Rat epidermal keratinocytes (REKs) were originally obtained from Dr. Vincent Hascall (Cleveland Clinic, Cleveland, Ohio, United States). REKs are spontaneously immortal but retain the capacity to form an organotypic rat epidermis when grown in an air-liquid interface (Maher et al., 2005; Au et al., 2020). REKs express transcripts for eight connexins, but only express high levels of Cx43 protein with trace levels of Cx26 detectable in overgrown cells (Maher et al., 2005; Au et al., 2020). Therefore, Cx43-ablated REKs (Cx43KO REKs) were generated using CRISPR/Cas9 technology as described and characterized earlier (Au et al., 2020). REKs, Cx43KO REKs, and normal rat kidney (NRK) cells were grown in sterile T25 (Fisher Brand, Burlington, ON, Canada, Cat# FB012935) or T75 flasks (Fisher Brand, Cat# FBO12937) containing Dulbecco’s modified Eagle’s medium (DMEM; ThermoFisher, Burlington, ON, Canada, Cat# 11960) fortified with 10% (v/v) fetal bovine serum (FBS; ThermoFisher, Cat# 12483-020), 2 mM L-glutamine (ThermoFisher, Cat# 25030-081), and 100 U/mL penicillin/streptomycin (ThermoFisher, Cat# 15140-122). Cells were cultured at 5% CO₂ and 37°C (unless otherwise stated) and subcultured when 80%–90% confluent into 35 mm plastic culture dishes or 35 mm 6-well plates using 0.25% (w/v) trypsin-ethylene-diamine-tetraacetic acid (Trypsin-EDTA; ThermoFisher, Cat# 25200056). Connexin-deficient Neuro-2a (N2a) cells were used for electrophysiology experiments and cultured under the same conditions outlined above.

The human Cx30.3-BFP expression vector driven by a CMV promoter was custom ordered from VectorBuilder (Chicago, IL, United States). Employing XhoI and HindIII restriction sites, the BFP tag was replaced by fusing Cx30.3 in-frame to mox-GFP (AddGene, Watertown, MA, United States, Cat# 68070). Plasmids encoding GFP-tagged Cx30.3 mutants (G12D, T85P, and F189Y) were special ordered and purchased from NorClone Biotech Laboratories (London, ON, Canada) and confirmed by sequencing. Red fluorescent protein (RFP) tagged Cx30.3, Cx26, Cx30, and Cx43 were described earlier (Berger et al., 2014; Au et al., 2020; Beach et al., 2020). The myc-DDK (FLAG)-tagged Cx30.3 expression vector driven by a CMV promoter was purchased from Origene (Rockville, MD, United States, Cat# RC204406). This plasmid was outsourced to NorClone Biotech Laboratories (London, ON, Canada) for the generation of the G12D, T85P, and F189Y mutants via site-directed mutagenesis. Resulting plasmids encoding Cx30.3 mutants were verified by sequencing.

REKs and Cx43KO REKs were cultured and grown to ∼50%–60% confluence and transiently transfected by mixing 0.1–1 μg of DNA constructs encoding GFP- or FLAG-tagged Cx30.3, G12D, T85P, and F189Y and 2 μl Lipofectamine 2000 (ThermoFisher, Cat# 11668019) or Lipofectamine 3000 (ThermoFisher, Cat# L3000015) in 200 μl of Opti-Mem reduced serum (ThermoFisher, Cat# 31985-070). This mixture was gently mixed and incubated at room temperature for 20 min (for Lipofectamine 2000) or 5 min (for Lipofectamine 3000) before being added dropwise to the REKs. In some cases, REKs were co-transfected with 1 μg of GFP-tagged Cx30.3 or Cx30.3 mutants with 1 μg of constructs encoding Cx30.3-RFP, Cx26-RFP, Cx30-RFP or Cx43-RFP (Berger et al., 2014; Au et al., 2020). On occasion, REKs or Cx43KO REKs were transfected with a plasmid encoding Cx43-GFP as a control. All transfected cells were used in experiments 24–48 h later.

