TMEM43 was initially identified in silico as one of the 368 transmembrane (TMEM) proteins with unknown functions [1]. TMEM43 was shown to be expressed throughout the mouse body, including the cochlea, heart, eye, brain, and kidney [2]. Mutations in TMEM43 have been studied in human diseases such as arrhythmogenic right ventricular cardiomyopathy type 5 (ARVC5) in the heart [3-6] and auditory neuropathy spectrum disorder (ANSD) in the cochlea [2]. A p.(Ser358Leu) missense mutation in the TMEM43 has been shown to alter intercalated disc and Cx43 protein expression [4]. Moreover, microelectrode array recordings from HL-1 Cardiac Muscle Cells revealed that TMEM43-p.(Ser358Leu) expressing cells show reduced conduction velocity and asynchronous beating rhythms compared to TMEM43-WT expressing cells [4]. Similarly, a p.(Arg372Ter) nonsense mutation in the TMEM43 disrupts connexin-linked function in the cochlear supporting cells, which is to maintain inner ear homeostasis via recycling and regulating intracellular K⁺ [2, 7, 8]. Indeed, the TMEM43-p.(Arg372Ter) mutant knock-in mice display a significantly reduced passive conductance current from cochlear glia-like supporting cells (GLSs) [2]. Collectively, TMEM43 is suggested to be essential for mediating background conductance and electrical communication in heart and cochlear cells.

The passive conductance channels, which give rise to a low membrane resistance and linear I-V relationship, are known as the unique property that defines mature astrocytes in the brain [9-11]. Our group has formally revealed the molecular identity of the astrocytic passive conductance channels as a dimer of TWIK-1 and TREK-1 subunits of the two-pore potassium (K2P) channels [12]. In the brain, passive conductance channels are critically involved in K⁺ ion homeostasis [13] and volume regulation [14, 15]. In the cochlea, the endocochlear potential of +80 mV is generated by maintaining a high K⁺ concentration in the endolymph, and this potential is critical for the activation of hair cells [16]. Therefore, various channel proteins in GLSs should be thoroughly examined to understand their role in establishing and maintaining K⁺ ion homeostasis and their associated role in hearing and speech discrimination. Furthermore, the TMEM43-mediated large passive conductance current in GLSs was reminiscent of the typical passive conductance current observed in hippocampal astrocytes [2]. Accordingly, we hypothesized that there is a possibility of TMEM43 protein interacting with one of the K2P channels in the cochlea.

There are 15 known members of the K2P family, each containing two-pore (P) loops and four transmembrane (TM) domains [17]. They have a unique extracellular cap structure formed by two TM1-P1 linkers, containing a cysteine for dimerization [18, 19]. Interestingly, this cysteine residue is not present in the TASK-1/3/5 subfamily [19], suggesting a disulfide-bridge-free binding in the TASK channels. In addition, the K2P channels are regulated by diverse stimuli such as oxygen tension, pH, lipids, mechanical stretch, neurotransmitters, and G protein-coupled receptors [20]. Among the various 15 K2P channels, we searched for candidate K2P subunits that would most likely interact with TMEM43. In the end, we narrowed it down to KCNK3 (TASK-1), which exhibits the most similar properties as the TMEM43-involved currents, sensitive to external acidosis and displaying linear rectification [2, 21]. Although TMEM43 is expected to play critical roles in mammals, studies on this protein are sparse. In this study, we provide a piece of additional information regarding TMEM43 protein that it interacts with one of the K2P channels, KCNK3 (TASK-1), in the cochlea.

Animals

All experimental procedures were conducted according to protocols approved by the directives of the Animal Care and Use Committee of the Institutional Animal Care and Use Committee of IBS (Daejeon, Republic of Korea). C57BL/6 mouse pups of Postnatal day 5~8 (p5~p8) were purchased from Raon Bio (Youngin, Republic of Korea) or DBL (Eumseong, Republic of Korea) and were sacrificed on the delivered day.

