Evidence for the Presence of a Free C-Terminal Fragment of Cx43 in Cultured Cells

Abstract

Migration of the gap junction protein connexin 43 (Cx43) in SDS-PAGE yields 2 to 4 distinct bands, detectable in the 40–47 kDa range. Here, we show that antibodies against the carboxy-terminal domain of Cx43 recognized an additional 20-kDa product. This protein was detected in some culture cell lysates. The presence of the 20-kDa band was not prevented by the use of protease inhibitors (Complete® and phenylmethylsulfonyl fluoride (PMSF), 1–5 mM). The band was absent from cells treated with Cx43-specific RNAi, and from those derived from Cx43-deficient mice, indicating that this Cx43-immunoreactive protein is a product of the Cx43 gene. Treatment of CHO cells with cyclosporin A caused a reduction in the amount of full-length Cx43 and a concomitant increase in the amount of the 20-kDa band. Overall, our data show that a fraction of the Cx43-immunoreactive protein pool within a given cell may correspond to a C-terminal fragment of the protein.

Keywords:

• connexin
• Connexin 43
• Cx43
• gap junction
• Cx43 C-terminal fragment

INTRODUCTION

Connexin 43 (Cx43) is the most abundant gap-junction protein in the heart and other tissues. Immunochemical techniques have been used extensively to characterize the presence and abundance of this protein in a variety of cell systems. It is commonly accepted that Western immunoblots using anti-Cx43 antibodies reveal up to 4 discernible bands, all clustered within the 40–47 kDa range, with each one corresponding to a different phosphorylation state of the protein (Lampe and Lau 2000; Musil et al. 1990). Several investigators have reported that additional forms of immunoreactive Cx43 migrate on SDS-PAGE in the 30–32 kDa range (Giblin and Christensen 1997; Hofer and Dermietzel 1998, Zahs et al. 2003). The source and significance of these bands remain to be determined, though their existence has often been dismissed as consequent to sample degradation (Yeager and Gilula 1992; Zahs et al. 2003). Here, we present evidence indicating that a high-mobility band can be detected in some (but not all) Cx43-expressing cell lines. We further show that the existence of this band is not consequent to protein degradation after cell lysis, but, rather, represents a native Cx43 form that is detectable by antibodies to the carboxyl terminus of the molecule and that can be regulated by selected factors. The possible functional implications and mechanisms of production of this alternative Cx43 form remain to be determined. However, evidence from other laboratories suggests that C-terminal (CT) fragments of Cx43 may participate in important regulatory functions such as cell proliferation and growth (Dang et al. 2003; Moorby and Patel 2001).

METHODS

Cell Culture

The cell cultures used in the present study were obtained from American Type Culture Collection (ATCC) except AT84 which was a gift from Dr. Edward Shillitoe (SUNY Upstate Medical University). NRK (Normal rat kidney cells; CRL-6509), CHO (Chinese hamster ovary cells), bEnd3 (Brain endothelial cells; CRL-2299) and HeLa (adenocarcinoma epithelial cells; CCL-2) cells were cultured in DMEM supplemented with 10% FBS, 100 units/mL penicillin and 100 μ g/mL streptomycin. Cx43-GFP stably transfected HeLa cells were a gift from Dr. Feliksas F. Bukauskas (Bukauskas et al. 2001). These cells express Cx43 with eGFP fused to the C-terminus. AT84 cells (murine squamous cell carcinoma, (Schultz-Hector et al. 1993) were grown in RPMI 1640 with 10% FBS, penicillin and streptomycin. All the cultures were maintained at 37°C and 5% CO₂. Bone marrow-derived macrophages were cultured as previously described (Tomaras et al. 1999).

RNAi Treatment

RNAi silencing constructs were developed using Invitrogen's Block-It™ technology. The constructs used were (both positive and negative strands): 598-GGAUUGAAGAACACGGCAA, 1079-CCG CAAUUACAACAAGCAA and 1247-GCCCUUA GCUAUCGUGGAU. The number designates the position within the Cx43 cDNA sequence. The control sequence, CCGAUUACAACGAAACCAA, was based on siRNA 1079 but was altered to no longer bind Cx43. RNAi oligonucleotides were introduced into bEnd3 cells using Lipofectamine 2000 following the manufacturer's recommendations.

