Connexin Expression and Cell Coupling Fail to Reverse the v-src Transformed Growth Characteristics of a Cx43–/– Cell

Abstract

Gap junctions, composed of connexins, have been shown to suppress transformation in a variety of malignancies and transformed cell types. In addition, transforming factors such as the src oncogene have been shown to directly phosphorylate some connexins (e.g., Cx43) and inhibit coupling. To investigate the role of gap junctions in cell transformsation by v-src, we utilized a clonal cell line derived from Cx43 knockout mice (KoA) that was immortalized, but not transformed. Transfection by v-src induced a marked transformed phenotype characterized by growth in low serum and anchorage-independent conditions. Subsequent transfections by Cx43, Cx32 or vector alone were then tested for their effects on growth. Activity of pp60^{v - src} was confirmed in all transfectants as well as the ability of pp60^{v - src} to phosphorylate Cx43 in several clones. Despite the documented effect of pp60^{v - src} on Cx43 channel closure, modest coupling was still retained in many of the Cx43 and Cx32 transfectants. However, none of the four Cx43 transfected clones showed significant inhibitory effects on proliferation in either anchorage-independent or low serum growth conditions. Of the Cx32 clones, only one in five showed effects on growth in both assays, which was the same ratio observed for the control transfectants. Thus, based on the levels of expression achieved, which were comparable to endogenous levels in established cell lines, neither Cx43 nor Cx32 serve as effective suppressors of the transformed growth phenotype of this v-src expressing cell line.

Keywords:

• Connexins
• gap junctions
• transformation
• v-src

Introduction

Gap junctions are the only direct mediators of the exchange of low molecular weight metabolites between cells of metazoans. As such, they have been implicated in the maintenance of homeostasis, the coordination of tissue responses, and as negative regulators of cell growth (36). The linkage of gap junctions to growth suppression of tumors and transformed cells dates back to 1966 with the demonstration that liver cancer cells showed inhibition of electrical coupling as compared to normal hepatocytes (22). Since then, numerous investigators have documented a reduction of connexins in various tumor types, as well as in many transformed and virally infected cell lines (reviewed in (25, 29). Tumor promoters such as phorbol esters, pesticicides, and phenobarbitol have also been shown to decrease connexin expression in some systems (25), as have several oncogenes including SV-40, middle T antigen, v-ras, v-mos, v-fps, and v-src (15, 36). Conversely, some antineoplastic agents such as retinoids, vitamin D, and carotenoids were shown to increase gap junction intercellular communication (GJIC) (35). Even more directly, when transformed cell lines were transfected with connexins, suppression of several transformed growth characteristics was evident (7, 12, 24, 26, 39).

A specific role for Cx43 as a tumor suppressor has been proposed in numerous cases due to its reduced expression in many tumor types, including those derived from breast, ovary, prostate, and colon (9, 14, 18, 21) and in various tumor cell lines derived from liver, ovary, lung, and brain, as well as chemically and virally transformed epithelial cells (4, 14). Furthermore, transfection of Cx43 has reduced the transformed phenotype in lung carcinoma cells, glioma cells, rhabdomyosarcoma cells, mammary carcinoma cells, and kidney epithelial cells (reviewed in (32)). Transfection with other connexins has also shown suppression of transformation in a tissue-specific manner, such as Cx26 in cervically derived HeLa cells (26) and Cx32 in liver tumor cells (7). Consistent with the latter finding, Cx32 knockout mice show an increase in hepatic tumors (28).

In many of the above studies, correlations were established between coupling levels and growth suppression, although, controversy has recently arisen over whether intact channel function is necessary for the growth suppressive effect of connexins in all cell systems. In support of the intercellular channel function of gap junction growth regulation, rat liver epithelial cells lose their ability to communicate with nontumor cells as their transformation progresses (27). Inhibition of GJIC through expression of dominant negative constructs also correlated with increases in the tumorigenicity of rat bladder carcinoma cells, even though the protein is still present (2, 17). However, in contrast to these studies, it has been claimed that Cx43 and Cx26 inhibit tumor growth of breast cancer cells in nude mice by a mechanism that is independent of GJIC, at least as estimated by dye transfer between cells (2002). In fact, transfection of the C-terminal tail of Cx43 alone appeared to result in localization to the nucleus, and a significant decrease in the growth of HeLa cells (6). This observation suggests that signaling roles associated with cytoplasmic domains of connexins may have effects independent of the channel function of these proteins. However, there is no evidence, outside of eye lens tissue, to indicate that these domains are cleaved in situ in any programmed way that would be needed to allow for translocaion to the nucleus.

