A truncated isoform of Connexin43 caps actin to organize forward delivery of full-length Connexin43

Abstract

While membrane proteins such as ion channels continuously turn over and require replacement, the mechanisms of specificity of efficient channel delivery to appropriate membrane subdomains remain poorly understood. GJA1-20k is a truncated Connexin43 (Cx43) isoform arising from translation initiating at an internal start codon within the same parent GJA1 mRNA and is requisite for full-length Cx43 trafficking to cell borders. GJA1-20k does not have a full transmembrane domain, and it is not known how GJA1-20k enables forward delivery of Cx43 hemichannels. Here, we report that a RPEL-like domain at the C terminus of GJA1-20k binds directly to actin and induces an actin phenotype similar to that of an actin-capping protein. Furthermore, GJA1-20k organizes actin within the cytoplasm to physically outline a forward delivery pathway for microtubule-based trafficking of Cx43 channels to follow. In conclusion, we find that the postal address of membrane-bound Cx43 channel delivery is defined by a separate protein encoded by the same mRNA of the channel itself.

Figure 1.

The presence of GJA1-20k results in the formation of actin puncta. (A) Schematic of GJA1 plasmid designs used in overexpression experiments. Internal translation of truncated isoforms does not occur in plasmids where methionines (green) are mutated to leucines (red). (B) Representative live-cell confocal images of the actin cytoskeleton, visualized by LifeAct-mCherry, in HeLa cells expressing either 0.5 µg of GJA1-20k-GFP or GST-GFP; scale bars = 5 µm. (C) Representative live-cell confocal images of the actin cytoskeleton, visualized by LifeAct-mCherry, in HeLa cells expressing either GJA1-WT (1 µg), GJA1-43k-GFP (1 µg) + GST-GFP (0.5 µg), or GJA1-43k-GFP (1 µg) + GJA1-20k-GFP (0.5 µg); scale bars = 5 µm. (D) Quantification of actin puncta density of all HeLa transfection conditions. Data are presented as mean ± SD (n = 100–200 cells per group from four independent experiments). ****P < 0.0001 by two-tailed Mann–Whitney U test (GST-GFP and GJA1-20k) or by one-way ANOVA with multiple comparisons with Bonferroni’s post hoc test (GJA1-WT, GJA1-43k, and GJA1-43k + GJA1-20k).

Figure 2.

Endogenous GJA1-20k expression results in the formation of actin puncta. (A) Western blot comparison of exogenous (ex.) GJA1-20k-GFP in HeLa cells and endogenous (end.) Cx43 and GJA1-20k in non-transfected (NT) C33A cells. HeLa cells were either NT or transfected with 0.5, 1, or 2 µg of GJA1-20k-GFP (predicted band size for GJA1-20k-GFP = 47 kDa). (B) Representative live-cell confocal imaging of actin, visualized by LifeAct-mCherry, in a wild-type C33A cell; scale bars = 10 µm. (C) Representative live-cell confocal imaging of actin, visualized by LifeAct-mCherry, in C33A cells after treatment with either control siRNA or siRNA-targeting GJA1 alone or with post siRNA knockdown expression of 0.5 µg of either GST-GFP, GJA1-43k-GFP, or GJA1-20k-GFP; scale bars = 10 µm. (D) Quantification of actin puncta density for all C33A transfection cell conditions. Data are presented as mean ± SD (n = 50–65 cells per group from three independent experiments). ns = P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-way ANOVA with multiple comparisons with Bonferroni’s post hoc test. Source data are available for this figure: SourceData F2.

Figure 3.

