If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Institut Curie, Normal and Pathological Development of Melanocyte, Orsay, France
Affiliations
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
10 The first two authors contributed equally to this work.
Flavie Luciani
Footnotes
10 The first two authors contributed equally to this work.
Affiliations
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
Department of Dermatology and Pediatric Dermatology, National Reference Centre for Rare Skin Disorders, Hôpital Saint-André Bordeaux, Bordeaux, FranceINSERM U1035, University of Bordeaux, Bordeaux, France
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
Department of Dermatology and Pediatric Dermatology, National Reference Centre for Rare Skin Disorders, Hôpital Saint-André Bordeaux, Bordeaux, FranceINSERM U1035, University of Bordeaux, Bordeaux, France
Department of Dermatology and Pediatric Dermatology, National Reference Centre for Rare Skin Disorders, Hôpital Saint-André Bordeaux, Bordeaux, FranceINSERM U1035, University of Bordeaux, Bordeaux, France
11 Lionel Larue and Véronique Delmas are co-last authors.
Affiliations
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
11 Lionel Larue and Véronique Delmas are co-last authors.
Véronique Delmas
Footnotes
11 Lionel Larue and Véronique Delmas are co-last authors.
Affiliations
Institut Curie, Normal and Pathological Development of Melanocytes, Orsay, FranceCNRS UMR3347, Orsay, FranceINSERM U1021, Orsay, FranceEquipe labellisée, Ligue Nationale contre le Cancer, Orsay, France
Vitiligo is the most common depigmenting disorder resulting from the loss of melanocytes from the basal epidermal layer. The pathogenesis of the disease is likely multifactorial and involves autoimmune causes, as well as oxidative and mechanical stress. It is important to identify early events in vitiligo to clarify pathogenesis, improve diagnosis, and inform therapy. Here, we show that E-cadherin (Ecad), which mediates the adhesion between melanocytes and keratinocytes in the epidermis, is absent from or discontinuously distributed across melanocyte membranes of vitiligo patients long before clinical lesions appear. This abnormality is associated with the detachment of the melanocytes from the basal to the suprabasal layers in the epidermis. Using human epidermal reconstructed skin and mouse models with normal or defective Ecad expression in melanocytes, we demonstrated that Ecad is required for melanocyte adhesiveness to the basal layer under oxidative and mechanical stress, establishing a link between silent/preclinical, cell-autonomous defects in vitiligo melanocytes and known environmental stressors accelerating disease expression. Our results implicate a primary predisposing skin defect affecting melanocyte adhesiveness that, under stress conditions, leads to disappearance of melanocytes and clinical vitiligo. Melanocyte adhesiveness is thus a potential target for therapy aiming at disease stabilization.
Abbreviations
DDR1
discoidin domain receptor tyrosine kinase 1
Ecad
E-cadherin
H202
hydrogen peroxide
INTRODUCTION
Vitiligo is the most common chronic acquired depigmenting disorder, with a prevalence of ∼1% in the world population (70 million people are affected) resulting from the loss of melanocytes from the basal epidermal layer (
). The progression of depigmentation in vitiligo is unpredictable, and there are no markers specific for the various steps of the disease. The etiology of vitiligo remains poorly understood, but progressive disease is clearly characterized by inflammatory infiltrates at the margin of lesions, suggesting an immune-mediated acceleration phase common to all forms of vitiligo (
). The inflammatory cells are mostly T cells usually found adjacent to the basal membrane where melanocytes are located. Several studies have detected serum antibodies directed against melanocyte antigens in vitiligo patients, and a correlation has been found between such antibodies and disease activity (
). Genetic investigations have implicated at least 30 susceptibility loci, encoding melanocyte components, including MC1R, TYR, and OCA2, immunoregulatory proteins targeting the innate and adaptive immunity, including HLA, PTPN22, and NLRP1, and cell adhesion molecules, including DDR1 (
). The underlying epidermal defect is poorly characterized. However, the imbalance between oxidant/antioxidant systems in the epidermis of vitiligo patients and high levels of oxidative stress markers in their peripheral blood are both consistent with the involvement of oxidative stress (
In vivo and in vitro evidence for hydrogen peroxide (H2O2) accumulation in the epidermis of patients with vitiligo and its successful removal by a UVB-activated pseudocatalase.
