Aquaporin-3 in the epidermis: more than skin deep

Abstract

The skin is essential for terrestrial life. It is responsible for regulating water permeability and functions as a mechanical barrier that protects against environmental insults such as microbial infection, ultraviolet light, injury, and heat and cold, which could damage the cells of the body and compromise survival of the organism. This barrier is provided by the outer layer, the epidermis, which is composed predominantly of keratinocytes; keratinocytes undergo a program of differentiation to form the stratum corneum comprising the cornified squame “bricks” and lipid “mortar.” Dysregulation of this differentiation program can result in skin diseases, including psoriasis and nonmelanoma skin cancers, among others. Accumulating evidence in the literature indicates that the water-, glycerol-, and hydrogen peroxide-transporting channel aquaporin-3 (AQP3) plays a key role in various processes involved in keratinocyte function, and abnormalities in this channel have been observed in several human skin diseases. Here, we discuss the data linking AQP3 to keratinocyte proliferation, migration, differentiation, and survival as well as its role in skin properties and functions like hydration, water retention, wound healing, and barrier repair. We also discuss the mechanisms regulating AQP3 levels, localization, and function and the anomalies in AQP3 that are associated with various skin diseases.

INTRODUCTION

The water channel aquaporin-3 (AQP3) is the most abundant aquaporin in the outer epithelial layer of the skin, the epidermis [reviewed in (62)], although it is not the only such aquaporin; there is accumulating evidence that this channel plays a key role in the epidermis. Therefore, knowledge about the mechanisms regulating its expression and function is necessary for a complete understanding of epidermal and skin function. However, the exact role of AQP3 in the epidermis is not entirely clear, as it is likely involved in multiple aspects of skin biology. After introducing the epidermis and its structure and discussing the mechanisms that determine AQP3 transport activity, this review will discuss the data supporting AQP3’s involvement in various aspects of epidermal and skin function as well as in several skin diseases.

SKIN, EPIDERMAL STRUCTURE, AND THE BURDEN OF SKIN DISEASE

The skin is the body’s first line of defense against the external environment. For this reason, the skin is responsible for providing a physical and water-impermeable barrier for the body as it responds to a variety of stresses such as sunlight, microorganisms, or physical trauma (70). The skin is composed of three distinct layers: the hypodermis, the dermis, and the epidermis. The hypodermis is the deepest layer of skin that is mainly comprised of adipose tissue and is used for fat storage and insulation of the body. The dermis is located in between the epidermis and hypodermis and consists of connective tissue composed of extracellular matrix and fibroblasts; the dermis also contains blood vessels, nerves, hair follicles, and sweat glands, among other features. Finally, the epidermis, the outermost layer of the skin, is comprised of a variety of cell types, such as melanocytes, Langerhans cells, Merkel cells, and keratinocytes. Melanocytes synthesize the pigment melanin to protect the skin from ultraviolet damage, which results from absorption of light energy by cellular chromophores, leading to oxidative stress and possible DNA mutations (69). Langerhans cells are part of the skin’s immune system, serving as antigen-presenting dendritic cells, and Merkel cells assist in tactile sensation by the skin (47). However, keratinocytes, which form the skin barrier, are the predominant cell type found in the epidermis, constituting approximately 90% of the epidermal cells. These cells form the four layers of the epidermis starting with the stratum basale, the deepest layer of the epidermis, secured into position via hemidesmosomes adhering to a collagenous basement membrane. The next layer is termed the stratum spinosum, followed by the stratum granulosum and the most superficial layer, the stratum corneum (Fig. 1).

Fig. 1.

The epidermis. Shown is a schematic of normal epidermis, illustrating its various layers. These include the basal layer, or stratum basale, that attaches to the basement membrane at the junction of the epidermis with the dermis, the spinous layer, or stratum spinosum, which often comprises several cell layers particularly in humans, the granular layer or stratum granulosum and the cornified layer or stratum corneum, with the outermost layer exposed to the environment.

The layers of the epidermis are formed via proliferation of keratinocytes in the stratum basale, followed by migration of these daughter cells outwards towards the stratum spinosum, stratum granulosum, and stratum corneum, during which process they undergo a differentiation program to result in their final maturation. This differentiation process is extremely important for normal epidermal turnover and function, as well as wound healing, and is initiated by basal progenitor cells in the stratum basale. Basal progenitor cells have significant mitotic ability, allowing continuous proliferation of the basal layer to replace cells lost to the environment (45). In addition to characteristics of proliferation such as mitotic figures and DNA synthesis, this layer also expresses the immature keratins, keratin 5 and 14, which provide mechanical stability to the cells (45). The newly synthesized cells undergo differentiation and migrate upward into the stratum spinosum epidermal layer (45). While basal cells express keratins 5 and 14, spinous cells express mature keratins 1 and 10 (45). In addition, there is an upregulation of desmosomes, as well as increased keratin production, which results in increased cell-to-cell adhesion in the stratum spinosum, giving this layer its namesake spine-like histological appearance (56). As cells differentiate and move into the stratum granulosum layer, they continue to increase keratin production while also beginning to take on a more flattened structure (56). The granular cells are characterized by their keratohyalin granules (56), which consist of keratins, profilaggrin, loricrin, and other cornified envelope proteins that play an important role in the formation of the stratum corneum (45). The contents of these stratum granulosum granules are cross-linked by transglutaminase to the plasma membrane generating the tough cornified envelopes that characterize the squames, which together with lipids secreted in the form of lamellar bodies (see below) form a functional barrier to the external environment, the stratum corneum (45). The stratum corneum is comprised of tightly packed, large, flat, nonviable cells arranged in layers that can range from 15 to 100 layers thick, depending on the location of the skin on the body (56). Granular cells also contain lamellar bodies comprising polar lipids, glycosphingolipids, free sterols, phospholipids, and hydrolytic enzymes (45). The contents of these lamellar bodies are secreted as the cell moves into the stratum corneum; the lipids are then metabolized and organized to form the water permeability barrier that is so important for terrestrial life.