GJA1-20k inhibits actin polymerization. (A) Representative live-cell confocal imaging of actin, visualized by LifeAct-mCherry, in a HeLa cell expressing 0.5 µg of CP plasmid; scale bars = 10 µm. (B) Quantification of actin puncta density for HeLa cells transfected with GST-GFP, CP, or GJA1-20k-GFP. Data are presented as mean ± SD (n = 50–100 cells per group from three independent experiments). P > 0.05, ***P < 0.001, ****P < 0.0001 by one-way ANOVA with multiple comparisons with Bonferroni’s post hoc test. (C) Representative TIRFM images (t = 15 min after initiation of polymerization) of cell-free Alexa Fluor 488–labeled actin in the presence of increasing concentrations of purified CP (alpha and beta subunits); scale bar = 5 µm. Reduction in actin polymerization is quantified by total fluorescence (A.U.) (D) Data are presented as mean ± SEM (n = 3 experiments per concentration). **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-way ANOVA with comparison to the actin only group with Bonferroni’s post hoc test. (E) Representative TIRFM images (t = 15 min after initiation of polymerization) of cell-free Alexa Fluor 488–labeled actin in the presence of increasing concentrations of GJA1-20k; scale bars = 5 µm. Reduction in actin polymerization is quantified by total fluorescence (A.U.) (F) Data are presented as mean ± SEM (n = 4–5 independent experiments per concentration) **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-way ANOVA with comparison to the actin only group with Bonferroni’s post hoc test.

Figure S1.

GJA1-20k does not decorate the sides of actin filaments and inhibits actin depolymerization. (A) F-actin (2 μM, 5% pyrene labeled) was depolymerized with the addition of 10 μM latrunculin A (LatA) in the presence (cross marks) or absence (circles) of 0.9 μM GJA1-20k. F-actin without LatA does not depolymerize (triangles). (B) In high-speed pelleting assays with F-actin (0–30 μM), the amount of GJA1-20k (0.7 μM) in supernatant fractions does not change as the concentration of F-actin is increased, which indicates the lack of side binding to actin filaments. Co-sedimentation results are shown as a gel (top) and densities which were quantified and plotted (bottom). The intensity of GJA1-20k band in the supernatant lacking actin was set as 1 (100%). Source data are available for this figure: SourceData FS1.

Figure 4.

Structural modeling of the interaction between GJA1-20k and actin. (A) Illustration of Cx43 and GJA1-20k featuring the translation start sites for both the GJA1-20k and GJA1-11k isoforms and the sequence of the final nine amino acids (RPRPDDLEI). (B) Modeling of GJA1-20k (PDB ID: 1R5S) docked to actin. In tan: GJA1-20k ATTRACT models docked using F-actin (PDB ID: 2ZWH), in light blue: GJA1-20k models docked using G-actin (PDB ID: 1J6Z), highlighted in purple: RPRPDDLEI tail sequence of GJA1-20k, red and black text in zoomed in box: hotspot residues as defined by Robetta computational alanine scanning. Hotspots are identified by calculating the free energy change (ΔΔG > 1) after a residue is mutated to alanine. (C) Chart of hotspot residues on actin (red) and GJA1-20k (light blue) as defined by Robetta computational alanine scanning. (D) RPEL domain sequences of Cx43, MAL, MRTF-A, and Phactr1. (E) ATTRACT docking modeling of GJA1-20k (light blue and tan) compared with known RPEL domains of Phactr1 and MRTF-A. In orange: MAL-RPEL2 (PDB ID: 2V52), in pale blue: Phactr1-RPEL1 (PDB ID: 4B1W), in silver: Phactr1-RPEL2 (PDB ID: 4B1X), in yellow: Phactr1-RPEL3 (PDB 1D: 4B1Y); in light green: Phactr1-RPELN (PDB 1D: 4B1U), and in dark blue: full Phactr1 RPEL domain (PDB 1D: 4B1Z).

Figure 5.

