|Home | About | Journals | Submit | Contact Us | Français|
Ca2+-activated BK channels modulate neuronal activities including spike frequency adaptation and synaptic transmission. Previous studies found that Ca2+ binding sites and the activation gate are spatially separated in the channel protein, but the mechanism by which Ca2+ binding opens the gate over this distance remains unknown. By studying an Asp to Gly mutation (D434G) associated with human syndrome of generalized epilepsy and paroxysmal dyskinesia (GEPD), we show that a cytosolic motif immediately following the activation gate S6 helix, known as the AC region, mediates the allosteric coupling between Ca2+ binding and channel opening. The GEPD mutation inside the AC region increases BK channel activity by enhancing this allosteric coupling. We found that Ca2+ sensitivity is enhanced by increases in solution viscosity that reduce protein dynamics. The GEPD mutation alters such a response, suggesting that a less flexible AC region may be more effective in coupling Ca2+ binding to channel opening.
Ion channels are proteins that conduct ions across cell membranes. Channel proteins regulate membrane excitation and ionic homeostasis in response to cellular signals, and mutations of ion channels often induce serious, sometimes life-threatening diseases(Ashcroft, 2000). Structural studies in recent years have greatly advanced our understanding of the basic working modules of ion channels including the activation gate (Yellen, 1998) and sensors to various stimuli such as voltage (Bezanilla, 2008; Swartz, 2008) and chemical ligands (Jiang et al., 2002; Unwin, 2003; Xia et al., 1998; Zagotta et al., 2003). However, the structural basis for the energetic coupling between sensors and the activation gate still remains elusive. Considering the fact that many disease-causing mutations are outside of the gate and sensors, it is particularly important to understand the molecular mechanism of the energetic coupling, i.e. the structural components of the allosteric activation pathways, and how they are modified by channel mutations. Activated by both membrane depolarization and increases in intracellular Ca2+ concentration ([Ca2+]i), large-conductance, voltage and calcium-activated K+ (BK) channels become one of the best systems to address this question (Cui et al., 2008). Recently, we characterized a mutation D434G in human BK channels (hSlo1, GenBank accession number, GI: 26638649) which was identified from patients with nervous disorders of coexistent generalized epilepsy and paroxysmal dyskinesia (Du et al., 2005). This epilepsy/dyskinesia mutation (hD434G) significantly increases Ca2+ sensitivity of BK channels under physiological conditions. More interestingly, the hD434G mutation resides in a cytosolic motif which is important for the allosteric coupling between Ca2+ binding and channel activation (Jiang et al., 2002; Krishnamoorthy et al., 2005), but outside of the putative Ca2+ binding sites (Figure 1A), implying its role in the allosteric Ca2+-dependent activation. Thus, this mutation provides us a unique opportunity to uncover the structural basis and dynamic nature of the coupling between Ca2+ binding and BK channel opening.
BK channels are composed of four identical alpha subunits encoded by the Slo1 gene (Atkinson et al., 1991; Shen et al., 1994). Each Slo1 subunit contains a membrane-spanning domain, which includes the voltage sensor and the ionic pore with the activation gate, and a large cytoplasmic C-terminal domain where two putative Ca2+ binding sites, D367 and the Ca2+ bowl, have been identified (Schreiber and Salkoff, 1997; Xia et al., 2002) (Figure 1A). In addition, the channel contains a third site that binds both Ca2+ and Mg2+ at high (mM) concentrations (Shi et al., 2002; Xia et al., 2002; Yang et al., 2008), which is not shown in Figure 1A because under the experimental conditions used in this paper this site will not be occupied. The atomic structure of the BK channel has not been solved. However, the X-ray crystallographic structure of KV1.2 channel (Long et al., 2005) has been used as a model for the membrane-spanning domain of BK channels, while the RCK (regulator of K+ conductance) domains of the MthK channel (Jiang et al., 2002) are suggested as a model for the cytoplasmic domain (Cui et al., 2008). A recent three-dimensional structure of BK channels from electron cryomicroscopy is consistent with these homology models (Wang and Sigworth, 2009). Structural and functional studies suggested that Ca2+ binding activates the channel through an allosteric mechanism (Magleby, 2003). At the molecular level, the allosteric mechanism is suggested to be similar as that for the MthK channel such that Ca2+ binding first alters the conformation of the cytosolic domain, which then opens the activation gate by pulling the S6 transmembrane segment (Jiang et al., 2002; Niu et al., 2004). To date, no structural component in the cytosolic domain that changes conformation during Ca2+ dependent activation has been identified in BK channels, and the nature of such conformational changes is not known. The hD434G mutation in BK channels causes an increase in the activity of BK channels at the same membrane potentials and [Ca2+]i’s as compared to the wild type (WT) hSlo1 channels. This change may be the basis for the association of the mutation with the neurological disorders (Brenner et al., 2005; Du et al., 2005). To investigate how this mutation uniquely enhances channel function, we studied the hD434G–equivalent D369G in mouse Slo1 (mSlo1) BK channels (Figure 1). These studies identified the N-terminus of the RCK1 domain containing 76 amino acids, a region including the secondary structures βA-αC (Jiang et al., 2002) and thus named the AC region, as part of a structural basis connecting Ca2+ binding to channel opening. The results suggested a conformational and dynamical change of the AC region during BK channel activation that is altered by the mD369G mutation. These results also provide a novel allosteric mechanism for how ligand binding opens activation gates in ion channels.
