Biochemical assessment of α-α-subunit interactions of Nav1.5 in a heterologous expression system

Voltage-gated sodium channel; Na v1.5; dimer; oligomerization; protein 48 aggregation; heterologous expression system 49 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 3

50 Human voltage-gated sodium (Na+) channels (VGSCs) represent a family of proteins 51 consisting of one pore-forming α-subunit (Nav1.x) and two distinct auxiliary β-subunits 52 (Navβ) (Shen et al., 2019) The pore- forming α -subunit is comprised of four 53 transmembrane domains (I–IV), each of which has six transmembrane segments (S1–54 S6), where S1–S4 represent the voltage- sensor and S5 and S6 form the pore (Li et 55 al., 2021). N- and C-terminals, as well as linkers between the transmembrane domain 56 (hereafter named L1, L2, and L3) are intracellular (Li et al., 2021). The most 57 distinguishable feature of VGSCs is their fast permeation of Na + current ( INa) in 58 response to membrane depolarizations of few millivolts (mV) , followed by rapid 59 inactivation. Therefore, VGSCs are first in line to respond to an initial depolarization 60 and, hence, are the key regulators of cellular excitability. Within various tissues and 61 cell types (e.g., neurons, neuromuscular junctions, and cardiomyocytes), VGSCs have 62 been observed to form clusters (Eshed-Eisenbach and Peles, 2021). The clustering of 63 VGSCs is believed to enlarge their functional output, i.e. , increase membrane 64 responsiveness to electrical stimuli and conduction (Dixon et al., 2022; Eshed-65 Eisenbach and Peles, 2021) . However, it remains unclear whether the clustering of 66 VGSCs is supported by direct interactions between its subunits. 67 While Navβ-subunits were shown to homo- and hetero-oligomerize (Bouza and Isom, 68 2017), α-subunits were thought to present as singular protein units clustered together, 69 mainly due to specific membrane targeting (Marchal and Remme, 2022). Not long ago, 70 this dogma was challenged by findings reporting direct α –α-subunit interactions . 71 Specifically, human Nav1.1, Nav1.2, Nav1.5, and Nav1.7 were shown to homodimerize 72 when heterologously expressed without Na vβ-subunits (Clatot et al., 2012, 2017; 73 Rühlmann et al., 2020; Mercier et al., 2012; Zheng et al., 2021; Iamshanova et al., 74 2024). Importantly, the bacterial α-subunit of VGSC was previously demonstrated to 75 assemble as a dimer -of-d imers, consequently forming a functional ion-conducting 76 homotetrameric channel (Payandeh et al., 2011). Domain-swapping tetramerization of 77 the bacterial α-subunit of VGSC was reportedly achieved by hydrophobic, polar, and 78 charged amino acid residues mainly located within the S5 and S6 helices (Payandeh 79 and Minor, 2015) . Higher-order oligomers were not observed for the prokaryotic α-80 subunit of VGSC until a recent study suggested inter-channel dimerization may occur 81 through the voltage-sensing domain in the resting state (Sumino et al., 2023) . In 82 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 4 contrast, human Na v1.5 reportedly dimerizes through its intracellular L1 region 83 between domains I and II (Clatot et al., 2017) . L1 is a subject of multiple post -84 translational modifications, in particular, the α–α-subunit interaction region described 85 for Na v1.5 ( i.e., Arg493-Arg517) is comprised of at least one methylation site and 86 several phosphorylation sites (Marionneau and Abriel, 2015) . However, nothing is 87 known yet about the dependence of Na v1.5 dimerization on its activation state or 88 relationship to post-translational modifications. 89 Biosynthesis of Na v1.5 is a multistep fine -tuned process starting with SCN5A 90 transcription, followed by the production of nascent peptide in the endoplasmic 91 reticulum (ER), its maturation, and subsequent anchoring to the plasma membrane 92 (Dong et al., 2020). Typically, faulty and nonfunctional channels are trapped in the ER, 93 triggering the unfolded protein response (UPR) that leads to the degradation of 94 unwanted protein cargo (Wiseman et al., 2022). UPR is especially important for adult 95 cardiomyocytes, where native Na v1.5 is abundant, to maintain cellular functionality 96 because their regenerative potential is significantly low (Liu and Dudley, 2018). Folding 97 of Nav1.5 is regulated by the ER quality control machinery, comprised of chaperones 98 and folding enzymes (Dong et al., 2020). As the nascent polypeptide chain enters the 99 ER, a core glycan is added to an Asn at a specific N-glycosylation site (Wiseman et 100 al., 2022). Monoglucosylated N-linked glycans are substrates for the lectin chaperones 101 calnexin and calreticulin, which assist in oxidative folding (Wiseman et al., 2022) . 102 Another family of ER-residing folding enzymes is the protein disulfide isomerase 103 family, members of which catalyze disulfide formation, isomerization, or reduction 104 between juxtaposed Cys (Wiseman et al., 2022). Thus, rapid detection and elimination 105 of unfolded nascent peptides in the ER are complicated due to their shared properties 106 with folded unmature proteins, such as exposed hydrophobic patches or aggregation-107 prone regions (Wiseman et al., 2022) . Accordingly, accumulated misfolded proteins 108 can aggregate into larger structures , leading to ER stress and UPR activation 109 (Wiseman et al., 2022) . Given the complexity of membrane protein folding, one can 110 hypothesize that in a heterologous expression system, where the host ER machinery 111 m ay not be fully tailored for proper Na v1.5 processing, at least some portion of 112 overexpressed channels may aggregate. Importantly, at this point , misfolded Nav1.5 113 aggregates are difficult to distinguish from true Na v1.5 oligomers because they also 114 contain direct α-α-subunit interactions. 115 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 5 Taken together, the aim of our study was to critically assess α-α-subunit interactions 116 of Nav1.5 in a heterologous expression system, striving for a better understanding of 117 the VGSC dimerization phenomenon. 118 Experimental procedures 119 cDNA constructs 120 Template cDNA constructs and their derivatives are listed in Table 1. 121 Table 1. Description of cDNA constructs used in this study. 122 cDNA construct Transcript (Genbank) Source pcDNA3.1-CMV Empty vector V86020, Invitrogen (USA) pMB1-SV40 a gift from Jordan Green (Addgene plasmid # 127639; http://n2t.net/addgene:127639 ; RRID:Addgene_127639) (USA) pCAGEN-UbC a gift from Connie Cepko (Addgene plasmid # 11155; http://n2t.net/addgene:11155; RRID:Addgene_11155) (Matsuda and Cepko, 2004) pcDNA3.1-CMV-SCN5A-WT NM_000335.5 GenScript, NJ, USA pMB1-SV40-SCN5A-WT pCAGEN-UbC-SCN5A-WT pcDNA3.1-3xFLAG-SCN5A-WT pcDNA3.1-3xHA-SCN5A-WT pcDNA3.1-1xmEGFP-SCN5A-WT pFN217K-LgBiT(Nter)-SCN5A-WT Cloned from pcDNA3.1-CMV- SCN5A-WT into NanoBiT® CMV Flexi BiBit ready vectors (ID# CS1603B33) according to Flexi® vector systems technical manual #TM254 from Promega (Switzerland), as previously described (Iamshanova et al., 2024) pFC219K-SCN5A-WT-LgBiT(Cter) pFN218K-SmBiT(Nter)-SCN5A- WT pFC220K-SCN5A-WT- SmBiT(Cter) pF5K-SCN5A-WT CMV/SmBiT-CA/BlastR NM_002730.4 NanoBiT® CMV PPI control vectors (ID# CS1603B54) from Promega (Switzerland) CMV/LgBiT-R2A/HygR NM_004157.4 BiBiT-RI/SmBiT-CA/LgBiT- R2A/BlastR NM_002730.4 and NM_004157.4 CMV/HaloTag®-SmBiT HM157289.1 1xGFP-Tpr NM_003292.3 a gift from Larry Gerace (Addgene plasmid # 35024; .