Rat Epidermal Keratinocytes as an Organotypic Model for Examining the Role of Cx43 and Cx26 in Skin Differentiation

Abstract

In order to characterize connexin expression and regulation in the epidermis, we have characterized a rat epidermal keratinocyte (REK) cell line that is phenotypically similar to basal keratinocytes in that they have the ability to differentiate into organotypic epidermis consisting of a basal cell layer, 2–3 suprabasal cell layers, and a cornified layer. RT-PCR revealed that REK cells express mRNA for Cx26, Cx31, Cx31.1, Cx37, and Cx43, which mimics the reported connexin profile for rat tissue. In addition, we report the expression of Cx30, Cx30.3, Cx40, and Cx45 in rat keratinocytes, highlighting the complexity of the connexin complement in rat epidermis. Furthermore, 3-dimensional analysis of organotypic skin revealed that Cx26 and Cx43 are exquisitely regulated during the differentiation process. The life-cycle of these connexins including their expression, transport, assembly into gap junctions, internalization, and degradation are elegantly depicted in organotypic epidermis as keratinocytes proceed from differentiation to programmed cell death.

Keywords:

• connexin
• gap junction
• epidermis
• keratinocytes
• organotypic culture
• internalization

INTRODUCTION

Gap junctions are specialized channels that form between two adjacent cells to allow ions and molecules of less than 1 kDa to pass from one cell to another (1), a process known as gap junctional intercellular communication (GJIC). Gap junctions are thought to be important in cell proliferation, differentiation, and homeostasis. There are 21 human and 20 mouse connexin genes that have been cloned and sequenced (2). Virtually every tissue and cell type expresses a specific subset of this large connexin family of gap junction proteins (3). Epidermal keratinocytes temporally and spatially express as many as 10 different connexins localized to various layers of the differentiating epidermis. For example, adult human palm epidermis has been shown to express Cx26, Cx30, Cx31, Cx32, Cx40, Cx43, and Cx45 at the protein level, and mRNA for Cx30.3, Cx31.1, and Cx37 (4). Adult rat epidermis expresses Cx26, Cx31, Cx31.1, Cx37, and Cx43 but not Cx32 (5, 6).

Keratinocytes undergo continual differentiation during epidermal renewal and GJIC is thought to be essential for regulated cell function in wound healing, carcinogenesis and skin disease prevention (7, 8). Mutations in Cx26, Cx30, Cx30.3, and Cx31 result in human hyperproliferative skin diseases (9, 10, 11, 12). Similarly, the existence of transgenic mouse models supports the essential role of connexins in the epidermis. For example, transgenic mice with a Cx43 C-terminal truncation die shortly after birth due to barrier dysfunction of the epidermis (13), and transgenic mice that express the human Cx26 D66H mutation driven off a keratin promoter show similar symptoms to human Vohwinkel's syndrome (11, 14). The function of multiple connexins in the epidermis has yet to be elucidated, and although transgenic mouse models provide important insights, connexin deletion or mutations can lead to major developmental abnormalities (14, 15). Therefore, in order to examine the function and regulation of connexins in epidermal differentiation, we sought a model system where keratinocytes could be grown in a manner that would recapitulate in vivo characteristics.

We have acquired an immortal rat epidermal keratinocyte (REK) cell line that can be grown, passed, and manipulated in one-dimensional monolayer culture, and importantly, induced to differentiate at a liquid/air interface into stratified organotypic epidermis (16). Our results support the conclusion that REKs are able to differentiate, stratify, and cornify into a tissue with all the morphological features and differentiation markers of normal epidermis. We show that multiple connexins are coexpressed in REK cells, and that Cx26 and Cx43 are differentially regulated. We also found that gap junction internalization into vesicle-like compartments appears to precede keratinocyte terminal differentiation and cornification. Overall, REK cell-derived organotypic epidermis is an excellent system to examine the role of connexins in epidermal differentiation and is a suitable physiological environment to examine the mechanisms associated with connexin assembly and turnover.

