Journal of Molecular Medicine

, Volume 86, Issue 2, pp 221–231

Aquaporin-3 facilitates epidermal cell migration and proliferation during wound healing

Authors

  • Mariko Hara-Chikuma
    • Departments of Medicine and Physiology, Cardiovascular Research InstituteUniversity of California
    • Departments of Medicine and Physiology, Cardiovascular Research InstituteUniversity of California
Original Article

DOI: 10.1007/s00109-007-0272-4

Cite this article as:
Hara-Chikuma, M. & Verkman, A.S. J Mol Med (2008) 86: 221. doi:10.1007/s00109-007-0272-4

Abstract

Healing of skin wounds is a multi-step process involving the migration and proliferation of basal keratinocytes in epidermis, which strongly express the water/glycerol-transporting protein aquaporin-3 (AQP3). In this study, we show impaired skin wound healing in AQP3-deficient mice, which results from distinct defects in epidermal cell migration and proliferation. In vivo wound healing was ~80% complete in wild-type mice at 5 days vs ~50% complete in AQP3 null mice, with remarkably fewer proliferating, BrdU-positive keratinocytes. After AQP3 knock-down in keratinocyte cell cultures, which reduced cell membrane water and glycerol permeabilities, cell migration was slowed by more than twofold, with reduced lamellipodia formation at the leading edge of migrating cells. Proliferation of AQP3 knock-down keratinocytes was significantly impaired during wound repair. Mitogen-induced cell proliferation was also impaired in AQP3 deficient keratinocytes, with greatly reduced p38 MAPK activity. In mice, oral glycerol supplementation largely corrected defective wound healing and epidermal cell proliferation. Our results provide evidence for involvement of AQP3-facilitated water transport in epidermal cell migration and for AQP3-facilitated glycerol transport in epidermal cell proliferation.

Keywords

AQP3Wound healingCell migrationCell proliferationGlycerol transport

Abbreviations

AQP

aquaporin

MAPK

mitogen-activated protein kinase

Introduction

Cutaneous wound healing is a multi-step process that involves several cell types, including epidermal keratinocytes, fibroblasts, endothelial cells, and peripheral nerve cells. Re-epithelialization is a crucial step during wound healing, which involves the migration and proliferation of keratinocytes from the surrounding epidermis and appendages such as hair follicles and sweat glands (reviewed in [1, 2]). After injury, keratinocytes at the wound margin begin to proliferate behind actively migrating cells, resulting in a dense, hyperproliferative epithelium seen as migrating cell sheets at the wound margins. These events are thought to be regulated mainly by growth factor receptors, integrins, extracellular matrix, and matrix metalloproteases (reviewed in (35]).

The aquaporins (AQPs) are a family of small, integral membrane proteins that transport water and, in some cases, water and glycerol (‘aquaglyceroporins’). Phenotype analysis of mice lacking individual AQPs has proven their anticipated involvement in the urinary concentrating mechanism and exocrine glandular fluid secretion (reviewed in [6]) and has elucidated various unanticipated roles of AQPs, such as in brain edema [7], neural signal transduction [8], and fat metabolism [9]. The mechanisms underlying each of these phenomena were attributed to AQP water or glycerol transporting functions. The newly described AQP11 may be functionally distinct from other water and/or glycerol transporting AQPs, as AQP11 gene disruption produces a severe phenotype with renal vacuolization and cyst formation [10].

Recently, we reported a novel cellular role for AQPs in cell migration, which was initially demonstrated in endothelial cells and transfected cells [11] and subsequently in brain astroglial cells [12], kidney proximal tubule cells [13], tumor cells [14], and corneal epithelium cells [15]. We suggested that AQP-facilitated water influx into dynamic cellular protrusions (lamellapodia) at the leading edge of migrating cells occurs in AQP-dependent cell migration.

In skin, AQP3 is expressed in plasma membranes of the basal epidermal cell layer [1618]. Compared to wild-type mice, AQP3 null mice have relatively dry skin, reduced skin elasticity, and delayed recovery of barrier function after removal of the stratum corneum [17, 19]. We suggested that these defects were consequent to the absence of AQP3-facilitated glycerol transport, resulting in reduced stratum corneum and epidermal cell glycerol content [19, 20]. We also described delayed wound repair in AQP3 null mice in a hairless genetic background, although no information was available at that time about possible cellular mechanisms.

The goals of this study were to characterize and investigate possible mechanisms of AQP3-dependent wound healing in skin. We tested the hypothesis that AQP3-facilitated water and/or glycerol transport is involved in epidermal cell migration and proliferation during repair of skin wounds. For these studies, we utilized AQP3 null mice and primary keratinocyte cultures derived from mouse skin and siRNA-treated human keratinocyte cultures. Our results provide strong evidence for distinct defects in AQP3-dependent epidermal cell migration and proliferation, which account for impaired in vivo would healing in the AQP3 null mice. Mechanistic analysis provided evidence for involvement of AQP3-facilitated water transport in epidermal cell migration and for AQP3-facilitated glycerol transport in epidermal cell proliferation. Our data suggest pharmacological modulation of AQP3 as a possible therapy to accelerate wound healing in traumatic, burn, and other forms of injury.

