Channel HCN4 mutation R666Q associated with sporadic arrhythmia decreases channel electrophysiological function and increases protein degradation

Mutations in the hyperpolarization-activated nucleotide-gated channel 4 (HCN4) are known to be associated with arrhythmias in which QT prolongation (delayed ventricular repolarization) is rare. Here, we identified a HCN4 mutation, HCN4-R666Q, in two sporadic arrhythmia patients with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia. To determine the functional effect of the mutation, we conducted clinical, genetic, and functional analyses using whole-cell voltage-clamp, qPCR, Western blot, confocal microscopy, and co-immunoprecipitation. The mean current density of HEK293T cells transfected with HCN4-R666Q was lower in 24 to 36 h after transfection and was much lower in 36 to 48 h after transfection relative to cells transfected with wildtype HCN4. Additionally, we determined that the HCN4-R666Q mutant was more susceptible to ubiquitin-proteasome system–mediated protein degradation than wildtype HCN4. This decreased current density for HCN4-R666Q could be partly rescued by treatment with a proteasome inhibitor. Therefore, we conclude that HCN4-R666Q had an effect on HCN4 function in two aspects, including decreasing the current density of the channel as a biophysical effect and weakening its protein stability. Our findings provide new insights into the pathogenesis of the HCN4-R666Q mutation.

Mutations in the hyperpolarization-activated nucleotidegated channel 4 (HCN4) are known to be associated with arrhythmias in which QT prolongation (delayed ventricular repolarization) is rare. Here, we identified a HCN4 mutation, HCN4-R666Q, in two sporadic arrhythmia patients with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia. To determine the functional effect of the mutation, we conducted clinical, genetic, and functional analyses using whole-cell voltage-clamp, qPCR, Western blot, confocal microscopy, and co-immunoprecipitation. The mean current density of HEK293T cells transfected with HCN4-R666Q was lower in 24 to 36 h after transfection and was much lower in 36 to 48 h after transfection relative to cells transfected with wildtype HCN4. Additionally, we determined that the HCN4-R666Q mutant was more susceptible to ubiquitin-proteasome system-mediated protein degradation than wildtype HCN4. This decreased current density for HCN4-R666Q could be partly rescued by treatment with a proteasome inhibitor. Therefore, we conclude that HCN4-R666Q had an effect on HCN4 function in two aspects, including decreasing the current density of the channel as a biophysical effect and weakening its protein stability. Our findings provide new insights into the pathogenesis of the HCN4-R666Q mutation.
Sinus bradycardia is defined as a heart rate less than 60 beats per minute (bpm) in adults (1). Asymptomatic sinus bradycardia is usually harmless and is often a sign of good physical conditioning, whereas symptomatic sinus bradycardia is associated with sick-sinus syndrome and can be a lifethreatening condition (1). Accordingly, permanent pacing is required for the majority of patients with symptomatic sinus bradycardia (2)(3)(4).
Here, we identified a novel heterozygous mutation HCN4-R666Q, which was located in the non-cAMP-binding region of CNBD, in two sporadic patients with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia. We hypothesized that HCN4-R666Q mutation was associated with the clinical phenotype of two sporadic patients and explored the mechanism of channel dysfunction caused by the mutation.

