of the preceding membrane potential. The data points were fitted with the Boltzmann function: I ϭ 1/(1 ϩ e ͑Vm Ϫ V0. 5 V m is the test potential, V 0.5,inact is the half-maximum voltage for pseudo-steady-state inactivation, k inact is the slope factor of the curve and e is the Euler's number. The liquid junction potential calculated according to Ref. 23 was 5.4 mV for extracellular solution containing 10 mM Ca 2ϩ and 5.0 mV for extracellular solutions containing 2 mM Ca 2ϩ . Unless stated otherwise, the liquid junction potential was not corrected. Constructs for Photometric FRET Experiments-For all FRET constructs, we used monomeric enhanced YFP (A206K) or enhanced CFP (A206K) (24,25) to avoid homo-and heterodimerization. For simplicity, we refer to the different fluorescent proteins as "YFP" and "CFP" throughout the text. All FRET constructs were cloned into the pcDNA3 expression vector (Invitrogen).
Measurements of single-cell FRET based on aggregate (nonspatial) fluorescence recordings were performed using threecube FRET as described previously (26,27). The notation and abbreviations used follow the definitions of Refs. 26 and 27. The degree of FRET in an individual cell was quantified using the FRET ratio (FR), which is defined as the fractional increase in YFP emission caused by FRET. The FR was calculated using Fluorescence measurements for the determination of S FRET , S CFP , and S YFP were performed in cells coexpressing CFPtagged and YFP-tagged peptides or intramolecular FRET constructs dually labeled by CFP and YFP using the following parameters: S CFP , excitation at 440 Ϯ 7.5 nm and emission at 483 Ϯ 16 nm (donor excitation; donor emission); S FRET , excitation at 440 Ϯ 7.5 nm and emission at 542 Ϯ 13.5 nm (donor excitation; acceptor emission); and S YFP , excitation at 505 Ϯ 7.5 nm and emission at 542 Ϯ 13.5 nm (acceptor excitation; acceptor emission). R D1 , R A1 , and R D2 are experimentally predetermined constants from measurements applied to single cells expressing only CFP-or YFP-tagged molecules. These constants are used to correct for bleed-through of CFP into the YFP channel (R D1 ), direct excitation of YFP by CFP excitation (R A1 ), and the small amount of CFP excitation at the YFP excitation wave-length (R D2 ) as described by Refs. 26 and 27. Throughout all FRET experiments, the CFP:YFP ratio was between 1:10 and 20:1, which can be expected to yield reliable results (supplemental Fig. S1) (5,28).
Western Blots of HEK293 Cells-Recombinant CFP-tagged human CaBP4 mutations (CaBP4 R216X and CaBP4 E267fs) and wild type CaBP4 were expressed in HEK293 cells using the calcium phosphate transfection method. Western blot protocol was performed as published previously (29) using an antibody against GFP (Clontech).
Data Analysis and Statistics-Electrophysiology data were analyzed using Clampfit 10.0 (Molecular Devices) and Origin 7.5 (MicroCal). FRET data were analyzed using Microsoft Office Excel software (Microsoft). Data plotting, curve fitting, and statistical analysis were performed using Origin 7.5 (MicroCal). All values are presented as mean Ϯ S.E. for the indicated number n of experiments. An unpaired Student's t test was performed for the comparison between two groups. Significance was tested by analysis of variance followed by Dunnett's test if multiple comparisons were made. Values of p Ͻ 0.05 were considered significant. The determination of molar CFP/YFP ratios and the determination of the window conductances are described in supplemental methods.

