Cell surface glycans in the human stratum corneum: distribution and depth‐related changes

During the formation of the stratum corneum (SC) barrier, the extracellular spaces of viable epidermis, rich in glycans, are filled with a highly organized lipid matrix and the plasma membranes of keratinocytes are replaced by cornified lipid envelopes. These structures comprise cross‐linked proteins, including transmembrane glycoproteins and proteoglycans, covalently bound to a monolayer of cell surface ceramides. Little is known about the presence and distribution of glycans on the SC corneocytes despite their possible involvement in SC hydration, cohesion and desquamation. In this work, we visualized ultrastructurally and quantified the distribution of glycans on the surface of native and delipidated corneocytes. The cells were harvested at different depths of the SC, allowing us to define the relationship between the distribution of various glycans, proteoglycans and glycoproteins, and other changes occurring in SC. At the cell periphery, we found a correlation between the depth‐related alterations of corneodesmosome glycoproteins and α‐d‐mannosyl and N‐acetyl‐d‐glucosamine‐labelling patterns. Elimination of the terminal sugars, α‐linked fucose and α‐(2,3) linked sialic acid, was less abrupt, but also the initial extent of their peripheral distribution was overall lower than that of concanavalin A and wheat germ agglutinin lectin‐detected glycans. Diffuse labelling of heparan sulphate glycosaminoglycans disappeared completely from the outermost corneocytes, whereas that of several simple carbohydrates could be detected at all SC levels. Our results suggest that specific glycan distribution may participate in the progressive changes of SC, as it evolves from the SC compactum to the SC disjunctum, towards desquamation.

keratinocytes and recovered by a 5-nm-thick lipid envelope. Elements composing CE, including keratinocyte transmembrane proteins, are cross-linked via covalent bonds by transglutaminases 1, 3 and 5. 3 SC may be considered as a depositary of the epidermis' past, as it retains traces of the tissue activity preceding cornification. 4 The extracellular spaces of viable epidermis, rich in glycans and notably in hyaluronic acid, are filled upon cornification with lipids organized in several molecular layers. This hydrophobic matrix highly contributes to the relative impermeability of the SC barrier. 1,5 Despite the rich literature concerning glycan expression on the surface of viable keratinocytes, the presence and distribution of glycans on the SC corneocytes are scarcely studied. [6][7][8][9][10][11] Yet, sugar moieties may play an important functional role in the SC formation, cohesion and desquamation. [12][13][14] On the other hand, deglycosylation of glucosyl ceramides is essential for the SC lipid maturation and permeability barrier formation. 15 Studies dealing with the identification of epidermal glycans used lectin-labelling technique and showed the presence of Nacetyld-glucosamine, sialic acid, αd-mannosyl, αd-glucosyl and d-galactosyl terminal sugars in the basal layer of the epidermis and N-acetylgalactosamine in the upper epidermal layers. 11,[16][17][18] Several epidermal proteoglycans bearing heparan sulphate glycosaminoglycans (GAGs), such as syndecans, epican, perlecan and glypican, many of them showing differentiation-dependent expression, were detected in human epidermis. 14 In their vast majority, these studies indicate a drastic reduction of these labellings within SC. However, such reduction was reported less pronounced in the case of in vitro-reconstructed epidermis. 6,7,11 Moreover, detection of cell surface carbohydrates is largely altered in disorders of epidermal proliferation, characterized by incompletely differentiated SC. [19][20][21][22] Various hypotheses may explain these observations. (i) The absence of glycan moieties in the SC results from a considerable glycosidase activity at the bottom of the SC. 23 For example, βd-glucosidase that transforms glucosyl ceramides into ceramides and heparinase 1 that degrades heparan sulphates 24 are both present at the interface between the living and cornified epidermal layers. (ii) Extracellular lipid matrix that covers corneocyte surface masks the glycan moieties and results in a false-negative outcome of the in situ detection. 25 (iii) The occurrence of technical artifacts or bias cannot be excluded. 26 In the present study, we aimed at addressing this issue using an ex vivo approach, in which native and delipidated samples of normal human SC have been examined. The presence and distribution of various glycan moieties on corneocytes and changes in the glycan expression along the consecutive phases of SC maturation are herein described.

