Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Gi/o-coupled muscarinic receptors co-localize with GIRK channel for efficient channel activation

  • Michihiro Tateyama ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing

    tateyama@nips.ac.jp

    Affiliations Division of Biophysics and Neurobiology, Department of Molecular and Cellular Physiology, National Institute for Physiological Sciences, Okazaki, Japan, Department of Physiological Sciences, School of Life Science, SOKENDAI (The Graduate University for Advanced Studies), Hayama, Japan

    ⨯
  • Yoshihiro Kubo

    Roles Conceptualization, Supervision, Writing – review & editing

    Affiliations Division of Biophysics and Neurobiology, Department of Molecular and Cellular Physiology, National Institute for Physiological Sciences, Okazaki, Japan, Department of Physiological Sciences, School of Life Science, SOKENDAI (The Graduate University for Advanced Studies), Hayama, Japan

    ⨯

Abstract

G protein-gated inwardly rectifying K+ (GIRK) channel regulates cellular excitability upon activation of Gi/o-coupled receptors. In Gi/o-coupled muscarinic M2R, the intracellular third loop (i3) is known as a key domain for Gi/o coupling, because replacement of i3 of Gq-coupled muscarinic M1R with that of M2R enables the chimeric receptor (MC9) to activate the GIRK channel. In the present study, we showed that MC9, but not M1R, co-localizes with the GIRK channel and Gαi1 by Förster resonance energy transfer (FRET) analysis. When M1R was forced to stay adjacent to the channel through ligation with short linkers, M1R activated the GIRK channel. FRET analysis further suggested that the efficacy of channel activation is correlated with the linker length between M1R and the GIRK channel. The results show that co-localization is an important factor for activating the GIRK channel. In contrast, for MC9 and M2R, the GIRK channel was activated even when they were connected by long linkers, suggesting the formation of a molecular complex even in the absence of a linker. We also observed that replacement of 13 amino acid residues at the N-terminal end of i3 of MC9 with those of M1R impaired the co-localization with the GIRK channel as well as channel activation. These results show that localization of the receptor near the GIRK channel is a key factor in efficiently activating the channel and that the N-terminal end of i3 of M2R plays an important role in co-localization.

Citation: Tateyama M, Kubo Y (2018) Gi/o-coupled muscarinic receptors co-localize with GIRK channel for efficient channel activation. PLoS ONE 13(9): e0204447. https://doi.org/10.1371/journal.pone.0204447

Editor: Mark S. Shapiro, University of Texas Health Science Center, UNITED STATES

Received: May 9, 2018; Accepted: September 7, 2018; Published: September 21, 2018

Copyright: © 2018 Tateyama, Kubo. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and its Supporting Information files.

Funding: This work was supported partly by research grants from the Japan Society for the Promotion of Science to Michihiro Tateyama (17K08557) and Yoshihiro Kubo(17H04021).

Competing interests: The authors have declared that no competing interests exist.

Introduction

G protein-gated inwardly rectifying potassium (GIRK) channel is a key protein that regulates cellular excitability and is activated by interacting with free Gβγ released from the heterotrimeric Gαβγ complex upon the activation of G protein-coupled receptors (GPCRs) [1]. The conformational rearrangement between the GIRK channel and Gβγ has been observed upon the receptor activation on the living cell membrane by Förster or bioluminescence resonance energy transfer analyses (FRET or BRET, respectively) [2, 3] and the GIRK-Gβγ complex formation was demonstrated by structural analyses [4]. The GIRK channel is also known to bind to Gαi and Gαo in their resting and active forms [5–8], suggesting that Gβγ released from Gαi or Gαo immediately interacts with and activates the GIRK channel. In addition, several Gi/o-coupled receptors (Gi/o-Rs) have been reported to localize near Gi/o by FRET and BRET analyses [9–11]. Therefore, Gi/o-Rs, Gi/o and the GIRK channel have been suggested to form a ternary pre-signaling complex to effectively activate the GIRK channel.

Gi/o-Rs activate the GIRK channel through releasing Gβγ from pertussis toxin (PTX)-sensitive Gαi/o [1, 12]. When the receptors are highly expressed, the GIRK channel is activated by receptors that stimulate PTX-resistant G protein, such as the Gq-coupled muscarinic receptors (M1R and M3R) or Gs-coupled adrenergic receptors (β1-AR and β2-AR) [13, 14]. When surface expression of M1R is very high, the chance of M1R-Gi/o coupling may increase [13] and/or the distance between M1R and GIRK channel may be shortened, which could allow Gβγ released from Gq to activate the channel. Interestingly, Gs-coupled β2-AR has been reported to functionally and physically interact with GIRK channels [15, 16]. In addition, the rate of GIRK channel activation was thought to correlate with the co-localization of Gi/o-Rs with the channel [17]. We thought that the distance between GPCR and the GIRK channel is a critical factor for effective channel activation.

In this study, we first showed by FRET analyses that a chimeric receptor of M1R and M2R (MC9) [18], which activates the GIRK channel [19], locates adjacent to the channel and Gαi1 in transfected HEK293T cells. Next, we examined whether Gq-coupled M1R activates the GIRK channel when these molecules are forced to stay close to each other by ligating them with short linkers. The key amino acid residues in complex formation for activating the GIRK channel were further investigated by analyzing the changes in co-localization of MC9 mutants with the GIRK channel.

