In vitro and in silico characterization of the inhibition of Kir4.1 channels by aminoglycoside antibiotics

Abstract

Background and Purpose

Aminoglycoside antibiotics are positively charged molecules that are known to inhibit several ion channels. In this study, we have shown that aminoglycosides also inhibit the activity of Kir4.1 channels. Aminoglycosides inhibit Kir4.1 channels by a pore‐blocking mechanism, plugging the central vestibule of the channel.

Experimental Approach

Patch‐clamp recordings were made in HEK‐293 cells transiently expressing Kir4.1 channels to analyse the effects of gentamicin, neomycin and kanamycin. In silico modelling followed by mutagenesis were realized to identify the residues critical for aminoglycosides binding to Kir4.1.

Key Results

Aminoglycoside antibiotics block Kir4.1 channels in a concentration‐ and voltage‐dependent manner, getting access to the protein from the intracellular side of the plasma membrane. Aminoglycosides block Ki4.1 with a rank order of potency as follows: gentamicin ˃ neomycin ˃ kanamycin. The residues T128 and principally E158, facing the central cavity of Kir4.1, are important structural determinants for aminoglycosides binding to the channel, as determined by our in silico modelling and confirmed by mutagenesis experiments.

Conclusion and Implications

Kir4.1 channels are also target of aminoglycoside antibiotics, which could affect potassium transport in several tissues.

Abbreviations

EGFP

enhanced green fluorescence protein

EK

potassium equilibrium potential

KCNQ4

potassium voltage‐gated channel subfamily Q member 4

Kir

inwardly rectifying potassium channels

MET

mechano‐electric transducer channel

MMPBSA

molecular mechanics Poisson–Boltzmann surface area

POPC

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine

POPE

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphoethanolamine

POPS

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphoserine

RMDS

root mean square deviation

TRPV1

transient receptor potential cation channel subfamily V member 1

What is already known

• Aminoglycoside antibiotics are known to block several ion channels by different mechanisms.

What does this study add

• Aminoglycosides inhibit Kir4.1 channels by a pore‐blocking mechanism.

• A residue on the pore cavity (E158) is critical for this block.

What is the clinical significance

• Some aminoglycoside side effects may arise from inhibition of Kir4.1 channels.

1. INTRODUCTION

Aminoglycoside antibiotics (AGAs) are highly efficacious drugs, particularly against Gram‐negative bacteria (Krause, Serio, Kane, & Connolly, 2016). However, these drugs carry significant risk of nephrotoxicity (Mingeot‐Leclercq & Tulkens, 1999; Wargo & Edwards, 2014) and ototoxicity (Jiang, Karasawa, & Steyger, 2017; O'Sullivan et al., 2017; Selimoglu, 2007). Aminoglycoside antibiotics enter the inner ear by crossing the blood‐labyrinth barrier through a yet undefined mechanism. It is proposed that on its route from the blood‐labyrinth barrier to the cochlear hair cells, aminoglycoside antibiotics cross the endothelial cells of strial capillaries, cells of the stria vascularis and the intrastrial space, the endolymph and finally enter the cochlear hair cells through the apical membrane (Li & Steyger, 2011; Steyger & Karasawa, 2008). Although cochlear hair cells are the primary site of damage induced by aminoglycoside antibiotics on the inner ear, the stria vascularis is also susceptible to this group of drugs (Forge & Fradis, 1985; Kusunoki et al., 2004; Xiong et al., 2011).

Aminoglycoside antibiotics are polycationic molecules that are known to inhibit several ion channels including Ca²⁺‐activated K⁺ channels (Nomura, Naruse, Watanabe, & Sokabe, 1990), Ca²⁺ channels (Pichler et al., 1996; Zhou & Zhao, 2002), Na⁺ channels (Zhou & Zhao, 2002), TRPV1 (Raisinghani & Premkumar, 2005), acid‐sensing ion channels (Garza, Lopez‐Ramirez, Vega, & Soto, 2010), mechano‐electric transducer (MET) channels (Kimitsuki & Ohmori, 1993) and Kv7.4 (KCNQ4) (Leitner, Halaszovich, & Oliver, 2011). Some of these channels are present in the inner ear. For example, in hair cells, aminoglycoside antibiotics not only block the MET channels but also enter the cells through this protein (Marcotti, van Netten, & Kros, 2005). In addition, the inhibition of Kv7.4 channels has been proposed to contribute on the aminoglycoside antibiotics‐induced outer hair cell degeneration (Leitner et al., 2011). Furthermore, on the route followed by aminoglycoside antibiotics to the hair cells, several other ion channels and transporters exist (Mittal et al., 2017), which could be affected by these drugs.

