A Novel Route for Connexin 43 to Inhibit Cell Proliferation: Negative Regulation of S-Phase Kinase-associated Protein (Skp 2)

The connexin genes express proteins forming gap junction channels within the membranes of two adjacent cells. By means of these channels, small molecules (M_r below ∼1000, such as ions, second messengers, and metabolites) can be directly transferred from one cell to neighbor cells (1, 2). This family contains at least 14 members of connexin proteins (2), which are usually named according to the predicted molecular weight.

A large number of studies have indicated the tumor-suppressing character of certain connexin genes, such as Cx43,² Cx32, and Cx26 (reviewed in Ref. 3). However, the mechanisms by which connexin genes inhibit tumor growth still remain obscure. Two reasons could be considered; first, it seems that the tumor-suppressing effects of connexin genes largely depend on the connexin species and cell types used (4). Therefore, the molecular basis for connexin genes to inhibit cell proliferation is likely complicated. There may not exist a uniform pathway for all connexins to inhibit the tumor growth. Second, limited evidence was reported to show that connexin genes causally changed the core machinery controlling cell proliferation, i.e., the cell cycle (3).

Cx43 is one of the most frequently investigated connexin proteins that show growth inhibition effects, not only because it is widely distributed among different tissues (5) but also because a decrease or loss of both Cx43 expression and the Cx43-modulated gap junction function is usually observed in a variety of tumors (see Ref. 6 for details). The tumor suppressing role of Cx43 has been confirmed by the observations that immortalized MEFs derived from the Cx43 −/− mice showed several biological properties of transformed cells (7), and transfection of Cx43 inhibited the proliferation of many tumor cell lines (see Ref. 6 for details). Cx43 inhibited tumor growth probably via secretion of diffusible factors (8), altered gene expression (9, 10), and modulation of apoptosis (6), or cell cycle (11, 12, 13, 14, 15). Recently, we and others have shown that Cx43 increased expression of p27 (13, 14), with a cyclin-dependent kinase inhibitor playing the central role in the cell cycle progression (16). We also reported that the increased level of p27 occurred largely because of the reduced degradation of the proteins, but not because of changes in the p27 mRNA (14). However, the question as to the mechanisms responsible for the lowered p27 degradation remained unanswered.

The ubiquitin-dependent proteolysis is thought to be mainly responsible for p27 degradation (17, 18). Ubiquitination of p27 requires three enzymes: the ubiquitin-activating enzyme E1, the ubiquitin-conjugating enzyme E2, and the ubiquitin ligase E3. The E3 ligase is composed of Skp1, Nedd 8-modified Cul1/Cdc53, Rbx/Roc1, and the F-box protein Skp2. Thus, it is also called the SCF^Skp2 complex (19, 20). Phosphorylation of p27 on the threonine residue 187 (T187) is required for its degradation (21, 22, 23, 24). Once p27 is phosphorylated on T187, it will be recruited to the F-box protein Skp2 of the SCF^Skp2 complex with the aid of Cks1 (25, 26, 27, 28, 29). Then, multiple ubiquitin molecules will be attached to the lysine residues of p27, and finally the ubiquitinated p27 proteins will be degraded by the 26 S proteasome (19). The F-box protein Skp2 plays a central role in p27 ubiquitination by targeting the T187-phosphorylated p27 proteins (25, 26, 27).

Here, we report that Cx43 reduced the expression level of Skp2 via posttranscriptional mechanisms. Skp2 is needed for Cx43 to increase the level of p27. We also provide evidence that Skp2 and p27 are required for Cx43 to inhibit cell proliferation.

U2OS, HeLa, C33A, T98G, and COS-7 cells were cultivated in DMEM containing 10% serum. Human oral squamous cell carcinoma (HSC) cells were grown in MEM containing 5% serum. NOBs were cultivated in αMEM containing 10% serum. Culture of MEFs was performed as described previously (30). Stable cell line generation was performed as described previously (14). The cell pool of the vector and Cx43-stably transfected cells was used in this study.

