Time-pregnant NMRI (Janvier Labs, Saint Berthevin, France), FoxG1^cre/WT and Elp3^loxp/loxp mice backcrossed in 129/SvJ genetic background were housed under standard conditions. This study was carried out in accordance with the recommendations of the guidelines of the Belgian Ministry of Agriculture in agreement with European Community Council Directive for the care and use of laboratory animals of 22 September 2010 (2010/63/EU) and approved by the local ethics committee. The generation of the conditional Elp3 knock-out mouse required breeding of Elp3^loxp/loxp mice with FoxG1^cre/WT mice (Hébert and McConnell, 2000), as previously described (Laguesse et al., 2015). The following primers were used for genotyping FoxG1 and Cre recombinase: 5′-GCC GCC CCC CGA CGC CTG GGT GAT-3′, 5′-TGG TGG TGG TGA TGA TGA TGG TGA TGC TGG-3′ and 5′-ATA ATC GCG AAC ATC TTC AGG TTC TGC GGG-3′.

Total RNA was extracted from cortices of E14.5 embryos. RNA extraction was performed using the All prep DNA/RNA/protein kit (Qiagen, Hilden, Germany). All RNA samples were treated with DNAse I (Roche, Basel, Switzerland). Synthesis of cDNA was performed on total RNA, which was reverse-transcribed with SuperScript III reverse transcriptase (ThermoFisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Resulting cDNA was used for quantitative PCR, using Faststart Universal SYBR Green Master (Roche). Thermal cycling was performed on an Applied Biosystem 7900HT Fast Real-Time PCR detection system (Applied Biosystems, Foster city, CA, USA). The quantity of each mRNA transcript was measured and expressed relative to Glyceraldehyde-3-Phosphate deshydrogenase (GAPDH). The following primers were designed with Primer3 software: GJA1 forward 5′-GGA CTG CTT TCT CTC ACG TC-3′ and GJA1 reverse 5′-GAG CGA GAG ACA CCA AGG AC-3′; GAPDH forward 5′-GCA CAG TCA AGG CCG AGA AT-3′ and GAPDH reverse 5′-GCC TTC TCC ATG GTG GTG AA-3′.

Human Glioblastoma (U87) and Adenocarcinoma (Hela) cells were cultured in DMEM medium supplemented with Bovine fetal serum (FBS) 10%. Mouse Neuroblastoma cells (N2A) were cultured in DMEM supplemented with FBS 10% and glutamine 2 mM. Human embryonic kidney HEK-293 lentiX cells (Clontech, Moutain View, CA, USA) and HEK-293 cells were cultivated in DMEM supplemented with FBS 10%, glutamine 1%. To generate HEK-293 stably expressing ELP3-FLAG proteins, cells were transfected with pIRES-Elp3-puro construct and selected in 1 μg/ml puromycin (Sigma Aldrich, St Louis, MO, USA). Cells were maintained in selecting media for 3 weeks and surviving cells were used for experiment after transgene expression confirmation. Cell transfections were performed using lipofectamine 2000 according to manufacturer’s protocol (ThermoFisher Scientific). Cells were lysed or fixed 48 h after transfection. Trichostatin A (TSA; 5 mM in DMSO, Sigma-Aldrich) was added to the medium at a final concentration of 1 μM for 4 h before cell fixation or lysis.

ORFs encoding human Elp3 were cloned into pIRESpuro (Clontech) with a FLAG tag at the C terminus. Flag-HDAC6 in pcDNA3 has been described previously (Viatour et al., 2003). All constructs were sequence verified. Cx43 was subcloned from the clone image MC205621 (Origene, Rockville, MD, USA) by high fidelity PCR using NheI-Cx43 forward primer and EcoRV-reverse primer and inserted into the pCAGGS-IRES-GFP vector. To obtain Cx43-4KR, directed mutagenesis was performed on the pCAGGS-Cx43-IRES-GFP (Agilent, Santa Clara, CA, USA), replacing K9R, K162R, K234R and K264R. Plasmids DNA were prepared using a Plasmid Endofree Maxi Kit (Qiagen).