Despite the extensive literature documenting a role for connexins in growth suppression, the mechanism for this action has yet to be elucidated, largely because the cause of transformation in most tumor cells remains unknown. We sought to circumvent this problem by utilizing cells transformed with the v-src oncogene, as it activates a well-defined signal transduction pathway that leads to cell transformation. Thus, any transformation suppressive effects of connexins could be interpreted in the context of known mitogenic pathways. Another rationale for studying connexins in the context of v-src transformation is that expression of v-src has been shown to rapidly uncouple cells (3, 5), suggesting that interruption of intercellular coupling might be necessary for src transformation. pp60^{v - Src}-induced closure of gap junctions composed of Cx43 has been shown to occur via a “ball and chain” mechanism where the C-terminal domain of Cx43 occludes the pore (38). In Xenopus oocytes coinjected with Cx43 and v-src (34), and in v-src infected mammalian cells subsequently transfected with Cx43 (20), direct tyrosine phosphorylation by pp60^{v - src} was found to be necessary for channel closure. However, in acute gating by pp60^{v - src} in Xenopus oocytes already expressing Cx43 channels, and in temperature-sensitive LA25 cells in which pp60^{v - src} function was induced by a change to the permissive temperature, downstream effectors of pp60^{v - src} signaling such as MAPK were shown to be necessary for channel closure (38).

As a starting point, we utilized a KoA cell line, derived from the brain of embryonic Cx43 knockout mice, which are immortalized, but not transformed, and show minimal endogenous coupling (Goldberg, GS, personal comm.). Following selection of a v-src transformed clone of these cells, the effect of transfection with Cx43 and Cx32 was investigated. The latter was chosen as it lacks phosphorylation sites for pp60^{v - src} and MAPK and is, therefore, not prone to channel closure by pp60^{v - src} (34). We demonstrated that pp60^{v - src} was active in all of the transfectants, resulting in tyrosine phosphorylation of Cx43 in several of the Cx43 transfected clones. Despite this effect, transfectants of both Cx43 and Cx32 showed modest levels of intercellular coupling in the presence of v-src. However, no consistent effects on cell growth were observed for Cx43 clones. Modest decreases were found in most of the Cx32 clones in terms of anchorage independent growth, but these did not correlate with protein or coupling levels and were less than the effects seen in one of the control transfectants.

MATERIALS AND METHODS

Cells and Transfections

Cells derived from the forebrain, midbrain, and hindbrain of Cx43 knockout mice (31) at embryonic stage 18 (KoA) were utilized for this study (Goldberg, GS, personal comm.) These cells were immortalized by passing crisis, but did not express characteristics of transformed cells. The cells were transfected with the v-srconcogene in a pBABE vector and initially maintained as a pool. Clones were isolated from this pool by drug selection with 1 μ g/ml puromycin (Sigma, St. Louis, MO) and tested for transformation by growth in DMEM (Sigma, St. Louis, MO) either in the presence of 1% fetal bovine serum (Invitrogen, Carlsbad, CA) or under anchorage-independent conditions (see below). On the basis of its robust transformed phenotype in both assays, the KoAsrc#15 clone was selected for further transfection with full length Cx43 or Cx32 in a pIREShygro vector (Clontech, Palo Alto, CA), or by vector alone as a negative control. Transfection was achieved using Lipofectamine plus (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Clones were selected with 400 μ g/ml hygromycin (Sigma, St. Louis, MO) and maintained in 200 μ g/ml hygromycin.

Cell Growth Rates Under Control and Reduced Serum Conditions

Different transformed and nontransformed KoA clones were seeded at 1× 10⁴ cells/well in 12-well plates (Becton Dickinson, Franklin Lakes, NJ) with DMEM culture medium containing 10% fetal bovine serum (FBS). For assays in reduced serum, the medium was removed after 24 h, cells were rinsed with phosphate buffered saline (PBS), and DMEM with 1% FBS was added. Cells were seeded in triplicate in all experiments, and data shown in Table 1 are drawn from one or two independent experiments. At each time point (approximately every 24 hours), the triplicates were treated with 0.25% trypsin (Invitrogen, Carlsbad, CA) and collected for counting in a Beckman Coulter Counter (Beckman, Fullerton, CA). The reciprocals of the slopes in plots of log₂ (cell number) vs. time were used to calculate doubling time.