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 6 http://n2t.net/addgene:35024; RRID:Addgene_35024) (Frosst et al., 2002) pCMV6-SCN1BB NM_199037.5 #RC222161, Origene (USA) pCIH-SCN1B NM_001037.5 a gift from Simon Tate pCIH-SCN2B NM_004588.5 pFBM-SCN3B NM_018400.4 pCIH-SCN4B NM_174934.4 pcDNA3.1-3xFLAG-SCN5A-I450X truncated NM_000335.5, stop codon after Ile450 GenScript, NJ, USA pcDNA3.1-3xFLAG-SCN5A- I450X-allCys-mut same as previous with Cys139Ala, Csy145Ala, Cys182Ala, Cys260Ala, Cys280Ala, Cys326Ala, Cys335Ala, Cys341Ala, Cys373Ala pcDNA3.1-1xFLAG-SCN5A- R535X truncated NM_000335.5, stop codon after Arg535 pcDNA3.1-3xFLAG-SCN5A- R535X-extraCys-mut same as previous with Cys280Ala, Cys326Ala, Cys335Ala, Cys341Ala pcDNA3.1-3xFLAG-SCN5A- R535X-extraCys-mut same as previous with Cys139Ala, Cys145Ala, Cys182Ala, Cys260Ala, Cys280Ala, Cys326Ala, Cys335Ala, Cys341Ala, Cys373Ala pcDNA3.1-3xHA-SCN5A-NTD MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSANFLLPRGTSSFRRFTRESLAAIEKRMAEKQ ARGSTTLQESREGLPEEEAPRPQLDLQASKKLPDLYGNP PQELIGEPLEDLDPFYSTQKTFIVLNKGKTIFRFSATNALYV LSPFHPIRRAAVKILVHS* pcDNA3.1-3xHA-SCN5A-DI MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSLFNMLIMCTILTNCVFMAQHDPPPWTKYVEY TFTAIYTFESLVKILARGFCLHAFTFLRDPWNWLDFSVIIMA YTTEFVDLGNVSALRTFRVLRALKTISVISGLKTIVGALIQS VKKLADVMVLTVFCLSVFALIGLQLFMGNLRHKCVRNFTA LNGTNGSVEADGLVWESLDLYLSDPENYLLKNGTSDVLLC GNSSDAGTCPEGYRCLKAGENPDHGYTSFDSFAWAFLAL FRLMTQDCWERLYQQTLRSAGKIYMIFFMLVIFLGSFYLVN LILA* pcDNA3.1-3xHA-SCN5A-L1 MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSVVAMAYEEQNQATIAETEEKEKRFQEAMEM LKKEHEALTIRGVDTVSRSSLEMSPLAPVNSHERRSKRRK RMSSGTEECGEDRLPKSDSEDGPRAMNHLSLTRGLSRTS MKPRSSRGSIFTFRRRDLGSEADFADDENSTAGESESHH TSLLVPWPLRRTSAQGQPSPGTSAPGHALHGKKNSTVDC NGVVSLLGAGDPEATSPGSHLLRPVMLEHPPDTTTPSEE PGGPQMLTSQAPCVDGFEEPGARQRALSAVSVLTSALEE LEESRHKCPPCWNRLAQRYLIWECCPLWMSIKQGVKLVV MDP* pcDNA3.1-3xHA-SCN5A-DII MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSFTDLTITMCIVLNTLFMALEHYNMTSEFEEML QVGNLVFTGIFTAEMTFKIIALDPYYYFQQGWNIFDSIIVILS LMELGLSRMSNLSVLRSFRLLRVFKLAKSWPTLNTLIKIIGN SVGALGNLTLVLAIIVFIFAVVGMQLFGKNYSELRDSDSGL LPRWHMMDFFHAFLIIFRILCGEWIETMWDCMEVSGQSLC LLVFLLVMVIGNLVVLNLFLALL* .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 7 pcDNA3.1-3xHA-SCN5A-L2 MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSLSSFSADNLTAPDEDREMNNLQLALARIQRG LRFVKRTTWDFCCGLLRQRPQKPAALAAQGQLPSCIATP YSPPPPETEKVPPTRKETRFEEGEQPGQGTPGDPEPVCV PIAVAESDTDDQEEDEENSLGTEEESSKQESQPVSGGPE APPDSRTWSQVSATASSEAEASASQADWRQQWKAEPQ APGCGETPEDSCSEGSTADMTNTAELLEQIPDLGQDVKD PEDCFTEGCVRRCPCCAVDTTQAPGKVWWRLRKTCYHI VEHSW* pcDNA3.1-3xHA-SCN5A-DIII MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSFETFIIFMILLSSGALAFEDIYLEERKTIKVLLEY ADKMFTYVFVLEMLLKWVAYGFKKYFTNAWCWLDFLIVD VSLVSLVANTLGFAEMGPIKSLRTLRALRPLRALSRFEGM RVVVNALVGAIPSIMNVLLVCLIFWLIFSIMGVNLFAGKFGR CINQTEGDLPLNYTIVNNKSQCESLNLTGELYWTKVKVNF DNVGAGYLALLQVATFKGWMDIMYAAVDSRGYEEQPQW EYNLYMYIYFVIFIIFGSFFTLNLFI* pcDNA3.1-3xHA-SCN5A-L3 MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSGVIIDNFNQQKKKLGGQDIFMTEEQKKYYNA MKKLGSKKPQKPIPRPLNKYQGFIFDIVTKQA* pcDNA3.1-3xHA-SCN5A-DIV MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSFDVTIMFLICLNMVTMMVETDDQSPEKINILA KINLLFVAIFTGECIVKLAALRHYYFTNSWNIFDFVVVILSIV GTVLSDIIQKYFFSPTLFRVIRLARIGRILRLIRGAKGIRTLLF ALMMSLPALFNIGLLLFLVMFIYSIFGMANFAYVKWEAGID DMFNFQTFANSMLCLFQITTSAGWDGLLSPILNTGPPYCD PTLPNSNGSRGDCGSPAVGILFFTTYIIISFLIVVNMYIAII* pcDNA3.1-3xHA-SCN5A-NTD MYPYDVPDYAGYPYDVPDYAGYPYDVPDYAGGGGSGG GGSGGGGSLENFSVATEESTEPLSEDDFDMFYEIWEKFD PEATQFIEYSVLSDFADALSEPLRIAKPNQISLINMDLPMVS GDRIHCMDILFAFTKRVLGESGEMDALKIQMEEKFMAANP SKISYEPITTTLRRKHEEVSAMVIQRAFRRHLLQRSLKHAS FLFRQQAGSGLSEEDAPEREGLIAYVMSENFSRPLGPPSS SSISSTSFPPSYDSVTRATSDNLQVRGSDYSHSEDLADFP PSPDRDRESIV* pUC19-mScn5a-WT NM_001253860.1 a gift from Thomas Zimmer pcDNA3.1-mScn5a-WT Cell culture and transfection 123 The human embryonic kidney cell line tsA201 (ECACC Cat# 96121229, 124 RRID:CVCL_2737) and monkey kidney cell line COS-7 (ECACC Cat# 87021302, 125 RRID:CVCL_0224) were cultured up to 20 passages at 37°C with 5% CO 2 in 126 Dulbecco’s Modified Eagle’s Medium (41965, Gibco TM, Thermo Fisher Scientific, 127 USA), supplemented with 2 mM L-glutamine (G7513, Sigma, USA), 50 U/mL penicillin-128 streptomycin (15140122, Gibco) and 10% heat-inactivated fetal bovine serum (10270-129 106, Lot 2440045, Gibco). When needed, cells were split using phosphate- buffered 130 saline (PBS ; 10010-015, Gibco) and 0.05% Trypsin- EDTA (25300- 054, Gibco). 131 Mycoplasma contamination status was tested weekly with a PCR Mycoplasma Test 132 Kit I/C (PK -CA91-1096, Promokine, PromoCell GmbH, Germany). cDNA constructs 133 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 8 were introduced into cells by transfection with LipoD293 TM (SL100668, SignaGen ® 134 Laboratories, USA). Cells were harvested 48 hours after transfection unless stated 135 otherwise. 136 Protein extraction from cellular monolayers 137 Pre-washed (PBS) adherent monolayers of cells were scraped in 2 mL of cold PBS 138 (pH 7.4) and pelleted by centrifugation for 5 minutes at 200 × g at 4°C. Cell pellets 139 were lysed in lysis buffer (50 mM NaCl, 50 mM imidazole/HCl, 2 mM 6-aminohexanoic 140 acid, 1 mM EDTA, pH 7) with the addition of cOmplete tablets EDTA -free 141 (04693132001, Roche, Switzerland), 0.5 mM Na 3VO4, 0.5 mM NaFl, 10 μg/mL 142 aprotinin, 10 μg/mL leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 1% digitonin 143 (D141, Sigma) for 1 hour at 4°C. Lysates were centrifuged at 16,100 × g for 15 minutes 144 at 4°C. The obtained supernatant was taken for further analysis. Protein 145 concentrations were determined by Coo Assay Protein Dosage Reagent (UPF86420, 146 Uptima, Israel). 147 Protein extraction from mouse heart 148 According to the Swiss Federal Animal Protection Law, all animal experiments were 149 performed and approved by Bern’s Cantonal Veterinary Administration (license BE88 150 2022). This investigation conforms to the Guideline for the Care and Use of Laboratory 151 Animals, published by the US National Institutes of Health (NIH publication no. 85-23, 152 revised 1996). Mice were housed in a controlled, specific pathogen -free environment 153 (23 ± 1°C; humidity 60%; lights on 06:00 AM – 06:00 PM; food and water available ad 154 libitum; enriched environment) with a maximum of 6 mice per cage. 155 Female and male C57BL/6JRj mice between 20- 40 weeks old were deeply 156 anesthetized using ketamine/xylazine injection (intraperitoneal; one time; 200/20 157 mg/kg body weight) with heparin 25000IE/5 ml (4 ml/kg body weight) to avoid blood 158 coagulation. Mice were sacrificed by a thoracotomy (median sternotomy) to access 159 the heart as quickly as possible. Then the heart was rapidly excised and taken for 160 further experimental procedures. 161 For the isolation of cardiomyocytes, the excised heart was cannulated and mounted 162 in a Langendorff system for retrograde perfusion at 37°C. The heart was perfused free 163 of blood (few minutes) with nominally Ca 2+-free solution containing (mmol/L): 135 164 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 9 NaCl, 4 KCl, 1.2 MgCl2, 1.2 NaH2PO4, 10 HEPES, 11 glucose, pH 7.4 (NaOH). Next, 165 the heart was perfused with the same solutions with 50 µM Ca2+ and 275U collagenase 166 type II (275U/mg, CLS-2, Worthington, NJ, USA) till digestion. Following digestion, the 167 ventricle was transferred to the aforementioned buffer with 100 µM Ca 2+, minced into 168 small pieces and triturate to liberate single ventricular myocytes by gentle pipetting, 169 and filtered through a 100 µm nylon mesh. After supernatant was removed, 170 cardiomyocytes were lysed in 1x lysis buffer, 1% digitonin for 1 hour at 4°C. 171 For the homogenization procedures, the excised heart was washed three times in five 172 volumes of ice-cold PBS and put in one volume of 2x lysis buffer. For homogenization 173 with Benchtop Bioprep-24 (Hangzhou Allsheng Instruments, China) , 30 silica beads 174 of 1.4 mm diameter (531607, Milian) and 10 silica beads of 2.8 mm diameter (531608, 175 Milian) were added. The homogenization was performed in 6 cycles at 4260 rpm for 5 176 seconds (7m/s linear speed) with an interruption of 30 seconds at 4°C between the 177 cycles. For homogenization with VD12 Polytron (VWR), the heart was subjected to 2 178 cycles of 5 seconds of homogenization on ice. 179 After homogenization, digitonin was added to make a final concentration of 1% and 180 samples were left to lyse for 1 hour at 4°C. Lysates were centrifuged at 16,100 × g for 181 15 minutes at 4°C. The obtained supernatant was taken for further analysis. Protein 182 concentrations were determined by Coo Assay Protein Dosage Reagent (UPF86420, 183 Uptima). 