The gap junction protein connexin43 is degraded via the ubiquitin proteasome pathway.

We investigated the degradation of the gap junction protein connexin43 in E36 Chinese hamster ovary cells and rat cardiomyocyte-derived BWEM cells. Treatment of E36 cells with the lysosomotropic amine, primaquine, for 16 h doubled the amount of connexin43 detected by immunoblotting and modestly increased the half-life of connexin43 in pulse-chase studies, suggesting that the lysosome played a minor role in connexin43 proteolysis. In contrast, treatment with the proteasomal inhibitor N-acetyl-L-leucyl-L-leucinyl-norleucinal led to a 6-fold accumulation of connexin43 and increased the half-life of connexin43 to 9 h. The role of ubiquitin in connexin43 degradation was examined in an E36-derived mutant, ts20, which contains a thermolabile ubiquitin-activating enzyme, E1. E36 and ts20 cells grown at the permissive temperature contained similar amounts of connexin43 detectable by immunoblotting. Heat treatment dramatically reduced the amount of connexin43 detected in E36 cells, while connexin43 levels in heat-treated ts20 cells did not change. E36 cells that were heat-treated in the presence of N-acetyl-L-leucyl-L-leucinyl-norleucinal did not lose their connexin43. Pulse-chase experiments showed the reversibility of the block to connexin43 degradation in ts20 cells that were returned to the permissive temperature. Finally, sequential immunoprecipitation using anti-connexin43 and anti-ubiquitin antibodies demonstrated polyubiquitination of connexin43. These results indicate that ubiquitin-mediated proteasomal proteolysis may be the major mechanism of degradation of connexin43.

We investigated the degradation of the gap junction protein connexin43 in E36 Chinese hamster ovary cells and rat cardiomyocyte-derived BWEM cells. Treatment of E36 cells with the lysosomotropic amine, primaquine, for 16 h doubled the amount of connexin43 detected by immunoblotting and modestly increased the half-life of connexin43 in pulse-chase studies, suggesting that the lysosome played a minor role in connexin43 proteolysis. In contrast, treatment with the proteasomal inhibitor N-acetyl-L-leucyl-L-leucinyl-norleucinal led to a 6-fold accumulation of connexin43 and increased the half-life of connexin43 to ϳ9 h. The role of ubiquitin in con-nexin43 degradation was examined in an E36-derived mutant, ts20, which contains a thermolabile ubiquitinactivating enzyme, E1. E36 and ts20 cells grown at the permissive temperature contained similar amounts of connexin43 detectable by immunoblotting. Heat treatment dramatically reduced the amount of connexin43 detected in E36 cells, while connexin43 levels in heattreated ts20 cells did not change. E36 cells that were heat-treated in the presence of N-acetyl-L-leucyl-L-leucinyl-norleucinal did not lose their connexin43. Pulsechase experiments showed the reversibility of the block to connexin43 degradation in ts20 cells that were returned to the permissive temperature. Finally, sequential immunoprecipitation using anti-connexin43 and anti-ubiquitin antibodies demonstrated polyubiquitination of connexin43. These results indicate that ubiquitin-mediated proteasomal proteolysis may be the major mechanism of degradation of connexin43.
Gap junctions are plasma membrane structures that contain groups of channels that allow the passage of ions and small molecules between adjacent cells (1). While their microscopic appearance suggests that gap junctions might be stable structures, in fact, studies have suggested that they are quite dynamic. The turnover and degradation of gap junctions may have profound consequences for intercellular interactions in several physiological or pathological situations. Examples are as follows: hepatocyte gap junctions are rapidly degraded, and cells uncouple after partial hepatectomy followed by gap junction resynthesis associated with liver regeneration (2); ventricular myocyte interconnections at gap junctions are remodeled and redistributed in the zones bordering healed myocardial infarcts in a manner that may predispose to reentrant arrhythmias (3), and gap junctions in uterine myometrium are synthesized rapidly at term and degraded rapidly after delivery leading to uncoupling of these smooth muscle cells (4).
