Aspartate Substitutions Establish the Concerted Action of P-region Glutamates in Repeats I and III in Forming the Protonation Site of L-type Ca2+ Channels*

Hydrogen ions reduce ion flux through voltage-gated Ca2+ channels by binding to a single protonation site with an unusually high pK a . Recent evidence localizes the protonation site to the same locus that supports high affinity Ca2+ binding and selectivity, a set of four conserved glutamate residues near the external mouth of the pore. Remaining controversy concerns the question of whether the protonation site arises from a single glutamate, Glu-1086 (EIII), or a combination of Glu-1086 and Glu-334 (EI) working in concert. We tested these hypotheses with individual Glu → Asp substitutions. The Glu → Asp replacements in repeats Iand III stood out in two ways. First, in both EID and EIIID, protonation was destabilized relative to wild type, whereas it was unchanged in EIID and stabilized in EIVD. The changes in affinity were entirely due to alterations in H+ off-rate. Second, the ratio of protonated conductance to deprotonated conductance was significantly closer to unity for EID and EIIID than for wild-type channels or other Asp mutants. Both results support the idea that EI and EIII act together to stabilize a single titratable H+ ion and behave nearly symmetrically in influencing pore conductance. Neutralization of EIII by alanine replacement clearly failed to abolish susceptibility to protonation, indicating that no single glutamate was absolutely required. Taken together, all the evidence supports a model in which multiple carboxylates work in concert to form a single high affinity protonation site.

The pore of voltage-gated Ca 2ϩ channels is capable of high affinity interactions with either calcium or hydrogen ions (1)(2)(3). These interactions are functionally important for the pore's exquisite Ca 2ϩ selectivity (4) and its sensitivity to blockade by acidification of the external medium in physiological or pathological settings (5)(6)(7). The kinetics of proton block and unblock of single channels can be resolved in recordings of individual blocking events with monovalent ions as charge carriers (2,3,8,9). Under such experimental conditions, half-maximal block by H ϩ occurs at pHϳ 8.5, three to four pH units greater than the pK a of an individual glutamate carboxylate (3,8,10). All evidence supports the idea that block by protons occurs at the same locus where selectivity for Ca 2ϩ takes place, where carboxylic acid side chains of a set of conserved glutamate residues come into close proximity (8, 10 -14). However, there is sharp disagreement about the nature of the titratable group (8,10). Klöckner et al. (10) proposed that the protonation site consisted of only a single glutamate, Glu-1086, in repeat III (here designated simply as EIII). This conclusion was based on a series of glutamate to alanine replacements in which only EIIIA appeared to abolish the ability of protons to cause flickering block of unitary flux. In contrast, Chen et al. (8) suggested that the protonation site arises from multiple glutamate side chains, acting in concert (Fig. 1A). In one particular version of this hypothesis, carboxylate groups from EI and EIII were proposed to share the titratable H ϩ , in a stable hydrogen-bonding configuration (Fig. 1B). Although more complex than that of Klöckner et al. (10), this model readily accounts for the existence of only a single binding site for H ϩ , not one for each Glu or Glu pair. It also provides an explanation for the finding of halfblock at pH Ͼ8, in line with the well documented properties of carboxylic acid-carboxylate complexes (15,16), not at pH Ͻ5, characteristic for a single carboxylate.
Aspartate substitutions for glutamate offered a useful approach for testing these conflicting models (e.g. Refs. 17 and 18). Unlike Glu 3 Gln replacements, which alter side chain charge, or Glu 3 Ala mutations, which change both charge and side chain bulk, Glu 3 Asp substitutions preserve the acidity of the relevant groups while merely decreasing side chain length. Thus, aspartate substitutions would be expected to perturb the protonation of the channel in relatively subtle yet informative ways. Indeed, we found that each of the four aspartate mutants displayed a unique phenotype distinct from the wild-type channel and that their behavior in all cases conformed closely to the predictions of the model in Fig. 1. Therefore, this study provides new evidence for the coordinated interaction of multiple glutamates in forming the protonation site of L-type Ca 2ϩ channels.