In 2013, more than 85 million Americans were seen by a medical professional for skin diseases (53). The United States healthcare system alone is estimated to spend approximately 75 billion dollars annually to treat skin diseases (53). Additionally, the prevalence of skin disease increases to almost 50% in Americans 65 years of age or older (53). Moreover, it is estimated that the population over the age of 65 will almost double in the United States in the next 30 years, leading to an inevitable increase in annual healthcare costs for skin disease treatment (29). Therefore, it is essential to develop better therapies for skin diseases to decrease annual spending of healthcare systems on this issue. Skin diseases such as psoriasis and nonmelanoma skin cancer are characterized by hyperproliferation and abnormal differentiation of epidermal keratinocytes, making an understanding of the regulation of these processes necessary to develop better treatments for and decrease annual healthcare costs for skin-related diseases. As will be discussed in subsequent sections, accumulating data in the literature point to a likely key role for the water channel AQP3 in various skin disorders [reviewed in (63) and see below], suggesting the importance of information about how the localization and function of this channel are regulated for the development of effective treatment strategies for skin pathologies.

REGULATION OF AQP3’S EXPRESSION, LOCALIZATION, AND FUNCTION

AQP3 is a member of the aquaporin family of (to date) 13 water channels (AQP0–AQP12); aquaporins span the membrane six times to form a pore that permits selective permeation of water. The selectivity filter is thought to be formed by three amino acids, asparagine, proline, and alanine (i.e., the NPA motifs), in loops B and E between the second and third and the fifth and sixth transmembrane domains, respectively. AQP3, however, belongs to the subfamily of aquaglyceroporins that transport small molecules in addition to water. Specifically, AQP3 has been shown to transport water, glycerol, and hydrogen peroxide (17, 18, 32, 33, 59, 81). In addition, AQP3 is expressed throughout the body, including in the kidney collecting duct and, as discussed below, in the skin.

AQP3 is expressed in keratinocytes in the basal layer, where it can be found in the plasma membrane as well as in the intracellular compartment. In keratinocytes of the stratum spinosum, AQP3 is localized almost exclusively to the plasma membrane, outlining the periphery of each cell in the multiple layers comprising the human stratum spinosum (Fig. 2). This in situ expression is consistent with the finding in vitro that strongly driving basal keratinocytes to late differentiation results in downregulation of AQP3 mRNA levels (85), but providing a more physiological differentiation stimulus allowing good expression of early differentiation markers also results in (glycosylated) AQP3 protein expression (63).

Fig. 2.

Aquaporin-3 (AQP3) immunoreactivity in the epidermis. Shown is immunohistochemical analysis of AQP3 staining in the epidermis of two different normal human subjects. Skin tissue collected as a result of breast reduction or abdominoplasty surgery was formalin-fixed and paraffin-embedded; these tissue blocks were obtained from the pathology archives after a determination of non-human subjects research was received from the Augusta University Institutional Review Board. Sections were stained using an antibody recognizing AQP3 (Novus Biologicals, Littleton, CO) and counterstained with hematoxylin, as described previously (78). Note that in the basal layer, AQP3 (brown staining) is detected both in the plasma membrane and in an intracellular compartment, whereas in the spinous layer, AQP3 stains almost exclusively in the plasma membrane. Scale bars represent 20 µm.

The mechanisms regulating this expression pattern in keratinocytes and the epidermis are beginning to be defined. For example, we have recently shown that AQP3 expression can be induced by inhibition of histone deacetylases (HDACs) in keratinocytes and mouse skin ex vivo (12). HDACs constitute a family of enzymes represented by type I (HDAC1, -2, -3, and -8), type II (HDAC4, -5, -6, -7, -9, and -10), type III (sirtuins 1 through 7), and type IV (HDAC11), which remove acetyl groups added by histone acetyltransferases (HATs) from histones and other proteins. Lysine acetylation of histones generally results in a relaxed chromatin structure because of reduced histone association with DNA, leading to enhanced gene transcription. Thus, by increasing histone acetylation, HDAC inhibitors can promote gene expression. In addition to their effects on histones, HATs and HDACs also can regulate the lysine acetylation of nonhistone proteins, including transcription factors, and this posttranslational modification can often alter the activity of these proteins. Given the broad substrate specificity of HATs and HDACs, it has been proposed that these enzymes be renamed lysine acetyltransferases and lysine deacetylases. In our recent study, we identified that in keratinocytes HDAC3 in particular is involved in regulating AQP3 expression, such that selective HDAC3 inhibition increases AQP3 mRNA and protein levels (12). This result suggests that under basal conditions HDAC3 suppresses AQP3 expression in epidermal keratinocytes. Indeed, overexpression of HDAC3 was found to reduce, and knockdown of HDAC3 to enhance, HDAC inhibitor-stimulated AQP3 mRNA and protein levels (12). However, the mechanism by which HDAC3 inhibition increases AQP3 expression is not entirely clear.