The turnover of gap junctions and their constituent proteins also appears to be rapid in normal adult organs and in cultured cells. Fallon and Goodenough (5) determined that hepatic gap junctions have a half-life of about 5 h in vivo. Further turnover studies were facilitated by the molecular cloning of the subunit gap junction proteins (connexins) and the development of specific antibodies. Use of these antibodies for immunoprecipitation of radioactively labeled connexins in pulse-chase experiments has shown that connexin43 (Cx43), 1 connexin26, connexin32, and connexin45 have half-lives of 2-3 h in cultured cells (6 -9). Compared with many other membrane proteins, gap junction proteins turn over rather rapidly. These findings suggest that even under normal conditions, synthesis and degradation of gap junctional channels are very dynamic processes and may provide major mechanisms for the regulation of intercellular coupling and potential remodeling of cellular connections.
The major pathways of protein degradation include lysosomal proteolysis, ubiquitin-dependent lysosomal proteolysis (autophagy), ubiquitin-dependent proteasomal proteolysis, and ubiquitin-independent proteasomal proteolysis. Lysosomes are predominantly involved in the degradation of internalized extracellular materials and receptors. Ubiquitination may regulate the proteolysis of many short-lived and abnormal cellular proteins (10). The proteasome may degrade proteins in a nonubiquitin-dependent manner (11)(12)(13), and ubiquitin is involved in autophagy (14).
To elucidate mechanisms of gap junction degradation, we characterized Cx43 turnover and degradation in E36 Chinese hamster ovary cells, in rat cardiac myocyte-derived BWEM cells, and in a mutant of the E36 cells (ts20). The ts20 cells have a temperature-sensitive defect in the ubiquitin-activating protein, E1, and fail to degrade short-lived or abnormal proteins at the restrictive temperature (10).

EXPERIMENTAL PROCEDURES
Materials-All reagents were obtained from Sigma unless otherwise noted. The protease inhibitors aprotinin, pefabloc, leupeptin, EDTA, antipain, bestatin, and E-64 were purchased from Boehringer Mannheim Biochemicals. A rabbit antiserum directed against a synthetic peptide representing amino acids 252-271 in Cx43 was produced previously and has been extensively characterized (15). Mouse monoclonal antibodies (IgG1) against Cx43 (amino acids 252-270) or against ubiquitin were purchased from Chemicon (Temecula, CA). * This work was supported by National Institutes of Health Grants HL45466 and EY08368. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a fellowship from the American Heart Association (Missouri Affiliate). To whom correspondence should be addressed: Div. Cell Culture-E36 and ts20 cells were provided by Dr. Alan L. Schwartz (Washington University School of Medicine). E36 and ts20 cells were grown at 30.5°C in minimum essential medium supplemented with 4.5 g of glucose/liter (Life Technologies, Inc.), 10% fetal calf serum (JRH Biosciences, Lenexa, KS), 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.). BWEM cells were grown at 37°C in 50% Dulbecco's modified Eagle's medium, 50% Hams' F-12 medium buffered with HEPES (Life Technologies, Inc.) supplemented with 4% fetal calf serum (JRH Biosciences, Lenexa, KS) and 100 units/ml penicillin and 100 g/ml streptomycin (Life Technologies, Inc.).
Heat Treatment and Metabolic Labeling of Cells-Cells were heattreated for 1 h at 44°C in normal medium buffered with HEPES and transferred to 39.5°C and incubated for 2 h. In some of the immunoblotting experiments, the heat-treated cells were co-incubated with the proteasomal inhibitors 20 M ALLN, 20 M ALLM (13), or 200 M primaquine (16). In metabolic labeling experiments and pulse-chase experiments, cells were labeled for 2 h in methionine-depleted normal media containing [ 35 S]methionine (100 Ci/ml) at the indicated temperature. Cells were chased in normal media supplemented with 2 mM methionine at the indicated temperature.
Ubiquitin Conjugation Assay-The ubiquitin conjugation activity in cellular extracts was used as a measure of E1 activity. Lysates were prepared from parental E36 cells or ts20 cells in ice-cold 20 mM Tris-HCl, pH 7.2, 0.25% Triton X-100, and 2 mM dithiothreitol, collected, incubated on ice for 20 min, and then centrifuged. Equal aliquots of protein as determined by the Bio-Rad protein assay (17) were incubated with 125 I-ubiquitin at 30.5°C in the presence or absence of added ATP. These aliquots were analyzed as outlined in Handley-Gearhart et al. (18) to measure ubiquitin conjugation activity.