EXPERIMENTAL PROCEDURES
Materials-The cDNAs for rabbit ␣ 1C and ␣ 2 -␦ subunits of L-type Ca 2ϩ channels were kindly provided by Dr. T. Tanabe (University of Tokyo), and the cDNA for rabbit ␤ 2b subunit was a gift from Drs. V. Flockerzi and F. Hofmann (Technische Universität, Mü nchen, Germany). FPL 64176 1 was purchased from RBI (Natick, MA); 2 H 2 O was from Aldrich, and other chemicals were from Sigma. * This work was funded by research grants from the National Institutes of Health, the Silvio Conte-NIMH Center for Neuroscience Research at Stanford, and the Mathers Foundation (to R. W. T.), and a postdoctoral fellowship from the American Heart Association (California Affiliate) (to X.-H. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Expression of Ca 2ϩ Channels in Xenopus Oocytes-Wild-type and mutant L-type channels were expressed in the subunit combination ␣ 1C ␤ 2b ␣ 2 in Xenopus oocytes, as described previously (8,12,19). Expression of the L-type Ca 2ϩ channels was confirmed by the appearance of FPL64176-sensitive Ba 2ϩ current in whole oocyte recordings using the two-electrode voltage clamp. The single glutamate to aspartate mutations in the P-region of L-type Ca 2ϩ channels were generated as described previously (12). In all mutant channels, the 5Ј-untranslated region of ␣ 1C cDNA was truncated to increase the expression level (8).
Single Channel Recordings-Unitary currents carried by WT and mutant L-type channels were recorded as described previously (8,20). Briefly, cell-attached patch-clamp recordings were performed on stripped oocytes bathed in 100 mM K ϩ solution (in mM, 100 KCl, 10 HEPES, 10 EGTA, 11 MgCl 2 , pH 7.2 with KOH). All the recordings were performed with high KCl solution (in mM, 150 KCl, 5 EDTA, 5 HEPES) prepared in 2 H 2 O. Under these conditions, the average time in the protonated state was 2-3-fold longer than that measured in K ϩ -H 2 O and Ͼ5-fold longer than that found with Na ϩ -H 2 O (2, 3,8,9,21). The choice of recording conditions proved essential for resolving the rapid kinetics of the protonation events in some mutant channels (e.g. EIIIA), in which the mean dwell time in the protonated state was reduced more than 5-fold compared with WT channels. In all experiments, the L-type Ca 2ϩ channel agonist FPL 64176 (5 M) was included in both pipette and bath solutions to prolong channel openings.
The unitary currents were recorded using an Axopatch 200 amplifier (Axon Instruments, Foster City, CA), filtered at 5 kHz, and sampled at 25 kHz. Membrane potential was held at Ϫ100 mV and stepped to various test voltages ranging from Ϫ40 to Ϫ120 mV. In some experiments, a brief prepulse to a strongly positive voltage was applied prior to the test pulse to facilitate channel opening. Single channel records were leak subtracted off-line. All experiments were conducted at room temperature (21-23°C).
Data Analysis-Single channel traces were analyzed as described previously (8). In brief, long channel openings from multiple sweeps were manually selected and pooled together for generation of the openstate all-points amplitude histograms (bin size 0.2 pA). The resulting peaks on the histogram were fitted with Gaussian functions. For kinetic analysis, current records were idealized using a two-point crossing criterion with the threshold set halfway between h and l conductance levels. Dwell time distributions of deprotonated (h) and protonated (l) states were generated based on idealized records and fit with a single exponential. Corrected time constants for the conductance states were calculated from the mean dwell times using a method to correct for missed events and instrumental dead-time (80 s) (8,22).

RESULTS
We carried out a systematic analysis of the effects of Glu 3 Asp substitutions on each of the four P-region glutamates as a critical test of the hypothesis that the protonation site in L-type Ca 2ϩ channels is formed by multiple P-region glutamates working in concert. This approach capitalizes on the ability of single channel recordings to resolve even modest changes in onor off-rates for protonation or in the conductance of protonated and deprotonated conductance states. Fig. 2 compares the in-teraction of H ϩ ions with wild-type Ca 2ϩ channels ( Fig. 2A) and channels where an aspartate is substituted for the conserved glutamate in repeat I (EID) (Fig. 2B). For openings of single wild-type channels, individual protonation/deprotonation events were detected as abrupt changes in the unitary current signal between a high conductance level corresponding to the deprotonated channel (␥ h ϭ 132 Ϯ 10 pS, n ϭ 6) and a low conductance level representing the protonated channel (␥ l ϭ 47 Ϯ 4 pS, n ϭ 6). These conductance levels appeared as two distinct peaks in the accompanying open-state all-points histogram ( Fig. 2A, right panel). The peaks were roughly the same size with K ϩ -2 H 2 O recording solution at pH 8.5, indicating that the open channel divided its time nearly equally between being deprotonated and protonated. The balance was shifted toward the high conductance state (ϳ85%) at pH 9.75 and away from the high conductance state (17%) at pH 7.5, as expected for a protonation reaction with pK a ϳ8.5 (8).