Table 1 Summary of clonal analysis of KoAsrc transfectants

Cell clone	Relative connexin protein level	% Cells coupled	% of 1st order cells coupled	% with 2nd order coupling	Doubling time in low serum (% of pIRES#5)	Anchorage independent growth at day 6–7 (% of pIRES#5)
KoA	Undetectable	0	—	—	200	5.0
KoAsrc15 pIRES#5	Undetectable	0	—	—	100	100
KoAsrc15 pIRES#1	Undetectable	0	—	—	101.72	94.7 (0)
KoAsrc15 pIRES#3	Undetectable	0	—	—	88.79	126.1 (+)
KoAsrc15 pIRES#10	Undetectable	0	—	—	134.48 (-)	11.2 (-)
KoAsrc15 Cx43#2	Intermediate	51.9	63.2	33.3	81.1 (0)	97.0 (0)
KoAsrc15 Cx43#6	Low	42.9	52.5	33.3	110.98 (0)	60.8 (-)
KoAsrc15 Cx43#27	Intermediate	0	—	—	64.02 (+)	102 (0)
KoAsrc15 Cx43#29	High	29.3	33.6	22.2	72.49 (+)	127.7 (+)
KoAsrc15 Cx32#5	Low	0	—	—	88.87 (0)	89.8 (0)
KoAsrc15 Cx32#18	High	56.6	84.8	16.7	103.85 (0)	62.7 (-)
KoAsrc15 Cx32#19	Intermediate	5.0	75.0	0	98.36 (0)	42.2 (-)
KoAsrc15 Cx32#20	Low	0	—	—	124.28 (-)	23.8 (-)
KoAsrc15 Cx32#26	Intermediate	0	—	—	103.40 (0)	28.8 (-)

*In addition to the calculated values, (—) indicates strong reduction, (−) modest reduction, (0) no significant change; (+) increase in comparison to the KoAsrc15 pIRES#5 control clone.

Anchorage-Independent Cell Growth Assay

Cells were seeded at a concentration of 1× 10⁴ (Figures 5B, 6B) or 2 × 10⁴ (Figures 1C, 4B) per well on 24-well plates (Becton Dickinson, Franklin Lakes, NJ) in DMEM with 10% FBS. The plates were precoated with 10 mg/ml Poly-HEMA (Sigma, St. Louis, MO) which prevents cell adhesion (10). Every three days, cells were pipetted into microfuge tubes, centrifuged at 5,000 rpm, rinsed with PBS, treated with 0.25% trypsin and counted using the Beckman Coulter Counter. Comparisons of cell density were determined at the estimated growth peak of the experiment (usually at day 6–7).

Figure 1 Assays to compare growth properties for clones of KoAsrc. From a pool of v-src transfected KoA cells, clones were drug-selected and tested for growth in DMEM with 10% FBS (A), or 1% FBS (B). Anchorage-independent growth was tested in normal (10%) serum on Poly-HEMA coated plates (C). All assays were performed in triplicate (data are shown as mean +/− standard deviation). Growth of individual clones was compared to that of the non transformed KoA cells and the original pool of v-src infected cells. Significant variation in growth characteristics among the clones was evident, suggesting a great degree of heterogeneity in the pool. Hence, clone 15, showing the most consistent transformed behavior in all assays, was selected for further study.

Figure 4 Testing for the presence of Gap Junctional Intercellular Communication. Functional expression of gap junctions was tested by a preloading assay, where transfer of the gap junction permeable dye (calcein) from donor cells labeled with DiI, to the neighboring recipient cells was measured. Fluorescent images are shown for DiI (A) and Calcein (B), with corresponding phase contrast images (C). KoAsrc15 pIRES clone 5 serves as a negative control (top panels). Arrows indicate cells that were preloaded in the Calcein and phase contrast images. Examples of the better coupled Cx43 (clone 2-middle panels) and Cx32 (clone 18-bottom panels) transfected clones are shown.