Results
HCN4 mutation R666Q was found in two sporadic patients with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia Two female patients (40 years; 1-II.1 and 45 years; 2-II.1; Fig. 1A) suffering from palpitation and syncope were admitted to Fuwai Hospital as outpatients. On arrival, the heart rate of two patients was slowed down (patient A: 46 bpm, patient B: 54 bpm), but their blood pressure was normal (patient A: 115/ 88 mm Hg, patient B: 110/80 mm Hg). According to the two patients' descriptions, they were the only child in their families. Patient A had a son and an aborted fetus, while patient B had no children. Panel sequencing showed that the two patients with clinical manifestations of sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia carried HCN4-R666Q (c.1997G > A, NM_005477) mutation in which glutamine was substituted for arginine. In addition, no other significant mutations in arrhythmia-related genes were observed in the two patients ( Table 1). Analysis of UK biobank (www.https://genebass.org/) suggested that the HCN4 p. Arg666Trp mutation was associated with susceptibility to atrial fibrillation (p = 9.7e-4), indicating the potential significance of this residue in HCN4 (Fig. 1B). Holter monitoring showed the two patients have lower average (patient A: 48 bpm, patient B: 52 bpm) and minimum (patient A: 44 bpm, patient B: 46 bpm) heart rates. 24-h Holter recordings were analyzed by a computer system and confirmed by an electrophysiologist. The patients were diagnosed as having sinus bradycardia with average heart rate <60 bpm concurrent with QT prolongation with QT-intervals >0.44 s (Patient A: QTintervals = 0.52 s; the corrected QT (QTc) corrected QT-intervals = 0.51 s; Patient B: QT-intervals = 0.48 s; QTc = 0.51 s) accompanied by short bursts of ventricular tachycardia as assessed by 24-h Holter monitoring and 12-lead ECG ( Fig. 1, C-F). Patient A had premature ventricular complexes (Fig. 1E) and two patients had no history of coronary heart disease or drug-induced QT prolongation. No abnormal changes were observed in plasma ion concentrations of two patients, including sodium, potassium, magnesium, and calcium. Echocardiography examination of two patients showed no structural heart disease (Table S1). DNA sequencing, 12-lead ECG, and Doppler echocardiography results for the two patients' parents and patient A's son showed no mutation or structural heart disease were observed (Figs. S1 and S2 and Table S1).
In silico analysis showed the mutation was pathogenic and evolutionarily conserved Sanger sequencing (Fig. 1G) showed that arginine 666 was replaced with a glutamine (R666Q) in the highly conserved CNBD of the HCN4 subunit (Fig. 1H). This mutation was not found among the 1000 genomes and EPS6500 database, and no minor allele frequency was observed in the general population in the Exome Aggregation Consortium database (21). Furthermore, the HCN4-R666Q mutation was predicted to have a high pathogenicity index in silico analysis with Mutation Taster (disease causing), PROVEAN (deleterious), FATHMM (damaging), and PolyPhen-2 (possibly damaging). A multiple alignment (ClustalX2.0.10) showed that arginine 666 was conserved across a wide range of mammals, including Homo sapiens, dog, rabbit, zebrafish, chicken, and mouse (Fig. 1I).

Reduced current density of HCN4-R666Q mutant channels
To investigate electrophysiological properties caused by the HCN4-R666Q mutation, we performed whole-cell voltageclamp recordings of HEK293T cells transfected with only wildtype HCN4 or HCN4-R666Q (homomeric) or cells that were cotransfected with wildtype and HCN4-R666Q (heteromeric). A series of hyperpolarizing steps were applied to activate functional HCN4-mediated currents ( Fig. 2A). The results showed that the amplitudes and density of wildtype HCN4 currents were significantly larger than the current amplitudes recorded from cells expressing mutant homomeric or heteromeric HCN4-R666Q channels in both 24 to 36 h and 36 to 48 h after transfection (Fig. 2, A and B). The mean current density of homomeric and heteromeric mutant channels and wildtype channels in 24 to 36 h and 36 to 48 h after transfection were also calculated. Interestingly, the mean current density of mutant channels was lower in 24 to 36 h after transfection and much lower in 36 to 48 h (Fig. 2C) compared to wildtype HCN4, respectively. The voltage for half maximal activation (V 1/2 ), which was calculated by fitting the activation curve of HCN4 currents using the Boltzmann equation, was similar for both wildtype and homomeric HCN4-R666Q channels, suggesting that the HCN4-R666Q mutation did not influence channel activation kinetics (Fig. 2, D and E and Table S2). The voltage sensitivity of activation, k (slope of the simplified Boltzmann equation) also showed no significant difference between homomeric mutant and wildtype channels (Table S2).
HCN4-R666Q mutation did not affect cAMP binding to the CNBD Modulation of HCN4 channels is mediated by cAMP binding to the CNBD. Since residue R666 was located in the HCN4 CNBD region, we tested whether the HCN4-R666Q mutation affected cAMP binding to the CNBD. 1 mM the cAMP analog 8-Br-cAMP (Sigma-Aldrich) was included into patch pipettes. After breaking into the whole-cell mode, the cells 24 h after transfection were held for 10 min at a resting membrane potential of -30 mV to stabilize the intracellular 8-Br-cAMP concentration. The values of V 1/2 of activation curves for wildtype and mutant HCN4-R666Q channels were shifted about 10 mV toward the depolarization direction (Fig. 2, F and G), whereas k values of the activation curves showed no significant difference between wildtype and homomeric HCN4-R666Q channels in the presence of 8-Br-cAMP (Table S3), indicating HCN4-R666Q mutation did not affect the cAMP dependence of channel activation. This was presumably due to HCN4-R666Q was located in the non-cAMP-binding region of HCN4 CNBD.

