Single Channel Recordings Reveal Differential β2 Subunit Modulations Between Mammalian and Drosophila BKCa(β2) Channels

Large-conductance Ca2+- and voltage-activated potassium (BK) channels are widely expressed in tissues. As a voltage and calcium sensor, BK channels play significant roles in regulating the action potential frequency, neurotransmitter release, and smooth muscle contraction. After associating with the auxiliary β2 subunit, mammalian BK(β2) channels (mouse or human Slo1/β2) exhibit enhanced activation and complete inactivation. However, how the β2 subunit modulates the Drosophila Slo1 channel remains elusive. In this study, by comparing the different functional effects on heterogeneous BK(β2) channel, we found that Drosophila Slo1/β2 channel exhibits “paralyzed”-like and incomplete inactivation as well as slow activation. Further, we determined three different modulations between mammalian and Drosophila BK(β2) channels: 1) dSlo1/β2 doesn’t have complete inactivation. 2) β2(K33,R34,K35) delays the dSlo1/Δ3-β2 channel activation. 3) dSlo1/β2 channel has enhanced pre-inactivation than mSlo1/β2 channel. The results in our study provide insights into the different modulations of β2 subunit between mammalian and Drosophila Slo1/β2 channels and structural basis underlie the activation and pre-inactivation of other BK(β) complexes.


Introduction
Large-conductance Ca 2+ -and voltage-activated potassium (MaxiK or BK) channels are widely distributed in mammalian tissues [1]. As a voltage and calcium sensor [2], the open probability of BK channel increases with both membrane depolarization and increased intracellular calcium concentration [3][4][5][6][7], thereby playing a critical role in modulating many significant study provided insights into the different modulations of β2 subunit between mammalian and Drosophila Slo1/β2 channels and structural basis underlie the activation and pre-inactivation of other BK(β) complexes.