N2a cells used for electrophysiology experiments were cultured and grown to ∼50%–70% confluence and transiently transfected by combining separate tubes containing 1) 1 μg of DNA constructs encoding GFP-tagged Cx30.3, G12D, T85P, F189Y, and 2) free GFP in 400 μl of Opti-Mem, and 2 μl X-tremeGENE HP DNA transfection reagent (MilliporeSigma, Oakville, ON, Canada Cat# 6366244001) in 400 μl of Opti-Mem and incubated at room temperature for 30 min (Yue et al., 2021; Jaradat et al., 2022). Growth media was aspirated, and cells were incubated with the transfection cocktail for 5 h at 37°C. Subsequently, the transfection cocktail was replaced with fresh serum-containing culture media and cells grown at 37°C and 5% CO₂ overnight and used for electrophysiology experiments the following day as outlined below.

Control cells or cells expressing GFP- or FLAG-tagged Cx30.3, Cx30.3 mutants or co-expressing Cx30.3 mutants together with RFP-tagged Cx30.3, Cx26, Cx30, or Cx43 were plated on sterile glass coverslips (ThermoFisher, Cat# 1254580). Cells were grown to ∼80% confluence, rinsed twice with 1x phosphate-buffered saline (PBS) (ThermoFisher, Cat# 10010-023) and fixed in ice-cold 80% methanol/20% acetone (v/v) solution for 15 min at 4°C. After rinsing in PBS, cells were blocked with 2% (w/v) bovine serum albumin (BSA; MilliporeSigma, Cat# A2153) for 30 min at room temperature. Cells were labelled for 45 min at room temperature with a 1:200 dilution of the following primary antibodies: mouse monoclonal anti-FLAG (MilliporeSigma, Cat# F3165), anti-PDI (Enzo Life Science, Cat# ADI-SPA-891F), anti-E-cadherin (BD Biosciences, Cat# 610182) or rabbit monoclonal anti-Cx43 (MilliporeSigma, Cat# C2619), anti-FLAG (Cell Signaling, Cat# 14793) and anti-BiP/GRP78 (MilliporeSigma, Cat# G8918). Cells were then washed with PBS and incubated for 45 min in the secondary antibody Alexa Fluor 555-conjugated goat monoclonal anti-mouse (ThermoFisher, Cat# A32727) or anti-rabbit (ThermoFisher, Cat# A32732) IgG (1:800 dilution in 2% BSA), or in some cases, AlexaFluor 488-conjugated donkey anti-mouse (ThermoFisher, Cat# A21202) or donkey anti-rabbit (ThermoFisher, Cat# A21206) IgG (both used at 1:800 dilution). All antibodies were diluted as indicated in blocking solution. All coverslips were then stained with Hoechst 33342 (1:1000 dilution in distilled water; ThermoFisher, Cat# H3570) for 1 min to demarcate the nuclei and mounted onto glass microscope slides using Airvol mounting solution. In some cases, cells were only labelled with Hoechst 33342.

Cells were imaged on a Zeiss LSM800 confocal microscope equipped with Zen2.3 software (Zeiss International, Oberkochen, Germany, www.zeiss.com) and a ×63 oil immersion objective lens. When imaging samples for direct comparisons, all images were acquired and presented using identical imaging conditions (laser strength, pinhole, and contrast) to ensure fluorescence intensity could be directly compared. For gap junction quantification, a third-party investigator blinded to the treatment quantified the percent of gap junction forming cell pairs. Gap junction plaques were defined as a linear or punctate green fluorescence signal of approximately 0.2 μm in length, situated at the interface between two transfected cells. Experiments were repeated at least three times, with 30 or more cells or cell-cell interfaces analyzed for each group in each repeat.

