alpha 1D (Cav1.3) subunits can form l-type Ca2+ channels activating at negative voltages.

In cochlea inner hair cells (IHCs), L-type Ca(2+) channels (LTCCs) formed by alpha1D subunits (D-LTCCs) possess biophysical and pharmacological properties distinct from those of alpha1C containing C-LTCCs. We investigated to which extent these differences are determined by alpha1D itself by analyzing the biophysical and pharmacological properties of cloned human alpha1D splice variants in tsA-201 cells. Variant alpha1D(8A,) containing exon 8A sequence in repeat I, yielded alpha1D protein and L-type currents, whereas no intact protein and currents were observed after expression with exon 8B. In whole cell patch-clamp recordings (charge carrier 15-20 mm Ba(2+)), alpha1D(8A) - mediated currents activated at more negative voltages (activation threshold, -45.7 versus -31.5 mV, p < 0.05) and more rapidly (tau(act) for maximal inward currents 0.8 versus 2.3 ms; p < 0.05) than currents mediated by rabbit alpha1C. Inactivation during depolarizing pulses was slower than for alpha1C (current inactivation after 5-s depolarizations by 90 versus 99%, p < 0.05) but faster than for LTCCs in IHCs. The sensitivity for the dihydropyridine (DHP) L-type channel blocker isradipine was 8.5-fold lower than for alpha1C. Radioligand binding experiments revealed that this was not due to a lower affinity for the DHP binding pocket, suggesting that differences in the voltage-dependence of DHP block account for decreased sensitivity of D-LTCCs. Our experiments show that alpha1D(8A) subunits can form slowly inactivating LTCCs activating at more negative voltages than alpha1C. These properties should allow D-LTCCs to control physiological processes, such as diastolic depolarization in sinoatrial node cells, neurotransmitter release in IHCs and neuronal excitability.

Using D-LTCC-deficient mice, we have previously demonstrated that inward currents through ␣1D form LTCCs with biophysical and pharmacological properties distinct from C-LTCCs (4,6). These include a more negative range of current activation and slower current inactivation during depolarizations, allowing these channels to mediate long lasting Ca 2ϩ influx during weak depolarizations. Such properties allow LTCCs to control tonic neurotransmitter release in hair cells (5,10), diastolic depolarization in the sinoatrial node (11), and electrical excitability of neurons (12)(13)(14)(15).
In addition to these different biophysical properties, it has also been postulated that D-LTCCs may exhibit a lower sensitivity to DHP channel blockers (4,16). However, this has never been proven or quantified in a comparative study of the DHP sensitivity of D-and C-LTCCs. At present it is also unclear to which extent these biophysical and pharmacological differences are determined by the ␣1D subunit itself, by known accessory Ca 2ϩ channel subunits, or by other Ca 2ϩ channelassociated proteins.
By analyzing the biophysical and pharmacological properties of cloned human ␣1D splice variants in tsA-201 cells, we provide evidence that most of the biophysical differences described above are determined by ␣1D. We also demonstrate that alternative splicing of exon 8 is critically affecting the expression of functional ␣1D protein in tsA-201 cells. Our analysis also revealed that, in functional experiments, D-LTCCs display lower sensitivity for DHPs than ␣1C, despite similar affinity for the DHP binding pocket.

Membrane Preparation and (ϩ)-[ 3 H]Isradipine
Binding-Membranes from tsA-201 cells transfected with 4.5 g of ␣1, 3.5 g of ␣2-␦, 2.5 g of ␤1a or ␤3 subunit, and 4.5 g of pUC18 carrier DNA in 10-cm culture dishes were prepared as described (24). Binding experiments with (ϩ)-[ 3 H]isradipine were performed in a final assay volumes of 0.5 or 1 ml (24). Experimental details are given in the legend to Fig. 5. Kinetic experiments were performed in the absence of added Ca 2ϩ . Nonspecific binding was determined in the presence of 1 M unlabeled isradipine. Serial dilutions of drugs were made in Me 2 SO (final Me 2 SO concentration Ͻ 1% (v/v)).