Figure 5 Effects of Cx43 on KoAsrc growth suppression. Several Cx43 clones, representing a range of protein expression and coupling levels, were assayed for growth suppressive properties in comparison to KoA parental cells and a vector control (KoAsrc15 pIRES, clone 5), of which three are shown here (clones 2, 6, and 27). Growth was tested in low serum (1% FBS) (A) or on an anchorage-independent substrate (B). The controls are compared to KoAsrc15 Cx43 clone 29 in panels (C) growth in low serum and (D) on anchorage-independent substrate. Growth responses varied, but only one clone (clone 6) showed some degree of reduced growth in both assays compared to the vector control (see Table 1)

Figure 6 Effects of Cx32 on KoAsrc growth suppression. Five Cx32 clones, also representing a wide range of protein expression and coupling levels (clones 5, 18, 19, 20, and 26), were assayed as in Figure 5 for growth in low serum (1% FBS) (A) or on anchorage-independent substrate (B). Other than clone 20, there was little difference between the Cx32 and control transfectant in low serum growth (A). However, in anchorage-independent growth most clones (with the exception of clone 5) showed delayed or reduced growth (B), although this did not correlate with levels of Cx32 expression or coupling (Table 1).

Western Blots

Total cell lysates were prepared by adding 1 ml RIPA buffer (PBS, 1%NP40, 0.5% sodium deoxycholate, 0.1% SDS, 10 mg/ml PMSF, 10 μg/ml Na₃VO₄, aprotinin, leupeptin, pepstatin) to 100 mm dishes and scraping with rubber policemen. Lysates were passed through a needle 18 times to shear the DNA and spun at 15,000 rpm for 20 min. Protein estimations were performed on the supernatant using the BioRad DC protein assay kit (BioRad, Hercules, CA) with BSA as a standard (Sigma, St. Louis, MO). 5 μg of protein per sample was separated on a 12% SDS gel and electroblotted onto Immobilon-P membranes (Millipore, Bedford, MA). The blots were probed with monoclonal anti-Cx43 antibody at 1:1000 (BD Biosciences, San Jose, CA), monoclonal anti-Cx32 antibody at 1:500 (Zymed, San Francisco, CA), monoclonal anti-actin antibody at 1:2000 (Sigma), and polyclonal anti-Phospho-Src antibody (Tyr416) at 1:1000 (Cell Signaling, Beverly, MA). Blocking and antibody incubations were performed in 1% BSA in Tris Buffered Saline (TBS) with 0.1% Tween (Sigma). Washes were performed in TBS with 0.1% Tween. Antibody binding by a secondary antibody conjugated to HRP (Cell Signaling) was detected in conjunction with ECL Plus (Amersham Biosciences, Piscataway, NJ) according to the manufacturer's instructions. For the purpose of quantification, the blots were scanned in the Storm phosphorimager (Amersham Biosciences), after ECL Plus treatment. The antibody-bound protein of interest was quantitated using Image Quant software (Amersham Biosciences) according to manufacturer's instructions. Blots were stripped in stripping buffer (100 mM 2-mercaptoethanol, 2% w/v SDS, 62.4 mM Tris-HCl, pH 6.7) for 30 min at 50°C with mild agitation. Actin levels were quantitated as above and served as a loading control.

Immunoprecipitation

Cells were lysed in RIPA buffer (described above) and 100 μg of each lysate was incubated overnight at 4°C with 10 μg polyclonal anti-Cx43 antibody-agarose conjugate (Santa Cruz Biotechnology, Santa Cruz, CA). Following centrifugation, the pellet was rinsed with RIPA buffer, boiled in sample buffer and loaded onto an SDS gel. Electrophoresis and blotting were performed as described above. Membranes were hybridized to a monoclonal anti-Cx43 antibody at 1:1000 (as above), stripped, and reprobed with either a phospho-tyrosine monoclonal antibody at 1:2000 (Cell Signaling), or an anti-phosphoCx43 antibody at serines 279 and 282 at 1:200 (Santa Cruz Biotechnology).

Dye Transfer Assay to Measure Cell Coupling

Coupling of connexin transfected cells was assayed using the preloading method of fluorescent dye transfer (11). The donor cells were incubated with two fluorescent dyes; calcein AM (5 mM) and DiI (10 mM) (Molecular Probes, Eugene, OR), for 20 min in isotonic glucose. These preloaded donor cells were then plated with unlabeled cells at a ratio of 1:72 and allowed to settle for 5–6 hours. This time point was optimized using a range of 4–8 hours. Several coupling parameters were measured: the percentage of preloaded cells passing dye to at least one receiver cell; the percentage of first-order cells receiving dye (i.e., those immediately adjacent to the DiI labeled cell), and; the percentage of coupled cells that showed second-order coupling. The latter included cells that passed dye to third or fourth order recipients. Typically, two to three donor cells were visualized per field using a Zeiss Axiophot 10 photomicroscope, equipped with UV epifluorescence to measure green (Calcein) and red (DiI) fluorescence. Calcein and DiI have excitation wavelength peaks of 494 nm and 549 nm, respectively, and emission wavelength peaks of 517 nm and 565 nm, respectively. Data was recorded on a micromax CCD camera (Princeton Instruments, Trenton, NJ) and stored digitally using Metamorph software (Universal Imaging Corp., West Chester, PA). The data from two to four experiments, whereby approximately 100 cells were counted per clone are averaged in Table 1.