The OMA (Ortholog MAtrix) phylogenomic database (https://omabrowser.org/) was queried for the available primary sequences of Cx30.3 and all seven β-connexins from various vertebrate species. Multi sequence alignment was then performed using Clustal Omega in Jalview version 2.11.2.0 (https://jalview.org/). Sequence logos were then generated for the span of nine amino acids flanking the G12, T85, or F189 residues using the WebLogo sequence logo generator (https://weblogo.berkeley.edu/) (Schneider and Stephens, 1990; Crooks et al., 2004; Bai, 2016).

The pathogenicity of missense mutants was predicted using several algorithms, including SIFT (Kumar et al., 2009), SNPs & GO (Calabrese et al., 2009), MutPred2 (Pejaver et al., 2020), PolyPhen-2 (Adzhubei et al., 2010), and CADD (Rentzsch et al., 2021). The sequences of the missense mutants were input into the algorithms and compared against the wild type Cx30.3 reference sequence taken from UniProt (UniProt ID Q9NTQ9; http://uniprot.org) (Bateman et al., 2021). The genome aggregation database (gnomAD; https://gnomad.broadinstitute.org/) was also queried to determine whether the three mutants were common genetic polymorphisms (Karczewski et al., 2020). Allele frequencies for the G12D (allele ID 20046), T85P (allele ID 20045) and F189Y (allele ID 20048) were taken from the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/).

REKs engineered to transiently express FLAG-tagged Cx30.3 or Cx30.3 mutants were subjected to SDS-PAGE and Western blotting as previously described (Au et al., 2020; Lucaciu et al., 2022). Blots were labeled with rabbit anti-FLAG (1:3000) and mouse anti-GAPDH (1:5000; MilliporeSigma, Cat# MAB374) antibodies overnight at 4°C. Primary antibodies were removed, blots were washed, and then labeled with goat anti-mouse IgG AlexaFluor 680 (1:10000) (ThermoFisher, Cat# A-21057) and goat anti-rabbit IgG DyLight™ 800 (1:10000, ThermoFisher, Cat# SA5-10036) for 45 min at room temperature. Secondary antibodies were removed, blots washed, and protein bands were subsequently visualized using the LI-COR Odyssey infrared imaging system. In some cases, control and cells treated with 2 μg/mL of tunicamycin (MilliporeSigma, Cat# 11089-65-9) for 24 h or cells engineered to express Cx30.3 or mutants were immunoblotted for BiP/GRP78 (1:2000) to probe for UPR activation.

Transiently transfected N2a cells were re-plated onto glass coverslips and incubated for 2 h at 37°C and 5% CO₂. Coverslips were transferred to a recording chamber on a fluorescent microscope (Leica DM IRB) and subsequently washed and bathed in extracellular solution (140 mM NaCl, 2 mM CsCl, 2 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES, 4 mM KCl, 5 mM D-Glucose, 2 mM pyruvate, pH 7.4). Glass micropipettes were pulled with a PC-100 puller (Narishige, Amityville, NY, United States) configured to yield a patch pipette resistance in the range of 2–3 MΩ, and back-filled with intracellular solution (130 mM CsCl, 10 mM EGTA, 0.5 mM CaCl₂, 3 mM MgATP, 2 mM Na₂ATP, and 10 mM HEPES, pH 7.2). Isolated transfected cell pairs, identified via fluorescence from the covalently bound GFP tag, were selected for dual whole-cell patch clamp and a voltage clamp at 0 mV was applied to each cell. In an alternating fashion, one cell in the pair was held at a constant holding potential of 0 mV and junctional currents (I _j ) were recorded in this cell during the application of 7 s long trans-junctional voltage (V _j ) pulses (at ±20, ±60, and ±100 mV) to the other cell of the pair as previously described (Xin et al., 2010; Tong et al., 2014; Yue et al., 2021; Jaradat et al., 2022; Lucaciu et al., 2022). The I _j recorded during the ±20 mV V _j pulse was used to calculate the macroscopic junctional conductance (G _j = I _j ÷ V _j ) (Musa et al., 2004; Bai and Cameron, 2016).