Immunofluorescent Labeling of Cultured Cells-Cells were cultured in plastic chamber microscope slides (Nunc, Naperville, IL), fixed in 50% methanol, 50% acetone for 2 min at room temperature, and permeabilized in 1% Triton X-100/PBS. Cells were incubated in a primary antibody (mouse monoclonal anti-Cx43) at 1:400 dilution overnight at 4°C and secondary antibodies (CY3-conjugated goat anti-mouse IgG) (Jackson Immunoresearch, West Grove, PA) at 1:800 dilution for 1 h at room temperature with intervening washes. The cells were viewed on a Nikon Optiphot microscope at using a 60ϫ (NA 1.4) objective and epifluorescence (21).
Immunoprecipitation-Cultures of ts20 cells and E36 cells were processed for immunoprecipitation as outlined in Laing et al. (21). Cells were metabolically labeled with [ 35 S]methionine and then were lysed by boiling in RIPA buffer, (PBS containing 1% Triton X-100, 0.6% SDS, 100 /ml aprotinin, 0.5 mg/ml pefabloc, 0.5 g/ml leupeptin, 0.5 mg/ml EDTA, 50 g/ml antipain, 50 g/ml bestatin, 1 g/ml E-64, 10 mM NaF, 10 mM NaVO 4 ). This material was then centrifuged at 14,000 ϫ g for 15 min to remove the insoluble material. In other experiments, immunoprecipitated material was resolubilized by boiling in 1 ml of RIPA buffer for 3 min and reacted with monoclonal antibodies directed against ubiquitin, and these antibody-antigen complexes were precipitated with immobilized anti-mouse IgG (Pierce, Rockford, IL).
Densitometric Analysis-Densitometric images were generated by using a Dage-MTI CCD72 camera (Dage) and digitized with a Matrox MVP image processing board. Gray scale values were obtained by using an FL-4000 (Georgia Instruments). The relative amount of Cx43 in each experiment was calculated by subtracting the background gray scale value from the gray scale value in each experiment. In pulse-chase experiments, the gray scale value was normalized to unchased values (which were assigned the value of 1). The first-order decay constant (k) was calculated for the different treatments from the first-order decay curves of the form y ϭ e Ϫkt were generated with the program ENZFIT-TER. The half-life of the protein was determined using the formula t1 ⁄2 ϭ 0.693/k (9).

RESULTS
To determine possible mechanisms of Cx43 degradation, we treated E36 or BWEM cells with several different protease inhibitors and determined Cx43 levels by immunoblotting and densitometry. (Similar results were seen with both cell lines; those obtained using E36 cells are shown in Fig. 1). Cells were treated for 16 h with either the lysosomotropic amine, primaquine (200 M), or two inhibitors of neutral cysteine proteases, ALLN (20 M) and ALLM (20 M). ALLN and ALLM are similar and can both inhibit calpain and cathepsin D; however, at the concentrations used, only ALLN inhibits proteasomal degradation (13). The electrophoretic mobilities of Cx43 were similar in control and treated samples. The amount of Cx43 was modestly increased (2-fold) in E36 cells treated with primaquine (200 M) as compared with control cultures (Fig. 1, lanes 1 and 2). E36 cells treated with ALLM contained similar amounts of Cx43 to control cultures (1.2 times as much by densitometry) (Fig. 1,  lanes 1 and 4). In contrast, E36 cells treated with ALLN contained 6 times as much Cx43 as control cultures (Fig. 1, lane 3). These data suggested that proteasomal degradation might be the major route of Cx43 proteolysis.
To further test mechanisms of Cx43 degradation, cells were labeled with [ 35 S]methionine for 2 h and then were chased with fresh medium containing 2 mM methionine in the presence or absence of protease inhibitors. The cells were harvested and Cx43 was immunoprecipitated and analyzed by electrophoresis and fluorography. Fig. 2 shows the results of a representative pulse-chase experiment in control E36 cells or primaquinetreated E36 cells. Cx43 was detected as a doublet at 42 and 44 kDa, and the amount of [ 35 S]methionine-labeled Cx43 diminished throughout the labeling period in both control and primaquine-treated cultures. Densitometric quantitation of these immunoprecipitations showed that the half-life of Cx43 was 2.5 h in the control and 3 h in the primaquine-treated cultures. While the turnover of Cx43 differed only mildly, examination of total [ 35 S]methionine-labeled proteins by SDS-PAGE and fluorography confirmed that the primaquine had inhibited the degradation of many other cellular proteins (data not shown).