). The site at which vitiligo macules begin is frequently determined by local conditions, such as continuous pressure or repeated friction (Koebner’s phenomenon) (
). In keratinocytes in depigmented vitiligo lesions, the expression of two molecules implicated in cell–cell adhesion, DDR1 (discoidin domain receptor tyrosine kinase 1, also known as CD167a), and E-cadherin (Ecad), appeared to be weaker than normal (
). Cadherins are a family of Ca2+-dependent transmembrane proteins mediating specific homophilic cell–cell adhesion. Ecad is the major adhesion molecule mediating melanocyte–keratinocyte interactions. Ecad is essential for both the establishment of epithelial structures during development and numerous dynamic processes such as differentiation, polarity, proliferation, and migration during morphogenesis and homeostasis (
Major role of epidermal growth factor receptor and Src kinases in promoting oxidative stress-dependent loss of adhesion and apoptosis in epithelial cells.
), implicating eventually Ecad in vitiligo pathogenesis.
We report that numerous melanocytes in nonlesional areas of the skin in vitiligo patients present membranous Ecad deficiency correlated with mislocalization of melanocytes. We also show that epidermal melanocytes lacking/reducing Ecad are poorly resistant to both mechanical and chemical stresses. The molecular alteration to Ecad in melanocytes might constitute a predictive factor for resistance to stress conditions in vitiligo patients and may help to optimize patient management.
RESULTS
Altered distribution of Ecad at the membrane of melanocytes from vitiligo patients
We studied the distribution of Ecad in epidermal melanocytes in samples from 10 healthy controls and from nonlesional skin distant from the depigmentation macules, mainly the buttock area, in 29 vitiligo patients (Supplementary Table S1 online ). Depigmented skin from vitiligo patients did not contain melanocytes (data not shown). Melanocytes were classified into three types according to the presence and discontinuity of the Ecad labeling: homogeneous (type 1), heterogeneous (type 2) labeling at the cell surface, and no labeling (type 3) (Figure 1a). The percentage of melanocytes showing each of the three types of staining pattern was recorded for each biopsy (Figure 1b and Supplementary Figure S1 online); ∼2,000 melanocytes were analyzed, with a mean of 50 melanocytes per biopsy. All three types of labeling were observed in melanocytes in the epidermis of healthy controls and vitiligo patients. However, a larger proportion of melanocytes in vitiligo patients showed abnormal Ecad labeling at the membrane. In all, 43% of vitiligo melanocytes and 78% of control melanocytes had an homogeneous staining for Ecad and were classified as type 1. Confocal microscopy with immunofluorescence labeling of Ecad was used to confirm these findings and did not show any difference with routine inverted analysis (data not shown). Furthermore, the epidermis was reconstructed in standard medium conditions with melanocytes derived from controls and vitiligo patients; most (82%) vitiligo melanocytes but few (22%) control melanocytes lacked Ecad (Figure 1d and e). β-Catenin, the major cytoplasmic partner of Ecad at the plasma membrane, was similarly discontinuously distributed in, or absent from, melanocyte membranes in vitiligo patients (45 vs. 73% for controls) (Figure 1c and Supplementary Figure S2 online). The amount of nuclear β-catenin was similarly low in melanocytes from vitiligo patients and healthy controls. Thus, the distribution and/or abundance of the cell–cell adhesion molecule, Ecad, and β-catenin were abnormal in clinically normal skin of vitiligo patients. Ecad staining at the membrane of keratinocytes appeared similar in pigmented areas of vitiligo patients and in healthy controls (Figure 2a). It has been established that IL1β is expressed in all epidermal layers of lesional and perilesional vitiligo skin (
). To assess whether altered Ecad in human melanocytes in vivo is associated with the activation of the pro-inflammatory cytokine IL1β, we performed IL1β staining (Supplementary Figure S3a and b online). IL1β expression was similar in nonlesional vitiligo and control skin, suggesting that there is no increased (or reduced) activity of the NLRP inflammasome in nonlesional vitiligo epidermis. However, we detected an infiltration of CD3-positive cells in nonlesional vitiligo epidermis that was not present in normal epidermis (Supplementary Figure S3d and c online).