The behavior of the EID mutant differed from WT in several respects (Fig. 2B). Although transitions between high and low conductance states were retained, both levels of unitary current were increased relative to their WT values, as illustrated in the all-points histograms (Fig. 2B, right panel) and in analysis of the unitary current-voltage relationships. Over the range between Ϫ50 to Ϫ120 mV, the slope conductance of the protonated state (␥ l ) was significantly increased in the EID mutant relative to WT (94 Ϯ 10 pS, n ϭ 4 versus 46 Ϯ 9 pS, n ϭ 6, p Ͻ 0.001). As for the deprotonated state (␥ h ), unitary currents were consistently greater in EID than in control over the same voltage range. The slope conductances also may have increased slightly relative to WT, although not in a statistically significant way (150 Ϯ 5.3 pS, n ϭ 3 in EID versus 132 Ϯ 10 pS, n ϭ 6 in WT, p Ͼ 0.05). In addition to the increase in single channel current levels, the EID mutation also altered the balance between protonated and deprotonated states (Fig. 2B, right). In the case of the mutant channels, the proportion of time spent in the protonated state was only 31 Ϯ 1.5% (n ϭ 3) at pH 8.5, significantly less than the ϳ50% observed in WT channels (49.5 Ϯ 3%, n ϭ 6, p Ͻ 0.001). This change in the EID mutant corresponded to a ϳ2-fold reduction in the affinity of H ϩ ions for the protonation site relative to WT. The decrease of pK a in the EID mutant was due primarily to acceleration of the off-rate for protons, as revealed by kinetic analysis of single channel records (Fig. 2D). The time constant for the protonated state was reduced significantly, from 0.50 Ϯ 0.001 ms in WT to 0.23 Ϯ 0.001 in EID (p Ͻ 0.001). On the other hand, the time constant for the deprotonated state remained unchanged in EID compared with WT ( h ϭ 0.42 Ϯ 0.03 ms in EID versus 0.48 Ϯ 0.005 ms in WT, p Ͼ 0.05, unpaired t test). The present data establish that protonation can be destabilized without altering the charge of the side chain, merely by shortening it by a methylene group (ϳ1.4 Å), in strong support of the importance of EI in the formation of the native protonation site.
Since the Glu 3 Gln mutants in both repeats I and III displayed a very similar phenotype (8), it was of great interest to compare the behavior of EID to that of the Glu 3 Asp mutation in repeat III (EIIID) (Fig. 3). In EIIID, both h and l current levels were increased relative to control. Once again, the slope conductance of the low conductance state (␥ l ϭ 95 Ϯ 2.1 pS, n ϭ 3) was more than double the WT value, whereas the high conductance of EIIID (␥ h ϭ 161 Ϯ 4.5 pS, n ϭ 3) was only slightly increased if at all relative to WT (p Ͼ 0.05). The percent blockade at pH 8.5 was 33 Ϯ 2.7% (n ϭ 5), corresponding to a 2-fold reduction in proton affinity relative to WT. The altered pK a was due to an accelerated rate of deprotonation (k off , Fig.  3D, right). In all respects, the properties of the EIIID mutant were very much like those of EID, suggesting that the side ␣ 1C ; EIVD, glutamate to alanine mutation at position 1446 of rabbit ␣ 1C ; EIVD, glutamate to aspartate mutation at position 1446 of rabbit ␣ 1C ; pS, picosiemens. 6 (n ϭ 5), a significant increase compared with WT (p Ͻ 0.05, unpaired t test). On the other hand, EIID did not differ from WT in either the degree of proton block or in the kinetics of protonation and deprotonation. The proportion of channels in the l-state (49.5% Ϯ 0.7, n ϭ 5, pH 8.5) (Fig. 4B) was close to that found in WT, and the histograms of unblocked or blocked times (Fig. 4D) coincided perfectly with WT distributions. Evidently, shortening the side chain length of EII does not perturb the proton binding affinity, even though it can certainly influence the flux rates through high and low conductance states.