184 Co-immunoprecipitation 185 Anti-FLAG® M2 Magnetic Beads (M8823, Sigma) and Pierce TM anti-HA Magnetic 186 Beads (88836, Thermo Scientific) were used at a proportion of 1 μL of beads per 40 187 μg of total protein lysate. Protein lysates were diluted in Tris -buffered saline [TBS, 20 188 M Tris, 0.1 mM NaCl, pH 7.6 (HCl-adjusted )] and mixed with 0.05% TBS -Tween-20-189 prewashed antibody -coupled beads at 4°C overnight. After thoroughly washing the 190 beads with TBS, the co-immunoprecipitated protein complex was eluted with 4× LDS 191 sample buffer (NP0007, Invitrogen, USA ). Samples were then analyzed with an 192 immunoblotting technique (see below). 193 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 10 Cell surface biotinylation 194 Adherent cell monolayers were gently washed three times with cold PBS (pH 8) and 195 incubated, while slowly rocking, with 1 mg/mL EZlink ™ Sulfo-NHS-SS-Biotin (21331, 196 Thermo Fisher Scientific) in PBS (pH 8) for 45 minutes at 4°C. Afterward, excess biotin 197 was quenched by washing with 50 mM Tris -HCl (pH 8), followed by several washes 198 with PBS (pH 7.4). Extracted protein lysates were mixed with Streptavidin Sepharose 199 High-Performance Beads (GE Healthcare, USA) at 4°C overnight. After thoroughly 200 washing the beads with washing buffer [ 0.1% bovine serum albumin ( BSA), 0.001% 201 Tween-20, PBS, pH 7.4], the biotinylated fraction was eluted with 4× LDS sample 202 buffer and analyzed with an immunoblotting technique (see below). 203 Sodium dodecyl sulfate –polyacrylamide gel 204 electrophoresis (SDS-PAGE) and immunoblotting analysis 205 Protein lysates containing LDS sample buffer with or without 100 mM dithiothreitol 206 (DTT) were loaded onto 4%–8% or 4%–12% Tris-acetate acrylamide gels and run at 207 60V in Tris -acetate running buffer (50 mM Tris, 50 mM Tricine, 0.1% SDS, pH 8.3). 208 Afterward, proteins were transferred to a nitrocellulose membrane using a Trans-Blot 209 Turbo system (1704158, Bio-Rad, USA). After validation of successful protein transfer 210 with Ponceau S solution (0.1% Ponceau S, 12.5% acetic acid), membranes were 211 blocked with 5% BSA in TBS-T (TBS, 0.1% Tween-20) for 1 hour at room temperature 212 and then incubated with primary antibody dilutions at 4°C overnight (Table 1). After 213 incubating membranes with secondary antibodies for 1 hour at room temperature, 214 infrared fluorescent signals were revealed with a LI-COR Odyssey Infrared Imaging 215 System (LI -COR Biosciences, USA ) or FUSION FX7 Imaging System (Witec , 216 Germany) and quantified with ImageJ software (Rasband, W.S., US National Institutes 217 of Health, Bethesda, MD, USA). 218 Primary and secondary antibodies used in this study are listed in Table 2. 219 Table 2. Description of primary and secondary antibodies used in this study. 220 Antigen Dilution Host Epitope RRID Source FLAG 1:1000 mouse DYKDDDDK AB_262044 F1804, Sigma-Aldrich (USA) GFP 1:1000 mouse n/a AB_390913 11814460001, Roche (Switzerland) GFP 1:1000 rabbit Full-length denatured and non-denatured TurboGFP and CopGFP n/a AB514, Lot#51402250467, Evrogen (Russia) .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 11 HA 1:500 mouse YPYDVPDYA n/a ENZ-ABS118-0500, Enzo Life Sciences (USA) HA 1:1000 rabbit YPYDVPDYA AB_2810986 ab137838, Abcam (UK) Nav1.5 1:1000 rabbit DTVSRSSLEMSPLAPV n/a generated by Pineda (Germany) Navβ1 1:200 rabbit CTGVQVAE n/a a gift from Isabelle Decosterd (Lausanne, Switzerland) Navβ2 1:1000 rabbit n/a n/a Navβ3 1:200 rabbit CSAVPVEE n/a Navβ4 1:1000 rabbit n/a n/a E90110A, Enogene, a gift from Isabelle Decosterd (Lausanne, Switzerland) α1-Na+/K+ ATPase 1:500 mouse Full length native α1- Na+/K+ ATPase AB_306023 ab7671, Abcam (UK) α1- syntrophin 1:1000 rabbit SGRRAPRTGLLELRAC n/a generated by Pineda (Germany) α-actin 1:1000 rabbit SGPSIVHRKCF AB_476693 A2066, Sigma-Aldrich calnexin 1:1000 rabbit C-terminus of human calnexin (amino acids 573-592) AB_476845 C4731, Sigma-Aldrich IgG mouse 1:20,000 goat Reacts with the heavy and light chains of mouse IgG1, IgG2a, IgG2b, and IgG3, and with the light chains of mouse IgM and IgA. AB_10956588 926-68070, LI-COR Biosciences (Germany) IgG rabbit 1:20,000 goat Reacts with the heavy and light chains of rabbit IgG, and with the light chains of rabbit IgM and IgA. AB_621843 926-32211, LI-COR Biosciences Protein-protein interaction assay in living cells 221 The NanoLuc® Binary Technology NanoBiT® (Promega, USA) allowed us to monitor 222 protein-protein interactions between two Na v1.5-WT α -subunits in living cells as 223 described previously (Iamshanova et al., 2024). The luminescent signal was obtained 224 with a Nano-Glo® Live Cell Assay (N2012, Promega) and normalized to the cell 225 quantity represented by the fluorescent signal obtained with a CellTiter-Fluor® Cell 226 Viability Assay (G6082, Promega). S ignals were detected with a GloMax® Explorer 227 Multimode Microplate Reader (GM3500, Promega). 228 Data and statistical analysis 229 Data are represented as the mean ± SEM or SD. Data normality was tested using the 230 Shapiro-Wilk test. Statistical significances of normally distributed data were calculated 231 with ordinary one- way ANOVA and Tukey’s multiple comparisons test s. In case s 232 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 12 where data were normalized to a control condition, the statistical significance was 233 calculated with a one-sample two-tailed t-test with hypothetical mean value = 1 (Prism 234 version 8.4.3; GraphPad, CA, USA). Exact p -values are indicated in the figures. 235 Neither randomization nor blinding were performed in this study. 236

237 Unlike in native tissue extracts, heterologous expression of 238 NaV1.5 introduces α-α-subunit interactions 239 Previous studies reported the α-α-subunit interactions of Nav1.5 (Clatot et al., 2017, 240 2012; Iamshanova et al., 2024) . It is important to note that the “true” Nav1.5 241 dimers/oligomers and the aggregates of Nav1.5 would both possess α-α-subunit 242 interactions. Therefore, hereafter, we use the definition of oligomer alongside protein 243 aggregate, as those two are indistinguishable by most biochemical methods. 244 Here, we confirmed that differently tagged Nav1.5 proteins co-immunoprecipitated in 245 a het erologous expression system represented by tsA201 cells (Fig. 1A). In living cells, 246 the dimerization of Na v1.5 was preserved (Iamshanova et al., 2024) . Because 247 interactions through the N-N termini of Na v1.5 consistently produced the brightest 248 signal (Iamshanova et al., 2024) , they were used as a default configuration for the 249 NanoBiT assay in this study. Based on results showing that untagged Na v1.5 250 competed for its interaction with the luminescence-producing pair of Nav1.5-LgBiT and 251 Nav1.5-SmBiT (Fig. 1B , C) but not the control luminescence- producing pair of well -252 known interacting proteins PRKAR2A -LgBiT and PRKACA- SmBiT (Fig. S1), we 253 concluded that the observed dimerization was specific to Nav1.5. 254 To resolve Nav1.5 dimers at the expected molecular weight of ~460 kDa, we performed 255 non-reducing 4%– 8% gradient SDS -PAGE. To deduce whether the observed high 256 molecular weight band represented the Na v1.5-Nav1.5 complex rather than an 257 aspecific band recognized by the antibody used during immunoblotting, we co-258 expressed non-tagged Nav1.5 with green fluorescent protein (GFP) -tagged Nav1.5. 259 Accordingly, a ~460-kDa band was revealed with both antibodies, against Nav1.5 and 260 against GFP, while differences in the sizes of corresponding monomeric Na v1.5 261 proteins (~30 kDa for GFP tag) confirmed the validity of the constructs used (Fig. 1D). 