dSlo1/β2 channel doesn't have complete inactivation
When studying the modulation of Slo1 by β2 subunit, it is important to saturate the β2 expression to get stable results. Our previous study has demonstrated that in human embryonic kidney (HEK) 293 cell expression system, the α:β2 transfection ratio at 1:0.2 was good enough to maintain saturate β2 binding for every mSlo1/β2 channel. For instance, when the transfection ratio of α:β2 was 1:0.2, all of the random tests of mSlo1/β2 channel currents exhibited rapid and complete inactivation, which indicated that the binding rate between α and β2 was about 100% (Fig 1A and 1B), and at the same 1:0.2 ratio in the mSlo1/Δ3-β2 channel (the inactivation ball FIW was deleted), the V 50 of all the random tests had obvious left shifts to −30 mV from approximately +25 mV (mSlo1) in 10 μM Ca 2+ , which suggested that the binding rate between α and β2 was also approximately 100% ( Fig 1B) [49,50]. hSlo1/β2 and hSlo1/Δ3-β2 shared the same results ( Fig 1B). However, in the dSlo1/β2 channel, when the transfection ratio of α:β2 was 1:0.2, only a few (about 7%) random tests exhibited inactivation or brief openings (Fig 1B), thereby indicating that the binding rate between dSlo1 and β2 was much lower than that in the mSlo1/β2 channel.
To test this hypothesis, we increased the β2 transfection amount (α:β2) to 1:1, 1:4, and 1:8. As the number of transfected β2 subunits increased, we obtained more examples with inactivation or brief openings, and when the transfection ratio of dSlo1:β2 was 1:4, the β2 binding ratio was almost saturated at 85%. However, for dSlo1:Δ3-β2, all of the random tests exhibited slow activation even when the transfection ratio was 1:0.2, which suggested that the β2 binding rate was again about 100% ( Fig 1B). These results indicate that the inactivation ball FIW in the dSlo1/β2 channel might serve as a retention signal that could block the binding of β2. All statistical results were listed in Fig 1B. We further confirmed the electrophysiology results by immunofluorescence imaging. Green fluorescent protein (GFP, green) and Myc (EQKLISEEDL, red) were inserted into the C-terminal of dSlo1 and the extracellular loop of β2 subunit, respectively, as fluorescent tags to facilitate our fluorescent analyses. The fused fluorescent proteins did not alter the gating properties of the wild-type (WT) channels (S1 Fig). According to the results of electrophysiological experiments, we selected two representative dSlo1-GFP:β2-72Myc transfection ratios of 1:1 and 1:4, and measured the membrane fluorescence intensities of β2-72Myc (Fig 1B and 1C). After normalizing the fluorescent intensity of dSlo1-GFP/β2-72Myc [1:4] against the electrophysiology result for dSlo1/β2 [1:4], there was no significant difference between the fluorescent intensity of dSlo1-GFP/β2-72Myc [1:1] and the electrophysiology result for dSlo1/β2 [1:1] (Fig 1B). These results suggested that the FIW was a retention signal in the Drosophila Slo1/β2 channel, but not in mammalian Slo1/β2 channels. β2(K33E,R34D,K35E) generated slow activation in dSlo1/Δ3-β2 channel, but not in mSlo1/Δ3-β2 channels It was interesting to notice that compared with WT dSlo1 channel currents, the currents of dSlo1/β2 channel was much smaller in amplitude and filled with "paralyzed"-like brief openings ( Fig 1A). After deleting the inactivation ball FIW of β2 subunit, the dSlo1/Δ3-β2 channel exhibited dramatically slower activation (Fig 1A). To elucidate the underlying behavior of the Macroscopic currents of mSlo1, mSlo1/β2, mSlo1/Δ3-β2, dSlo1, dSlo1/β2, and dSlo1/Δ3-β2 obtained from inside-out patches in the presence of 10 μM Ca 2+ according to the protocol indicated. Scale bars represent 50 ms and 2 nA for mSlo1/x channels, and 200 ms and 0.5 nA for dSlo1/x channels. The α:β transfection ratio was 1:8 for dSlo1/ β2, and 1:0.2 for other combinations.