Electrophysiological Recordings-Whole cell patch clamp experiments were carried out at room temperature using an Axopatch 200B amplifier (Axon Instruments, Foster City, CA). Currents were recorded at sampling rates of 5 or 25 kHz, low pass-filtered at 2 or 5 kHz, and recorded directly onto a personal computer equipped with pCLAMP version 7.0. Borosilicate glass pipettes had a resistance of 2-4 megohms when filled with internal solution. Capacitance and series resistance compensations of 60 -80% were used. The voltage error due to uncompensated series resistance was 1.2 Ϯ 0.6 mV. The average membrane time constant was 83 Ϯ 6 s. Unless stated otherwise, Ba 2ϩ currents (I Ba ) through LTCCs were measured with the following solutions: Pipette solution (in mM): 135 mM CsCl, 10 mM Cs-EGTA, and 1 mM MgCl 2 , adjusted to pH 7.4 with CsOH; bath solution (in mM): 15-20 mM BaCl 2 , 10 HEPES, 150 mM choline Cl, and 1 mM MgCl 2 , adjusted to pH 7.4 with CsOH. Run-down of I Ba was typically less than 10% in 3 min. All voltages were corrected for a liquid junction potential of Ϫ10 mV. Leak and capacitative currents were measured using hyperpolarizing pulses. Raw currents were corrected for linear leak currents. The voltagedependence of activation was determined from IV curves obtained by step depolarizations from a holding potential of Ϫ90 mV to various test potentials. IV curves were fitted according to Equation 1.
V rev is the extrapolated reversal potential of I Ba , V is the membrane potential, I is the peak current, G max is the maximum conductance of the cell, V 0.5,act is the voltage for half-maximal activation, and k act is the slope factor of the Boltzmann term. The time course of current activation was fitted using an exponential function.
I͑t͒ ϭ A 0 ϫ exp(Ϫt/ 0 ) ϩ C (Eq. 2) is the current at time t after the depolarization, A 0 is the steady state current amplitude with the respective time constant of activation, 0 , and C represents the remaining steady state current. The rate of inactivation was assessed by both the percentage of current that had inactivated after 2 s, and by fitting the current traces to the biexponen-tial function, yielding time constants for a fast ( fast ) and a slow ( slow ) component.
Voltage dependence of inactivation under quasi-steady state conditions was measured using a multi step protocol to account for run-down. A control test pulse (60 ms to the voltage of peak current, V max ) was followed by a 2-s step to Ϫ90 mV, followed by a 5-s conditioning step, and a subsequent test pulse to V max . The start-to-start interval was 20 s at a holding potential of Ϫ90 mV. Inactivation during the 5-s conditioning pulse was calculated as follows.
% inactivation ϭ ͑1 Ϫ ͑I Ba,test /I Ba,control )) ϫ 100 (Eq. 4) I Ba,control and I Ba,test are the current amplitudes at V max before and after the 5-s conditioning pulse, respectively. Inactivation curves were fitted to a Boltzmann equation as described (25). Effects of DHPs were monitored continuously using 0.1-Hz depolarizing pulses to V max . DHPs were dissolved in the external recording solution from a 10 mM stock solution in dimethyl sulfoxide and perfused through a microcapillary onto cells using a gravity-driven perfusion system. Only cells exhibiting stable currents (run down Ͻ15% during the first 60 s) were used for analysis of DHP effects. The DHPs isradipine and BayK8644 were employed as their racemic mixtures.
Transfected cells were visualized as cotransfected GFP fluorescence Ca 2ϩ currents in IHCs were measured as previously described in cochleae isolated from 2-4-day-old animals (4).
Immunoblotting and preparation of affinity-purified sequence-directed antibodies was carried out as described (4, 26) using antibodies described in the legend to Fig. 1.