REKs were subcultured into 35 mm glass-bottom dishes (Matsunami, Cat# D1123OH) and transfected as described earlier. Cells were loaded with CellTrace™ Calcein Red-Orange, AM (5 μg/mL; ThermoFisher, Cat# C34851) dye solution diluted in 1 mL of OptiMEM reduced serum for 10 min at 37°C and 5% CO₂. The cells were then washed three times with 1x PBS, covered with 2 mL of Opti-Mem and transferred to a live cell imaging apparatus on the Zeiss LSM800 confocal microscope heated to 37°C. A single cell in a small cluster or pair of transfected cells was selected and imaged with a ×63 oil immersion objective prior to photobleaching with the Zeiss LSM800 confocal microscope with Zen2.3 software. The selected fluorescent cells were photobleached to reduce the fluorescence by ∼80% intensity using a 561 nm laser, and fluorescence recovery was imaged every second for 2 min. Fluorescence recovery was quantified using the Time Series Analyzer v3 Plugin for ImageJ (NIH, imagej.nih.gov/ij/) as previously described (Au et al., 2020; Lucaciu et al., 2022). Fluorescence recovery was measured by dividing the fluorescence at each 1-s interval and the level of fluorescence immediately after photobleaching by the initial fluorescence intensity before photobleaching, measured as a function of time. The area under the curve (AUC) was calculated for each bleached cell.

The membrane permeability of wild type and mutant Cx30.3-expressing keratinocytes to propidium iodide (PI) was assessed as previously described with some modifications (Tong et al., 2007; Lucaciu et al., 2022). Briefly, Cx43KO REKs were seeded in six-well plates with a glass coverslip and transfected with 0.1 µg plasmid DNA encoding FLAG-tagged Cx30.3 and Cx30.3 mutants as described above. As controls, one well was untreated (denoted “control”) while another well was treated with all transfection reagents except for plasmid DNA (denoted “lipofectamine”). 24 h after transfection, the growth media was aspirated, and cells were gently washed twice with either divalent cation-free extracellular solution (DCF-ECS; 140 mM NaCl, 5.4 mM KCl, 2 mM EGTA, 10 mM HEPES, 25 mM D-Glucose, pH 7.35, osmolarity adjusted to 298 mOsm/L) or extracellular solution containing divalent cations (ECS; 140 mM NaCl, 5.4 mM KCl, 1.4 mM MgCl₂, 2 mM CaCl₂, 10 mM HEPES, 25 mM D-Glucose, pH 7.35, osmolarity adjusted to 298 mOsm/L). Subsequently, cells were incubated in DCF-ECS or ECS supplemented with 150 µM propidium iodide (PI; 668.5Da; ThermoFisher, Cat# P1304MP) and 2.5 μg/mL calcein, AM (ThermoFisher, Cat# C3100MP) for 15 min at 37°C and 5% CO₂. The dye solution was then aspirated, and cells were washed gently three times with ECS. Cells were then transferred to a Zeiss LSM800 confocal microscope and imaged using a ×10 objective lens. Five random frames (∼400,000 µm²) were captured for each condition. The number of PI-positive cells per frame were quantified using the Cell Counter plugin (https://imagej.nih.gov/ij/plugins/cell-counter.html) for ImageJ (https://imagej.nih.gov/ij/).

Transfection efficiency was quantified in parallel for each condition using cells plated on the coverslips in each dish. Prior to dye loading, the coverslips were removed, and cells were fixed in 10% (v/v) neutral buffered formalin (VWR, Mississauga, ON, Canada, Cat# 10790-714) for 20 min at room temperature. Cells were then permeabilized with 0.1% Triton X-100 (MilliporeSigma, Cat# 9002-93-1) for 10 min at room temperature before immunolabeling for FLAG as described above. Five random fields of view (∼25,500 µm²) were captured using a 40× water-immersion lens and the number of total cells (identified by Hoechst 33342-positive nuclei) and FLAG-positive cells were quantified using the Cell Counter plugin for ImageJ. At least 180 total cells were quantified per coverslip to estimate transfection efficiency.