262 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 13 Therefore, we concluded that the ~460-kDa band represented the dimer of Na v1.5 263 proteins, which could also be represented by the SDS-resistant aggregate. 264 Because our non-reducing SDS-PAGE yielded robust results for detecting the putative 265 Nav1.5 dimer/aggregated, we decided to apply this method to native tissues. Using a 266 C57BL/6 mouse model with two untagged wild-type alleles for SCN5A, we performed 267 three different types of protein extraction on excised hearts: tissue homogenizer, 268 benchtop homogenizer, and direct lysis of isolated cardiomyocytes. Importantly, 269 regardless of the protein extraction method used, only monomers of Na v1.5 were 270 detected (Fig. 1E). To verify that our result was not caused by species specificity, we 271 carried out non-reducing SDS-PAGE on protein lysates from tsA201 cells in parallel 272 expressing mouse and human Na v1.5 proteins (Fig. 1F). Immunoblotting analysis 273 revealed the Na v1.5-specific bands corresponding to the expected size of its 274 monomeric state (between 171- kDa and 238-kDa of protein standard bands) as well 275 as its putative dimeric/aggregate state (~460-kDa of protein standard band) observed 276 in Fig. 1D (Fig. 1F). Interestingly, application of a reducing agent, 100 mM DTT, 277 decreased the intensity of putative Nav1.5 dimer/aggregate band and increased the 278 amount of Nav1.5 monomers (Fig. 1F). This could be due to the disruption of Na v1.5 279 dimers into monomers, suggesting the involvement of covalent disulfide bridges. 280 Therefore, we concluded that these putative Nav1.5 dimers/aggregates were at least 281 partially sensitive to the reducing agent and were absent in non -diseased mouse 282 cardiomyocytes. 283 Nav1.5 dimers are present at the plasma membrane, but are 284 not primarily formed by disulfide bonds 285 Functional Nav1.5 proteins are typically targeted to the plasma membrane. To assess 286 whether the interacting α -subunits were trapped intracellularly , we performed 287 a biotinylation assay. Our results suggest that both monomers and putative 288 dimers/aggregates of Nav1.5 were present at the cell surface when heterologously 289 expressed in tsA201 (Fig. 2A). 290 Since Cys bridges are typically extracellular, this finding was in line with our previous 291 suggestion that disulfide bridges may play a role in linking two Nav1.5 proteins 292 together. However, c onsidering that α -α-subunit interactions could also be 293 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 14 represented by the aggregated misfolded proteins, we also assessed the possibility of 294 non-native disulfide bridges formed via intracellular Cys. 295 The full-length Nav1.5 (UniProtKB: Q14524-1) has 42 Cys residues in total (Fig. 2B). 296 Importantly, a previous study reported on the existence of a direct α-α-subunit 297 interaction site between Arg493-Arg517 (Fig. 2B) (Clatot et al., 2017). Like that study, 298 we used the truncated Na v1.5 proteins , Nav1.5-Arg535X and Nav1.5-Ile450X, to 299 additionally investigate the role of Arg493- Arg517 for the direct interaction (Fig. 2B). 300 Moreover, to identify which Cys could be responsible for the intermolecular disulfide 301 bonds between Nav1.5 proteins, we mutated Cys into Ala. For this, we created Nav1.5-302 Arg535X constructs with only 4 extracellular Cys (Nav1.5-Arg535X-extraCys-mut) and 303 all 10 Cys being mutated (Nav1.5-Arg535X-allCys-mut). Similarly, we mutated all 9 304 Cys of Nav1.5-Ile450X (Nav1.5-Ile450X-allCys-mut). Co-immunoprecipitation analysis 305 between full-length Nav1.5 and the aforementioned truncated Nav1.5 proteins revealed 306 that mutating only 4 extracellular Cys or all 10 Cys was insufficient to abolish the ability 307 of Na v1.5-Arg535X protein to interact with the full-length Na v1.5 (Fig. 2 C). 308 Furthermore, Na v1.5-Ile450X, lacking the presumed α-α-subunit interaction site 309 between Arg493-Arg517, also preserved its binding with the full-length Nav1.5, and 310 this binding was not affected by the absence of Cys (Fig. 2C). Therefore, we concluded 311 that Arg493-Arg517 was not responsible for linking two Na v1.5 proteins. Moreover, 312 although Nav1.5 multimers/aggregates could have been at least partially supported via 313 di sulfide bridges , as observed in Fig. 1F , they were not the main cause of its 314 oligomerization/aggregation. 315 Nav1.5 dimers are unaffected by co -expression with Na vβ-316 subunits 317 Navβ-subunits, particular ly Navβ2 and Na vβ4, are known to form intermolecular 318 disulfide bridges within themselves as well as with α -subunits of VGSCs [reviewed in 319 (Iamshanova et al., 2023) ]. Notably, a previous study proposed that α -α-subunit 320 interactions of Nav1.5 were regulated by the presence of Navβ1 (Mercier et al., 2012). 321 Although we and other groups reported the self -interactions of Nav1.5 proteins in the 322 absence of Navβ-subunits (Iamshanova et al., 2024; Clatot et al., 2017) , we decided 323 to revise this question in this study . First, we compared the ratios of Na v1.5 324 dimers/monomers with and without individual co-expression of Navβ1B, Navβ1, Navβ2, 325 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 15 Navβ3, and Na vβ4 subunits (Fig. 3A, C). Second, we analyzed the amount of 326 haemagglutinin (HA)-tagged Nav1.5 that co-immunoprecipitated with FLAG-Nav1.5 in 327 the absence and presence of Na vβ-subunits co- expression (Fig. 3B, D). Third, we 328 estimated levels of the putative Nav1.5 dimers between the aforementioned conditions 329 in living cells (Fig. 3 E). Because we did not observe an effect in any experimental 330 procedure, we concluded that Na vβ-subunits are indeed dispensable for the homo-331 oligomerization/aggregation of Nav1.5 in our heterologous expression system. 332 Transmembrane domains strongly interact with full -length 333 Nav1.5 and reduce its protein level 334 Because our data did not confirm the exclusivity of the previously described Arg493-335 Arg517 site for α -α-subunit interactions of Nav1.5 (Fig. 2 C) (Clatot et al., 2017) , we 336 decided to revise the interaction sites by co -immunoprecipitation analysis between 337 different parts of Nav1.5 and the full-length protein (Fig. 4A). Strikingly, the presence 338 of transmembrane domains I –IV decreased total levels (both putative 339 dimers/aggregates and monomers) of full -length Nav1.5 (Fig. 4A , B). This could be 340 explained by increased aspecific hydrophobic interactions between the large 341 transmembrane domains of Na v1.5 that would lead to protein aggregation and 342 activation of UPR with subsequent proteasomal degradation (Fig. 4 C). Indeed, 343 although the total protein quantity was diminished, the interaction between 344 transmembrane domains I–IV and full-length Nav1.5 remained very strong (Fig. 4B). 345 Furthermore, we detected potent interactions between full -length Na v1.5, its 346 intracellular N-terminal domain (NTD), and the linker between domains I and II (named 347 L1) (Fig. 4B). Overall, these results indicate that the heterologusly expressed Na v1.5 348 proteins interact with each other not due to a specific amino acid sequence but rather 349 within multiple regions. 350 Nav1.5 dimerization levels are regulated by its protein 351 expression level and core N-linked glycosylation in the ER 352 All α -α-subunit interactions of VGSC s were previously shown in a heterologous 353 ov erexpression system . In this case, there is a possibility that the overload and/or 354 unsuitability of protein folding machinery could lead to the aggregation of ectopic 355 proteins (Fig. 5A). Thus, we decided to assess the oligomerization/aggregation of 356 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 16 Nav1.5 while controlling the level of its overexpression. To achieve this, we compared 357 the ratios of the putative reducing agent -sensitive Nav1.5 dimers /aggregates to its 358 monomers produced under promoters with different expression strengths (Fig. 5B). 