Ultraviolet A (UVA) (320–400 nm) radiation is known to cause cutaneous aging and skin cancer. We studied the effect of UVA (365 nm) radiation on the human epidermis by focusing on keratinocyte gap junction-mediated intercellular communication (GJIC). We observed a dose-dependent 10-fold decrease in GJIC induced by UVA in normal human keratinocytes. This decrease in GJIC was associated with time-dependent internalization of connexin43 (Cx43). UVA radiation also damaged the actin cytoskeleton, as shown by microfilament disappearance. Importantly, the decrease in GJIC was transient when keratinocytes were irradiated with 10 J/cm2UVA, with a return to baseline values after 8 h. Concomitantly, Cx43 was relocalized and the actin cytoskeleton was restored. UVA irradiation and 12-O-tetradecanoylphorbol 13-acetate (TPA) treatment activated protein kinase C and reduced GJIC. However, Cx43 localization and phosphorylation were differently regulated by the two treatments. This suggests that at least two different pathways may mediate the observed fall in GJIC. These findings identify keratinocyte GJIC as a new UVA target that might sensitize human skin to photoaging and cancer formation.
- skin physiology
- actin cytoskeleton
ultraviolet (UV) radiation is an energy source for life on earth that also has detrimental effects on animal life. The UV spectrum of solar radiation that reaches the earth's surface comprises UVB (280–320 nm) and UVA (320–400 nm). UVA has deleterious effects on the human epidermis by generating reactive oxygen species that can degrade lipids, proteins, and DNA and give rise to photo-oxidized products (26). The resulting oxidative stress can damage cellular components and change the pattern of gene expression, leading to photoaging and serious skin diseases such as cancer (20).
The human epidermis is a multilayered, cohesive tissue with a unique functional architecture. It forms the primary barrier to the outside environment. Keratinocytes are the main component of the epidermis. Cells from the basal layer proliferate and then differentiate in the upper epidermal layer, where they acquire new biological properties. The symbiotic equilibrium between cell growth and differentiation in the epidermis points to an intimate spatial relationship in this tissue (31). Cell-cell communication via gap junctions appears to participate in epidermal homeostasis.
Cell-cell communication occurs in the epidermal and dermal compartments of normal fully differentiated human skin. In the epidermis, gap junction-mediated intercellular communication (GJIC) usually involves groups of about a dozen keratinocytes (25). Gap junctions are clustered at cell-cell contacts and participate in local information share between cells. Each cell contributes a hemichannel (connexon) composed of six connexin (Cx) subunits that cross the plasma membrane four times and have their NH2 and COOH termini in the cytoplasm (6). Gap junctions permit direct exchanges and sharing of ions and small molecules (≥1,000 Da), and the properties of gap junction channels are determined by the Cx component (11). Gap junctions are formed by aggregation, at the plasma membrane, of numerous Cx oligomers into plaque. Phosphorylation is a crucial step of Cx assembly into gap junctions, channel gating, and turnover. With the exception of Cx26, all the connexins are phosphoproteins.
Cx43 is the main connexin expressed in the interfollicular epidermis (7) and in all epidermal adnexae. Cx43 is also detected in the dermis, although at a much lower level. There is a significant increase in Cx43 expression when keratinocytes enter the suprabasal layer, suggesting that this increase is an important feature of the complex program of keratinocyte differentiation (24). Cx43 is phosphorylated on serine/threonine and tyrosine residues. Cx43 phosphorylation is associated with positive and negative regulation of gap junction communication (4). Moreover, Cx43 incorporation into gap junctions, as assessed by the acquisition of Triton X-100 insolubility, is closely related to its phosphorylation state (30).
Alterations in Cx have been associated with various skin disorders. In psoriasis, the keratinocyte hyperproliferation is linked to a massive increase in Cx26 expression (12). There are also reports of two rare diseases associated with mutations in Cx genes. Mutations in Cx31 are responsible for an autosomal dominant disease, erythokeratoderma variabilis (32), and Cx26 for a form of palmoplantar keratoderma associated with hearing loss (14). So far, no mutations have been found in Cx43. However, abnormal Cx43 localization has been observed in an autoimmune acantholytic disease, Cx43 being internalized in cells that have lost cell-cell contacts (8). Moreover, Cx seem to be required for various mechanisms that fall into general classes such as speed, synchrony, symbiosis, and stimulus/suppression (27).