Western Blot

Cells were grown to 80–90% confluency in 100-mm cell culture dishes. After 24 hours the cells were briefly rinsed in ice-cold phosphate buffer saline (PBS; pH 7.4) and harvested by scraping. In some experiments 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 μ g/mL aprotinin and 1 X Complete® protease inhibitor cocktail (with EDTA; Roche) were added to the scraping buffer. The cells were pelleted by centrifugation for 5 min at 14,000 rpm. The cell pellet was suspended in freshly prepared extraction buffer containing 50 mM Tris (pH 8.0), 150-mM NaCl, 0.02% sodium azide, 1% Triton X-100, 1-mM PMSF, 1-μ g/mL aprotinin and 1 X Complete® protease inhibitor. Where noted calpain inhibitors I and II (50 and 100 μ M; Calbiochem) or caspase inhibitor I (20 μ M; Calbiochem) were added. The cells were sonicated and incubated on ice for 30 min. The total cell lysate was centrifuged at 14,000 rpm for 10 min at 4°C. The supernatant was transferred to a fresh tube and concentrated Laemmli sample buffer (Laemmli 1970) was added. The samples were incubated at 37°C for 10 min prior to loading of gels. The proteins were separated on 4–20% Tris glycine mini-gels (Invitrogen) and transferred to nitrocellulose membranes, where they were blocked in 5% non-fat milk in pH 7.4 PBS with 0.1% Tween 20™ for 1 hour with gentle shaking. Cx43 immunodetection was carried out using one of the following three antibodies: rabbit anti-Cx43 C-terminus (AB1728, Chemicon), monoclonal anti-Cx43 C-terminus (CX43CT, Fred Hutchinson Cancer Research Center (FHCRC)) or monoclonal anti-Cx43 N-terminus (Cx43NT1, FHCRC). Primary antibody was diluted in pH 7.4 PBS containing 5% non-fat milk and 0.1% Tween-20™. Protein bands were visualized using ECL substrate (Pierce, Rockville IL) and exposed to film. Blots were then stripped and reprobed for β -actin (A5441, Sigma).

Isolation of Triton X-100–Soluble and Insoluble Fractions

In order to isolate the Triton X-100–soluble and insoluble fractions of Cx43, the cells were prepared in a manner essentially the same as described above except that the cells were not sonicated prior to Triton-X lysis. After 30 min on ice, the cell lysate was centrifuged at 14,000 rpm for 10 min and the supernatant was treated as above. The pellet was resuspended in lysis buffer, briefly sonicated and concentrated Laemmli sample buffer was added prior to heating and PAGE electrophoresis.

RESULTS

Detection of a 20-kDa Protein Fragment by Immunoblot with Antibodies Directed to the CT Domain of Cx43

Figure 1 shows the Cx43-immunoreactive bands detectable in lysates prepared from the following cell lines: NRK (normal rat kidney epithelial), bEnd3 (brain endothelial), AT84 (oral epithelial) and CHO (Chinese hamster ovary cells). Protein extraction was performed in the presence of Complete® protease inhibitor, 1 mM PMSF, and 1 μ g/mL aprotinin to prevent proteolytic degradation during sample processing. Equal amounts of protein were loaded in each lane. As shown in Figure 1(A), a Cx43 polyclonal antibody targeted to the CT domain of the protein recognized a cluster of bands in the 40–45 kDa range. In addition, an immunoreactive product of higher mobility, migrating into the ∼ 20-kDa range, was detected. The amount of this high-mobility form relative to that of the full-length protein varied from one cell line to another, being most prominent in AT84 and bEnd3 cells (65% and 55% of the total immunoreactive protein, respectively), less apparent in NRK cells (13% of the total) and was nearly absent in CHO cells. Moreover, the bands were sharply defined and no “smearing” or laddering was apparent, suggesting that this was a single product rather than non-specific degradation of the sample. After stripping and reprobing the membrane, the same bands were detected by a separate antibody, also targeted to the CT domain of Cx43 (monoclonal Cx43-CT; see Figure 1(B)). However, as shown in Figure 1(C), when a blot of the same samples was run in parallel and screened with a monoclonal antisera to the N-terminal domain of Cx43 (Cx43NT1, FHCRC) only the full-length protein was detected. These results strongly suggested that the high-mobility band corresponded to a Cx43 fragment containing the carboxyl terminal domain of the protein.