359 According to our results, the Nav1.5 dimer/monomer ratio was directly proportional to 360 the level of heterologous protein expression (Fig. 5 C). Interestingly, the strongest 361 promoter, cytomegalovirus (CMV), also led to the production of monomeric Na v1.5 of 362 a smaller size compared with monomeric Nav1.5 proteins produced with SV40 and 363 Ubiquitin C (UbC) promoters (Fig. 5B). This encouraged us to further investigate 364 whether a slight change in the size of monomeric Na v1.5 under strong level s of 365 expression was associated with the insufficiency of post -translational modifications 366 resulting from saturated protein folding machinery (Fig. 5 A). One of the first post -367 translational modifications is N-glycosylation, which occurs while the nascent peptide 368 is being synthesized to assist with correct protein folding in the ER (Fig. 5D). Both 369 native and heterologous Na v1.5 were shown to be N-glycosylated (Mercier et al., 370 2015). Therefore, we used an inhibitor of N-core glycosylation, tunicamycin (TUN), to 371 test its importance for the heterologous oligomerization/aggregation of Nav1.5 (Fig. 372 5D). TUN prevents the addition of oligosaccharides to nascent polypeptides, leading 373 to their misfolding and accumulation in the ER (Fig. 5D). This, in turn, induces ER 374 stress followed by activation of the UPR and proteasomal degradation (Fig. 5D). To 375 constrain the latter and preserve Na v1.5 dimers/aggregates possibly evoked due to 376 the affected N-glycosylation, we used the cell-permeable proteasomal inhibitor MG132 377 (Fig. 5D). Because the presence of TUN and MG132 might significantly alter the total 378 amount of Na v1.5 protein, we compared the differences between Nav1.5 379 dimer/monomer ratios under various conditions by normalizing them to treatment with 380 the highest total protein amount, i.e. MG132 alone. Accordingly, our results confirmed 381 t hat TUN alone induced the degradation of heterologous protein and reduced the size 382 of the band corresponding to monomeric Na v1.5, whereas the addition of MG132 383 rescued the amount of heterologous Nav1.5 (Fig. 5E, F). Furthermore, the amount of 384 non-reduced Nav1.5 monomers decreased under the condition of combined TUN and 385 MG132 treatment, while the reduced Na v1.5 monomers tended to increase (Fig. 5E, 386 F). This result could signify that the combination of TUN with MG132 facilitated the 387 accumulation of Nav1.5 oligomers compared with MG132 treatment alone. Altogether, 388 our results suggest that a high production rate of heterologous Nav1.5 might saturate 389 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 17 the ER folding machinery, and this, in turn, as observed with impaired N-glycosylation, 390 could promote its dimerization/aggregation. 391

392 By using non-reducing SDS-PAGE, we were able to resolve putative Nav1.5 dimers in 393 the heterologous expression system represented by tsA201 cells, but not in native 394 mouse cardiomyocytes. These putative Nav1.5 dimers were shown to be partially 395 sensitive to the reducing agent, depending on the ER stress and the presence of 396 transmembrane domains. T his could signify that the apparent dimers of Nav1.5 are 397 likely being represented by the SDS-resistant aggregates, supported by the non-native 398 disulfide bonds. 399 W e show that these dimers /aggregates were unaffected by Navβ-subunits but rather 400 relied on the level of SCN5A overexpression. We identified that strong interactions 401 occurred between the full-length Nav1.5 and its NTD, DI, L1, DII, DIII, and DIV in our 402 cell model. 403 Our study validates the occurrence of di merized human VGSC α-subunits, formerly 404 reported by us and other groups (Mercier et al., 2012; Clatot et al., 2012, 2017; 405 Salvage et al., 2022; Rühlmann et al., 2020; Iamshanova et al., 2024; Salvage et al., 406 2020). In addition to the previously described homodimerization site of Nav1.5 at L1 407 (Clatot et al., 2017) , our group previously observed close proximity between the full-408 length protein and its truncated NTD (Wang et al., 2020) . In the present study, we 409 confirm direct interactions between full -length Na v1.5 and its NTD , as well as L1. 410 Recently, another inter-channel dimerization of a VGSC α-subunit was reported and 411 suggested to take place through its voltage-sensing domain (Sumino et al., 2023). In 412 support of these observations, our results show that all four transmembrane domains 413 (i.e., DI, DII, DIII, and DIV) were indeed able to strongly interact with full-length Nav1.5. 414 Interestingly, two-distinct pools of N-glycosylated Nav1.5 channels were reported for 415 the human embryonic kidney cell line HEK293T (RRID:CVCL_0063), a heterologous 416 expression system highly similar to ours (Mercier et al., 2015) . A previous study 417 proposed the co-existence of fully mature Na v1.5 channels that underwent post-cis-418 Golgi glycosylation together with partially mature N-glycan-restricted Nav1.5 proteins 419 (Mercier et al., 2015) . Even though the latter did not undergo terminal glycosylation, 420 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 18 both Nav1.5 pools were shown to traffic to the plasma membrane (Mercier et al., 2015). 421 The authors suggested that the unconventional Golgi -independent pathway of 422 membrane transport c ould be potentially used for the clearance of Na v1.5 proteins 423 accumulating in the ER to prevent ER stress (Mercier et al., 2015) . Similarly, we 424 suggest a possible link between Na v1.5 dimerization/aggregation and its post-425 translational modifications, such as N-glycosylation. According to our results , 426 preventing early N -glycosylation of nascent polypeptides with TUN while blocking 427 proteasomal degradation with MG132 tended to promote Na v1.5 428 dimerization/aggregation. We also show that Nav1.5 dimers/aggregates reached the 429 plasma membrane. In line with these findings, Förster resonance energy transfer 430 studies demonstrated that Na v1.5 α-subunits interacted at the cell surface and that 431 these interactions occurred before protein trafficking to the plasma membrane (Clatot 432 et al., 2017). 433 Previously, Navβ1 was reported to mediate α -α-subunit interactions between Na v1.5 434 proteins (Mercier et al., 2012) . However, our current data demonstrate that 435 dimerization/aggregation of Na v1.5 was unaffected by the presence of Navβ1, its 436 alternative splicing form Navβ1B, or subunits Navβ2, Navβ3, and Navβ4. This was not 437 surprising to us, as other studies showing oligomerization of human VGSC α-subunits 438 did not use co- expression of Na vβ-subunits ( Clatot 2012, Clatot 2017, Rhulmann 439 2020). In addition, another study concluded that clusterization of Na v1.5 is possible 440 with and without Na vβ3 (Salvage et al., 2020) . Unexpectedly, the α -α-subunit 441 interaction site previously described for Na v1.5 as Arg493-Arg517 region at L1 was 442 not proven to be exclusive in our hands (Clatot et al., 2017) . Previously, the authors 443 demonstrated that truncated GFP -tagged constructs Na v1.5-Ile450X and Na v1.5-444 Asn470X were un able to co- immunoprecipitate with full -length HA -tagged Na v1.5 445 (Clatot et al., 2017) . However, according to our results, Nav1.5-Ile450X strongly 446 i nteracted with full -length Nav1.5. Indeed, such an interaction could be mediated via 447 NTD and DI, parts of Nav1.5-Ile450X that by themselves were shown to strongly bind 448 to full-length Nav1.5 in our current work. 449 Another discrepancy with our study comes from the previously shown migration of 450 human Na v1.5 and Na v1.7 as monomers in both