Materials and methods

Constructs and expression system

The chimeric MC9 receptor was constructed by replacing the last eight residues of the 5th transmembrane (TM) and most of i3 of M1R with those of M2R (M2R-i3) as previously reported [18]. YFP was fused at the C-tail of M1R, M2R, and MC9 (M1R-YFP, M2R-YFP, and MC9-YFP, respectively). To control the subunit composition of the GIRK channel, GIRK1 and GIRK2 were ligated (GIRK1/2) [20], by a glycine-rich linker of 34 amino acids (a.a.) (see S1 Method). For FRET analysis, CFP was attached to the C-tail of GIRK1/2 (GIRK1/2-CFP). Fluorescent protein (FP) was inserted into mouse Gαi1 between Ala121 and Glu122 (Gαi1-FP) with a short linker (A121-SGGGS-V2…K212-SGGGS-E122) [21] or into mouse Gαq between Phe123 and Glu124 with the SGGGS linker. M1R-YFP and MC9-YFP were ligated to GIRK1/2 or GIRK1/2-CFP by junctional glycine-rich linkers, with varying lengths of 34, 100, 265, and 535 a.a. residues (S1 Method). The chimeric Gαqi5 was constructed by replacing the last five residues of Gαq with those of Gαi1 [22]. Each construct was subcloned into the pcDNA3.1(-) expression vector and the DNA sequence was confirmed. The cDNA construct of CFP fused with the PH domain (CFP-PH) was kindly gifted from Dr. Jalink [23]. HEK293T cells were transfected with the plasmid DNAs (μg for 4 x 104 cells) of M1R-YFP (0.8), MC9-YFP constructs (0.8), M2R-YFP (1.2), GIRK1/2 (0.7), GIRK1/2-CFP (1.2), Gα-FPs (1.1), Gαqi5 (1.0), Gβ1 (0.8), Gγ2 (0.5), M1R, MC9 or M2R tandem constructs (1.2), CFP-PH (1.1) or prestin-YFP (1.0) [24] using LipofectAMINE 2000 (1 μL, Invitrogen, Carlsbad, CA, USA). The transfected cells were seeded onto cover glasses or glass bottom dishes 8–24 h after the transfection, as previously reported [25]. G protein subunits were not additionally transfected for the electrophysiological experiments and FRET analysis between receptor-YFPs and GIRK1/2-CFP, in order to analyze the interaction between receptor and endogenous Gi/o. Experiments were carried out 24–72 h after transfection. For PTX treatment, the cells were cultured with PTX (300 ng/mL) for more than 16 h and then the experiments were performed.

Electrophysiology

Macroscopic membrane currents were recorded from cells expressing fluorescent constructs by the whole cell patch clamp technique using Axopatch 200B amplifiers, Digidata1332A, and pClamp 9 software (Molecular Devices, San Jose, CA, USA). After establishing the whole cell configuration, the cell was held at -80 mV and ramp pulses (from -120 to +40 mV for 400 s) were applied every 5 s. The bath solution was composed of the following: 140 mM NaCl, 4 mM KCl, 1 mM CaCl2, 0.3 mM MgCl2, 10 mM HEPES (pH 7.4 adjusted with NaOH). The internal solution was composed of: 130 mM KCl, 5 mM Na2-ATP, 3 mM EGTA, 0.1 mM CaCl2, 10 mM HEPES, 4 mM MgCl2, 0.3 mM GTP-Li3 (pH 7.3 adjusted with KOH). An agonist of muscarinic receptors, oxotremorine M (oxo-M), was applied by using fast perfusion system (VC-77SP, Warner Instruments, Hamden, CT, USA) as previously described [26]. To record the GIRK channel current, Na+ ions in the bath solution were substituted with K+ to enhance the driving force of K+ and oxo-M was applied in 140 mM K+ bath solution. The average of the holding current amplitude for 100 ms before applying the ramp pulse was measured and then normalized to the cell capacitance to calculate the current density (I). The basal current density (I0) before application of oxo-M (10 μM) was subtracted from the maximum value (Imax) after agonist application to evaluate the agonist-induced current density (ΔImax). To analyze the decay of the GIRK current, the ratio of the agonist-induced current density before washout of the agonist (ΔIlast) to (ΔImax) was calculated. We also analyzed the onset of GIRK channel activation, with the application of oxo-M for 5 s controlled by the combination of Clampex9 and VC-77SP. The time to half-maximum current amplitude (t1/2) was then measured for each trace.

Imaging and FRET analysis

Fluorescent images were obtained from cells expressing FP-fused constructs using a total internal reflection fluorescence (TIRF) microscope (Olympus, Tokyo, Japan) as previously described [25]. MetaFluor imaging software (Molecular Devices, Sunnyvale, CA, USA) was used to control the excitation of CFP and YFP and to acquire images. For FRET analysis, five images were acquired before and after acceptor bleaching, which was performed by excitation of YFP with a 515-nm laser line for 1 min (emission intensity of YFP was decreased by less than 5% in the bleaching procedure). Averaged emission intensities of CFP (ICFP) and YFP (IYFP) before and after acceptor bleaching were measured in each cell by subtracting their background intensities and were then normalized to the cell size. To calculate FRET efficiency, the increase in ICFP after acceptor bleaching was normalized to the total ICFP after bleaching in each cell. To make the CFP emission images, five CFP images before and after acceptor bleaching were averaged and their subtraction was performed by using MetaMorph software (ver 6, Molecular Devices).

To evaluate the activation of Gqi5 by MC9 constructs, the intensity of CFP-PH (ICFP-PH) was measured under TIRF illumination. Cells expressing the fluorescent constructs were continuously perfused with bath solution by gravity at a rate of approximately 3 mL/min, and various concentrations of oxo-M were applied by changing the perfusion solution. The baseline ICFP-PH was the averages of ICFP-PH measured from 13 images before oxo-M application. The maximal decrease in ICFP-PH upon application of oxo-M (ΔICFP-PH) was normalized to the baseline ICFP-PH in each cell.

Statistical analysis

All data are expressed as the means ± S.E., with n indicating the number of data. Statistical significance between two groups was examined by unpaired Student’s t-test and that between more than two groups was tested by one-way analysis of variance (ANOVA) followed by Tukey’s test. Values of p≤0.05 were considered statistically significant (*:0.01<p≤0.05, **:0.001<p≤0.01, ***:p≤0.001, n.s.:p>0.05). Pearson correlation coefficients were calculated using Microsoft Excel2013.

Results

i3 of M2R is a key structure in activating and staying adjacent to the GIRK channel