One group of transcription factors shown to be involved in upregulating AQP3 is the p53 family. Indeed, the AQP3 promoter possesses a conserved p53 response element that can also be bound and activated by p73 (86). In addition, pfithrin, an inhibitor of p53 and p73, has been shown to inhibit, and p53 overexpression to enhance, HDAC inhibitor-induced AQP3 expression (12). These data provide evidence of a role for p53 transcription factors in increasing AQP3, although it is not yet clear whether the enhanced AQP3 expression in response to HDAC inhibition is the result of a generalized increase in gene transcription via effects on chromatin structure, through stimulation of the activity of one or more transcription factors including p53 or another family member [perhaps p63 as in head and neck squamous cell carcinoma (27)], or a combination of these processes.

Peroxisome proliferator-activated receptors (PPARs) are nuclear hormone receptors that are involved in the regulation of differentiation, development, and metabolism; they also play a key role in regulating AQP3 expression. Feingold and colleagues (43) demonstrated that agonists of PPARγ and PPARδ, but not PPARα, upregulated AQP3 mRNA and protein levels in human keratinocytes; topical application of the PPARγ agonist also increased AQP3 expression in mouse skin in vivo. Activators of several other nuclear hormone receptors, including agonists of liver X receptors (LXR), retinoic acid receptors (RAR), and retinoid X receptors (RXR), also exerted similar effects on AQP3 expression (43). In agreement with these findings, we reported that PPARs mediate, at least in part, the HDAC inhibitor-induced elevation in AQP3 levels; PPAR antagonists prevent the HDAC inhibitor-dependent increase in AQP3 expression (82). In addition, PPARγ overexpression enhances the HDAC inhibition-stimulated AQP3 mRNA and protein levels (82). Again, however, whether HDAC inhibition is simply increasing the accessibility of the AQP3 gene promoter to the transcription machinery or modifying PPAR acetylation to stimulate its transcriptional activity is at present unknown.

AQP3 itself is also posttranslationally modified, and it seems possible that these posttranslational modifications might affect AQP3’s function and/or localization. For example, AQP3 is known to be glycosylated, resulting in an approximately 28 kDa unglycosylated form and an approximately 40 kDa glycosylated protein. AQP3 has been previously shown to be N-glycosylated in rat kidney (3), although this has not yet been confirmed in epidermal keratinocytes. It has also been reported that AQP3 possesses a single possible N-linked glycosylation site at asparagine no. 141 (15), but again whether this is the amino acid glycosylated in keratinocytes has not been studied to date. Although effects of posttranslational modifications on AQP3 localization have not yet been investigated, by analogy with aquaporin-2 (AQP2) (40), proper glycosylation may be required for membrane localization. Studies to examine this hypothesis are obviously needed.

We have previously shown that in mouse keratinocytes AQP3 protein levels and mRNA expression are decreased by differentiating agents, including 1,25-dihydroxyvitamin D₃ and high-calcium-containing (1 mM) medium (85). These decreased levels are accompanied by a reduction in glycerol uptake. However, glycerol uptake is increased by a moderate calcium-containing medium (125 µM calcium) (87) known to promote a more complete program of keratinocyte differentiation (84). We subsequently observed that this 125 µM calcium concentration decreased the non-glycosylated approximately 28 kDa form of AQP3 but increased the glycosylated, roughly 40 kDa AQP3 (63). If glycosylation promotes plasma membrane localization as it does for AQP2 (40), this result is consistent with the observed distinct membrane localization of AQP3 in the suprabasal layers of the skin in human (78) and mouse (63) skin in situ. On the other hand, Feingold and colleagues (43) have reported essentially no effect of elevated Ca²⁺ on AQP3 protein levels in human keratinocytes. Nevertheless, these authors demonstrated that the effect of PPAR agonists to significantly increase AQP3 levels was limited to the glycosylated form of the protein (43), suggesting that also in human keratinocytes differentiating agents increase glycosylated AQP3 levels.