Because gap junctions participate in epidermal homeostasis, we postulated that keratinocyte GJIC could be a target of UVA. We thus studied the expression of Cx43, one of the major Cx expressed in epidermal keratinocytes, after exposure of primary cultured human keratinocytes to UVA.
MATERIALS AND METHODS
Human keratinocyte primary culture.
All cell culture products were from GIBCO Life Technologies (Cergy Pontoise, France).
For cell isolation, human skin fragments were obtained from abdominal plastic surgery (in accordance with French bioethics legislation), rinsed twice in phosphate-buffered saline (PBS), and then treated for 30 s with 70% ethanol to sterilize the preparation. The fragments were rinsed in PBS and cut into thin strips 1.5 to 2 cm2long. To remove the epidermis from the dermis, the fragments were incubated overnight in 0.25% trypsin and 1% gentamicin. The detached epidermis was then homogenized, filtered, and centrifuged. Epidermal cells were resuspended, counted, and plated in Dulbecco's minimal essential medium containing 10% fetal calf serum, 10 ng/ml mouse epidermal growth factor (Sigma, Saint Louis, MO), 10−9 M choleric toxin (Sigma), and 0.4 μg/ml hydrocortisone (Sigma). This medium was replaced 24 h later with keratinocyte serum-free medium (GIBCO). Cells were cultured at 37°C in 5% CO2/air.
Functional assay of keratinocyte GJIC.
Keratinocytes were grown to confluence on 60-mm plastic culture dishes (Corning Costar, Cambridge, MA). Microinjection experiments were performed at room temperature using a Fluovert Leica Epifluorescence inverted microscope. A 10% solution of lucifer yellow (Sigma) in 0.33 M LiCl was microinjected into cells via a glass micropipette connected to an Eppendorf air injector (Eppendorf, France) (1, 18,23). To control the microinjections, the entire procedure was monitored under a phase-contrast lens. At the end of the injections (5 min), the culture medium was removed and the cells were fixed with 4% paraformaldehyde for 10 min. The cultures were then rinsed twice with PBS. Injected cells were identified, and neighboring fluorescent cells were counted. Cells were then photographed with a high-performance charge-coupled device camera (COHU) coupled to a computer and analyzed with Perfectimage software (Iconix). Keratinocyte transfer capacity was determined by measuring the surface of fluorescent cells and expressed in relative units of lucifer yellow surface diffusion.
Ultraviolet A (365 nm) irradiation.
Confluent keratinocytes were irradiated with UVA 365 nm at a dose of 10 J/cm2 for 50 min (Bio-Sun RMX 3W, Vilber-Lourmat, France). Cells were irradiated in PBS to avoid secondary toxicity due to culture medium modifications. After removal of PBS and addition of culture medium, irradiated cells were incubated at 37°C in 5% CO2/air for various times. Unirradiated PBS-treated keratinocytes were used as controls.
Measurement of cell viability and lipid peroxidation.
Cell survival was studied by measuring mitochondrial activity with the XTT reagent kit [sodium 3′-[1-phenyl-amino-carbonyl) 3,4-tetrazolium] bis(4-methoxy-6-nitro) benzenesulfonic acid] (Boehringer Mannheim). This assay is based on cleavage of the tetrazolium salt XTT into a soluble orange formasan dye, mainly by the mitochondrial dehydrogenase activity of living cells. Formasan formation was quantified with a spectrophotometer at 450 nm by using a Fluostar microtiter plate reader (BMG). Keratinocytes were plated in 96-well plates at a density of 104/well and grown to confluence. After irradiation at various UVA doses (see above), cells were incubated with XTT for 2 h at 37°C in 5% CO2/air, and absorbance was then determined. Results were analyzed with Biolise software and data were expressed as mean optical density (OD) ± SD.