In contrast, the turnover of Cx43 was significantly prolonged by the proteasomal inhibitor ALLN in both cell lines. The amount of Cx43 immunoprecipitated from E36 cells chased for 3-6 h in the presence of 20 M ALLN was not significantly different from that at the beginning of the chase period (data not shown). Experiments with the chase period extended to 24 h showed that the Cx43 protein had a half-life of 9 h in BWEM cells (Fig. 3). Taken together, these immunoblot and pulsechase experiments suggested that there was limited involvement of lysosomal proteolysis but a major role of the proteasome in Cx43 degradation in these cells.
The role of the ubiquitin-proteasome system in Cx43 degradation was examined in heat-treated ts20 and E36 cells. Heat  4). These cells were harvested, and 5-g aliquots of each were resolved on a 10% polyacrylamide gel and transferred to Immobilon-P. These membranes were probed with the anti-Cx43 monoclonal antibody and detected by chemiluminescence. treatment of many cells, including the E36 cells, leads to a stress-induced degradation of many cellular proteins. But heat treatment (44°C for 1 h followed by 2 h at 39.5°C) of ts20 cells inactivates the thermolabile E1 enzyme and thereby inactivates the ubiquitin-conjugation system (10), as we confirmed by an ATP-dependent ubiquitination assay in lysates of cultures of both E36 and ts20 cells (data not shown). Cellular lysates of heat-treated E36 and ts20 cells were analyzed by immunoblotting with a monoclonal antibody directed against Cx43 (Fig. 4). Untreated ts20 and E36 cells contained similar amounts of Cx43 detected as a major band of 42 kDa (Fig. 4,  lanes 1 and 2). Much less Cx43 was detected in heat-treated E36 cells (Fig. 4, lane 4). Inactivation of ubiquitination apparently prevented the loss of Cx43 in ts20 cells, since heat-treated ts20 cells retained a substantial fraction of the original amount of Cx43 (Fig. 4, lane 3). The loss of Cx43 was blocked in E36 cells that were heat-treated in the presence of ALLN (Fig. 4, lane 5), but not ALLM (Fig. 4, lane 6). These results indicate that the proteolysis of Cx43 in heat-treated E36 cells is dependent on ubiquitination and the proteasome. In addition to these differences, the mobility of some of the Cx43 was reduced in heat-treated cells, suggesting changes in phosphorylation may accompany heat-treatment.
To examine the cellular location of Cx43 in these cells, cultures of E36 and ts20 cells were heat-treated, fixed, and permeabilized, and Cx43 was detected by immunofluorescence (Fig. 5). At permissive temperatures (30.5°C), moderate amounts of Cx43 staining were found at appositional membranes between either ts20 or E36 cells (Fig. 5, A and C) as expected for a gap junctional protein. Heat treatment led to some reduction of the amount of Cx43 detected in the E36 cells; this residual staining was still apparent at the cell membrane (Fig. 5B). After heat treatment of the ts20 cells, the distribution of immunoreactive Cx43 appeared substantially different; while some of the staining appeared to be at appositional membranes, substantial cytoplasmic staining was also detected (Fig. 5D). This staining appeared to be diffuse within the cytoplasm.
We performed further pulse-chase experiments to confirm the reversibility of the accumulation of Cx43 induced by heat treatment, as would be predicted by the reversibility of E1 inactivation. Cultures of ts20 cells were incubated at 44°C for 1 h, followed by incubation in the presence of [ 35 S]methionine in methionine-depleted media for 2 h at 39.5°C. One culture was then harvested for immunoprecipitation, while parallel cultures were chased in normal media for 3 h at 39.5 or at  1 and 2) or heat-treated (lanes 3-6) ts20 (lanes 1 and 3) or  E36 (lanes 2, 4, 5, and 6)  Cultures of ts20 or E36 cells were heat-treated (44°C for 1 h followed by 2 h at 39.5°C) or maintained at 30.5°C, followed by harvesting preparation of cell lysates and analysis by immunoblotting with a monoclonal antibody directed against Cx43.

