Downregulation of Connexin 43 Expression by High Glucose Reduces Gap Junction Activity in Microvascular Endothelial Cells

Microvascular endothelial cells derived from rat epididymal fat pads ascertained positive for von Willebrand protein by immunofluorescence microscopy were used in this study (20). Cells were grown in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum (Sigma, St. Louis, MO) and antibiotics. To determine the effect of high glucose on Cx43 expression and GJIC activity, these cells were grown in normal (5 mmol/l) or high (30 mmol/l) d-glucose medium for 9 days to confluency.

Cells exposed to normal or high-glucose medium were washed with PBS and lysed in buffer containing 10 mmol/l Tris, pH 7.5 (Sigma), 1 mmol/l EDTA, and 0.1% Triton X-100 (Sigma). Cellular protein content in the cell extract was measured by bicinchoninic acid protein assay method (Pierce, Rockford, IL). After addition of an equal volume of 2× sample buffer and denaturation at 95°C for 5 min, the extracts (containing 30 μg protein) were electrophoresed at 260 V for 2 h. Molecular weight standards (Bio-Rad, Richmond, CA) were included in separate lanes in each gel. After electrophoresis, the proteins were transferred onto nitrocellulose membrane (Bio-Rad) according to the procedure of Towbin et al. (21) using a semidry apparatus with Towbin buffer system. Western blot analysis was performed to examine the steady-state levels of Cx37, Cx40, and Cx43 expression in these cells. The proteins transferred to the nitrocellulose membrane were detected with the Immun-Star Chemiluminescent Protein Detection System (Bio-Rad) and rabbit anti-mouse Cx37, Cx40 antiserum (Alpha Diagnostic, San Antonio, TX), and mouse anti-rat Cx43 antiserum (Chemicon, Temecula, CA). Briefly, the membrane was blocked with 5% nonfat dry milk for 1 h and incubated with mouse anti-rat Cx43 antibody solution (1:500) in 0.2% nonfat milk for 1 h. The blot was washed with Tris-buffered saline containing 0.1% Tween-20 and then incubated with antibody solution containing goat anti-mouse IgG antibody conjugated with alkaline phosphatase enzyme (Sigma) for 1 h. The membrane was washed as above, applied to the Immun-Star chemiluminescent substrate, and exposed to X-ray film (Fuji, Tokyo, Japan). The amounts of protein loaded in the gel lanes were confirmed through duplicate gels that were prepared simultaneously. A similar procedure was followed for detection of Cx37 and Cx40. For treatment of the samples with alkaline phosphatase (AP) to assess the phosphorylation of Cx43, the samples were, in part, preincubated with AP (10 U/ml calf intestine phosphatase; Boehringer Mannheim, Indianapolis, IN) for 20 min at 30°C and then subjected to Western blot analysis, as described above. Densitometric analysis of the luminescent signal was performed at nonsaturating exposures with a laser scanning densitometer.

To study the distribution pattern and relative amounts of Cx43, immunofluorescence staining for Cx43 was performed on RME cells grown in normal or high-glucose medium. Briefly, cells grown to confluency were fixed in ice-cold methanol for 15 min, washed in PBS, and treated with 2% BSA for 15 min to block nonspecific antibody binding. The cells were then incubated overnight at 4°C in a moist chamber with a monoclonal mouse anti-rat Cx43 antibody (Chemicon) diluted 1:100 in PBS containing 2% BSA. After three PBS washes, the cells were incubated for 1 h with fluorescein isothiocyanate-conjugated goat anti-mouse IgG (Sigma) diluted 1:100 in PBS containing 2% BSA. After three PBS washes, coverslips were mounted in Slow-Fade (Molecular Probes, Eugene, OR). Negative control samples were processed in exactly the same way as those in the experimental groups except that the primary antibody was omitted. The cells were viewed and photographed using a confocal microscope (LSM510; Zeiss, Göttingen, Germany) equipped with LSM510 software (version 2.01; Zeiss). The punctate fluorescence counts were assessed at the site of contact between adjacent cells within a different and random area of 27 × 27 μm. Counts obtained from cells hybridized without primary antibody were used as background and subtracted from counts obtained from cells grown in normal or high-glucose medium.