Phosphorylation may also be a posttranslational modification that regulates AQP3’s localization and/or function, again by analogy with AQP2. In particular, AQP2 trafficking to the apical membrane in cells of the kidney is regulated through signal transduction-mediated changes in vesicle trafficking [reviewed in (75)]; AQP3 has also been reported to translocate to the plasma membrane upon exposure of keratinocytes to osmotic stress (24). Signaling pathways involving cAMP/protein kinase A (PKA), Ca²⁺/calmodulin, protein kinase C, Akt, and mitogen-activated protein kinases have been reported to regulate this trafficking of AQP2 (38, 44, 51, 76). Mutation of serine 256 in AQP2 to unphosphorylatable alanine inhibits, and to phosphorylation-mimicking aspartate enhances, basal AQP2 apical membrane localization (76), suggesting the likelihood that phosphorylation regulates this process. On the other hand, a basolateral targeting sequence (YLLR) has also been reported in AQP3 from multiple species (15), so there may be mechanisms other than posttranslational modifications that can regulate AQP3 plasma membrane localization as well.

A final posttranslational modification that may affect AQP3’s localization or function is lysine acetylation. In the kidney, AQP3 was shown to be acetylated on lysine 282 (41), although the histone acetyltransferase responsible for adding this acetyl group and the HDAC mediating its removal were not determined. As discussed above, we have previously shown that HDAC inhibitors can induce AQP3 expression, enhancing mRNA levels as well as the levels of both the unglycosylated and glycosylated AQP3 protein (12). Interestingly, certain HDAC inhibitors seemed to elicit a greater increase in glycosylated AQP3 protein than was expected from their effect on unglycosylated protein levels (12), suggesting the possibility that the promotion of lysine acetylation by HDAC inhibition might exert actions on AQP3 in addition to its effects on expression. Clearly, additional studies are warranted to define the posttranslational modifications of AQP3 and the role of these modifications on the channel’s localization and function.

AQP3’S ROLE IN KERATINOCYTE PROLIFERATION AND MIGRATION

A series of elegant studies by Verkman and colleagues (30, 34–36, 60) have suggested an important role for AQP3 in keratinocyte proliferation, although perhaps not under basal conditions in vivo (30), since in the global AQP3-knockout mouse model, Hara et al. (30) demonstrated similar epidermal thickness and layer number as well as proliferation (as measured by radiolabeled thymidine incorporation into DNA) compared with wild-type mice. However, this group demonstrated that AQP3-knockout mice exhibit reduced tumor formation in a two-stage mouse model of carcinogenesis (36), in which topical application of a carcinogen is followed by administration of a tumor promoter. Consistent with this finding, the tumor promoter 12-O-tetradecanoyl phorbol 13-acetate (TPA) produced less epidermal thickening and a smaller increase in the number of proliferating cells in AQP3-knockout versus wild-type mice (36). The mechanism underlying this reduced proliferation seems to relate to decreased cellular ATP levels caused by AQP3 deficiency, possibly because of reduced cellular uptake and metabolism of glycerol (36). Indeed, AQP3-knockout mice exhibit a decreased epidermal glycerol content (30). Keratinocyte proliferation induced by wounding also seems to be reduced by AQP3 deficiency (see below) (35), and in both cases supplementation with glycerol can correct the defect (35, 36). Finally, AQP3 overexpression was also found to increase glycerol uptake, keratin 5 and 14 expression, and cell growth in normal human keratinocytes, whereas AQP3 knockdown inhibited proliferation in response to the cytokine CCL17 (60). A similar inhibition of proliferation (as indicated by a reduced number of BrdU-positive cells) was observed in vivo in AQP3-knockout mice treated with retinoic acid (34). Thus, the results of Verkman and colleagues have suggested a pro-proliferative role for AQP3 in keratinocytes (35, 36) and a pro-tumorigenic role of this channel in skin cancer (36, 77). Similar results were obtained by Guo et al. (28), who found that siRNA-mediated AQP3 knockdown decreased human keratinocyte proliferation and increased the expression of several differentiation markers (keratin 10, involucrin, and filaggrin, markers of early, intermediate, and late differentiation, respectively). A pro-proliferative role for AQP3 has also been found in several cancers [reviewed in (57)], and it has been reported that AQP3 can promote an epithelial to mesenchymal transition in certain cell types, such as gastric cancer cells (10).

AQP3 also plays a role in migration, although in this case its transport of water, rather than other small molecules, like glycerol, appears to be the key. Hara-Chikuma and Verkman (35) demonstrated that knockdown of AQP3 in normal human keratinocytes decreased glycerol uptake, scratch wound healing, and fetal bovine serum-induced migration in vitro. Keratinocytes isolated from AQP3-knockout mice also showed reduced migration compared with cells originating from wild-type mice (35). Interestingly, adenoviral-mediated expression of either aquaporin-1, a classical aquaporin with minimal glycerol transport activity, or AQP3 was able to restore normal migration (35). This result suggests that AQP3’s transport of water mediates its effect on migration in keratinocytes, although a role for hydrogen peroxide transport is also possible (see below) (1, 59).