Figure 1E-cadherin (Ecad) staining is altered in vitiligo melanocytes. (a) Staining of Ecad (green) and Trp2 (red, specific of melanocytes) of the epidermis. A simplified sketch shows the merge. Type 1: Ecad staining is homogeneously distributed in melanocyte; type 2: Ecad staining is heterogeneous; and type 3: Ecad staining is absent. Scale bars=20 μm. The dotted lines represent the basal membrane. Bars=20 μm. (b, c) Mean percentages of control (blue) and vitiligo patient (red) melanocytes displaying type 1, 2, and 3, and types 2 and 3 (2–3) labeled with Ecad and β-catenin (bcat)-specific antibodies. (d) Staining of Ecad (green) and Melan A (red) of reconstructed epidermis made with control keratinocytes and either control (RE mel crtl) or nonlesional vitiligo (RE mel vit) melanocytes. Bars=20 μm. (e) Mean percentages of control (blue) and vitiligo (red) melanocytes displaying Ecad type 1 or type 2–3. **P<0.01, ***P<0.001, ****P<0.0001.
Figure 2The amount of E-cadherin (Ecad) correlates with the numbers of melanocytes of the upper layer of the epidermis. (a) Human epidermis from controls (Ctrl) and vitiligo patients (Vit) labeled with anti-Ecad and Trp2 antibodies and DAPI (4',6-diamidino-2-phenylindole; blue). Basal melanocytes are indicated by white arrowheads and suprabasal melanocytes by white arrows. Bars=20 μm. (b) Numbers of melanocytes (Mc) in control (blue) and nonlesional vitiligo skin sections (red). (c) Inverse correlation between the number of suprabasal melanocytes and Ecad; Pearson's correlation coefficient of -0.56. Red dots indicate vitiligo patients and blue dots indicate controls. (d) Number of keratinocytes between two melanocytes in the epidermis of controls and in nonlesional skin of vitiligo patients. Each individual score is represented as a dot (blue=control, red=vitiligo). On the right, the number of scores of each value is indicated. Means of the number of keratinocytes are represented by horizontal bars: 6±0.3 for controls and 9±0.4 for vitiligo. **P<0.01, ****P<0.0001; NS, nonsignificant.
Alteration of melanocyte distribution in the epidermis of nonlesional skin of vitiligo patients
During the analysis of Ecad in melanocytes in the skin of vitiligo patients, we observed that numerous melanocytes were not located at the basal layer but in the suprabasal layers of the epidermis (Figure 2a). We therefore evaluated melanocyte distribution in the epidermis. Although the total numbers of melanocytes in vitiligo and control epidermis were similar, ∼25% of the melanocytes in vitiligo epidermis were located in the suprabasal layers compared with 5% in controls (Figure 2b). We used anti-cleaved caspase 3 antibody to evaluate the viability of suprabasal melanocytes from 8 vitiligo and 4 control skin biopsies; 46% of suprabasal vitiligo melanocytes but only 9% of the controls were apoptotic (Supplementary Figure S4 online). This indicates that vitiligo melanocytes, once they are detached from the basal membrane, are more prone to cell death. The numbers of suprabasal melanocytes in buttock and thigh skin from four healthy controls and eight vitiligo patients were plotted against the percentage of Ecad type 1 melanocytes (Figure 2c). There was an inverse correlation between the numbers of melanocytes in the upper layer of the epidermis and homogeneous staining of Ecad in melanocytes. These observations suggest that low Ecad abundance in melanocytes correlates with melanocyte detachment from the basal membrane of the epidermis. We analyzed melanocyte distribution in the basal layer of the epidermis in buttock and thigh skin from four healthy controls and four vitiligo patients (Figure 2d). The number of keratinocytes between two basal melanocytes was larger (9.0±0.4) in vitiligo epidermis than in control epidermis (6.0±0.3). This indicates that despite the normal pigmentation in the nonlesional vitiligo biopsies, the ratio between the numbers of melanocytes and keratinocytes was abnormal. Therefore, the epidermal-melanin unit in vitiligo patients is modified, but insufficiently, to have a visible effect on skin pigmentation.