Lentivirus production and lentiviral infections were performed as previously described (Creppe et al., 2009). Briefly, HEK-293 lentiX cells were transfected with the lentiviral packaging vectors VSVG and R8.91 and the pLL3.7 shELP1 or pLL3.7 shSCR using Fugene6 (Promega, Madison, WI, USA) in Opti-MEM medium. Twenty-four hours after transfection, medium was changed to DMEM-FBS 10%. Seventy-two hours after transfection, supernatant containing the viral particles was collected and passed through 0.22 μm filter. The supernatant was then used to infect N2A cells two times consecutively for 6 h with Polybrene® (Sigma-Aldrich) added at 5 μg/mL. Efficacy of infection was determined by GFP expression.

Embryonic brains (E14.5) were dissected in 0.1 M phosphate-buffered saline pH7.4 (PBS) and were fixed at 4°C in 4% paraformaldehyde (PFA) for 1 h. Fixed samples were cryoprotected overnight in 20% sucrose in PBS at 4°C, embedded in OCT Compound (VWR International, Leuven, Belgium) and sectioned (12 μm) onto slides (SuperFrost Plus, VWR International) using a cryostat. Cells were fixed at RT in 4% PFA for 15 min and rinsed three times with PBS. Frozen cryosections and fixed cells were washed three times in PBS-Triton 0.1% (PBST) and blocked for 1 h at room temperature in PBST containing 10% donkey serum (Jackson Immunoresearch Laboratories, West Grove, PA, USA). Sections were incubated overnight at 4°C with the following primary antibodies: anti-Elp3 (1:1000, gift from J. Svejstrup, Cancer Research UK London Research Institute, South Mimms, UK), anti-Cx43 (1:500, rabbit, Abcam, Cambridge, UK), anti-Cx43 IF1 (1:500, mouse, Max Planck institute, Munchen, Germany, Sosinsky et al., 2007), anti-GFP (1:500, goat, Abcam). After washing, sections were incubated for 1 h at room temperature with either anti-mouse, anti-rabbit, or anti-goat secondary antibodies coupled to Rhodamine-redX or FITC (Jackson Immunoresearch Laboratories). Nuclei were counterstained with Hoechst 33342 (1:1000, ThermoFisher Scientific), washed in PBST and coverslipped using Aqua Polymount (Polysciences Inc, Washington, DC, USA). The slides were stored in the dark at 4°C. For images analyses, sections were analyzed by confocal microscopy using A1Ti confocal microscope (Nikon) and ImageJ software. Images of cortical slices were acquired with a 40× objective with a z-interval of 1 μm (z-stack = 5 images) or with a 60× objective with a z-interval of 1 μm (z-stack = 5 images). The quantifications of intracellular and membrane fluorescent signal intensities in Hela cells were conducted using ImageJ software.

Dissected E14.5 embryonic cortices were incubated in lysis buffer (50 mM Tris-HCl pH 7.4, 450 mM NaCl, 1% triton, 10 mM NaF, 1 mM Na₃VO₄ and proteases inhibitors, Roche). Cells were lysed in another lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% triton, 10 mM NaF, 1 mM Na₃VO₄ and proteases inhibitors, Roche). Proteins were extracted by centrifugation (10,000 g) for 10 min at 4°C, and quantified using BCA method (Pierce). Immunoprecipitation (IP) was carried out using the following antibodies at 1/250: anti-HA (rabbit, Santa Cruz Technology SCT, Santa Cruz, CA, USA), anti-HA (mouse, SCT), anti-Flag (mouse, Sigma-Aldrich), anti-ELP1 (Close et al., 2006), anti-ELP3 (gift from J. Svejstrup, Cancer Research UK London Research Institute, South Mimms, UK), anti-Cx43 (rabbit, Abcam), followed by 1 h incubation with protein A/G plus agarose beads. Anti-HA IP was carried out as control. Beads were washed out six times with the lysis buffer. Protein samples were then mixed with loading buffer and incubated at 95°C for 5 min, then separated by SDS-PAGE and transferred to 0.45 μm PVDF membranes (Millipore, Billerica, MA, USA). Membranes were blocked 1 h in a solution containing non-fat milk, then incubated O/N at 4°C with the following antibodies anti-ELP1 (Close et al., 2006), anti-Cx43 (Abcam), anti-pan-acetylated lysine (Cell signaling technology, Danvers, MA, USA). HRP-conjugated antibodies were applied for 1 h at RT (conjugated anti-mouse, anti-rabbit, GE Healthcare, Waukesha, WI, USA). Membranes were developed with the ECL chemiluminescent reagent (Thermo scientific, Rockford, IL, USA) using Hyperfilm ECL (GE Healthcare). ImageJ was used for optical density quantification.