AQP3’S ROLE IN SKIN WOUND HEALING

As might be expected from its ability to modulate proliferation and migration, AQP3 also plays an important role in skin wound healing. For example, AQP3-knockout mice exhibit delayed wound healing of full-thickness skin wounds (35). This effect is accompanied by decreased keratinocyte proliferation (as measured by BrdU) during wound healing, and again, this deficiency can be rescued by supplementation with glycerol (35). Support for a role in wound healing is also provided by data demonstrating that AQP3 expression is reduced in the wounds of diabetic rats with impaired wound healing (73), although these data are only correlative. Unfortunately, however, rodents are not ideal models for human skin wound healing, as rodent wounds heal predominantly by contraction, whereas human wounds heal primarily by re-epithelialization, so whether AQP3 is also involved in human skin wound healing is unclear. Nevertheless, the data showing the ability of AQP3 to promote proliferation and migration of human keratinocytes argue for the likely importance of this channel in human skin wound healing as well.

AQP3’S ROLE IN KERATINOCYTE DIFFERENTIATION AND ITS ASSOCIATION WITH PHOSPHOLIPASE D2

We and others have also obtained data to support a role for AQP3 in keratinocyte differentiation. One of the first events to occur as basal keratinocytes move into the first differentiated layer of the stratum spinosum is growth arrest, and there is evidence consistent with an antiproliferative effect of AQP3. Thus, we found that AQP3-transported glycerol, but not equivalent amounts of xylitol or sorbitol (as osmolar controls), inhibited keratinocyte proliferation in vitro (6). In addition, in basal cell carcinoma, a nonmelanoma skin cancer characterized by excessive proliferation, AQP3 immunoreactivity was greatly reduced in the lesions compared with the overlying normal-appearing epidermis (78). Furthermore, in another nonmelanoma skin cancer, squamous cell carcinoma, AQP3 immunoreactivity was shown to outline the cells in those regions of the lesion that did not stain for Ki67, a marker of proliferation, but to be nearly absent in cells that were positive for this proliferative marker (78). Also, we have shown that co-expression of AQP3 with reporter constructs in which the promoters of keratinocyte differentiation markers drive the expression of luciferase results in elevated reporter enzyme activity, suggesting that AQP3 is able to promote differentiation (6). In subsequent experiments, we also demonstrated that AQP3 re-expression in AQP3-knockout keratinocytes increased the mRNA and protein levels of differentiation markers either alone and/or in combination with an elevated calcium concentration (13). Similarly, Kim and Lee (48) found that knockdown of AQP3 inhibited the expression of keratin 10 in keratinocytes induced to differentiate by elevated calcium levels. These authors also found that AQP3 promoted the survival of keratinocytes, since siRNA-mediated knockdown resulted in reduced cell viability (48). Finally, Guo et al. (28) demonstrated an initial increase in AQP3 protein expression with cell contact-initiated human keratinocyte differentiation followed by a decrease, consistent with an association between AQP3 levels and early differentiation. On the other hand, Hara-Chikuma et al. (34) found no effect of siRNA-mediated knockdown of AQP3 on the levels of differentiation markers in human keratinocytes induced to differentiate by an elevated calcium concentration. They also observed no difference in the basal expression of keratinocyte differentiation markers in AQP3-knockout mouse epidermis (34) [although this group also did not find a difference in proliferation under basal conditions (30)].

The ability of AQP3 to induce keratinocyte differentiation may depend on its interaction with another cellular protein, the lipid-metabolizing enzyme phospholipase D2 (PLD2). In previous studies, our group has demonstrated a physical and functional association between AQP3 and PLD2 (5, 6, 85, 87). Thus, AQP3 and PLD2 are colocalized in and can be co-immunoprecipitated from caveolin-rich membrane microdomains in keratinocytes (85). Our recent results indicate that this interaction is direct, rather than through a scaffolding or adaptor protein, since AQP3 and PLD2 can be co-immunoprecipitated from Sf9 insect cells only when both proteins are expressed (5). In addition, the functional association of these two proteins results in the PLD-mediated formation of phosphatidylglycerol (PG) in keratinocytes induced to differentiate in response to elevated extracellular calcium levels (87). We proposed that PLD2 converts the AQP3-transported glycerol to PG via its well-known transphosphatidylation reaction (Fig. 3) and subsequently demonstrated that manipulations to increase this AQP3/PLD2/PG signaling pathway promote keratinocyte differentiation (6). Thus, as mentioned above, co-expression of AQP3 increases the promoter activity of keratinocyte differentiation markers (6), and AQP3 re-expression in AQP3-knockout keratinocytes upregulates several differentiation markers (13). Importantly, the effect of this AQP3 re-expression on differentiation is inhibited by PLD2 inhibitors or co-expression of a lipase-dead PLD2 mutant (13), suggesting the importance of PLD2 activity for the differentiative action of AQP3. Finally, we have shown that direct provision of PG can also stimulate keratinocyte differentiation (6), although different PG species appear to exert different effects (80). PG also inhibits keratinocyte inflammatory mediator expression induced by pathogen-associated molecular patterns (PAMPs), microbial components that can activate pattern recognition receptors of the innate immune system, as well as several antimicrobial peptides, endogenous proteins released by endangered keratinocytes that function as danger-associated molecular patterns to also stimulate these receptors (11, 14).

Fig. 3.