MATERIALS and METHODS

Cell Culture

Rat epidermal keratinocytes (REKs) are a spontaneously immortalized newborn rat epidermal keratinocyte cell line originally described by Baden and Kubilus (16). REKs, NRK cells over-expressing Cx26 (NRKv26), NRK cells over-expressing Cx30, and HBL-100 human mammary tumor cells were cultured in high glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mM L-glutamine, 10% fetal bovine serum, 100 U/ml penicillin and 100 μ g/ml streptomycin (all from Invitrogen, Burlington, ON). Cells were passed using 0.25% trypsin-EDTA at 37°C for 5–30 minutes. For immunofluorescent labeling of monolayer cultures, REK cells were grown on glass coverslips coated with or without collagen type I (Sigma, Oakville, ON) in phosphate buffered saline (PBS). Cells destined for Western blot analysis were grown to confluence in 60 mm tissue culture dishes.

In order to grow organotypic epidermis, 24 mm transwell filter inserts containing 3 μ m pores (Falcon, VWR, Canada) were coated with 1 ml of rat tail type I collagen as previously described (17). REK cells were plated directly onto the collagen matrix at 3× 10⁵ cells per well and growth medium was placed in both the upper and lower chambers. Three days after plating the cells, the medium from the upper chamber was removed to expose the cells to the liquid/air interface, and the cells were cultured for an additional 11 days with daily lower chamber medium changes.

Preparation of Cryosections

Two-week-old organotypic epidermis was peeled from the collagen matrix, rolled up, and cut in half. Each half was embedded in Optimal Cutting Temperature compound (OCT; Tissue-Tek) by freezing in a cup of isopentane submerged in liquid nitrogen. Embedded samples were stored at -80°C until sectioning. 12 μ m sections were cut with a cryostat and mounted on Superfrost PLUS slides (VWR).

Histological Staining

In order to observe epidermal structure, cryosections were fixed in 3.7% formaldehyde for 20 min and rinsed in tap water. Samples were stained with 1% haematoxylin for 5 min, rinsed in 1% hydrochloric acid in 70% ethanol and further stained with 1% eosin for another 5 min. Finally, the organotypic tissue samples were progressively rinsed in 90% and then 100% ethanol followed by two rinses in xylene and then mounted in cytoseal (Canadawide Scientific, Ottawa, ON).

RT-PCR

Total RNA was collected from confluent REK monolayer cultures using Trizol (Invitrogen) as per manufacturer's instructions. Connexin cDNA was amplified using rat or mouse connexin-specific primers with the one-step RT-PCR kit (Qiagen, as per manufacturer's instruction). RT-PCR primer sequences for Cx30.3, Cx31, Cx33, Cx45, Cx46, and Cx50 were previously published (18). The following primer sequences were used:

• Cx26 : sense, 5′ CGGAAGTTCATGAAGGGAGAGAT 3′;

• antisense, 5′ GGTCTTTTGGACTTTCCTGAGCA 3′;

• Cx30: sense, 5′ AATGTGGCCGAGTTGTGTTA 3′;

• antisense, 5′ CCAAGGCCCAGTTGTCAC 3′;

• Cx31.1 : sense, 5′ ATATACCCTCCCTTCTATGGT 3′;

• antisense, 5′ TCACAGAATGGTTTTCTTCA 3′;

• Cx32: sense, 5′ CTGCTCTACCCCGGCTATGC 3′;

• antisense, 5′ CAGGCTGAGCATCGGTCGCTCTT 3′;

• Cx36: sense, 5′ GGGGTGCTGCAGAACACAGAGA 3′;

• antisense, 5′ ACCACACAAATGCCGCTCACA 3′;

• Cx37: sense, 5′ GGCTGGACCATGGAGCCGGT 3′;

• antisense 5′ TTCTGGCCACCCTGGGGGGC 3′;

• Cx40: sense, 5′ CTGGCCAAGTCACGGCAGGG 3′;

• antisense, 5′ TTGTCACTGTGGTAGCCCTGAGG 3′;

• Cx43: sense, 5′ GGCTGCTCCTCACCAACGGCT 3′;

• and antisense, 5′ AGGTCATCAGGCCGAGGC CTG 3′.