Lipid peroxidation due to UVA exposure was measured with the thiobarbituric acid-reactive substances (TBARS) method. Keratinocytes were plated in petri dishes, 105 cells per dishes, and grown to confluence. After irradiation, supernatants (900 μl) were collected and 90 μl of butylated hydroxytoluene (BHT) were added (19), followed by 1 ml of 0.375% TBA solution in 0.25 M HCl containing 15% trichloroacetic acid. The mixture was heated to 80°C for 15 min, cooled on ice, and extracted with butanol. The organic phase was collected for fluorescence analysis (eexc = 515 nm, eem = 550 nm). Fluorescence was measured with a Shimadzu 1501 spectrofluorophotometer (Shimadzu, Kyoto, Japan). TBARS were expressed in malondialdehyde (MDA) equivalents with tetraethoxypropane as standard. Data were normalized to the protein content of the same cells collected by scraping. Data are expressed as the mean ± SD for three petri dishes.
Cx43 protein assay.
Cx43 was quantified by Western blotting. Cellular total protein (50 μg) was separated in 12% polyacrylamide gel and transferred to nitrocellulose membranes (Bio-Rad). After blocking with 10% dried milk in Tris · HCl, pH 7.6, 150 mM NaCl, 0.5% Tween 20 (TBS-T) solution, the membranes were washed and incubated with 1 μg/ml anti-Cx43 polyclonal antibody (Zymed Laboratory, San Francisco, CA) in TBS-T buffer. After being washed with TBS-T buffer, the membranes were incubated with horseradish peroxidase (HRP)-labeled secondary antibody (1:5,000), and the blots were developed with the ECL enhanced chemiluminescence detection kit (Amersham, Little Chalfton, UK).
Cell surface Cx43 expression was quantified in biotinylation assays on freshly cultured human keratinocytes. Cells were rinsed three times in cold PBS and then incubated for 20 min with 0.5 mg/ml biotin (Pierce) in PBS containing 1 mM MgCl2 and 0.5 mM CaCl2. After incubation, cells were rinsed three times with 50 mM glycine/PBS. Lysis buffer [400 μl, 0.25 mM saccharose, 2 mM MgSO4, 0.4% Triton, 40 mM PMSF, 0.5 mg/ml leupeptin, and 2 mM 1,4-dithiothreitol (DTT) in 5 mM Tris · HCl, pH 7.5] was then added, and cells were collected by scraping. Protein content was quantified and samples (100 μg) were subjected to cell surface protein purification as follows. Lysates were incubated for 1 h with avidin (monomeric)-agarose (Sigma). After incubation and purification, the protein extract was subjected to Western blot.
Subcellular fractionation of keratinocytes was performed to isolate Triton X-100-soluble and -insoluble Cx43. The procedure used solubilized Cx43 that was not clustered in plaques, whereas plaque-associated Cx43 was quantitatively recovered in the insoluble pellet. Keratinocytes were rinsed three times with chilled incubation buffer (136.8 mM NaCl, 5.36 mM KCl, 0.336 mM Na2HPO4, 0.345 mM KH2PO4, 0.8 mM MgSO4, 2.7 mM CaCl2, and 20 mM HEPES, adjusted to pH 7.5) containingN-ethylmaleimide and PMSF and scraped in 1 ml of incubation buffer (30). Cell pellets were suspended in 500 μl of incubation buffer supplemented with leupeptin (Sigma) and soybean trypsin inhibitor (Sigma). After Triton X-100 (1% final concentration) was added, cells were centrifuged at 100,000 g for 50 min at 4°C. The supernatant was collected as the soluble fraction and the pellet as the insoluble fraction. Samples were then immunoblotted as described above. Immunoblots were scanned densitometrically using Gel 3.0 image analysis software (Iconix).
Cx43 immunolocalization and actin labeling.
Keratinocytes were cultured to confluence on coverslips. After treatment, cells were rinsed twice with cold PBS and fixed with 3% paraformaldehyde for 15 min at room temperature and then permeabilized and blocked with 0.1% Triton and 3% bovine serum albumin in PBS (1 h at room temperature). After three washes with PBS, cells were incubated with a rabbit polyclonal antibody against Cx43 (Zymed) diluted 1:100 in TBS-T for 2 h at room temperature. Cy3-coupled goat anti-rabbit (Jackson Laboratories, West Grove, PA) diluted 1:200 in 1% dried milk/TBS-T solution was used as the secondary antibody. The actin network was labeled with phalloidin-FITC (Sigma) diluted 1:500 in PBS and incubated for 30 min with the cells. Cells were observed with a Zeiss LSM 510 confocal microscope equipped with argon (488 nm) and helium-neon (543 nm) lasers. Images were acquired and stored on a computer.