reduced and non- reduced states 451 (Rühlmann et al., 2020). Based on these observations, the authors argued against the 452 existence of disulfide bonds between α-subunits but not against non- covalent bonds 453 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 19 (Rühlmann et al., 2020). However, our results indicate that Nav1.5 dimers/aggregates 454 were reducing agent-dependent, even though disulfide bonds did not appear to be the 455 primary cause of α -α-subunit interactions . One possible explanation could be 456 technical: we us ed Tris-acetate SDS-PAGE in contrast to SDS-urea-PAGE used in 457 previous work. As a strong protein denaturant , urea might assist in protein unfolding 458 and/or stabilization of the unfolded state. If this is the case and non- native disulfide 459 bridges appear during partial protein refolding as a result of our experimental 460 procedures, then it would not explain the absence of Na v1.5 dimers /aggregates in 461 native cardiac tissues/cells when using the same non- reducing Tris -acetate SDS -462 PAGE protocol. Interestingly, for the 30 proteins analyzed, no correlation was found 463 between the extent of urea- induced protein unfolding and the presence/absence of 464 disulfide bridges in the native form (Candotti et al., 2013) . Additionally, an other 465 explanation could be experimental: cellular machinery providing for post -translational 466 modifications, including glycosylation, is significantly different between 467 the heterologous expression system used in this study, human tsA201 cells, and 468 Xenopus laevis oocytes used in previous work (Rühlmann et al., 2020). 469 Strengths/limitations 470 Importantly, most of our data were obtained in a heterologous expression system 471 using a protein overexpression approach. Thus, it is possible that the observed 472 interactions and effects on Na v1.5 dimerization/aggregation were largely affected by 473 non-native ectopic gene expression. 474 Another limitation of our study is that by using non-reducing Tris-acetate SDS-PAGE, 475 we were only able to resolve putative Nav1.5 dimers that primarily depended on the 476 disulfide bonds. Therefore, whenever possible, we performed co-immunoprecipitation 477 analysis as an additional confirmation of the α-α-subunit interactions. 478 Implications 479 Biochemical interactions between α -α-subunits of Na v1.5 were largely used to 480 describe/explain the dominant-negative effect (DNE) of SCN5A (reviewed in 10). DNE 481 was shown for both native and heterologous cell models (Doisne et al., 2021; O’Neill 482 et al., 2022; Keller et al., 2005) . Accordingly, it was hypothesized that DNE occurred 483 due to direct interactions between wild-type and mutant Nav1.5 because the trafficking, 484 t urnover, and function of the interacting pair would be largely affected (Sottas and 485 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 20 Abriel, 2016). However, a direct link between DNE and Na v1.5 has yet to be shown, 486 as disruption of the protein-protein interaction between mutant and wild-type channels 487 should lead to the rescue of INa. 488 Our data suggest a link between Na v1.5 dimerization/aggregation and ER stress. 489 Although we did not detect Na v1.5 dimers in non- diseased mouse cardiomyocytes, 490 such α-α-subunits interactions may occur after ER stress induction. Indeed, fever, 491 pharmacological agents, and age might promote ER stress and are known risk factors 492 for the manifestation of Brugada syndrome, a cardiac disease primarily attributed to 493 SCN5A loss-of-function variants (Keller et al., 2005). 494

495 Taken together, our data raise caution regarding challenges related to investigating 496 the oligomerization of Nav1.5 in a heterologous expression system. Therefore, further 497 work needs to be carried out, specifically, using an in vivo model where both alleles of 498 SCN5A are genetically modified to produce native Nav1.5 proteins with different tags. 499 Supporting information 500 Figure S1 with its legend shows analyzed data to complement Figure 1. 501 Data availability statement 502 Data are available in the article itself and its supplementary materials. 503 Conflict of interest 504 The authors declare no potential conflict of interest. 505 Acknowledgments 506 Illustrations were created with BioRender.com. 507 Author contributions 508 O.I, J.S.R., and H.A. conceptualization. O.I., A.F.H., S.S., A.S., and S.G. investigation. 509 O.I., A.F.H., and S.S. formal analysis. O.I., J.S.R., and H.A. writing – original draft. 510 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 21 Funding and additional information 511 This work was funded by the Swiss National Science Foundation [SNF 512 310030_184783] to H.A. 513 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 22

514 Bouza, A.A., and L.L. Isom. 2017. Voltage- Gated Sodium Channel β Subunits and 515 Their Related Diseases. 423–450. 516 Candotti, M., A. Pérez, C. Ferrer -Costa, M. Rueda, T. Meyer, J.L. Gelpí, and M. 517 Orozco. 2013. Exploring Early Stages of the Chemical Unfolding of Proteins at 518 the Proteome Scale. PLoS Comput Biol. 9. doi:10.1371/journal.pcbi.1003393. 519 Clatot, J., M. Hoshi, X. Wan, H. Liu, A. Jain, K. Shinlapawittayatorn, C. Marionneau, 520 E. Ficker, T. Ha, and I. Deschênes. 2017. Voltage- gated sodium channels 521 assemble and gate as dimers. Nat Commun . 8:1–14. doi:10.1038/s41467-017-522 02262-0. 523 Clatot, J., A. Ziyadeh-Isleem, S. Maugenre, I. Denjoy, H. Liu, G. Dilanian, S.N. Hatem, 524 I. Deschênes, A. Coulombe, P. Guicheney, and N. Neyroud. 2012. Dominant -525 negative effect of SCN5A N-terminal mutations through the interaction of Nav1.5 526 α-subunits. Cardiovasc Res. 96:53–63. doi:10.1093/cvr/cvs211. 527 Dixon, R.E., M.F. Navedo, M.D. Binder, and L. Fernando Santana. 2022. Mechanisms 528 and Physiological Implications of Cooperative Gating of Clustered Ion Channels. 529 Physiol Rev. 102:1159–1210. doi:10.1152/physrev.00022.2021. 530 Doisne, N., M. Grauso, N. Mougenot, M. Clergue, C. Souil, A. Coulombe, P. 531 Guicheney, and N. Neyroud. 2021. In vivo Dominant-Negative Effect of an SCN5A 532 Brugada Syndrome Variant. Front Physiol . 12:1– 13. 533 doi:10.3389/fphys.2021.661413. 534 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 23 Dong, C., Y. Wang, A. Ma, and T. Wang. 2020. Life Cycle of the Cardiac Voltage-535 Gated Sodium Channel NaV1.5. Front Physiol . 11:1 –11. 536 doi:10.3389/fphys.2020.609733. 537 Eshed-Eisenbach, Y., and E. Peles. 2021. The clustering of voltage- gated sodium 538 channels in various excitable membranes. Dev Neurobiol . 81:427– 437. 539 doi:10.1002/dneu.22728. 540 Frosst, P., T. Guan, C. Subauste, K. Hahn, and L. Gerace. 2002. Tpr is localized within 541 the nuclear basket of the pore complex and has a role in nuclear protein export. 542 Journal of Cell Biology. 156:617–630. doi:10.1083/jcb.200106046. 543 Iamshanova, O., A.F. Hämmerli, E. Ramaye, A. Seljmani, D. Ross -Kaschitza, N. 544 Schärz, M. Essers, S. Guichard, J.S. Rougier, and H. Abriel. 2024. The 545 dispensability of 14-3- 3 proteins for the regulation of human cardiac sodium 546 channel Nav1.5. PLoS One. 19. doi:10.1371/journal.pone.0298820. 547 Iamshanova, O., J.-S. Rougier, and H. Abriel. 2023. Exploring the Oligomerization of 548 Nav1.5 and Its Implication for the Dominant -Negative Effect. Bioelectricity. 5:1–549 11. doi:10.1089/bioe.2023.0014. 550 Keller, D.I., J.S. Rougier, J.P. Kucera, N. Benammar, V. Fressart, P. Guicheney, A. 551 Madle, M. Fromer, J. Schläpfer, and H. Abriel. 2005. Brugada syndrome and 552 fever: Genetic and molecular characterization of patients carrying SCN5A 553 mutations. Cardiovasc Res. 67:510–519. doi:10.1016/j.cardiores.2005.03.024. 554 Li, Z., X. Jin, T. Wu, X. Zhao, W. Wang, J. Lei, X. Pan, and N. Yan. 2021. Structure of 555 human Nav1.5 reveals the fast inactivation -related segments as a mutational 556 hotspot for the long QT syndrome. Proc Natl Acad Sci U S A . 