RESULTS

Transformation of KoA Cells by v-src and Clonal Selection

Assays were performed to compare the growth rates of various KoA cells transfected with v-src. The goal was to identify a clone with a relatively high expression of pp60^{v - src} and a consistent transformed phenotype. Parental KoA cells showed robust exponential growth in 10% serum after a lag phase of approximately 50 h, but showed minimal growth in 1% serum (Figure 1A and B). All pp60^{v - src} clones tested showed a similar lag phase in 10% serum, but an enhanced rate of exponential growth, with clone #15 showing the most rapid growth. Saturation density was difficult to measure for the pp60^{v - src} clones since, at confluence, the cells became rounded up and lost adherence to the plate, in contrast to the KoA parental cells that packed very tightly and remained adherent as a monolayer.

In reduced serum media, the KoAsrc pool showed a slight increase in growth compared to the untransformed parental line. Comparison of individual clones derived from the pool showed significant variability, with three of five growing slower than parental KoA cells. Clone #14 had a high slope in the initial log phase of growth, but then failed to continue growing after 100 h. Only clone #15 showed both robust exponential growth in low serum (approximately 20% of the growth rate observed in 10% FBS) and also reached saturation in low serum (at slightly less than half the density seen in 10% serum). As the difference of growth rates between parental KoA cell line and KoAsrc#15 is higher in low serum, as compared to normal serum, low serum assay was used for further studies.

The ability of the cells to grow in the absence of substrate anchorage was assessed by plating the cells on dishes coated with polyHEMA (10). In this assay, the parental KoA cells showed no cell growth, whereas the KoAsrc pool showed significant growth that peaked around day 8 and decreased by day 12 (Figure 1C). As seen in the growth in 1% serum, some KoAsrc clones grew at slower rates than the pool and some faster, but in this assay, all clones grew significantly more than the parental KoA cells. In general, the faster clones grew initially, the earlier they reached peak numbers before decreasing. As was the case in low serum, clone #15 showed the fastest initial growth, the highest peak counts. Based on these assays, KoAsrc#15 was selected as the clone with the most robust transformed phenotype, and was used for subsequent connexin transfections.

Transfected Clones Show Varying Connexin Levels

KoAsrc15 cells were transfected with either Cx43 or Cx32 in the pIRES vector, or with the vector alone as a control. A total of 42 clones were obtained from the Cx43 transfection. The clones varied greatly in their Cx43 expression levels with most of the clones expressing levels comparable to the endogenous levels of Cx43 in NRK cells (19). Clones were classified as low (18 clones: e.g., clone 6), intermediate (16 clones: e.g., clones 2 and 27), or higher expressors (8 clones: e.g., clone 29) (Figure 2A, Table 1).

Figure 2 Connexin transfected clones show a range of expression levels. Western blots of cell lysates from confluent cultures show connexin expression levels for four representative Cx43 transfectants (from a total of 42) (A) and five representative Cx32 transfectants (from a total of nine) (B). Blots were stripped and reprobed with actin as a control for loading. Cx43 transfected C6 glioma cells (C643) and Cx32 transfected HeLa cells (Hela 32H) were included as positive controls. Significant variation in the expression levels of both connexins was evident in the different clones.

For Cx32 transfections, nine clones were obtained. These varied from low (6 clones: e.g., clones 5 and 20) to intermediate (clones 19 and 20) to higher expressors (clone 18) (Figure 2B, Table 1). Several of the clones obtained from Cx32 and Cx43 transfection were selected for further analysis to investigate the effects of a range of protein expression levels on transformed growth.

pp60^{v −src} is Functional in All Transfectants and Cx43 is Tyrosine and Serine Phosphorylated in Multiple Cx43 Clones

Assessment of pp60^{v - src} expression in each of the connexin transfectants was achieved by two independent mechanisms. Most directly, Western blots, using an antibody specific for pp60^{v - src} phosphophorylation at tyrosine 416 served to assess levels of functional pp60^{v - src} in all of the clones (Figure 3A). The blots were normalized to actin as a loading control. The levels of pp60^{v - src} autophosphorylation relative to KoAsrc15 ranged from 1.53 to 1.93 for pIRES clones, 1.57 to 1.79 for Cx32 clones, and 1.06 to 1.58 for Cx43 clones. It is interesting to note that all of the transfectants showed a higher pp60^{v - src} activity than the parental KoAsrc15 cells, and ensured that any decrease in transformed phenotype caused by connexin expression was not due to changes in v-src activity.