Loss of Ecad in melanocytes reduces proliferation and induces vacuole formation
To assess the function of Ecad in epidermal melanocytes, mice with Ecad-deficient melanocytes (ΔEcad mutants) were generated by mating Tyrosinase-Cre recombinase mice (Tyr::Cre) with Ecad mice harboring loxP sites flanking exons 6–10 (EcadF/F). The invalidation of Ecad in melanocytes has no major consequence on the coat color of the hairy part of the mice; P-cadherin is the major cadherin in follicular melanocytes (Figure 3a). In the tail and paws, most melanocytes are epidermal and Ecad is the main cadherin. In 83% of the mutant mice, the absence of Ecad was associated with lower than normal pigmentation of the tail, with a depigmented tip, and paws. This phenotype is because of the low numbers of epidermal melanocytes in the ΔEcad mutants due to proliferative defect of these cells during development (Figure 4). However, there was no mislocalization of ΔEcad melanocytes within the epidermis: the cells were found in the basal membrane at the epidermal–dermal junction (Figure 3b). Ultrastructural analysis revealed that many ΔEcad mutant melanocytes contained numerous large cytoplasmic vacuoles, not present in wild-type melanocytes (Figure 3c). These vacuolar ΔEcad melanocytes are reminiscent of the vitiligo melanocytes in pigmented skins (Figure 3d) confirming previous findings (
). However, despite these similarities between ΔEcad and vitiligo melanocytes, ΔEcad mutant tails did not develop depigmented macules with age. This may be simply because the life span of mice is too short (a maximum of 2 years): 2/3 of cases of human vitiligo appear during adulthood when they reach 20–30 years of age.
Figure 3Loss of E-cadherin (Ecad) favors murine melanocyte detachment after skin friction. (a) Hair and tail pigmentation in wild-type (WT) mice and mice lacking Ecad in melanocytes (ΔEcad). Dramatic depigmentation zone (arrow). (b) Immunostaining of tail skin of controls and ΔEcad mice in a Dct::LacZ genetic background using anti-Ecad (green) and β-galactosidase (βgal; red) antibodies. Scale bar=10 μm. Basement membrane is indicated as a dotted line. Bars=20 μm. (c, d) Transmission electron micrographs of WT and ΔEcad epidermal mouse melanocytes. Bars=2 μm (c) and human control and vitiligo epidermal melanocytes from nonlesional skin (d). Note the large vacuoles, black arrows, in the cytoplasm of murine ΔEcad and human pre-vitiligo melanocytes. (e) Photographs of WT and ΔEcad tails before (d0) and after repeated friction (d200). Tails were brushed 5 days a week for 28 weeks. Boundary between the brushed (B) and nonbrushed (NB) parts of the tail is indicated. (f) Numbers of melanocytes found in NB and B parts for each genotype. Tail sections 866, 980, 624, and 550 corresponding to WT-B, WT-NB, ΔEcad-B, and ΔEcad-NB, respectively, were generated from three WT and three ΔEcad mice. Histograms represent the ratio between the number of melanocytes found in NB and B parts and that in the NB part for each genotype, as percentages. (g) Location of melanocytes within the epidermis of the B and NB epidermis. Note that some melanocytes were found in the suprabasal epidermis after friction. Bars=20 μm. (h) Percentage of detached melanocytes (Mc). The number of sections per mouse tail and the number of mice are given in f. **P<0.01, ***P<0.001, ****P<0.0001; NS, nonsignificant.
Figure 4Loss of E-cadherin (Ecad) affects melanoblast proliferation during embryogenesis specifically in the epidermis. (a) Transverse sections of wild-type (WT) and ΔEcad tails of p10 (postnatal day 10) mice in a Dct::LacZ genetic background. Melanocytes are identified as β-galactosidase-positive cells. Bars=50 μm. (b) Scatter dot plot reporting the number of melanocytes (Mc) in sections of WT and ΔEcad tails at p10. Each dot represents the average of four sections from the same mouse. Five mice of each group were studied. (c) Number of melanoblasts in the epidermis and the dermis of WT and ΔEcad mouse tails at E15.5. The number of melanoblasts in the epidermis and dermis found in mouse tails at E15.5 was determined for four WT and three ΔEcad embryos. The numbers of melanoblasts in the epidermis, but not in the dermis, were smaller in ΔEcad compared with WT embryos. (d) Proliferation rate of Dct::lacZ-positive cells at E15.5 in the tail, as evaluated by measuring BrdU incorporation as previously described (
). Approximately 70 sections, derived from two to four embryos from independent litters, were analyzed for each genotype. No apoptotic melanoblasts were detected at E15.5 in either mutants or controls (data not shown). *P<0.05, ****P<0.0001; NS, nonsignificant.