Inwardly rectifying potassium (Kir) channels, a class of K⁺ channels that conduct larger inward currents at membrane voltages negative to the K⁺ equilibrium potential (EK) than outward currents at positive voltages to it (Hibino et al., 2010), have a critical role in the inner ear for hearing (Chen & Zhao, 2014; Hibino et al., 1997). Kir4.1 is the predominant isoform in the inner ear and the only detectable Kir subunit in the stria vascularis (Hibino et al., 1997), located specifically at the apical membrane of intermediate cells (Ando & Takeuchi, 1999). Kir4.1 channels participate in the generation and maintenance of the positive endocochlear potential and high K⁺ concentration in the endolymph (Ando & Takeuchi, 1999; Hibino et al., 1997; Marcus, Wu, Wangemann, & Kofuji, 2002; Takeuchi, Ando, & Kakigi, 2000). Loss of function mutations on Kir4.1 can induce hearing loss, as has been detected in several patients with SeSAME/EAST syndrome (Bockenhauer et al., 2009; Freudenthal et al., 2011; Scholl et al., 2009). Therefore drugs that compromise channel function could also compromise hearing. Here, we found that aminoglycoside antibiotics inhibited Kir4.1 channels. Based on our patch‐clamp experiments combined with mutagenesis, molecular dynamics simulations and ligand docking, we propose that these drugs block the permeation pathway of Kir4.1 in a concentration‐ and voltage‐dependent manner, decreasing potassium outflow. The inhibition of Kir4.1 channel function could be involved in the early hearing impairment induced by aminoglycosides.

2. METHODS

2.1. Molecular biology, cell culture and transfection

Rat Kir4.1 cDNA (98% amino acid sequence identity to human Kir4.1 cDNA) (Kindly provided by Dr. Colin G. Nichols, Washington University, St. Louis, MO, USA) was subcloned into the pcDNA3.1(+) plasmid (Invitrogen, Carlsbad, CA, USA). T128A and E158N mutations were made using the QuikChange Site‐Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) and confirmed by direct DNA sequencing. HEK‐293 cells (ATCC® CRL‐1573™, RRID:CVCL_0045) were grown in 60‐mm tissue culture dishes (Corning, Corning, NY, USA) at 37°C in a humidified air atmosphere containing 5% CO₂. Cells were maintained in DMEM (Gibco, Grand Island, NY, USA) supplemented with 10% (v/v) FBS (Corning Life Sciences, Manassas, VA, USA) and 1% (v/v) antibiotic‐antimycotic solution (Sigma‐Aldrich). cDNAs encoding Kir4.1 wild type (WT) or mutant channels were transiently transfected in HEK‐293 cells using the Lipofectamine 2000 reagent (Invitrogen). The cDNA of the enhanced green fluorescence protein (EGFP) was co‐transfected with the cDNA of interest to detect the successfully transfected cells.

2.2. Electrophysiological recordings

Macroscopic current recordings in HEK‐293 cells were performed at room temperature (22–24°C) by using the whole‐cell and inside‐out configurations of the patch‐clamp technique. Data acquisition and generation of voltage‐clamp pulse protocols were carried out using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA) and a Digidata 1440A interface (Molecular Devices) controlled by the pCLAMP 10 software (Molecular Devices, RRID:SCR_011323). Currents were filtered with a four‐pole Bessel filter at 1 kHz and digitized at 10 kHz. The specific voltage‐clamp protocols used are described in Section 3 or figure legends. Micropipettes were pulled from borosilicate glass capillary tubes (World Precision Instruments, Sarasota, FL, USA) on a programmable puller (Sutter Instruments, Novato, CA, USA) and had resistances of 1.5–2.5 MΩ when filled with the internal solution. For whole‐cell recordings, the external solution contained (in mM): 140 NaCl, 4 KCl, 1 MgCl₂, 10 HEPES, 1.8 CaCl₂ and 10 glucose (pH was adjusted to 7.4 with NaOH). The pipette solution contained (in mM): 130 KCl, 1 MgCl₂, 1 EGTA, 5 K₂ATP and 10 HEPES (pH adjusted to 7.2 with KOH). The currents are represented as the current sensitive to block by 2 mM BaCl₂ (Figure S1A). Inside‐out patches were recorded by using a Mg²⁺‐ and polyamine‐free solution on both sides of the patch containing the following: 123 mM KCl, 5 mM K₂EDTA, 7.2 mM K₂HPO₄ and 8 mM KH₂PO₄, pH 7.2. To prevent current rundown, K⁺‐fluoride, Na⁺‐vanadate and K⁺‐pyrophosphate were added (Huang, Feng, & Hilgemann, 1998). The pH 5.0 condition was sufficient to abolish any detectable currents through Kir channels and offline subtraction of the pH 5.0 currents was used to subtract endogenous and leak currents (Figure S1B). An agar‐KCl bridge was used to ground the bath. Solutions were applied using a Fast‐Step Perfusion System (VC‐77SP Warner Instruments, Hamden, CT, USA).

Dose‐dependent inhibitions were measured by sequential application of aminoglycosides to the bath solution. The fractional block of current (f) was plotted as a function of drug concentration ([D] and the data were fitted with a Hill equation: $f=1/\left\{1+\left({\mathrm{IC}}_{50}\right)/{\left[\mathrm{D}\right]}^{n_{\mathrm{H}}}\right\}$, to determine the IC₅₀ and the Hill coefficient, n _H.