Primary neuron culture

C57BL/6 pups of embryonic day 15 were used. Embryos were dissociated from amniotic sac, and fetuses were put into a sterile petri dish with HBSS. They were decapitated, and skull and meninges were removed on ice. Only cortex was collected in 50 ml tube with HBSS and washed 3 times with 20 ml of HBSS to clear the debris. The cortex was incubated in 5 ml HBSS containing 0.1% trypsin for 5 min in a 37℃ water bath (shaking every 1 min). Afterward, 5 ml FBS (1X volume) was added to the tube, and the cortex was dissociated by pipetting. The cells were collected using a cell strainer (40 μm) and centrifuge (800 g, 3 min). The supernatant was removed, and the cell pellet was resuspended in Neurobasal A medium (10888022, Gibco). 1×10⁴ cells were plated on 0.1 mg/ml Poly-L-Ornithine hydrobromide (PLO) (P3655, Sigma-Aldrich)-coated 24 well plates and incubated in the incubator (37℃, 5% CO₂). The neuron culture was maintained by replacing the half of medium with fresh medium: Neurobasal A medium+2% B27 (17504-044, Gibco)+1% Glutamax (35050-061, Thermofisher Scientific)+0.5% P/S (250 ul/50 ml) (SV30010, Hyclone) every 2~3 days.

Plasmids

TMEM43 (Myc-DDK-tagged)-Human transmembrane protein 43 (TMEM43) (NM_024334.2) was purchased from OriGene (RC200998), KCNK3 (Myc-DDK-tagged) (TASK-1) (NM_002246.3) was purchased from OriGene (RC215155), GJB2 (NM_004004) Human Tagged ORF Clone was purchased from OriGene (RC202092), and Cx30-msfGFP was purchased from Addgene (69019). The coding sequences of these genes were cloned into CMV-MCS-IRES2-EGFP or CMV-MCS-IRES2-dsRed plasmid vector using BglII/XmaI sites. TMEM43 truncation mutations were made using the EZchange site-directed mutagenesis kit (EZ004S, Enzynomics).

Heterologous expression of TMEM43 on HEK293T cell lines

Human embryonic kidney (HEK) 293T cells were purchased from ATCC (CRL-3216). The cell line has been tested for mycoplasma contamination. HEK293T cell was cultured in DMEM (10-013, Corning) supplemented with 10% heat-inactivated fetal bovine serum (10082-147, Gibco) and 10,000 units/ml penicillin-streptomycin (15140-122, Gibco) at 37℃ in a humidified atmosphere of 95% air and 5% CO₂. According to the manufacturer’s protocol, the transfection of expression vectors was performed with Effectene Transfection Reagent (Effectene, 301425, Qiagen). One day before performing the experiments, HEK293T cells were transfected with each DNA 1 μg per 35 mm dish.

Co-immunoprecipitation (IP) and western blot

DNA transfected HEK293T cells, or cochlear tissues of an average of 20 mice (p5~p8) per sample were extracted and homogenized with Pierce^TM IP Lysis Buffer (87787, Thermo Fisher Scientific) and Halt^TM Protease and Phosphatase.

Inhibitor Cocktail (100X) (78446, Thermo Fisher Scientific). Equal amounts of precleared cell lysates were incubated with mouse anti-TASK-1 (KCNK3) monoclonal (1 μg, ab186352, Abcam) antibody or control IgG (1 μg) overnight. Protein A/G-Agarose beads (20422, Thermo Fisher Scientific) were added to the mixtures and further incubated with rotation for 1 h at 4℃, followed by a wash with lysis buffer. Bound proteins were eluted from the beads with SDS-PAGE sample buffer, and western blotting was performed with rabbit anti-FLAG (1:500, 2368s, Cell Signaling) or rabbit anti-TMEM43 (1:100, NBP1-84132, Novus).

Duolink^® proximity ligation assay (PLA)

Duolink^® In Situ Red Starter Kit Mouse/Rabbit (DUO92101, Sigma-Aldrich) and far-red detection kit (DUO92013, Sigma-Aldrich) were used. On the experimental day, cochlea tissue (p5~p8) was obtained, fixed with 4% paraformaldehyde, and washed with 0.3% Triton-X100 containing PBS. After blocking, the sample was incubated with rabbit anti-TMEM43 polyclonal (1:100, NBP1-84132, Novus) and mouse anti-TASK-1 (KCNK3) monoclonal (1:100, ab186352, Abcam) at 4℃, overnight. The next day, the sample was incubated with anti-rabbit MINUS and anti-mouse PLUS probe, ligase, and polymerase sequentially. DNA strands participate in rolling circle DNA synthesis only when two probes are in close proximity (<40 nm). Fluorescent-labeled complementary oligonucleotide probes were observed under Zeiss confocal microscopy.