Statistics-All data are presented as mean Ϯ S.E. for the indicated number of experiments. Statistical significance was determined by unpaired Student's t test. Data were analyzed using Clampfit (Axon Instruments) and Origin ® 5.0 (Microcal).

RESULTS
We cloned two different cDNAs encoding full-length ␣1D subunits ( Fig. 1A) from human pancreatic tissue, containing alternatively spliced exons 8A or 8B (␣1D 8A , ␣1D 8B ), which results in a six-amino acid difference in the pore region of repeat I. Their sequence is identical to a previously cloned cDNA (9) but both splice variants lack exons 32 and 44. Heterologous expression in tsA-201 cells together with auxiliary subunits yielded DHP-sensitive L-type currents for ␣1D 8A (100 of 154 patched GFP-expressing cells gave measurable I Ba ) but not for ␣1D 8B . Similarly, no current was measured under our experimental conditions after transfection with rat ␣1D cDNA (18), which also contains exon 8B. To determine if the absence of Ca 2ϩ current resulted in the expression of a non-functional subunit or was due to the absence of ␣1D protein, ␣1D subunit expression was quantified by immunoblot analysis of transfected tsA-201 cell membranes (Fig. 1B). A full-length form of ␣1D protein was only detected for ␣1D 8A but not for rat and human ␣1D 8B . After transfection of cells with ␣1D 8B , ␣1D immunoreactivity was only associated with polypeptides smaller than the expected full-length form (Fig. 1), suggesting that ␣1D 8B protein underlies proteolytic degradation in tsA-201 cells. As a cloning artifact was ruled out by DNA sequencing (see "Experimental Procedures"), our data suggest that the six-amino acid residue difference prevents effective ␣1D Ca 2ϩ subunit expression in tsA-201 cells.
Next we tested if ␣1D 8A could give rise to L-type currents with the characteristics described recently in chick and mouse cochlea (4,5,16). These include low activation threshold, fast activation kinetics, slow inactivation, and lower apparent DHP antagonist sensitivity. After transfection using Ca 2ϩ phosphate precipitation, the ␣1D current density was similar to ␣1C-a mediated currents (Table I). With 15 mM Ba 2ϩ as the charge carrier, the threshold of activation for ␣1D 8A was found at about 15 mV more hyperpolarized potentials (Ϫ45.7 Ϯ 0.5 mV, n ϭ 38) as compared with ␣1C-a (Ϫ31.5 Ϯ 0.5 mV, n ϭ 16) ( Table I, Fig. 2A). The maximum of the current-voltage rela-tionship (V max ) as well as the potential of half-maximal I Ba activation (V 0.5,act ) was also shifted significantly to hyperpolarized potentials without affecting the slope of the activation curve (Table I). Depolarizations to V max revealed that I Ba through ␣1D 8A showed a monoexponential activation time course about 3-fold faster than for ␣1C-mediated currents (Table I, Fig. 2B). As for ␣1C-a, the speed of ␣1D 8A activation increased at more positive voltages. More rapid activation of ␣1D 8A was consistently found over a voltage range from Ϫ30 to ϩ30 mV (Fig. 2C).
To test if ␣1D 8A also mediates the slowly inactivating L-type currents observed in IHCs, we compared current inactivation during long test pulses with the inactivation time course of (largely ␣1D-mediated; Ref. 4) L-type currents in IHCs and heterologously expressed ␣1C. During a 5-s depolarizing test pulse from a holding potential of Ϫ90 mV to V max , 90 Ϯ 2.2% (n ϭ 6) of ␣1D 8A current inactivated (Fig. 3A). In contrast, only 9.3 Ϯ 4.6% (n ϭ 6) of I Ba inactivated during a 5-s depolarizing pulse in IHCs. Although inactivation of ␣1D 8A current was faster than in IHCs, it was significantly slower than for ␣1C-a (99 Ϯ 1.3%, n ϭ 4, p Ͻ 0.05; Fig. 3, A and B). Fig. 3A shows normalized currents of ␣1C-a and ␣1D 8A , which both exhibit biexponential inactivation. Slower inactivation of ␣1D 8A was due to a 1.5-fold increase of the time constant for the slowly inactivating component (Fig. 3B)  ␤2a subunits stabilize slow inactivation of Ca 2ϩ channel ␣1 subunits (21,27). We therefore investigated if the slower inactivation time course of ␣1D currents of IHCs can be obtained by coexpression with rat or bovine ␤2a subunits (together with ␣2␦ subunits). As shown in Fig. 3C, current inactivation during 2-s depolarizing test pulses was slower upon coexpression of rat or bovine ␤2a than upon coexpression with ␤3 but could not account for the slow inactivation of ␣1D in IHCs.