Figure 3 Functionality of pp60^{v - src} in KoAsrc transfectants. (A) Western blots show levels of pp60^{v - src} autophosphorylation at site Y416 in lysates from KoAsrc15 clones transfected with pIRES, Cx32, and Cx43. The blots were stripped and reprobed with actin for normalization. (B) Cell lysates from Cx43 transfectants were immunoprecipitated with a polyclonal Cx43 antibody and probed with a monoclonal Cx43 antibody (upper panel). The blot was stripped and reprobed with an antibody to Phosphotyrosine (middle panel), and PhosphoCx43 at serines 279 and 282 (lower panel).

In the case of Cx43 transfectants, the effectiveness of pp60^{v - src} in targeting gap junctions was assessed by probing an immunoprecipitate of Cx43 with antibodies to Cx43, phosphotyrosine, and phosphoserine on Western blots (Figure 3B). NRK cells, expressing Cx43 but untransformed by v-src, were used as a negative control, showing no detectable tyrosine or serine phosphorylation of Cx43. However, most of the KoAsrc15Cx43 transfectants show phosphorylation by both tyrosine and serine (clones 2, 6, and 27), with the exception of clone 29. Thus, while pp60^{v - src} is active in all connexin transfected clones, in clone 29 it fails to induce or maintain Cx43 phosphorylation either directly on tyrosine residues or indirectly by MAPK activation and subsequent phosphorylation of serine residues. Another Cx43 v-src transformed clone also failed to show tyrosine and serine phosphorylation (data not shown). Cx32 lacks sites for tyrosine phosphorylation and has been demonstrated to be incapable of phosphorylation by pp60^{v - src} (34). Since phosphorylation of Cx43 at Y265 has been demonstrated to cause chronic channel closure (20), and phosphorylation of serine residues 255, 279, and 282 by MAPK was shown to induce acute Cx43 channel closure (38), assays were performed to test if gap junctional coupling varied between these clones.

Coupling Does not Correlate with Connexin Expression in KoA Cells

Preloading dye transfer assays were performed to measure functional gap junction communication in the connexin transfectants. No coupling was observed for KoAsrc15 pIRES#5, the vector control clone (Figure 4A, Table 1). In contrast, several KoAsrc15 Cx43 and two KoAsrc15 Cx32 clones showed modest coupling as illustrated in Figure 4B and 4C. Coupling levels were measured by three criteria: % of cells transferring dye to at least one other cell, % of first order neighbors coupled, and the % of cells transferring beyond the first order of cells (Table 1). In general, coupling efficiency was lower than seen in many published cell lines due to the “spindle-like” morphology of the cells. Even at high cell densities, contact between cells was restricted to the intersection of fine extended processes and not via the cell bodies. For Cx43, there was a poor correlation between coupling and protein levels. One high and one intermediate expressing clone showed significant coupling (clones 29 and 2). However, clone 27, which also had intermediate protein levels, showed no coupling. Most notably, a very low expressing clone (clone 6) showed high levels of coupling. This clone, as well as clone 2, showed atypical morphology with multiple processes extending from a wide cell body (Figure 4B). Clone 6 was an exception to the overall trend in which six other low Cx43 expressing clones showed no detectable dye coupling and a more normal morphology (data not shown). Interestingly, despite readily detectable tyrosine phosphorylation of Cx43 in clones 2, 6, and 27, coupling is eliminated in the latter only.

For the Cx32 clones, a better correlation between protein levels and coupling was found, with clone 18, expressing the highest level of Cx32, also showing the greatest extent of coupling. The two intermediate expressing clones showed little or no coupling, (clones 19 and 26, respectively), and the low expressing clones tested showed no detectable dye coupling in this assay (Table 1).

Effects of Connexins on Cell Growth do not Correlate Well with Protein or Coupling Levels

Assays for growth under low serum and anchorage-independent conditions showed no dramatic decreases in growth for any of the Cx43 clones compared to the vector control (KoAsrc15 pIRES#5) (Figure 5A, Figure 5B). Only clone 6 showed a significant reduction in growth in polyHEMA, but this was not evident over several experiments in low serum growth (Table 1). While clone 6 has low protein expression, it shows some of the highest coupling levels. However, clone 2, which has a very similar phenotype of low connexin levels, high coupling, and distinct morphology, showed no decrease in growth compared to the vector control in either assay.