Figure 1 Western blot analysis of Cx43-immunoreactive proteins. The following cell lines were probed: CHO (Chinese hamster ovary), AT84 (murine oral cancer cell line), bEnd3 (brain endothelial cell) and NRK (normal rat kidney). Panels A and B display results obtained after immunoprobing the same membrane with two different antibodies: polyclonal anti–Cx43-CT and monoclonal anti–Cx43-CT. For panel C, samples from the same cell lines were tested for immunoreactivity to a monoclonal anti–Cx43-NT antibody. Panel D is an immunoblot of rat heart, brain and ovary extracts probed using the polyclonal anti–Cx43-CT antibody. Note the 20-kDa band present in the extracts of NRK, AT84 and bEnd3 cells.

While the 20-kDa Cx43-immunoreactive protein was detected in a number of cell lines, we failed to identify it in Western blots from freshly isolated tissue using either the polyclonal or the monoclonal antisera against the C-terminus of Cx43. Indeed, as shown in Figure 1(D), full-length Cx43 was abundantly present in extracts from rat heart, ovary and brain, but no alternative high-mobility bands were observed. Overall, these results indicate that a Cx43-immunoreactive 20-kDa protein is present in some (but not all) cell lines and absent in the tissue preparations tested.

It is important to note that all cell and tissue extracts were protected using conventional protocols for immunoblotting of Cx43 (Lampe et al. 2006; Loo et al. 1999; Shao et al. 2005). Samples were prepared using Complete® protease inhibitor cocktail with additional PMSF and aprotinin, were kept in ice and the total extraction procedure was completed in less than 30 min. The lack of smearing surrounding either the full-length or the 20-kDa protein in the Western blots, the fact that the amount of the 20-kDa protein detected varied between cell lines and that these experiments were performed in the presence of protease inhibitors in standard concentrations argued against the idea that this fragment was the result of nonspecific degradation of the Cx43 protein during processing. Yet, to further assess whether the samples were adequately protected from protease digestion after lysis, we performed additional experiments varying the protease inhibition at different stages of sample processing. Figure 2 shows the results. The experiment was conducted on bEnd3 cells, given that these cells present a large amount of the ∼ 20-kDa form (Figure 1). Figure 2(A) demonstrates that the absence of inhibitors, the presence of inhibitors in the lysis and extraction phase only or the presence of inhibitors in the cell scraping, lysis and extraction phases had no effect on the accumulation of the 20-kDa band. To emphasize this point, one sample, shown in Figure 2(A), was directly lysed in Laemmli sample buffer and immediately heated to denature proteases. This was without any effect on the accumulation of the 20-kDa fragment. Protection of Cx43 samples has often centered on the presence of the serine esterase inhibitor, PMSF, during sample preparation (Manjunath et al. 1984, 1985). Figure 2(B) shows the results of varying the concentration of PMSF used in the lysis and extraction step of sample preparation. Clearly, 5 mM of PMSF, five times the amount conventionally used in Cx43 studies (Yeager and Gilula 1992) did not prevent the production of the 20-kDa fragment. In fact, densitometry analysis of the data presented in Figure 2(B) indicates that in the absence of PMSF, the high-mobility band represented 53% of the total immunoreactive protein and when PMSF was present in excess, the same fraction corresponded to 54% of the total protein. Finally, we performed a similar study in which additional protease inhibitors were added during all steps of the procedure. Figure 2(c) demonstrates that the addition of either calpain inhibitors (calpain inhibitor I, 50 μ M; calpain inhibitor II, 100 μ M) or a caspase inhibitor (caspase inhibitor I, 20 μ M) had no effect on the accumulation of the 20-kDa band. All of these studies were performed in the presence of Complete® protease inhibitor (Roche) and the trypsin inhibitor aprotinin. Similar results were obtained from experiments performed on NRK and AT84 cell lines (which endogenously express Cx43) as well as HeLa and N2A cells that were transiently transfected with the Cx43 gene (data not shown). In additional experiments, cells were directly lysed in Laemmli buffer, whereas in others, the concentration of Complete® was increased. In spite of these manipulations, a 20-kDa band of similar density was still detected. Together with results presented in Figure 1, these data strongly support the contention that formation of the ∼ 20-kDa band is not an artifact resulting from Cx43 degradation after lysis.