Application of an agonist oxo-M increased the amplitude of the inward current density in cells transfected with the GIRK channel and Gi/o-coupled M2R, but not with the channel and Gq-coupled M1R (Fig 1A and 1B, left and center panels). The amplitude of the agonist-induced GIRK channel current density (ΔImax) was decreased to almost null by treating the cells with PTX (300 ng/mL) (Fig 1C left panels), showing that the effect of M2R is mediated by the activation of Gi/o. Replacement of the i3 of M1R with that of M2R (M2R-i3) enabled the chimeric receptor (MC9) to activate the GIRK channel, as previously reported [19]. The effect of MC9 was inhibited by PTX treatment (Fig 1C right panels) and the speed of current increase induced by the MC9 activation did not differ from that by M2R (S1 Fig), indicating that M2R-i3 is a critical structure for Gi/o coupling and GIRK channel activation. We then examined whether or not MC9 locates adjacent to the GIRK channel by analyzing FRET from GIRK1/2-CFP to receptor-YFP. In HEK293T cells expressing these constructs, acceptor bleaching resulted in the increases in intensities of CFP (ICFP) (Fig 2A). Interestingly, the FRET efficiency from GIRK1/2-CFP to MC9-YFP was larger than that to M1R-YFP (Fig 2B left panels). As the surface expression level of the FP-tagged constructs were not different (S1 Table), MC9 but not M1R was suggested to co-localize with the GIRK channel. The FRET efficiency did not change by application of oxo-M (Fig 2B), suggesting that activation of receptor does not change the co-localization. We also analyzed the FRET from Gαi1-CFP to receptor-YFP in HEK293T cells transfected together with Gβ1 and Gγ2. The FRET efficiency from Gαi1-CFP to MC9-YFP was higher than that to M1R-YFP both in the absence and presence of the agonist (Fig 2B bottom bars). The results support pre-coupling of MC9 with Gαi1 and show that pre-coupling is not changed by MC9 activation. The interaction between GIRK1/2 and Gαi1 showed higher FRET efficiency than that between GIRK1/2 and Gαq (Fig 2B, right panel), which is consistent with previous reports [5–8]. As a negative control for the FRET analysis, we chose a member of an anion-transporter family, prestin [24], since prestin is not expected to selectively interact with the GIRK channel and G protein. FRET values from CFP tagged constructs to prestin-YFP were similar to those to M1R-YFP (Fig 2), suggesting that M1R does not specifically interact with GIRK channel and Gαi1. Taken together, MC9, GIRK, and Gi/o are adjacent to each other and may form a molecular complex.

thumbnail
Download:
Fig 1. Chimeric MC9 receptor activates GIRK channel.

GIRK channel was activated by M2R-YFP (left) and MC9-YFP (right) but not by M1R-YFP (middle). (A) Schematic diagrams in upper panels depict the tested muscarinic receptors and GIRK1/2 channel. Shown in the middle panels are the current traces recorded in 140 mM KCl bath solution. Cells were held at -80 mV and the ramp pulse (-120 to 40 mV for 400 ms) was applied every 5 s. The black bars on the traces indicate the timing of agonist application. Basal and maximal current amplitude was measured at the holding potential before (a) and during (b, red) oxo-M application. (B) Expanded traces corresponding to “a” (black lines) and “b” (red lines) are shown in middle panels. The ramp pulse protocol is shown above the expanded trace. The agonist-induced current densities at a holding potential of -80 mV (ΔImax) was measured. (C) Summary of ΔImax is shown as bars. Numbers of data are shown in parentheses. ***:p≤0.001, n.s.:p>0.05.

https://doi.org/10.1371/journal.pone.0204447.g001

thumbnail
Download:
Fig 2. Chimeric MC9 receptor remains in proximity of GIRK channel.

(A) Representative CFP emission images before (left panel) and after (middle panel) acceptor bleaching, and their subtraction (right panel). Cells expressing GIRK1/2-CFP and MC9-YFP were excited by 442 nm laser under TIRF illumination. YPF was bleached by exposing strong 515 nm laser line for 1 min. White bars indicate 20 μm and cyan scale bars represent fluorescence intensity. (B) FRET efficiency. FRET efficiency was analyzed in the absence or presence of oxo-M (10 μM) and calculated by the acceptor bleaching method and is shown as bars. Notes below the bar indicate each combination for FRET analysis. Prestin-YFP was thought to reflect non-specific FRET from GIRK-CFP or Gαi1-CFP to YFP-tagged membrane protein. Numbers of data are shown in the parentheses. *:0.01<p≤0.05, **:0.001<p≤ 0.01, ***:p≤0.001, n.s.:p>0.05.

https://doi.org/10.1371/journal.pone.0204447.g002

Distance between receptor and GIRK1/2 channel is a key determinant for channel activation

The FRET analysis raised a question about the functional significance of co-localization. To address this question, Gq-coupled M1R was forced to stay adjacent to the GIRK channel by connecting them with a short linker composed of 34 glycine-rich amino acid (a.a.) residues (Fig 3A). Application of oxo-M (10 μM) elicited a rapid increase in the amplitude of the inward current in cells transfected with M1R-YFP-34-GIRK1/2 (Fig 3B red lines). This effect remained after PTX treatment (Fig 3C, Table 1), suggesting that Gβγ released from PTX-resistant Gαq activated the channel. The amplitude of the GIRK channel current decreased by approximately 30% at 25 s after application of oxo-M, which may have resulted from negative regulation induced by Gq signaling [27]. When the linker length was elongated, the GIRK channel was not activated; ΔImax was decreased in accordance with the increase in the linker length of the M1R tandem constructs (Fig 3D open circles). In contrast, the application of oxo-M evoked a rapid increase in the inward current, even when MC9-YFP was connected to GIRK1/2 by a long linker of 535 a.a. residues (Fig 3E and 3F). The response was totally mediated by PTX-sensitive Gi/o (Fig 3G, Table 1). Similar results were observed when the linker length was 100 or 265 a.a. (Fig 3H, Table 1), indicating that the effect of MC9 does not change by the linker length. ΔImax in the MC9 tandem constructs were 30% smaller than that in non-linker combination, but the difference was not statistically significant (Fig 3H). The ligation of the receptor and the GIRK channel may have a slight inhibitory effect on the expression and/or activation of the GIRK channel. These results suggest that MC9, but not M1R, is in proximity of the GIRK channel, even when the linker is long, presumably by forming a molecular complex.

thumbnail
Download:
Fig 3. M1R activates GIRK channel when they are ligated with short junctional linkers.