TABLE I Biophysical properties of D-and C-LTCCs expressed in tsA-201 cells
The half-maximal voltage for activation (V 0.5,act ), the slope for activation (k act ), the maximum of the current-voltage relationship (V max ), and the slope for inactivation (k inact ) were obtained by fitting the data as described under "Experimental Procedures." The activation threshold is defined in the legend to Fig. 2. The time course of current activation was fitted using a single exponential function. act is given for I Ba elicited by test pulses to V max . Data are given as mean Ϯ S.E. *, statistically significant difference (p Ͻ 0.05). antagonist effects on ␣1D and ␣1C-a currents under identical experimental conditions. We therefore measured the inhibition of both channel types by the DHP antagonist isradipine at various concentrations and at different holding potentials. At Ϫ90 mV holding potential, 300 nM isradipine completely inhibited ␣1C-a current (98.3 Ϯ 1.7%, n ϭ 3) but only 30 -40% of ␣1D 8A (Fig. 4). The concentration dependence of block revealed that isradipine sensitivity for ␣1D 8A was 8.5-fold lower (Fig.  4B) at this holding potential. Isradipine sensitivity increased by an order of magnitude at a more positive holding potential (Ϫ50 mV, Fig. 4B), demonstrating a voltage-dependent mechanism of DHP block as described previously for ␣1C-a currents (28,29).
␣1D 8A Ca 2ϩ channels coexpressed with ␤3 and ␣2␦ in tsA-201 cells also exhibited the typical current modulation by the Ca 2ϩ channel activator BayK8644 (Fig. 4, C and D) similar to its actions in IHCs (4). BayK8644 increased the maximal I Ba , produced a slight hyperpolarizing shift in the IV relationship and slowed current deactivation of the tail current induced by repolarization of the cell to the holding potential (Fig. 4D). The extent of maximal I Ba stimulation was more pronounced in tsA-201 cells as compared with IHCs (7.5 Ϯ 1.8 -fold, n ϭ 6), versus 2.9-fold; Ref. 4).

DISCUSSION
Here we present the first detailed patch clamp analysis of the biophysical properties of whole cell currents through Ca 2ϩ channels formed by ␣1D subunits after heterologous expression in mammalian cells. Our studies revealed that these subunits can form L-type Ca 2ϩ channels with properties similar to D-LTCCs previously described in IHCs (4). We clearly demonstrate lower DHP antagonist sensitivity, more rapid activation kinetics, and slower inactivation for the ␣1D 8A splice variant as compared with ␣1C-a currents. Another novel finding was that the lower DHP antagonist sensitivity is not due to lower binding affinity for the DHP binding pocket. Instead, it is only  Table I. A, normalized IV curves for ␣1D 8A and ␣1C-a. The activation threshold, determined as the test potential at which 5% of maximal I Ba was activated, was Ϫ45.7 Ϯ 0.5 mV (filled circles) and Ϫ31.5 Ϯ 0.5 mV (open circles) for ␣1D 8A and ␣1C, respectively. B, the kinetics of current activation was determined by depolarizing pulses to V max . Representative traces for ␣1D 8A and ␣1C-a show monoexponential activation time courses with time constants ( act ) of 0.41 and 1.45 ms, respectively. C, voltage dependence of act for ␣1D 8A (filled circles) and ␣1C-a (open circles). act was determined for currents elicited by test pulses to the indicated potentials. *, statistically significant difference (p Ͻ 0.05).