Figure 2 Protease inhibitors did not alter the production of the 20-kDa fragment of Cx43. Cx43-immunoreactive proteins were detected by Western blot in bEnd3 cell extracts. Samples were lysed in the presence of 1X Complete® protease inhibitor (with EDTA), 1 μ g/mL aprotinin and 1 mM PMSF. Panel A: the protease inhibitors were omitted, added only in the lysis and extraction step or added to both the cell scraping buffer and lysis buffer. In the last lane the cells were scraped in the presence of protease inhibitors and directly lysed in Laemmli sample buffer. In panel B the concentration of PMSF was varied from 0 to 5 mM. In Panel C the cells were scraped and lysed in buffers containing Complete®, aprotinin and PMSF as well as calpain inhibitors I and II (50 and 100 μ M, respectively) or caspase inhibitor II (20 μ M). The relative density of the 20-kDa band was not affected by the presence or absence of protease inhibitors.

The 20-kDa Band is a Product of the Cx43 Gene

The experiments above show that the ∼ 20-kDa band is immunodetected by antibodies directed to the carboxy-terminal (but not the N-terminal) domain of Cx43. Here, we show that this band is directly associated with the presence of the Cx43 gene and its transcript. bEnd3 cells were subjected to treatment with three different siRNA constructs for 48 hours; each siRNA was targeted to a different region of the coding sequence (see details in Methods). The cells were then lysed and immunoblotted with anti–Cx43-CT antibody. While the effectiveness of the three siRNAs varied, the reduction of the intensity of the 43-kDa band was accompanied by a significant decrease in the intensity of the high-mobility band detected at ∼ 20 kDa (Figure 3). The control treatment with Lipofectamine 2000 alone or with a control siRNA sequence did not affect the intensity of either band, further supporting the notion that the loss of the Cx43 transcript was directly responsible for the loss of the 20-kDa Cx43-immunoreactive protein.

Figure 3 siRNA silencing of Cx43 proteins reduces the production of both the full-length and 20-kDa fragment of Cx43. Western blot analysis was performed for Cx43 in bEnd3 cells treated with either lipofectamine alone, lipofectamine with a control siRNA construct or with three different Cx43-specific siRNA constructs (identified as 598, 1079 and 1247, based on their position in the cDNA relative to the 5′ end). Notice that the decrease in the density of the 20-kDa band paralleled the change in the density of the 43-kDa protein, thus suggesting that both proteins are the product of expression of the Cx43 gene. After probing for Cx43 the blot was stripped with SDS and 2-mercaptoethanol and reprobed for actin.

As an alternative approach, we demonstrated that the loss of the Cx43 gene led to loss of both the full-length protein and its associated “small form” (i.e., the ∼ 20-kDa protein detected by the Cx43 antibody). Due to the fact that the 20-kDa band could not be detected from tissue preparations (see Figure 1(d)), these experiments were conducted in 7-day primary cultures of bone marrow–derived macrophages obtained from neonatal mice that were either wild-type, or lacking the Cx43 gene in either one (Cx43 +/–) or both alleles (Cx43 –/–). Cx43 expression was induced by treatment with bacterial lipopolysaccharide (1 μ g/mL, 24 hrs). As can be seen in Figure 4, the density of both Cx43 full length and 20-kDa bands decreased as a result of the loss of one Cx43 allele (lane labeled “Cx43 +/–”) and neither protein was detectable in the cells harvested from the Cx43 –/-animals. These results further support the notion that the 20-kDa “small form” band is the result of the expression of the Cx43 gene.