For cell synchronization at G₀-G₁ phase, U2OS cells were 0% serum-starved for 72 h. Then the cells were released into the S phase by readdition of 10% serum for 20 h. Cells were also arrested at the G₁-S-phase boundary, S phase, and G₂-M phase by treatment with 100 μm mimosine for 42 h, 1 mm thymidine for 24 h, and 1.25 μm nocodazole for 16 h, respectively. At the indicated times, cells were either subjected to cell cycle analyses or immunoblotted.

The antisense phosphorothioate oligodeoxynucleotides of rat Cx43, human p27, and SKP2 have the sequences of: ACTCCAGTCACCCATGTCTG (31), TGGCTCTCCTGCGCC (32) and TCCTGTGCATAGCGTCCGCAGGCCC (33), respectively. Cells were transfected using the FuGENE 6 reagent (Roche Molecular Biochemicals, Indianapolis, IN), as recommended by the manufacturer.

Cell count assay was performed as described previously (14). For colony formation, 4 × 10⁶ U2OS cells were initially seeded into 100-mm dishes. After 20 h, cells were transfected with equal amounts of DNA and cultured for 48 h. Then the cells were split at the ratio of 1:10 dishes and cultured in the presence of 500 μg/ml G418 for 2 weeks. The cell colonies were counted using the Giemsa solution (Merck).

For BrdUrd incorporation assay, cells were plated onto 35-mm glass-covered dishes. After 20 h, the cells were transfected with various DNAs together with one-fourth of the amount of pCMV β-gal. After 32 h of transfection, cells were labeled with 10 μm BrdUrd for an additional 16 h. Then the cells were fixed and stained with rabbit anti-β-gal antibodies (Cappel, ICN Biomedicals, Aurora, OH) followed by the FITC-conjugated antirabbit IgG (Sigma, St. Louis, MO). The BrdUrd incorporation was then detected using 5-Bromo-2′-deoxy-uridine Labeling and Detection kit (Roche). At least 200 β-gal positive cells were counted for each sample, and each experiment was performed in duplicate.

The Student’s t test was performed for statistical evaluation of a difference; P < 0.001 was considered to be statistically significant.

Total cellular RNA was extracted by using the Isogen RNA isolation reagent (Nippon Gene Co., Tokyo, Japan) according to the manufacturer’s instructions. Equal amounts of total RNA were denatured, size-fractionated on 1% agarose formaldehyde gels and transferred to Hybond N⁺ nylon membranes (Amersham, Little Chalfont, Buckinghamshire, United Kingdom). The human SKP2, p27, and GAPDH cDNAs were labeled with [α-³²P]dCTP (Amersham) using the Rediprime random prime labeling system (Amersham), and hybridization was carried out in 50% formamide at 55°C overnight.

Anti-p27 (C-19), anti-Skp2 (H-435), anti-Ubiquitin (FL-76), and anti-HA (F-7) were from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse anti-Cx43 (C13720, clone 2) was purchased from Transduction Laboratories (Lexington, KY). Rabbit anti-phospho-p27 was from Zymed (San Francisco, CA). Anti-FLAG (m5) monoclonal antibody was from Sigma.

Immunofluorescence staining, Western blot analyses, and immunoprecipitation were performed as described previously (14, 34).

Recently, we and others have reported that transfection of Cx43 elevated the level of the cyclin-dependent kinase inhibitor p27 (13, 14), a key cell cycle regulator. We also showed that, although no changes in the mRNA level of p27 were observed, degradation of this protein was inhibited in the Cx43 stably transfected osteosarcoma U2OS cells (14). The p27 degradation was reported to be mainly regulated by the ubiquitin-mediated proteolysis (17, 18). Therefore, to understand how Cx43 inhibited the degradation of p27, we examined whether Cx43 affected p27 ubiquitination or not. As reported (13, 14), the level of p27 was elevated in the Cx43 stably transfected U2OS cells (Fig. 1,A, Lanes 1–3). Treatment with the proteasome-specific inhibitor LLnL induced p27 accumulation (Fig. 1,A; compare Lanes 4–6 with Lanes 1–3), indicating that the 26S proteasome regulates, at least in part, p27 degradation in U2OS cells. Then we examined the ubiquitinated p27 proteins by immunoprecipitating the cell lysates with the anti-p27 antibodies and followed by immunoblotting with the anti-ubiquitin antibodies. The results demonstrate that the proteins with M_r between 75,000 and 105,000 were clearly less in the Cx43-transfected cells than in the parental and vector-transfected cells (Fig. 1 B, the arrow bands). Because it has been reported that the ubiquitinated p27 proteins have similar molecular weights (17), these bands may represent the ubiquitinated p27 proteins. Therefore, these data suggest that Cx43 inhibited the p27 ubiquitination.