FIG. 5. Immunofluorescence analysis of Cx43 in untreated (A and C) or heat-treated (B and D) ts20 (A and B) and E36 (C and D) cells.
Cultures of ts20 or E36 cells were heat-treated (as in Fig. 3) or maintained at 30.5°C. Cultures were fixed and permeabilized and stained with a monoclonal antibody directed against Cx43 as detected by indirect immunofluorescence (bar ϭ 50 m). 30.5°C. After the chase period, cells were harvested and lysed in RIPA buffer, and Cx43 was immunoprecipitated and analyzed by SDS-PAGE and fluorography. The cells chased at 39.5°C (Fig. 6, lane 3) contained comparable amounts of Cx43 (42 and 44 kDa bands) to the sample from the start of the chase (Fig. 6, lane 1), while the cells chased at 30.5°C contained only 10% as much Cx43 (44 kDa band only) confirming that degradation of Cx43 was restored (Fig. 6, lane 2).
In order to detect evanescent polyubiquitinated forms of Cx43, we performed immunoprecipitation experiments using E36 cells that were metabolically labeled for 16 h with [ 35 S]methionine in the presence of ALLN to inhibit proteasomal activity. These cells were harvested in RIPA buffer and sequentially immunoprecipitated with antibodies directed against Cx43 followed by an anti-ubiquitin monoclonal antibody. The material immunoprecipitated by anti-Cx43 antibodies was detected as two major bands of 42 and 44 kDa after an overnight exposure (Fig. 7, lane 1). In contrast, subsequent immunoprecipitation with the anti-ubiquitin antibody precipitated a number of bands between 44 and 70 kDa (Fig. 7, lane 2) detected only after a much longer exposure (3 weeks). These bands were not seen when the immunoprecipitation was performed in the presence of an excess of competing ubiquitin (Fig. 7, lane 3). These results appeared similar to the ladders of ubiquitinated polypeptides detected with other substrates of the ubiquitin/ proteasomal apparatus (22). DISCUSSION The major conclusion of this study is that the degradation of Cx43 is dependent upon ubiquitin and the proteasome. This conclusion is supported by several observations of E36, BWEM, and ts20 cells. First, Cx43 accumulated in E36 or BWEM cells treated with the proteasomal inhibitor ALLN, due to decreased turnover of the protein. Second, while in the normal E36 cells heat treatment induced degradation of Cx43, in heat-treated ts20 cells containing a thermolabile ubiquitination pathway, Cx43 remained abundant as assessed by immunoblotting, immunofluorescence, and immunoprecipitation. Third, the heatinduced degradation of Cx43 was blocked by the proteasomal inhibitor ALLN but not by equal concentrations of the related protease inhibitor ALLM. While calpains are inhibited by similar concentrations of both of these neutral cysteine protease inhibitors, the proteasome is inhibited much more effectively by ALLN than by ALLM (13). Fourth, the blocked Cx43 degradation in heat-treated ts20 cells was reversible as assessed by pulse-chase experiments in which returning heat-treated ts20 cells to 30.5°C restored the proteolysis of Cx43. Fifth, polyubiquitinated Cx43 conjugates were isolated from ALLN-treated E36 cells. These could serve as substrates for the proteasome. The rather long exposure time required for their detection and the necessity for inclusion of ALLN, suggested that the ubiquitinated forms of Cx43 may constitute only a small fraction of the total Cx43 pool and may have only a very short lifespan. Sixth, in ts20 cells stably transfected with a cDNA encoding the human E1 enzyme (18), we have found that Cx43 does not accumulate after heat-treatment, 2 confirming that the block to degradation in the ts20 cells is due to the lability of this enzyme.