All samples were resolved on 1.5% (for Cx30) or 2% agarose gel for analysis.

Microinjection

To determine whether REKs were gap junction coupled, confluent monolayer cultures were pressure-microinjected with 5% Lucifer yellow (Molecular Probes) using an Eppendorf microinjection system and a Leica inverted epifluorescent microscope as we have previously described (19). At least 20 cells were microinjected and examined for dye spreading.

Immunofluorescence

Immunofluorescence and confocal microscopy were used to examine Cx43, Cx26, keratin 14, involucrin, and loricrin localization in REK cells as modified from our previous studies (19). Monolayer cultured cells were plated on 12 mm glass coverslips and fixed with an ice-cold solution of 80% methanol and 20% acetone at 4°C for 15 min. Both cryosections and fresh organotypic cultures for whole mount optical sectioning were fixed with 3.7% formaldehyde overnight at room temperature. Standard antibody labeling protocols were used as previously described (19) except that 0.1% Triton-X-100 was added to the blocking, primary antibody and secondary antibody solutions to increase antibody penetration into the tissue. Connexin localization was analyzed using rabbit anti-Cx43 (1:500, Sigma-Aldrich, St. Louis, MO) or rabbit anti-Cx26 (1:50, Zymed Laboratories Inc., San Francisco, CA) antibodies. Antibodies to keratin 14 (dilution 1:200, Lab Vision Co. Fremont, CA), involucrin (dilution 1:50, Covance Research Products, Denver, PA), and loricrin (dilution 1:100, Covance Research Products), were used to assess the degree of differentiation in organotypic epidermis. Secondary anti-mouse or anti-rabbit antibodies conjugated to Texas red or fluorescein isothiocyanate (FITC) (1:100, Jackson Laboratories, Westgrove, PA) were used. Finally cell nuclei were stained for 15 mins with Hoechst 33342 (10 μ g/ml) in PBS. The samples were mounted in Airvol or Vectashield (Vector Laboratories, Burlington, ON).

Immunolabeled monolayer cultures of keratinocytes were imaged using a 63X oil objective lens mounted on a Zeiss LSM 510 META system as previously described (19). For imaging of intact organotypic samples a 40X water objective lens was used. Organotypic epidermis (approximately 50 μ m thick) was optically z-sectioned (1 μ m optical slices) using a Zeiss LSM 510 microscope and images were prepared using Adobe Photoshop and Corel Draw software.

SDS-PAGE and Western Blot Analysis

Confluent monolayer cells were collected and lysed in ice-cold triple detergent lysis buffer containing 50 mM Tris/HCL pH 8.5, 150 mM NaCl, 0.2% sodium azide, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate and protease inhibitor (Roche). Organotypic tissue was lysed in ice-cold urea lysis buffer as described (17) plus 0.1% SDS, 1% NP-40 and 1% deoxycholic acid and protease inhibitor (Sigma). The lysates were sonicated (unless blot analysis was for Cx26, in which case the samples were sheared with a 25 G needle and syringe) and centrifuged to remove debris from the lysates. Protein concentrations were calculated by Bradford assay (BioRad) or Bicinchonic Acid (BCA) assay (Pierce Biotechnology Inc., Rockford IL). Equal amounts of protein (10–30 μ g) were resolved by 10% (or 12% for Cx26 analysis) SDS-PAGE, transferred to nitrocellulose paper and immunoblotted with rabbit anti-Cx43 (1:10,000, Sigma-Aldrich), rabbit anti-Cx26 (1:500, Zymed Laboratories Inc.), and mouse anti-GAPDH (1:10,000, Chemicon) antibodies. Secondary antibodies were conjugated to horseradish peroxidase (1:10,000, Sigma-Aldrich) and specific antibody binding was detected using the West-Pico chemiluminescence detection reagents (Pierce Biotechnology Inc.)