The aquaporin-3-phospholipase D2 (AQP3-PLD2) signaling unit. AQP3 and PLD2 are physically and functionally associated such that AQP3 transports glycerol (Glyc) to PLD2, which then transphosphatidylates phosphatidylcholine (PC), exchanging the choline (Chol) headgroup of PC for glycerol, to produce phosphatidylglycerol (PG) and choline.

AQP3’S ROLE IN STRATUM CORNEUM HYDRATION

Strong evidence exists for an important role of AQP3 in skin hydration. One of the first phenotypes determined by Verkman and colleagues in the AQP3-knockout mice was that of reduced stratum corneum hydration compared with the wild-type control, as measured by changes in skin surface conductance and stratum corneum water content (30, 55). The difference between wild-type and AQP3-knockout mice could be abolished by exposure of the mice to low (10%) humidity (55) and was not accompanied by differences in stratum corneum morphology, thickness, lipid content, or levels of a variety of metabolites, including amino acids, lactic acid, glucose, or urea (30). However, a decrease in stratum corneum and epidermal glycerol content was observed (30). Indeed, restoring epidermal glycerol levels by topical, intraperitoneal, or oral administration of glycerol but not other osmotically active agents, including xylitol, propanediol, or urea, corrected the stratum corneum hydration defect; indeed, epidermal glycerol levels correlated with stratum corneum hydration (31). In addition, AQP3 is regulated in a circadian pattern, and skin hydration correlates with these cyclical AQP3 levels (58).

The importance of glycerol in skin function is also supported by a study in the asebia mouse model. These mice, due to a mutation in the gene encoding stearoyl-CoA desaturase-1, exhibit sebaceous gland hypoplasia and, therefore, deficient production of sebum (23), an oily secretion rich in triglycerides and other lipids that can be acted upon by lipases in the skin to release metabolites such as glycerol. The defect in sebum production results in a reduction in epidermal glycerol content and abnormal stratum corneum hydration, which in turn leads to hyperkeratosis (epidermal thickening), epidermal hyperplasia, and mast cell activation (23). Again, this phenotype can be corrected by topical administration of glycerol but not urea or water (23). Together, these results support a key physiological role for glycerol in skin function as well as the involvement of AQP3 in maintaining epidermal glycerol content and stratum corneum water-holding capacity.

Changes in AQP3 expression may also mediate the xeroderma (skin dryness) that is observed in diabetes. Thus, Ikarashi et al. (42) demonstrated that streptozotocin (STZ)-induced diabetes in mice resulted in downregulation of skin AQP3 mRNA and protein levels concomitant with a reduction in dermal water content (as well as transepidermal water loss; see below). The decrease in expression did not appear to be the result of elevated serum glucose levels, as no changes were seen one week after STZ injection, despite the fact that blood glucose was increased (42). However, one week later (or two weeks after STZ injection), AQP3 mRNA and protein levels were reduced at comparable (high) blood glucose levels. In addition to AQP3, the expression of several genes involved in regulating the circadian rhythm (Bmal1, Clock and Dbp) were reduced at two weeks (but not one week) after STZ (42), suggesting disruption of the circadian rhythm as a possible mechanism by which STZ-induced diabetes contributes to AQP3 downregulation. In summary, there is strong evidence to support a role for AQP3 in regulating skin hydration, leading to extensive research by industry to define novel agents to induce AQP3 expression, with the goal of improving hydration and other aspects of skin function [e.g., (24, 66)].

THE ROLE OF AQP3 IN THE EPIDERMAL WATER PERMEABILITY BARRIER

AQP3 has also been found to be important in the function of the epidermal water permeability barrier. The function of this permeability barrier can be assessed by monitoring transepidermal water loss (TEWL). Upon disruption of the barrier, for example, by repeated application and removal of tape to the skin (tape stripping), TEWL increases; as the permeability barrier is repaired, TEWL gradually returns to basal levels [reviewed in (22)]. Verkman and colleagues demonstrated that AQP3-knockout mice exhibit a delay in the repair of their water permeability barrier in comparison with wild-type mice (30). Furthermore, oral glycerol administration restored the impaired barrier recovery of AQP3-knockout mice and even accelerated recovery in wild-type mice (30). Consistent with these results, our group showed that a transgenic mouse in which AQP3 was overexpressed in the epidermis under the control of the keratin 1 promoter exhibited accelerated water permeability barrier repair (63).

AQP3 AND SKIN DISEASES

Accumulating evidence also points to a likely important role of AQP3 in various skin diseases, although there are often discrepant results found concerning whether AQP3 is increased, decreased, or mislocalized. For example, in psoriasis, AQP3 levels have been reported to be decreased or mislocalized in psoriatic lesions (50, 67, 78). In contrast, microarray and RNA-seq studies have indicated increased mRNA levels in psoriasis (9, 74). Although these data might suggest a possible divergence between mRNA and protein levels in psoriatic lesions, it seems more likely that the increased expression observed is due to the fact that proliferating keratinocytes comprise a greater proportion of psoriatic epidermis than of the normal epithelium, and AQP3 expression is decreased with later differentiation (85). Indeed, this interpretation is supported by results in a mouse model of psoriasis, in which AQP3 protein levels are elevated to a similar extent to the increase observed in the levels of keratins 5 and 14, which mark proliferating basal keratinocytes (33). Also consistent with the idea of a dysregulation of AQP3 in psoriasis, monomethyfumarate, an active metabolite of the dimethylfumarate in the psoriasis treatment Fumaderm, was found to upregulate AQP3, with particular efficacy to increase the plasma membrane-localized form of the protein (39).