Experiments were carried out on mSlo1 rather than hSlo1. The mSlo1 channel is homologous to hSlo1; aside from the different lengths at the N- and C- termini, 99% amino acids are identical between the two channels (Butler et al., 1993; Du et al., 2005). The D369G mSlo1 channels exhibit the same phenotype as D434G hSlo1 such that at the same voltage and physiological [Ca2+]i the mutant channels activated more than the WT channels (Figure 1B). We studied the mSlo1 channels over a more complete range of [Ca2+]i’s from nominal 0 (~0.5 nM) to the near saturating 111.5 µM, and the results revealed that the mutation enhances Ca2+ sensitivity of channel activation (Figures 1C–F). The increase of [Ca2+]i shifted the voltage range of the conductance-voltage (G–V) relations of the mutant channels to more negative voltages as compared to those of the WT channels (Figures 1C and 1D). The G–V relation of BK channel activation can be approximated by fitting the data to the Boltzmann function with two independent parameters: V1/2 (voltage at half-maximum conductance) and z (proportional to the steepness of the curve). A plot of V1/2 vs. [Ca2+]i clearly shows that Ca2+ binding shifts the G–V relation to more negative voltage ranges, and that the shifts for the mD369G mutant channels are larger than that for the WT channels (Figure 1E). To further quantitatively estimate the change of Ca2+ sensitivity caused by the mutation, we fit the G–V relations of both mutant and WT channels at various [Ca2+]i to the MWC model of BK channel activation (Cox et al., 1997; Magleby, 2003) (Figures 1C and D) and list the parameters (Figure 1F), where Ko and Kc are dissociation constants for Ca2+ binding at the open and closed conformations, respectively. The mD369G mutation decreased both Ko and Kc, consistent with an enhancement of Ca2+ sensitivity. Such a change in Ca2+ sensitivity would increase open probability of the channel by 1.4 – 5 times during a neuronal action potential with an estimated amplitude of ~30 mV and local [Ca2+]i between 2–10 µM (Figures 1C and 1D). Previous studies suggested that the increased BK channel activity may be prominent at the peak of neuronal action potentials, where [Ca2+]i increases due to Ca2+ influx via voltage dependent Ca2+ channels, to reduce the duration and increase the frequency of action potentials and result in seizures (Brenner et al., 2005).
The shift of G–V relations in response to changes in [Ca2]i is an effective measurement of Ca2+ sensitivity of BK channels (Cui et al., 1997), which has been the primary method for investigating the gating mechanism of BK channels including the identification of Ca2+ binding sites (Schreiber et al., 1999; Xia et al., 2002). However, since BK channels are activated by both voltage and Ca2+ (Cui et al., 2008), this method does not directly reveal the mechanistic properties of Ca2+ dependent activation. To eliminate the influence of voltage dependent activation, we further examined Ca2+ sensitivities of the WT and D367G mSlo1 using limiting slope measurements at extremely negative voltages where BK channels open spontaneously and independently of voltage sensor movements (Horrigan et al., 1999). A patch containing hundreds of BK channels is held at negative voltages, where the open probability of the channels is so small that only rare openings of single channels are observed (Figure 2A). The open probability of the channels increases in the presence of Ca2+, resulting in marked increase of single channel openings. Figure 2 shows that mutation mD369G enhances Ca2+ sensitivity of the channel. At −140 mV, the number of single channel opening events of both WT and mD369G increases with increasing [Ca2]i, but more prominently for mD369G (Figure 2A). The Po-V relations at various [Ca2]i become flat at negative voltages (Figure 2B and C), indicating that channel opening at these voltages no longer depends on voltage sensor movements. The Ca2+ dependences of Po at −140 mV for WT and mD369G channels (Figure 2D) clearly show that mutation mD369G enhances Ca2+ sensitivity. The fit of the data by the Hill equation (solid lines) results in a microscopic dissociation contant of 15 and 8 µM and a Hill coefficient of 2.8 and 3.5 for WT and mD369G, respectively. The larger Hill coefficient for mD369G indicates a higher cooperativity among Ca2+ binding sites in the mD369G channels, thus mutation mD369G enhances the allosteric coupling among Ca2+ binding sites, possibly via an enhanced allosteric coupling between Ca2+ binding sites and the activation gate.