Reduced protein levels of HCN4-R666Q mutant
Electrophysiological analysis showed that HCN4-R666Q channels reduced the current density more significantly in 36 to 48 h after transfection, so we determined whether this change was resulted from the alteration of mRNA or protein levels. There were no significant differences in the HCN4 mRNA levels between HEK293T cells expressing only wildtype or HCN4-R666Q mutant channels at different time points (12,24,36,48 h) after transfection (Fig. 3A). However, total protein and membrane protein levels at 48 h and 72 h after transfection were significantly decreased in HEK293T cells expressing homomeric HCN4-R666Q compared to those expressing wildtypeHCN4 (Fig. 3, B and C). The localization of the wildtype and homomeric HCN4-R666Q mutant protein by confocal microscopy also suggested that protein levels of HCN4-R666Q were decreased compared to wildtype (Fig. 3D).
Increased degradation of HCN4-R666Q protein was regulated by ubiquitin-proteasome system To understand the potential mechanism of the altered protein levels of HCN4-R666Q, we examined the half-life of wildtype and HCN4-R666Q protein. In HEK293T cells treated with the protein synthesis inhibitor cycloheximide (CHX) (31), the half-life of HCN4-R666Q mutant channel protein was significantly reduced compared with wildtype channels (19.1 ± 2.59 h versus 10.7 ± 1.98 h, p = 0.011; Fig. 4A), indicating the HCN4-R666Q protein was less stable and more rapidly degraded than wildtype.
Next, we investigated the pathway mediating HCN4-R666Q protein degradation. In transfected HEK293T cells treated with autophagy inhibitors, HCN4-R666Q protein levels were not rescued, although p62 and the ratio of LC3-I to LC3-II were accumulated (Fig. 4B). This result suggested that the autophagy pathway was successfully inhibited, and HCN4 mutant protein levels were not regulated by autophagy.
We next examined whether HCN4 protein were affected by the ubiquitin-proteasome system (UPS). MG132 affected HCN4 protein levels dose-dependently, and the optimal concentration was 5 μM (Fig. 4C). Furthermore, an immunecoprecipitation assay indicated that HCN4 was ubiquitinated, and ubiquitination levels of HCN4-R666Q mutant protein were higher than that of wildtype ( Fig. 4D). Finally, we examined whether the SUMOylation, which was the reversible covalent attachment of a small ubiquitin-related modifier (SUMO) protein to a target protein, affected HCN4 protein levels. Immunoblot analysis showed that the expression levels of four SUMO isoforms (SUMO 1-4) were not significantly different between HCN4-R666Q mutant and wildtype (Fig. 4E). Taken together, these results indicated that the UPS pathway was considered as the potential mechanism of reduced HCN4-R666Q protein levels.