Recently, it has been shown that the F-box protein Skp2 regulates ubiquitination of p27 (25, 26, 27). Moreover, Skp2 expression correlates inversely with p27 in human tumors (35, 36, 37, 38). We, therefore, asked whether the elevation in the p27 level induced by Cx43 correlates with a decrease of Skp2 expression or not. The immunoblot analyses revealed that the Cx43-transfected cells expressed a lower level of Skp2 accompanied by a higher level of p27 (Fig. 1,C, Lanes 1–3). Introduction of human SKP2 reduced the p27 proteins to nearly undetectable levels (Fig. 1,C, Lane 5). Although transient transfection of human p27 hardly affected expression of Skp2 (Fig. 1,C, Lane 4), stable expression of p27 significantly reduced the level of Skp2 (data not shown). No changes in the levels of Skp1, Cul1, and Cks1 were observed (data not shown). Because ubiquitination of p27 by Skp2 requires phosphorylation of p27 on the threonine residue 187 (T187) and the subsequent interaction of the T187 phosphorylated p27 with Skp2 (25, 26, 27), we examined the levels of T187-phosphorylated p27 proteins using the specific antibodies. Unlike p27 and Skp2, the cellular level of T187-phosphorylated p27 was not affected by either Cx43 or Skp2 (Fig. 1,C, Lanes 1–3, 5). We further examined the interaction of Skp2 with p27. The immunoprecipitation-Western analyses demonstrated that few Skp2 proteins interacted with p27 in the Cx43-transfected cells (Fig. 1 D). These data suggest that Cx43 inhibited ubiquitination of p27, probably by reducing the level of Skp2, but not by affecting the phosphorylation of p27.

A characteristic of Skp2 expression is its periodicity. It expresses at very low level in the early G₁ phase but begins to increase in the late G₁ phase and reaches the peak in the S phase (39, 40). We have previously shown that Cx43 inhibited cell cycle transition from G₁ to S phase (14). Thus, one may ask whether the observed decrease in the Skp2 level was just a consequent result of the cell cycle arrest in the G₁ phase induced by Cx43. Given that this is true, no difference in the Skp2 level would be observed when the parental, vector-, and Cx43-transfected cells were simultaneously synchronized into the same cell cycle phases. To answer this question, we synchronized cells into different cell cycle phases and investigated the expression pattern of Skp2. When cells were serum-starved for 3 days, no Skp2 was detected (Fig. 2,A, Lanes 1–3). At 20 h after readdition of 10% serum, >70% cells entered the S phase (Fig. 2,B), and Skp2 was readily detected (Fig. 2,A). However, the level of Skp2 was clearly lower in the Cx43-transfected cells (Fig. 2,A, Lanes 4–6), although the difference of the cell population in the S phase was quite small. We also arrested cells at the G₁-S phase boundary, S phase, and G₂-M phase using the specific inhibitors. The flow cytometry data confirmed that the parental, vector-, and Cx43-transfected cells were arrested at the same cell cycle phases (Fig. 2,C). As reported (39, 40), the Skp2 proteins fluctuated with the different cell cycle phases for these three cell lines (Fig. 2,C). However, the levels of the Skp2 proteins were clearly lower in the Cx43-transfected cells than those in the parental and vector-transfected cells, irrespective in which cell cycle phases the cells were arrested (Fig. 2 C, compare Lanes 3, 6, 9, and 12 with Lanes 1, 4, 7, and 10 and Lanes 2, 5, 8, and 11). These data demonstrate that the reduction of Skp2 was cell cycle-independent, suggesting that it was not the consequent result, but rather a probable cause, of the slowed cell cycle progression induced by Cx43, because reduced expression of Skp2 correlated with an elevated level of p27, a protein that controls the G₁-S transition (16). In conclusion, these data suggest that Cx43 increased the level of p27 probably by inhibiting expression of Skp2.