RESULTS
To characterize the functional effect of CaBP4 on Cav1.4 channels, we first tested the effect on CDI. We compared Ca 2ϩ and Ba 2ϩ currents through Cav1.4 in transfected HEK293 cells in the absence of CaBP4 (Fig. 1A), which is not endogenously expressed (supplemental Fig. S2), and in the presence of overexpressed CaBP4 (Fig. 1B). Wild type channels did not display any CDI under both conditions (Fig. 1, A and B). Truncation of

CaBP4 Regulates Cav1.4 Channels
the ICDI domain in the distal C terminus of Cav1.4 unmasked CDI, which is caused by endogenous CaM (4) (Fig. 1C). In contrast, CaBP4 completely abolished CDI in Cav1.4⌬ICDI channels (Fig. 1D). This finding indicated that both CaBP4 and the ICDI domain have equivalent functions with regard to their effect on CDI. Furthermore, these observations suggested that CaM and CaBP4 stabilize the channel in a different conformational state that gives rise to the absence of CDI in the case of CaBP4 and to the presence of CDI in the case of CaM. To study the effect of CaBP4 on voltage-dependent gating of Cav1.4, we determined activation and inactivation curves for Ca 2ϩ and Ba 2ϩ currents in the absence and presence of CaBP4 (Figs. 2 and 3). From these experiments, it was evident that CaBP4 shifted the activation curve of Cav1.4 to more hyperpolarized potentials (Fig. 3A, Table 1) and increased the steepness of the activation curve (lower k act ; Table 1). Voltage-dependent inactivation gating was characterized using a pseudo-steady-state inactivation protocol (Fig. 2B) rather than true steady-state inactivation for reasons outlined in detail under "Experimental Procedures". The protocol is well suited to characterize overall Cav1.4 channel inactivation consisting of open state inactivation and a small amount of closed state inactivation. These two components were not further dissected. Using this protocol, CaBP4 dramatically reduced pseudo-steady-state inactivation (Fig. 3, B and C). Together the effects of CaBP4 on activation and inactivation lead to a pronounced increase of the window conductance caused by the overlap of the activation and the inactivation curve (Fig. 3C, Tables 1 and 2). We determined the window conductance for Cav1.4 channels in extracellular recording solution containing 2 mM Ca 2ϩ , which is close to the physiological extracellular Ca 2ϩ concentrations (supplemental Fig. S3). As compared with standard extracellular recording solution containing 10 mM Ca 2ϩ , the window conductance was shifted to more hyperpolarized potentials. We used the experimentally determined voltage shift to correct the window conductances obtained in experiments using standard recording solutions (10 mM extracellular Ca 2ϩ ; supplemental Fig. S4). A comparison revealed that in the presence of CaBP4, the predicted window conductance for the physiological conditions of 2 mM Ca 2ϩ was more than 2-fold larger than in the absence of CaBP4 (supplemental Figs. S3-S5). To account for the physiological voltage range observed in photoreceptors (less than Ϫ30 mV (30 -34)), we also calculated window conductances at Ϫ30 and Ϫ40 mV (supplemental Fig. S5). We found that the window conductance was 2.2-or 1.7-fold higher in the presence of   OCTOBER 19, 2012 • VOLUME 287 • NUMBER 43

JOURNAL OF BIOLOGICAL CHEMISTRY 36315
CaBP4 at Ϫ30 or Ϫ40 mV, respectively. Together our findings indicated that CaBP4 markedly increased overall Cav1.4 channel availability. This effect was equally pronounced for Ca 2ϩ and Ba 2ϩ currents, indicating that this effect is not Ca 2ϩ dependent (Fig. 3C).
Similar experiments were carried out for Cav1.4⌬ICDI channels. Surprisingly, in the absence of the ICDI domain, CaBP4 had no effect on voltage-dependent activation and inactivation and window currents (Fig. 4, A and B, supplemental Fig. S4, C and D, Tables 1 and 2). Moreover, voltage-dependent activation and inactivation of Cav1.4⌬ICDI channels were very similar to that of wild type Cav1.4 channels in the presence of CaBP4. In contrast, in the absence of CaBP4, Cav1.4⌬ICDI channels displayed CDI (Fig. 1C). In summary, in the absence of the ICDI domain, CaBP4 has no effect on voltage-dependent gating (Fig. 4) but blocks CDI (Fig. 1D). Together these observations indicate that CaBP4 modulates the functional impact of the ICDI domain in a complex fashion. With respect to voltagedependent gating of Cav1.4 channels, the ICDI domain and CaBP4 have opposing functional effects. The ICDI domain shifts the activation curve of Cav1.4 to more depolarized potentials and increases inactivation, whereas CaBP4 shifts the activation curve of Cav1.4 to more hyperpolarized potentials and decreases inactivation. With respect to the inhibition of the CDI, the ICDI domain and CaBP4 have equivalent functional effects.