Replacement of the pore glutamate in repeat IV also produced changes in single channel behavior, some significantly different from the other Glu to Asp substitutions (Fig. 5). In EIVD, the slope conductances of protonated and deprotonated states were both increased compared with the wild-type channel. Estimates of the slope conductances ␥ l (74 Ϯ 7.3 pS, n ϭ 4) and ␥ h (174 Ϯ 16 pS, n ϭ 4) were ϳ60 and ϳ30% greater than WT values, respectively (p Ͻ 0.05 for both). More unusual was the finding that the EIVD mutation increased the proton affinity rather than weakening it. As illustrated in Fig. 5B, the percentage of time spent in the protonated state at pH 8.5 was 61 Ϯ 1.5%, corresponding to approximately 50% increase in affinity relative to WT. The increased proton affinity was well accounted for by prolongation of the intervals in the protonated state (Fig. 5D, right). The time constants for l in EIVD was 0.75 Ϯ 0.01 ms (n ϭ 7), an increase of 50% above WT (0.5 Ϯ 0.001 ms, p Ͻ 0.001). In contrast, h was 0.51 Ϯ 0.005 ms (n ϭ 7), no different than WT (0.48 Ϯ 0.005 ms, p Ͼ 0.46).
Our observations with the Glu to Asp mutations support the idea that both EI and EIII act in concert to form a single proton binding site. This conclusion seems at odds with Klöckner et al. (10), who reported that replacement of EIII with alanine completely abolished susceptibility to proton block and concluded that EIII was the sole determinant of the proton binding site. Accordingly, we re-examined the behavior of the EIIIA mutant, using 2 H ϩ instead of H ϩ as blocking ion to slow deprotonation rates and improve resolution of current transitions (Fig. 6). At pH 8.5, EIIIA displayed flickery transitions between two conductances (Fig. 6A, left), a predominant h level and less fre-quent l level. The transitions were seen more clearly on an expanded time scale (lower left trace, Fig. 6A). Increasing the external proton concentration to pH 7.5 shifted the balance toward the lower conductance but left the h and l current levels unchanged. More quantitative analysis with all-points histograms indicated that occupancy of the protonated state was ϳ20% at pH 8.5 and ϳ50% at pH 7.5. These results are perfectly consistent with a conventional protonation reaction with the low conductance level corresponding to the protonated state. The pK a of the reaction was ϳ7.5, about 1 log unit more acidic than that of WT channels under the same experimental conditions. Thus, the behavior of EIIIA is actually quite similar to that of EIA (pK a 7.4 in K ϩ -H 2 O (10), corresponding to an expected value of 7.8 in K ϩ -2 H 2 O). In both cases, the Glu 3 Ala mutation spares the susceptibility to protonation while reducing the H ϩ affinity by 5-10-fold relative to WT.
The conductance of the protonated state of the EIIIA mutant was 112 Ϯ 10 pS, a ϳ2.5-fold increase relative to native ␣ 1C channels, whereas the conductance of the deprotonated state (ϳ150 pS) was little changed relative to WT. This sharply narrowed the difference between h and l conductances and may have contributed to the initial difficulties in detecting proton block in EIIIA. Transitions were more readily identified in our experiments because the transitions were slowed by use of 2 H 2 O rather than H 2 O, and the recordings from the EIIIA mutant were performed at a more negative potential (Ϫ70 mV rather than Ϫ40 mV) to increase driving force and unitary current size. The previously published records for EIIIA (10) display a significant degree of flickering in the open channel current, hinting at the existence of transitions between a predominant h level and a less prevalent l level, as is evident in our recordings (Fig. 6). DISCUSSION We have carried out critical tests of two different models of the protonation site within the pore of L-type Ca 2ϩ channels. One model identified EIII (Glu-1086) as the unique proton acceptor site, solely responsible for the appearance of the low conductance state (10). This hypothesis has become untenable since it is now clear that replacement of this residue by alanine actually spares responsiveness to H ϩ block (Fig. 6). When our results are considered along with previous evidence (10), it is evident that each of the four Glu 3 Ala substitutions shares the ability to destabilize the titratable proton. Estimated shifts in the pK a of the Glu 3 Ala mutants relative to wild-type ranged from ϳ0.9 to ϳ0.3 pH units, in the order EIII Ͼ EII Ϸ EI Ͼ EIV. The differences between the various Glu 3 Ala constructs are subtle enough to make it difficult to decipher the roles of the individual glutamates on the basis of these mutants alone.