RESULTS

Growth and Architecture of REK Organotypic Epidermis

To assess the role of connexins in epidermal differentiation we obtained and cultured a rat keratinocyte cell line that retained the properties of differentiation and stratification in culture when grown at a liquid/air interface in a transwell chamber (Figure 1A) as described previously. Haematoxylin and eosin staining of organotypic epidermis revealed that within 14 days the resulting organotypic epidermis stratified into approximately 4–5 layers which included a single layer of basal cells, a 2–3 layer thick suprabasal region and a cornified layer (Figure 1B).

Figure 1. Organotypic model of REK differentiation. (A) Rat epidermal keratinocytes (REKs) can be grown as an undifferentiated monolayer or at a liquid/air interface that supports 3-dimensional differentiation into cell layers that mimics the epidermis. (B) Haematoxylin and eosin staining of 2-week-old organotypic epidermis revealed that the keratinocytes differentiated into 3–4 vital cell layers plus a large cornified layer. Bar = 100 μ m.

REKs Express Multiple Connexins and are GJIC-Competent

Keratinocytes have been previously shown to express multiple connexins in vivo (5, 6, 20, 21, 22). To characterize connexin expression in the REK cell line and to assess whether they possess a connexin profile similar to keratinocytes in vivo, we examined the mRNA expression for 14 different connexin family members. RT-PCR, using rat specific connexin primers revealed that REK monolayer cultures express mRNA for Cx26, Cx30.3, Cx31, Cx31.1, Cx37, Cx40, Cx43, and Cx45 but not Cx32, Cx33, Cx36, Cx46, or Cx50 (Figure 2A). In subsequent experimentation where the annealing conditions were changed to 53°C we also detected a light band in our REK sample (Figure 2B, 2nd lane, arrow) that was extracted and confirmed by sequencing to be Cx30. The Cx30 message was also detected in NRK cells overexpressing Cx30 (⁺ control). Consequently, we showed that REKs express mRNA for at least nine members of the connexin family. To determine whether REKs were assembling one or more of these connexins into functional channels, confluent monolayer cultures of REKs were microinjected with Lucifer yellow. Dye was observed to spread to several adjacent cells indicating that REK cells were functionally coupled (Figure 2C).

Figure 2. REK cells express mRNA for nine different connexins and are well coupled. Total RNA was extracted from confluent monolayers of REKs and subjected to RT-PCR using 14 connexin-specific primers. mRNA for Cx26, Cx30.3, Cx31, Cx31.1, Cx37, Cx40, Cx43, Cx45 (A), and Cx30 (B) was detected in REKs. While Cx30 message was not detected at the annealing temperature of 55°C and 30 cycles (A) it was detected when an annealing temperature of 53°C and 35 cycles were used (B, arrow denotes position of Cx30 which was verified by band extraction and sequencing). Confluent monolayers of REKs were microinjected with Lucifer yellow to assess the GJIC status. (C) REKs readily transferred dye to the surrounding cells. Combined phase contrast/fluorescent image and asterisks denotes the microinjected cell (C, right panel). “+control” in B was RNA extracted from Cx30 over-expressing NRK cells, and “-control” was a water control. Bar = 40 μ m.

Cx43 and Cx26 are Differentially Expressed in the Stratified REK Organotypic Epidermis

In order to examine whether Cx43 and/or Cx26 protein was present in REKs and to assess their localization patterns we first examined monolayers of REKs. Cx43 gap junction plaques were observed between adjoining cells (Figure 3A, arrows), however, Cx26 was detected only in stratified piles of overgrown monolayer cultures in which differentiation had begun (Figure 3B, arrows). The presence of Cx26 only in stratified subpopulations of REKs, would suggest that Cx26 is normally silent in REKs, and mechanisms related to keratinocyte differentiation are required for its expression. Western blot analysis of monolayer cultures confirmed the presence of Cx43 and absence of Cx26 (Figure 3C). However, in overgrown monolayers of REKs the levels of both Cx26 and Cx43 were elevated with respect to GAPDH, indicating that the initiation of REK differentiation enhances the expression of both Cx26 and Cx43 (Figure 3C).