Our data implicate ubiquitination and the proteasome in the degradation of a very hydrophobic plasma membrane protein, the gap junction protein Cx43. While ubiquitin modification has been implicated in a variety of cellular processes including regulating gene expression, cell cycle and division, the cellular stress response, modification of cell surface receptors, DNA repair, import of proteins into mitochondria, uptake of precursors into neurons, and the biogenesis of ribosomes, mitochondria, and peroxisomes, its exact role in most of these processes is unclear (23). In contrast, the role of the ubiquitin modification in the proteasomal pathway of protein degradation has been extensively characterized, and polyubiquitination is a signal for the proteasome to recognize proteins marked for degradation (23). Numerous cytoplasmic and nuclear proteins involved in signal transduction and cellular replication are substrates of the ubiquitin-proteasome pathway including Nmyc, c-myc, c-fos, and p53 (24). Several plasma membrane proteins, including the c-kit protein (25), Ste6 (26), and the T-cell antigen receptor (27), are modified by polyubiquitin moieties. Mori et al. (28) have shown that proteolysis of another plasma membrane protein, the platelet-derived growth factor receptor, is dependent upon polyubiquitination. Taken together, these studies suggest that ubiquitin plays a role in degradation of proteins from many different compartments of the cell.
Our findings imply that polyubiquitination is the signal for Cx43 degradation, but the structural determinants of Cx43 ubiquitination are currently unknown. Cx43 ubiquitination is unlikely to be governed by the N-end rule (30); it contains "stabilizing" residues as its first two amino acids (methionine and glycine), and N-terminal microsequencing (30) indicates that the amino-terminal residue in Cx43 isolated from rat 2 J. G. Laing and E. C. Beyer, unpublished observations. tissues is glycine. Cx43 might possibly be recognized by a similar ubiquitin protein ligase (E3) and ubiquitin-carrier protein (E2) to those needed to catalyze the proteolysis of another non-N-end rule protein, p53, which is a well known substrate of the ubiquitin proteasome pathway (31). Some proteins that have short half-lives contain a sequence rich in proline, glutamic acid, serine, and threonine residues called a PEST sequence (32). As discussed by Laird et al. (8), potential PEST sequences are present in the Cx43 protein. These sequences may correlate with rapid degradation of PEST box bearing proteins and have recently been shown to play a role in the ubiquitination and degradation of Cln3 cyclin by the ubiquitinconjugating enzyme Cdc34 (33).
We believe that the ubiquitin proteasome pathway may also be a major mechanism for the proteolysis of gap junctions and connexins in other systems. Our data demonstrate involvement of ubiquitination and the proteasome in the normal turnover of Cx43 in two different cell lines and in the accelerated degradation induced by heat stress. Studies of the proteolysis of connexin32 have suggested that this protein may be digested by calpains (34). A Lewis lung carcinoma cell line major histocompatibility complex class I antigen contained an octapeptide differing only at a single position from a portion of mouse connexin37 (35). These cells contained and expressed both mutant and wild-type connexin37 genes, but only the mutant connexin37 peptide was associated with the major histocompatibility complex class I complex (35). Thus, this mutant con-nexin37 protein was proteolyzed by the ubiquitin proteasome pathway (13). Immunohistochemical studies of cultured cardiac myocytes and heart tissue using anti-ubiquitin conjugate antibodies indicated the presence of ubiquitin conjugates in the intercalated disc (36), further suggesting that components of intercellular junctions may be substrates of the proteasome.
Because we observed a 2-fold Cx43 accumulation and a minor prolongation of its half-life in the presence of primaquine, we cannot exclude any role for the endosomal/lysosomal pathway in Cx43 degradation. But, in contrast to our observations, the half-life of P-selectin, a membrane protein that is degraded in the lysosome, is extended significantly by lysosomotropic amines (37). Therefore, we have concluded that the lysosome has only a minor role in Cx43 proteolysis. While we observed accumulation of Cx43 within the cytoplasm of heat-shocked ts20 cells, this staining appeared diffuse, not vesicular. In contrast, several previous morphological studies reported vesicular structures containing morphologically identifiable gap junctions called annular gap junctions; such studies suggested that gap junctions were internalized by endocytosis. Treatments that lead to a loss of morphologically identifiable gap junctions, such as tissue dissociation, ischemia, or anoxia in the liver increase the abundance of annular gap junctions (2). In some tissue culture systems, these vesicles appear to be clathrin coated and associated with actin (38), phagolysosomes (39), or multi-vesicular "complex structures" (40). Further biochemical analyses in such systems will be required to test the generality of our conclusions.