Cx43 mRNA measurement.
Cx43 mRNA was measured by means of quantitative PCR. Total keratinocyte RNA was prepared using a kit (Promega). Equivalent quantities of RNA were assayed by real-time RT-PCR using an RNA amplification kit (SYBR Green, Roche, France). Melting curve analysis of PCR products was used to separate the higher melting point-specific PCR product from nonspecific products. For quantification, following the elongation step in each PCR cycle and just before fluorescence reading, the reaction temperature was stepped up to the temperature required to melt nonspecific PCR products. The primers used to amplify a 400-base pair (bp) fragment of the Cx43 gene were TCTGAGTGCCTGAACTTGC (sense) and TTATCTCAATCTGCTTCAAGTGC (antisense). As an endogenous control for PCR, we used 18S rRNA to normalize Cx43 mRNA. The amplicon generated from the 18S rRNA gene is 187 bp long. Equivalent quantities of RNA were assayed by real-time RT-PCR using the Taqman Ribosomal RNA Control Reagent VIC probe (Applied Biosystems). All experiments were done with a Roche light cycler.
PKC activity assay.
One hour after irradiation, cells were rinsed with cold PBS and scraped into lysis buffer (as described above). PKC activity was assayed by measuring P32 transfer to a synthetic peptide and measured with a PKC enzyme assay system (Amersham Pharmacia). Radioactivity was measured with a Beckman LS6000IC scintillation counter (Beckman).
Human keratinocyte sensitivity to UVA irradiation.
Confluent cells in 96-well plates were submitted to UVA doses ranging from 5 to 20 J/cm2. Cell viability (XTT reduction) values in irradiated and nonirradiated cultures are shown in Fig.1. Cell viability was not altered by UVA, regardless of the dose. These results were confirmed by the Trypan blue exclusion test (data not shown). A UVA dose of 10 J/cm2 was used in subsequent experiments.
UVA irradiation modifies GJIC.
GJIC was investigated by dye transfer microinjection assay. Confluent keratinocytes were exposed to increasing UVA doses and microinjected 2 h later to determine their capacity to transfer lucifer yellow. As illustrated in Fig. 2 A, control cells possessed a transfer capacity of 11.3 ± 3 relative units (rU). GJIC values fell as the UVA dose increased. A dose of 10 J/cm2 diminished GJIC twofold (6.4 ± 2 rU), and a dose of 20 J/cm2 led to a threefold fall (3.3 ± 2 rU). UVA irradiation thus induced a dose-dependent decrease in the transfer capacity of confluent keratinocytes.
As shown in Fig. 2 B, the transfer capacity of control keratinocytes fell from 12.9 to 6.95 rU during the 24 h of culture. Thirty minutes after the end of irradiation with 10 J/cm2 UVA, keratinocytes showed a 10-fold decrease in GJIC 1.3 ± 0.3 rU vs. 12.9 ± 2 rU (control). Interestingly, 8 h after the end of the irradiation, GJIC had recovered to 9.2 ± 4 rU, a value similar to the control (9.4 ± 3 rU). However, GJIC values were more variable after irradiation.
Thus UVA irradiation at a dose of 10 J/cm2 induced a transient reduction in cell-cell communication by keratinocytes. Recovery occurred gradually over 8 h.
Cx43 immunolocalization after UVA irradiation.
Cx43 typically appeared as intense fluorescent spots forming plaques at keratinocyte cell-cell contacts and remained in this location in control cells (Fig. 3, left). Cx43 detection was performed after irradiation at 10 J/cm2for 50 min, as shown in Fig. 3 (right). Cx43 localization varied after irradiation. Immediately after the end of UVA irradiation (t0), Cx43 was localized at the cell-cell contacts, but some Cx43 plaques were also seen in the cytoplasm (Fig. 3). No internalization was detected in control cells.