2.3. Molecular modelling and ligand docking

A crystal structure of Kir2.2 channel (PDBID: 3SPI) was used as a template to build the homology model of Kir4.1 channel as described in our previous reports (Arechiga‐Figueroa et al., 2017). The Induced Fit module of Schrödinger Program (Farid, Day, Friesner, & Pearlstein, 2006) was used for the docking of gentamicin, kanamycin and neomycin into the structure of Kir4.1 channel. The centre of the grid box was set to a cluster of residues, T128 and E158 of the channel. Five docking simulations were performed for each compound and the final docked poses were selected based on docked binding energies.

2.4. System set‐up and molecular dynamics simulations

CHARMM‐GUI Membrane Builder webserver (http://www.charmmgui.org/?doc=input/membrane) (Jo, Kim, & Im, 2007; Jo, Kim, Iyer, & Im, 2008; Jo, Lim, Klauda, & Im, 2009; Wu et al., 2014) was used to immerse the drug‐Kir4.1 channel complexes in explicit lipid bilayer of POPC, POPE, POPS and cholesterol with molecular ratio of 25:5:5:1 (Leal‐Pinto et al., 2010) and then the complexes were put in a water box of (118 Å × 118 Å × 170 Å); 150 mM KCl were added into the system with the K⁺ ions located in the selectivity filter as obtained from the crystal structure. Each of the three systems involved ~179,000 atoms in the molecular dynamics simulations. Amber 16 program was used for molecular dynamics simulations (Case et al., 2018). The Antechamber module of AmberTools (RRID:SCR_018497) was used to generate the parameters of the compounds using general AMBER force field (GAFF). The partial charges for compounds were calculated by using ab initio quantum chemistry at the HF/6‐31G* level (GAUSSIAN 16 program, RRID:SCR_014897). The CHELPG charge‐fitting scheme was used to calculate partial charges on the atoms (Breneman & Wiberg, 1990). The tleap module was used to neutralize the drug‐Kir4.1 channel complexes by adding additional K⁺ or Cl⁻ ions. The FF14SB, LIPID17,and GAFF2 force fields were chosen for protein, mixed lipid membrane and drugs, respectively.

The PMEMD.CUDA program in AMBER16 was used to conduct the simulations. Each of the drug‐Kir4.1 channel system was run for 100 ns molecular dynamics simulations. The molecular dynamics simulations were performed with periodic boundary conditions to produce isothermal‐isobaric ensembles (NPT, constant‐temperature, constant‐pressure ensemble). The pressure was regulated using the isotropic position scaling algorithm with the pressure relaxation time set to 2.0 ps. Long range electrostatics was calculated using the particle mesh Ewald (PME) method (Darden, Darrin, & Pedersen, 1993) with a 10 Å cut‐off. Prior to production runs, energy minimization of 2,000 steps of the steepest descent followed by 3,000 steps of conjugate gradient were carried out on the system. Subsequently, the systems were heated from 0 to 303 K using Langevin dynamics with the collision frequency of 1 ps. During the heating, the receptor complexes were position‐restrained using an initial constant force of 500 kcal·mol⁻¹·Å⁻² and weakened to 10 kcal·mol⁻¹·Å², allowing lipid and water molecules to move freely. Then, the systems went through 5 ns equilibrium molecular dynamics simulations. Finally, a total of 100 ns production molecular dynamics simulations were conducted and coordinates were saved every 100 ps for analysis. A 4‐fs timestep by employing hydrogen mass repartition algorithm for system solutes was used to accelerate the molecular dynamics simulations (Hopkins, Le Grand, Walker, & Roitberg, 2015).

The Molecular Mechanics Poisson‐Boltzmann Surface Area (MMPBSA) binding free energy calculations were implemented by Amber16 and Ambertools17 (Case et al., 2018). The detailed gentamicin, kanamycin, neomycin and Kir4.1 channel interactions were analysed using LIGPLOT program (Wallace, Laskowski, & Thornton, 1995).

2.5. Data and statistical analysis

The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2018). Patch‐clamp data were processed by using Clampfit 10 (Molecular Devices) and then analysed in Origin 8 (OriginLab Corp., Northampton, MA, USA) and GraphPad Prism 5.0 (GraphPad Software Inc., La Jolla, CA, USA, RRID:SCR_002798). Data are presented as mean ± SEM and the number of cells or patches obtained from independent experiments is indicated by n. We did not perform power analysis, randomization, or blinding of the samples under recording. All experiments were performed with the cell or patch number of n ≥ 5, which is considered to be sufficient for the evaluation of statistical difference in our electrophysiological recordings. Data with a basal current out of the range between 0.2 and 4 nA, except for T128A mutant, were excluded from the data set,and this caused the variable group sizes. Comparison of multiple groups was performed by one‐way ANOVA followed by Tukey's or Dunnett's test where appropriate. Post hoc tests were carried out only if F was significant and there was no variance in homogeneity. P values <0.05 were considered as statistically significant.

2.6. Materials

Gentamicin sulfate, kanamycin disulfate and neomycin trisulfate were purchased from Sigma‐Aldrich (St. Louis, MO, USA). All drugs were dissolved in water to make stock solutions of 100 mM. The stock solutions were diluted in bath solution to the final concentrations required for patch‐clamp recordings.