FIG. 3. Inactivation properties of the L-type Ca 2؉ channel ␣1D
8A . An I Ba was elicited by 5 s depolarizing pulses from a holding potential of Ϫ90 mV to V max . Representative traces of ␣1D 8A and ␣1C-a (coexpressed with ␤3 and ␣2␦) were fit to a biexponential decay yielding the following time constants for the fast ( fast ) and the slow ( slow ) component (in seconds); the contribution of the fast and slow components to total I Ba are given in parentheses: ␣1D 8A , 0.18 (40%) and 1.7 (48%); ␣1C, 0.16 (69%) and 1.05 (30%). *, statistically significant difference (p Ͻ 0.05). B, inactivation time constants for the fast ( fast ) and the slow ( slow ) component (in seconds). Data are means Ϯ S.E. for n ϭ 4 -6. *, statistically significant difference (p Ͻ 0.05). C, effects of ␤-subunit expression on the inactivation properties of GFP-␣1D 8A . Current inactivation was measured during 2-s depolarizing pulses to V max for rat ␤3, rat ␤2a, and bovine ␤2a: 82.8 Ϯ 6.5% (n ϭ 4); 34.5 Ϯ 3.2%, n ϭ 8; 38.1 Ϯ 10.2%, n ϭ 4). Slowing of inactivation was significant (p Ͻ 0.05; asterisk) only for rat ␤2a. D, voltage dependence of Ca 2ϩ channel inactivation during 5-s depolarizing pulses. Inactivation was determined as described under "Experimental Procedures." Half-maximal inactivation potential (V 0.5,inact ) and the corresponding slope parameters (k inact ) of ␣1D 8A (filled circles) and ␣1C-a (open circles) are given in Table I. evident in the presence of membrane potential and therefore must involve a voltage-dependent mechanism affecting DHP antagonist sensitivity in functional studies. We were also able to demonstrate that the heterologous expression of ␣1D in mammalian cells is critically affected by six amino acids in the pore region of repeat I.
Our previous experiments with mice lacking class D Ca 2ϩ channels clearly revealed that ␣1D subunits are required to form low voltage-activated LTCCs with the above described properties in mouse cochlear IHCs. However, from these experiments it remained unclear if all of these properties are inherent to ␣1D or require the presence of a yet unidentified Ca 2ϩ channel subunit.
Heterologous expression of ␣1D 8A together with ␣2␦ and ␤3 (or ␤2a) resulted in I Ba with an activation threshold below Ϫ40 possible that alternative splicing in this region also affects the activation threshold of ␣1D. In contrast to our constructs, a previously cloned human ␣1D (9) contains exon 32 (and exon 44). Unfortunately, its biophysical and pharmacological properties have not been directly compared with ␣1C under identical experimental conditions. Stable expression of this cDNA in HEK-293 cells (together with ␤3 and ␣2␦, 20 mM Ba 2ϩ as charge carrier) yielded channel currents with V 0.5,act of ϩ1.8 mV and a V max of about ϩ20 mV (33). However, correction for junction potential would shift these values to more negative potentials by about 23 mV, resulting in activation parameters nearly identical to our ␣1D 8A subunit. Minor effects of exon 32 on channel gating cannot be ruled out and will require a more detailed analysis.
The only property not mimicked by our splice variant was the very slow inactivation of IHC currents during a depolarizing pulse. ␣1D 8A currents inactivate slower than ␣1C-a but faster than I Ba in IHCs. As compared with ␣1C-a, the slower inactivation of ␣1D 8A can be explained by an increase of the contribution and the time constant of a slow component of I Ba inactivation. It remains to be determined if alternative splicing or the presence of another auxiliary subunit is responsible for the slow inactivation of class D currents in IHCs.