Figure 3. Cx43 is expressed in monolayer REK cultures and Cx26 is up-regulated following REK differentiation. Cx43 was detected by immunofluorescence in all cells grown in monolayer (A, arrows) and these findings were confirmed by Western blot (C). Cx26 was only detected in stratified piles of overgrown monolayer cells (B, arrows). Western blot analysis of monolayers and overgrown monolayers of REKs, as well as organotypic epidermis, revealed the differential regulation of Cx26 and Cx43 with respect to each other and GAPDH (C). NRK cells over-expressing Cx26 (NRKv26) were used as a positive control for both Cx43 and Cx26 while HBL-100 cells were found to be devoid of both Cx43 and Cx26 (C). Bar = 20 μ m.

In order to characterize Cx43 and Cx26 localization in REK organotypic epidermis, samples were cyrosectioned, immunofluorescently labeled and compared to molecular markers of keratinocyte differentiation including keratin 14, involucrin, and loricrin. Cx43 was found to be abundant in the basal and suprabasal layers with diminishing evidence for its expression as the cells approached the cornified layer (Figure 4A, arrows). While keratin 14 is characteristically used to denote basal keratinocytes, the slow turnover kinetics of keratin intermediate filaments has resulted in it also being found in more differentiated suprabasal cells in vivo (23, 24). Consequently, we found that keratin 14 was localized to both basal and suprabasal layers of organotypic epidermis (Figure 4B, Figure 4F, red). To assess whether the organotypic epidermis was exhibiting properties of terminal differentiation, sections were double immunolabeled for Cx43 and either involucrin or loricrin (Figure 4C, Figure 4D). Then, Cx43 was observed in areas expressing involucrin (Figure 4C, arrows) but not loricrin (Figure 4D, arrowhead). Interestingly, Cx26 was detected in differentiated suprabasal layers only (Figure 4E, Figure 4F, Figure 4G, Figure 4H, arrows) and was particularly evident in areas expressing loricrin (Figure 4H, arrowhead). Western blot analysis of organotypic epidermis confirmed the presence of both Cx43 and Cx26 (Figure 3C). It was also notable that the level of Cx26 constituted a large percentage of the total Cx26 and Cx43 levels found in organotypic epidermis when compared to monolayer cultures (Figure 3C).

Figure 4. Cx43 is expressed in the basal and suprabasal cells of organotypic epidermis while Cx26 is only expressed in differentiated suprabasal cells. Cryosections of organotypic cultures revealed that Cx43 was abundant in the basal layer and suprabasal layers of the epidermis (A, arrows). Cx43 was localized to the same layers as keratin 14 (B, arrows), and involucrin (C, arrows). Cx43 was not localized in loricrin positive cells (D, arrowhead). Cx26 was localized to the upper suprabasal cell layers (E, F arrows) and epidermal layers expressing involucrin and loricrin (G, H, arrowhead). Dots represent the basal limits of the epidermis. The area between the dashed lines represents the cornified layer. Hoechst (blue) = nuclei. Bar = 50 μ m.

Spatial Localization of Cx43 and Cx26 in Organotypic Epidermis

High-resolution analysis of Cx43 and Cx26 localization in organotypic epidermis was further assessed by optical imaging and 3-dimensional analysis (Figure 5). Cx43 gap junction plaques were seen in the basal layer and suprabasal layers of organotypic epidermis until approximately 30 μ m above the level of the extracellular matrix support (Figure 5C). Upon close examination of the basal cells, scattered Cx43 gap junctions were seen between these densely packed cells, but there was also evidence of diffuse cell surface fluorescence suggesting that there was a population of Cx43 that was transported to the cell surface but had not clustered into gap junction plaques (Figure 5E, arrowheads). In the suprabasal cell layers, where keratinocyte differentiation had begun, the cells became more flattened and less densely packed with increased numbers of Cx43 gap junction plaques at the cell surfaces. In the upper suprabasal layers, where cells were reaching terminal differentiation, there were prominent Cx43 gap junction plaques between cells, however, there was a major increase in the number of intracellular Cx43-positive vesicle-like structures (Figure 5F, arrows). Finally, in presumably apoptotic cells that resided just under the stratum corneum there was considerably reduced evidence of Cx43 (Figure 5C, 40 μ m section).