To further examine the interaction between CaBP4 and Cav1.4, FRET experiments were carried out using YFP-tagged variants of Cav1.4 C terminus and CFP-CaBP4 (Fig. 5). We found that CaBP4 interacts with the C terminus of both Cav1.4 wild type and Cav1.4⌬ICDI (Fig. 5, A and B). Truncation of the C terminus downstream of the IQ motif (EF-IQ) did not significantly change FRET when coexpressed with CaBP4 as compared with the complete C terminus of Cav1.4. In contrast, there was no FRET between the EF-hand motif and CaBP4. One important possibility would be that CaBP4 binds to the IQ motif in the C-terminal tail of Cav1.4. Binding of other CaBPs and of CaM to the IQ motif of high voltage-activated and low voltage-activated Ca 2ϩ channels including Cav1.4 has been reported (6,27,(35)(36)(37). To test this possibility in our experimental setting, we performed FRET experiments using a mutant of the C terminus of Cav1.4, in which five amino acid residues within the IQ motif are replaced by alanines (1.4/5A). The FRET between CaBP4 and 1.4/5A was very low. The most likely interpretation of this finding is that CaBP4 does not interact with 1.4/5A, indicating that CaBP4 and CaM share the same binding domain on the C terminus of Cav1.4 (Fig. 5B).

TABLE 1 Voltages for half-maximum activation (V 0.5,act ) and slope values (k act ) from the measurements shown in Figs. 2-4 and 6 and supplemental Fig. S3, respectively
* ϭ significantly different compared to measurements of Cav1.4 (with 10 mM Ca 2ϩ or 10 mM Ba 2ϩ in the extracellular solution). # ϭ significantly different compared to measurements of Cav1.4 in presence of CaBP4 (with 10 mM Ca 2ϩ or 10 mM Ba 2ϩ in the extracellular solution). p Ͻ 0.05 for one symbol (* or #), p Ͻ 0.01 for two symbols, p Ͻ 0.001 for three symbols. n ϭ number of cells.

CaBP4 Regulates Cav1.4 Channels
We next tested the possibility whether CaBP4 also associates with the N terminus of Cav1.4 channels. For Cav1.2 and Cav1.4 channels, CaM or CaBP1 binding to the N terminus has been reported (14,39,40). However, we found that the FRET signal between CaBP4 and the N terminus of Cav1.4 was very low (Fig.  5B), indicating weak or no interaction. Alternatively, the positioning or the angle between the fluorescent tags may be unfavorable for FRET.
To assess whether CaBP4 interferes with binding of the ICDI domain, we compared the FRET signal between the ICDI domain and the C terminus of Cav1.4 lacking the ICDI domain in the absence and in the presence of CaBP4 (Fig. 5C). These experiments revealed that binding of the ICDI domain was impaired in the presence of CaBP4. In agreement with these coexpression experiments, CaBP4 significantly decreased intramolecular FRET for the C terminus of Cav1.4 that was Nand C-terminally tagged with YFP and CFP, respectively (Fig.  5C). A direct interaction between CaBP4 and the ICDI domain is not likely because the FRET signal between CaBP4 and 1.4/5A was very low (Fig. 5B). This finding indicates that besides the IQ motif, other CaBP4 binding regions within the C terminus including the ICDI domain are unlikely.
We next set out to analyze the functional consequences of two mutations in CaBP4, which are associated with autosomal recessive forms of human congenital retinal disease CSNB2 (Fig. 5A). The first mutation (CaBP4-R216X) results in a truncated CaBP4 protein lacking the C-lobe containing EF-hands 3 and 4 (13). In the second mutation (CaBP4-E267fs), the last residue of EF-hand 4 (Glu-267) is exchanged to valine followed by a frameshift elongating the protein (12). Western blot experiments confirmed the expression of mutant CaBP4 variants in HEK cells (Fig. 5D). In FRET experiments, we found that CaBP4-E267fs and CaBP4-R216X interact with the C terminus of Cav1.4 (Fig. 5E). The FRET signal of both mutant CaBP4s was even higher than that of wild type CaBP4. The finding that CaBP4-R216X interacts with the C terminus of Cav1.4 is very interesting because it shows that the N-lobe of CaBP4 is sufficient to interact with Cav1.4 with a FRET ratio (FR) even higher than wild type CaBP4. Furthermore, the presence of CaBP4-E267fs and CaBP4-R216X partially impairs binding of the ICDI domain (Fig. 5C).