To confirm the role of the Cx43 proteins in Skp2 and p27 expression, we used the specific antisense oligonucleotide of Cx43 to block the stably expressed exogenous Cx43 proteins and then examined expression of the endogenous Skp2 and p27. Compared with the parental and vector-transfected cells, the Cx43-transfected cells expressed lower levels of Skp2 and higher levels of p27 (Fig. 3,A, Lanes 2–4). The Cx43 antisense oligonucleotide reversed the effect of Cx43 on expression of Skp2 and p27 (Fig. 3,A, Lanes 4 and 5), whereas a sense oligonucleotide did not (Fig. 3,A, Lanes 4 and 8). The p27 antisense oligonucleotide showed no effect on the Cx43-reduced Skp2 expression, whereas it significantly suppressed the Cx43-induced p27 accumulation (Fig. 3,A, Lanes 4 and 6). However, the Skp2 antisense oligonucleotide that inhibited expression of the endogenous Skp2 proteins further increased the level of p27 in the Cx43-transfected cells (Fig. 3 A, Lanes 4 and 7). These data confirmed the role of Cx43 in the reduced expression of Skp2. They argue that the Cx43-induced p27 accumulation was likely caused by inhibition of Skp2 expression.

Then, to determine whether Skp2 is a critical target for Cx43 to up-regulate p27, we asked whether ectopic expression of Skp2 is sufficient to abolish the effect of Cx43 on p27 accumulation. As expected, expression of the Flag-tagged mouse Skp2 completely inhibited the Cx43-induced p27 accumulation (Fig. 3,B, Lanes 3 and 6). Skp2 overexpression also blocked the Cx43-induced cell proliferation inhibition (see below, Fig. 7). These data suggest that Skp2 is important for Cx43 to up-regulate p27 and suppress cell proliferation. Interestingly, when cells were simultaneously transfected with the SKP2 gene, the Skp2 protein levels were still lower in the Cx43-transfected cells than those in the parental and vector-transfected cells (Fig. 3,B, Lanes 4–6). Immunoblotting with the anti-Flag antibodies confirmed this reduction (Fig. 3,B, Lanes 4–6). The low level of the exogenous Skp2 was not caused by the reduced transfection efficiency, because these three cell lines expressed essentially the same levels of β-gal proteins simultaneously transfected with the Flag-Skp2 (Fig. 3 B, Lanes 4–6). These data suggest that Cx43 also suppressed expression of the exogenous Skp2 proteins.

In addition, we found that the NOBs (from which U2OS cells are derived) expressed very high levels of endogenous Cx43 and p27 proteins, whereas Skp2 was nearly undetectable (Fig. 3,A, Lane 1). In contrast, the transformed osteosarcoma U2OS cells expressed undetectable levels of Cx43 accompanied by a very high level of Skp2 and a low level of p27 (Fig. 3,A, Lanes 2 and 3). However, when the transformed U2OS cells were forced to stably express levels of Cx43 proteins that were similar to those in NOBs, expression of Skp2 was significantly suppressed, whereas p27 was elevated to the comparable level of that in NOBs (Fig. 3,A, Lanes 1–4). Moreover, proliferation of the Cx43-stably transfected U2OS cells was quite close to that of NOBs, and both were significantly less than that of the transformed U2OS cells (Fig. 3,C). Furthermore, proliferation of the Cx43-transfected U2OS cells was nearly restored to the level that was similar to that of the transformed U2OS cells in the presence of the Cx43 antisense oligonucleotide (Fig. 3 C). These data indicate that the effect of Cx43 may not be an artificial result attributable to a massive expression of the exogenous Cx43 proteins, but rather quite close to what happens in the normal physiological conditions.