To assess the ability of sub-physiological temperature incubation to improve protein folding and trafficking, cells expressing Cx30.3 or EKVP-linked mutants were incubated at 26°C for 48 h, fixed and subsequently immunolabelled as described earlier. To assess the potential for rescue of EKVP-linked mutants with chemical chaperones, connexin or mutant expressing cells were treated with 5 mM 4-phenylbutyrate (4-PBA; MilliporeSigma, Cat# 1716-12-7), or 500 μM tauroursodeoxycholic acid (TUDCA; MilliporeSigma, Cat# 14605-22-2). Drug treated cells were fixed and immunolabelled as described earlier. Connexin or mutant expressing cells treated with 10% (v/v) glycerol Caledon Laboratories, Georgetown, ON, Canada Cat# 5350-1) for 24 and 48 h were grown in glass-bottom 35 mm dishes and live-imaged as described earlier.

All data are shown as mean ± standard error of the mean (SEM) and analyzed with either a Mann-Whitney test for two group analysis or a one-way or two-way ANOVA test with multiple comparisons for analysis of more than two groups. Normality tests were performed for all data to determine if parametric or non-parametric analysis should be used. All statistical analysis was performed using GraphPad Prism Version 8 (GraphPad Software, San Diego, California, United States, www.graphpad.com).

Keratinocytes express multiple connexins including Cx30.3, yet surprisingly no studies have previously investigated the ability of human Cx30.3 to assemble into gap junctions when expressed in keratinocytes. Here, we show that GFP-, RFP- and FLAG-tagged Cx30.3 all assembled into gap junctions in Cx43KO REKs similar to GFP-tagged Cx43 (Figure 1, arrowheads). It is notable that GFP- and RFP-tagged Cx30.3 tended to form variable numbers of perinuclear inclusion bodies suggesting that the large fluorescent protein tags may attenuate the efficiency for Cx30.3 to form gap junctions. On the other hand, the small FLAG tag appeared to allow ectopically expressed Cx30.3 to form a greater number of gap junctions although punctate intracellular deposits of Cx30.3 were also identified in some cells. Collectively, we concluded that Cx30.3 is a gap junction forming connexin when expressed in keratinocytes in keeping with the vast majority of connexins that have been studied.

Bioinformatic analysis revealed that all three gene mutations were absent from the genome aggregation database (gnomAD) suggesting that the mutations that lead to G12D, T85P, and F189Y occur at extremely low frequencies and have not been detected in healthy human subjects. Multi sequence alignment analysis revealed that the G12, T85, and F189 residues are highly conserved in Cx30.3 across 86 species, in all seven human β-connexins, and in 675 β-connexins across species (Supplementary Figure S1) suggesting that there is strong natural selection pressure on these positions. In the case of T85, this position can tolerate a closely related serine residue in place of threonine, or in some cases the threonine may be in position 86 instead (Supplementary Figure S1). Subsequent in silico analysis of the pathogenicity of the G12D, T85P, and F189Y mutants using four different algorithms almost unanimously predicted these mutants to be pathogenic, probably damaging, or affecting protein function (Figure 2A). The only exception was the G12D mutant that was predicted to be tolerated by the SIFT algorithm, however, this mutant scored directly at the threshold of being tolerated (threshold score = 0.05) and was predicted to be pathogenic or probably damaging by all other algorithms.