How does the mD369G mutation perturb the molecular mechanism of Ca2+ dependent activation to result in a gain-of-function? Previous studies showed that the mutation of each of the two putative Ca2+ binding sites, D367A and 5D5N (five consecutive Asp residues at 897–901 mutated to Asn), eliminates part of Ca2+ dependent activation and the sum of the effects is close to the total high-affinity Ca2+ sensitivity of the channel. These results suggest that the channel is activated by two separate Ca2+ dependent pathways involving either D367 or the Ca2+ bowl, with only a weak cooperativity between the two pathways (Qian et al., 2006; Sweet and Cox, 2008; Xia et al., 2002) (Figure 3A). Therefore, to answer the above question, we first investigated whether the mD369G mutation affects each of the activation pathways, and then addressed how a particular pathway is affected. We found that eliminating the function of one pathway by D367A also abolishes the difference in Ca2+-induced G–V shift between the mD369G mutant and the WT channels, since V1/2 changes with [Ca2+]i nearly in parallel for the two channels (Figures 3B and 3C). This result indicates that the mD369G mutation no longer enhances Ca2+ sensitivity when the pathway associated with D367 is disabled. On the contrary, when the function of the Ca2+ bowl pathway is eliminated by the 5D5N mutation, the mD369G mutation continues to enhance Ca2+ sensitivity in a similar way as in the WT channel (Figures 3D and 3E). Taken together, these results suggest that the epilepsy/dyskinesia mutation specifically affects the Ca2+ dependent activation pathway associated with D367.
Then, what are the targets in the D367-specific pathway that are subject to modification by mD369G to result in an increase in Ca2+ sensitivity? Based on the structural model of the RCK1 domain (Jiang et al., 2002), D367 appears to be located some distance from the activation gate. Thus the D367-specific activation pathway may involve a large structure that includes many residues (Figure 3A). Since mutation mD369G is located close to D367, it is possible that the mutation may alter local interactions of the D369 side chain with the putative Ca2+ binding site to enhance Ca2+ sensitivity (Wang et al., 2009). On the other hand, it is also possible that mD369G may alter the conformation of the activation pathway and allosterically enhance Ca2+ sensitivity (Figure 2D). To distinguish between these two possibilities, we mutated D369 to conservative (D369E) and neutral (D369A, N, W, and P) amino acids with various side chain sizes and examined the effects of these mutations on Ca2+ sensitivity. Among all these mutations, only mD369G increased Ca2+ sensitivity, with V1/2 of the G–V relations shifted to more negative voltage ranges (Figure 3F). The effects of other mutations were not as pronounced and varied at different [Ca2+]i’s, increasing channel activation at some concentrations, but decreasing activation at other concentrations. These results suggest that the side chain of D369 is not part of the Ca2+ binding site (also see (Shi et al., 2002; Xia et al., 2002)), and it may not contribute to Ca2+ sensitivity by the short range electrostatic or van der Waals interactions with the putative Ca2+ binding site. Rather, the specific effect of mD369G in increasing Ca2+ sensitivity may derive from the ability of glycine to adopt a wide range of main-chain dihedral angles and make protein structure flexible at its site, which may alter the conformational changes of the D367-specific pathway during channel gating. Consistent with this suggestion, the G–V relation of the mD369G mutant channels shifted a small but significant amount to more negative voltages (ΔV1/2 = 8.8 ± 3.8 mV (mean of difference ± standard error of difference, n = 6 for WT and 20 for mD369G), p = 0.029 in unpaired Student’s t-test) at 0 [Ca2+]i (Figure 1E), indicating that mD369G can alter channel function by a change in the structure that links to the activation gate, instead of by merely changing Ca2+ binding. Furthermore, although mD369G no longer enhances Ca2+ sensitivity of the channel with the D367A mutation, it alters channel activation independently of [Ca2+]i, shifting the G–V relation at all [Ca2+]i’s (Figures 3B and 3C). It is possible that D367A eliminated Ca2+ binding to the putative site but did not destroy the rest of the activation pathway down-stream from Ca2+ binding, which was subsequently altered by mD369G mutation to affect channel activation.
To examine if the flexibility of the channel protein matters to gating and if mD369G alters the flexibility of the channel protein, we studied the WT and mutant channels in intracellular solutions with increased viscosity. The dynamics of a protein in solution are intimately coupled to the dynamics of the solvent; the fluctuation amplitudes and relaxation rates, i.e. flexibility, of proteins can be reduced by increases in solution viscosity (Ansari et al., 1992; Beece et al., 1980; Frauenfelder et al., 2009). Previous studies showed that the increases in solution viscosity may affect the moving parts of voltage dependent Na+ channels during gating to reduce activation rates (Fumio, 2000; Kukita, 1997). In our experiments, increasing the viscosity of intracellular solution by addition of sucrose (measured viscosity = 0.906 ± 0.003 mPa•s and 23.1 ± 0.1 mPa•s, mean ± s.e.m., n = 12 at 0 and 2 M sucrose, respectively) slows down the time course of activation and deactivation (Figures 4A and 4B) and enhances Ca2+ dependent activation (Figures 4C and 4D) of both the WT and mD369G channels. In 200 µM [Ca2+]i, the reduction of the deactivation rate is smaller as compared to the enhancement of activation rate (Figure 4B), and since BK channel activation can be approximated by a two-state voltage dependent activation mechanism in saturating [Ca2+]i’s (Cui et al., 1997), the changes in time courses of the current are consistent with the changes in G–V relations. Adding 9 M glycerol in the intracellular solution (measured viscosity = 20.2 ± 0.1 mPa•s, mean ± s.e.m., n = 12), which resulted in a similar increase of viscosity as 2 M sucrose, also caused similar changes in BK channel activation (Figures 4E and 4F). On the other hand, addition of 2 M urea to intracellular solution, which alters viscosity little (measured viscosity = 0.906 ± 0.003 mPa•s and 0.987 ± 0.005 mPa•s, mean ± s.e.m., n = 12 at 0 and 2 M urea, respectively), had no effect on channel activation (Figures 4E and 4F), excluding the possibility that the changes in BK channel activation with sucrose were due to changes in osmolarity or other unknown effects associated with high concentrations of solutes. These results suggest that gating of BK channels is affected by the flexibility of the channel protein.