The glutamate to aspartate mutations provided new perspectives on the protonation site. They are particularly useful in providing relatively subtle changes in the positions of key side chains without altering their net charge. Thus, Glu 3 Asp replacements provide information complementary to that derived from glutamine substitutions, which alter net charge with little change in geometry, thereby highlighting the importance of electrostatic effects on permeation (8). We used the aspartate mutants to test another model in which EI and EIII jointly coordinate the titratable H ϩ ion (8). Shortening an amino acid side chain by a single methylene group (ϳ1.4 Å), while maintaining its carboxylic acid terminus, must be regarded as a relatively mild structural alteration. Yet, each of the Glu 3 Asp replacements was sufficient to cause a significant alteration in channel behavior. These effects took the form of changes in the balance between protonated and deprotonated states in some cases and changes in their conductance levels in all cases. Fig. 7 summarizes equilibrium and kinetic data for the protonation reaction, studied at pH 8.5. Of all the Glu 3 Ala constructs, only EID and EIIID displayed a significant reduction in the degree of H ϩ inhibition, indicating a weakening of proton affinity (Fig. 7A). This supports the hypothesis that EI and EIII play essential roles in formation of the protonation site (8). The alteration in proton affinity is entirely accounted for by a doubling of the off-rate (Fig. 7B), with no detectable change in on-rate (Fig. 7C). The constancy of k on is not surprising since this rate is thought to be dominated by bulk aqueous diffusion of H ϩ and its interaction with the negative charge of the protonation site (23), neither of which are expected to change with the Glu 3 Asp mutation. The speeding of k off also makes good sense in terms of a protonation site formed by oxygen atoms of opposing carboxylate groups. Shortening either side chain would sharply restrict the set of configurations consistent with retention of the H ϩ , thereby hastening deprotonation. Along these lines, one would expect that protonation would be further destabilized in the construct EID-EIIID. This in fact was observed. The protonated state in the double mutant became even briefer than in EID or EIIID, to the point where dwell times were not completely resolved (data not shown).
In contrast to the very similar behavior of EID and EIIID, aspartate replacements in repeats II and IV each produced a different effect. EIID displayed the same degree of proton block as WT (Fig. 7A), with no changes in either k on or k off (Fig. 7, B and C). This seems compatible with the previous hypothesis, which proposed that EII has no direct interaction with the titratable H ϩ , only an indirect effect through hydrogen bonding with EI and EIII. Since the Glu 3 Asp replacement caused no alteration of side chain charge, it is not surprising that it gave a very different effect from the Glu 3 Gln mutation at the same position, which resulted in a ϳ10-fold reduction in proton affinity. One might have expected that shortening the carboxylate side chain in domain II would cause some change in pK a , merely through disturbance of the positioning of EI and EIII. On the other hand, it is possible that a slight tilting of EI and EIII would be sufficient to accommodate the Glu 3 Asp substitution or that other compensatory rearrangements take place, leaving the configuration of the proton binding site essentially unaltered.
Unlike the other three constructs, the EIVD mutant displayed the striking property of accentuating the inhibition at pH 8.5, indicating an increase in proton affinity. This arose from a slowing of k off but, once again, no change in k on (Fig. 7). The stabilization of the bound proton might be partly attributed to a through-space electrostatic interaction of bound H ϩ ions with the carboxylate side chain of EIV, if it has been correctly pictured as projecting toward the cytoplasmic end of the pore (8,24). Shortening the side chain would tend to bring the negative charge of its head group closer to the bound proton, thus promoting its retention. The conductances of the deprotonated (h) and protonated (l) states provided valuable information about the local environment within the channel pore (Fig. 8). In general, conductances of the aspartate mutants significantly exceeded those of WT channels. The low conductance was increased by Glu 3 Asp replacements at each of the four positions. A simple interpretation is that the various side chains protrude into the pore and thereby influence the energetic profile that a permeating ion encounters. Replacing a glutamate side chain with a smaller aspartate side chain would tend to widen the aperture for passage of monovalent ions. The Glu 3 Asp substitutions can be contrasted with the Glu 3 Gln replacements, which did not significantly increase ␥ l relative to the protonated state of WT (8). This makes sense since glutamine is isosteric with glutamate and the Glu 3 Gln substitution may be expected to leave the geometry of the permeation path unchanged.