Figure 5. Subcellular localization of Cx26 and Cx43 in stratified organotypic epidermis. Optical sectioning of immunolabeled 3-D organotypic epidermis was used to resolve Cx43 and Cx26 subcellular localization. The schematic (A) of the epidermis depicts the relative position of the optical slices for Cx43 immunolabeled epidermis while the schematic (B) represents the relative position of the optical slices for Cx26 immunolabelled epidermis. Cx43 was observed throughout the basal and suprabasal cells (C, arrows). Cx43 was diffusely localized to, or near, the cell surface (E, arrowheads) and was also observed as punctate gap junction plaques in basal keratinocytes (C, E). Cx43 was more readily observed as punctate gap junctions in suprabasal cells where intracellular vesicles also became apparent (F, arrows). Cx26 was not detected in basal cells (D, 10 μ m) but was prominent as punctate gap junctions in suprabasal cells (D, arrows). Cx26 was detected in intracellular vesicles as cells approach the cornified layer (G, arrows). Hoechst (blue) = nuclei. Bar = 20 μ m.

As suggested from the immunolabeled cryosections, Cx26 expression was not detected in basal keratinocytes and detected only in the suprabasal layers of optical slices that were taken 25 μ m above the collagen substrate (Figure 5D). This localization is consistent with Cx26 being restricted to the upper suprabasal layers in adult rat epidermal tissue (5). Although considerably less frequent, Cx26-positive intracellular vesicles, similar to those seen for Cx43, were detected in keratinocytes below the cornified layer (Figure 5G, arrows). As expected, there was a loss of Cx26 staining in the cornified layer where the cells were dead or dying (Figure 5D, 35 μ m section).

DISCUSSION

In 1983, Baden and Kubilus isolated a REK cell line that maintained the ability to differentiate and stratify into an in vivo-like epidermis (16). Importantly, when REKs were grown as organotypic cultures, the architecture and epidermis thickness was found to be similar to that found in rat epidermis in vivo (5). We found that the connexin expression profile of these REKs included all the connexins reported from in vivo studies (5, 6) plus four connexins not previously reported. RT-PCR analysis revealed that REKs express mRNA for at least nine connexins: Cx26, Cx30, Cx30.3, Cx31, Cx31.1, Cx37, Cx40, Cx43, and Cx45. Conversely, REKs did not express mRNA for Cx32, Cx33, Cx36, Cx46, or Cx50. In rat epidermis, mRNA for Cx26, Cx31, Cx31.1, Cx37, and Cx43 has been previously reported (5, 6) confirming that the REK model encompasses the connexin profile of rat epidermis. While Cx30, Cx40, and Cx45 have been reported in mouse and human epidermis (4, 20, 25), to our knowledge, this is the first report of Cx30, Cx30.3, Cx40, and Cx45 in rat epidermis. It remains to be determined if the expression of these connexins can be demonstrated in rat epidermis in vivo.

Cx43 and Cx26 have differential expression patterns in the epidermis that change during development, wound healing, and disease (26). In monolayer cultures of REKs, Cx43 was readily detected at the protein level and localized to gap junction plaques. In contrast, Cx26 was rarely detectable in monolayer unless the keratinocytes began to stratify. In organotypic epidermis, Cx43 was detected in basal and suprabasal cells while Cx26 was present in the more differentiated suprabasal cells, mimicking the expression characteristics of these connexins in rat epidermis (5), human organotypic epidermis (27, 28), human epidermis (4, 29, 30), and mouse epidermis (31). This data provides a strong correlation between Cx26 expression and epidermal stratification, raising the possibility that either the differentiation of the keratinocyte stimulates Cx26 expression, or Cx26 expression enhances keratinocyte differentiation.