The functional impact of the CaBP4 mutations was analyzed in electrophysiological experiments (Fig. 6). After coexpression of CaBP4-R216X or CaBP4-E267fs together with wild type Cav1.4 channels, Ca 2ϩ currents did not display CDI (not shown). In contrast, in coexpression experiments of mutated CaBP4s together with Cav1.4⌬ICDI, slowly developing CDI could be observed, which was pronounced for CaBP4-R216X and subtle but detectable for CaBP4-E267fs (Fig. 6, A-C). The appearance of slow CDI in coexpression experiments of Cav1.4⌬ICDI channels and mutant CaBP4 suggests that CaM can access the effector site for CDI at least in some of the channels. We next analyzed the effect of CaBP4 mutations on voltage-dependent gating of Cav1.4 (Fig. 6, D and E). In contrast to wild type CaBP4, both CaBP4 mutations did not shift voltagedependent activation to more hyperpolarized potentials. For CaBP4-R267fs, the activation curve was even shifted by about 6 mV to more depolarized potentials in experiments using Ca 2ϩ as charge carrier (Fig. 6E, Tables 1 and 2). Slope factors of both CaBP4 mutations were significantly higher than those observed in the presence of CaBP4 and very similar to that observed in the absence of CaBP4 (Table 1). CaBP4 mutants, like wild type CaBP4, functionally antagonize the inactivation process (Fig. 6,  D and E). However, the antagonizing effect of mutant CaBP4s was significantly less pronounced as compared with wild type CaBP4. For CaBP4-R216X, this effect was only observed using Ba 2ϩ as charge carrier (Fig. 6D). Together the observed effects in CaBP4 mutants significantly decreased the predicted window conductance as compared with wild type CaBP4 (2-or 3-fold for CaBP4-R216X at Ϫ40 or Ϫ30 mV, respectively and 3or 4-fold for CaBP4-E267fs at Ϫ40 or Ϫ30 mV, respectively;  Fig. S3, respectively * ϭ significantly different compared to measurements of Cav1.4 (with 10 mM Ca 2ϩ or 10 mM Ba 2ϩ in the extracellular solution). # ϭ significantly different compared to measurements of Cav1.4 in presence of CaBP4 (with 10 mM Ca 2ϩ or 10 mM Ba 2ϩ in the extracellular solution). p Ͻ 0.05 for one symbol (* or #), p Ͻ 0.01 for two symbols, p Ͻ 0.001 for three symbols. n ϭ number of cells.

DISCUSSION
Here, we describe the functional effects of CaBP4 on wild type Cav1.4 channels. The main effect of CaBP4 is a profound increase in Cav1.4 channel availability. We propose the following mechanism to explain this finding (Fig. 7). In wild type Cav1.4 channels, the ICDI domain shifts the activation curve to the right and induces voltage-dependent inactivation. CaBP4 antagonizes these effects. In line with this hypothesis, in the presence of CaBP4, the ICDI domain is functionally silent. Furthermore, in Cav1.4 channels lacking the ICDI domain, CaBP4 has no effects on Cav1.4 voltage dependence. These observations suggest that Cav1.4 channels lacking the ICDI domain are    (30 -34). Besides the effects on voltage-dependent channel availability, CaBP4 also blocks CDI. This effect is only evident in Cav1.4 channels lacking the ICDI domain. With respect to its inhibitory effect on CDI, CaBP4 and the ICDI domain are functionally equivalent.
In FRET experiments, we find that CaBP4 tightly associates with the IQ motif of Cav1.4 channels and that CaBP4 is able to displace CaM from binding to the IQ motif at physiological conditions. In line with this interpretation, we find that CaBP4 can very efficiently regulate the functional properties of Cav1.4 channels in HEK293 cells in which CaM is endogenously expressed at high levels. It is very likely that binding of CaBP4 induces a conformation different from the conformation in the presence of CaM. This difference could be the reason for differential effects of CaBP4 and CaM on CDI in Cav1.4⌬ICDI channels.