). We therefore induced Koebner’s phenomenon on mouse tails. Tails of wild-type and ΔEcad mice were subjected to repeated brushing for up to 30 weeks: the proximal part was brushed, and as control the distal part was not brushed (Figure 3e). Depigmentation was detected within 1 month of brushing and increased with time (Figure 3e and not shown). In ΔEcad mouse tails, the depigmentation correlated with a significant reduction in the numbers of melanocytes (Figure 3f). In brushed tail skin, melanocytes were observed in suprabasal layers; there were three to four times more suprabasal melanocytes in ΔEcad mice than in wild-type mice (Figure 3g and h). Therefore, Ecad appears to be required for the maintenance of melanocytes at the basal membrane and contributes to resistance to mechanical stress applied to the skin.
Oxidative stress affects membranous Ecad and induces transepidermal migration of melanocytes in reconstituted epidermis
) in nonlesional vitiligo and control biopsies using 4-hydroxynonenal (4-HNE) as a marker of lipid peroxidation. We found a strong 4-HNE signal in biopsies from six of seven cases of nonlesional vitiligo, and no signal in biopsies from any of the five controls (Figure 5a and b). In parallel, all these biopsies were stained for Ecad: the amount of Ecad was altered in all nonlesional vitiligo biopsies even in the sample in which lipid peroxidation was not abundant. To evaluate the oxidative stress on Ecad, the nonaggressive human melanoma cell line, MNT1, was exposed to hydrogen peroxide (H202). This leads to the loss of cell–cell contacts as observed by phase contrast analysis (data not shown), resembling an epithelial–mesenchymal transition phenomenon (
). No cell death was observed under these conditions (data not shown). The level of Ecad decreased substantially at the cell membrane (Figure 5c–h). Confocal microscopy confirmed these results (data not shown). We generated reconstructed epidermis with keratinocytes from healthy controls and human melanocytes transduced with a control vector (Ctrl) or with a vector silencing Ecad expression (shEcad). After 2 weeks in culture, the reconstructed epidermis samples were treated with 0.025% H202 for 6 hours before analysis. Western blotting confirmed that there was less Ecad in shEcad cells than the controls (data not shown). The number of melanocytes at the basal layer in reconstructed epidermis after 2 weeks in culture was similar for shEcad-treated and -untreated samples (Figure 5i), and the number of detached melanocytes was similar (Figure 5j). We then evaluated the consequence of a reduction of Ecad combined with oxidative stress on the detachment of melanocytes (Figure 5i and k). The numbers of detached melanocytes was significantly higher in the presence of H202 and a low concentration of Ecad (Figure 5i–k). This result is in agreement with the previously described detachment of vitiligo melanocytes upon H202 treatment (
). In conclusion, the reduction of Ecad in melanocytes and oxidative stress both act on the detachment of melanocytes from the basal membrane.
Figure 5Loss of E-cadherin (Ecad) from the melanocyte membrane and melanocyte detachment in reconstructed human epidermis after exposure to hydrogen peroxide (H202). (a, b) Immunostaining of control (a) and vitiligo (b) biopsies using anti-4-HNE (red) and Trp2 (green) antibodies. DAPI (4',6-diamidino-2-phenylindole; blue). Bars=20 μm. (c–h) Human MNT-1 melanoma cells incubated with (f–h) or without (c–e) 0.025% H202 for 6 hours and labeled with anti-Ecad (green), Melan A (red) antibodies, and DAPI (blue). (c–h) Ecad, (d, g) Ecad+DAPI, (e, h) Ecad+Melan A. Bars=10 μm. (i, j) Percentages of basal (i) and detached (j) melanocytes (Mc) in reconstructed epidermis with normal Sh Ecad (ShEcad, red) or Puro (Ctrl, blue) transduced melanocytes. Reconstructed epidermis was treated or not with 0.025% H202 for 6 hours. (k) Reconstructed epidermis with ShEcad transduced or Ctrl melanocytes incubated with or without 0.025% H202 for 6 hours and labeled with anti-Ecad (green) and melan A antibodies (red). Ecad labeling is weak at the melanocyte membrane in ShEcad samples without H202 (arrows) and in Ctrl melanocytes after H202 treatment. Ecad labeling was totally absent from ShEcad melanocyte-exposed H202 (star). The dotted line indicates the epidermis–dermis junction. Bars=20 μm. *P<0.05; NS, nonsignificant.