Statistics for dual comparisons were generated using unpaired two-tailed Student’s t-tests. Statistical analyses were performed using graphPad Prism 5.0 Software (GraphPad software Inc., San Diego, CA, USA). Values are presented as mean ± SEM *p < 0.05, **p < 0.01, ***p < 0.001 for all statistics.

Cx43 is expressed in neurons and progenitors of the developing cortex of the rat brain, with high levels of expression in the VZ/SVZ and reduced levels in the CP (Elias et al., 2007; Qi et al., 2016). We analyzed the expression of Cx43 in E14.5 embryonic mouse brain and observed a punctate staining throughout the cortical wall, with a stronger labeling of the VZ cells (Figures 1A–H). The different subunits of Elongator are expressed in migrating projection neurons, but are also found in the VZ/SVZ progenitor cells (Creppe et al., 2009; Laguesse et al., 2015). We thus immunolabeled embryonic cortices to detect Cx43 and Elp3, and we observed a strong co-expression of both proteins in cortical progenitor cells that are lining the ventricle (Figures 1B,F–H). Co-IP experiments on microdissected tissue from the cerebral cortex of E14.5 embryos demonstrated an in vivo interaction between Cx43 and both Elp1 and Elp3 (Figures 1I–L). We also generated a HEK293 cell line stably expressing flag-Elp3, and we detected a comparable interaction between flag-ELP3 and Cx43 in vitro (Figure 1M). This interaction was further confirmed in a neuroblastoma cell line (Figure 1N).

In the developing mouse heart, Cx43 assembles into GAP junctions expressed at the intercalated discs that physically delimitate cardiomyocytes, and Cx43 acetylation has been shown to control such localization (Colussi et al., 2011). To determine whether Cx43 is acetylated in the developing mouse cortex, we performed western blot to detect acetylated lysine(s) in Cx43 immunoprecipitate from E14.5 mouse cortical extracts (Figure 2A). The acetylation level of Cx43 but not its expression (Figure 2B (inputs), Figure 2C) was reduced in the cortex of E14.5 Elp3cKO embryos (breeding of Elp3loxlox mice with FoxG1: Cre mice (Hébert and McConnell, 2000), as previously described (Laguesse et al., 2015)), as compared to WT embryos (Figure 2B). Altogether, these results show that Elongator is required for the proper acetylation of Cx43 in the mouse developing cerebral cortex.