One hour after irradiation, most Cx43 was localized in the cytoplasm and surrounding the nucleus (Fig. 3). This Cx43 relocation occurred during the 2 h after irradiation. Five hours after irradiation, the amount of Cx43 at cell-cell contacts partially recovered and Cx43 labeling was also more intense in the cytoplasm, suggesting de novo Cx43 synthesis. Eight hours after irradiation, the localization of Cx43 was similar to that in control cells.
Actin cytoskeleton modifications after UVA irradiation.
Immediately after irradiation (t0), the actin fiber organization was similar to that of control cells and stress fibers were visible. One hour after irradiation, the actin cytoskeleton was markedly modified, as shown in Fig. 3. The stress fibers had disappeared and numerous round spots were visible, indicating actin filament depolymerization. The latter apparently began 1 h after the end of irradiation and peaked at 2 h.
Actin reorganization was also accompanied by a change in cell morphology. Irradiated keratinocytes lost their polyhedral shape and became rounded 2 h after irradiation. Five hours after irradiation, stress fibers reappeared, and at 8 h the actin cytoskeleton displayed some stress fibers at the edge of the cell membrane and the polyhedral cell shape had been recovered.
Thus actin microfilaments were depolymerized by UVA radiation, leading to the disappearance of stress fibers and changes in cell morphology. These phenomena were transient, with cells recovering their normal pattern of actin organization 8 h after irradiation. UVA radiation thus alters both the cytoskeleton and gap junction communication in keratinocytes.
Measurement of Cx43 expression at the cell membrane.
To confirm and extend the immunolocalization data, cell surface Cx43 expression was measured. The total amount of Cx43 before and after UVA irradiation was similar (Fig.4 A). Cx43 was present in two forms: phosphorylated (46 kDa) and unphosphorylated (43 kDa). UVA irradiation specifically increased the proportion of the unphosphorylated form (Fig. 4 A).
Biotin labeling allows the detection of accessible membrane proteins and precludes the proteins trapped at cell junctions. In control keratinocytes, no Cx43 was labeled by biotin, probably because Cx43 was trapped in stable plaque (Fig. 4 B). Immediately after irradiation, a significant amount of Cx43 became accessible to biotin, suggesting a loosening of cell-cell interactions. Two hours after irradiation, a small amount of Cx43 remained detectable at the membrane. After 4 h, Cx43 membrane expression increased, with no further augmentation at 8 h. When the same experiment was repeated in the presence of the protein synthesis inhibitor cycloheximide (Fig.4, B and C), no Cx43 was observed at 4 h, suggesting that the emergence of Cx43 is due to de novo synthesis.
Gap junction plaque assembly after UVA irradiation.
To determine the precise cytoplasmic location of Cx43, we took advantage of the low detergent solubility of Cx43 trapped in gap junction plaque to separate the cytoplasmic (Triton X-100-soluble) and plaque (detergent-insoluble) fractions. As shown in Fig. 4 D, the distribution of Cx43 varied as a function of time after irradiation. The cytoplasmic fraction increased after irradiation, reaching a maximum after 3 h. Values then declined until 8 h, after which they remained constant. Cx43 in the plaque fraction behaved in the opposite way. After irradiation, most Cx43 protein was found in the insoluble fraction. After 3 h, there was a decrease in this fraction, which then increased to a maximum at around 4–5 h after irradiation and subsequently fell again. Thus UVA radiation induced Cx43 translocation from plaque to the cytoplasm. Recovery took between 2 and 5 h.
Cx43 mRNA content in control and irradiated keratinocytes.
Cx43 mRNA levels were determined using quantitative RT-PCR. When measured immediately after UVA irradiation (10 J/cm2), Cx43 mRNA levels were slightly higher than the control (Fig.5). Cx43 mRNA levels were also increased 4 h after irradiation and remained high thereafter (4–8 h). This was in keeping with the observed pattern of change in Cx43 protein levels (Fig. 4).
PKC activity in irradiated keratinocytes.
Kinase activation is required for the phosphorylation of Cx43, but uncontrolled phosphorylation can also lead to altered connexon function. PKC activity in a total keratinocyte protein extract is shown in Fig. 6 A. UVA irradiation stimulated PKC activity in confluent keratinocytes. When measured 1 h after the end of irradiation with increasing doses of UVA, PKC activity increased in a dose-dependent manner (Fig. 6 A). A 60% increase in PKC activity was observed in cells irradiated with 10 J/cm2 UVA relative to control cells.