To address the mechanism by which Cx43 reduced the protein level of Skp2, we first examined the mRNA level of SKP2. No change in the SKP2 mRNA was observed (Fig. 4 A), indicating that no transcriptional alterations may contribute to the decrease of Skp2 induced by Cx43.

Because the F-box proteins have been shown to be themselves degraded (41, 42, 43), we then assessed the protein stability of Skp2. Treatment with the protein synthesis inhibitor cycloheximide for 1, 2, or 4 h hardly affected the levels of Skp2 proteins in the parental and vector-transfected cells (Fig. 4,B, Lanes 1–8). However, in the Cx43-transfected cells, the Skp2 proteins declined in a time-dependent manner in the presence of cycloheximide (Fig. 4,B, Lanes 9–12). Conversely, the degradation of p27 was inhibited (Fig. 4,B). Quantitative analyses confirmed that Skp2 degraded more quickly in the Cx43-transfected cells (Fig. 4 C). The pulse-chase analysis also showed that Cx43 promoted Skp2 degradation (data not shown). These data suggest that Cx43 suppressed Skp2 expression via an increase in the protein instability.

Recently, the self-ubiquitination mechanism has been proposed for the degradation of the F-box proteins (41, 42, 43). Therefore, to understand how Cx43 enhanced Skp2 degradation, we examined whether or not Cx43 promoted ubiquitination of Skp2 by detecting the endogenous ubiquitinated Skp2 proteins. The results demonstrate that the ubiquitinated Skp2 proteins were higher in the Cx43-transfected cells (Fig. 5,A, Lanes 1–3, and B, Lanes 2–4; the asterisk bands). Treatment with LLnL increased the levels of the ubiquitinated Skp2 proteins (Fig. 5, A and B, the asterisk bands); however, the ubiquitinated Skp2 proteins were still higher in the Cx43-transfected cells (Fig. 5 A and B, the asterisk bands). These data suggest that Skp2 was itself ubiquitinated and Cx43 promoted the Skp2 ubiquitination.

To further confirm that Cx43 increased Skp2 ubiquitination, we transiently transfected cells with the HA-tagged ubiquitin and examined the ubiquitination of Skp2. The results showed that the ubiquitinated Skp2 proteins were readily detected when cells were transfected with HA-ubiquitin; however, the levels of these proteins were clearly higher in the Cx43-transfected cells (Fig. 5,C, Lanes 1–3, the asterisk bands). Treatment with LLnL further increased the levels of these ubiquitinated Skp2 proteins (Fig. 5,C, compare Lanes 4–6 with 1–3, the asterisk bands), but they were still significantly higher in the Cx43-transfected cells (Fig. 5 C, Lanes 4–6, the asterisk bands. No smear bands were observed, indicating that no multiubiquitination of Skp2 may have happened for HA-ubiquitin). These data suggest that Cx43 reduced the level of Skp2, at least in part, by increasing the ubiquitination of the Skp2 proteins.

Cx43 has been reported to inhibit the growth of many tumor cell lines, but the mechanisms were quite different. Thus, we wonder whether regulation of Skp2 and p27 by Cx43 is only specific to U2OS cells or not. Various cell lines were stably transfected with Cx43, and expression of p27 and Skp2 was examined. The results showed that Cx43 reduced the protein level of Skp2 in the tested cell lines; as a result, p27 was increased (Fig. 6). These data indicate that Cx43 has the capacity to overcome, to some extent, the negative effect of Skp2 on p27 expression in certain cell lines.