The results show that mutation mD369G alters the responses of activation and deactivation kinetics to the increased viscosity (Figure 4B) and causes a reduced shift of the G–V relation as compared to that in WT when the [Ca2+]i is increased from 0 to the saturating 200 µM (Figures 4C and 4D). Thus, the enhancement of Ca2+ sensitivity due to an increase of viscosity for mD369G channels is less than that for WT. Unlike mD369G, mutation D369A, which does not alter Ca2+ sensitivity (Figure 3F), has no effect on the response of BK channels to sucrose (Figure 4D). These results support that mD369G alters the flexibility of the BK channel protein to potentiate Ca2+ sensitivity.
Results in Figure 3 have shown that mD369G specifically alters the Ca2+ dependent activation pathway associated with D367 (Figure 3C), but not the pathway associated with the Ca2+ bowl (Figure 3E). Our unpublished results also indicate that mD369G does not affect Mg2+ dependent activation (Diez-Sampedro et al., 2006). To examine if a change of solution viscosity affects any specific metal dependent activation pathways, we measured channel activation with each of the mutations E374A, D367A and 5D5N, which abolished metal binding to the Mg2+ binding site (Shi et al., 2002), the Ca2+ binding site at D367 (Shi et al., 2002; Xia et al., 2002) and the Ca2+ binding site at the Ca2+ bowl (Schreiber and Salkoff, 1997), respectively. D367A, but not E374A or 5D5N, abolished the response of BK channels to sucrose (Figures 4E and 4F), suggesting that only the Ca2+ dependent activation pathway associated with D367 specifically depends on the dynamics of the channel protein for the coupling between Ca2+ binding and channel opening, which is consistent with the result that mD369G specifically affects the same activation pathway (Figure 3).
It is important to note that both the increase of viscosity, which reduces the overall protein dynamics, and the mD369G mutation, which enhances the local peptide flexibility, potentiates Ca2+ sensing. This result suggests that the local peptide flexibility enhanced by mD369G may reduce the dynamics in other parts of the protein allosterically, consistent with the results in Figures 1E and and3C.3C. Taken together, these results suggest that mD369G alters the flexibility and conformation of the channel protein to potentiate the allosteric coupling between Ca2+ binding at the D367 site and channel opening.
To demonstrate more directly that mD369G affects the allosteric coupling between Ca2+ binding and channel opening, we find structural perturbations that affected the coupling in the D367-specific activation pathway and then test if such perturbations also alter the effect of mD369G. In search of such perturbations, we performed a mutation-scan in the AC region, which is a crucial gateway for the allosteric coupling between Ca2+ binding and channel opening because it physically connects the activation gate to the rest of the cytosolic domain (Figure 1A), and is important in determining Ca2+ sensitivity (Krishnamoorthy et al., 2005). Furthermore, both the putative Ca2+ binding site D367 and the mutation mD369G are situated within the AC region so that the mutation may alter its structure to affect Ca2+ sensitivity. In this experiment, most residues were mutated individually to Ala and some to other amino acids (Figure 5A). The loss of Ca2+ sensitivity of the channel due to each mutation is quantified by the change in the G–V shift (ΔV1/2) when [Ca2+]i changes from nominal 0 to the near saturating 99.3 – 111.5 µM (i.e. ΔΔV1/2(ΔCa 2+ ) = ΔV1/2 WT - ΔV1/2 Mutant). We classify a ΔΔV1/2(ΔCa 2+ ) of more than ± 20 mV as significant since the G–V relation of BK channels often exhibits variations within a voltage range of ≤ 20 mV. Ten mutations are found to reduce Ca2+ sensitivity significantly (indicated by asterisks in Figure 5A and Supplemental Figure S1), including D362A and D367A that have been previously reported (Shi et al., 2002; Xia et al., 2002). These mutations are located in different secondary structures within the AC region, including αA, αB, βC, and the inter loop between αA and βB that contains Asp-Arg-Asp-Asp at positions 367–370 and is called the DRDD loop (Figure 5A). Consistent with D367 being part of a Ca2+ binding site, mutations in the DRDD loop resulted in large reductions in Ca2+ sensitivity; and D367A reduced the G–V shift by 105 ± 4 mV (mean of difference ± standard error of difference, n = 6 for WT and 5 for D367A) as compared to the 182 ± 3 mV (n = 6) total shift in WT mSlo1 in response to an increase of [Ca2+]i from 0 to 111.5 µM, equivalent to a ~60% reduction of Ca2+ sensitivity. The functionally important mutations in other secondary structures are separated by large distances from the DRDD loop in the structural model of the RCK1 domain (see Figures 7 and and8).8). These residues are important in the D367-specific pathway (see below), but cannot be part of the Ca2+ binding site because of their long distances from D367 and because they are mostly hydrophobic (Figure 5). The long distances from D367 also suggest that it is unlikely for all these residues to make direct interactions with the putative Ca2+ binding site. Therefore, at least some of these mutations reduced Ca2+ sensitivity by perturbing the allosteric coupling between Ca2+ binding and channel opening.