Proteasome inhibitor treatment rescued reduction of HCN4-R666Q protein levels and current density
To explore the therapeutic potential of targeting HCN4-R666Q mutation, we determined the effect of proteasome inhibitor MG132 on rescuing the decreased protein levels and current density of HCN4-R666Q. HEK293T cells were transfected with wildtype or the homomeric HCN4-R666Q mutant and then treated with MG132 for 12 h. The half-life of the wildtype and HCN4-R666Q protein both were prolonged in the presence of MG132 (29.0 ± 4.88 h and 17.2 ± 0.92 h, respectively; p = 0.014) (Fig. 5A). Interestingly, in the presence of MG132 treatment, HCN4-R666Q protein levels were similar to that seen for wildtype in the absence of MG132 treatment (Fig. 5B). Consistently, decreased mutant channel current density was obviously rescued by treatment of transfected cells with MG132 for 12h (Fig. 5, C-F). However, there was no significant difference in V 1/2 values after MG132 treatment, whereas the k was different apparently after MG132 treatment (Table S4).

Discussion
In the present study, we described the HCN4-R666Q mutation in two sporadic patients with sinus bradycardia, QT prolongation and short bursts of ventricular tachycardia. No other pathogenic mutations in the known arrhythmia-related genes were found in the two patients, suggesting that the loss of function of the HCN4 mutation might be associated Voltage-gated sodium channel α subunit (Nav1.5) Voltage gated L-type calcium channel alpha 1 subunit Sodium channel beta 4 subunit SNTA1 Syntrophin-alpha 1 CACNB2 Voltage-dependent L-type calcium channel subunit beta-2 PKP2 Plakophilin-2 RANGRF Ran guanine nucleotide release factor TMEM43 Transmembrane protein 43 RYR2 Ryanodine receptor 2 TRDN Triadin GPD1L Glycerol-3-phosphate dehydrogenase 1-like protein CACNB2 Voltage-dependent L-type calcium channel subunit beta-2 SCN1B Sodium channel subunit beta-1 SCN3B Sodium channel subunit beta-3 HCN4 Hyperpolarization-activated cyclic nucleotide-gated channel 4 CASQ2 Calsequestrin-2 JUP Junction plakoglobin HCN4-R666Q mutation in sporadic arrhythmia with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia. We found the HCN4-R666Q mutation decreased the channel electrophysiological function and increased its protein degradation, both resulted in reduced current density of HCN4 channels. We also showed that proteasome inhibitor might has therapeutic benefits to mitigate the effects of this HCN4 mutation (Fig. 6). Two individuals in UK biobank (www.https://genebass.org/) carried HCN4-R666Q mutation, which has not been studied. The allele frequency of the mutation in the Non-Finnish European population was 3.72e-6. Interestingly, another mutation at residue 666, HCN4 p. Arg666Trp (R666W), appears to be associated with the susceptibility to atrial fibrillation (p = 9.7e-4). The allele frequency of HCN4-R666W in the Non-Finnish European population was 2.42e-5, indicating the significance of position 666 in HCN4. Several lines of evidences further suggested that the HCN4-R666Q mutation had functional consequences and was not simply a silent polymorphism. First, within the general population, the HCN4-R666Q variant was not found in the 1000 genomes, EPS6500, or the Exome Aggregation Consortium databases. Second, the mutation occurred in a segment that was highly conserved throughout evolution. Third, this substitution decreased electrophysiological function that underlies the clinical phenotype.
Nearly all reported HCN4 mutations result in reduced current density of HCN4 channels that leads to slow heart rates. HCN4 channels are directly regulated by cAMP, which binds to CNBD of HCN4 channels and elicits a positive shift in the voltage dependence of activation. Notably, HCN4-R666Q had no dramatic impairments in cAMP activation in our study. Alternatively, the current reduction of HCN4-R666Q might be caused by a pronounced gating defect of HCN4-R666Q channels, which was independent of the cAMP modulation.
Aside from channel biophysical property, channel function was determined by the channel abundance at the plasma membrane. Our results revealed that mRNA levels were decreased for both groups as transfection time goes on. Previous studies have demonstrated that protein synthesis and degradation define protein turnover rates (32,33). Immunoblotting and confocal microscopic analysis suggested that HCN4-R666Q led to an increase in the degradation of protein, suggesting that plasma membrane abundance was reduced. Similar decreasing protein levels was reported for HCN4-A485V mutation linked to symptomatic sinus bradycardia (17). Because surface expression of normal HCN4 channels was much decreased by the co-expression of mutant HCN4 channel, the HCN4-R666Q mutation serves as a dominant-negative suppression, which was also observed in previous studies on HCN4 channel loss of function mutations (1,8,9,17,19).