The data in Fig. 2,B demonstrated that Cx43 reduced the levels of both the endogenous and the exogenous Skp2 proteins. Skp2 has been reported to be able to promote cell proliferation (26, 33); we, therefore, asked whether or not Cx43 could attenuate the Skp2-induced cell proliferation. The colony formation analysis showed that both Cx43 and p27 significantly inhibited the colony formation ability of U2OS cells, albeit the effect of p27 was stronger than that of Cx43 (Fig. 7,A). In contrast, Skp2 significantly promoted U2OS cells to form colonies (Fig. 7,A). However, this effect of Skp2 was partially or strongly inhibited by cotransfection with Cx43 or p27 (Fig. 7, A and B). Similar results were also obtained from the BrdUrd incorporation assay (Fig. 7,C). Parallel immunoblotting showed that Cx43 reduced Skp2 expression in these cells (data not shown). These data suggest that, although Skp2 abolished the inhibitory effects of Cx43 on cell proliferation (Fig. 7, B and C; compare bars of vector, Cx43, and Cx43 with F-Skp2), Cx43 also attenuated the Skp2-enhanced cell proliferation (Fig. 7, B and C; compare bars of Skp2 and Cx43 with F-Skp2).

Finally, to determine the roles of Skp2 and p27 in the Cx43-induced cell proliferation inhibition, we examined the effects of Cx43 on cell proliferation of the p27 −/− and SKP2 −/− MEFs. The BrdUrd incorporation assay showed that Cx43 only weakly suppressed cell proliferation of the p27 −/− MEFs (Fig. 8, bars 3 and 4), whereas it significantly inhibited cell proliferation of the p27 +/− MEFs (Fig. 8, bars 1 and 2). Reintroduction of human p27 significantly inhibited cell proliferation of the p27 −/− MEFs (Fig. 8, bars 3 and 5), and Cx43 enhanced this effect (Fig. 8, bars 5 and 6). Cx43 hardly showed any effect on cell proliferation of the SKP2 −/− MEFs (Fig. 8, bars 9 and 10), but it significantly inhibited cell proliferation of the SKP2 +/− MEFs (Fig. 8, bars 7 and 8). Interestingly, when the mouse SKP2 gene was reintroduced into these SKP2 −/− cells, the inhibitory effects of Cx43 on cell proliferation reappeared (Fig. 8, bars 11 and 12), and the inhibition extent was comparable with those observed in U2OS cells (Fig. 7, B and C). These data suggest that both Skp2 and p27 are required for Cx43 to inhibit cell proliferation, although the results that Cx43 still induced very slight decreases in cell proliferation of both the p27 −/− and the SKP2 −/− MEFs indicate that additional minor mechanisms may also be involved in the Cx43-inhibited cell proliferation.

In addition, the effect of Cx43 on p27 accumulation correlates well with the absence or presence of the SKP2 gene. Cx43 obviously increased the levels of p27 proteins in the p27 +/−, the p27 reintroduced p27 −/−; and the Skp2 reintroduced SKP2 −/− MEFs (Fig. 8, compare Lanes 1, 2; 5, 6; and 11, 12), whereas it weakly and hardly did so in the SKP2 +/− and in the SKP2 −/− MEFs, respectively (Fig. 8, Lanes 7–10). These data demonstrate that the Cx43-induced p27 accumulation is largely dependent on Skp2, confirming the critical role of Skp2 in the Cx43-induced p27 accumulation.

In this study, we have shown that Cx43 reduced expression of the human F-box protein Skp2 that regulates the ubiquitin-dependent p27 degradation, through which Cx43 induced p27 accumulation and cell proliferation inhibition. First, Cx43 induced a reduction in the level of Skp2, probably by increasing the protein instability of Skp2. This reduction correlates with an increase of p27 and a decrease of cell proliferation. Second, the antisense oligonucleotide of Cx43 almost completely blocked effects of Cx43 on expression of the endogenous Skp2 and p27, whereas the Skp2 antisense oligonucleotide further enhanced it. Third, ectopic expression of Skp2 abolished the Cx43-induced p27 accumulation and cell proliferation inhibition. Fourth, Cx43 failed to reduce cell proliferation and elevate the p27 level in the SKP2 −/− MEFs, whereas it did so in the SKP2 +/+ cells. Thus, we suggest a new route for Cx43 to inhibit tumor growth by negative regulation of Skp2 and positive regulation of p27.