Since mD369G may enhance Ca2+ sensitivity by potentiating the allosteric coupling between Ca2+ binding and channel opening, do these perturbations also alter the effect of mD369G? Before answering this question, we examined if these mutations, like mD369G, are able to affect the activation pathway associated with D367 specifically. Figure 5B shows that the combined mutations in αA and αB with 5D5N (F359A/L387A/F391A/5D5N) reduced Ca2+ sensitivity more than 5D5N alone. Since 5D5N completely abolishes the Ca2+ sensitivity derived from the Ca2+ bowl (Schreiber and Salkoff, 1997; Xia et al., 2002), the additional reduction in Ca2+ sensitivity suggests that the mutations in αA and αB may perturb the activation pathway associated with D367. Consistent with these results, the combined mutations in αA, αB, and βC with D367A (F359A/D362A/H365A/L387A/F391A/T396A/F400A/D367A) did not reduce Ca2+ sensitivity more than D367A alone (Figure 5C), indicating that these mutations all specifically affect the activation pathway associated with D367; once the upstream pathway is disabled by mutation D367A, additional mutations perturbing the coupling between Ca2+ binding and channel opening were no longer able to affect Ca2+ sensitivity. With more combinations of mutations, we further studied how these mutations in each individual secondary structure perturb Ca2+ dependent activation (Figure 5D and Supplemental Figure S2). The Ca2+ sensitivity is measured as the total G–V shift induced by a change in [Ca2+]i from 0 to saturation (Figure 5A), which represents the free energy change during Ca2+ dependent channel opening (ΔGCa 2+ = ΔzeV1/2) (Cui and Aldrich, 2000), where e is the elementary charge and z is similar at various [Ca2+]i (Figures 1C and 1D) and largely unchanged by mutations (Figure 5A). Therefore, we assumed that for any combination of mutations, if the mutations do not affect the same molecular process, the sum of the effects on Ca2+ sensitivity by the individual mutations should be the same as that of the combined mutations (Horovitz and Fersht, 1990). Mutations in the DRDD loop (H365A/D367A/R368A) reduced Ca2+ sensitivity that is indistinguishable from that of the single mutations D367A, indicating that these mutations all affect the same molecular process. Contrarily, the reduction in Ca2+ sensitivity was nearly additive by mutations in αB or βC (L387A/F391A or T396A/F400A) (Figure 5D), suggesting that in αB or βC, each mutated amino acid may independently contribute to local interactions that are important for channel gating. However, the combined mutations in αB and βC (L387A/F391A/T396A/F400A) reduced no more or even less Ca2+ sensitivity than the mutations in αB alone (L387A/F391A), indicating that αB and βC also cooperate as part of a common molecular process in Ca2+ dependent activation. When the mutations in αA and αB were combined (F359A/D362A/L387A/F391A) they had a larger effect than mutations in individual secondary structures (F359A, D362A, or L387A/F391A); nonetheless, the effect was less than the summed effects of all individual mutations. Therefore, while αA, αB and βC may be involved in localized interactions, they also cooperate as part of a common molecular process contributing to Ca2+ dependent activation. Taken together, these results suggest that all perturbations in the AC region that reduce Ca2+ sensitivity (Figure 5A) affect the activation pathway associated with D367, and they all are allosterically connected such that one perturbation can affect the outcome of other perturbations.
We then examined how the above mutations alter the effect of mD369G by combining mD369G with mutations in αA (F359A), αB (L387A/F391A) or βC (T396A/F400A) that are likely to be located long distances away from D369 (Figure 6). Interestingly, while the mutations in αA, αB or βC reduce total Ca2+ sensitivity, they either enhanced or did not significantly alter the effect of mD369G (Figures 6A–D). Compared to the WT mSlo1, the mD369G mutation shifted the G–V relation maximally around 1.8 µM [Ca2+]i; however, the mutation in αA altered the profile of mD369G effects and increased G–V shift, with the largest shift at 111.5 µM [Ca2+]i (Figures 6A and 6D). The mutations in αB or βC did not alter the profile of mD369G effects, nor the maximal G–V shifts (Figures 6B–D). Thus, the effect of mD369G is determined by the mutations in αA, αB and βC within the AC region. Since these mutations are all allosterically connected in perturbing the D367-specific Ca2+ dependent activation pathway (Figure 5), any of these mutations would alter most allosteric interactions within the AC region to provide a unique network of allosteric connections that determines the effects of mD369G on Ca2+ sensitivity. Reciprocally, mD369G also affects most allosteric interactions in the AC region to enhance Ca2+ sensitivity. Figure 6E shows that in the presence of F359A, mD369G reduced the response of channel activation to the increase of viscosity, suggesting that F359A does not alter the mechanism for the function of mD369G such that mD369G may still reduce the flexibility of the AC region even though the allosteric pathway is altered by F359A. Based on all the results presented in this paper, we conclude that mD369G alters the conformation and dynamics of the channel protein allosterically to potentiate Ca2+ sensing via the pathway associated with D367, and the AC region is an important structural component of this pathway.