Vitiligo is a complex skin disorder involving a chain of events leading ultimately to the loss of melanocytes. We report a thorough analysis of clinically normal pigmented skin of vitiligo patients showing that Ecad, which has a key role for the adhesion between melanocytes and keratinocytes, is abnormally distributed in the membrane. The low Ecad levels in melanocytes are associated with a suprabasal location of melanocytes and a decreased ratio of melanocytes to keratinocytes in the basal layer of the epidermis. In stressed murine melanocytes lacking Ecad and in vitiligo epidermis the detached melanocytes are not adequately replaced in the basal layer, possibly because of an exhaustion of melanocyte renewal, as proposed (
). We suggest that altered Ecad distribution in melanocytes precedes vitiligo pathogenesis but remains silent until melanocyte renewal becomes a limiting factor for normal pigmentation of the skin (Supplementary Figure S5 online ). We also demonstrated that Ecad is required for melanocyte resistance to mechanical and oxidative stress, establishing a link between silent, cell-autonomous defects in vitiligo melanocytes and known environmental stressors accelerating disease expression.
The loss of Ecad from the surface of melanocytes could be due to: (1) a cell lineage–specific underexpression of Ecad in melanocytes of vitiligo patients, (2) a selective downregulation of Ecad at the interface between melanocytes and keratinocytes, or (3) a general reduction of Ecad abundance in all Ecad-expressing cells (including all epithelial cells and, of course, keratinocytes) in vitiligo patients. Consequences of this loss are apparent only in melanocytes because the physiological abundance of Ecad at the cell–cell contact is much lower than in keratinocytes. Correct cadherin function requires appropriate amounts of cadherin–catenin at cell–cell contacts, maintained by a complex balance between synthesis, recycling, and degradation. Ecad can be downregulated by the expression of negative transcriptional regulators (well known in the context of epithelial–mesenchymal transition) and by epigenetic and/or allogenetic regulations (
). However, none of these events have been identified in vitiligo pathophysiology. Similarly, no polymorphisms have been found in the Cdh1 locus or its regulators despite recent genome-wide association studies with a large cohort of patients (
). We confirmed this result with a larger cohort from Iceland (not shown). In the context of vitiligo, the proteins aquaporin (AQP3) and CCN3-DDR1 have been identified as regulating the amount of Ecad at the cell membrane (
). Consistent with the presence of Ecad at the membrane requiring AQP3 and DDR1, the epidermis in lesional skin contains abnormally low levels of all of Ecad, AQP3, and DDR1 (
). DDR1 has a key role in the adhesion of melanocytes to the basal membrane, and it would be informative to evaluate DDR1 expression in nonlesional vitiligo epidermis particularly in melanocytes (
). Possibly, several intrinsic factors, each slightly abnormal in vitiligo patients, and/or with extrinsic factors act in combination or even synergistically affect Ecad expression.
There is mounting evidence implicating oxidative stress in vitiligo pathogenesis (
). Melanocytes are much more sensitive to oxidative stress than keratinocytes that contain larger amounts of antioxidants and can transfer H2O2 to melanocytes (
). High H2O2 concentration destabilizes Ecad and β-catenin complexes at the membrane disrupting cell–cell adhesion. Both these notions are consistent with our observation of low Ecad levels in melanocytes but not keratinocytes in clinically normal skin in vitiligo patients. This possibility is also coherent with the gradual loss of Ecad initially only from melanocytes of normally pigmented skin and subsequently from melanocytes and keratinocytes in perilesional and depigmented skin of vitiligo patients (this paper and (
)). These observations are consistent with the decline of Ecad abundance in melanocytes as an early event in vitiligo pathogenesis. Altogether, one may believe that oxidative stress occurs in melanocytes in which Ecad is already abnormal and that the two phenomena cooperate to the loss of Ecad from the cell membrane and melanocyte detachment from the basal membrane.