2.7. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in the IUPHAR/BPS Guide to PHARMACOLOGY http://www.guidetopharmacology.org and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander et al., 2019).

3. RESULTS

3.1. Aminoglycosides inhibit Kir4.1 channels from the cytoplasmic side

Several drugs have been reported to block Kir4.1 channels acting from the intracellular side of the plasma membrane (Arechiga‐Figueroa et al., 2017; Furutani, Ohno, Inanobe, Hibino, & Kurachi, 2009; Marmolejo‐Murillo, Arechiga‐Figueroa, Cui, et al., 2017; Marmolejo‐Murillo, Arechiga‐Figueroa, Moreno‐Galindo, et al., 2017; Rodriguez‐Menchaca, Arechiga‐Figueroa, & Sanchez‐Chapula, 2016). Thus, the effects of gentamicin (Figure 1a), neomycin (Figure 1d) and kanamycin (Figure 1g) were first examined using the inside‐out configuration of the patch‐clamp technique under symmetrical K⁺ concentrations and in the absence of endogenous blockers (polyamines and Mg²⁺). Using this configuration, drugs were directly applied to the cytoplasmic side of the plasma membrane. Figure 1b,e,h shows representative Kir4.1 current traces in control conditions and after the application of aminoglycoside antibiotics at different concentrations. From a holding potential of −100 mV, currents were elicited by 1‐s depolarizing pulses to +80 mV, followed by repolarizing pulses to −100 mV. In control conditions, large outward and inward currents were recorded, as expected by the absence of the endogenous blockers. Application of aminoglycoside antibiotics‐induced a concentration dependent block of the outward current, with no effect on the inward current (Figure 1b,e,h). The fractional block of Kir4.1 currents elicited at +80 mV was plotted as a function of aminoglycoside antibiotics concentrations. By fitting these data with the Hill equation, we obtained an IC₅₀ for gentamicin of 6.2 ± 0.4 μM with a Hill slope (n _H) of 1.1 ± 0.06 (Figure 1c), for neomycin IC₅₀ = 63.8 ± 4.5 μM, n _H = 1.2 ± 0.05 (Figure 1f) and kanamycin IC₅₀ = 76.8 ± 8.2 μM, n _H = 0.96 ± 0.06 (Figure 1i).

FIGURE 1.

Effect of aminoglycosides on Kir4.1 channels recorded in excised inside‐out patches. (a) Chemical structure of gentamicin. (b) Representative Kir4.1 current traces recorded in the absence (control) and presence of gentamicin at increasing concentrations (3–300 μM). From a holding potential of −100 mV, currents were elicited by 1‐s depolarizing pulses to +80 mV, followed by repolarizing pulses to −100 mV. (c) Concentration–response relationship for Kir4.1 current inhibition by gentamicin at +80 mV (n = 7). (d) Chemical structure of neomycin. (e) Representative Kir4.1 current traces recorded in the absence (control) and presence of neomycin (30–1,000 μM). (f) Concentration–response relationship for Kir4.1 current inhibition by neomycin at +80 mV (n = 7). (g) Chemical structure of kanamycin. (h) Representative Kir4.1 current traces recorded in the absence (control) and presence of kanamycin (10–1,000 μM). (i) Concentration–response relationship for Kir4.1 current inhibition by kanamycin at +80 mV (n = 8)

To compare the apparent binding/unbinding rates of aminoglycoside antibiotics to Kir4.1, monoexponential fits of the current traces like those shown in Figure 2a were used to calculate the time constant (τ) of apparent binding (τ_block) and unbinding (τ_unblock) for 30 μM gentamicin, 300 μM neomycin and 300 μM kanamycin. These selected concentrations inhibit 77–83% of Kir4.1 currents. The mean time constants for apparent binding and unbinding of aminoglycoside antibiotics to Kir4.1 are shown in Figure 2b,c, respectively. The τ_block obtained for gentamicin was 99.2 ± 3.8 ms, 231.5 ± 9.7 ms for neomycin and 319.5 ± 18.8 ms for kanamycin (Figure 2b). For the unbinding of aminoglycosides, the τ_unblock was 30.4 ± 2.7 ms for gentamicin, 16.1 ± 0.5 ms for neomycin and 11.3 ± 0.7 ms for kanamycin (Figure 2c). Altogether, the obtained IC₅₀, τ_block and τ_unblock results suggest that aminoglycoside antibiotics block Kir4.1 channels with a rank order of affinity as follows:‐ gentamicin ˃ neomycin ˃ kanamycin.

FIGURE 2.

Kinetics of block and unblock of Kir4.1 currents by aminoglycosides. (a–c) Representative Kir4.1 current traces in the presence of 30 μM gentamicin, 300 μM neomycin and 300 μM kanamycin, respectively. Cells were held at −100 mV and stepped to +80 mV followed by a repolarization to −100 mV. (b) Time constants of block of Kir4.1 currents by gentamicin (n = 7), neomycin (n = 7) and kanamycin (n = 8). (c) Times constants of unblock of Kir4.1 currents by gentamicin (n = 7), neomycin (n = 7) and kanamycin (n = 8). ^* P < 0.05, significantly different as indicated, one‐way ANOVA with Tukey's multiple comparison test

We further examined the effect of aminoglycoside antibiotics at different voltages (Figure 3). From a holding potential of −100 mV, test pulses from −100 to +100 mV in 20 mV increments were applied every 10 s. Representative current traces elicited with this protocol are shown for the control condition (Figure 3a,d,g) and after the application of gentamicin (Figure 3b), neomycin (Figure 3e) and kanamycin (Figure 3h). All drugs block Kir4.1 channels in a voltage‐dependent manner (Figure 3c,f,i).