The RPEL-like tail of GJA1-20k is sufficient to inhibit actin polymerization. (A) Representative immunofluorescence images of actin (phalloidin) and nuclei (Hoechst) in HeLa cells transfected with 0.5 µg of GJA1-11k-V5; scale bar = 10 µm. (B) Representative western blotting of actin in nuclear and cytoplasmic subcellular fractions of HeLa cells transfected with 0.5 µg of either GJA1-11k-V5 or a GFP-V5–negative control. Lamin-B1 was used as a loading control for the nuclear fraction and tubulin was used as a loading control for the cytoplasmic fraction. (C) Quantification of western blots of actin density in the nuclear and cytoplasmic fraction of HeLa cells transfected with either GJA1-11k-V5 or GFP-V5, normalized to either Lamin-B1 (nucleus) or tubulin- (cytoplasm) loading controls. Data are presented as mean ± SEM (n = 5 separate independent fractionation experiments) *P < 0.05 by two-tailed Mann–Whitney U test. (D) Representative TIRFM images of cell-free Alexa Fluor 488–labeled actin in the presence of increasing concentrations of the αCT11 peptide (RPRPDDLEI) or a reverse control (IELDDPRPR); scale bar = 5 µm. Reduction in actin polymerization is quantified by total fluorescence (A.U.) (E) Data are presented as mean ± SEM (n = 3–5 independent experiments per concentration). ns = P > 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by multiple t tests comparing the αCT11 peptide to the reverse control for each concentration with Holm–Sidak correction for multiple comparisons. No significant difference was found between the actin only group and the reverse control groups (all concentrations), ns = P > 0.05 by one-way ANOVA with multiple comparisons. (F) Representative TIRFM images of Alexa Fluor 488–labeled actin (green) in the presence of 250 nM of Alexa Fluor 555–labeled αCT11 (red) or reverse control (red) to show colocalization (merge); scale bar = 5 µm, images taken 2–3 min after the initiation of polymerization. Images were quantified by using Pearson’s correlation to measure colocalization. (G) Data are presented as box and whisker plots with boxes showing median, 25th, and 75th percentile, with whiskers spanning the 10th to 90th percentile (n = 35 frames from three independent experiments) *P < 0.05, ****P < 0.0001 by two-tailed Mann–Whitney U test. Source data are available for this figure: SourceData F5.

Figure 6.

The RPEL-like motif of GJA1-20k facilitates Cx43 trafficking. (A) Representative frame of a live-cell video of a HeLa cell transfected with LifeAct (0.5 µg) to visualize the cytoskeleton and EB1-YFP (0.5 µg) tracks to visualize dynamic microtubule movement. These cells were also expressing GJA1-WT (1 µg) with no fluorescent tag in order to express full-length Cx43 and its downstream isoforms, or GJA1-43k to express full-length Cx43 in the absence of GJA1-20k; scale bars = 10 µm. White arrows indicate location of intracellular actin puncta. (B) Quantification of EB1 local track thickness (n = 13–15 videos, 1 cell per video). *P = 0.0222 by two-tailed Mann–Whitney U test. (C) Quantification of actin puncta density for cells in A (n = 15 and 13 cells for GJA1-WT and GJA1-43k). Data are presented as mean ± SD. ***P < 0.001 by two-tailed Mann–Whitney U test. (D–G) Representative images of Cx43-NT, N-cadherin, and DAPI in HeLa cells transfected with GJA1-WT or GJA1-ΔαCT11 with or without GJA1-20k (GST used as control). Cx43 is located on cell border–labeled N-cadherin (arrowheads) but αCT11 mutation resulted in less trafficking (asterisks). Scale bars, 10 and 5 µm (insets). Cx43 localization is quantified as Cx43 intensity at cell border (E), total Cx43 (F), and Cx43 ratio of border to total (G). n = 47 (GJA1-WT + GST), 45 (GJA1-WT + GJA1-20k), 49 (GJA1-ΔαCT11 + GST), and 47 (GJA1-αCT11 + GJA1-20k) cells from three independent experimental repeats. ns = P > 0.05, *P < 0.05, **P < 0.01, ****P < 0.0001 by two-way ANOVA with Tukey’s multiple comparisons test.

Figure S2.