How can the functional effects of CaBP4 on Cav1.4 voltage gating mechanistically and structurally be explained? Our FRET experiments demonstrate that CaBP4 decreases binding of the ICDI domain to the proximal C terminus of Cav1.4. In structural terms, one possible explanation could be that CaBP4 partially displaces the ICDI domain. In line with this idea is the observation that the FRET signal between the ICDI domain and C terminus of Cav1.4 is only reduced but does not equal zero (FR ϭ 1). Partial departure of the ICDI domain could selectively abolish the effects of the ICDI domain on voltage-dependent Cav1.4 channel gating. Our results that CaBP4 mutants both interact with the C terminus of Cav1.4 but lack most of the effects of wild type CaBP4 suggest that both lobes of CaBP4 need to be present and act in concert to produce a conformational change that regulates voltage-dependent activation and inactivation.
Our model also gives insight into the pathomechanism of CaBP4 mutants identified in CSNB2 patients. A common feature of CaBP4 mutants is a significant reduction of overall channel availability as compared with wild type CaBP4. We find that in CaBP4 mutants, there was no negative shift of the voltagedependent activation curve and no change in the slope of the activation curve. This suggests that CaBP4 mutants, although bound to the channel, cannot antagonize the effect of the ICDI domain on voltage-dependent activation. In CaBP4-R216X, this could be because the functional C-lobe is missing. In the absence of the C-lobe, CaBP4-R216X could indeed interact with the channel, but would not induce the conformational change of Cav1.4-CaBP4 complex, which is required for the interference with the ICDI domain. In CaBP4-E267fs, both lobes are present, but the function of the C-lobe could be affected by the exchange of the last amino acid residue in the fourth EF-hand and the additional nonsense sequence. Interestingly, for CaBP4-E267fs, even a significant shift of the activation curve toward more positive potentials was observed. A possible explanation is that the CaBP4 mutant stabilizes the channel in a conformational state that intensifies the interaction with the ICDI domain, shifting the activation curve even toward more positive potentials. Alternatively, the interaction of the ICDI domain and the channel could be intensified by completely displacing endogenous CaM (Fig. 7). In both CaBP4 mutants, voltage-dependent inactivation was more pronounced as compared with wild type CaBP4. This indicates that mutant CaBPs can only partially induce the conformational change required for the full wild type CaBP4 effect. In total, the overall window currents and the physiological relevant window currents are both significantly reduced. These effects are more The ICDI domain and CaBP4 have opposing functional effects on voltage-dependent gating of Cav1.4 channels. The ICDI domain shifts the activation curve of Cav1.4 to more depolarized potentials and increases inactivation (black curve: ϩICDI), whereas CaBP4 shifts the activation curve of Cav1.4 to more hyperpolarized potentials and decreases inactivation (green curve: ϩCaBP4). In addition, with respect to the inhibition of the CDI, ICDI and CaBP4 are functionally equivalent. The voltage axis is in arbitrary units; the physiological voltage range of operation in photoreceptor cells is shown in green. The intensity of interaction between the ICDI domain and Cav1.4 is represented by the gray triangle. The black activation curve corresponds to intermediate interaction strength of the ICDI in the presence of endogenous CaM. The green activation curve indicates weak interaction between the ICDI and Cav1.4. The purple activation curve represents tight interaction between the ICDI and the channel, which may be observed in the absence of CaBP4 and endogenous CaM. Boxes summarize the conditions under which the respective activation curve is observed. ϩICDI refers to the wild type Cav1.4 channel; ϪICDI refers to truncated channels lacking the ICDI domain. Furthermore, in the boxes, the presence or absence of endogenous CaM or CaBP4 is given. OCTOBER 19, 2012 • VOLUME 287 • NUMBER 43 pronounced in CaBP4-R216X. The appearance of slow CDI in Cav1.4⌬ICDI channels coexpressed with mutant CaBP4 suggests that CaM can access the effector site for CDI at least in some of the channels.

CaBP4 Regulates Cav1.4 Channels
Finally, the effect of CaBP4 on CDI may be relevant in a Cav1.4 variant lacking the ICDI domain. It has been shown that Cav1.4 splice variants lacking the ICDI domain exist under physiological conditions (38). In these channels, CDI is predicted to be switched off by the action of CaBP4. Similar regulation has been observed for Cav1.3 channels in inner hair cells of the auditory system (35).