TPA effect on GJIC, Cx43 localization and content, and actin organization.
TPA was used as a strong inhibitor of gap junctional communication (Fig. 6 B). Indeed, treatment with 100 ng/ml TPA for 2 h led to GJIC disruption, as described in other cell types. Transfer capacity in TPA-treated keratinocytes was at least three times lower than in control cells, as shown in Fig. 8B, a fall comparable with that induced by UVA. This TPA-induced loss of gap junction communication was accompanied by Cx43 translocation from cell-cell contacts to the cytoplasm, as shown in Fig. 6 C. Almost all the keratinocytes displayed Cx43 internalization around the nucleus (Fig. 6 C) after TPA treatment. This phenomenon was associated with profound actin reorganization (Fig. 6 C). TPA-treated keratinocytes contained no stress fibers, but an intense and thin actin network was seen at the cell borders. This was in contrast with what was observed in irradiated keratinocytes (Fig.6 C) where the actin cytoskeleton was disorganized. Depolymerization or breakage of microfilaments was observed in and at the cell border. Indeed, the cytoskeleton was modified but was not totally disorganized in TPA-treated keratinocytes, the main event being microfilaments reorganization.
TPA-treated keratinocytes contained a much more abundant soluble Cx43 pool and less insoluble Cx43 than control cells. TPA-treated keratinocytes contained larger amounts of hyperphosphorylated Cx43 (Fig. 6 D) than control cells and UVA-irradiated cells.
In this study, we found that UVA (365 nm) irradiation disrupts GJIC between normal human keratinocytes. This reduction in GJIC was not due to a cytotoxic effect of irradiation, as cell viability was unaffected. We did, however, observe an increase in lipid peroxidation (TBARS) in keratinocytes exposed to UVA at 15 and 20 J/cm2(results not shown). The increase was quite small relative to a previous study (28). When cells are irradiated in culture medium (such as DMEM without phenol red), the lipid peroxidation rate is high (results not shown). We therefore performed UVA irradiation in PBS, conditions under which no cytotoxicity was observed.
The reduction in keratinocyte communication via gap junctions was dependent on the intensity of UVA irradiation. Importantly, the reduction was transient at a dose of 10 J/cm2, and transfer capacity started to recover within 30 min after the end of irradiation, reaching control levels at 8 h. Surprisingly, both irradiated and control cells showed reduced GJIC after 24 h of culture at confluence. It has been shown that GJIC is reduced in fully differentiated keratinocytes (3). UVA irradiation has also been reported to disrupt GJIC in the V79 Chinese hamster fibroblast cell line (1), albeit at a dose three times higher than that used here. The latter authors did not study the reversibility of this effect.
The actin cytoskeleton is markedly altered by irradiation. Irradiation with a mixed UVB/UVA source has been reported to disrupt microfilament organization in keratinocytes (33). In our experiments, with UVA only, the keratinocyte microtubule network did not show any particular modifications (results not shown). In contrast, the actin network was disrupted 1 and 2 h after UVA irradiation of confluent keratinocytes, but this effect was reversible several hours after the irradiation. In cells naturally exposed to UVA, such as lens epithelial cells, both microtubule and actin networks are sensitive to depolymerization after UVA exposure (21). UVA possesses the energy required to enter the dermis and also damages fibroblasts. In particular, disintegration of actin filaments by UVA has been reported in V79 Chinese hamster fibroblasts (2). Among the cells that were damaged by UVA, keratinocytes demonstrated a capacity for recovery of the actin cytoskeleton after limited UVA radiation.
Cx43 immunolocalization indicated that the keratinocyte GJIC uncoupling observed after UVA irradiation was linked to Cx43 internalization from gap junction plaques to the cytoplasm. Interestingly, Cx43 returned to cell-cell contacts several hours after irradiation, in parallel to GJIC recovery and actin cytoskeleton reorganization. Cx43 located in both the cytoplasm and at cell-cell contacts appeared to be newly synthesized. Indeed, the rise in the soluble (cytoplasmic) Cx43 fraction preceded the return of Cx43 to plaques. The increase in Cx43 mRNA levels after irradiation, and results obtained with a protein synthesis inhibitor, also suggested that de novo synthesis of Cx43 protein was required for the recovery of gap junction function after UVA radiation. Further experiments will be necessary to determine the influence of UVA radiation on the dynamic process that leads from new Cx43 synthesis to gap junction plaque formation and on Cx43 gene transcription and Cx43 promoter activity.
