Molecular Localization of Ion Selectivity Sites within the Pore of a Human L-type Cardiac Calcium Channel*

A highly conserved position of negatively charged amino acids is present in the 552 segments of the 55-56 linker regions among calcium channels. We report here that replacing Glu residues at this position alters the ion selectivity of the human cardiac calcium channel. in motif I or in motif TI1 with Lys produced mutant calcium channels that per-meated sodium ions 10-fold more effectively than bar- ium ions. More conservative changes such as substitution of G1u10B6 with Gln or substitution of Glu13s7 with Ala also increased sodium permeation through the mutant calcium channels. Sodium currents through the mutant calcium channels could be modulated by dihydropyridines and blocked by external divalent cations. These results suggest that G1u334, G1u10B6, and Glu1387 are part of a ring of glutamate residues formed in the pore-lining SS1-SS2 region and are critical in determining ion selectivity and permeability of a human cardiac calcium channel.

A highly conserved position of negatively charged amino acids is present in the 552 segments of the 55-56 linker regions among calcium channels. W e report here that replacing Glu residues at this position alters the ion selectivity of the human cardiac calcium channel.

Substituting G~u~~~ in motif I or Glu10B6 in motif TI1
with Lys produced mutant calcium channels that permeated sodium ions 10-fold more effectively than barium ions. More conservative changes such as substitution of G1u10B6 with Gln or substitution of Glu13s7 with Ala also increased sodium permeation through the mutant calcium channels. Sodium currents through the mutant calcium channels could be modulated by dihydropyridines and blocked by external divalent cations. These results suggest that G1u334, G1u10B6, and Glu1387 are part of a ring of glutamate residues formed in the pore-lining SS1-SS2 region and are critical in determining ion selectivity and permeability of a human cardiac calcium channel. ~~~ ~ ~ ~ Calcium channels are members of the structurally homologous superfamily of voltage-gated ion channels. In Na+ and Kt channels, the SS1-SS2 region of the S5S6 linker region has been postulated to line the channel pore (1, 2) and therefore play a major role in ion selectivity. Single amino acid substitutions in this region altered ion selectivity (3-61, conductance (7, 81, as well as toxin and Cd2+ sensitivity (9-11). Analysis of the amino acid sequences of the SS2 and flanking regions of the four repeating domains (motifs) of Ca2+ channels (12)(13)(14)(15)(16) reveals a highly conserved position of negatively charged Glu residues (Fig. 1 idues occupy equivalent positions in motifs I and 11, but a positively charged residue, Lys, occupies the equivalent position in motif I11 and a neutral residue, Ala, occupies the equivalent position in motif IV (Fig. 1). These 2 residues, Lys1422 and Ala'714, proved to be critical in determining the ion selectivity of the Na+ channel (4). In Ca2+ channels these positions are occupied by Glu in all four motifs. We investigated whether the homologous region in the human cardiac Ca2+ channel lines the channel pore and is critical in determining ion selectivity. Our hypothesis is that these 4 Glu residues may form a Ca2+ binding site or ion selectivity filter in the pore of the Ca2+ channel.
We examined the permeation of Ba2+ and Na+ ions through the wild-type normal human cardiac Ca2+ channel and through mutant channels in which Glu was substituted by Lys (E334K in motif I, E1086K in motif 111) or by Ala (E1387A in motif IV).
In addition, a double mutant (E1086K,E1387A) was examined in which 2 Glu residues were substituted to mimic the Na+ channel at the equivalent positions. We also altered the charge of a residue within the SS2 region (D1391K), which may be positioned near the extracellular mouth of the pore.