GIRK channel activation in tandem constructs of M1R-YFP (A–D) and of MC9-YFP (E–H). (A) and (E) Schematic diagram of constructs. Shown are the tandem constructs of receptor-YFP and GIRK1/2 channel, which were ligated by junctional linkers. (B) and (F) GIRK channel currents are shown in the upper panel. Whole cell currents were recorded from cells expressing M1R-YFP-34-GIRK1/2 (B) or MC9-YFP-535-GIRK1/2 (F) in 140 mM KCl bath solution. Application of oxo-M (10 μM for 25 s) is indicated as black bars on the upper traces. Time scale of current traces corresponding to “a” (before application of oxo-M, black lines) and to “b” (after application of oxo-M, red lines) is expanded and shown in the lower panel. The ramp pulse protocols are shown above the expanded traces in the bottom. (C) and (G) GIRK channel current in the PTX-treated cell. Cells expressing M1R-YFP-34-GIRK1/2 (C) or MC9-YFP-535-GIRK1/2 (G) were treated with PTX (300 ng/mL). (D) and (H) Correlation between ΔImax and the linker length. The amplitudes of agonist-induced GIRK channel current density (ΔImax) are plotted as a function of the length of the linker residues. Circles represent ΔImax of the M1R constructs (D, open circles: tandem constructs, filled circle: no linker). Squares represent ΔImax of the MC9 constructs (H, open squares: tandem constructs, filled square: no linker). Numbers of data are shown in the parentheses. *:0.01<p≤0.05, **:0.001<p≤ 0.01, ***:p≤0.001, n.s.:p>0.05.

https://doi.org/10.1371/journal.pone.0204447.g003

thumbnail
Download:
Table 1. Effects of PTX treatment on the GIRK channel activation of M1R-YFP and MC9-YFP tandem constructs.

https://doi.org/10.1371/journal.pone.0204447.t001

Next, the FRET efficiency between GIRK1/2-CFP and M1R- or MC9-YFP was measured for each tandem construct. FRET efficiency in the M1R tandem construct was decreased in accordance with the increase in the linker length (Fig 4A open diamonds), whereas that in the MC9 construct was not changed by the change of the linker length in the presence or absence of the agonist (Fig 4B triangles). For M1R, FRET efficiency was well-correlated with ΔImax (Fig 4C, Pearson correlation coefficient is 0.98). The results suggest that receptor-GIRK signaling tightly depends on their distance. For MC9, the FRET efficiency and ΔImax did not change when the linker length was changed (Fig 4C, Pearson correlation coefficient is 0.53). These results support that MC9, but not M1R, forms a complex with the GIRK channel for efficient channel activation.

thumbnail
Download:
Fig 4. Distance from M1R to GIRK channel correlates with efficacy of channel activation.

(A) and (B) FRET analyses. Schematic diagrams in the top panel illustrate the tandem construct for the FRET analysis. Symbols represent FRET efficiency in each tandem construct in the absence of oxo-M (A, open diamonds for M1R tandem constructs, B, open triangles for MC9 constructs) or presence of oxo-M (B, filled triangles for MC9 constructs). Dashed lines indicate FRET efficiency when the receptor-YFP and GIRK1/2-CFP were co-expressed (no linker, data form Fig 2B). Numbers of data are shown in the parentheses. ***:p≤0.001, n.s.:p>0.05 (C) Correlation between ΔImax and FRET efficiency. ΔImax is plotted as a function of the FRET efficiency for each tandem construct (open symbols) or each combination (no linker, filled symbols). The number of junctional linker residues is indicated on each symbol. The results of M1R tandem constructs are expanded in the inset to highlight the correlation between FRET efficiency and the GIRK channel activation.

https://doi.org/10.1371/journal.pone.0204447.g004

M2R forms a complex with GIRK channel

M2R-i3 is the key structure in complex formation with the GIRK channel and Gi/o (Figs 1 and 2). Thus, the results of the M2R tandem constructs connected to the GIRK channel by various linker residues were expected to be similar to those observed in MC9 constructs (Figs 3H and 4B). As expected, ΔImax as well as FRET efficiency were not changed with the increases in the linker length (Fig 5A and 5B) and ΔImax was not positively correlated with FRET efficiency (Fig 5C, Pearson correlation coefficient is -0.79). GIRK current increase induced by M2R was completely inhibited by the PTX treatment (Table 2), indicating that the effects were mediated by the activation of Gi/o in the M2R tandem constructs. The FRET efficiency from GIRK to M2R was smaller than that to MC9, which may be due to the difference in the C-terminal tail of the receptor at which YFP is tagged. Gi/o was suggested to be located in proximity to M2R as well as the GIRK channel; the FRET efficiency from Gαi1-CFP to M2R-YFP was 2.8 ± 0.5 (n = 12, p = 0.001 compared to FRET from Gαi1-CFP to prestin-YFP in Fig 2B). Taken together, these results indicate that M2R, Gi/o, and the GIRK channel form a molecular complex.

thumbnail
Download:
Fig 5. M2R forms a molecular complex with GIRK channel.

(A) GIRK current density. Schematic diagram in the top panel illustrates the tandem construct. Symbols represent the amplitude of the agonist-induced GIRK current density (ΔImax) recorded from cells transfected with M2R-YFP tandem constructs (open squares) or co-transfected with M2R-YFP and GIRK1/2 (no linker, filled square). Numbers of data are shown in the parentheses. n.s.:p>0.05 (B) FRET analysis. Schematic diagram in the top panel illustrates the construct for the FRET analysis. Symbols represent FRET efficiency in each tandem construct (open squares). The dashed line indicates FRET efficiency when the M2R-YFP and GIRK1/2-CFP were co-expressed (no linker). Numbers of data are shown in the parentheses. (C) Correlation between ΔImax and FRET efficiency. ΔImax is plotted as a function of FRET efficiency for each tandem construct (open squares) or a combination with no linker (filled square). The number of junctional linker residues is indicated on each symbol.

https://doi.org/10.1371/journal.pone.0204447.g005

thumbnail
Download:
Table 2. Inhibitory effects of PTX treatment on the GIRK channel activation by M2R-YFP tandem constructs.

https://doi.org/10.1371/journal.pone.0204447.t002

Dual roles of M2R-i3 in the activation of and co-localization with GIRK

We then evaluated the residues in M2R-i3 responsible for activation of and co-localization with the GIRK channel and activation of Gi/o. It was previously reported that most of the long M2R-i3 is not necessary for Gi/o coupling [28]. Indeed, the replacement of the long chain in the middle of M2R-i3 (K221-P379) with SGGGS did not decrease ΔImax and the FRET efficiency (ΔImax = 75.1 ± 12.5 pA/pF, n = 10; FRET = 5.5 ± 0.9%, n = 15, cf Figs 1 and 2). Thus, the proximal N- and distal C-terminal residues of M2R-i3 may be critical regions; the charged residues at the proximal N-terminal end of M2R-i3 as well as Val and Thr residues at the distal C-terminal end have been suggested to play important roles in Gi/o coupling [19, 28, 29]. Residues at the N-terminal end were replaced with those of M1R (MC9A-YFP and MC9B-YFP, Fig 6A) and also double mutations (V387A/T388A) were introduced at the C-terminal end. The emission intensities of YFP (IYFP) fused to these constructs under TIRF illumination did not differ (IYFP in S2 and S3 Tables), indicating that their surface expression levels were similar.

thumbnail
Download:
Fig 6. N-Terminal end of M2R-i3 plays important roles in the activation of and co-localization with GIRK channel.