To characterize the ectopic expression and cellular localization of Cx30.3 mutants, we generated and expressed FLAG-tagged Cx30.3 and EKVP mutants in REKs and counterstained the keratinocytes for the adherens junction protein E-cadherin as a proxy marker of cell-cell interfaces (Figure 2B). FLAG-tagged Cx30.3 efficiently assembled into gap junctions while Cx30.3 mutants were frequently found within intracellular compartments (Figure 2B). However, repeatedly a population of Cx30.3 mutants were found to also form a few gap junctions (Figure 2B, see T85P as an example). The ability of the FLAG-tagged Cx30.3 mutants to form gap junctions was found to be reduced by ∼50%–70% when compared to wild type Cx30.3 (Figure 2C). Western blots further revealed that none of the Cx30.3 mutants were particularly unstable or rapidly degraded as all mutants were found at comparable levels to wild type Cx30.3 (Figures 2D, E). However, the F189Y mutant exhibited slightly slower electrophoretic mobility (Figure 2D).

To characterize the subcellular localization of wild type Cx30.3 and EKVP-linked mutants, Cx43-ablated REKs were genetically engineered to express GFP-tagged or FLAG-tagged Cx30.3 or EKVP-linked G12D, T85P, and F189Y Cx30.3 mutants. While wild type Cx30.3 assembled into gap junctions (or intracellular structures resembling inclusion bodies), all three mutants were found localized to a reticular network that was continuous with the nuclear envelope reminiscent of the ER (Figure 3A) or found in paranuclear locations. Further immunolabelling for the ER-lumen resident protein, PDI, revealed that both GFP- and FLAG-tagged G12D, T85P, and F189Y mutants were substantially retained in the ER (Figure 3A), a conclusion supported when the Pearson’s correlation coefficient (PCC) was calculated (Figures 3B, C). To further interrogate the paranuclear compartment housing a population of FLAG-tagged Cx30.3 mutants, mutant expressing cells were immunolabeled for the resident Golgi apparatus protein, GM130 (Figures 4A, B). Here distinct populations of all three mutants were found to localize to the Golgi apparatus. Collectively, we concluded that whether the Cx30.3 mutants were GFP- or FLAG-tagged they exhibited an impaired ability to traffic through the secretory pathway and form prototypical gap junctions, although this impairment was more pronounced when using GFP-tagged mutants.

To assess if the Cx30.3 mutants could form functional gap junction channels when expressed in the absence of other connexins, GFP-tagged mutants were expressed in GJIC-deficient N2a cells. Paired cells that both expressed either wild type or Cx30.3 mutants were subject to patch-clamp electrophysiology and junctional currents were recorded (Figure 5A). Quantification of the gap junctional coupling conductance (G _j ) revealed that wild type Cx30.3 formed functional intercellular channels but all three Cx30.3 mutants and the GFP (negative control) did not (Figure 5B). Cell coupling percentage analysis revealed that ∼70% of N2a cell pairs expressing wild type Cx30.3 were electrically coupled compared to 0% of control or mutant expressing N2a cells (Figure 5C). Confirmation that the Cx30.3 mutants failed to form gap junctions beyond that observed in controls was confirmed when fluorescence recovery after photobleaching was performed on calcein red-orange loaded Cx43KO REK pairs and cell clusters that expressed wild type Cx30.3 or the various Cx30.3 mutants (Figure 5D). Area under the curve was quantified and revealed that the expression of wild type Cx30.3, but not the EKVP-linked Cx30.3 mutants, significantly increased the amount of dye recovery (Figure 5E), suggesting the mutants were not forming functional gap junction channels capable of dye transfer beyond that seen in controls.

Since the Cx30.3 mutants were primarily retained in the ER of REKs, we sought to determine whether BiP/GRP78 levels were elevated as one measure of ER stress. Qualitative immunofluorescence analysis revealed no apparent upregulation of BiP/GRP78 levels compared to neighbouring control cells that did not express the mutants, while BiP/GRP78 expression was clearly induced in keratinocytes treated with tunicamycin (Figure 6A). Quantitative immunoblot analysis showed no evidence of increased BiP/GRP78 levels (Figures 6B, C).