A major source of skin stress is frequent exposure to pressure and friction. The mechanisms converting mechanical forces into a biological signal in the skin have not been characterized, but our findings clearly demonstrate that, under such stress conditions, melanocyte attachment to the basal layer is dependent on Ecad expression. Here, we identify a comprehensive molecular mechanism for the melanocytorrhagy theory, based on melanocyte detachment following standardized friction of the skin of vitiligo patients (
). Detached melanocytes in the suprabasal layer might translate various stresses into an immune reaction, e.g., via the heat shock protein HSP70, consistent with the larger numbers of Langherans cells in the vicinity of melanocytes and previous findings (
). In normal pigmented skin of vitiligo patients, the loss of Ecad may generate a damage-associated molecular pattern, providing, in turn, the initiating danger signal that activates innate immunity. It seems unlikely that the innate immunity is activated through the inflammasome production of IL1β because IL1β expression was similar in the epidermis of vitiligo patients and controls (Supplementary Figure S3a and b online); the findings from analysis of the mouse model presented were also consistent with this conclusion (data not shown). However, adaptive immunity, assessed according to the presence of CD3-positive cells infiltrating the dermis, was induced in vitiligo patients but not controls (Supplementary Figure S3c and d online). Nevertheless, the link between altered Ecad expression and the immune component of vitiligo remains to be established.
In conclusion, we identified a functional alteration of cell–cell adhesion mediated by Ecad in vitiligo epidermis. Preliminary experiments reveal similar loss of Ecad from segmental vitiligo melanocytes. This suggests that impaired melanocyte adhesiveness is a common first step to melanocyte loss in all types of vitiligo. As found for filaggrin in another major chronic inflammatory skin disease, atopic dermatitis, a link from a structural defect in the epidermis to an inflammatory sequence can now be plausibly suggested. Indeed, early-onset vitiligo is associated with atopic diathesis (
). Our findings suggest that the magnitude of the Ecad defect in vitiligo may be predictive of resistance to stress conditions and may help patient management if correlated to disease severity. In addition to treatments aiming to stabilizing vitiligo including immune-mediated skin inflammation, an approach to treating this disease also involves strategies to improve melanocyte adhesiveness to the basal membrane.
Materials and Methods
Skin biopsies
Patients provided signed informed consent before the study that was approved by the relevant local ethics committees (Bordeaux CNIL #1545937 v.0 and Rabat Hospitals). The mean age of vitiligo patients and controls was 46 and 43 years, respectively (range 20–67-years for both). The mean duration of the disease for the vitiligo group was 7 years (range 1–20 years). There were no significant differences between controls and patients for age or sex, although the proportion of males was slightly higher in the control group. Punch skin biopsies (4 mm) were obtained from adult patients with generalized vitiligo from clinically normal skin distant from depigmented lesions and from normal adults undergoing plastic surgery (see Supplementary Table S1 online). Skin biopsies were fixed in 4% paraformaldehyde, incubated in 30% sucrose, then in 30% sucrose with 50% optimal cutting temperature compound (OCT), and finally in 100% OCT.
Cell culture and epidermal reconstruction
Exponentially growing melanoma cells were seeded at a density of 106 cells per well in 12-well plates as described previously (
). After 20 hours, the medium was replaced with a fresh medium with or without 0.025% H202 (Sigma, Saint Quentin, France) and samples incubated for 6 hours. For epidermal reconstruction, human control or vitiligo melanocytes and control keratinocytes were cultured from skin samples obtained from adults undergoing plastic surgery or from children’s foreskin obtained at circumcision as described earlier (
). ShEcad (5′-AAGATAGGAGTTCTCTGATGCCTCGAGGCATCAGAGAACTCCTATCTT-3′) was expressed from the lentiviral puromycin vector (pLKO.1–18 bp stuffer and pLKO puro shEcad). Transduced melanocytes were selected by incubation with 7.5 μg ml−1 puromycin for 1 week, and epidermal reconstructions were performed as described previously (
). A total of 38 epidermal reconstructions were carried out including 11 treated with 0.025% H202 for 6 hours.