β2 Modulations on mSlo1 and dSlo1 channels, we performed single channel recordings of different mSlo1 and dSlo1 channel combinations (Fig 2). The currents were recorded with a Ca 2+ concentration of 10 μM in inside-out mode at +100 mV as indicated. A -180 mV (100 ms) pre-pulse was used to eliminate the possibility of inactivation [49]. Fig 2B showed the single-channel recordings of dSlo1/β2. Unlike the mSlo1/β2 channel, which exhibited fast and complete inactivation (τ i = 15.1 ms), mainly long openings and rare brief openings (Fig 2A " Ã "), dSlo1/β2 exhibited slower inactivation (τ i = 166.2 ms) and much more brief openings, as well as a detectable steady-state current after an average of 100 consecutive single-channel traces, which were consistent with the macroscopic currents shown in Fig 1A. However, after the application of 0.2 mg/ml trypsin [47], the brief openings and inactivation were cut off gradually, whereas the current amplitude and opening time duration were greatly increased, thereby exhibiting similar characteristics to the dSlo1 alone channel (Fig 2A and 2B). This result indicated that the "paralyzed"-like brief openings came from the N-terminal of β2 since trypsin could digest the N-terminal of β2 [47,49].
Then we confirmed the slow activation of dSlo1/Δ3-β2 at single channel level. The dSlo1/ Δ3-β2 single channel currents exhibited obvious hysteresis in terms of the activation rate (τ a = 1446.3 ms) and the burst openings made the channel open in an unusual manner so it appeared "paralyzed" as well ( Fig 2C). After applying 0.2 mg/ml trypsin, the burst openings and slow activation were also eliminated gradually, and the current eventually appeared to be more similar to that of dSlo1 alone (Fig 2A and 2C). These results suggested that the slow activation might also have been attributable to the N-terminal of β2.
To precisely determine the key residues for the slow activation, we systematically shortened the length of β2 N-terminal and then tested their effects on the slow activation (Fig 3). The macroscopic currents of dSlo1 with different β2 N-terminus lengths were shown in Fig 3A. After truncating the first 3 or 30 amino acids from the N-terminus of β2, dSlo1/Δ3-β2 and dSlo1/Δ30-β2 both clearly exhibited slow activation, but when the first 35 amino acids were cut off, the dSlo1/Δ35-β2 activation time constant was obviously accelerated to the level of dSlo1 channel activation (Fig 3A and 3B), which suggested that the amino acids β2(D32,K33,R34, K35,T36) located at positions 32-36 of the N-terminus of β2 played a key role in generating the slow activation. The three positively charged amino acids β2(K33,R34,K35) were mainly considered. When they were mutated to negative charges, dSlo1/Δ30-β2(K33E,R34D,K35E) exhibited the same activation rate as the dSlo1/Δ35-β2 channel (Fig 3A and 3B), thereby indicating that β2(K33,R34,K35) was responsible for the slow activation. To exclude the possibility that Δ30-β2(K33E,R34D,K35E) did not express, we tested the dSlo1/β2(K33E,R34D,K35E) currents. Although we didn't see clear inactivation, the currents also showed "paralyzed"-like brief openings and the amplitude was much smaller than the dSlo1 control currents from the same bench of cells (S2 Fig), which demonstrated that β2(K33E,R34D,K35E) did express well.
Single channel analysis of dSlo1/Δ3-β2 burst openings indicated that even if the inactivation ball FIW was deleted from β2, the remaining inactivation chain could still enter the channel pore to block the channel opening by half ( Fig 3C). When the first 30 amino acids were cut The voltage ranges of the test potential were from -160 mV to +160 mV for mSlo1, mSlo1/β2, and mSlo1/Δ3-β2, and from -160 mV to 140 mV for dSlo1, dSlo1/β2 and dSlo1/Δ3-β2. (B) β2 binding percentages of a series of α/β2 channels at different α:β transfection ratios. Gray bars were obtained from patch clamp recordings and the red bars from immunofluorescence imaging. The mSlo1/β2 channel currents were analyzed based on patches with entire inactivation, mSlo1/Δ3-β2 based on the G-V relationship left shift, dSlo1/β2 based on inactivation or briefly opening recordings, and dSlo1/Δ3-β2 based on slow activation. dSlo1-GFP/β2-72Myc at [1:1] (N = 60) and at [1:4] (N = 61) were analyzed based on surface immunofluorescence images of β2-72Myc and then normalized against the dSlo1/β2 [1:4] results obtained from patch clamp recordings. The error bars represent the standard error of the mean (SEM). (C) Immunofluorescence imaging of dSlo1-GFP/β2-72Myc at different α:β transfection ratios of 1:1 (left) and 1:4 (right). Scale bar represents 10 μm.
doi:10.1371/journal.pone.0163308.g001 from the N-terminal of β2, the chain length might not be sufficiently long to enter the pore, so the single channel recordings indicated long openings as normal dSlo1 but the slow activation remained (Fig 3D) since the β2(K33E,R34D,K35E) was still working.
A comprehensive macroscopic activation or inactivation currents were also performed to confirm the time constants of the averaged single channel recordings (S3 Fig).