Construction and packaging of shRNA–lentiviral vectors

For Task-1 gene silencing, three candidates for mouse Task-1 shRNA (GenBank accession no. NM_010608) sequences were cloned into lentiviral pSicoR vector using XhoI/XbaI sites, as previously described [22]. The candidate shRNA containing pSicoR vector was packaged into high-titer lentivirus by the IBS virus facility (Republic of Korea). The target regions of the 3 candidate Task-1 shRNAs are: 5’-ggtgaacgcatcaacaccttc-3’, 5’-ggactttcttccaggcctatt-3’ and 5’- gccttcagcttcgtgtacatc-3’. Candidate #2 (5’-ggactttcttccaggcctatt-3’) was used in the study. For control shRNA, scrambled sequence 5’-tcgcatagcgtatgccgtt-3’ was inserted in place of shRNA sequence. Knockdown efficiency was determined by RT-PCR data analysis with ImageJ. We analyzed the transcripts of Task-1 and Gapdh by primer pairs as follows: the forward primer (5'- atcaccgtcatcaccaccat-3') and the reverse primer (5'-acgaaggtgttgatgcgttc-3') for Task-1, and the forward primer (5'-tgatgacatcaagaaggtggtgaag-3') and the reverse primer (5'-tccttggaggccatgtaggccat-3') for Gapdh. When calculating the knockdown efficiency, the virus infection efficiency was considered by dividing the knockdown efficiency with infection efficiency.

Cochlear organotypic culture and shRNA treatment

Neonatal mouse (C57/BL6, P5) cochlear turns were isolated in ice-cold sterile Hanks’ Balanced Salt Solution (HBSS). Cochlear segments were attached on Cell-Tak (354240, Corning) coated coverslips and incubated overnight in DMEM/F12 medium containing 1% fetal bovine serum, 5 µg/ml ampicillin, B27 and N2 (37°C, 5% CO₂ humidified incubator). Upon confirming stable attachment, the tissues were treated with either control or shRNA carrying lentivirus diluted in culture medium (1:1,000) for 48 h. The medium was then replaced with a fresh one. After an additional 24 h of incubation, the tissues were subjected to further examination.

Immunocytofluorescence on cochlear tissue

The virus infection on cultured cochlea tissue was confirmed via the presence of mCherry signal. For immunostaining, cochlea tissue was fixed in ice-cold 4% paraformaldehyde, for 10 min and incubated in blocking/permeabilizing buffer (PBS with 5% goat serum and 0.25% Triton-X100). Then, the preparations were incubated overnight at 4°C with rabbit anti-TMEM43 polyclonal (1:100, NBP1-84132, Novus) and mouse anti-TASK-1 (KCNK3) monoclonal (1:100, ab186352, Abcam) diluted in the blocking/permeabilizing buffer. After 3 washes, the cochlear turns were reacted with donkey anti-rabbit Alexa Fluor 594 (1:500, 711-585-152, Jackson ImmunoResearch) donkey anti-mouse Alexa Fluor 488 (1:500, 715-545-150, Jackson ImmunoResearch) in blocking/permeabilizing buffer for 1 h at room temperature. Samples were then rinsed once with blocking/permeabilizing buffer and twice with PBS. Using Fluorsave reagent (345789, Calbiochem), the tissues were mounted on glass slides and covered with a coverslip. Specific immunolabeling was examined under Zeiss confocal microscope. No immunoreactivity was found when the primary antibodies were omitted.

Electrophysiological recording in the cochlear supporting cells

Cultured cochlea tissue infected with lentivirus was used. Recordings were done in HEPES buffer containing (mM): 144 NaCl, 5.8 KCl, 1.3 CaCl₂, 2 MgCl₂, 10 HEPES, 0.7 NaH₂PO₄, and 5.6 D-glucose (pH 7.4 was adjusted with NaOH). Glia-like supporting cells located below the inner hair cell layer were whole-cell patch clamped, and current traces were elicited by 1-sec ramps descending from +100 mV to -100 mV with -60 mV holding potential. Recording electrodes (7~11 MΩ) supplemented with (mM): 126 K-Gluconate, 5 HEPES, 0.5 MgCl₂, and 10 BAPTA (pH adjusted to 7.3 with KOH) were advanced through tissue under positive pressure. Slice chamber was mounted on the stage of an upright Hamamatsu digital camera viewed with a 60X water immersion objective with infrared differential interference contrast optics using Imaging Workbench Software. Electrical signals were digitized and sampled with Digidata 1320A and Multiclamp 700B amplifier (Molecular Devices) using pCLAMP 10.2 software. Data were sampled at 10 kHz and filtered at 2 kHz. Glass pipettes were pulled from a micropipette puller (P-97, Sutter Instrument), and all experiments were conducted at room temperature 20~22℃.