GJA1-20k without RPEL motif results in less actin puncta formation compared with GJA1-20k. (A) Representative images of actin puncta formation in HeLa cells transfected with either GJA1-20k or GJA1-20kΔαCT11. Scale bars = 20 and 5 µm (inset). (B) Quantification of actin puncta density in A (n = 113, 127 cells, GJA1-20k, and GJA1-20kΔαCT11, respectively, from three independent experimental repeats). Data are presented as mean ± SD. ****P < 0.0001 with two sample Mann–Whitney U test.

Figure S3.

GJA1-20k expression by GJA1-WT and GJA1-43k plasmids. Western blot comparison of ex. GJA1-20k-YFP in HeLa cells transfected with 1 μg of either GJA1-WT or GJA1-43k. The result shows that the GJA1-WT plasmid expresses the GJA1-20k isoform, whereas the GJA1-43k plasmid does not. Source data are available for this figure: SourceData FS3.

Figure 7.

GJA1-20k rescues gap junction permeability in C33A cells with GJA1 knockdown. (A) Images of cells scraped loaded with Lucifer yellow dilithium salt and rhodamine dextran. Each image is the mean of 30–34 individual images from each corresponding group. Scale bar = 100 μm. (B) Lucifer yellow spreading from the scraping lines from images in A. (C) Normalized areas of dye-coupled cells. Data were calculated by subtracting dextran-positive area from Lucifer yellow–positive area, the result of which was then divided by cut length. n = 30–34 images per group from three independent experimental repeats. Data are presented as box and whisker plots, with boxes showing median, 25th, and 75th percentile, with whiskers spanning the 10th to 90th percentile. ns = P > 0.05, **P < 0.01 by one-way ANOVA with Bonferroni’s post hoc test and multiple comparisons between each group with GJA1 siRNA. (D) Schematic flowchart summarizing the conclusion from Fig. 6 and Fig. 7. (E) Cx43 vesicles are trafficked on microtubules to the cell border to form Cx43 gap junctions. GJA1-20k expression directly reorganizes intracellular actin to produce an increased amount of actin puncta and stabilized F-actin fibers. Actin puncta, where Cx43 vesicles dock, serve as microtubule rest stops and F-actin fibers orient microtubules toward the cell border. Together, these populations of actin, directly produced by GJA1-20k’s functioning as an actin-capping protein, pattern microtubules to increase Cx43 trafficking, and the subsequent amount of Cx43 gap junctions at the cell border.

Figure S4.

Generation of local thickness. Demonstration of how EB1 thickness is calculated from EB1 microtubule tracks (GJA1-WT and GJA1-43k images shown are from Fig. 6 A). Representative images from HeLa cells transfected with either GJA1-WT or GJA1-43k to demonstrate how thickness is calculated from EB1 microtubule tracks. The Trackmate plugin was used to track EB1 particle movements (gray). Next, the Local Thickness plugin generates a distance map (center panel) and then the distance ridge is calculated by removing redundant points from the thickness map. The function Distance Ridge to Local Thickness was then performed to create a distance map of the thickest sections of the image (right panel). Intensity was measured from the right panel images to quantify EB1 microtubule track thickness.

Figure S5.

GJA1-20k rescued microtubule track density in C33A cells with GJA1 knockdown. (A) Representative images visualizing microtubule movement tracked by EB1 over one-minute time-lapse imaging, interval = 1s. Scale bar = 5 µm. (B) Quantifications of microtubule track local thickness (n = 35, 27, 24, and 23 cells from Ctrl siRNA, GJA1 siRNA, GJA1 siRNA + GST, and GJA1 siRNA + GJA1-20k, respectively, from three independent experimental repeats). (C) Representative images of C33A cells stained with tubulin Tracker Deep Red. Scale bar = 10 µm. (D) Quantifications of total tubulin density by measuring areas occupied with tubulin Tracker divided by cell areas (n = 92, 64, 83, and 84 cells from Ctrl siRNA, GJA1 siRNA, GJA1 siRNA + GST, and GJA1 siRNA + GJA1-20k, respectively, from three independent experimental repeats). Data are presented by mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 with the Kruskal–Wallis test with Dunn’s post hoc test.