In each experiment, RNA from cells grown in normal medium was processed in parallel with RNA from cells grown in high-glucose medium. RT reactions were performed in 20 μl volume, with 1 μg RNA, 200 units of reverse transcriptase, 2.5 μmol/l random hexamers, 1 mmol/l of each dNTPs, 5 mmol/l MgCl₂, PCR buffer, and RNase inhibitor for 10 min at room temperature followed by 40 min at 42°C. At the end of RT, samples were heated to 95°C for 5 min, cooled on ice, and treated with RNase H (1 unit) for 15 min at 37°C. PCR was designed to measure the level of Cx43 expression relative to the expression of an endogenous internal standard gene, β-actin. To prevent quantitative inaccuracies deriving from competitive effects and different efficiency and ranges of amplification of the two cDNAs, the Cx43 and β-actin cDNAs generated in the same RT reaction were amplified in separate tubes containing increasing volumes of the RT reaction (1, 2, and 4 μl) to document amplification in the linear region for each cDNA. The primers used to amplify Cx43 (5′-GAATCCTGCTCCTGG-3′ and 5′-GATGCTGATGATGTAG-3′) and β-actin (5′-ATGGATGACGATATCGCT-3′ and 5′-ATCACAATGCCAGTGGTA-3′) were designed from the European Molecular Biology Laboratory (EMBL) sequences (accession no. NM 012567 for Cx43 and J00691 for β-actin). The specificity of the PCR was enhanced by using the “hot start” approach (22). The PCR containing the appropriate aliquot of RT material with primers, 0.2 μmol/l each, 2.5 U AmpliTaq DNA polymerase (Boehringer Mannheim), MgCl₂ (1.8 mmol/l), and PCR buffer in 50 μl volume was performed in a DNA thermal cycler (Hybaid, Middlesex, U.K.) using the following cycle conditions: denaturation for 1 min at 95°C, annealing for 1 min at 54°C, and extension for 2 min at 72°C. PCR was performed with 25 cycles for Cx43 and 27 cycles for β-actin.

PCR products were always resolved on the same gel (1.0% agarose) containing 0.05 μl/ml GelStar (BMA, Rockland, ME) together with molecular weight markers (100-bp DNA ladder; Gibco, Grand Island, NY). The gel was photographed with Positive/Negative Instant Film (665; Polaroid, Cambridge, MA). Signal intensity was quantitated at nonsaturating exposure of the Polaroid film with a Zeineh soft laser scanning densitometer (Biomed Instruments, Chicago, IL). The densitometric values of the PCR products generated from increasing volumes of RT reaction were averaged to yield the Cx43 and β-actin signals for each sample (densitometric units per microliter of RT reaction).

Total RNA was extracted from rat embryo cells grown in normal or high-glucose medium using RNAgents Total RNA Isolation System (Promega) following the manufacturer’s instructions. Electrophoresis was performed with 10 μg RNA samples in 1% denaturing agarose gel containing 2.2 mol/l formaldehyde. The RNA was transferred onto a nylon membrane (Micron Separations, Westboro, MA) and hybridized to a biotin-labeled rat Cx43 cDNA and to a β-actin RNA probe (Kirkegaard & Perry Laboratories [KPL], Gaithersburg, MD). The biotin-labeled Cx43 probe was generated with Detector Random Primer DNA Biotinlyation kit (KPL) using biotin-N₄-dCTP. Prehybridization and hybridization were carried out at 42°C for Cx43 probe and at 65°C for β-actin probe. After hybridization, the membranes were washed and subjected to AP-Streptavidin and CDP-Star detection system using RNA Detector Northern Blotting Kit (KPL). Briefly, this chemiluminescent detection system relies on the use of AP, a reporter enzyme that catalyzes CDP-Star, a 1,2-dioxetane chemiluminescent substrate, during the last step of the detection system to produce luminescent signals. The chemiluminescent signal was then detected by exposing to Fuji X-ray film. Densitometry was performed at nonsaturating exposures. Statistical analysis was performed on the Cx43/β-actin ratios and recorded for the normal and high-glucose groups.

We used the scrape-loading dye transfer (SLDT) technique (23,24) to assess GJIC activity. Briefly, cells were grown in normal or high-glucose medium to confluency and rinsed three times with PBS containing 0.01% Ca²⁺ and 0.01% Mg²⁺ (Ca²⁺,Mg²⁺-PBS). An aliquot of 1.5 ml PBS containing 0.05% Lucifer yellow CH (Molecular Probes) was added to the medium, and several scrapes (cuts) were made on the monolayer using a surgical scalpel. The cells were incubated for 3 min at room temperature in the dye solution and then rinsed three times with Ca²⁺,Mg²⁺-PBS to remove any background fluorescence. The cells were then fixed with 1 ml of 4% formalin and photographed using a confocal microscope (LSM510). The dye-coupled cells on either side of the scrape line were counted in random areas to evaluate the GJIC activity in RME cells grown in normal or high-glucose medium.