| Skin samples
Normal abdominal human skin from anonymous healthy female donors (aged between 20 and 50 years) was obtained during plastic surgery procedures according to the French regulations and Declaration of Helsinki act. A written consent was obtained from the patients.

| Tape stripping
Strandard D-squame self-adhesive discs (CuDerm Corporation, Dallas, TX, USA) were applied onto the excised skin with a constant pressure for 10 seconds, then removed and stored in hermetic vials at −80°C. At least twenty consecutive tape strippings were performed on the same zone by the same operator to reduce variability. Tape strippings were also performed on delipidated skin. To this end, before each stripping step, a plastic ring was placed on the top of a sample and filled with a 1:1 mixture of chloroform/ether (v/v), left for 5 minutes before being removed, and the freshly delipidated surface was then tape-stripped.

| Isolation of epidermal sheets and their delipidation
Superficial skin strips were obtained from the surgically excised tissue using a dermatome set at the depth of 0.8 mm. The epidermis was dissociated from dermis by placing the samples, dermal side down, into 2.5 mg/mL dispase in PBS (Gibco, Paisley, UK) for 4 hours at 37°C. Fragments of the isolated epidermis were divided into two lots. One was quickly washed in PBS and stored at −80°C. The other part was delipidated by immersion of the samples in a mixture of ether/chloroform (1:1) for 4 hours under constant stirring. Delipidated epidermis samples were also stored at −80°C. Isolated epidermal sheets were used for pre-embedding detection of glycans using histochemical methods.

| Lectins
Colloidal gold-labelled lectins (10-nm granules) used in this study are as follows: Concanavalin A (Con A) that recognizes αd-mannosyl and αd-glucosyl monosaccharides, Maackia amurensis (MAA) that recog-  (i) 10 mU of heparinase III in 100 mmol/L sodium acetate, 0.5 mmol/L calcium acetate buffer, pH 7.1, at 37°C for 24 hours; (ii) 500 mU of chondroitinase ABC in 50 mmol/L Tris-HCl buffer, pH 7.5, containing 50 mmol/L NaCl, 2 mmol/L CaCl 2 and 0.01% bovine serum albumin, at 37°C for 24 hours; (iii) 500 mU of hyaluronidase in 50 mmol/L sodium acetate buffer, pH 5, containing 75 mmol/L NaCl, at 37°C for 24 hours; (iv) consecutive digestions with hyaluronidase, as described, followed by a 24 hours digestion at 37°C with a mixture of 10 mU of heparinase III and 500 mU of chondroitinase ABC in 100 mmol/L sodium acetate, 0.5 mmol/L calcium acetate buffer, pH 7.1. All the enzymes and buffers were purchased from Sigma. The presence of the RHT staining was evaluated on ultrathin tissue sections routinely counterstained with uranyl acetate.

| Pre-embedding labelling with lectins and antibodies for transmission electron microscopy
Double-labellings with colloidal gold-labelled lectins (a direct labelling) and with antibodies followed by appropriate immunogold conjugates (indirect immunolabelling) were performed on native and delipidated D-squame strips. After labelling, the samples were fixed in 2% glutaraldehyde, washed, dehydrated in graded ethanol series and embedded in epoxy resins. Ultrathin cross-sections were counterstained with uranyl acetate before transmission electron microscopy (TEM) examination.  The results were expressed as a mean width of the marginal zone ± standard deviation (SD) and compared using ANOVA oneway test.

| Glycans at the corneocyte surface can be detected with RHT after SC delipidation
Ruthenium hexamine trichloride failed to stain corneocytes in native isolated epidermis (data not shown). Only after delipidation, a thin layer of RHT staining material appeared on the free surface of corneocytes (Fig. 1). The stain presented an approximately 10-nm-thick, deep-dark, uniform line superimposed over the lipid envelopes. The "cores" of corneodesmosomes exposed through mechanical disruption were also coloured ( Fig. 1a).
Ruthenium hexamine trichloride staining profile changed when the chemical was applied onto the samples predigested with enzymes targeting GAG. After digesting for 24 hours with heparinase III or chondroitinase ABC, that degrade heparan and chondroitin sulphates, a disruption of the cell coat stained with RHT was observed (Fig. 1b,   c). No major change of the surface staining pattern was noticed after hyaluronidase digestion (not shown). Only after consecutive digestions with hyaluronidase followed by heparinase III and chondroitinase ABC, RHT stain on exposed corneocyte surfaces was absent (Fig. 1d).