Materials and methods

AQP3 null mice

AQP3 null mice (CD1 genetic background) were generated by targeted gene disruption as described [21]. Protocols were approved by the UCSF Committee on Animal Research.

Human keratinocyte cell cultures

Neonatal human keratinocytes (NHK, Cascade Biologics, Portland, OR, USA) were grown in keratinocyte basal medium (KBM) supplemented with 0.2% bovine pituitary extract, 5 μg/ml insulin, 0.18 μg/ml hydrocortisone, 5 μg/ml transferrin, and 0.2 ng/ml epidermal growth factor (EGF). NHK were transfected with AQP3 siRNA or RNA-induced-silencing-complex-free siRNA (Dharmacon, Lafayette, CO, USA) at 40–50% confluence using Lipofectamine 2000™ (Invitrogen Life Technologies, Carlsbad, CA, USA). AQP3 expression was confirmed by immunofluorecence and immunoblotting using a commercial polyclonal AQP3 antibody (Chemicon, Temecula, CA, USA). Experiments were done at 48–72 h after transfections.

Mouse keratinocyte cell cultures

Full-thickness skin from 1- to 3-day-old mice was incubated in dispase II (5 U/ml, Roche, Nutley, NJ, USA) for 7 h at 4°C. The epidermis was separated from the dermis, cut into fragments, and treated with 0.25% trypsin and 0.1% ethylenediaminetetraacetic acid (EDTA) for 10 min. Cells were seeded at a density of 105 per cm2 on collagen type I plates (BD Biosciences, Bedford, MA, USA) and cultured in keratinocyte growth medium (Cambrex, East Rutherford, NJ, USA) at 37°C under 5% CO2. Recombinant adenoviruses encoding rat AQP1 and AQP3 (AQP1-Ad and AQP3-Ad) were custom generated by ViraQuest (North Liberty, IA, USA). Keratinocytes from AQP3 null mice were infected with AQP1-Ad or AQP3-Ad at 1,000 pfu/cell. Experiments were done at 48–72 h after viral infection.

Water and glycerol permeability measurements

Water permeability was measured by calcein fluorescence quenching, as described [13]. Calcein fluorescence was monitored in response to changing perfusate osmolalities from 300 (phosphate-buffered saline, PBS) to 450 mosM (PBS containing 150 mM mannitol). Glycerol permeability was measured by 14C-glycerol uptake for 90 s or 3 min, as described [9].

In vivo wound healing

Two full-thickness punch biopsies extending through the epidermis and dermis (diameter, 5 mm) were done on the backs of wild-type and AQP3 null mice. Wound repair, which was expressed as the percentage of initial wound area, was monitored daily. In some experiments, mice were given water containing glycerol ad libitum (5% for wild-type and 1% for AQP3 null mice) for 3 days before wounding, along with standard solid mouse chow as described [20].

In vitro wound healing assay

In vitro wound healing was assayed in confluent cell monolayers as described [13]. Cells were stained with phallotoxins conjugated to Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA) to visualize the F-actin and quantify lamellapodia. In some experiments, BrdU was incorporated for 8 h after wounding, and cells were stained with BrdU antibody. BrdU-positive cells were counted within 250 μm from the wound margin and expressed as the percentage of BrdU-positive cells [with all cells identified by 4′,6-diamidino-2-phenylindole (DAPI) staining].

Transwell migration assay

Cell migration was assayed using a modified Boyden chamber containing a polycarbonate Transwell membrane filter (5 μm pore size for mice keratinocytes, 8 μm pore size for NHK; Corning Costar, Cambridge, MA, USA) coated with collagen type I. Cells (105) were deposited on the upper chamber in KBM containing 1% fetal bovine serum (FBS). The lower chamber contained KBM with 10 or 1% (control) FBS. Cells were incubated for 6 h at 37°C in 5% CO2/95% air. After scraping non-migrated cells from the upper surface of the membrane with a cotton swab, the migrated cells remaining on the bottom surface were stained with 0.5% crystal violet. The stained insert was washed thoroughly, dissolved with 1% Triton X-100, and absorbance (O.D.) at 595 nm was measured.

Histology

To measure cell proliferation, mice were injected intraperitoneally with BrdU (100 μg/g body weight; Sigma-Aldrich, St. Louis, MO, USA) at 1 h before sacrifice. Paraffin-embedded sections were stained with hematoxylin and eosin, BrdU antibody (Abcam, Cambridge, MA, USA), or AQP3 antibody (Chemicon). BrdU-positive cells were counted in epidermis to 500 μm from the wound edge.