Figure 4 Cells derived from Cx43 knock-out mice do not produce either the full-length or 20-kDa Cx43-immunoreactive proteins. Western blot analysis was performed for Cx43-immunoreactive proteins expressed in bone marrow–derived macrophages. Cells were harvested from newborn mice that were either wild-type (+/+), heterozygous (+/–) or homozygous-null (–/–) for the Cx43 gene. After 7 days in culture the cells were treated with 1 μ g/mL lipopolysaccharide. After 24 hours, cells were lysed and Cx43 was analyzed by immunoblotting. Notice the parallel decrease in the density of the 43-kDa and the 20-kDa band in the cells from the heterozygote animal and the absence of both bands in cells obtained from the Cx43-null animal. These results further support the notion that the 20-kDa band is the product of the Cx43 gene.

Transfection experiments further demonstrated that the high-mobility band is the product of the Cx43 gene and not dependent on the antibody used for detection. HeLa cells were stably transfected with a cDNA coding for a chimeric Cx43-enhanced green fluorescent protein (GFP), where the enhanced GFP (EGFP) was concatenated in-frame at the carboxy-terminal end of Cx43. To further assess that our results were not dependent on the use of a given antiserum, expression of the exogenous gene was detected using an anti-GFP antibody. As shown in Figure 5(a), both the full-length Cx43-EGFP product, as well as a form of lower molecular weight, could be identified by immunoblot. The form with higher mobility displayed an apparent molecular weight of ∼ 45 kDa, compatible with the expression of the ∼ 20-kDa fragment concatenated to the 25-kDa EGFP molecule. In a parallel experiment HeLa cells were transiently transfected with Cx43-GFP and probed with either anti-GFP or anti-Cx43. As can be seen in Figure 5(B), both antibodies detect the same bands at 68 and 45 kDa. These results are consistent with the observations resulting from cells natively expressing Cx43 and indicate that the immunoreactive ∼ 20-kDa band is the product of the Cx43 gene.

Figure 5 Cx43-GFP chimeric proteins present 2 immunoreactive forms of Cx43. Panel A: GFP-immunoreactive proteins were detected in HeLa cells that were either untreated (lane 1) or stably transfected with a construct expressing a Cx43-GFP fusion protein (lane 2). GFP was fused to the C-terminal of Cx43. Notice the presence of both a 68-kDa band and a 45-kDa fragment only in cells that were transfected with the fusion construct. Panel B: a Cx43-GFP construct was prepared in pEGFP-N1, which fused Cx43 to GFP just after the last amino acid in Cx43. This was transiently transfected into HeLa cells and the proteins were analyzed by immunoblot. The proteins were detected with either anti-GFP or anti-Cx43 antisera.

The 20-kDa Fragment is in the Triton X-100–Soluble Fraction

Oligomerization and the assembly of Cx43 protein into gap junction plaques is associated with a shift of Cx43 protein from a Triton X-100-soluble to -insoluble fraction. In Figure 6, the 20-kDa fragment was shown to be within the Triton X-100–soluble fraction in both bEnd3 cells and AT84 cells. bEnd3 cells, which are communication-deficient, did not have a significant amount of either the full-length or 20-kDa fragment of Cx43 in the insoluble fraction. AT84 cells had Triton X-100–insoluble Cx43 but that fraction had only a trace amount of the 20-kDa fragment (representing less than 1% of the Cx43-immunoreactive protein). NRK cells did not have a significant amount of 20-kDa protein in the soluble fraction and there was no detectable 20-kDa protein in the insoluble fraction. It is not possible to determine from this experiment whether the 20-kDa protein is produced exclusively from the Cx43 protein in the soluble fraction or whether this form is produced from the insoluble protein and becomes more soluble upon its production.