FIGURE 3.

Voltage‐dependent block of Kir4.1 currents by aminoglycosides. (a, b) Representative Kir4.1 current traces recorded in the absence (a) and presence of 30 μM gentamicin (b). For elicit the currents, test pulses from −100 mV to +100 mV in 20 mV increments were applied every 10 s, from a holding potential of −100 mV. (c) Current–voltage relationships from Kir4.1 currents in control conditions and after gentamicin application (3, 30 and 300 μM, n = 7). (d, e) Representative Kir4.1 current traces recorded in the absence (d) and presence of 300 μM neomycin (e). (f) Current–voltage relationships from Kir4.1 currents in control conditions and after neomycin application (30, 100 and 300 μM, n = 6). (g, h) Representative Kir4.1 current traces recorded in the absence (g) and presence of 300 μM kanamycin (h). (i) Current–voltage relationships from Kir4.1 currents in control conditions and after kanamycin application (100, 300 and 1,000 μM, n = 7)

The effects of aminoglycoside antibiotics were also examined using the whole‐cell configuration of the patch‐clamp technique. In this configuration, aminoglycoside antibiotics were applied from the external side of the plasma membrane. Figure 4a,b shows representative current traces elicited by a depolarizing pulse to −30 mV, followed by a repolarization to −120 mV, from a holding potential of −80 mV. In control conditions, a time‐dependent decrease in the outward current was observed, reflecting the inward rectification induced by intracellular blocking cations (Figure 4a,b). After 20 min recording under control conditions, the current amplitude remained stable, with a slight slowing of the current kinetics, which is likely due to the partly washout of the endogenous blockers (Figure 4a,c). For these experiments, we used drugs concentrations that blocked >80% of Kir4.1 currents in the inside‐out configuration. Representative current traces before and after application of 300 μM kanamycin are shown in Figure 4b. In this condition, aminoglycoside antibiotics blocked Kir4.1 channels with less potency, reflecting the poor membrane permeability of these drugs. After 20 min application, 100 μM gentamicin decreased the Kir4.1 current at −30 mV by 20 ± 2.2%, 300 μM neomycin by 22 ± 3.5% and 300 μM kanamycin by 18 ± 2.0% (Figure 4c). Temporal courses of Kir4.1 current inhibition by aminoglycoside antibiotics are shown in Figure S2. Noticeably, in the whole‐cell configuration, delivering aminoglycoside antibiotics through the patch pipette increases its potency of block. The dialysis of 100 μM gentamicin through the patch pipette inhibits Kir4.1 channels by 69.3 ± 3.5% at −30 mV (Figure S3).

FIGURE 4.

Effect of aminoglycosides on Kir4.1 channels recorded with the whole‐cell configuration. (a) Representative current traces recorded under control conditions. Currents were elicited by 2‐s depolarizing pulses to −30 mV followed by repolarizing pulses to −120 mV, from a holding potential of −90 mV. Traces at the beginning of the experiment (time 0) and after 20 min are shown. (b) Kir4.1 current traces at time 0 and after 20 min of perfusing 300 μM of kanamycin in the bath solution. (c) Normalized Kir4.1 currents recorded at −30 mV after 20 min in control (n = 7), 100 μM gentamicin (n = 6), 300 μM neomycin (n = 7) and 300 μM kanamycin (n = 8). ^* P < 0.05, significantly different from control; one‐way ANOVA with Dunnett's multiple comparison test

3.2. In silico determination of the aminoglycosides binding site in Kir4.1 channels

Gentamicin, kanamycin and neomycin were docked into the binding site of Kir4.1 channel, which was predicted previously for several Kri4.1 blockers (Arechiga‐Figueroa et al., 2017; Furutani et al., 2009; Marmolejo‐Murillo, Arechiga‐Figueroa, Cui, et al., 2017; Marmolejo‐Murillo, Arechiga‐Figueroa, Moreno‐Galindo, et al., 2017; Rodriguez‐Menchaca et al., 2016); 100 ns molecular dynamics simulations were performed based on the predicted drug‐Kri4.1 channel complexes by molecular docking simulations. The overall root mean square deviation (RMSD) of Cα atoms of Kir4.1 channel plots suggested that all the drugs possess stable and static binding with Kir4.1 (Figure 5a–c). The final conformations (after 100 ns molecular dynamics simulations) of the three drugs interacted with the Kir4.1 channel through multiple hydrogen bonds formed between the drug and mostly T128 and E158 residues, plugging the central cavity of the channel (Figure 5d–f, Table S1). The predicted 3D models of drug‐Kir4.1 channel complexes are shown in Figure 6. For quantitative estimation of the binding affinity of aminoglycoside antibiotics and Kir4.1 channel, we calculated Molecular Mechanics Poisson‐Boltzmann Surface Area (MMPBSA) binding free energy. The last 10 ns (90–100 ns) extracted from the molecular dynamic trajectories were used for the MMPBSA calculations of drug‐channel complexes. The results showed that gentamicin‐Kir4.1 channel has the lowest binding energy, while kanamycin‐Kir4.1 channel has the highest binding energy, which was well correlated to the IC₅₀ of the three drugs (Figure 7).