(A) Amino acid sequences of i3 of mutant and chimeric constructs of MC9-YFP (black: M1R sequences, red: M2R sequences). (B) Summary of ΔImax. Circles represent agonist-induced GIRK channel current density (ΔImax) upon application of oxo-M (10 μM). (C) Summary of FRET efficiency between MC9-YFP constructs and GIRK1/2-CFP in the absence of the agonist. FRET efficiency was calculated by acceptor bleaching (see experimental procedures). Fluorescent intensities of YFP and CFP in each combination under TIRF illumination were measured and are summarized in S2 Table. (D) Correlation between ΔImax and FRET efficiency. ΔImax is plotted as a function of FRET efficiency for each MC9-YFP construct (open circles). (E) Activation of Gqi5 by MC9-YFP constructs. Schematic diagram of monitoring of the activation of Gqi5 is shown in the upper panel. Under TIRF illumination, application of oxo-M decreased the intensity of CFP tethered at the PH domain (ICFP-PH), as shown in lower traces in the absence of PTX treatment (open circles). The inhibitory effects of PTX (filled circles) are summarized in Table 3. (F) Summary of agonist-induced decrease in ICFP-PH (ΔICFP-PH). Circles represent ΔICFP-PH upon application of oxo-M (10 μM). Numbers of experiments are indicated in parentheses. *:0.01<p≤0.05, **:0.001<p≤ 0.01, n.s.:p>0.05.

https://doi.org/10.1371/journal.pone.0204447.g006

We first measured ΔImax and FRET from GIRK1/2-CFP to chimeric and mutant receptor-YFP. Although it was not statistically significant, ΔImax in VT/AA mutants was 30% smaller than that in MC9-YFP (Fig 6B), possibly due to the impairment of the activation of Gi/o. Interestingly, ΔImax in MC9B-YFP was significantly smaller than that in MC9-YFP (Fig 6B). Similarly, the FRET efficiency from GIRK1/2-CFP to MC9B-YFP, but not to MC9-VT/AA-YFP, was smaller than that to MC9-YFP (Fig 6C). ΔImax and FRET values appeared to be correlated in MC9-YFP chimeric constructs (Fig 6D): Pearson correlation coefficient is 0.79 (+VT/AA) and 0.97 (-VT/AA). Next, we examined the effect of the chimeric and double mutation on the Gi/o activation by using Gαqi5, which is activated by Gi/o-Rs and stimulates Gq-phospholipase C (PLC) signaling [22]. As the last five a.a. residues of Gαqi5 are derived from Gαi1 and include the ADP-ribosylation site, Gαqi5 is inhibited by PTX. The activity of Gαqi5 signaling was monitored as the decrease in fluorescent intensity of the CFP-tagged PH domain (CFP-PH) under TIRF illumination (Fig 6E upper panel). Upon PIP2 hydrolysis, CFP-PH translocates from the membrane into the cytosol, where TIRF illumination cannot reach [23, 26], resulting in a decrease in the fluorescent intensity of CFP-PH (ICFP-PH). Indeed, activation of MC9-YFP constructs decreased ICFP-PH in the presence of Gαqi5 (Fig 6E, lower left open circles). The amplitude of the decrease in ICFP-PH (ΔICFP-PH) was almost null when cells were treated with PTX (Fig 6E filled circles and Table 3), suggesting that ΔICFP-PH may reflect the efficacy of the receptor to activate Gi/o. As previously reported in studies of M2R [28, 29], VT/AA attenuated Gαqi5 activation: ΔICFP-PH in the VT/AA mutant was smaller than that in MC9-YFP (Fig 6F). In contrast, ΔICFP-PH was not changed by chimeric mutations of MC9A and MC9B (Fig 6F). These results suggest a possibility that all MC9 constructs are equally not co-localized with Gαqi5 and/or PLC. Another possibility is that there is a difference in the extent of co-localization with Gαqi5 between the MC9, MC9A and MC9B, and that the difference of the co-localization does not change the extent of PLC activation. If this is the case, co-localization might not be critical for the PLC activation, highlighting the significance of co-localization for the GIRK channel activation. We also analyzed FRET from Gαi1-CFP to MC9B-YFP and found that co-localization with Gαi1 appeared to be attenuated, but the difference was not significant (S3 Table). Taken together, at least the 13 a.a. residues at the N-terminal end of M2R-i3 play important roles in co-localization with the GIRK channel as well as channel activation, while Val and Thr residues at the C-end are required for Gq signaling by Gαqi5 activation.

thumbnail
Download:
Table 3. Inhibitory effects of PTX treatment on the activation of Gqi5 or GIRK channel currents by MC9-YFP constructs.

https://doi.org/10.1371/journal.pone.0204447.t003

Coupling properties of MC9B and MC9-VT/AA

The oxo-M concentration dependence of MC9B-YFP and MC9-VT/AA-YFP was then investigated. Examination of the concentration-ΔICFP-PH relationship (Fig 7A) revealed that EC50 did not differ between MC9 and MC9B. In the case of VT/AA mutant, 100 μM oxo-M was not the saturating concentration which made the estimation of EC50 impossible and support that VT/AA mutations impaired the Gαqi5 activation. The concentration-ΔImax of MC9-VT/AA-YFP showed that EC50 of the mutant was 5-fold larger than that of MC9-YFP (Fig 7B). The lower affinity may represent impaired Gi/o activation. In the case of MC9B, the EC50 was 2-fold larger than that of MC9 (Fig 7B filled circles). These results indicate that the coupling efficacy between MC9B and the GIRK channel was reduced. As the speed of the GIRK channel activation is another index of signaling efficacy, we analyzed the activation speed by using a motor-driven fast-perfusion system to control agonist application. The time to reach the half-maximal current (t1/2) upon activation of MC9B was longer than that of MC9 (Fig 7C), consistent with the results showing that GIRK and MC9B are distant from each other [17]. Taken together, the N- and C-terminal ends of M2R-i3 were shown to have different roles in receptor-GIRK signaling.

thumbnail
Download:
Fig 7. Properties of MC9B and MC9-VT/AA receptors.