We treated keratinocytes with TPA to investigate the role of Cx43 phosphorylation in GJIC. The effect of TPA has been ascribed to a 10-fold increase in PKC activity mainly in membrane-associated fractions (10). UV irradiation is also a potent activator of PKC in keratinocytes (16). We observed a UVA dose-dependent increase in PKC activity, with a two- to threefold rise after exposure to 5–20 J/cm2. In contrast to TPA, UVA mainly induced PKC activity in the cytosolic fraction (17). In many cell types, Cx43 phosphorylation seems to regulate its trafficking to the membrane, as well as subsequent formation of gap junction plaques, single-channel behavior, and degradation. These events have been reported to be TPA sensitive and, therefore, regulated by PKC (13). In the current study, Cx43 was also sensitive to TPA. We observed GJIC inhibition, coupled with Cx43 internalization and actin cytoskeleton alterations, both after UVA irradiation and after TPA treatment. However, the level of hyperphosphorylated cytoplasmic Cx43 increased in TPA-treated keratinocytes, whereas it remained constant in UVA-irradiated cells. Determination of the PKC isoforms stimulated after UVA irradiation could help in understanding the signal transduction pathways involved. TPA is known to activate PKCα and δ isoforms by inducing their translocation from the soluble to the particulate fraction. UVB, on the other hand, increases soluble PKCα and δ activities (5). Our results suggest that, despite the similar reduction in GJIC, the signaling cascade triggered by UVA is not strictly identical to that activated by TPA.
Gap junctions play a key role in tissue homeostasis, as demonstrated by the Cx null phenotype in mice (27). Homozygous Cx43 null mice do not survive after birth due to myocardial malformation (22). Cx43-/- fibroblasts established from Cx43-/- mice acquire a transformed phenotype (15), and these animals have a high incidence of spontaneous and chemically induced liver tumors (29). Correct expression of Cx isoforms, therefore, seems to protect against carcinogenesis. Moreover, the role of UVA in solar carcinogenesis is now generally accepted (34). We demonstrate here that human keratinocytes exposed to UVA irradiation display a transient decrease in GJIC, which corresponds to Cx43 internalization from plaque, as sometimes observed in tumor promotion (9). This decrease may represent a defensive mechanism against UVA irradiation. Indeed, the transient rupture of intercellular communication might prevent damaged cells from emitting deleterious signals to their neighbors. Multiple exposure to UVA irradiation, as occurs in vivo, could exhaust cellular repair possibilities and favor the emergence of initiated cells. Further studies will be necessary to understand, for instance, the role of PKC and/or oxygen radicals in the cytoskeleton disruption and Cx43 internalization observed after UVA irradiation. A study by L'Oreal Research indicated that the human face receives about 1,200 J/cm2 UVA irradiation annually (59th Annual Meeting of the American Academy of Dermatology), and the dose of 10 J/cm2 used in our experiments would thus be comparable to sunlight exposure for 2 days. Investigations of Cx43 localization after UV exposure of human skin in situ could provide valuable information on GJIC regulation in this tissue.
In conclusion, we report that gap junctions are a target of UVA irradiation in the epidermis. This could participate in the emergence of the epidermal photoaging phenotype and, after extreme or chronic exposure, might contribute to the development of epidermal cancer.
We thank C. Klein for his help in the confocal microscopy experiments and for careful reading of the manuscript. We also thank Prof. J. Chambaz and G. Mazzoleni for helpful discussions.
This study was supported by LVMH, Research and Development division.
Address for reprint requests and other correspondence: N. Provost, Laboratoire de Pharmacologie Cellulaire EPHE-INSERM U505, 15 rue de l'Ecole de Médecine, 75006 Paris, France (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published September 4, 2002;10.1152/ajpcell.00205.2002
- Copyright © 2003 the American Physiological Society