FIGURE 5.

Molecular dynamics simulation results on the predicted Kir4.1‐aminoglycoside antibiotics (AGAs). (a–c) The Cα atoms RMSD (root mean square deviation) based on the molecular dynamics simulation trajectories (100 ns) of gentamicin (a), kanamycin (b) and neomycin (c) with Kir4.1 channels. (d–f) Schematic depiction of the main interactions of gentamicin (d), kanamycin (e), neomycin (f) with Kir4.1 channels. The images were generated using LIGPLOT program. A distance between donor and acceptor of less than 3.5 Å is considered as a hydrogen‐bond and a 4.1 Å distance between two hydrophobic atoms is considered a hydrophobic interaction

FIGURE 6.

Molecular models of Kir4.1 channel with docked drugs. (a–c) The Kir4.1 channel model is shown in NewCartoon presentation (subunits A, B, C and D are in orange, green, blue and purple, respectively). Lateral (top) and top (bottom) views of the preferential orientation of gentamicin (a), kanamycin (b) and neomycin (c) in the pore. Drugs are drawn in Licorice (lateral view) and VDW sphere (top view)

FIGURE 7.

Correlations between MMPBSA predicted binding energies and IC₅₀s of gentamicin‐Kir4.1, neomycin‐Kir4.1 and kanamycin‐Kir4.1 complexes (r ² = 0.9902)

3.3. T128A and E158N mutants reduced Kir4.1 sensitivity to aminoglycosides

Our in silico approach identified numerous hydrogen bonds formed between the aminoglycoside antibiotics and T128 and E158 residues of Kir4.1 channels. Therefore, we generated the Kir4.1‐T128A and Kir4.1‐E158N point mutations to test for the potency of aminoglycoside antibiotics on excised inside‐out patches. For comparison, we selected aminoglycoside antibiotics concentrations blocking ˃80% of wild type (WT) Kir4.1 channel current. Figure 8a,c,e shows representative currents of Kir4.1‐WT, Kir4.1‐T128A and Kir4.1‐E158N mutant channels before and after the application of 100 μM gentamicin (Figure 8a), 300 μM neomycin (Figure 8c) and 300 μM kanamycin (Figure 8e). At +80 mV, 100 μM gentamicin blocks 86.9 ± 0.7% of Kir4.1‐WT, 84.2 ± 0.8% of Kir4.1‐T128A and 10.2 ± 2.2% of Kir4.1‐E158N currents (Figure 8b). Neomycin (300 μM) blocks 82.9 ± 0.9% of Kir4.1‐WT, 52.3 ± 2.2% of Kir4.1‐T128A and 12.5 ± 2.0% of Kir4.1‐E158N currents (Figure 8d). Finally, 300 μM kanamycin blocks 82.2 ± 3.0% of Kir4.1‐WT, 50.9 ± 2.0% of Kir4.1‐T128A and 15.1 ± 3.2% of Kir4.1‐E158N currents (Figure 8f). These results confirm the importance of E158 and T128 residues for the binding of aminoglycoside antibiotics to Kir4.1 channels to block the potassium outflow.

FIGURE 8.

Mutagenesis analysis of the Kir4.1 block by aminoglycosides. (a) Representative current traces of wild type (WT) and mutant (Kir4.1‐T128A and Kir4.1‐E158N) channels before and after 100 μM gentamicin application. Currents were elicited by 1‐s depolarizing pulses to +80 mV followed by repolarizing pulses to −100 mV, from a holding potential of −100 mV. (b) Percentage block of the currents by 100 μM gentamicin (WT, n = 7; T128A, n = 9; E158N, n = 7). (c) Representative current traces of WT, T128A and Kir4.1‐E158N channels before and after 300 μM neomycin application. (d) Percentage block of the currents by 300 μM neomycin (WT, n = 7; T128A, n = 6; E158N, n = 6). (e) Representative current traces of WT, T128A and Kir4.1‐E158N channels before and after 300 μM kanamycin application. (d) Percentage block of the currents by 300 μM kanamycin (WT, n = 8; T128A, n = 8; E158N, n = 6). ^* P < 0.05, significantly different as indicated, one‐way ANOVA with Tukey's multiple comparison test

3.4. Aminoglycosides weakly inhibit Kir1.1 channels

Kir1.1 channel is a weak inward rectifier which share ~75% homology to Kir4.1 channel in their amino acid sequence. However, the most important amino acid (Kir4.1‐E158) for aminoglycoside antibiotics‐induced block is occupied by an asparagine (N171) in Kir1.1 channels. Thus, we tested the effects of aminoglycoside antibiotics on Kir1.1 channels using the inside‐out configuration of the patch‐clamp technique under symmetrical K⁺ concentrations and in the absence of endogenous blockers. Figure 9a–c shows representative Kir1.1 current traces in control conditions and after the application of 300 μM gentamicin (a), 300 μM neomycin (b) and 300 μM kanamycin (c). The percentage of Kir1.1 currents inhibition by aminoglycoside antibiotics is shown in Figure 9d. Gentamicin blocked 20.0 ± 3.0%, neomycin 22.9 ± 2.5% and kanamycin 12.63 ± 2.0% of Kir1.1 currents at +80 mV. This result contributes additional evidence of the importance of the negatively charged E158 residue on the inhibition induced by aminoglycoside antibiotics.