(A) Oxo-M concentration-ΔICFP-PH relationships. ICFP-PH was decreased upon application of various concentrations of oxo-M in cells transfected with MC9-YFP constructs. ΔICFP-PH is plotted as a function of oxo-M concentration (MC9-YFP, open circles; MC9B-YFP, filled circles; MC9-VT/AA-YFP, diamonds). EC50 value is 0.82 ± 0.20 μM (n = 6) for MC9-YFP and 0.99 ± 0.32 μM (n = 5) for MC9B-YFP (p>0.05). EC50 was not calculated for MC9-VT/AA-YFP because of the small changes of ICFP-PH. (B) Oxo-M concentration-ΔImax relationships. The amplitude of the GIRK channel current density (ΔImax) induced by various concentrations of oxo-M is plotted as a function of oxo-M concentration (symbols are indicated in A). EC50 value is 0.22 ± 0.06 μM (n = 4) for MC9-YFP, 0.59 ± 0.10 μM (n = 5) for MC9B-YFP (p>0.05, vs MC9-YFP) and 1.16 ± 0.13 μM (n = 5) for MC9-VT/AA-YFP (p≤0.001 vs MC9-YFP). (C) Activation kinetics of GIRK channel current upon oxo-M application. Traces represent the GIRK channel current recorded from cells expressing MC9-YFP constructs and GIRK1/2. The application of oxo-M (10 μM) was controlled by a rapid perfusion system. Bars in right represent the half time to maximum (t1/2). Numbers of experiments are indicated in parentheses. *:0.01<p≤0.05, n.s.:p > 0.05.

https://doi.org/10.1371/journal.pone.0204447.g007

Discussion

We showed that M2R-i3 enables the chimeric MC9 to locate adjacent to Gαi1 and the GIRK channel, that Gq-coupled M1R activates the GIRK channel when it is in proximity to the channel, and that the distance between the receptor and GIRK channel is a key determinant in effective channel activation.

Interaction between Gi/o-coupled muscarinic receptors, GIRK and Gi/o

Localization of MC9 or M2R in proximity to the GIRK channel was shown by FRET analysis and experiments using tandem constructs (Figs 2–5). These results were consistent with those of the previous FRET study of metabotropic receptor GABABR, Gi/o, and GIRK [10]. As suggested for other Gi/o-Rs [9–11], MC9 and M2R are likely to be pre-coupled with Gi/o (Fig 2B). Because Gαi/o binds to the GIRK channel either in GTP-bound form or GDP-bound form [5–8, 30], the Gi/o-GIRK coupling is thought to play important roles in the ternary complex formation. In fact, replacement of the helical domain of Gαq with that of Gαi was reported to enable the chimeric G protein to interact with the GIRK channel and the Gq-coupled M1R to activate the channel in cells treated with PTX [30]. Therefore, M2R and MC9 were suggested to localize near the GIRK channel by pre-coupling with Gi/o. In the case of MC9 chimeric mutants, MC9B showed significant impairment of the activation of and co-localization with the GIRK channel (Fig 6). Although it was not statistically significant in MC9B, the co-localization with Gαi1 was attenuated (S3 Table). The lack of statistical significance (MC9B-YFP vs MC9-YFP, S3 Table) was possibly due to the small FRET efficiency between Gαi1-CFP and YFP tagged constructs relative to the background level. We think that the proximal N-terminal residues of M2R-i3 is a key region for the pre-coupling of M2R and MC9 with Gi/o. However, another possibility that the proximal N-terminal residues may play some roles in direct interaction with the GIRK channel cannot be excluded.

FRET efficiency between MC9, Gi/o, and the GIRK channel was not changed by receptor activation (Figs 2B and 4B). These results were different from those of previous FRET studies, in which agonist-induced increases in the FRET efficiency was shown between Gq-Rs and Gαq [25, 26] or between adenosine receptor A1R and Gi1 [31]. Similar FRET increases have been reported between adrenergic receptors (α2A-AR and β1-AR) and G protein [32, 33]. As revealed in a recent single-molecule imaging study of G proteins and α2A-AR or β1-AR, receptors and G proteins dynamically associate with each other [34]. These reports supported the notion that the activated receptors associate with and activate G protein, which is detected as agonist-induced increases in FRET between receptor and G protein. In contrast, agonist-induced FRET increases were not detected in several Gi/o-Rs, which were suggested to pre-couple with Gi/o [9–11]. These reports are consistent with the results showing that pre-coupling is not changed by receptor activation. Interestingly, an agonist-induced decrease in FRET was observed between the opioid receptor and Gαi1 [9]. Therefore, the receptor-G protein interaction may differ depending on the receptor type.

Difference in efficacy of GIRK channel activation between M1R and MC9

M1R activates the GIRK channel only when it is in proximity of the channel by linking them with a short linker (Fig 3), but ΔImax upon activation of M1R was lower than that of MC9 (Fig 3D and 3H). As FRET efficiency did not differ between the M1R and MC9 constructs ligated with the GIRK channel by a short linker (Fig 4A and 4B), the relative distance from M1R to the channel is likely to be similar to that from MC9. The weak effect of M1R on the GIRK channel activation may be because of the low abundance of the endogenous Gαq subunit, as co-expression of Gαs markedly increased the GIRK channel current amplitude for β2-AR [16]. We thus co-expressed Gαq with M1R-YFP-100-GIRK1/2, but co-expression failed to increase the amplitude; ΔImax was 6.4 ± 4.5 pA/pF (n = 3). This may be because of the inhibitory effects of Gαq on the GIRK channel [27, 35–37]. In fact, co-expression of Gαq with M2R-YFP-100-GIRK1/2 inhibited the effect of M2R on the GIRK channel; ΔImax was 3.6 ± 2.5 pA/pF (+Gαq, n = 4) and 78.2 ± 15.9 pA/pF (-Gαq, n = 5, P = 0.002). Therefore, the weak effect of M1R on the GIRK channel activation can be explained by the inhibitory effects of Gαq on the GIRK channel.