Cx43 has been considered as a tumor-suppressing gene (3). However, how Cx43 mediates tumor growth has not been fully characterized. Because Cx43 knockout mice did not survive beyond birth because of conotruncal heart defects (44), we could not observe tumor development using this knockout mouse model. Thus, understanding the tumor-suppressing role of Cx43 will have to depend on other techniques. Transfection of Cx43 has been reported to inhibit growth of many tumor cells (see Ref. 6 for details), but the mechanisms are complicated. Recently, increasing evidence suggests that cell cycle regulation may be one of the most convincing means for Cx43 to inhibit tumor growth (11, 12, 13, 14, 15). We and others have shown that increased expression of p27 was responsible for Cx43 to inhibit the growth of human osteosarcoma U2OS cells (14), mouse lung carcinoma E9, and neoplastic rat liver WB-aB1 cells (13). Here, we showed that the elevated level of p27 was mainly attributable to the reduced expression of Skp2. Furthermore, effects of Cx43 on Skp2 and p27 were observable in several human tumor cell lines. They indicate that the regulation of Skp2 and p27 may be one of the routes for Cx43 to inhibit tumor growth, at least in certain cell lines. It has been well documented that p27 plays a central role as a negative regulator in the cell cycle progression (16). By controlling the level of p27, Skp2 has been recently recognized as an important regulator of the cell cycle progression, too (33, 45). Thus, the present study links Cx43 with the core cell-cycle regulators controlling cell proliferation. Because the exact molecular mechanisms by which Cx43 promoted the Skp2 ubiquitination are currently unclear, we could not exclude the possibility that Cx43 showed these effects with the aid of some intermediate entities.

In mice, p27 is haploinsufficient for tumor suppression (46). Reduced expression of p27 has been observed in a variety of human tumors. Low p27 protein levels correlate well with high-grade and decreased survival in human cancers (reviewed in Refs. 47, 48, 49). Recently, it has been demonstrated that the reduced level of p27 is, at least in part, the result of increased expression of Skp2 (35, 36, 37, 38). In this regard, increasing the level of p27 or inhibiting expression of Skp2 will suppress tumor growth. Like p27, a loss or decrease of Cx43 expression has been observed in many tumors (see Ref. 6 for details). Moreover, reduced expression level of Cx43 correlates well with the high grade of some tumors (50, 51, 52, 53). Thus, Cx43 expression was suggested to be useful as an adjunctive marker of progression for some specific tumors. Here, we show that transfection of Cx43 reduced expression of Skp2 and increased p27 levels in human osteosarcoma, ovarian carcinoma, glioblastoma, and monkey kidney cells. It seems that Cx43 shows an inverse expression pattern with Skp2. Moreover, when the normal osteoblasts became transformed, the Cx43 proteins disappeared, whereas Skp2 appeared; consequently, p27 was reduced. Thus, we suggest that Cx43 may inhibit tumor growth by controlling the level of Skp2 and, consequently, increasing the level of the tumor suppressor p27 in certain tumors. To prove this further, it would be worthwhile to examine expression of Cx43, p27, and Skp2 in human tumors, through which we could tell whether or not Cx43, like p27, also shows an inverse expression with Skp2 in a wide spectrum of human tumors.

As far as we know, this is the first report to show that a protein, thought to form cell-cell gap junctions, regulated ubiquitination. Mutation analyses show that this inhibitory effect of Cx43 on Skp2 is independent of the Cx43-modulated gap junctions. The cytoplasmic domain of Cx43 is sufficient to inhibit Skp2.³ Thus, these data reveal a novel function of the gap junction protein Cx43. A further investigation of how Cx43 promoted ubiquitination of Skp2 is needed. That will help us to further understand the tumor-suppressing role of Cx43.

We thank Michele Pagano, Mark Ewen, Junjie Chen, Masaaki Ikeda, and Woodring E. Wright for helpful discussion and suggestions; Junjie Chen and Masaaki Ikeda for critical reading of the manuscript; and Michele Pagano (New York University, New York, NY), Masaaki Ikeda (Tokyo Medical and Dental University, Tokyo, Japan), and Yu Xiong (University of North Carolina at Chapel Hill, Chapel Hill, NC) for reagents.