For the Cx32 transfected clones of src transformed KoA cells, a broad range of results was obtained. In the low serum assay (Figure 6A), only 1 clone (clone 20) showed suppression of growth that, when averaged over several experiments, did not approach the levels of the untransfromed KoA cells. In anchorage-independent conditions (Figure 6B) four of five clones (clones 18, 19, 20, and 26) showed decreased growth when averaged over several experiments (Table 1). Interestingly, greater decreases in polyHEMA growth were observed in clones with lowest levels of both Cx32 protein and intercellular coupling (clones 19, 20, and 26) than were evident in clone 18, with the highest protein and coupling levels. The exception was clone 5 that had lower levels of protein and coupling, and showed no growth inhibition in either of the assays. Overall, the data as outlined in Table 1, suggest that Cx43 transfection had no significant effect on growth suppression, whereas some Cx32 clones show a modest reduction in growth, but only in the anchorage-independent assay. However, this growth suppression does not correlate with protein or coupling levels.

This raised the issue of the degree to which clonal variation among transfectants may affect growth phenotypes. This was independently assessed by comparing several “empty” vector control clones in the same growth assays. Most of the clones showed similar or higher growth rates than the KoAsrc15 parental clone. However, KoAsrc15 pIRES 10 showed a dramatic decrease in growth by both assays (Figure 7A, 7B) that was greater than any of the connexin transfected clones analyzed in Figures 5 and 6. Table 1 summarizes the experimental results of several trials for all the clones described above and serves to emphasize the lack of overall correlation between connexin expression, function, and growth suppressive properties in this cell line.

Figure 7 Effects of pIRES vector controls on KoA growth suppression. Vector (pIRES) alone transfected control KoAsrc15 clones (clones 1, 3, 5, and 10) were assayed for effects on growth in low serum (1% FBS) (A) or on anchorage-independent substrate (B) as in Figures 5 and 6. KoA parental cells were tested as well as the KoAsrc15 clone that had been used in the connexin expression studies. While most clones showed no significant difference as compared to KoAsrc15, clone 10 was a notable exception in that it displayed markedly reduced growth in both assays that actually exceeded the growth suppression seen in any of the connexin transfectants shown in Figures 5 and 6.

DISCUSSION

The original design of this study was to develop a defined system to determine the molecular mechanism by which gap junctional coupling or connexin expression mediates tumor suppression. This first required a cell line with minimal endogenous coupling so that different connexins could be expressed exogenously. Secondly, v-src was chosen as a means of transformation since it has a defined mechanism of action that operates at an early step in the mitogenic signaling cascade (enabling the testing of many downstream potential targets for connexin mediated growth suppression); and the oncogene has been shown to target some gap junctions and mediate their closure. Thirdly, connexins needed to be efficiently expressed in the cells.

The immortalized KoA cell-line derived from 18-day embryonic Cx43 knockout mice brains met the first criterion, and were effectively transformed by v-src, meeting the second criterion. Although the levels of connexin expression varied significantly among the clones, most of the Cx43 clones showed expression comparable to endogenous expression in NRK cells, and most Cx32 clones showed expression comparable to the levels in HeLa 32 (Figure 2). It is interesting to note that some of the clones with low expression levels (KoAsrc15 Cx43 6 and KoAsrc15 Cx32 20) showed the most growth suppression among the connexin transfectants examined for this study (Table 1). However, while some clones (particularly in the case of Cx32) did show modest growth restriction, most notably in terms of anchorage dependence, this did not correlate well with protein levels or coupling and, ultimately, even greater levels of growth suppression were seen in one control transfectant, emphasizing the importance of considering clonal variation that can influence many other factors, including cell adhesion, growth factor receptors, and other factors that affect growth rates.