Similar considerations may apply to the high conductance state. Since glutamine replacements in domains I and III completely eliminate the high conductance state, these can only be directly compared for Glu 3 Asp and Glu 3 Gln replacements in domains II and IV (8). Here again, Glu 3 Asp replacements caused a clear increase in ␥ h relative to WT (Fig. 8B), and the corresponding glutamine substitutions left ␥ h unaltered (8).
To focus specifically on the incremental effects of protonation on channel conductance, we compared the ratio of low to high conductance values (␥ l /␥ h ) for the various mutants (Fig. 8C). In EID and EIIID, this ratio was significantly closer to unity than in WT or the other aspartate mutants. The similarity between EID and EIIID provides further support for the idea that EI and EIII act as nearly equal partners in coordinating the titratable proton (8). For example, one might imagine that binding of H ϩ stabilizes EI and EIII side chains in a symmetrical carboxyl-carboxyl configuration, thereby impeding monovalent ion flux. The high value of ␥ l /␥ h indicates that shortening the side chain of either EI or EIII has a disproportionate effect on the conductance of the protonated state. If the protonated EI-EIII complex posed the most severe impediment to monovalent ion flux (e.g. in forming the narrowest aperture), Glu 3 Asp replacement would be expected to yield a relatively large degree of relief (e.g. in widening such an aperture). Open-state all-points amplitude histograms of EIIIA mutant at pH 8.5 (left) and pH 7.5 (right) were each fitted as the sum of two Gaussian functions (solid curves), with individual Gaussians indicated by dotted curves. Note that the balance between the two peaks is shifted in favor of the l-state at pH 7.5 compared with pH 8.5. The all-points amplitude histogram of the closed state is included for comparison, along with the corresponding Gaussian fit (solid curve). C, unitary current-voltage relationships of the EIIIA mutant. Current amplitudes arising from h conductance state (open symbols) and l conductance state (closed symbols) at various testing potentials are shown. Current amplitudes for each state were no different at pH 8.5 (triangles) and at pH 7.5 (circles). The solid lines represent linear fits to the data obtained at pH 8.5.

FIG. 7. Comparison of the effects of Glu 3 Asp substitutions on the potency of proton block.
A, comparison of the percent inhibition of WT channels and of the four Glu 3 Asp mutants at pH 8.5. Relative to WT, the reductions of proton block in EID and EIIID mutants, and the potentiation of block in EIVD mutant are all statistically significant (p Ͻ 0.05). B, the effect of Glu 3 Asp substitution on the time constant of the l conductance (protonated) state ( l ). Estimates of l for EID, EIIID, and EIVD are significantly different than the value for WT channels. C, lack of effect of Glu 3 Asp mutations on time constants of the h-state ( h ). No statistically significant differences are detected between WT and mutant channels. The time constants for h-state ( h ) and l-state ( h ) were calculated from the mean dwell times using a method to correct for instrumental dead-time and missed events (22). * denotes that data are statistically different (p Ͻ 0.05) from that of WT.
On the other hand, aspartate substitution at the EII and EIV positions increases ␥ l and ␥ h to nearly the same degree, suggesting a more general effect in favoring current flow. In the case of EIV, this fits well with the proposal that its side chain is positioned downstream of the point of greatest pore constriction (8,24,25).
It is interesting to compare the properties of L-type Ca 2ϩ channels with the behavior of other systems where multiple carboxylates are involved in H ϩ titration. The pore of the cyclic nucleotide-gated (CNG) channel also contains four glutamates, one from each of four monomers, which appear to cluster in pairs to form two independent protonation sites (pK a ϳ7.6) (11). Although mutagenesis has not yet been reported at the level of individual glutamates, replacement of all four glutamates with aspartate was found to greatly weaken the H ϩ affinity and to obscure the appearance of discrete conductance levels (11). Another interesting case is the photosynthetic reaction center of the purple bacterium, a well characterized H ϩ transporter, whose Q B cluster contains a tightly coupled set of three glutamates. The carboxylate side chains of these residues strongly interact, giving a pK a as basic as 8.2 depending on oxidation state (16). In the Q B cluster, two Glu residues seem to share the titratable proton, whereas a third remains fully protonated over a wide experimental range, very much similar to the behavior of L-type Ca 2ϩ channels described here. In both of these systems, as in voltage-gated Ca 2ϩ channels, the protonation site arises from multiple glutamate residues, each originating from a distinct region of the transmembrane protein.
The effects of the aspartate mutations in L-type channels provide a compelling basis for further clarification of how such carboxylate head groups might work in concert.