Figure 6 The 20-kDa fragment of Cx43 is in the Triton X-100–soluble fraction of Cx43. Cells were extracted with Triton X-100 to separate soluble and insoluble fractions. Cells were lysed in extraction buffer with 1% Triton X-100, Complete® protease inhibitor, aprotinin and PMSF. After incubation for 30 min on ice, the cells were centrifuged and the soluble protein was removed. The insoluble protein pellets were sonicated in lysis buffer and the Cx43 protein was solubilized in Laemmli sample buffer. The cells tested were NRK, bEnd3 and AT84. When present, the majority of the 20-kDa fragment was in the soluble fraction.

The Production of the 20 kDa Fragment Can Be Regulated

Further support to the notion that the 20-kDa fragment is not a product of degradation after lysis comes from the observation that its presence can be regulated by exogenous agonists. Figure 7 shows results obtained from CHO cells that were treated overnight with cyclosporin A (CsA). This drug is a well-known immunosuppressant that can also alter the regulation of Cx43, particularly under ischemic conditions (Cruciani et al. 1999; Li and Nagy 2000). CHO cells were treated with 50 μ M CsA overnight. As shown in Figure 7(A), CsA caused both a decrease in the density of the band corresponding to the full-length protein and an increase in the density of the high-mobility fragment. Quantitative densitometry indicated that the density of the full-length band was decreased by 39% following treatment with CsA; in contrast, we observed a five-fold increase in the density of the 20-kDa fragment. Densitometry analysis revealed that, before treatment, the 20-kDa band amounted to 9% ± 3.2% (± SEM; n = 3) of the total immunodetectable Cx43, while after treatment the 20-kDa band was 44% ± 2.2% of the total protein (Figure 7(B)). Overall, the effect of CsA on the level of expression of the Cx43-immunoreactive 20-kDa fragment supports the notion that this product is generated in vivo and can be regulated as part of the biological modulation of the Cx43 protein.

Figure 7 The production of the 20-kDa fragment of Cx43 may be regulated in some cells. Western blot analysis of Cx43-imunoreactive proteins expressed in CHO cells that were either maintained in control conditions (untreated) or stimulated with 50 μ g/mL cyclosporin A overnight. Panel A shows an example of the results and panel B depicts a plot summarizing the cumulative results. Notice the increase in the density of the 20-kDa band and the decrease in the density of the band corresponding to the 43-kD protein. For the data in panel B, the density of the 20-kDa band was measured relative to the total amount of Cx43-immunoreactive protein. Actin was used as a loading control. The results suggest that the amount of the 20-kDa CT fragment of Cx43 that is present in the cells can be regulated by exogenous factors.

DISCUSSION

Early studies on the biochemical identification of Cx43 stressed the importance of proper protease inhibitors during the extraction of the protein (Manjunath et al. 1985; Manjunath et al. 1987a, 1987b; Yeager and Gilula 1992). All of the samples used for the experiments presented in Figure 1 were prepared in buffers where PMSF (1 mM), Complete® protease inhibitor (1X, based on the manufacturer's recommendation) and aprotinin (1 μ g/mL) were added immediately prior to use. Application of these inhibitors at the concentrations stated is consistent with current practice in a number of laboratories (e.g., Loo et al. 1999; Shao et al. 2005; Singh et al. 2005). Moreover, protein extraction protocols were performed within 30 min after initiation of cell lysis and samples were kept in ice to further avoid degradation. Yet, as a further test to ensure that the high-mobility band reported in this study was not an artifact of sample preparation, experiments were repeated in the presence of up to 5 mM PMSF. Complete® protease inhibitor has been shown to effectively inhibit serine, cysteine and metalloproteases (Roche, Inc.). As shown in Figure 2(C) the introduction of additional protease inhibitors targeting calpain and caspase also had no effect on the accumulation of the 20-kDa band. In fact, to this point, alterations in the protease inhibitors, time and conditions of lysis did not alter the detection of the 20-kDa band.