FIGURE 9.

Effect of aminoglycosides on Kir1.1 channels recorded in excised inside‐out patches. (a–c) Representative Kir1.1 current traces recorded in the absence (control) and presence of 300 μM gentamicin (a), 300 μM neomycin (b) and 300 μM kanamycin (c). Currents were elicited by 1‐s depolarizing pulses to +80 mV followed by repolarizing pulses to −100 mV, from a holding potential of −100 mV. (d) Percentage block of the currents by 300 μM gentamycin (n = 5), 300 μM neomycin (n = 5),and 300 μM kanamycin (n = 5)

4. DISCUSSION AND CONCLUSIONS

In the present study, we evaluated the effects of the aminoglycoside antibiotics gentamicin, neomycin and kanamycin on Kir4.1 channels heterologously expressed in HEK‐293 cells. We observed the following:‐ (1) aminoglycoside antibiotics block Kir4.1 channels in a concentration‐ and voltage‐dependent manner from the cytoplasmic side of cells, (2) The obtained rank order of block was gentamicin ˃ neomycin ˃ kanamycin and (3) E158 residue located in the central cavity of Kir4.1 is crucial for aminoglycoside antibiotics‐mediated inhibition, with an important contribution of T128 residue for neomycin and kanamycin. Mutations on these residues alter the sensitivity of Kir4.1 channels to aminoglycoside antibiotics block.

Aminoglycoside antibiotics are known to inhibit several ion channels through different mechanism. The aminoglycoside antibiotics neomycin, dibekacin, ribostamycin and kanamycin inhibited the Ca²⁺‐activated K⁺ channels from synaptosomal membranes incorporated in planar lipid bilayers by blocking the ion‐conduction pathway (IC₅₀ in the mM range) (Nomura et al., 1990). Neomycin also inhibited the high‐threshold Ca²⁺ currents (IC₅₀ = 3.69 μM) and tetrodotoxin‐resistant Na⁺ currents (IC₅₀ = 1.2 mM) from dorsal root ganglion neurons by a pore‐blocking mechanism (Zhou & Zhao, 2002). In contrast, neomycin inhibited the TRPV1 channel activity (IC₅₀ = 411 nM) by allosteric binding and altering channel gating, but no indication of pore‐block was found (Raisinghani & Premkumar, 2005). In outer hair cells, neomycin, gentamicin and kanamycin inhibited the potassium conductance (IC₅₀ = 590 μM for neomycin), mediated by Kv7.4 channels, by depleting phosphatidylinositol 4,5‐biphosphate (Leitner et al., 2011), a required lipid for the function of Kv7 channels (Suh & Hille, 2002; Zhang et al., 2003).

Our results provide evidence that aminoglycoside antibiotics are pore‐blockers of Kir4.1 channels. In the absence of intracellular blockers (polyamines and Mg²⁺), that is, excised inside‐out patches, aminoglycoside antibiotics‐induced block was voltage dependent, the outward component of Kir4.1 current was strongly inhibited (increasing with progressive membrane depolarizations), whereas the inward component was barely affected (Figure 3). In addition, the inward current rise‐time upon repolarization was delayed by aminoglycoside antibiotics, which reflects its unbinding from the inner cavity of the channels. Moreover, we found that aminoglycoside antibiotics get access to its binding site in Kir4.1 channels from the cytoplasmic side of the plasma membrane, that is, they need to enter the cells prior to inhibit Kir4.1. We observed a reduced aminoglycoside antibiotics potency when they were applied from the extracellular side in whole‐cell experiments (Figure 4) compared to when they were applied directly to the internal side of the membrane in excised inside‐out patches (Figure 1). However, these results are not surprising as it has been shown that these drugs enter the cells through nonselective cation transporters (Karasawa, Wang, Fu, Cohen, & Steyger, 2008; Marcotti et al., 2005; Steyger & Karasawa, 2008), which probably are not endogenously expressed in HEK‐293 cells, or by endocytosis (Hashino, Shero, & Salvi, 1997), also unlikely in our system. In support to this hypothesis, the potency of gentamicin under whole‐cell configuration increases if it is dialysed through the patch pipette (Figure S3) compared to bath application (Figure 4). Altogether, these results also demonstrate that aminoglycoside antibiotics can inhibit Kir4.1 channels even in the presence of endogenous blockers (polyamines and Mg²⁺). Importantly, our in silico modelling identifies a binding site for aminoglycoside antibiotics in the pore vestibule of Kir4.1. The binding site is flanked by residues T128 and E158 (Figure 5), which agrees with our mutagenesis results (Figure 8). Therefore, our data suggest that aminoglycoside antibiotics acts by plugging the conduction pathway, as has already been suggested for other drugs (Arechiga‐Figueroa et al., 2017; Furutani et al., 2009; Marmolejo‐Murillo, Arechiga‐Figueroa, Cui, et al., 2017; Marmolejo‐Murillo, Arechiga‐Figueroa, Moreno‐Galindo, et al., 2017; Rodriguez‐Menchaca et al., 2016). In addition, Kir1.1 channel which possess an asparagine (N171) in the position equivalent to E158 on Kir4.1, is weakly affected by aminoglycoside antibiotics, supporting the importance of this negative residue.