Data analysis and statistical analysis

Off-line analysis was carried out using Clampfit version 10.4.1.10 and GraphPad Prism version 7.02 software. When comparing between two samples, the significance of data was assessed by Student’s two-tailed unpaired t-test. One-way ANOVA with Kruskal-Wallis test with Dunn's posthoc test was used to compare 3 samples, as the data did not pass the normality test. Significance levels were given as: N.S. p>0.05, *p<0.05, **p<0.01, ***p<0.001 and ^#p<0.0001.

TMEM43 interacts with TASK-1 in the in vitro system

In order to examine whether TMEM43 interacts with TASK-1, human TMEM43 and TASK-1 genes were cloned into an internal ribosome entry site (IRES2) vector. The IRES sequences can translate two independent proteins under the same promoter in a cap-independent mechanism via recruiting the eukaryotic ribosome to the mRNA [23]. Unlike fluorescent-tagged proteins, which may have disturbed protein trafficking or interaction, the fluorescent reporter molecules are expressed without affecting TMEM43 or TASK-1 protein in the IRES system. Using this tool, we heterologously expressed TMEM43 and TASK-1 together in the HEK293T cell line and performed a co-immunoprecipitation (IP) assay (Fig. 1A). The cell lysate was pulled down with TASK-1 antibody and blotted with anti-FLAG to detect TMEM43 protein. A positive band appeared at the size of TMEM43 protein, indicating that TMEM43 protein was immunoprecipitated with TASK-1 (Fig. 1B). As the previous study has shown interaction of TMEM43 with Cx26 and Cx30 proteins [2], we next examined if TASK-1 also interacts with the connexins (Fig. 1A). The positive western blot band for connexin protein was only present when the TASK-1 was co-expressed with Cx26 but not with Cx30 (Fig. 1C). This result demonstrates that TASK-1 interacts with Cx26 protein.

To further examine the direct interaction of TMEM43 and TASK-1 in the in vitro system, we implemented a Duolink proximity ligation assay (PLA). This tool allows visualizing protein-protein interaction with high specificity and sensitivity; hybridization of oligonucleotide arms of the PLA probes creates a template for rolling circle amplification when the epitopes of the target proteins are in a close proximity (<40 nm) (Fig. 1D) [24]. Consistent with the co-IP results, we observed a positive PLA signal with TMEM43 and TASK-1 antibodies in the TMEM43 and TASK-1 co-expressed cells, but not with TMEM43 antibody alone (Fig. 1E), highly indicating close proximity of the two proteins. Taken together, these results demonstrate that TMEM43 and TASK-1 can bind in the heterologous expression system.

The intracellular loop domain of TMEM43 is necessary for TASK-1 interaction

We next investigated the interacting domain of TMEM43 that is required for interaction with TASK-1. We predicted protein structure of TMEM43 with AlphaFold, the state-of-the-art deep neural network-based method with atomic level accuracy (Fig. 2A) [25]. The bioinformatic analyses predicted TMEM43 protein with 4 TM domains and a large intracellular loop (Loop1) containing an intramembrane (IM) domain [2, 3, 25]. Based on the predictions, we firstly made a deletion mutation of Loop1, which lies between TM1 and TM2 region, and linked the two TM with three GGS linkers (Fig. 2B). The co-IP assay revealed that the interaction between TMEM43-ΔLoop1 and TASK-1 was disrupted (Fig. 2C). We next targeted the IM domain (Fig. 2B) and found that the interaction between TMEM43-ΔIM with TASK-1 was intact (Fig. 2D). These findings indicate that TMEM43 intracellular loop (Loop1) is necessary for TASK-1 interaction. The exact binding domain at TMEM43 Loop1, other than IM domain, should be further examined.

