|Home | About | Journals | Submit | Contact Us | Français|
A family of tissue-specific auxiliary β subunits modulates large conductance voltage- and calcium-activated potassium (BK) channel gating properties to suit their diverse functions. Paradoxically, β subunits both promote BK channel activation through a stabilization of voltage sensor activation and reduce BK channel openings through an increased energetic barrier of the closed-to-open transition. The molecular determinants underlying β subunit function, including the dual gating effects, remain unknown. In this study, we report the first identification of a β1 functional domain consisting of Y74, S104, Y105, and I106 residues located in the extracellular loop of β1. These amino acids reside within two regions of highest conservation among related β1, β2, and β4 subunits. Analysis in the context of the Horrigan-Aldrich gating model revealed that this domain functions to both promote voltage sensor activation and also reduce intrinsic gating. Free energy calculations suggest that the dual effects of the β1 Y74 and S104–I106 domains can be largely accounted for by a relative destabilization of channels in open states that have few voltage sensors activated. These results suggest a unique and novel mechanism for β subunit modulation of voltage-gated potassium channels wherein interactions between extracellular β subunit residues with the external portions of the gate and voltage sensor regulate channel opening.
BK channels have an established role in regulating vascular smooth muscle tone by hyperpolarizing the membrane and deactivating voltage-dependent Ca2+ channels (Kaczorowski et al., 1996; Gribkoff et al., 1997; Calderone, 2002). Increased opening of smooth muscle BK channels is conferred by the β1 auxiliary subunit (Tanaka et al., 1997; Brenner et al., 2000b; Plüger et al., 2000). The important role of the β1 subunit has been demonstrated in β1 knockout mice, which display reduced BK channel opening, increased vascular tone, and hypertension (Brenner et al., 2000b; Plüger et al., 2000). In addition, two human β1 polymorphisms, each with a single amino acid change in the extracellular domain, have been associated with altered smooth muscle function. A gain-of-function polymorphism has been linked to a reduced incidence of hypertension (Fernández-Fernández et al., 2004). In addition, a polymorphism that moderately reduces channel opening has been associated with increased asthma severity (Seibold et al., 2008).
The BK channel pore-forming α subunit belongs to the six-transmembrane (TM) voltage-dependent K+ channel family (Atkinson et al., 1991). The β1 accessory subunit belongs to the two-TM BK channel auxiliary β subunit family consisting of four members, β1–β4, each having distinct tissue-specific expression (Knaus et al., 1994; Riazi et al., 1999; Wallner et al., 1999; Xia et al., 1999; Behrens et al., 2000; Brenner et al., 2000a). The β subunit structure contains a large extracellular loop, two TMs, and two small intracellular domains (Knaus et al., 1994; Orio et al., 2002). Recently, a distinct single TM protein, LRRC26 (leucine-rich repeat–containing protein 26), was also found to act as an accessory BK channel subunit in a prostate cancer cell line (Yan and Aldrich, 2010).
Mutagenesis of the α subunit has yielded a large number of insights into the structure of the pore-forming subunit of the BK channel (Lee and Cui, 2010). In contrast, the β structural determinants that modulate α subunit gating remain unclear. BK channel β subunits are apparently unrelated to other protein families. Therefore, identifying structural domains of β subunits by scanning mutagenesis is encumbered by the potentially large number of mutations that would be required. However, there is evidence that some modulatory effects of β subunit family members β1, β2, and β4 are conserved. These β subunits slow activation and deactivation gating kinetics (Behrens et al., 2000; Brenner et al., 2000a; Lippiat et al., 2003). In addition, these β subunits exert similar Ca2+-dependent effects on steady-state opening (Behrens et al., 2000; Brenner et al., 2000a; Lippiat et al., 2003). They increase channel opening in high Ca2+, effects that are accounted for by β subunit modulation of voltage sensor and Ca2+ binding (Bao and Cox, 2005; Orio and Latorre, 2005; Wang and Brenner, 2006; Wang et al., 2006; Sweet and Cox, 2009). β subunits also reduce channel opening in low intracellular Ca2+ by a reduction of intrinsic gating (channel opening independent of voltage sensor activation and Ca2+ binding; Orio and Latorre, 2005; Wang and Brenner, 2006; Wang et al., 2006). These functional similarities suggest that structural determinants underlying β subunit function may also be conserved (Orio et al., 2006). Therefore, key residues and functional domains may be uncovered by alanine substitution of conserved amino acids. Using this approach, we report here the identification of a novel β1 extracellular domain that is critical for modulation of voltage sensor activation and intrinsic gating of BK channels.