As shown in Figure 1, two different antisera to the C-terminus of Cx43 detected the presence of the 20-kD fragment; in contrast, antisera to the N-terminus of Cx43 failed to detect a corresponding N-terminal fragment. This result suggests rapid processing of the N-terminal complement, or at least, of the fragment containing the epitope for the N-terminal antibody. N-terminal processing of connexins has been previously demonstrated, though not for the particular case of Cx43 (Falk and Gilula 1998). Further studies will be necessary to assess the fate of this fragment and the pathways that may be involved in its degradation.

It is our contention that the 20-kDa fragment is a natural by-product of Cx43 processing in a normal, living cell. Such processing has previously been demonstrated for several proteins including E-cadherin (Marambaud et al. 2002), Notch (Hartmann et al. 2001; Selkoe and Kopan 2003; Xia and Wolfe 2003), amyloid precursor protein (APP) (Cupers et al. 2001; Xia and Wolfe 2003), CD74 (Becker-Herman et al. 2005) and Delta (LaVoie and Selkoe 2003). In many cases this proteolysis is regulated (Becker-Herman et al. 2005; Brown et al. 2000; Jung et al. 2003; Selkoe and Kopan 2003) and represents a critical cell-signaling role for the target protein. Often the C-terminal fragment is able to migrate to the nucleus where it can alter transcriptional activity (Becker-Herman et al. 2005; Berezovska et al. 2000; Landman and Kim 2004; LaVoie and Selkoe 2003; Selkoe and Kopan 2003). Our results demonstrate the presence of a Cx43-immunoreactive protein that migrates with higher mobility. Whether this small fragment is biologically relevant remains to be determined. Yet, published data give room for speculation. Indeed, Dang et al. (2003) recently showed that expression of the CT domain of Cx43 caused a reduction in the growth rate of HEK cells in culture, concurrent with a translocation of the Cx43-CT domain into the nuclear compartment. The effect of the Cx43 CT domain on growth rate was confirmed by Moorby and Patel, albeit in a different cell system (Moorby and Patel 2001). A separate study showed that, in several cell lines, the CT domain of Cx43 can regulate cell division, likely by altering the expression of Skp2 (Zhang et al. 2003). In all of these cases, the effect of the CT domain was independent of the formation of gap junctions. While the studies cited above did not indicate the presence of a naturally existing CT fragment, our results may lead to the hypothesis that the production of the endogenous 20-kDa protein may have a functional role in the control of cell growth. The latter would be a novel molecular mechanism to explain the well-established role of Cx43 in cell proliferation and oncogenesis.

We suggest that this Cx43 fragment exists as part of the Cx43 protein pool that can be detected by antibody targeted to the C-terminal domain. The latter has implications in the interpretation of experiments aimed at determining the subcellular localization of Cx43. Indeed, our results strongly suggest that, in some cell lines, a significant fraction of the protein that generates a fluorescent signal after Cx43 immunostaining is not the full-length molecule. In fact in some cells, as little as 40% of the Cx43-immunoreactive signal may represent full-length Cx43. Furthermore, the experiments shown in Figure 5 reveal that a fluorescent Cx43-EGFP fusion protein is also processed and fragmented, hence suggesting that at least some of the protein identified by fluorescent protein tagging may actually be a fragment of Cx43. The latter invites caution in the interpretation of experiments that rely exclusively on Cx43 visualization either by CT antibodies or by tagging with C-terminal fusion of fluorescent molecules.

In summary, our studies show the presence of a 20-kDa protein that immunoreacts with antibodies targeted to the CT domain of Cx43. The evidence indicates that this protein corresponds to the carboxyl-end fragment of Cx43. We further show that the abundance of this protein can be regulated exogenously, supporting the notion that this fragment is produced in living cells. The function of this fragment is yet unknown, though we speculate that it may be involved in processes that regulate cell division and growth. These findings may have implications in the understanding of the role of Cx43 in carcinogenesis.

This work was supported by NIH grants GM57691, HL080602 and HL39707.