EXPERIMENTAL PROCEDURES
Construction of Mutant Cuz' Channels-The full-length cDNA of the the plasmid pBluescript SK(-). Mutants E334K and E1086K were con-human heart Ca2+ channel a 1 subunit (hHT)' (16) was engineered in structed by first subcloning MunI(1732)/SphI(2718) or AflII(3923Y SphI(5598) fragments of hHT into M13mp18 or M13bm20, respectively, for production of single-stranded templates. The desired mutations were then made according to the method of Kunkel et al. (17). Cassettes carrying the desired mutations were then ligated into hHT in pBluescript SK(-). Other mutants in motif I11 and motif IV were synthesized within BstBI(4664)/BclI(5720) or BstBI(4664)/SphI(5598) cassettes, respectively, using the polymerase chain reaction (PCR; Hoffman-LaRoche) method (18). Oligonucleotides encoding BstBI or SphI recognition sites and carrying the designed base mismatches served as forward primers, and oligonucleotides covering BclI or BstBI sites presence of the desired mutation and then ligated into hHT to replace served as reverse primers. PCR products were sequenced to verify the the corresponding BstBI(4664)/BclI(5720) or BstBI(4664)/SphI(5598) segment. Double mutants E1086K,E1387A and E1086K,D1391K were constructed by replacing the AatII(5123)/Bcl1(5720) fragment from the mutant E1086K with the corresponding fragment from E1387A or D1391K. After mutagenesis, cassettes were sequenced to verify the presence of the desired mutation and to check for undesired random mutations. cRNAs specific for the wild-type Ca2+ channel al subunit, mutant Ca2+ channel a l subunits, skeletal muscle Ca2+ channel a2 subunit (19), and Ca2+ channel p subunit from human heart, designated pa (201, were synthesized by in vitro run-off transcription. Electrophysiology-Oocytes were injected with a total of 40 nl of cRNAfrom wild-type human heart a,-subunit or mutant a,-subunit (0.1 pg/pl) together with skeletal muscle a2 subunit (0.1 pg/pl) and heart a-subunit (0.1 pg/pl) and incubated in modified Barth's solution for 3-7 days. Whole cell currents were recorded with a two-microelectrode voltage-clamp amplifier (Axoclamp 2A). Voltage and current electrodes ( (16); rSK, rabbit skeletal muscle Cap+ channel (12); hB-D, human neuroendocrine Ca"+ channel (14); rB-BI, rabbit brain BI-2 Ca2+ channel (13); rB-N, rat brain N-type Ca2+ channel (15); pZ, rat skeletal muscle Na+ channel (26). The numbers of the amino acid residues are indicated on the right-hand side of each sequence. The proposed SS1 and SS2 segments are shadowed. Positionally conserved negatively charged amino acids of the Ca2+ channels and the corresponding amino acids of the Na+ channel are boxed. The numbers of the investigated Glu residues of the human cardiac Ca2 ' channel are indicated in each motif above the boxed Glu residues.
points. In uninjected oocytes, small endogenous outward currents are observed with depolarization to positive potentials in the Ba2 '-containing external solution. We have never observed inward currents in response to depolarizing pulses in the Na+-containing external solution with uninjected oocytes or oocytes coinjected with an and p subunit messages.