Localization in proximity is a key factor for activating GIRK channel

In MC9-YFP, the EC50 of GIRK channel activation was 4-fold lower than that of Gαqi5 activation (Fig 7). This can be interpreted as the functional role of receptor-GIRK channel co-localization; localization of MC9 in the proximity to the GIRK channel enhances the efficacy of receptor-GIRK channel signaling through increasing a probability for MC9 to associate with complex of Gi/o and GIRK channel (Fig 2C Gαi1 vs. Gαq). Interestingly, in the case of MC9B which impaired co-localization with the GIRK channel (Fig 6C), the difference of EC50 between ΔICFP-PH and ΔImax was attenuated. The EC50 of GIRK channel activation was 2-fold lower than that of Gαqi5 activation (Fig 7A and 7B), which may be results of the decreases in a probability for MC9B to associate with Gi/o-GIRK channel complex. In contrast, mutations at the C-terminal end of M2R-i3 (VT/AA mutation) impaired the activation of Gαqi5 (Figs 6F and 7A), consistently with previous reports in which VT/AA mutation of M2R disrupted the Gi/o activation [27, 28]. As for the GIRK channel activation, the double mutation did not abolish ΔImax but shifted the concentration-ΔImax relationship rightwards (Fig 7B). As co-localization with the GIRK channel was not affected (Fig 6C), impairment of Gi/o activation in the double mutant may be compensated by co-localization, which enhances receptor-GIRK signaling.

In conclusion, we showed that the Gi/o-coupled muscarinic receptors co-localize with the GIRK channel and Gαi1, which is mediated by at least 13 a.a. residues at the N-terminal end of M2R-i3, and that co-localization is another determinant of the efficacy of channel activation.

Supporting information

S1 Methods. Sequences of the glycine rich junctional amino acid residues.

https://doi.org/10.1371/journal.pone.0204447.s001

(DOCX)

S1 Fig. Activation speed of GIRK channel induced by M2R and MC9.

https://doi.org/10.1371/journal.pone.0204447.s002

(DOCX)

S1 Table. Fluorescent intensities of FPs fused at receptors, Gαi1 or GIRK1/2 and the FRET efficiency under TIRF illumination.

https://doi.org/10.1371/journal.pone.0204447.s003

(DOCX)

S2 Table. FRET efficiency between GIRK1/2 -CFP and MC9-YFP constructs under the TIRF illumination.

https://doi.org/10.1371/journal.pone.0204447.s004

(DOCX)

S3 Table. Fluorescent intensities of Gαi1-CFP and MC9-YFP constructs under the TIRF illumination.

https://doi.org/10.1371/journal.pone.0204447.s005

(DOCX)

Acknowledgments

We thank Drs. M. Lazdunski for mouse GIRK.2 cDNAs, T. Kubo for the porcine M1R and M2R cDNA, K. Nakajo for the junctional linker cDNAs, and K. Jalink for PH-CFP.