The most serious side‐effect of aminoglycoside antibiotics is the impairment induced to cells of the inner ear (Jiang et al., 2017; O'Sullivan et al., 2017; Selimoglu, 2007). Aminoglycoside antibiotics can cause both temporary and permanent damage to cells within the inner ear (Rutka, 2019). Previous reports have suggested that systemic aminoglycoside antibiotics follow an intrastriatal pathway to the endolymph and finally enter hair cells (Li & Steyger, 2011; Steyger & Karasawa, 2008; Wang & Steyger, 2009). Furthermore, it has been shown several cell types within the cochlea can uptake aminoglycoside antibiotics, although retention of these antibiotics varies from hours to months in a cell type‐specific manner. For example, in stria vascularis basal and intermediate cells aminoglycoside antibiotics are retained for ~24 h, whereas in the outer hair cells by several months (Imamura & Adams, 2003; Wang & Steyger, 2009).

Kir4.1 channels are expressed in the intermediate cells of stria vascularis (Ando & Takeuchi, 1999) and are involved in the generation of the positive endocochlear potential (Ando & Takeuchi, 1999; Hibino et al., 1997; Marcus et al., 2002; Takeuchi et al., 2000). A combination of kanamycin and furosemide has been reported to decrease the endocochlear potential during ~24 h after administration (Xiong et al., 2011). These short‐term effect on the endocochlear potential has been attributed to the actions of the loop diuretic on the stria vascularis; nevertheless, the inhibition of Kir4.1 could also be contributing to this phenomenon. Interestingly, the period of time aminoglycoside antibiotics are retained in the intermediate cells of the stria vascularis match with the period of endocochlear potential depression (Imamura & Adams, 2003; Wang & Steyger, 2009; Xiong et al., 2011).

Even though, additional evidence is needed to determine if Kir4.1 block is involved in the adverse effects induced by aminoglycoside antibiotics.

In conclusion, our present results have identified aminoglycoside antibiotics as inhibitors of Kir4.1 channels. The mechanism involves a concentration‐ and voltage‐dependent block by interacting with residues in the central vestibule of the channel. These findings provide information about a new off‐target of aminoglycosides antibiotics which could be clinically important as the block of Kir4.1 channels may be involved in the ototoxic and nephrotoxic effects induced by these drugs.

AUTHOR CONTRIBUTIONS

R.M.‐Z., M.D.‐R., B.V.‐A. and I.A.A.‐F. performed the in vitro experiments and data analyses. J.X. and M.C. performed the in silico analyses and interpreted the data. A.A.R.‐M. designed the study and data analyses. A.A.R.‐M. and M.C. wrote the paper and all authors revised the paper and confirmed the final version.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design and Analysis and as recommended by funding agencies, publishers and other organizations engaged with supporting research.

Supporting information

Fig. S1. Ba²⁺ and pH 5.0 sensitive currents. (A) Whole‐cell recordings in absence (control) and presence of 2 mM Ba²⁺. (B) Inside‐out recordings in absence (control) and presence of a pH 5.0 solution. Off‐line subtraction of the Ba²⁺ and pH 5.0 currents was used to subtract endogenous and leak currents.

Figure S2. Whole‐cell Kir4.1 currents inhibition by AGAs. (A‐C) representative time courses of Kir4.1 current inhibition by 100 μM gentamicin (A), 300 μM neomycin (B), and 300 μM kanamycin (C).

Figure S3. Inhibition of Kir4.1 currents by gentamicin dialyzed through the patch pipette. (A) representative temporal course of Kir4.1 current inhibition by 100 μM gentamicin dialyzed through the patch pipette. (B) Kir4.1 current traces at time 0 and after 20 min of recording with 100 μM of gentamicin in the patch pipette. (C) Normalized Kir4.1 currents recorded at −30 mV after 20 min in control (n = 7), and 100 μM gentamicin (n = 7).

Table S1 The hydrogen‐bonds formed between gentamicin‐Kir4.1 channel, kanamycin‐Kir4.1 channel and neomycin‐Kir4.1 channel.

Morán‐Zendejas R, Delgado‐Ramírez M, Xu J, et al. In vitro and in silico characterization of the inhibition of Kir4.1 channels by aminoglycoside antibiotics. Br J Pharmacol. 2020;177:4548–4560. 10.1111/bph.15214