118:1 –7. 557 doi:10.1073/PNAS.2100069118. 558 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 24 Liu, M., and S.C. Dudley. 2018. The role of the unfolded protein response in 559 arrhythmias. Channels. 12:335–345. doi:10.1080/19336950.2018.1516985. 560 Marchal, G.A., and C.A. Remme. 2022. Subcellular diversity of Nav1.5 in 561 cardiomyocytes: distinct functions, mechanisms and targets. Journal of 562 Physiology. 5:941–960. doi:10.1113/JP283086. 563 Marionneau, C., and H. Abriel. 2015. Regulation of the cardiac Na+ channel NaV1.5 564 by post -translational modifications. J Mol Cell Cardiol. 82:36 –47. 565 doi:10.1016/j.yjmcc.2015.02.013. 566 Matsuda, T., and C.L. Cepko. 2004. Electroporation and RNA interference in the 567 rodent retina in vivo and in vitro. Proc Natl Acad Sci U S A . 101:16 –22. 568 doi:10.1073/pnas.2235688100. 569 Mercier, A., R. Clément, T. Harnois, N. Bourmeyster, P. Bois, and A. Chatelier. 2015. 570 Nav1.5 channels can reach the plasma membrane through distinct N -571 glycosylation states. Biochim Biophys Acta Gen Subj . 1850:1215 –1223. 572 doi:10.1016/j.bbagen.2015.02.009. 573 Mercier, A., R. Clément, T. Harnois, N. Bourmeyster, J.F. Faivre, I. Findlay, M. 574 Chahine, P. Bois, and A. Chatelier. 2012. The β1-Subunit of Nav1.5 Cardiac 575 Sodium Channel Is Required for a Dominant Negative Effect through α-α 576 Interaction. PLoS One. 7:5–12. doi:10.1371/journal.pone.0048690. 577 O’Neill, M.J., A. Muhammad, B. Li, Y. Wada, L. Hall, J.F. Solus, L. Short, D.M. Roden, 578 and A.M. Glazer. 2022. Dominant negative effects of SCN5A missense variants. 579 Genetics in Medicine. 24:1238–1248. doi:10.1016/j.gim.2022.02.010. 580 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 25 Payandeh, J., and D.L. Minor. 2015. Bacterial voltage- gated sodium channels 581 (BacNaVs) from the soil, sea, and salt lakes enlighten molecular mechanisms of 582 electrical signaling and pharmacology in the brain and heart. J Mol Biol. 427:3 –583 30. doi:10.1016/j.jmb.2014.08.010. 584 Payandeh, J., T. Scheuer, N. Zheng, and W.A. Catterall. 2011. The crystal structure 585 of a voltage- gated sodium channel. Nature . 475:353– 359. 586 doi:10.1038/nature10238. 587 Rühlmann, A.H., J. Körner, R. Hausmann, N. Bebrivenski, C. Neuhof, S. Detro‐588 Dassen, P. Hautvast, C.A. Benasolo, J. Meents, J. Machtens, G. Schmalzing, and 589 A. Lampert. 2020. Uncoupling sodium channel dimers rescues the phenotype of 590 a pain‐linked Nav1.7 mutation. Br J Pharmacol. 0–1. doi:10.1111/bph.15196. 591 Salvage, S.C., K. Jeevaratnam, C.L.H. Huang, and A.P. Jackson. 2022. Cardiac 592 sodium channel complexes and arrhythmia: structural and functional roles of the 593 β1 and β3 subunits. Journal of Physiology. 0:1–18. doi:10.1113/JP283085. 594 Salvage, S.C., J.S. Rees, A. McStea, M. Hirsch, L. Wang, C.J. Tynan, M.W. Reed, 595 J.R. Irons, R. Butler, A.J. Thompson, M.L. Martin-Fernandez, C.L.H. Huang, and 596 A.P. Jackson. 2020. Supramolecular clustering of the cardiac sodium channel 597 Nav1.5 in HEK293F cells, with and without the auxiliary β3-subunit. FASEB 598 Journal. 34:3537–3553. doi:10.1096/fj.201701473RR. 599 Shen, H., D. Liu, K. Wu, J. Lei, and N. Yan. 2019. Structures of human Na v 1.7 600 channel in complex with auxiliary subunits and animal toxins. Science (1979) . 601 363:1303–1308. doi:10.1126/science.aaw2493. 602 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 26 Sottas, V., and H. Abriel. 2016. Negative- dominance phenomenon with genetic 603 variants of the cardiac sodium channel Nav1.5. Biochim Biophys Acta Mol Cell 604 Res. 1863:1791–1798. doi:10.1016/j.bbamcr.2016.02.013. 605 Sumino, A., T. Sumikama, M. Shibata, and K. Irie. 2023. Voltage sensors of a Na+ 606 channel dissociate from the pore domain and form inter -channel dimers in the 607 resting state. Nat Commun. 14. doi:10.1038/s41467-023-43347-3. 608 Wang, Z., S.H. Vermij, V. Sottas, A. Shestak, D. Ross -Kaschitza, E. V. 609 Zaklyazminskaya, A. Hudmon, G.S. Pitt, J.S. Rougier, and H. Abriel. 2020. 610 Calmodulin binds to the N-terminal domain of the cardiac sodium channel Nav1.5. 611 Channels. 14:268–286. doi:10.1080/19336950.2020.1805999. 612 Wiseman, R.L., J.S. Mesgarzadeh, and L.M. Hendershot. 2022. Reshaping 613 endoplasmic reticulum quality control through the unfolded protein response. Mol 614 Cell. 82:1477–1491. doi:10.1016/j.molcel.2022.03.025. 615 Zheng, Y., X. Wan, D. Yang, A. Ramirez -Navarro, H. Liu, J.D. Fu, and I. Deschênes. 616 2021. A Heart Failure-Associated SCN5A Splice Variant Leads to a Reduction in 617 Sodium Current Through Coupled- Gating With the Wild -Type Channel. Front 618 Physiol. 12. doi:10.3389/fphys.2021.661429. 619 620 621 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 27 Abbreviations 622 CMV cytomegalovirus CTD C-terminal domain DNE dominant-negative effect DTT dithiothreitol HA haemagglutinin INa Na+ current LDS lithium dodecyl sulfate Na+ sodium ion Nav1.x α-subunit of the voltage-gated sodium channel Navβ β-subunit of the voltage-gated sodium channel NTD N-terminal domain PBS Phosphate-buffered saline SDS-PAGE sodium dodecyl sulfate-polyacrylamide gel electrophoresis TBS Tris-buffered saline TUN tunicamycin UbC ubiquitin C UPR unfolded protein response VGSC voltage-gated sodium channel .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 28 Figure 1 . In a heterologous expression system , Nav1.5 proteins co - immunoprecipitate and interact in living cells. Na v1.5 dimers are expressed at .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 29 the cell surface and in the cytoplasm, can be visualized in non -reducing SDS- PAGE, and are disrupted by DTT. (A) Immunoblot of reduced total lysate (“INPUT”) and HA -specific immunoprecipitated (“IP:HA”) fractions from tsA201 cells 48 hours after transient co- expression of 3xHA -Nav1.5, 3xFLAG -Nav1.5, and 1xGFP -hTpr. Nuclear basket protein, which had a similarly size d plasmid, was used as a transfection control. Immunoprecipitation was performed with HA -coupled magnetic beads. Endogenous α1- syntrophin was used as a positive control for co- immunoprecipitation with Na v1.5, and α -actin was a negative control. (B) Schematic illustration of the NanoBiT assay, whereby different quantities of untagged Nav1.5 are co-expressed together with the luminesce-producing pair of Nav1.5-LgBiT and Nav1.5- SmBiT. Background signal was determined b y co-expression of Nav1.5-LgBiT with a non-interacting control, HaloTag-SmBiT. (C) In living cells, the presence of untagged Nav1.5 significantly decreased the luminescence produced by the interaction between Nav1.5-LgBiT and Na v1.5-SmBiT. This signifies that untagged Na v1.5 competed for the interaction between Na v1.5-LgBiT and Na v1.5-SmBiT, therefore the observed luminescence signal was Nav1.5-specific. Results of the NanoBiT assay are presented as the relative intensity of luminescence (indicating the level of protein- protein interactions) normalized to the fluorescence (indicating the number of cells) of tsA201 cells 48 hours after transfection. Each dataset was normalized to “+ empty vector” control. Data are presented as mean ± SEM from three biological replicates. Multiplicity-adjusted p-values, calculated with one-way ANOVA and post hoc Tukey’s multiple comparisons, are indicated in each panel. (D) Immunoblot of the non- reduced total lysate from tsA201 cells 48 hours after transient expression of untagged Nav1.5 alone and with 1xGFP-Nav1.5. White triangles indicate bands corresponding to monomeric untagged Na v1.5 and monomeric 1xGFP -Nav1.5. The difference in monomer sizes is affected by the GFP tag (~30 kDa). Thus, the upper bands (~460 kDa) revealed with anti -Nav1.5 and anti-GFP antibodies were considered to indicate dimers of Nav1.5. (E) Immunoblots of native Nav1.5 obtained from lysates of excised mouse hearts prepared using a benchtop homogenizer, direct lysis of isolated cardiomyocytes, and a tissue homogenizer. (F) Immunoblot of non- reduced and reduced total lysates of tsA201 cells 48 hours after transient co-expression of mouse and human Na v1.5. Dimers and monomers of Na v1.5 were revealed with an anti - Nav1.5 antibody. Endogenous calnexin was used as a loading control. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 30 Figure 2. Disulfide bridges are involved in oligomerization of Nav1.5 proteins but are not exclusive t o this process. (A) Immunoblot of non- reduced total lysate .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 31 (“INPUT”) and biotinylated (“CELL SURFACE”) fractions of wild-type (WT) and tsA201 cells stably expressing Nav1.5. Dimers and monomers of Na v1.5 were revealed with an anti-Nav1.5 antibody. Endogenous α1-Na+/K+-ATPase was used as a positive control for the biotinylated fraction. (B) Schematic illustration of full-length Nav1.5 with numbers of Cys corresponding to each domain indicated, and truncated Na v1.5 constructs with total number s of Cys. Nav1.5-R535X comprises the presumable α-α- subunit interaction site between Arg493-Arg517, while Nav1.5-I450X does not include this site . (C) Immunoblot of the reduced total lysate (“INPUT”) and HA -specific immunoprecipitated (“IP:HA”) fractions of tsA201 cells 48 hours after transient co- expression of 3xHA -Nav1.5, untagged Na v1.5, 1xFLAG -Nav1.5-R535X, 3xFLAG - Nav1.5-R535X-extraCys-mut, 3xFLAG -Nav1.5-R535X-allCys-mut, 3xFLAG- Nav1.5- I450X, and 3xFLAG -Nav1.5-I450X-allCys-mut. Ponceaus S staining was used as a loading control. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 32 Figure 3. Nav1.5 dimers are unaffected by co -expression of Navβ-subunits. (A) Immunoblot of non -reduced and reduced total lysate fractions of tsA201 cells stably expressing Nav1.5 after 48 hours of transient expression of Navβ1B, Navβ1, Navβ2, Navβ3, and Navβ4. Dimers and monomers of Na v1.5 were revealed with anti -Nav1.5. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 33 Endogenous calnexin was used as a loading control. Of note, Navβ1B is not expected to have an epitope for detection with anti-Navβ1. (B) Immunoblot of the reduced total lysate (“INPUT”) and FLAG -specific immunoprecipitated (“IP:FLAG”) fractions of tsA201 cells 48 hours after transient co- expression of 3xHA -Nav1.5 and 3xFLAG - Nav1.5 with and without Navβ1B, Navβ1, Navβ2, Navβ3, and Navβ4. Endogenous α - actin was used as a negative control. (C) Intensities of Nav1.5 dimers/monomers from (A) were normalized to calnexin and the control condition (“+ empty vector”). Data are presented as mean ± SEM from three biological replicates. Individual p -values, calculated with a one-sample two-tailed t-test with hypothetical mean value = 1, are indicated in each panel. (D) Intensities of Na v1.5-HA in (B) from “IP:FLAG” were normalized to their corresponding intensities from INPUT and further normalized to calnexin and the control condition (“+ empty vector”). Data are presented as mean ± SEM from three biological replicates. Individual p-values, calculated with a one-sample two-tailed t-test with hypothetical mean value = 1, are indicated in each panel. (E)

of the NanoBiT assay are presented as the relative intensity of luminescence (indicating the level of protein- protein interactions) normalized to fluorescence (indicating the number of cells) of tsA201 cells 48 hours after transfection. Each dataset was normalized to “+ empty vector” control. Data are presented as mean ± SEM from three biological replicates. The multiplicity -adjusted p-values, calculated with one-way ANOVA and post hoc Tukey’s multiple comparisons, are indicated in each panel. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 34 Figure 4. α-α-Subunit interactions occur between full-length Nav1.5 and all four transmembrane domains, the N-terminal domain, and the linker between domains I and II. (A) Immunoblot of non-reduced total lysate (“INPUT”) and reduced HA-specific immunoprecipitated (“IP:HA”) fractions of tsA201 cells 48 hours after .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 35 transient co-expression of 3xFLAG-Nav1.5 with untagged Nav1.5, 3xHA-Nav1.5-NTD, 3xHA-Nav1.5-DI, 3xHA-Nav1.5-L1, 3xHA-Nav1.5-DII, 3xHA-Nav1.5-L2, 3xHA-Nav1.5- DIII, 3xHA -Nav1.5-L3, 3xHA -Nav1.5-DIV, and 3xHA -Nav1.5-CTD. In total lysate , Nav1.5 dimers and monomers were revealed with an anti -FLAG antibody. The immunoprecipitated domains of Nav1.5 were revealed with an anti-HA antibody, while co-immunoprecipitated full-length Nav1.5 proteins were revealed with an anti-FLAG antibody. Endogenous α1-Na+/K+-ATPase was used as a loading control for the total lysate fraction. (B) Intensities of non-reduced Nav1.5 dimers and non-reduced Nav1.5 monomers from (A) were normalized to α1-Na+/K+-ATPase and the condition with co- expression of untagged full-length Nav1.5 (“+Nav1.5”). Data are presented as mean ± SEM from three biological replicates. Individual p-values, calculated with a one-sample two-tailed t-test with hypothetical mean value = 1, are indicated in each panel . (C) Schematic illustration of how hydrophobic interactions between transmembrane domains of Nav1.5 might promote protein aggregation and UPR. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 36 Figure 5. Nav1.5 dimerization is increased at higher expression levels and with ER stress induction. (A) Schematic illustration of how protein overexpression might overload ER folding machinery and trigger UPR. (B) Immunoblot of the non-reduced total lysate fraction of tsA201 cells after 48 hours of transient expression of Na v1.5 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 37 with promoters of different strengths. Dimers and monomers of Nav1.5 were revealed with anti-Nav1.5. Endogenous calnexin was used as a loading control. (C) Intensities of Nav1.5 dimers/monomers from (B) were normalized to calnexin and the condition with the strongest promoter (“+CMV -Nav1.5”). Data are presented as mean ± SEM from three biological replicates. Individual p-values, calculated with a one-sample two- tailed t -test with hypothetical mean value = 1, are indicated in each panel. (D) Schematic illustration of how TUN treatment ( prevents N- core glycosylation) and MG132 treatment (inhibits proteasomal degradation) affect protein folding and UPR. (E) Immunoblot of non-reduced and reduced total lysate fractions of tsA201 cells stably expressing Na v1.5 with the indicated combinations of TUN and MG132 treatments. Endogenous α-actin was used as a loading control. (F) Intensities of non- reduced Na v1.5 dimers/monomers , as well as non- reduced and reduced Na v1.5 monomers from (B) , were normalized to α -actin and the condition with inhibited proteasomal degradation (“+MG132+DMSO”). Data are presented as mean ± SEM from three biological replicates. Individual p-values, calculated with a one-sample two- tailed t-test with hypothetical mean value = 1, are indicated in each panel. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint 38 Figure S1. Nav1.5 does not affect the interaction between proteins of the control pair represented by PRKACA -PRKAR2A. Schematic illustration of the NanoBiT assay, when co-expressing different quantities of untagged Na v1.5 together with the control pair of PRKAR2A -LgBiT and PRKACA- SmBiT. Background signal was determined by co- expression of PRKAR2A -LgBiT with a non-interacting control, HaloTag-SmBiT. Results of the NanoBiT assay are presented as the relative intensity of luminescence (indicating the level of protein-protein interactions) normalized to the fluorescence (indicating the number of cells) of tsA201 cells 48 hours after transfection. Each dataset was normalized to “+ empty vector” control. Data are presented as mean ± SD from six technical replicates. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted October 3, 2025. ; https://doi.org/10.1101/2025.10.02.679760doi: bioRxiv preprint