The mouse BK α cDNA (GenBank/EMBL/DDBJ accession no. MMU09383) was modified to include the extended amino-terminal sequence (beginning MANG) encoded by the KCNMA1 gene. We found that the extra sequence causes an ~20-mV larger negative G-V shift by mouse β1 as compared with the truncated α subunit (initiating translation at the internal MDAL residues) that has been most often used in the past. The extended amino-terminal sequence was also added to the F315Y construct (Wang and Brenner, 2006), which we call F380Y in this study. With mouse β1 cDNA (Wang and Brenner, 2006) as a template, mutant β1 constructs were generated with a Quick-Change XL Site-Directed Mutagenesis kit (Agilent Technologies) and confirmed by sequencing.
α and various β1 cDNAs were cotransfected into HEK-293 cells (American Type Culture Collection) and studied 1–2 d after transfection. β1 subunits were subcloned in the mammalian expression vector pIRES2–enhanced green fluorescent protein (Takara Bio Inc.), which contains the enhanced green fluorescent protein gene that fluorescently labels transfected cells. A molar ratio of 1:6 α/β1 was used, aiming to saturate BK channels with β1 subunits. For β1 mutants with large effects on G-Vs, saturating β1 expression was experimentally verified (Fig. S2). In the event β1 expression did not reach saturation, reduced α/β1 molar ratios (1:12 and 1:24) were used to reach saturation.
Currents were recorded using the patch clamp technique in the inside-out configuration. The external recording solution contained 20 mM HEPES, 140 mM KMeSO3, 2 mM KCl, and 2 mM MgCl2, pH 7.2. Internal solutions contained 20 mM HEPES, 140 mM KMeSO3, and 2 mM KCl, pH 7.2. For the 60-µM free Ca2+ intracellular solution, Ca2+ was buffered with 5 mM nitrilo-triacetic acid. For nominally 0 Ca2+ (0.002 µM of free Ca2+), intracellular solution Ca2+ was buffered with 2 mM EGTA. Free [Ca2+] was measured using a Ca2+-sensitive electrode (Orion Research).
Open probability (Po) was estimated by steady-state macroscopic recordings (when Po > 0.05) and single-channel recordings in the same patch. nPo was determined from all-points amplitude histograms by the sum of open levels (k) multiplied by fractional time spent (Pk): nPo = ΣkPk. To estimate the number of channels in a patch (n), maximum macroscopic conductance (GKMax) was divided by single-channel conductance (gk) at the same voltage for tail current measurements (−80 mV). n = GKMax/gk. The steady-state data in 0 Ca2+ were fit to the Horrigan-Aldrich model (Horrigan and Aldrich, 2002) based on least-squares criteria. The following equations were used to estimate energetic changes associated with mutations (Ma et al., 2006):
Fig. S1 shows that β1 and β4 share common gating effects. Fig. S2 shows effects of alanine substitutions of four β1 residues on G-V relations at various α/β1 molar ratios. Fig. S3 compares mutant Po-V data in linear scale to best fits of the Horrigan-Aldrich model (see Table 2). Fig. S4 shows effects of I106A, Y74A, and Y105AY74A on free energies associated with various gating transitions. Online supplemental material is available at http://www.jgp.org/cgi/content/full/jgp.201110698/DC1.
Despite different physiological roles, both β1 and neuron-specific β4 subunits slow BK channel gating and modulate steady-state properties (Fig. S1, A and B). Compared with α alone channels, both β1 and β4 increase steady-state Po at high Ca2+ but reduce it at low Ca2+ (Fig. S1 B). These steady-state effects are largely accounted for by two Ca2+-independent mechanisms. These are a reduction in intrinsic gating and a negative shift of open-channel voltage sensor activation (Fig. S1 C; Bao and Cox, 2005; Wang and Brenner, 2006; Wang et al., 2006; Sweet and Cox, 2009). Similarly, data from a previous study suggest that the dual gating mechanisms also underlie β2 modulation of BK channel properties (Orio and Latorre, 2005).