References

  1. 1. Luscher C, Slesinger PA. Emerging roles for G protein-gated inwardly rectifying potassium (GIRK) channels in health and disease. Nat Rev Neurosci. 2010;11(5):301–15. pmid:20389305.
  2. 2. Richard-Lalonde M, Nagi K, Audet N, Sleno R, Amraei M, Hogue M, et al. Conformational dynamics of Kir3.1/Kir3.2 channel activation via delta-opioid receptors. Mol Pharmacol. 2013;83(2):416–28. pmid:23175530.
  3. 3. Riven I, Iwanir S, Reuveny E. GIRK channel activation involves a local rearrangement of a preformed G protein channel complex. Neuron. 2006;51(5):561–73. pmid:16950155.
  4. 4. Whorton MR, MacKinnon R. X-ray structure of the mammalian GIRK2-betagamma G-protein complex. Nature. 2013;498(7453):190–7. pmid:23739333.
  5. 5. Berlin S, Tsemakhovich VA, Castel R, Ivanina T, Dessauer CW, Keren-Raifman T, et al. Two distinct aspects of coupling between Galpha(i) protein and G protein-activated K+ channel (GIRK) revealed by fluorescently labeled Galpha(i3) protein subunits. J Biol Chem. 2011;286(38):33223–35. pmid:21795707.
  6. 6. Clancy SM, Fowler CE, Finley M, Suen KF, Arrabit C, Berton F, et al. Pertussis-toxin-sensitive Galpha subunits selectively bind to C-terminal domain of neuronal GIRK channels: evidence for a heterotrimeric G-protein-channel complex. Molecular and cellular neurosciences. 2005;28(2):375–89. pmid:15691717.
  7. 7. Mase Y, Yokogawa M, Osawa M, Shimada I. Structural basis for modulation of gating property of G protein-gated inwardly rectifying potassium ion channel (GIRK) by i/o-family G protein alpha subunit (Galphai/o). J Biol Chem. 2012;287(23):19537–49. pmid:22511772.
  8. 8. Peleg S, Varon D, Ivanina T, Dessauer CW, Dascal N. G(alpha)(i) controls the gating of the G protein-activated K(+) channel, GIRK. Neuron. 2002;33(1):87–99.
  9. 9. Audet N, Gales C, Archer-Lahlou E, Vallieres M, Schiller PW, Bouvier M, et al. Bioluminescence resonance energy transfer assays reveal ligand-specific conformational changes within preformed signaling complexes containing delta-opioid receptors and heterotrimeric G proteins. J Biol Chem. 2008;283(22):15078–88. pmid:18381293.
  10. 10. Fowler CE, Aryal P, Suen KF, Slesinger PA. Evidence for association of GABA(B) receptors with Kir3 channels and regulators of G protein signalling (RGS4) proteins. J Physiol. 2007;580(Pt 1):51–65. Epub 2006/12/23. pmid:17185339.
  11. 11. Nobles M, Benians A, Tinker A. Heterotrimeric G proteins precouple with G protein-coupled receptors in living cells. Proc Natl Acad Sci U S A. 2005;102(51):18706–11. pmid:16352729.
  12. 12. Ito H, Tung RT, Sugimoto T, Kobayashi I, Takahashi K, Katada T, et al. On the mechanism of G protein beta gamma subunit activation of the muscarinic K+ channel in guinea pig atrial cell membrane. Comparison with the ATP-sensitive K+ channel. J Gen Physiol. 1992;99(6):961–83. pmid:1640222.
  13. 13. Leaney JL, Dekker LV, Tinker A. Regulation of a G protein-gated inwardly rectifying K+ channel by a Ca(2+)-independent protein kinase C. J Physiol. 2001;534(Pt. 2):367–79. pmid:11454957.
  14. 14. Wellner-Kienitz MC, Bender K, Pott L. Overexpression of beta 1 and beta 2 adrenergic receptors in rat atrial myocytes. Differential coupling to G protein-gated inward rectifier K(+) channels via G(s) and G(i)/o. J Biol Chem. 2001;276(40):37347–54.
  15. 15. Lavine N, Ethier N, Oak JN, Pei L, Liu F, Trieu P, et al. G protein-coupled receptors form stable complexes with inwardly rectifying potassium channels and adenylyl cyclase. J Biol Chem. 2002;277(48):46010–9. pmid:12297500.
  16. 16. Lim NF, Dascal N, Labarca C, Davidson N, Lester HA. A G protein-gated K channel is activated via beta 2-adrenergic receptors and G beta gamma subunits in Xenopus oocytes. J Gen Physiol. 1995;105(3):421–39. pmid:7769382.
  17. 17. Schwarzer S, Nobles M, Tinker A. Do caveolae have a role in the fidelity and dynamics of receptor activation of G-protein-gated inwardly rectifying potassium channels? J Biol Chem. 2010;285(36):27817–26. pmid:20562107.
  18. 18. Kubo T, Bujo H, Akiba I, Nakai J, Mishina M, Numa S. Location of a region of the muscarinic acetylcholine receptor involved in selective effector coupling. FEBS Lett. 1988;241(1–2):119–25. pmid:3197827.
  19. 19. Ben-Chaim Y, Chanda B, Dascal N, Bezanilla F, Parnas I, Parnas H. Movement of ‘gating charge’ is coupled to ligand binding in a G-protein-coupled receptor. Nature. 2006;444(7115):106–9. Epub 2006/10/27. pmid:17065983.
  20. 20. Slesinger PA, Stoffel M, Jan YN, Jan LY. Defective gamma-aminobutyric acid type B receptor-activated inwardly rectifying K+ currents in cerebellar granule cells isolated from weaver and Girk2 null mutant mice. Proc Natl Acad Sci U S A. 1997;94(22):12210–7. pmid:9342388.
  21. 21. Kobrinsky E, Abrahimi P, Duong SQ, Thomas S, Harry JB, Patel C, et al. Effect of Ca(v)beta subunits on structural organization of Ca(v)1.2 calcium channels. PLoS One. 2009;4(5):e5587. Epub 2009/06/06. pmid:19492014.
  22. 22. Conklin BR, Farfel Z, Lustig KD, Julius D, Bourne HR. Substitution of three amino acids switches receptor specificity of Gq alpha to that of Gi alpha. Nature. 1993;363(6426):274–6. pmid:8387644.
  23. 23. van der Wal J, Habets R, Varnai P, Balla T, Jalink K. Monitoring agonist-induced phospholipase C activation in live cells by fluorescence resonance energy transfer. J Biol Chem. 2001;276(18):15337–44. Epub 2001/01/22. pmid:11152673.
  24. 24. Gleitsman KR, Tateyama M, Kubo Y. Structural rearrangements of the motor protein prestin revealed by fluorescence resonance energy transfer. Am J Physiol Cell Physiol. 2009;297(2):C290–8. pmid:19515900.
  25. 25. Tateyama M, Kubo Y. Binding of Gq protein stabilizes the activated state of the muscarinic receptor type 1. Neuropharmacology. 2013;65:173–81. pmid:23085334.
  26. 26. Tateyama M, Kubo Y. Analyses of the effects of Gq protein on the activated states of the muscarinic M3 receptor and the purinergic P2Y1 receptor. Physiol Rep. 2013;1(5):e00134. pmid:24303197.
  27. 27. Kobrinsky E, Mirshahi T, Zhang H, Jin T, Logothetis DE. Receptor-mediated hydrolysis of plasma membrane messenger PIP2 leads to K+-current desensitization. Nat Cell Biol. 2000;2(8):507–14. pmid:10934471.
  28. 28. Liu J, Conklin BR, Blin N, Yun J, Wess J. Identification of a receptor/G-protein contact site critical for signaling specificity and G-protein activation. Proc Natl Acad Sci U S A. 1995;92(25):11642–6. pmid:8524820.
  29. 29. Kostenis E, Conklin BR, Wess J. Molecular basis of receptor/G protein coupling selectivity studied by coexpression of wild type and mutant m2 muscarinic receptors with mutant G alpha(q) subunits. Biochemistry. 1997;36(6):1487–95. pmid:9063897.
  30. 30. Rusinova R, Mirshahi T, Logothetis DE. Specificity of Gbetagamma signaling to Kir3 channels depends on the helical domain of pertussis toxin-sensitive Galpha subunits. J Biol Chem. 2007;282(47):34019–30. pmid:17872944.
  31. 31. Tateyama M, Kubo Y. Stabilizing effects of G protein on the active conformation of adenosine A1 receptor differ depending on G protein type. Eur J Pharmacol. 2016;788:122–31. pmid:27321872.
  32. 32. Hein P, Frank M, Hoffmann C, Lohse MJ, Bunemann M. Dynamics of receptor/G protein coupling in living cells. EMBO J. 2005;24(23):4106–14. pmid:16292347.
  33. 33. Hein P, Rochais F, Hoffmann C, Dorsch S, Nikolaev VO, Engelhardt S, et al. Gs activation is time-limiting in initiating receptor-mediated signaling. J Biol Chem. 2006;281(44):33345–51. Epub 2006/09/12. pmid:16963443.
  34. 34. Sungkaworn T, Jobin ML, Burnecki K, Weron A, Lohse MJ, Calebiro D. Single-molecule imaging reveals receptor-G protein interactions at cell surface hot spots. Nature. 2017;550(7677):543–7. pmid:29045395.
  35. 35. Braun AP, Fedida D, Giles WR. Activation of alpha 1-adrenoceptors modulates the inwardly rectifying potassium currents of mammalian atrial myocytes. Pflugers Arch. 1992;421(5):431–9. pmid:1361052.
  36. 36. Koike-Tani M, Collins JM, Kawano T, Zhao P, Zhao Q, Kozasa T, et al. Signal transduction pathway for the substance P-induced inhibition of rat Kir3 (GIRK) channel. J Physiol. 2005;564(Pt 2):489–500. pmid:15731196.
  37. 37. Lei Q, Talley EM, Bayliss DA. Receptor-mediated inhibition of G protein-coupled inwardly rectifying potassium channels involves G(alpha)q family subunits, phospholipase C, and a readily diffusible messenger. J Biol Chem. 2001;276(20):16720–30. pmid:11279027.