This study demonstrates that mD369G enhances BK channel activity by altering the allosteric coupling between Ca2+ binding and gate opening. The experimental results that lead to this conclusion have also brought insights into the mechanism of such a coupling. Previous studies suggested that Ca2+ binds to two putative sites, the D367 site (Xia et al., 2002) and the Ca2+ bowl (Schreiber and Salkoff, 1997), in BK channels to activate the channel through two separate intra-molecular pathways, i.e., two separate subsets of conformational changes are responsible for coupling the activation gate to the two Ca2+ binding sites. However, aside from the putative Ca2+ binding sites, the structural basis of the two activation pathways was not clear. In this study, we show that the mD369G mutation and the mutations in the AC region primarily alter the Ca2+ dependence derived from the D367 associated activation pathway (Figures 3, ,4,4, ,6).6). These results further demonstrate that Ca2+ activates the channel through separate intra-molecular pathways, and the allosteric interactions within the AC region are part of the D367 specific pathway.
Our results suggest that the AC region is an important structural component in coupling Ca2+ binding to channel opening. Consistent with this role, homology modeling of the AC region in BK channels and its superimposition on the full gating ring of the MthK channel reveal that the Ca2+ sensitive mutations are distributed in a distinctive fashion from the periphery of the cytosolic domain to the center of the channel (Figures 7A and 7B). The DRDD loop, where the putative Ca2+ binding site and the mD369G mutation are located, is positioned at the outer edge of the channel molecule, and αA, αB and βC lie just beneath the central pore, making it possible for the allosteric connection within the AC region to couple Ca2+ binding to the activation gate. Sequence alignments show that the residues important for Ca2+ sensitivity in the AC region are conserved among Slo1 channels from fruit fly to human (Figure 7C), suggesting that all BK channels may share the same Ca2+ activation mechanism. Interestingly, while the DRDD loop is highly conserved among the Ca2+ activated Slo1 channels, other Slo families that are activated by intracellular H+ (Schreiber et al., 1998), Na+ (Yuan et al., 2003) and/or Cl− (Yuan et al., 2000) ions show variations in this region (Figure 7C). In addition, this loop is not present in the prokaryotic MthK or the E. coli K+ channels (Jiang et al., 2002; Jiang et al., 2001). Therefore, the residues in the Slo1 DRDD loop appear to be specifically important for Ca2+ sensing. In contrast, the sequences of αA, αB and βC are highly conserved among all Slo families. The residues corresponding to F359, L387, F391, T396 and F400 in mSlo1 are hydrophobic in these channels, and the residues corresponding to D362 are either negatively charged or polar. These comparisons suggest a common role for αA, αB and βC in the gating mechanism of eukaryotic channels that are sensitive to intracellular ions. It is possible that in all these channels αA, αB and βC modulate gate opening, but the modulation is allosterically controlled by various intracellular ions that specifically bind to their respective binding sites. A change in this allosteric mechanism such as the hD434G mutation can alter the function of these channels and result in human diseases.
An increase of viscosity in intracellular solutions enhances Ca2+ sensitivity of BK channels, suggesting that the dynamics of the channel protein is important in the allosteric pathway associated with D367. Two lines of evidence indicate that mD369G alters the dynamics of the channel protein to potentiate Ca2+ sensing, namely mD369G, but not other mutations of D369, enhances Ca2+ sensitivity (Figure 3F) and alters the responses of the channel to changes in solution viscosity (Figure 4D). To explore the possible mechanisms of how mD369G changes channel function, we performed molecular dynamics (MD) simulations for the AC region. Previous studies have suggested that the AC region may form an independent structural and functional unit in BK channels. First, it was demonstrated that a cleavage at 12 amino acids after the AC region does not prevent the resulting two divided peptides from forming functional channels, suggesting that the cytosolic domain in the N-terminal part of the divided Slo1 may be able to fold independently (Pico, 2003). Therefore, as the core of this cytosolic domain, the AC region may be treated as a structural unit in MD simulations. Second, we previously found that the entire AC region, but not any part of it, was responsible for the Ca2+ sensitivity difference between two Slo1 homologs, suggesting that the AC region also acted as a functional unit to modulate Ca2+ sensitivity (Krishnamoorthy et al., 2005). This idea is also supported by the results that mutations within the AC region are allosterically connected in perturbing Ca2+ sensitivity (Figure 5).