These prior findings led us to hypothesize that residues mediating gating modulation are conserved among β1, β2, and β4. To identify potential key gating residues and domains, we performed sequence alignment of mβ1, hβ1, bβ1, hβ2, mβ4, and hβ4 and identified two highly conserved segments (Fig. 1). Extracellular segment A consists of four identical residues (Q73YPC76), and segment B consists of five identical residues (C103SYIP107; Fig. 1). Segments A and B consist of the longest sequence of consecutive identical residues among these β subunits. To test the hypothesis that these conserved segments have important roles in gating modulation, we performed an alanine substitution mutagenesis of 13 identical residues within or neighboring segments A and B.
Mutant β1 (β1MT) subunits were expressed at saturating concentrations with wild-type α subunits using transient transfection in HEK-293 cells. BK currents were recorded using the inside-out patch clamp configuration at 60 µM Ca2+ (Fig. 2 A). Averaged G-V relations of αβ1MT were compared with α alone and αβ1WT channels (Fig. 2, B and C). Mutations such as V120A caused a small or no change in channel properties (Fig. 2, B and C). In contrast, mutations such as S104A reduced the size of the G-V shift (by ~30 mV; Fig. 2 B) and slowed the activation time constants (Fig. 2 C).
In addition to S104, we have identified three other positions that are important in mediating gating modulation. These include segment B residues Y105 and I106 as well as segment A residue Y74 (Fig. 3). Alanine substitutions of these residues reduce β1-mediated G-V shifts, increasing V1/2 by 28–44 mV (Fig. 3 A). These results are consistent with the hypothesis that segment A and B residues play an important role in gating modulation.
Similar effects of alanine substitution of Y74, S104, Y105, and I106 suggest that these residues may contribute to common functional interactions. To test this hypothesis, we examined whether these mutations have additive effects. If the two residues contribute to a common interaction, effects of the double mutation should not exceed both single mutations.
We first examined segment B mutations Y105A and I106A (Fig. 4 A). Indeed, the steady-state effect of the double mutant Y105AI106A (V1/2 of −48 ± 4 mV) is not significantly different from individual mutants Y105A (−45 ± 5 mV, P ≈ 0.6) or I106A (−49 ± 5 mV, P ≈ 0.9; Fig. 4 B, left). The nonadditive effects were also observed from activation kinetics. Compared with β1WT, both single mutations slow activation, with I106A having a greater effect than Y105A. This kinetic effect of Y105AI106A is not significantly greater than I106A (Fig. 4 B, right). At −20 mV, time constants for I106A and Y105AI106A are 10.1 ± 1.5 ms and 14.2 ± 2.0 ms, respectively (P ≈ 0.1). At −40 mV, they are 17.0 ± 1.3 ms and 21.4 ± 3.5 ms, respectively (P ≈ 0.3).
Next, we tested whether mutant effects of segment A and B residues Y74A and Y105A are additive (Fig. 4 C). The steady-state effect of Y105AY74A (V1/2 of −48 ± 3 mV) is not significantly different from Y105A (−45 ± 5 mV, P ≈ 0.6) even though it is greater than Y74A (−57 ± 3 mV, P ≈ 0.03; Fig. 4 D, left). Similarly, kinetic effects of the double mutation are not greater than those of Y105A (Fig. 4 D, right). At −20 mV, the activation time constant for Y105AY74A (7.2 ± 0.9 ms) is similar to Y105A (8.8 ± 1.0 ms, P ≈ 0.3) and Y74A (8.3 ± 0.7 ms, P ≈ 0.4). Finally, deactivation time constants are also not significantly different between Y105AY74A and Y105A. For example, at −200 mV, deactivation time constants are 1.3 ± 0.2 ms and 1.1 ± 0.1 ms, respectively (P ≈ 0.1).
The aforementioned results show that at 60 µM Ca2+, steady-state and kinetic effects of segment A and B mutations are nonadditive, suggesting that these residues functionally interact in modulating BK channel gating. The BK channel dual allosteric gating model includes a gate, four independent voltage sensors, and four Ca2+ sensors (Rothberg and Magleby, 2000; Horrigan and Aldrich, 2002). Channel opening is allosterically coupled to voltage sensor activation and Ca2+ binding. However, channels can open at a low probability independent of voltage sensor activation and Ca2+ binding (intrinsic gating). Previously, β1 subunits have been shown to confer modulatory effects on intrinsic gating, voltage sensor activation, and Ca2+ binding (Cox and Aldrich, 2000; Nimigean and Magleby, 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Wang and Brenner, 2006; Sweet and Cox, 2009). Because segments A and B are located in the extracellular region, we focused on how these domains modulate Ca2+-independent effects: intrinsic gating and voltage sensor activation.