Enhanced pre-inactivation in dSlo1/β2 channel
In previous studies, we demonstrated the two-step inactivation process for the mSlo1/β2 channel: C$O$O Ã $I [49]. During inactivation, the inactivation ball FIW swung randomly with the inactivation chain [45]. After the channel opened, two negatively charged residues located in the inactivation chain at β2(D16,E17) and two positively charged residues located in the intracellular mouth of the pore at mSlo1(K330,K331) would combine to form a pre-  inactivation site (PI site, O Ã state), after which the inactivation ball swung rapidly into the channel pore to block the conducting pathway and inactivate the channel. Hence, according to the opening time duration during the single channel behaviors, there were two distinct open states, i.e., relatively long opening (O state) and short-lived brief opening (O Ã state) [49]. In wild-type mSlo1/β2 channel, this brief opening happened rarely [47,49]. Occasionally, however, some brief openings were evident in our single channel recordings containing several mSlo1/β2 channels (Fig 4A, " Ã "). A mutation β2(W4E) was used to test the existence of the pre-inactivated state (O Ã ) of mSlo1/β2 based on the enhanced probability of brief openings in the steady-state compared with mSlo1/β2 (Fig 4A and 4B) [47,49]. And after we destroyed the PI site by mutating the two negatively charged residues on N terminal β2 to positively charged residues, mSlo1/β2(W4E,D16R,E17K) displayed only long openings ( Fig 4C).
Thus, we hypothesized that the brief openings seen in the single channel currents of dSlo1/ β2, which were similar to the mSlo1/β2(W4E) channel currents, were attributable to the preinactivation of the channels. Destruction of the PI site by trypsin could eliminate the brief openings of the dSlo1/β2 channel supported this idea (Fig 2B), and again, we performed single channel recordings of dSlo1/β2(D16R,E17K), which would destroy the PI site as well. The currents showed only long openings similar as dSlo1 channel currents (Figs 2 and 4E) as predicted. To confirm that β2(D16R,E17K) was expressed in the single channel currents, we further tested immunofluorescence imaging of dSlo1-GFP/β2(D16R,E17K)-72Myc (Fig 4F). Taken together, our results demonstrated that the pre-inactivation property was not only conserved in dSlo1/ β2 channel, but also enhanced without the enhancement of β2(W4E) (Figs 2B and 4D).