Data are expressed as means ± SD or SE. Comparisons between groups were performed with Student’s t test. A level of P < 0.05 was considered statistically significant.

Rat microvascular endothelial cells grown in high-glucose medium for 9 days exhibited reduced Cx43 expression. Cx43 protein level was significantly reduced (55.6 ± 16% of control; P = 0.003, n = 5); however, there was no effect on the expression of Cx37 or Cx40 protein levels (92 ± 20% and 102 ± 11% of control, respectively) in cells grown in high-glucose medium (Fig. 1). Western blot analyses and AP treatment revealed that Cx43 protein existed in multiphosphorylated forms, consisting of a nonphosphorylated form (P0) and two phosphorylated forms (P1 and P2) ranging in size between 43 and 45 kDa in cultured RME cells (Fig. 2A). AP treatment resulted in an increase in the density of the P0 form at the expense of the other two forms, P1 and P2 (Fig. 2A). In cells grown in high-glucose medium, all three forms showed a significant decrease in the intensity of the bands: P0, 73 ± 15% of control (P = 0.04); P1, 57 ± 16% of control (P = 0.008); and P2, 42 ± 22% of control (P = 0.006) (Figs. 2A and B). Overall, Cx43 (P0 + P1 + P2) protein level was reduced by the high-glucose conditions (55.6 ± 16% of control; P = 0.003, n = 5) (Figs. 2A and B).

The distribution and relative quantity of Cx43 expression in RME cells were determined using immunofluorescence microscopy. Cx43 localization could be ascertained from punctate “dot-like” (plaques) pattern at the sites of contact between adjacent cells (Figs. 3A and B). The intensity of Cx43 immunofluorescence was less in cells grown in high-glucose medium compared with cells grown in normal medium, and the number of plaques was less in cells grown in high-glucose medium (Figs. 3A and B). A semiquantitative analysis of Cx43 at cell-cell contacts, based on counts from plaques within defined areas, yielded a score of 54 ± 5 for cells grown in normal medium and 34 ± 3 for cells grown in high-glucose medium, indicating reduced Cx43 junctional protein at the cell membrane of cells grown in high-glucose medium (63% of control; P < 0.01, n = 5) (Fig. 3C).

Cx43 mRNA levels determined by RT-PCR or Northern blot analysis in RME cells grown in high-glucose medium indicated significant downregulation compared with cells grown in normal medium (Fig. 4). Figure 4A shows Cx43 and β-actin RT-PCR products after gel electrophoresis in which high-glucose conditions reduced Cx43 mRNA levels (68 ± 13% of control; P = 0.019, n = 5). The β-actin mRNA level used as an internal control showed no change (98% of control). The band intensity of the RT-PCR products increased linearly with the increasing amount of RT reaction (1, 2, and 4 μl) for both Cx43 and β-actin, indicating that comparisons between normal and high-glucose products were made within the exponential phase of the reaction (Figs. 4A and B). Figure 4B represents Cx43 mRNA levels generated from all volumes of RT reactions. Figure 4C shows relative levels of Cx43 and β-actin mRNA on a Northern blot. Cx43 levels were clearly reduced in cells grown in high-glucose medium compared with normal medium (58 ± 12% of control; P = 0.01), whereas the β-actin mRNA level was unchanged (Figs. 4C and D).

To understand the association between reduced Cx43 expression and GJIC activity, we assessed the ability of RME cells to transfer Lucifer yellow through gap junctions using the SLDT technique in corresponding cultures of cells with reduced Cx43 expression. The total number of dye-coupled cells on either side of the scrape line was clearly less in high-glucose medium compared with normal medium (Figs. 5A and B). When the number of dye-coupled cell layers were counted from the edge of the scrape, reduced dye transfer was evident; only three to four layers in high-glucose medium were positive compared with seven in normal medium (3.9 ± 0.6 vs. 6.5 ± 1.0; P < 0.01, n = 5), i.e., 60% of control (Fig. 5A, B, and C). The reduced GJIC activity also showed a striking correlation with the downregulation of Cx43 expression in high-glucose cells (r = 0.9).