The decreased channel electrophysiological function and the markedly reduced plasma membrane protein as the major causes of the mutant HCN4 channel dysfunction. Our results showed that the loss of function in HCN4-R666Q resulted from the reduced surface expression rather than impaired cAMP modulation. Unlike the previously identified HCN4-P257S mutation (5), which reduced the expression of mutant channels due to defects in protein trafficking, decreased in HCN4-R666Q at the cell membrane appear to be due to degradation of protein. Protein degradation is one crucial component governing protein turnover (34). It is well established that the lysosomal and proteasomal systems are the two major protein degradation pathways in eukaryotic cells (35)(36)(37). Autophagy induces degradation of cytoplasmic materials and organelles in lysosomes, which plays an important role in maintaining cellular homeostasis (38,39). The UPS plays an important role in the regulation of protein stability, which consists of ubiquitin-activating enzyme (E1), ubiquitinconjugating enzyme (E2), and ubiquitin ligase (E3) that delivers ubiquitin from E2 to the specific substrates (40,41). UPS pathway is involved in the turnover of multiple cellular proteins, enabling cells to dispose of biologically nonuseful proteins, including mutant, misfolded, and over accumulated proteins (42,43). Our results revealed that increased degradation of HCN4-R666Q protein was regulated by the UPS rather autophagy pathway. The UPS, which is the major degradation pathway of intracellular proteins including voltage-gated channels (44,45), is responsible for the degradation of most intracellular cytosolic, nuclear and membrane proteins, pivotal to both protein quality control and the regulatory degradation of normal proteins essential to virtually all cellular processes (46)(47)(48).
There is currently no targeted therapy for patients carrying HCN4 mutations. Treatment of HEK 293T cells with the proteasome inhibitor MG132 partly rescued the decreased protein and current density of the mutant. Previous study has showed that HCN4 between cardiomyocytes and HEK293 cells exhibits a similar difference in V1/2 values, indicating that the factor contributing to the difference in voltage dependence affects both isoforms equivalently (49). Therefore, our discoveries can be translated to myocytes, which suggested that modulation of proteasome activity could be further explored as potential targeted therapy for patients with this mutation.
Proteasome activity is required for diverse cellular processes, including transcriptional and epigenetic regulation. MG132, which is a specific proteasome inhibitor, plays an essential role in inducing cell cycle arrest and apoptosis (50)(51)(52). Proteasome inhibition enhances antiproliferative while dampening cell-proliferative gene expression programs (53). Furthermore, MG132 has been determined to have anticancer effects in U2OS cells (54). In addition, proteasome inhibitors are promising antitumor drugs with preferable cytotoxicity in malignant cells and have exhibited clinical efficiency in several hematologic malignancies (55). It may also reduce the occurrence of ventricular arrhythmia such as heart failure (56). Our study revealed that MG132 may partly rescue the arrhythmia caused by HCN4-R666Q mutation. However, the tolerance and sensitivity of MG132 was different in different types of cells. MG132 is expected to affect many cellular processes, so targeted delivery is important for its therapeutic benefits.
Our study had some limitations. Firstly, we verified the abnormal expressional levels only in the HEK293T cell model transfected with HCN4-R666Q, it should be better if one more kind of cell models was used. Secondly, the patients were not followed-up due to the patients' wishes, thus more clinical data were not received. Thirdly, the patch clamp experiment could only be performed within 48 h after transfection due to the current was unstable caused by poor cell status 72 h after transfection. Finally, HCN4 is the most abundant HCN isoform expressed in the human SA node and the atria. It would have been informative to determine if the HCN4-R666Q mutation segregated with QT prolongation, but the two patients refused our later follow-up.
In summary, the HCN4-R666Q mutation was found in two patients with sinus bradycardia, QT prolongation, and short bursts of ventricular tachycardia. This mutation resulted in a loss function of HCN4 channels that manifested as reduced I f current by both decreased channel electrophysiological function and increased protein degradation. The reduced I f current could be partly rescued by treatment with proteasome inhibitor. Our findings provided new insights into the pathogenesis of arrhythmias related to HCN4 mutations.