The transfer or removal of acetyl groups on lysine residues is mediated by two classes of enzymes: the lysine acetyltransferases (KATs) and lysine deacetylases (KDACs), also known as HDACs (Menzies et al., 2016; Simon et al., 2016). HDACs are grouped into four classes, depending on sequence homology. Classes I, II and IV are Zinc-dependent HDACs, whereas class III consists of NAD⁺-dependent sirtuins (SIRT1 and 2). HDACs play important roles in the regulation of transcription, but they also act on a large set of non-histones proteins, like transcription factors, translation-associated proteins, proteins involved in cytoskeleton regulation or cell signaling, to regulate several functions (Yao and Yang, 2011; Roche and Bertrand, 2016). In order to identify the enzyme responsible for Cx43 deacetylation, we treated cultured U87 cells with TSA, a potent inhibitor of class I, II and IV HDACs but not class III sirtuins (Codd et al., 2009), or DMSO as control. As shown in Figure 3A, TSA treatment of U87 cells significantly increased Cx43 acetylation levels, suggesting that the Cx43 deacetylase is a member of HDAC family. HDAC6 is found in the cytoplasm and promotes the deacetylation of multiple targets including α-tubulin, cortactin and HSP90 (Kovacs et al., 2005; Zhang et al., 2007; Li et al., 2011). We thus tested the ability of HDAC6 to promote Cx43 deacetylation. As shown in Figure 3B, HDAC6 overexpression reduced the acetylation of Cx43 in N2A cells, suggesting that HDAC6 is a deacetylase that targets Cx43. We next deciphered whether Elongator also promotes Cx43 acetylation in N2A cell line. For this purpose, we infected N2A cells with a lentivirus delivering shRNA against ELP1, because it has previously been reported to destabilize the Elongator complex (Petrakis et al., 2004; Close et al., 2006; Creppe et al., 2009). As shown in Figure 4A, ELP1 depletion in N2A cells resulted in decreased Cx43 acetylation level. We further showed that inhibition of HDACs activity was not sufficient to counteract the impact of the depletion of ELP1 on Cx43 acetylation (Figure 4B). Altogether, these results suggest that Elongator is necessary for proper acetylation of Cx43, a post-translational modification removed at least in part by HDAC6.

It has been previously shown that acetylated Cx43 delocalizes from the membrane and is mainly found in the cytoplasm and in the nucleus (Colussi et al., 2011). We thus analyzed the cellular localization of Cx43 in Hela cells before and after TSA treatment. We overexpressed Cx43 and showed that in basal conditions, Cx43 formed connexons at the contact points between cells (Figure 5A). However, TSA treatment delocalized Cx43 from the membrane to a more cytoplasmic position (Figures 5B,E), in line with previous results (Colussi et al., 2011). As Cx43 contains many lysine residues, we used the posttranslational modification database PHOSIDA to identify potentially acetylable lysines (Gnad et al., 2011). We identified four lysines (K) predicted to be acetylated with more than 90% confidence: K9, K162, K234 and K264. In order to generate a non-acetylable form of Cx43, we replaced the four corresponding lysines by arginine residues (R), as previously described (Li et al., 2002; Qiang et al., 2010; Colussi et al., 2011; Jiménez-Canino et al., 2016). Cx43-4KR was also found in connexons connecting Hela cells (Figures 5C,E), but unlike Cx43-WT, TSA treatment did not trigger its delocalization from the membrane, and Cx43-4KR was still found at the contact points between cells after TSA treatment (Figures 5D,E). This suggests that the cellular localization of Cx43 is regulated through acetylation, which triggers Cx43 delocalization from the membrane.

As Cx43 acetylation regulates its subcellular localization and is reduced in Elp3cKO embryonic cortices, we assessed the impact of the loss of Elp3 on Cx43 localization. We labeled Cx43 in E14.5 WT and Elp3cKO cortex with a specific antibody, which detects Cx43 only when present at the membrane (IF1 antibody; Sosinsky et al., 2007) and analyzed its distribution in the cortical wall. In WT cortex, we observed membrane Cx43 labeling throughout the cortical wall with some accumulation detected in the VZ/SVZ, as compared to the IZ and CP (Figures 6A–F). In the Elp3cKO cortex, we observed a reduced Cx43 labeling in the IZ/CP compared to the WT littermate (Figures 6G,H), which is likely the consequence of the reduced neuron population observed upon loss of Elp3 in cortical progenitors (Laguesse et al., 2015). However, as shown in Figures 6I–L, our results showed a stronger labeling of Cx43 in the VZ progenitor cells of Elp3cKO embryos compared to the WT cortex, without change of total Cx43 protein and mRNA expression (Figures 2B,C). This suggests that the reduced acetylation of Cx43 occurring upon loss of Elongator activity promotes its membrane localization.