To evaluate whether high glucose levels affected endothelial connexin expression and GJIC activity, we assessed mRNA and protein levels and, as a functional assay, examined dye transfer after scrape load in RME cells exposed to high-glucose medium. Of the three endothelial-specific connexins we studied, we found the expression of only Cx43 to be altered. Whereas Cx37 and Cx40 expression showed no change in RME cells grown in high-glucose medium, Cx43 expression was significantly reduced at both the mRNA and protein levels. In addition, immunofluorescence microscopy showed reduced numbers of Cx43 gap junctions at the site of contact between adjacent cells grown in high-glucose medium.

Alkaline phosphatase treatment resulted in almost complete removal of the P1 and P2 bands, with corresponding increase in intensity of the P0 band, suggesting that the P1 and P2 bands are the two forms of phosphorylated connexin, and that the P0 band is the unphosphorylated form of connexin protein. High-glucose conditions in our experiments showed overall reduction in expression of all three forms of Cx43. Consistent with these results, Granot and Dekel (25), using rat ovarian cells, also reported three forms of Cx43 with molecular weights between 43 and 45 kDa. When we examined RME cells grown in high-glucose medium with the SLDT assay in which Lucifer yellow transfer was measured after scrape-load, we found a significant reduction in GJIC activity compared with cells grown in normal medium. This reduced GJIC activity showed a strong correlation with the reduced Cx43 expression in high-glucose cells (r = 0.9). Similar to our results, Kuroki et al. (15) observed inhibition of GJIC activity in cultured vascular smooth muscle cells grown in high-glucose medium using the fluorescent dye transfer method. Taken together, these results indicate that Cx43 exists in multiphosphorylated forms and that the expression of both the nonphosphorylated and phosphorylated forms are downregulated in RME cells, with corresponding reduction in GJIC activity, under high glucose concentration.

In this study, we observed that high glucose impairs not only connexin expression in microvascular endothelial cells but also its functional activity. A link between reduced GJIC activity and reduced connexin gene expression has been previously reported in rat ovarian cells exposed to luteinizing hormone, which resulted in decreased Cx43 mRNA levels by 45%, with subsequent reduction in GJIC activity (25). Studies have shown Cx43 to interact with the NH₂-terminal domain of ZO-1, a component of the tight junction in endothelial cells (26). Because ZO-1 expression is reduced under high-glucose conditions (27), it is possible that the effect of high glucose on GJIC activity may be mediated via other junction protein components. Our findings indicate that reduced GJIC activity in RME cells exposed to high glucose is at least in part due to inhibition of Cx43 gene expression.

Reduced GJIC could contribute to diabetic vasculopathy by a number of mechanisms. Inhibition of GJIC activity induced by high glucose concentrations may impair diffusible transport of small molecules such as calcium ions necessary for cell proliferation and maintenance of cellular homeostasis. GJIC may also play a role in cell proliferation, cell differentiation, and apoptosis (28), a process known to occur in cultured endothelial cells and pericytes (29) and the loss of retinal capillary endothelial cells and pericytes in diabetic retinopathy (30–32). Reduced GJIC in retinal endothelial cells could affect other retinal cells: retinal astrocytes, Muller cells, and pericytes. These cells, together with the endothelial cells, maintain homeostasis of the BRB. Studies have shown that the surface retinal vessels of all calibers are in contact with the processes of astrocytes (33), and extensive gap junctions composed primarily of Cx43 (34) characterize these astrocytes. Dye transfer assay has revealed that rat retinal astrocytes are coupled to each other and to Muller cells (35). Because Cx43 reactivity has been detected at the interface of astrocytes and endothelial cells in co-cultures (36), reduced Cx43 expression by high glucose may lead to reduced GJIC between astrocytes and endothelial cells and between astrocytes and Muller cells. Because pericytes and endothelial cells communicate via gap junctions (37), express Cx43 mRNA (4), and have a high frequency of junctional contact between endothelial cells and pericytes (4), it is likely that high glucose-induced reduction of Cx43 expression could impair GJIC and contribute to retinal capillary cell death.

The cause of BRB breakdown in diabetes remains controversial, with some investigators reporting that BRB breakdown is mediated by altered membrane permeability and increased vesicle formation without tight junction alteration (38,39) and others that junction protein expression is altered in diabetes (40). Findings from this study indicate that high glucose levels reduce GJIC activity in microvascular endothelial cells by inhibiting Cx43 gene expression. Further studies are necessary to clarify the role of altered GJIC activity in the pathogenesis of diabetic retinopathy.

Work on connexin gene expression was supported in part by grants from the American Diabetes Association, Chartrand Foundation, the Macula Foundation, MA Lions Eye Research Foundation, the National Institutes of Health, and Research to Prevent Blindness.