Patients and clinical investigations
Existing Phenotype In 2013 to 2016, two female patients (40 years; Patient A and 45 years; Patient B) suffering from palpitations and syncope were admitted to Fuwai Hospital as outpatients.

Family history
According to the two patients' descriptions, both patients were normally healthy and had no medical history and familial cardiac events, sudden cardiac death, or other diseases and did not take drugs that could prolong QT.

Investigations
Patients were evaluated by clinical examination, 12-lead echocardiography, 24-h Holter monitoring, and Doppler echocardiography. The two patients' parents and patient A's son were evaluated by 12-lead ECG and Doppler echocardiography.

Follow-up
The patients were not followed-up in accordance with the patients' wishes. The study was approved by the ethics committee of Fuwai Hospital, Peking Union Medical College (Approval No. 2012-400), Beijing, and conformed to the principles outlined in the Declaration of Helsinki. Written informed consent was obtained from each patient.
To analyze the effect of the mutation in cells, we constructed plasmids with wildtype or mutated HCN4. Human HCN4 cDNA (GenBank: NM_005477) was subcloned into GV230 vector containing a C terminally fused enhanced green fluorescent protein. Subsequently, the HCN4-R666Q mutation was introduced using the QuikChange II Site-Direct Mutagenesis Kit (Stratagene) into GV230 vector with appropriate mutagenic primers. All constructs were verified by Sanger sequencing.

Electrophysiological recordings
HCN4 channel-mediated currents were recorded in HEK293T cells 24 to 48 h after transfection using conventional whole-cell patch clamp techniques at room temperature (21-23 C) as described (9,11,60). Signals were amplified using a HEKA EPC10 amplifier with the Patch-Master software (HEKA Instruments, Inc) Data were lowpass filtered at 2 kHz and acquired at 20 kHz. Patch electrodes had a resistance of 5 to 9 MΩ when filled with the pipette solution containing (in mmol/L) 115 KMeSO 4 , 1 NaCl, 0.1 CaCl 2 , 1 EGTA, 3 MgCl 2 and 10 Hepes (pH 7.2 with KOH). The bath solution contained (in mmol/L): 60 KCl, 75 NaCl, 10 glucose, 2 CaCl 2 , 1 MgCl 2 , 1.2 NaH 2 PO 4 , and 10 Hepes (pH 7.3 with NaOH). Intracellular and extracellular solution osmolarity was adjusted with glucose to 290 and 300 mOsmol/L, respectively. Series resistances were electronically compensated by 60%. To evaluate functional properties, HCN4 currents were evoked by hyperpolarizing steps ranging from -40 to -130 mV (holding potential: -30 mV, increment: 10 mV, interval: 5 s). For drug application experiments, the transfected HEK293T cells were treated with 5 μM MG132 (a proteasome inhibitor) for 12 h, and current density was determined by dividing the fully activated current at -120 mV at the end of the 5s pulse by cell capacitance. The voltage dependence of channel activation was analyzed by fitting the Boltzmann function (I -I min )/(I max -I min ) = 1/(1 -exp [(V -V 1/2 )/k]), where V 1/2 is the half-activation voltages, and k is the inverse slope factor.