Intrinsic gating is weakly voltage dependent and can be described by two free parameters (Horrigan and Aldrich, 2002). L0 represents the zero voltage value of the closed-to-open (C-O) equilibrium constant, and zL is its partial charge (Horrigan and Aldrich, 2002). To isolate mutant effects on intrinsic gating, we performed recordings using a modified α subunit F380Y. Previously, this modification has been shown to greatly increase channel opening in hslo1 (Lippiat et al., 2000) and mslo1 (Wang and Brenner, 2006). F380Y allows us to measure channel Po in ligand-unbound states (0 Ca2+) with voltage sensors residing in the resting state (the limiting slope). In αWT background, obtaining Po under these conditions is not technically feasible for the αβ1 channels (Po < 10−8; Wang and Brenner, 2006).
An example recording of α subunit F380Y in 0 Ca2+ over a range of voltages is shown in Fig. 5 A. The corresponding all-point histograms and estimated Po (Fig. 5 B) clearly show that the voltage dependence approaches a minimum between −120 and −220 mV. The estimated weak voltage dependence here (~0.25 e0) corresponds to the weak voltage dependence associated with intrinsic gating (zL; Horrigan and Aldrich, 2002). Fitting 0 Ca2+ limiting slope logPo-V relations to the Horrigan-Aldrich model (Horrigan and Aldrich, 2002), the two free parameters associated with intrinsic gating were estimated (L0 of 6.6 ± 0.9 e−2 and zL of 0.16 ± 0.01 e0; Fig. 6 A and Table 1). For αF380Y, the effect of β1WT on intrinsic gating is an ~15-fold reduction in L0 (4.3 ± 1.9 e−3, P < 0.001). However, zL is not significantly altered (0.20 ± 0.05 e0, P ≈ 0.5; Table 1).
L0 is significantly increased by segment A and B key mutations (Table 1). In the presence of Y105A, I106A, or Y74A, L0 are 1.9 ± 0.4, 4.4 ± 1.1, or 2.5 ± 0.5 e−2 (Table 1). The ~4-, 10-, and 6-fold increases of L0 (relative to αF380Yβ1WT channels) reflect a reduction in β1’s ability to decrease intrinsic gating. The results suggest an important role that Y105, I106, and Y74 play in β1 modulation of intrinsic gating.
Double mutation Y105AY74A was also examined to test whether effects of Y105A and Y74A on intrinsic gating were additive. If effects of Y105A and Y74A on intrinsic gating were additive, a significant increase in L0 would have been expected. Because the estimated L0 for αF380Yβ1Y105AY74A, 2.0 ± 0.3 e−2, is similar to the single mutations, the results suggest that segments A and B functionally interact in reducing intrinsic gating.
An increase in L0 predicts an increase in steady-state opening at all Ca2+. This is unlikely to be the sole effect of segment A and B mutants because the mutations displayed positive shifts of G-V relations. We therefore investigated whether these mutations also alter voltage sensor activation. Voltage sensor activation of unliganded channels is described by three free parameters: the partial charge associated with the resting-to-activated (R-A) transition (zJ) and the half-activation voltages for voltage sensors of open and closed channels (Vho and Vhc, respectively; Horrigan and Aldrich, 2002). The R-A equilibrium constant of closed, unliganded channel J0 (the zero voltage value of the R-A equilibrium constant) is a function of Vhc (J0 = 0.5/exp(zJVhc/kT)). The allosteric factor between the voltage sensor activation and gating (D) is a function of the difference between Vhc and Vho (D = 0.5/exp(zJ(Vhc − Vho)/kT); Horrigan and Aldrich, 2002).