Figure 8A plots root mean square fluctuation (RMSF) of the main chain Cα’s for the AC regions of the WT and mD369G mSlo1, as well as the AC regions of mutations D369E, A, N, W, and P that have little effect on Ca2+ sensitivity (Figure 3F). RMSF is the standard deviation of the movement of a residue around its mean position, i.e. residues with a higher RMSF are in a more flexible region of the protein. Consistent with the experimental data (Figures 1E, 3C, 3F, ,44 and 6A–D), mutation mD369G not only alters the local flexibility but also changes the overall dynamics of the AC region. The changes are most prominent in three areas of the AC region; the peak RMSF shifts in the DRDD loop and the maximum differences occur in the βA-αA linker and αB-βC (red, blue, and green shades, Figure 8A), corresponding to the same areas where the mutations reduce Ca2+ sensitivity with allosteric connections (Figure 5). Also correlating with the experimental results of channel activation (Figures 3F, 4E and 4F), the dynamics of the D369E, A, N, W, and P mutant AC regions do not show similar changes as that of mD369G. For a better visualization, the changes in dynamics caused by mD369G are also illustrated by decomposing the complex motion into simpler periodic orthogonal modes using principal component analysis and plotting the movements of the most significant mode (Figure 8B). The overall effect of mD369G is to reduce the flexibility in the βA-αA linker and αB-βC, causing the entire AC region to move as a more rigid entity that simply follows the motion of the DRDD loop (Figure 8B and Supplemental Movie S1), consistent with the result that both mD369G and an increase of solution viscosity enhances Ca2+ sensing (Figures 1 and and4).4). It is remarkable that the results of MD simulation and experiments correlate well in almost every aspect although the structure of the AC region is based on a homology model derived from the structure of MthK. This result is consistent with a number of previous studies that demonstrated the structural homology between the RCK1 domain in BK channels and the RCK domain of MthK (Cui et al., 2008; Wang and Sigworth, 2009). Thus, the MD simulation shows a plausible mechanism for mD369G to change the dynamics of the channel protein, which affects the allosteric coupling in the D367 activation pathway by altering the traverse of a dynamic substates ensemble or entropy in the free energy of channel gating (Kern and Zuiderweg, 2003; Popovych et al., 2006).
This dynamic based allosteric mechanism can be an important target for BK channel modulation. For instance, reported values of intracellular viscosity range from 1 mPa•s (the viscosity of pure water) to above a hundred mPa•s (Kuimova et al., 2009; Wandelt et al., 2005). Such a variation may be due to different samples and techniques of measurements as well as physiological states of the cell (Kuimova et al., 2009). Within this range of viscosity, Ca2+ sensitivity of WT BK channels increases with viscosity and can be equivalent or even larger than that of mD369G channels (Figure 4D). These results reveal a possibility that at certain physiological state, a high intracellular viscosity could be associated with epilepsy by altering the protein dynamics and potentiating Ca2+ sensitivity of BK channels. On the other hand, a reduction of intracellular viscosity reduces Ca2+ sensitivity of the mD369G mutant channels (Figure 4D), which may prevent the role of this mutation in epilepsy. These possibilities need to be explored by future investigations.
The mutations were made using overlap-extension PCR (polymerase chain reaction) from the mbr5 splice variant of mSlo1(GenBank accession number, GI: 347143) (Butler et al., 1993). The PCR-amplified regions of all the mutations were verified by sequencing. RNA was transcribed in vitro with T3 polymerase (Ambion, Austin, TX) and injected into Xenopus laevis oocytes (stage V-VI) with an amount of 0.05–50 ng each, followed by 2–5 days of incubation at 18 °C.
Inside-out patches were formed from oocyte membrane by borosilicate pipettes of 0.8–1.5 MΩ resistance. Macroscopic currents were recorded with an Axopatch 200-B patch clamp amplifier (Axon Instruments, Foster City, CA) and PULSE acquisition software (HEKA Electronik, Lambrecht, Germany). The current signals were low-pass-filtered at 10 kHz with the amplifier’s four-pole Bessel filter and digitized at 20-µs intervals. The pipette solution contains (in mM): 140 potassium methanesulphonic acid, 20 HEPES, 2 KCl, 2 MgCl2, pH 7.2. The nominal 0 µM [Ca2+]i solution contains (in mM): 140 potassium methanesulphonic acid, 20 HEPES, 2 KCl, 5 EGTA, 22 mg/L (+)-18-crown-6-tetracarboxylic acid (18C6TA), pH 7.2. The free [Ca2+] in the nominal 0 [Ca2+]i solution is about 0.5 nM. CaCl2 was added to a solution containing (in mM): 140 potassium methanesulphonic acid, 20 HEPES, 2 KCl, 1 EGTA, 22 mg/L 18C6TA, pH 7.2 to obtain the desired free [Ca2+], which was measured by a Ca2+ sensitive electrode (Thermo Electron, Beverly, MA). In experiments that changed viscosity, sucrose, glycerol, or urea was added to either the nominal 0 µM [Ca2+]i solution or the solution containing 200 µM free [Ca2+] (in mM): 140 potassium methanesulphonic acid, 20 HEPES, 2 KCl, 0.2 CaCl2, pH 7.2. 200 µM free [Ca2+] was used to ensure a saturating binding of Ca2+ to the channels. Viscosity was measured using a Brookfield DV-III ULTRA Programmable Rheometer (Brookfield Engineering Laboratories, Middleboro, MA) with a model 40 spindle. Viscosity was measured under two spindle rotation rates, 30 and 60 RPM, except for solutions with 2 M sucrose or 9 M glycerol, of which the viscosity was measured under 6 and 12 RPM. All the solutions are Newtonian fluids so that the viscosity does not change with spindle rotation rate. All the experiments were performed at room temperature (22–24 °C).