RESULTS AND DISCUSSION
We investigated the permeation of Ba2+ and Na+ ions through the wild-type and mutant Ca2+ channels. Fig. 2 shows a series of whole cell currents recorded from Xenopus oocytes injected with cRNA specific for the wild-type human cardiac Ca2+ channel (Fig. 2 A ) , the mutant E1086K (Fig. 2B), the mutant D1391K (Fig. 2C), and the mutant E1387A (Fig. 20). Expression of the wild-type cardiac Ca2+ channel resulted in inward voltage-activated currents with Ba2+ as the charge carrier (Fig. 2 A , parts a and c). The peak current-voltage relationship (Fig. 2 A , part c ) showed that the inward Ba2 current reached a peak value at +20 mV and that the extrapolated reversal potential was +68 mV (Table I). Replacement of external Ba2+ with Na+ resulted in a marked decrease in inward current and the appearance of an outward current at positive potentials ( Fig. 2A,parts b and e). The outward currents in the absence of external Ba2+ are most likely due to the unmasking of endogenous oocyte conductances for K I+.
The smaller inward current upon substitution of external Ba2+ with Na+ is in contrast to results obtained in single cardiac myocytes in which removal of all external divalent cations allows Na.+ ions to readily pass through Ca2+ channels (21,22). Addition of divalent cations, such as Ca2+ or Mg2+, in the micromolar range blocks monovalent permeation through the Ca2+ channel (21, 23). In the absence of external Mg2 , Na+ current through the wild-type Ca"' channel reached a magnitude of 1 pA after 3 days of incubation; however, oocyte membranes became unstable. Therefore all experiments utilizing Na+-containing external solutions also contained 1 mM Mg2+ to prevent rapid deterioration of.the oocytes. This concentration of Mg2 + partially blocks (21, 23) Na+ permeation through the Ca2+ channel, which accounts for the small inward current shown in Fig. 2A ( b and c). The ratio of peak inward current in Na+-containing versus Ba2+ -containing external solutions was 0.056 (2 0.01):l ( n = 5, Table I) for the expressed wild-type cardiac Ca2+ channels.
The mutant E1086K exhibited a small inward current in the presence of external Ba2+ (Fig. 2B), which reached a peak at 0 mV and then quickly reversed to become outward at more positive potentials. Replacing external Ba2~' with Na+ resulted in a dramatic increase in inward current, which reached a peak value at -10 mV and then quickly reversed. Similar results were seen for the mutant E334K (Table I)  Significantly different from wild-type human cardiac Ca2+ channel (hHT); p < 0.05.  1 FM ( B ) , and Cd2+ at 100 PM ( C ) on whole cell currents, in Na+-containing external solution, in oocytes expressing the mutant E1086KD1391K. All oocytes were co-injected with skeletal muscle a2 and heart p Ca2+ channel subunit cRNAs. The holding potential was -60 mV, and currents were elicited by depolarizing pulses from -50 to +30 mV, in 10-mV increments. Currents were recorded in Na+-containing external solution (A and B ) or Na+-containing solution that was EGTA-free ( C ) . Whole cell currents are shown before oocyte and due to partial block by external Mg2+, these ratios cannot be used as a quantitative estimate of ion permeability of the channel. Nevertheless, the data provide a qualitative index of the degree of change in selectivity of the channel between divalent and monovalent cations. In addition, mutants E334K and E1086K exhibited a shift in reversal potential to more negative potentials in the presence of Ba2+-containing external solutions (Table I), consistent with a n increase in permeation of monovalent cations through the mutant Ca2+ channels resulting in an increased outward flux of intracellular Kt. Therefore these mutations most likely induce a general loss of selectivity of the channel for divalent over monovalent cations.
The ratios of peak inward current, peak current potentials, and apparent reversal potentials in Na+-containing and Ba2+containing external solutions for all mutants tested are summarized in Table I. Briefly, more conservative changes such as substitution of G1u1OS6 with Gln (mutant E1086Q) or substitution of Glu13s7 with Ala (mutant E1387A) also increased the Na+ current relative to the Ba2+ current. Mutant E1086Q exhibited a peak inward current ratio of 1.15 (+ 0.ll):l ( n = 81, in the Na+-containing versus the Ba2+-containing external solution, which was less than that observed for the more drastic substitution of Lys for Glu at the same position. Also, the reversal potential for Ba2+ current was shifted to negative potentials to a lesser degree than that for mutant E1086K. Mutant E1387A exhibited only a slight increase in permeation of Na+ ions through the channel (Fig.  2 0 ) . Alteration of the charge of a residue within the SS2 region that may be positioned near the extracellular mouth of the pore, Asp1391 in motif IV (mutant D1391K), caused no change in Na' permeability or in the reversal potential as compared to the wild-type Ca2+ channel ( Fig. 2C and Table I). This suggests that this amino acid residue, Asp1391, may not contribute to the porelining region of the channel. When the mutations E1086K and E1387A were combined, the resultant double mutant (E1086K,E1387A), which mimics the Na' channel at the equivalent positions, exhibited a permeability to Na+ that was similar to that of mutant E1086K, but significantly greater than that of mutant E1387A ( Table I).
The double mutant E1086K,D1391K exhibited properties similar to the single mutant E1086K (Table I).
Modulation by dihydropyridines (DHP) is characteristic of L-type cardiac Ca2+ channels. Therefore we tested the effects of the DHP antagonist PN 200-110 and the DHP agonist BayK 8644 on the double mutant E1086K,D1391K (Fig. 3) as well as on all of the single point mutants (data not shown). Both inward and outward currents, in the Na+-containing extracellular solution, were reduced in the presence of 2 PM PN 200-110 (Fig. 3A). The peak inward current was reduced by 86.7 ? 2.1% ( n = 5). The agonist BayK 8644 (1 J~M ) increased peak inward current (Fig. 3B), in the Na+-containing solution, by 85.8 + 0.538 ( n = 4). DHP sensitivity provides strong evidence that the inward current in the presence of the Na+-containing extracellular solution, produced by expression of the double mu-tant E1086K,D1391K, as well as by the single mutants, is the result of Na' permeation through a mutant CaZ+ channel. In addition, the results suggest that the double mutation and the single mutations described induced no major conformational changes in channel structure outside of the pore region since the channel retained proper responses to dihydropyridines.
Cadmium is a potent blocker of native Ca2+ channels (24). The application of external Cd2+ (100 VM) blocked inward current in the double mutant E1086K,D1391K (Fig. 31, as well as in the single mutants (data not shown), in the Na+-containing extracellular solution (Fig. 3C). Cadmium at a concentration of 10 p~ produced a 37.5 ? 3.5% ( n = 3) decrease in inward current, whereas 100 p~ Cd2+ produced an 83.0 f 0.2% ( n = 3) block of the inward current. At similar concentrations of Cd2+, Na+ permeation is completely blocked in native cardiac Ca2+ channels (24).
The present study provides evidence that the SS1-SS2 regions of voltage-gated ion channels form at least part of the pore-lining region. In voltage-dependent Ca2+ channels, ion permeation occurs through multiple-ion occupancy states (22,25). The ability of the Ca2+ channels to discriminate between divalent and monovalent cations is determined by ion binding affinity and by ion-ion electrostatic interactions in the multiple occupancy state (22,25). In this study, substitution of positively charged or neutral residues for the negatively charged Glu residue, a t equivalent positions in motifs I, 111, and IV, modified the permeation properties such that the channel discriminated poorly between divalent and monovalent cations. Similar mutants in motif I1 are being tested, and preliminary results are consistent with the importance of the Glu residues. These results, taken together, strongly suggest that G~u~~~, G1u1086, and Glu1387 are part of a "glutamate ring" formed in the putative pore-lining SS1-SS2 region and are critical in determining Ca2+ ion binding and ion selectivity of the cardiac Ca2+ channel.