We obtained 0 Ca2+ Po over a wide range of voltages. Averaged logPo-V and Po-V relations data were fit to the Horrigan-Aldrich model (Fig. 7 and Fig. S3; Horrigan and Aldrich, 2002). In all these fits, zL and zJ were set as 0.2 e0 and 0.58 e0, respectively, to reduce the number of free parameters. The basis for our assumptions here are twofold. First, the finding that voltage sensors act as a source of gating charge for the opening transition (Ma et al., 2006) and the observation that the mutations do not alter zL imply that these mutations do not significantly alter zJ as well. Second, prior gating current data found that neither zL nor zJ are altered by β1 (Bao and Cox, 2005), suggesting that mutations likely do not alter zL or zJ either.
The fits to the logPo-V relations (Table 2) show that β1 stabilizes voltage sensor activation in both open and closed channels. There is a −55-mV shift of the half-activating voltage for the open-channel voltage sensor (Vho), similar to the −61-mV value estimated in the αWT background (Bao and Cox, 2005). αF380Y reports a −37-mV shift of the half-activating voltage for the closed-channel voltage sensor (Vhc), which is smaller than the −71-mV value estimated in the αWT background (Bao and Cox, 2005).
Y105A has little effect on closed-channel voltage sensor activation. This is evident from estimates of Vhc (76 vs. 73 mV) and the corresponding R-A equilibrium J0 (8.8 vs. 9.4 e−2; Table 2). However, the mutation destabilizes open-channel voltage sensor activation (13 vs. −10 mV), thereby reducing the allosteric coupling between voltage sensor activation and gating (D of 3.9 vs. 7.1; Table 2). Similarly, mutating I106 and Y74 also destabilize open-channel voltage sensor activation, increasing Vho by 29 mV and 24 mV, respectively (Table 2). Finally, Y105AY74A increases Vho by 21 mV (Table 2), suggesting that effects of single mutations are nonadditive. Therefore, segments A and B may functionally interact in stabilizing voltage sensor activation in open channels.
The aforementioned results indicate that the segment A-B domain mediates two gating effects: reduction of intrinsic gating and stabilization of voltage sensor activation in open channels. The fact that a single functional domain underlies both gating effects suggests obligatory coupling of the two. We have calculated mutant effects on the free energy changes associated with the C-O and R-A transitions to more closely examine how the segment A-B domain exerts the dual gating effects.
Fig. 8 A plots the 10-state gating scheme for unliganded channels. Channels reside in either closed (C) or open (O) states with zero to four activated voltage sensors (Horrigan and Aldrich, 2002). The effects of segment A-B mutations on the steady-state equilibrium constant and free energy associated with each transition were plotted in Fig. 8 (B and C, respectively). The Y105A mutation has little effect (1.1-fold, 0.04 kcal/mol) on the R-A equilibrium of closed channels (Fig. 8, B and C). However, the mutation increases the energetic barrier (0.6-fold, 0.31 kcal/mol) of the R-A transitions of open states (Fig. 8, B and C). The largest effects are increases of the early C-O transitions in which no or few voltage sensors are active (i.e., 5.6-fold, −1.02 kcal/mol for C0-O0; Fig. 8, B and C). These effects are quantitatively similar in mutations I106A, Y74A, and Y74AY105A (Fig. S3 and summarized in Table 3). From the perspective of the wild-type β subunit, a simple model that explains these results is one in which segment A-B interactions cause a destabilization of early open states (O0 to O2) but have little effect on closed states.
Our scan identified four residues critical for β1 gating modulation. Mutating these residues reduce β1-mediated negative G-V shift at 60 Ca2+, consistent with a partial disruption in β1 function. The experimental evidence does not suggest gain-of-function mutant effects. For example, gating parameters normally not altered by β1WT (such as zL) are also not altered by these mutations, and gating parameters normally altered by β1WT (such as L0 and Vho) are reduced by these mutations.
Previous studies suggest that β1 alters several gating parameters of BK channels (Cox and Aldrich, 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Orio et al., 2006; Wang and Brenner, 2006; Sweet and Cox, 2009). Our observations that no single mutation eliminated β1-mediated negative G-V shift or free energy changes suggest that additional key gating residues remain to be uncovered. This is true even for the effects on intrinsic gating and open-channel voltage sensor activation that mutants did not completely eliminate. Most likely, multiple independent interactions mediate β1 gating effects. These include Ca2+-dependent effects that are difficult to gauge using αF380Y and therefore are not addressed in this study.