Although connexins were not effective as growth suppressors in this oncogenically transformed system, they have been demonstrated to suppress the transformed phenotype in a variety of tumors and tumor cell lines (4, 9, 14, 18, 21, 22, 23). However, there are other examples where connexins do not mediate growth suppression in UVC-induced tumorigenic HeLa skin fibroblast hybrid cells (8), and even cases where connexins are associated with increased tumorigenicity (1). Interestingly, the results of another recent study investigating the potential for Cx43 to reverse the transformed phenotype of v-src transformed Cx43 knockout cells, has also documented the inability of Cx43 to affect growth in either low serum or anchorage-independence conditions (37). Finally, studies have shown that specific connexins can mediate growth suppression in a specific cell type while other connexin(s) have no effect. Typically, the connexins that are endogenous to a cell type seem most effective in growth suppression (e.g., Cx43 in glioma cells, presumably derived from astrocytes (16); Cx32 in SKHep1 hepatoma cells (7); and Cx26 in HeLa cells derived from cervical tissue (26)). Since the specific nature of the cell type from which the Cx43 knockout cells used in this study were derived is unknown, we do not know which connexins were endogenous. However, the lack of coupling in these KoA cells is consistent with the observations that many common cell types in the brain express Cx43 (e.g. fibroblasts, astrocytes), and that Cx43 is often the predominant connexin expressed by cultured cells, independent of the originally expressed isotypes (33).

Cx43 was a logical choice for transfection in this study, due to its tumor suppressor role in other brain-derived cell lines (e.g., C6 glioma cells (39), and its demonstrable efficiency in transporting several metabolites (13). However, Cx43 channels can also be closed by pp60^{v - src} (34) through phosphorylation of either tyrosine or serine sites (i.e., MAP kinase targets- (38)). Hence, Cx32 was also tested in this system, as it has been shown in several studies to not be affected by pp60^{v - src} expression, since it lacks the appropriate phosphorylation sites (34).

Surprisingly, despite the reports of pp60^{v - src} mediating Cx43 channel closure and evidence of pp60^{v - src} functionality, several Cx43 expressing clones in this study demonstrated intercellular coupling with an efficiency and frequency similar to Cx32 expression. The reason for this was not immediately clear, as pp60^{v - src} expression in the clone selected here (clone 15) was strong based on the transformed phenotype of the cells. Cx43 was also not dramatically over-expressed in these cells, making it unlikely to saturate the kinase activity of pp60^{v - src}. It remains possible that the KoA cell line may have particularly active phosphatases that either prevent Cx43 phosphorylation or negate the pp60^{v - src} mediated uncoupling of Cx43, while still permitting many of the other transforming effects of pp60^{v - src}.

Nonetheless, the data presented in this report show that connexins do not suppress the transformed phenotype of v-src transformed KoA cells to an extent greater than control transfectants. While the levels of expression of exogenous connexins was not high in this study, the levels were, in several clones, comparable to the endogenous levels seen in untransfected cells and many tissues in situ. In addition, in most cases, Cx43 expression was sufficient to overcome its gating by pp60^{v - src}. Of all the Cx43 and Cx32 clones, only one clone (Cx32 clone 20) showed significant growth suppression in both the assays averaged over several trials (Table 1). The number of Cx32 clones showing growth suppression, at least in the anchorage-independence assay (4 of 5), did exceed the frequency of growth suppression seen in Cx43 (1 of 5) and control transfectants (1 of 4), suggesting a significant effect. However, the magnitude of the effect of any individual Cx32 clone was significantly less than that observed for one of the vector control clones (pIRES clone 10). Furthermore, the growth suppression seen with the Cx32 clones did not correlate with protein levels, or functional expression of Cx32. It seems most likely that, in the case of chronic v-src transformation of these cells, Cx43 was unable to suppress the v-src transformed phenotype and Cx32 only modestly affected growth in a nonuniform manner. Given that Cx43 is a direct target of pp60^{v - src}, and cell coupling is rapidly inhibited in the acute phase of transformation of a cell by pp60^{v - src}, it does seem surprising that reconstituting coupling in these cells did not have more of an effect. However, it is possible that gap junction coupling may inhibit the initiation of cell transformation, yet be ineffective in reversing the chronically v-src transformed state. This serves to emphasize that suppression of the transformed phenotype by connexins is cell-type specific, and dependent on the nature of the transformation. The current study also serves to demonstrate the degree to which clonal variation itself can produce growth changes, and the need to establish specific effects of transformation of any gene through correlations of several independent clones and their corresponding controls.

ACKNOWLEDGEMENTS

We would like to thank Gregory Chomicz for technical assistance and Gary Goldberg for providing the original, nontransformed KoA cells, the KoA v-src infected pool, and for his critical input. This work was supported by National Institute of Health grant CA48049 (BJN) and by an NRSA Fellowship from the National Cancer Institute CA90062 (SNZ).