The tail current amplitudes at −50 mV were measured to determine the relative conductance. The conductance–voltage (G–V) curves were fitted with the Boltzmann function: , where G/G max is the ratio of conductance to maximal conductance, z is the number of equivalent charges, e is the elementary charge, V is membrane potential, V 1/2 is the voltage where G/G max reaches 0.5, k is Boltzmann’s constant, and T is absolute temperature. Error bars in this paper represent standard error of means (S.E.M.).
MWC model fits use the following equation:
Where Popen is channel’s open probability; L(0) is the equilibrium-state ratio of closed to open channels ([C]/[O]) in the absence of bound Ca2+ at 0 mV; z, e, V, k, and T are the same as above in Boltzmann function; Kc and Ko are the dissociation constants of Ca2+ in the closed and open states, respectively.
The limiting slope measurement is applied to determine the open probability at extremely negative voltages (Horrigan et al., 1999). The open probability determined by limiting slope measurement is combined with the corresponding G-V relation to construct a Po-V curve, which is fitted to the following HCA model (Horrigan et al., 1999):
Where zL is the charge associated with gate opening when all the voltage sensors are at their resting state. zJ is the charge associated with voltage sensor movements. L0 is the intrinsic open probability at V = 0 while all the voltage sensors are at their resting state. Vhc and Vho are the voltages for half of the voltage sensors to be at their activation state at the closed and the open conformations, respectively. All the fittings have fixed zL = 0.1, zJ = 0.57, and Vhc = 172 mV; L0 and Vho are optimized for the best fitting to reflect the Ca2+ dependence (Horrigan and Aldrich, 2002).
In Figure 2B and C the Po at −140 mV is obtained from the HCA model fittings, which is more accurate than that from direct measurements because the model fittings take into account all the Po-V data at a certain [Ca2+]i to eliminate experimental variations.
Sequence alignment of mSlo1 (GenBank accession number, GI: 347143) with MthK channel (1LNQ, GI: 2622639) and Kv1.2 channel (2A79, GI: 1235594) was performed using ClustalW (Thompson et al., 1994). The gaps in the sequence were predicted using the Protein Loop Optimization Algorithm (PLOP) (Jacobson et al., 2004). Homology models of the mSlo1 AC region (Gly341-Asp420) were generated using Modeller (Sali and Blundell, 1993).
The package GROMACS (Lindahl et al., 2001) was used for all MD simulations. The simulations were performed at 300K using an NpT ensemble, OPLS/AA force field and explicit SPC solvent. In all cases 2 fs time steps were used along with periodic boundary conditions, hydrogen bond constraints and Particle Mesh Ewald for the calculation of long-range electrostatics. The protocol for the simulation involved energy minimization, followed by heating to 300 K at intervals of 50 K, equilibration for 40 ps, and finally the 80 ns production run. The coordinates were saved every 50 ps for subsequent analysis. The 80 ns simulation is limited by computation capacities and is shorter than the time scale of BK channel gating that is at ~0.1 ms. Nevertheless, motions of the backbone on the ps–ns timescale have been shown experimentally as an important carrier of allosteric energy (Kern and Zuiderweg, 2003; Popovych et al., 2006).
The root mean squared fluctuation of every atom was found using the formula , where xi is the position of atom i in each simulation frame and xi is the average of x over all frames.
The trajectory obtained from our MD simulations was decomposed into orthogonal modes and the motion of the protein along the first few modes was studied. Cij, the correlation between the movement of an atom i and atom j was found for all N Cα atoms of the protein to form an N*N covariance matrix. , where M is a diagonal matrix containing the masses of the atoms. The covariance matrix was diagonalized to get N eigenvectors and corresponding eigenvalues. The MD trajectory was projected on the eigenvector having the largest eigenvalue to get the principal mode that captures the largest amplitude motion of the protein.
We thank Frank Horrigan, Toshi Hoshi, Urvi Lee, Lawrence Salkoff and Chris Lingle for helpful suggestions. We thank Dr. Jin-Yu Shao for allowing us to use the Rheometer. The mSlo1 clone was kindly provided by Lawrence Salkoff (Washington University, St. Louis, MO). This work was supported by National Institutes of Health Grant R01-HL70393 (J.C.). J.C. is an Associate Professor of Biomedical Engineering on the Spencer T. Olin Endowment.
G.K. and J.C. initiated the study. G.K., J.Y., G.Z. and J.C. made key experimental observations. A.S. and D.S. performed molecular modeling and MD simulations, which provided motivation for experiments in Figure 5. G.K., J.Y. and J.S. performed experiments in Figures 1, ,33 and and5.5. J.Y. and J.S. performed experiments in Figures 2, ,44 and and6.6. G.Z., H.Y. and K.D. also performed experiments in Figures 1 and and5.5. J.Y., G.K., A.S. and J.C. wrote the manuscript, with revisions from H.Y. and D.S‥