Our finding that mutations of β1 disrupt effects on intrinsic gating and voltage sensor activation is consistent with previous studies ascribing β1 modulation of BK channels to these two effects (Nimigean and Magleby, 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Wang and Brenner, 2006). Indeed, mutations of the BK channel voltage sensor have been suggested to specifically occlude the β1-mediated negative G-V shift (Yang et al., 2008). Based on the observation that single-residue mutations both reduce the energetic barrier for intrinsic opening and increase the energetic barrier for open-channel voltage sensor activation, a novel insight provided by the current study is that β1 modulation of these two aspects of gating is likely coupled. Because previous studies have identified single α mutations that alter both intrinsic opening and voltage sensor activation (Ma et al., 2006; Wang and Brenner, 2006), this coupling may reflect intrinsic properties of the pore-forming subunit rather than properties unique to β1 function.
The dual gating effects of the segment A-B domain likely contribute to the complex steady-state modulatory effects of β subunits (Fig. S1). Interactions involving the segment A-B domain transition the channel away from open states with relaxed voltage sensors (the OR state; Fig. 9). Thus, in the absence of calcium or membrane depolarization, β1 causes a relative destabilization of the OR state and “silences” the channel in the closed resting state (CR; Fig. 9). With sufficient depolarization and/or calcium, β1 promotes channel opening by facilitating transitions to the late opening states (the OA state; Fig. 9).
Our results suggest that the segment A-B domain is necessary for maintaining the regulatory effects of β1. The data do not exclude the possibility that this domain plays an indirect role, simply being required to maintain a particular β or α/β subunit structure necessary for β1 subunit gating effects. However, the fact that segments A and B are the most conserved domain led us to favor a simpler scenario that this domain mediates the modulatory effects directly.
Several studies suggest that the extracellular segments A and B may be positioned near the external vestibule of BK channels (Fig. 9). Extracellular residues of β subunits have been shown to affect charybdotoxin binding and instantaneous I-V relations (Hanner et al., 1998; Zeng et al., 2003; Chen et al., 2008). Interestingly, hβ2 lysine residues flanking segment B (Fig. 1) have been shown to confer outward rectification of BK currents (Chen et al., 2008). These results suggest close approximation between the outer vestibule of the channel and the β subunit extracellular domain (including the segment B region). Assuming the role of the segment A-B domain is direct, how do β subunit residues near the external mouth of BK channels modulate gating? There has been compelling evidence that suggests the selectivity filter is the activation gate for BK channels (Li and Aldrich, 2004; Piskorowski and Aldrich, 2006; Wilkens and Aldrich, 2006; Chen and Aldrich, 2011; Cox and Hoshi, 2011; Geng et al., 2011; Zhou et al., 2011). Proximity between the outer vestibule and the selectivity filter suggests that the segment A-B domain may be positioned near the gate. Our finding that the segment A and B mutations affect both voltage sensing and intrinsic gating presents the intriguing possibility that this domain may interact at a nexus for coupling open-channel voltage sensor to the selectivity filter/channel gate. Indeed, it will be interesting to determine whether physical interactions between the segment A-B domain and the pore-forming subunits are state dependent (i.e., interactions only occurring in open states; Fig. 9).
Alternatively, the segment A-B domain may interact with the pore-forming subunits indirectly, via other extracellular residues or other parts of the channel. Using chimera and deletion approaches, the intracellular domains of β1 have also been shown to be important for β1 function (Orio et al., 2006; Wang and Brenner, 2006). In addition, β1 TMs have been shown to lay interposed between voltage sensor domains of adjacent α subunits (Liu et al., 2010), suggesting a role these domains play in modulating voltage sensor activation. The relations among various β subunit domains and their relevant contributions remain an important question to be addressed in future studies.
In summary, this study identifies an important domain of β1 subunits that underlies, in part, modulation of BK/β1 channels. The reduced intrinsic opening conferred by the segment A-B domain and increased voltage sensor activation created a steeper activation response of BK channels associated with β1. The fact that human β1 polymorphisms with more moderate gating effects have been linked to decreased prevalence of hypertension and reduced pulmonary function (Fernández-Fernández et al., 2004; Seibold et al., 2008) highlights the potential physiological importance of the segment A-B domain in β1 and perhaps other β subunits.
We thank Robert Brenner, Vladislav Bugay, and Luke Whitmire for helpful discussions.
This work is supported by American Heart Association grant BGIA2390030 to B. Wang.
Kenton J. Swartz served as editor.