Microscopy and morphological studies
For immunofluorescence, cells or sections from patient biopsies or epidermal reconstruction were incubated overnight at 4 °C with primary antibody diluted in phosphate-buffered saline/-1% BSA (w/v). Antibodies were specific for Ecad (610182, BD Bioscience, San Jose, CA and M108, Takara, Fisher Scientific Biosciences, St Leon-Rot, Germany), β-catenin (610154, BD Bioscience), Melan A (Dako M7196, Dako, Trappes, France), Trp2 (sc-10452, Santa Cruz, Santa Cruz Biotechnology, Santa Cruz, CA), 4-HNE (HNE11-S, Alpha Diagnostics, Interchim, Montluçon, France), CD3 (Abcam ab699, Paris, France), IL1β (Abcam ab2105), and cleaved-caspase 3 (ASP175, Cell Signaling 9661, Danvers, MA); secondary antibodies were Alexa fluor A31570, A21206, A21208, A21202, A21432, and A21206 (Invitrogen, Carlsbad, CA). Fluorescence photomicrographs were obtained with a Leica-DM-IRB (Nanterre, France) inverted microscope or a TCS-SP5 confocal laser scanning microscope. Slides were read by two/three investigators who were blinded to the results, and the results diverged by between 0.5 and 5%. For transmission electron microscopy, the skin was fixed in 1.5% glutaraldehyde in phosphate-buffered saline, processed for transmission electron microscopy by standard methods, and observed under a Tecnai-12-biotwin transmission electron microscope TEM (Philips Optique Electronique SAS, Limeil Brévannes, France).
Mouse studies
Mice with a conditional deletion of the Ecad gene (Cdh1) were generated by mating Tyr::Cre transgenic mice with animals homozygous for a floxed allele of Ecad (ΔEcad) (
). All mice were backcrossed more than 10 times to C57BL/6 and housed in specific pathogen-free conditions at Institut Curie, conforming to French and European Union legislation. Skin friction was applied with a soft-bristled toothbrush for 2 min per day, 5 days a week for up to 30 weeks.
Melanocyte counts
Slides were scanned with an Olympus IX51 and analyzed with Olyvia software (Olympus, Rungis, France). Melanocytes in basal or suprabasal (spinous, granular layers, and stratum corneum) layers were counted in five independent skin sections and normalized per mm2. The number of keratinocytes between two melanocytes was determined in five skin sections for each patient. A minimum of 2,220 basal keratinocytes were analyzed for each group. For mouse skin, between 139 and 205 counts were performed for each genotype and condition.
Statistical analysis
Graphpad Prism 6 software (La Jolla, CA) was used for statistical analyses. Ordinary one-way analysis of variance and unpaired t-tests were used to assess the significance of differences (****P<0.0001, ***P<0.001, **P<0.01, and *P<0.05; NS, nonsignificant).
Acknowledgments
We thank the staff of the animal colony (Y Bourgeois, JD Mam, and H Harmange) and imaging facilities (F Cordelières) of the Institut Curie. We are grateful to Alfonso Bellacosa (FCCC, Philadelphia) for critically reading the manuscript. RYW is recipient of a fellowship from the MENRT. RYW was a recipient of MENRT and ARC. FL was recipient of fellowships from the MENRT, ARC, Ligue contre le cancer (Oise), and Société Française de Dermatologie (SFD). This work was supported by the Ligue Nationale Contre le Cancer (Equipe labellisée EL2012.LNCC/LL), INCa (2011-1-PL BIO-03-IC-1), and Labex CelTisPhyBio (ANR-11-LBX-0038).
Major role of epidermal growth factor receptor and Src kinases in promoting oxidative stress-dependent loss of adhesion and apoptosis in epithelial cells.
In vivo and in vitro evidence for hydrogen peroxide (H2O2) accumulation in the epidermis of patients with vitiligo and its successful removal by a UVB-activated pseudocatalase.
37319-X&id=gr1.jpg){kind=link}
37319-X&id=gr2.jpg){kind=link}
37319-X&id=gr3.jpg){kind=link}
37319-X&id=gr4.jpg){kind=link}
37319-X&id=gr5.jpg){kind=link}