Discussion
The β2 subunit diversifies the properties of the Slo1 α subunit, which raises many interesting and significant questions. Comparing the functional differences between homologous BK channels is a simple method for studying the BK channel gating, βx modulation, and pharmacological properties of the channel [24,48,52]. Using the xSlo1/β2 channel as a tool, in this study, we discovered three interesting functional differences regulated by β2 subunit between mammalian (mouse or human) and Drosophila Slo1/β2 channels.
Unlike the β1 subunit, which could be expressed freely on the membrane [53], a four-turn α-helical segment in the N-terminal of β2 subunit was shown to serve as a retention signal that could prevent surface expression of the β2 subunit [54]. When we doubled the inactivation ball FIW (double FIW or dFIW), "dFIW-β2" was retained firmly in the endoplasmic reticulum even when it was co-expressed with the mSlo1 α subunit and the BK currents were eliminated from the plasma membrane, which suggests that dFIW was an enhanced retention signal for the mSlo1/β2 complex [49]. However, for dSlo1/β2 channel, it appeared that FIW itself was a retention signal and dSlo1 and β2 binding (or interactions) on the plasma membrane could be adjusted by changing the amount of transfected β2 (Fig 1B and 1C). The detailed reason might also be related to the hydrophobicity of FIW.
The slow activation came from the three positively charged residues β2(K33,R34,K35) resided in the N terminal of β2 subunit (Figs 2 and 3). However, understanding why the lack of FIW dSlo1/Δ3-β2 currents led to a dramatically slower activation rate requires further research. It is interesting that a BK channel can be activated so slowly at a Ca 2+ concentration of 10 μM and dSlo1/Δ30-β2(K33,R34,K35) (127.92 ± 26.22 ms). Traces were fitted with a single exponential equation. (C-D) Single channel recordings and current histograms for the dSlo1/Δ3-β2 and dSlo1/Δ30-β2 channels. The currents between the blue lines were the raw data used for histogram analysis and the histogram were fitted with triple or double Gaussian equations.
doi:10.1371/journal.pone.0163308.g003 (τ a~1 s; Figs 2 and 3), and this may provide a more distinctive model for studying the Ca 2+ and voltage gating processes in the BK channel. A more precise interpretation of these behaviors may depend on the structure of the BK(β2) channel complex. For the enhanced pre-inactivation of dSlo1/β2 channel, it was reasonable to believe that the enhancement might be attributed to the different C-linker and AC regions between mammalian and Drosophila Slo1 channel. Based on the crystal structures of the BK channel gating ring (RCK domains) in both the open and closed states, as well as the recently reported cryo-electron microscopy structure of the BK(Slo2.2) channel, the C-linker and AC region are located closest to the transmembrane domain [33,[55][56][57] (S4 Fig). As a unique part of the Ca 2+ gating pathway, because they are located between the Ca 2+ binding RCK domain and channel pore opening domain, the Clinker and AC region are inevitably responsible for transferring the Ca 2+ gating force from the RCK domain to the channel pore [16,24,25]. Regarding the BK channel microstructure, any differences in the amino acid sequences will lead to variations in structure and thus the channel functions, especially when the different residues are charged or hydrophobic amino acids because this may allow new possibilities for the formation of electrostatic or hydrophobic interactions. Thus, multiple BK channel functions could be affected or even changed by electrostatic or hydrophobic interactions, such as the voltage dependence [58,59], single-channel conductance [60,61], gating dynamics [62][63][64][65], rectification characteristics [60,66,67], drug sensitivity [67][68][69], inactivation property [49], and membrane expression [49,54].
The pre-inactivation site was formed by the interaction between mSlo1(K330,K331) and β2 (D16,E17). In mammalian mSlo1 or hSlo1, there was two positively charged lysine residues at positions (K330,K331). By contrast, the corresponding position of Drosophila Slo1, dSlo1 (N327,K328), had only one positively charged residue. The single "arm" at dSlo1(N327,K328) may lack the ability to capture β2(D16,E17). Instead, β2(D16,E17) might interact with other positively charged residues within C-linker and AC region such as dSlo1(K334,R335) or dSlo1 (R377,K378) to form a "misplaced" PI site, thereby resulting in "paralyzed" inactivation (enhanced pre-inactivation) at either the single channel or macro-current level (Figs 1, 2 and  3). The "paralyzed" single channel recordings obtained for dSlo1/β2 also indicated that the pre-inactivation state (O Ã ) was enhanced without the help of β2(W4E) (Fig 4), so FIW might not interact as strongly with the pore as that found in the mammalian mSlo1 and the hSlo1 channel. Therefore, the mammalian (mouse and human) Slo1 channels and Drosophila Slo1 could have different pore structures. When FIW was removed from β2, the remaining inactivation chain could still enter the pore to reduce the channel opening by half (Fig 3). This model can be used as a tool to study the pore structure or size of the BK channel, as well as the interaction between the β2 N-terminal and pore region [46].Finally, we made a cartoon schematic to clarify the three different modulations of β2 subunit between mammalian and Drosophila BK Ca (β2) channels (Fig 5). Gray and orange bars indicated dSlo1 or mSlo1 and β2 subunit, respectively. Blue, orange, and pink circles indicated the α and β2 binding or interactions, β2 (K33,R34,K35), and β2(D16,E17) or pre-inactivation, respectively. Firstly, in dSlo1/β2 channel, the inactivation ball FIW is not enough for complete inactivation of dSlo1, and resulted in a much lower binding rate than mSlo1/β2 channel at the same α:β2 transfection ratio (Fig 1). Secondly, β2(K33,R34,K35) delayed the channel activation (pore opening, or voltage sensor movement, or the VSD-PD coupling), the channel currents exhibited obvious hysteresis in terms of the activation rate (Figs 1, 2 and 3). Thirdly, dSlo1/β2 channel had enhanced pre-inactivation than mSlo1/β2 channel without the help of β2(W4E) (Fig 4). The results in our study provided insights into the different modulations of β2 subunit between mammalian and Drosophila Slo1/β2 channels and structural basis underlie the activation and pre-inactivation of other BK(β) complexes. Gray and orange bars indicated dSlo1 or mSlo1 and β2 subunit, respectively. Blue, orange, and pink circles indicated the α and β2 binding or interactions, β2(K33, R34,K35), and β2(D16,E17) or pre-inactivation, respectively. The black circle indicated the enhanced pre-inactivation in dSlo1/β2 channel. Firstly, in dSlo1/β2 channel, the inactivation ball FIW is not enough for the complete inactivation, and may suppress the α and β2 binding or interactions. Secondly, β2(K33,R34,K35) delays the pore opening (or voltage sensor movement). Thirdly, dSlo1/β2 channel has enhanced pre-inactivation than mSlo1/β2 channel. doi:10.1371/journal.pone.0163308.g005