As keratinocytes undergo differentiation in vivo, they express a number of cytoskeletal proteins necessary for epidermal integrity. We used keratin 14 to denote basal and suprabasal keratinocytes (with preferential expression in the basal cells), involucrin to denote suprabasal cells, and loricrin to distinguish the upper suprabasal cells as previously described (32). The use of differentiation markers allowed us to compare the layers in which connexins were localized, the architecture of the epidermis, and the degree of cell differentiation. Importantly, similar to in vivo epidermis (31), organotypic epidermis expressed Cx43 in the basal and suprabasal layers but not in the subcorneum layer. Inversely, Cx26 could be detected only in the upper epidermal layers that express involucrin and loricirin, confirming the unique up-regulation of Cx26 in suprabasal cells of organotypic epidermis. Consequently, we concluded that organotypic epidermis generated from REKs exhibits properties similar to the in vivo environment, making it a suitable model for studying the role of connexins in epidermis differentiation.

Differential regulation of multiple connexins in the epidermis suggests that there is tight control over connexin expression, transport, assembly, and turnover. In basal cells, Cx43 was most evident at the cell surface in an organization that did not reflect gap junction plaques but rather it was found diffusely at the cell surface. The diffuse cell surface pattern of Cx43 expression could represent assembled gap junction channels that had not yet clustered into plaques, or possibly, undocked connexons or hemichannels which are known to exist at the cell surface (33, 34). While diffuse cell surface expression of Cx43 has been well documented in over-expressing cells (35), it is rare to immunodetect a subpopulation of Cx43 at the cell surface that is not assembled into gap junctions. As cells approach terminal differentiation, Cx43 is frequently found in circular structures that likely represent endosomal compartments or annular junctions (36, 37). The enrichment of intracellular Cx43 likely reflects the down-regulation of Cx43 prior to cell apoptosis as cornification of the epidermis begins. Finally, the minimal evidence of Cx43 in cells entering the state of cornification would indicate that protein degradation is clearing preexisting gap junctions and Cx43 from these cells. In the case of Cx26, it was not detected until the cells differentiated beyond the basal cells. While some diffused Cx26 staining was detected at the cell surface, gap junction-like plaque structures were most evident in suprabasal layers. This would suggest that Cx26 is rapidly assembled into gap junctions upon reaching the cell surface. Similar to Cx43, increased evidence of intracellular vesicles enriched in Cx26 could be detected, albeit not nearly as numerous as that seen for Cx43, in the subcornified strata of the epidermis. Given the variations in size of both Cx43 and Cx26 intracellular structures, it is possible that gap junction plaques are being internalized as annular junctions and endosomes prior to their targeting to lysosomes for degradation, as has been reported in other cell types (38).

The results of the present study demonstrate that REK organotypic epidermis is an excellent model for studying the effects of connexin expression on epidermal growth, differentiation, function and disease. In our study, four previously unreported connexins were found to be expressed in rat keratinocytes. Moreover, our studies suggest that during keratinocyte differentiation Cx43 and Cx26 undergo precise regulation that is not only governed at the level of gene expression but also at the level of assembly and turnover.

ACKNOWLEDGEMENTS

We would like to thank Dr. Vincent Hascall for the gift of the REKs. We are also grateful to Dr. Gerald Kidder for providing some of the RT-PCR primers. We also extend our gratitude to Dr. Hongling Wang for providing the schematic diagram of the epidermis. Finally, this work was supported by a Canadian Institutes of Health Research Training Award to AM, an NSERC award to TT, Canadian Institutes of Health Research Operating grant and a Canada Research Chair to DWL.

Authors contributed equally and share first authorship.