TMEM43 interacts with TASK-1 in the mouse cochlea

We subsequently examined TMEM43 interaction with TASK-1 in the cochlea tissue. The cochlea tissue was obtained from the inner ear of C57BL/6 pups (Fig. 3A). Consistent with the heterologous expression system results, immunoprecipitates from cochlear homogenates with anti-TASK-1 antibody showed a positive TMEM43 signal (Fig. 3B). Interestingly, the co-IP band size for TMEM43 was at around 60 kDa in the cochlea tissue (Fig. 3B), which was different from 47 kDa in the heterologous system (Fig. 1B). This discrepancy of size gap may arise from the absence of many proteins required for proper protein processing (such as chaperone and trafficking proteins) in cell lines but not in native environments. Furthermore, it has been demonstrated that co-translational and post-translational modification, in particular glycosylation, can also affect many proteins in membrane trafficking and functional gating properties [26, 27]. In fact, we have observed two TMEM43 bands, at size 43 kDa and 60 kDa from the cochlea lysates. However, after co-IP with anti-TASK-1 antibody, the major TMEM43 band appeared at 60 kDa but not at 43 kDa. This result is different from the result of the previous study [2], which showed TMEM43 band size at 43 kDa after co-IP with anti-Cx30 antibody. Therefore, these results together suggest that Cx30 likely interacts with the native TMEM43 (43 kDa) while TASK-1 preferentially associates with the post-translationally modified form of TMEM43 (60 kDa). Identification of these possible chaperones and membrane trafficking proteins or post-translational modifications of TMEM43 in GLSs awaits future investigations.

We next performed immunohistochemistry with the two antibodies to analyze TMEM43 and TASK-1 protein expression in the intact cochlea tissue. The confocal micrographs of the organs of Corti demonstrated a strong co-localization of TMEM43 and TASK-1 immunoreactivities in the cochlear GLSs (Fig. 3C). Furthermore, the Duolink PLA was done in the cochlea tissue using anti-TMEM43 and anti-TASK-1 antibodies. Similar to the in vitro results (Fig. 1E), a positive PLA signal was observed in the cochlea only when both antibodies were present (Fig. 3D). Thus, from the above ex vivo cochlea data, we could conclude that the endogenous TMEM43 protein can interact with the native TASK-1 protein in the mouse cochlea tissue.

TASK-1 is involved in the cochlear passive conductance current

TMEM43 has been reported to mediate the passive conductance current in the cochlear GLSs [2]. Accordingly, we tested if TASK-1 is also involved in the cochlear passive conductance current. To begin with, we developed a mouse Task-1 shRNA to genetically knockdown Task-1 mRNA. A total of 3 candidate shRNAs to target Task-1 were developed and packaged into lentivirus (Fig. 4A). We then infected the cultured neurons with each candidate shRNAs to reduce Task-1 expression (Fig. 4A). Among the 3 candidates, we chose candidate number 2, as both knockdown efficiency and virus expression were highest compared to the other two candidates (Fig. 4A). The Task-1 shRNA-carrying lentivirus could effectively infect cultured cochlea tissue and successfully ablate TASK-1 protein expression (Fig. 4B). In order to measure TASK-1 mediated current in the cochlear GLSs, we cultured cochlea tissue, infected with Task-1 shRNA-carrying lentivirus and did whole-cell patch-clamp recordings on the virus-infected GLSs (Fig. 4C). As a result, gene-silencing of Task-1 led to a significant loss of passive conductance current in the cochlear GLSs (Fig. 4D, E). These results provide possibility that TASK-1 is a potential contributor to the passive conductance current in GLSs.

In conclusion, our findings provide molecular and physiological lines of evidence manifesting that TMEM43 protein interacts with a K2P channel, TASK-1. The congruent results from Co-IP and Duolink PLA in both in vitro overexpression systems and ex vivo cochlea tissue strengthens our hypothesis that the TMEM43 protein may interact with a K2P channel in the cochlea. Furthermore, the electrophysiological recordings from cochlear GLSs revealed that TASK-1 also has chance to mediate the large K⁺ current in the cochlea. One of the major functions of cochlear GLSs, is to regulate cochlear K⁺ homeostasis via leak K⁺ channels and gap junction channels [28]. As this function of K⁺ homeostasis by GLSs is critical for hearing, we expect to see further studies regarding cochlear K2P channels. Lastly, because TMEM43 is ubiquitously expressed in mammals and has detrimental dominant-negative effects on human diseases, detailed research on this protein is required. With the new information provided from this study, we anticipate active investigations concerning TMEM43 protein on background passive conductance currents and human diseases.

This work was supported by the Institute for Basic Science (IBS), Center for Cognition and Sociality (IBS-R001-D2) to C.J.L. The graphical abstract was created with BioRender.com.