On the other hand, another group has reported that AQP3 levels seem unchanged in psoriasis (33). This group also demonstrated that AQP3-knockout mice exhibit reduced psoriasiform lesion development and epidermal hyperplasia compared with wild-type mice of the same strain background in two mouse models of psoriasis (33). This effect was attributed to both keratinocytes and T cells, since the response was not fully reconstituted in lethally irradiated AQP3-knockout mice using wild-type bone marrow or in lethally irradiated wild-type mice using AQP3-knockout bone marrow (33). The mechanism underlying this effect of AQP3 occurs through an ability of interleukin (IL)-23-induced tumor necrosis factor-α (TNFα) to activate NADPH oxidase-2 (NOX2) to produce hydrogen peroxide, which is thought to gain entry into the cells through AQP3 and to activate nuclear factor-κB (NF-κB) (33). Furthermore, these authors demonstrated a role for AQP3 in T cell chemokine-induced chemotaxis also through its ability to transport hydrogen peroxide (32), again suggesting a potential role for AQP3 in psoriasis. Therefore, these results are paradoxical in suggesting that psoriatic keratinocytes exhibit either decreased or unchanged AQP3 levels; nevertheless, the data support some involvement of this channel in the disease.

Also in the nonmelanoma skin cancers basal and squamous cell carcinoma, discrepant results can be found in the literature. For example, Hara-Chikuma and Verkman (36) reported “strong AQP3 expression in a plasma membrane pattern” in all of the 40 squamous cell carcinoma samples examined. In addition, these authors demonstrated that AQP3-knockout mice are resistant to tumor formation in the two-stage carcinogenesis model of squamous carcinoma development (36). Consistent with this idea, siRNA-mediated knockdown of AQP3 in the epidermoid carcinoma cell line A431 decreased proliferation and increased apoptosis of these cells (79). These authors also reported increased AQP3 levels in actinic keratoses (premalignant skin lesions), Bowen’s Disease (squamous cell carcinoma in situ), and squamous cell carcinoma, although in the representative photomicrographs shown, AQP3 upregulation was not obvious (79). On the other hand, our group found “patchy” AQP3 staining in squamous cell carcinoma, with lesional keratinocytes positive for AQP3 showing little or no staining for Ki67, a marker of proliferative cells, and those cells that were Ki67-positive exhibiting minimal AQP3 staining (78). Another group also indicated reduced AQP3 levels in squamous cell carcinoma (67). In another nonmelanoma skin cancer, our group previously observed reduced AQP3 levels in basal cell carcinoma lesions compared with the overlying, normal-appearing epidermis (78); similar results were reported by Seleit et al. (67). It seems possible that some of the disparate results observed may relate to posttranslational modifications of AQP3, which might then affect the ability of different antibodies to recognize the antigen, again suggesting the importance of knowledge about these AQP3 modifications for a complete understanding of its role in the skin.

AQP3 in keratinocytes also appears to play a role in vitiligo, a disease characterized by melanocyte loss. Thus, Kim and Lee (48) found that AQP3 levels are decreased in lesional vitiligo versus non-lesional skin. Further experiments demonstrated that AQP3 knockdown in normal human keratinocytes results in reduced levels of E-cadherin and activated phosphoinositide 3-kinase (PI3K); these changes in E-cadherin and PI3K were also observed in lesional vitiligo skin (48). AQP3 is apparently important in keratinocyte viability because AQP3 knockdown reduced the survival of normal human keratinocytes (48). Similar results were observed in a study of 40 patients with vitiligo versus 25 normal control subjects in that keratinocyte AQP3 levels were reduced and mislocalized in lesional vitiligo epidemis compared with the normal samples (2). Another group found a similar reduction and mislocalization in both lesional and perilesional vitiligo skin (20). Therefore, the data support a role for AQP3 in maintaining the viability of keratinocytes, which presumably then promote melanocyte survival, with reductions in keratinocyte AQP3 levels contributing to vitiligo pathology.

Atopic dermatitis, also known as eczema, is another skin disease in which AQP3 may play a role, but in this disorder, AQP3 levels have been reported to be both increased and decreased. Thus, Nakagashi et al. (60) demonstrated increased AQP3 mRNA and protein expression in atopic dermatitis lesions in human patients and two mouse models of the disease. This upregulation of AQP3 was accompanied by enhanced TEWL, and the results were supported by a second group’s finding of increased AQP3 gene expression in atopic dermatitis lesions (61). On the other hand, another group reported a reduction in AQP3 levels in atopic dermatitis that correlated with areas of lesional skin exhibiting spongiosis (8). Although its exact involvement is, therefore, unclear, these results nonetheless point to a role for AQP3 in atopic dermatitis.