Expression analysis
At 12, 24, 36, and 48 h after transfection with wildtype or HCN4-R666Q, HEK293T cells were washed three times with ice-cold PBS, and total RNA was isolated using Trizol (Invitrogen). Then, 1 μg of total RNA was reverse transcribed using Hifair II 1st Strand cDNA Synthesis Super Mix. Real-time PCR assays were run in triplicate and quantified using the Prism 7500 sequence-detection system (ABI) with Hieff qPCR SYBR Green Master Mix. Relative mRNA expression was calculated by the ΔΔCt method, and the values were normalized to the expression of actin beta (ACTB). Primers used: HCN4 forward 5 0 -TGGACACCGCTATCAAAGTGG-3 0 and HCN4 reverse 5 0 -CTGCCGAACATCCTTAGGGA-3 0 .
For immunoblotting, HEK293T cells at 36, 48, and 72 h after transfection were washed three times with ice-cold PBS, and total proteins were lysed with RIPA buffer and membrane proteins were extracted using a Minute Plasma Membrane Protein Isolation and Cell Fractionation Kit (Invent Biotechnologies). Protein samples were denatured by heating to 37 C for 30 min and then used for immunoblot analyses. ACTB was served as a total protein control and Na, K-ATPase (CST, 1:1000) was used as a membrane protein control. All data shown represented at least three independent experiments.

Immunofluorescence confocal microscopy analysis
Immunofluorescence assay was performed as described previously (8,21). At 36, 48, and 72 h after transfection, HEK293T cells were grown on microscope slides and fixed in 4% preheated paraformaldehyde for 10 min. The fixed cells were rinsed three times with PBS, permeabilized with 0.1% Triton X-100 (Sigma-Aldrich) for 15 min, blocked for 1 h in 10% goat serum, and then incubated with HCN4 antibody (Abcam) overnight at 4 C. The cells were then incubated with an Alexa Fluor 488-conjugated goat anti-rat antibody (Invitrogen) for 1 h in the dark and washed with PBS before DAPI staining and mounting in the dark. The cells were analyzed by fluorescence confocal microscopy (Leica). Identical parameters were used for image acquisition and analysis. A threshold was set for each image to eliminate background.

Determination of HCN4 protein half-life
Protein half-life assay was performed as described previously (31). At 36 h after transfection, HEK293T cells were incubated with 25 μg/ml CHX (an inhibitor of protein synthesis) (61)for 12 h. Then, total protein was collected at different time points (0, 5, 10, 12, 15, and 20 h) for immunoblot analysis to monitor the HCN4 protein expression levels. Band intensities for immunoblotted HCN4 protein were quantified using the Image J software (National Institute of Health) and normalized to controls. The relative levels of HCN4 protein in sample not treated with CHX was considered as initial level which were referred to as 1 unit.

Co-immunoprecipitation analysis
Co-immunoprecipitations were performed as previously described (66)(67)(68). Briefly, HEK293T cells were grown on 10 cm dishes and 48 h after transfection, washed three times with ice-cold PBS, and lysed with ice-cold IP lysis buffer containing 1× protease inhibitors for 30 min on ice followed by ultrasonic lysing for 1 min. The cell lysates were incubated for 60 min and then centrifuged at 10,000g at 4 C for 20 min. The lysates were precleared by incubation with protein A/Gagarose (Proteintech) for 1 h at 4 C. Specific antibodies were incubated overnight at 4 C followed by protein-A/G agarose beads incubation at 4 C for 4 h. After the beads were washed three times with washing buffer, the immunoprecipitation was heated in 2× sample buffer at 37 C for 30 min and then separated by SDS-PAGE. The separated proteins were analyzed via immunoblot.

Statistical analysis
Data are reported as the mean ± SD. Statistical analysis of the results of >2 groups was carried out by one-way ANOVA followed by the least significant difference test or Tukey test. Student's t test was performed for comparison of two groups only. p < 0.05 was considered statistically significant; 2-tailed tests were used for all analyses. Analysis of the Western blot bands was performed with ImageJ image analysis software. All measurements were obtained from distinct samples. Electrophysiological data were analyzed with patchmaster software. All statistical analyses were performed using GraphPad Prism 8.0.

Data availability
Raw data will be available upon request.
Supporting information-This article contains supporting information.