Constructs and mutations
To construct dSlo1-EGFP, EGFP in pEGFP-N1 (Clontech) was cut by two restriction enzymes and subcloned into pcDNA3.1(+) (Invitrogen). Next, dSlo1 was fused in-frame to the N-terminus of EGFP using appropriate restriction enzymes. The hβ2-72Myc construct was created by adding the sequence of Myc (EQKLISEEDL) between positions 72 and 73 in hβ2 via sequential overlap extension PCR and it was then subcloned into pcDNA3.1. Truncations of the β2 subunit, i.e., ΔFIW-hβ2 and Δ30-hβ2, were generated by removing amino acids from positions 2-4 and from positions 2-31, respectively. The mutations were introduced using a QuickChange Site-Directed Mutagenesis Kit (Stratagene). All of the constructs and point mutations were verified by direct DNA sequence analysis.

Immunofluorescence and confocal microscopy
At about 24 h after transfection, cells were fixed with 2% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min. After blocking with 2% bovine serum albumin (BSA) for 30 min, cells were incubated with mouse monoclonal anti-human Myc antibody (Abcam) at a dilution of 1:300 in 1% BSA (Abcam) for 3 h at room temperature (RT). The cells were then washed with PBS and incubated with rhodamine-conjugated goat anti-mouse IgG (H + L) (1:300) (Proteintech) for 1.5 h at RT. Nonspecific secondary antibodies were removed by washing with PBS, followed by a final soak in PBS. Cells were visualized by confocal laser scanning microscopy (Olympus IX71) using a 100× oil immersion lens (NA1.30). Parameter selection, sample scanning, and image acquisition were all controlled using Andor IQ 2 software.

Image analysis and statistics
In experiments with coexpressed dSlo1-EGFP (green) and hβ2-72Myc (red), four cases were observed: no fluorescence, red fringes, green cytoplasm, and green cytoplasm with red fringes. In the statistical analysis, we ignored cells with no fluorescence or those only with red fringes due to the non-expression or expression of only hβ2-72Myc in HEK293 cells. The fluorescent intensity of the red fringes was calculated using ImageJ 1.45 (Wayne Rasband, National Institutes of Health) and normalized against the intensity of the control group. The representative images shown in the figures were obtained with AutoQuantX2. The error bars represent the standard error of the mean (SEM).

Cell culture and transient transfection in HEK293 cells
HEK293 cells were cultured in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% fetal bovine serum (Gibco) and 100 U/ml penicillin and streptomycin in incubators at 37°C under 5% CO 2 . Cells were transferred to 24-well plates before transfection and then cotransfected with α and β subunits, and the reporter gene EGFP at a molar ratio of 1:4:1 for 4-6 h, and then transiently transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Cells were transferred to poly-D-lysine (Sigma) coated chambers for imaging or to prepare slides for electrophysiological experiments.

Patch clamp recording
All of the experiments were performed using excised patches in an inside-out recording configuration. Transfected HEK293 cells were prepared at 24 h after transfection in 160 K + solution containing (in mM): 160 MeSO 3 K, 2 MgCl 2 , 10 HEPES (pH 7.0). Patch pipettes were pulled from borosilicate glass capillaries with resistances of 2-4 megohms when filled with the pipette solution. Experiments were performed using a PC2C patch-clamp amplifier and its related software (InBio China), where the currents were typically digitized at 100 kHz and filtered at 5 kHz. For single channel recordings, experiments were performed using an EPC-9 patch-clamp amplifier and its related software (HEKA, Germany). The currents were digitized at 50 kHz and filtered at 15.7 kHz to reduce the impact of the filter settings [70]. During inside-out recordings, solutions with different Ca 2+ concentrations were applied to the membrane patches using a perfusion pipette containing eight solution channels. All of the experiments were performed at RT (22-24°C).

Data analysis
Macropatch recordings were analyzed using IGOR (Wavemetrics, Lake Oswego, OR), Clampfit (Axon Instrument, Inc.), and Sigmaplot (SPSS. Inc.). Unless stated otherwise, the data are expressed as the mean ± standard deviation (SD). The activation and inactivation time constants of the currents were fitted by a single exponential equation: I = a Ã exp(-(t/τ)) + c, where τ is the time constant, a is the current amplitude, and c is the current offset constant. Singlechannel recording data were analyzed using QuB (SUNY at Buffalo). Histograms were fitted by double or triple Gaussian equations: f = a Ã exp(-(x-b)^2/c) + e Ã exp(-(x-g)^2/h) + i Ã exp(-(x-j) 2/k), where a, e and i are count numbers, b, g and j are means, c, h and k are deviations.