| Different glycan moieties are present on superficial corneocytes of delipidated SC in various distribution patterns
In the light of results obtained with RHT staining, the cytochemical studies of corneocytes were performed on delipidated tissues.
Antibodies that recognize heparan sulphate and chondroitin sulphate GAG, as well as lectins recognizing specific sugar moieties, showed particular SEM labelling profiles on delipidated corneocytes from the SC disjunctum (Table S1). Each marker gave a reproducible labelling profile on various donors.
Antibodies to the protein cores of CD44, syndecan 1 and desmosealin proteoglycans, showed a very weak labelling dispersed at the corneocyte surface, with a slight tendency to cumulate at the cell margins. Labelling with the anti-heparan sulphate GAG antibody also followed this pattern, whereas antibody to chondroitin sulphate GAG and HABP failed to label superficial corneocyte surfaces (Fig. S1).
Antibodies to corneodesmosin and to the extracellular portion of desmoglein 1 showed a patchy labelling pattern concentrated at the corneocyte margins. The peripheral clusters visualized with the above-mentioned antibodies were mostly ovoid in shape with a longer diameter of 335±30 nm and a shorter one of 229±37 nm.
Labelling of the cell margins obtained with Con A, MAA, PWM, UEA and WGA was highly reminiscent of that detected with antibodies to corneodesmosome proteins (Figs 2 and S1). However, Con A, PWM and WGA showed an additional diffuse labelling dispersed over the cell central plate of the cells (Fig. 2c). With PNA and SNA lectins, the labelling was absent or very weak and uniformly distrib- whereas the central plate labelling was independent of any defined morphological features (Fig. 3a, b).
Enzymatic deglycosylation of the corneocyte surfaces with Sigma-Aldrich kit made impossible any subsequent lectin labelling, although it did not affect the detection of corneodesmosin. The use of lectin inhibitors considerably reduced the peripheral labelling intensity without completely blocking the central plate labelling (Fig. S2).

| Changes in the labelling profiles of different markers at various SC depths
In situ delipidation, performed on excised skin samples, resulted in a diminished number of D-squame strips that we were able to obtain (10.3±2; mean±SD; n=8), when compared to the total number of strips necessary to completely eliminate the native SC (22±1.7).
Instead, an increased number of corneocyte layers were harvested with each D-squame strip of the delipidated tissue.
Tape strips number two, n/2, and n, where n is the total number of strips collected from a given sample, were selected for labelling ( Fig. S3). The most striking depth-related feature observed was the statistically significant decrease in the corneocyte margin labelling with lectins and antibodies, accompanying the displacement of the cells towards the SC surface (Fig. 4). In this respect, measurements of the peripheral labelling width showed that the dynamics of corneodesmosin, Con A and WGA elimination was similar, with decreasing linear trend curves' slopes at 0.98, 1.11 and 1.03, respectively.
Compared to that, corneocyte margins labelled with UEA and MAA lectins that recognize terminal sugars were thinner in the deepest SC but presented "slower" decreasing slopes at 0.47 and 0.52, respectively (Fig. 4). Heparan sulphate GAG labelling was weaker than that The dispersed distribution pattern of the proteoglycans' protein cores was observed at all SC depths, although the labelling was more intense at the surface of the deepest corneocytes (data not shown).