The relatively weak channel block of I Ba through class D LTCCs by DHP antagonists was now directly confirmed in our experiments. We describe for the first time that DHP inhibition of ␣1D is voltage-dependent and favored at more positive voltages. As for ␣1C-a, this indicates higher affinity for inactivated channels (28,29) or induction of inactivated channel states (25). ␣1D 8A showed an approximately 10-fold lower sensitivity for block by the DHP antagonist isradipine than ␣1C-a. This finding nicely explains the absence of major side effects expected from block of class D channels in humans at therapeutic plasma concentrations. This includes the lack of bradycardic actions expected from the block of class D channels in sinoatrial node cells as well as the absence of hearing disturbances resulting from the block of this channel type in cochlear IHCs (4). Instead, DHPs preferentially block class C LTCCs in the cardiovascular system (34). The higher DHP sensitivity of the smooth muscle ␣1C splice variant (␣1C-b) as compared to the cardiac splice variant (␣1C-a; Ref. 28) in part explains the selectivity of DHPs for arterial smooth muscle, resulting in therapeutic antihypertensive effects in the absence of cardiodepression (34). The higher sensitivity of ␣1C-b is also not due to a higher affinity of DHP antagonists for its DHP binding pocket (28). Instead, it could be explained by differences in the voltage dependence of ␣1C block and is due to amino acid divergence between the two splice variants in transmembrane segment IS6. We propose that a similar mechanism must explain the lower DHP antagonist sensitivity of ␣1D 8A as compared with ␣1C-a because the affinity for the DHP binding pocket is the same in these subunits.
Although we and others (9) found evidence for alternative splicing in segment IS6 (exons 8A and 8B, respectively), the role of exon 8B for ␣1D function and DHP modulation could not be assessed. This was due to the absence of intact ␣1D protein and currents after expression of our human ␣1D construct in which exon 8A sequence was exchanged for exon 8B (␣1D 8B ). The inhibitory role of exon 8B on ␣1D expression is further supported by our inability to transiently express full-length rat ␣1D, which also contains exon 8B, despite the abundant presence of mRNA in the transfected cells. 2 In accordance with our interpretation, the functional expression of exon 8B containing ␣1D cDNAs isolated from rat brain (35) and chicken cochlea (36) has not yet been reported. Further studies must determine which of the 6 amino acids differing between these alternative exons account for this effect. It is possible that alternative splicing in IS6 serves as a molecular switch to modulate ␣1D subunit expression on the post-transcriptional level.
Our experiments provide convincing evidence that ␣1D 8Amediated L-type currents activate at slightly more positive voltages than recombinant T-type channels (37) but at more negative voltages than C-LTCCs as reported here. In this respect they resemble ␣1E-mediated currents (37), which were originally classified as low voltage-activated (38). Although we cannot exclude that alternative splicing or biochemical modulation (e.g. phosphorylation) also allows other L-type ␣1 subunit isoforms (including ␣1C) to activate at such "intermediate" voltages, our data provide direct evidence that D-LTCCs can account for the "low voltage-activated" L-type currents described previously in neurons (15). We have developed mutant mice in which the DHP sensitivity of ␣1C-subunits is dramatically reduced (39). Such animal models will allow direct determination of the role of D-LTCCs for L-type current components in various tissues.
Taken together, the evidence demonstrates that expression of ␣1D subunits should enable cells to slowly inactivating voltage-gated Ca 2ϩ influx in response to rather weak depolarizations. This property allows them to participate in important physiological functions such as tonic neurotransmitter release in IHCs (10) and control of diastolic depolarization in sinoatrial node (4). These properties also make them ideally suited to contribute to subthreshold Ca 2ϩ signaling (e.g. in hippocampal pyramidal cells; Ref. 14) and to the plateau potential underlying bistable membrane behavior (e.g. in motoneurons; Refs. 12 and 13).