Since several disease-linked connexin mutants have been linked to cell death and/or hyperactive or leaky hemichannels, we evaluated the effect of the mutants on cell membrane permeability to the normally impermeant fluorescent dye propidium iodide (PI). We found that keratinocytes expressing two of the three Cx30.3 mutants exhibited a non-significant trend towards an increase in PI uptake in the presence of extracellular divalent cations (Ca²⁺ and Mg²⁺) but significance was reached for the F189Y mutant (Figures 7A, C). However, when divalent cations were removed in an attempt to potentially induce the opening of hemichannels, we found that keratinocytes expressing all Cx30.3 mutants were significantly more susceptible to PI uptake compared to controls and keratinocytes expressing wild type Cx30.3 (Figures 7B, C). These findings were independent of transfection efficiency as all plasmids were similarly expressed in REKs (Figure 7D).

In the past, various drug and reduced-temperature approaches to improve ER function and protein folding have been employed to increase the intracellular transport and delivery of disease-linked mutants, including a cataract-linked Cx50 mutant, to the cell surface (Cortez and Sim, 2014; Uppala et al., 2017; Jara et al., 2018; Estabrooks and Brodsky, 2020) raising the notion that similar strategies may effectively increase the delivery of Cx30.3 mutants to the cell surface and their assembly into gap junctions. First, we found that incubating Cx43-ablated REKs engineered to express Cx30.3 or EKVP-linked mutants at 26°C for 48 h revealed no improvement in gap junction formation (Supplementary Figure S2). To assess the potential of chemical chaperones to rescue EKVP-linked mutants, Cx43-ablated REKs engineered to express Cx30.3 or EKVP-linked Cx30.3 mutants were treated with either 500 μM TUDCA (Supplementary Figure S3), 5 mM 4-PBA (Supplementary Figure S4), or 10% (v/v) glycerol (Supplementary Figure S5). All treatments failed to improve mutant or wild type Cx30.3 delivery to the cell surface and gap junction formation. These results suggest that conditions designed to improve ER function and protein folding were ineffective at improving the assembly of wild type Cx30.3 or Cx30.3 mutants into gap junctions.

Since Cx30.3 mutants are autosomal dominant and co-expressed with wild type Cx30.3 in patients with EKVP, we wanted to assess if co-expression of wild type Cx30.3 could rescue the assembly of Cx30.3 mutants into gap junctions. Thus, we transfected REKs with equal levels of plasmids encoding GFP-tagged mutants with RFP-tagged wild type Cx30.3 and assessed gap junction plaque levels in adjacent cells expressing both Cx30.3 mutants and wild type Cx30.3 (Figure 8A). Quantification of gap junctions that contained both Cx30.3 mutants and wild type Cx30.3 revealed that all three mutants could be effectively, but differentially, rescued to the cell surface where they intermixed with wild type Cx30.3 within gap junctions (Figure 8B).

Knowing that individual keratinocytes co-express multiple connexins during any stage of their differentiation within the epidermis, we investigated whether co-expressed connexins could intermix with Cx30.3 within the same gap junctions. Confocal imaging analysis revealed that not only did RFP-tagged Cx26 and Cx30 colocalize with GFP-tagged Cx30.3, but surprisingly the more sequence diverse Cx43 also intermixed with Cx30.3 when co-expressed in REKs (Figures 9A, B). Next, we addressed if these same connexin isoforms could selectively intermix with GFP-tagged Cx30.3 mutants and transdominantly rescue their assembly into gap junctions. Our findings revealed that the G12D (Figures 9C, D), T85P (Figures 9E, F) and F189Y (Figures 9G, H) could all be differentially rescued to the cell surface where they intermixed within gap junctions containing co-expressed Cx26, Cx30, and Cx43. These findings provide support to the notion that selective upregulation of keratinocyte connexins could possibly reduce the disease phenotype invoked by the Cx30.3 mutants.