| DISCUSSION
Ruthenium hexamine trichloride staining procedure, which is a fairly specific technique to detect cell surface GAGs, 27 did not modify the baseline RHT staining pattern, indicating that the dye targeted primarily cell surface GAGs other than HA.
However, the successive actions of hyaluronidase followed by heparinase III and chondroitinase ABC allowed for the complete elimination of the cell surface carbohydrates stained with RHT.
Our attempts to ultrastructurally localize HA using HABP failed to reveal this GAG either in the extracellular spaces of SC or at the surface of native and delipidated corneocytes (data not shown).
HA is mostly processed by hyaluronidase 1 during the transition between SG and SC 30 but is not completely degraded. In fact, Sakai et al. 31 reported that the chain length of SC HA was 100 times lower then epidermal HA chains and that the binding activity of HABP to the shorter chains was dramatically reduced.
To precise the distribution of different glycans at the surface of corneocytes, we used the labelling technique on tape strips adapted for SEM. Cell surface delipidation before each consecutive tape stripping allowed us to label the exposed surfaces of corneocytes GAGs is their interaction with other carbohydrates and proteins. 33 Here, the potential partners within the extracellular spaces could be cell-cell junction glycoproteins and hydrophilic molecules involved in the SC processing. The degradation of corneodesmosomes is essentially under the control of two families of proteases, kallikreins and cathepsins, 34,35 and their inhibitors. [36][37][38] They are synthetized in the living epidermis and delivered at the SG/SC interface via the lamellar bodies, together with the other hydrolytic enzymes including glycosidases, such as fucosidase α, mannosidase α and glucuronidase β. 39 However, it remains unclear where are these players, essential for processing of the SC lipids, saccharides, and structural proteins, located within the highly hydrophobic extracellular environment. We have hypothesized that proteinaceous components delivered into the intercorneocyte compartment become sequestered within the hydrophilic poaches present between the SC extracellular lipids. 40 In the SC compactum, lipids show a compact orthorhombic lateral organization, which may limit the accessibility of enzymes to their respective substrates. In the SC disjunctum, the less compact hexagonal lipid organization may facilitate the lateral displacement of hydrophilic poaches and, thus, the access of enzymes to corneodesmosomes. 40,41 Taking this into consideration, the extracellular space of SC cannot be considered as a single entity; it is a mosaic assembly of different micro-domains where the spatial organization of its components reflects its functionality.
Double-labelling with lectins and anti-corneodesmosin antibody proved that the peripheral clusters of lectin labelling corresponded to corneodesmosomes and that the central plate labelling was not related to any defined adhesion structure. The width of glycans' peripheral labelling significantly decreased during the displacement of corneocytes from the bottom to the top of SC. However, elimination of the terminal glycans (fucose and sialic acid) began already in the SC compactum, well before the elimination of mannose and Nacetylglucosamine. These observations could be related to the optimum pH of the glycosidases known to be delivered by the lamellar bodies. 39 Indeed, fucosidase is active at neutral pH, typical for the lowermost SC, whereas hexosaminidase and mannosidase require an acidic optimum pH that is only found in the upper SC. 42 Thus, our findings suggest the presence and activity of those glycosidases in the SC extracellular spaces.
The dynamics of corneodesmosome degradation deduced from the regression of the cell margins labelled with corneodesmosin and desmoglein 1, appeared to be concomitant with the reduction of mannose and N-acetyl-glucosamine peripheral labelling. Implication of glycans in the protection of corneodesmosomes from proteolysis was suggested by Walsh and Chapman in the early 1990s, but harsh experimental conditions of the study were very far from the physiological situation. 13 Although we have observed a striking parallelism between the glycan disappearance from the corneocyte surface and corneodesmosome degradation, our results do not permit to establish whether deglycosylation is a prerequisite to proteolysis of the junctions. Nevertheless, we can confirm the previously published reports indicating a progressive degradation of corneodesmosomes starting first at the central plate area of corneocytes. [43][44][45] The corneocyte central plate labelled with Con A, WGA and PWM did not evolve during the maturation of SC. We suppose that those glycan moieties could be remnants of glycoproteins that subsisted enzymatic degradation. Brysk et al. 12 developed an in vitro assay in which delipidated but not native corneocytes were shown to aggregate. As this phenomenon was abolished by pretreatment of delipidated corneocytes with lectins, the authors concluded that cell surface sugars participate in the cell-cell adhesion process in the absence of the lipid spacer. These observations remain in agreement with our present results, as Con A, WGA and PWM lectins as well as proteoglycan and heparan sulphate GAG labelling covered the entire surface of delipidated corneocytes. In fact, an increased corneocyte cohesion can also be observed in situ, after a complete delipidation of the SC, when cornified lipid envelopes are shown to stick to one another. 46,47 Our results indicate that glycans do persist on the cell surfaces in the SC and are mostly concentrated in corneodesmosomes. Provided the importance of glycans for tissue hydration and their potential protective role against premature proteolysis, we suggest that these sugar moieties may participate in the maintenance of SC barrier function and in the regulation of SC desquamation.

SUPPORTING INFORMATION
Additional Supporting Information may be found online in the supporting information tab for this article.