Recent studies have suggested a potential involvement of AQP3 in a few other skin diseases as well. Korany et al. (49) recently demonstrated reduced AQP3 levels in bullous pemphigus, a blistering skin disease in which autoantibodies to proteins mediating keratinocyte adhesion to the basement membrane induce keratinocyte loss of adhesion and death, resulting in blister formation. Soler et al. (71) have also suggested a role for AQP3 in pompholyx, also known as dyshidrotic eczema, a disease in which chronic vesicles appear on the soles and palms and cause pruritus and pain. These authors demonstrated that transgenic mice overexpressing both AQP3 and AQP10 in the epidermis under the control of the keratin 1 promoter exhibit enhanced acanthosis upon exposure to detergent-containing water and imiquimod, an inducer of the innate immune system (72). Therefore, they proposed that upregulation of these two channels in pompholyx, particularly in outer epidermal layers, might allow enhanced transepidermal water and glycerol loss exacerbated by exposure to water and the alkaline skin pH in pompholyx to result in skin dehydration (71). A recent study also suggests a downregulation of AQP3 levels in both lesional and perilesional skin in symmetrical acrokeratoderma (65), a disease in which symmetrical pigmented hyperkeratotic plaques form on distal (acral) regions of the skin but spare the palms and soles (26). Although perhaps not a disease, skin dryness associated with aging has also been linked to AQP3. Thus, data in human skin demonstrate decreased AQP3 in both extrinsically (sun-exposed) and intrinsically (chronologically) aged human skin (52, 68) and in mouse skin (42). This reduced AQP3 expression is thought to lead to decreased skin hydration (42). Aged skin also exhibits impaired keratinocyte proliferation and a resultant reduction in epidermal thickness (4, 7, 19, 25, 37, 64), leading to atrophy that contributes to poor wound healing, fragility, and an increased risk of skin tears [(16) and reviewed in (46)].

Finally, AQP3 in dermal fibroblasts, rather than in epidermal keratinocytes, may play a role in scleroderma/systemic sclerosis (SSc), a disease characterized by fibrosis of the skin and other organs. Thus, in a bleomycin-induced mouse model of scleroderma, AQP3 was found to be upregulated in mice injected with bleomycin compared with the control, as well as in fibroblasts isolated from the skin of these animals (54). The upregulation of AQP3 led to increased oxidative stress through its transport of hydrogen peroxide into the dermal fibroblasts of the bleomycin-treated mice, and this effect could be inhibited by AQP3 knockdown or treatment with the hydrogen peroxide-degrading enzyme catalase (54). The bleomycin-exposed fibroblasts also demonstrated a dose-dependent hydrogen peroxide-induced increase in transforming growth factor-β1 (TGFβ1) levels, which could again be inhibited by AQP3 knockdown or catalase treatment (54). Similar results were observed with collagen 1 and 3 expression. Finally, silencing of AQP3 in vivo in the bleomycin mouse model resulted in reduced dermal thickness/fibrosis, hydrogen peroxide levels and TGFβ1, and collagen 1 and collagen 3 expression (54). These results suggest that AQP3 might have a profibrotic role in the skin and the disease scleroderma. On the other hand, Yousefi et al. (83) demonstrated reduced mRNA and protein expression of AQP3 in dermal fibroblasts isolated from the skin of patients with SSc versus those from the skin of normal individuals. These SSc cells also showed a dysregulation of AQP3 expression in response to epidermal growth factor (EGF) and TGFβ1, with differences in the resulting induction of matrix metalloproteinase-1 expression (21). It is possible that AQP3 is downregulated in SSc patients in an attempt to reduce a profibrotic effect of this channel; however, it is also possible that the apparent discrepancies between these studies are due to species differences (mouse versus human) or model differences (acute bleomycin exposure versus chronic SSc). Nevertheless, the data suggest that AQP3 plays some role in skin fibrosis.

Together these accumulating data in the literature point to a key role for AQP3 in skin homeostasis, with its dysregulation contributing to multiple skin disorders. Perhaps it is not surprising that the expression of a channel protein that can transport multiple small molecules and is involved in so many cellular functions in the skin is abnormal in various skin diseases. The complexity of AQP3’s functions and its importance to understanding and possibly treating skin diseases in humans should make continued studies into this interesting channel a high priority in dermatologic research.

GRANTS

This work was supported in part by Veterans Affairs Merit no. CX001357 to W.B.B. K.A.H. was supported by National Institute of Diabetes and Digestive and Kidney Diseases Award nos. K01 DK105038 and R03 DK120503 and the pilot voucher program of no. P30 DK074038.

DISCLAIMERS

The contents of this article do not represent the views of the Department of Veterans Affairs or the US Government.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

W.B.B. and K.A.H. prepared figures; W.B.B. and L.A. drafted manuscript; W.B.B., L.A., J.W., and K.A.H. edited and revised manuscript; W.B.B., L.A., J.W., and K.A.H. approved final version of manuscript.