Cell size and adherence

Cell size was measured by transmission light microscopy at high magnification (250×) as described [11]. Cell adherence to collagen type I-coated wells (10 μg/ml, BD Biosciences, San Jose, CA, USA) or non-coated wells was measured as described [13].

Keratinocyte proliferation

Cell proliferation was assessed as [3H-methyl]-thymidine incorporation into DNA as described [22]. Total cellular DNA was determined using the Hoechst reagent (Sigma). DNA synthesis was expressed as 3H-thymidine incorporated per microgram DNA. BrdU incorporation into keratinocytes was detected using a BrdU Cell Proliferation Assay kit (Calbiochem).

Immunoblot analysis

To analyze MAPK signaling, cells were starved for >24 h and incubated with 12-O-tetradecanoylphorbol-13-acetate (TPA) and lysed with ice-cold lysis buffer (Cell Signaling Technology, Boston, MA, USA). The epidermis was separated by heat-split and homogenated in lysis buffer. The supernatant (14,000×g, 10 min, 4°C) was used for immunoblotting with antibodies against phospho-ERK, ERK, phospho-p38, p38, phospho-JNK (Cell Signaling Technology), JNK (Biosource, Camarillo, CA, USA), and β-actin (Sigma).

Phospho-p38 ELISA assay

Keratinocytes were incubated for 1 h with vehicle [0.01% dimethyl sulfoxide (DMSO)], 100 nM TPA, or 20 ng/ml EGF. Phospho-p38 content was measured by cellular activation of signaling enzyme-linked immunosorbent assay (ELISA) kit (SuperArray, Ann Arbor, MI, USA).

Statistical analysis

Statistical analysis was performed using the two-tailed Student’s t test or by analysis of variance (ANOVA).

Results

Delayed healing of cutaneous wounds in AQP3 null mice

We measured healing after the creation of full-thickness cutaneous wounds in wild-type (+/+) and AQP3 null (−/−) mice in a CD1 genetic background. Figure 1a (left) shows significantly delayed wound closure in AQP3 null mice, as quantified by wound area measurements, in agreement with our prior data obtained using hairless mice in a SKH1 genetic background [19]. Photographs of wound appearance are shown in Fig. 1a (right), just after wounding, and at 1 and 3 days later. Figure 1b (left) shows greater epidermal cell hyperplasia near the wound edge in wild-type than in AQP3 null mice at 3 days after wounding. Figure 1b (right) shows impaired epithelial closure at 8 days in AQP3 null mice. Figure 1c shows strong expression in AQP3 protein near wound edge, which was associated with epidermal cell hyperplasia.
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig1_HTML.gif
Fig. 1

Impaired wound healing in skin in AQP3 null mice. a Two full-thickness punch biopsies (diameter, 5 mm) were created on the backs of wild-type (+/+) and AQP3 null (−/−) mice. Wound area was measured daily. Left Percentage of initial wound area (SE, n = 5); p < 0.01 at 1–8 days. Right Representative photographs of wound size just after wounding, and at 1 and 3 days later. Bar 1 mm. b Hematoxylin and eosin staining of skin at 3 and 8 days after wounding. Bar 100 μm. epi Epidermis. Arrowhead Wound margin. Arrow Non-epithelialized area. c AQP3 immunostaining of mouse skin of indicated genotype before wounding and at 1 and 3 days after wounding. Bar 50 μm. Arrowhead Wound margin

Impaired in vitro wound healing in AQP3-deficient keratinocytes

Neonatal human keratinocytes (NHK) were used to investigate the involvement of AQP3 in wound repair. Immunolocalization confirmed AQP3 expression in the plasma membrane in control NHK (Fig. 2a). Transfection of siRNA-AQP3 into NHK, after optimization of conditions, consistently reduced AQP3 protein expression by ~95%. AQP3 immunostaining is shown in Fig. 2a (left) and immunoblot analysis in Fig. 2a (right). Osmotic water permeability, measured by calcein fluorescence quenching in response to osmotic challenge, was significantly reduced in the AQP3 knock-down NHK. Original fluorescence data are shown in Fig. 2b (left) and averaged water permeability rates in Fig. 2b (right). As shown in Fig. 2c, glycerol uptake was also reduced in the AQP3 knock-down cells. These data demonstrate effective AQP3 knock-down in human keratinocyte cell cultures.
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig2_HTML.gif
Fig. 2

Impaired wound healing in vitro in AQP3-knock-down human keratinocytes. aLeft AQP3 immunostaining of normal human keratinocytes (control) and AQP3-siRNA-treated cells (AQP3 RNAi). Nuclei stained with DAPI (blue). Bar 20 μm. Right AQP3 immunoblot of cell homogenates (20 μg protein/lane). b Osmotic water permeability of human keratinocyte cultures measured by calcein fluorescence quenching. Left Representative time course data showing responses to rapid changes in perfusate osmolality between 300 and 450 mOsm in PBS (pH 7.4). Right Reciprocal exponential time constants (τ−1) in six separate sets of measurements (SE, asterisk, p < 0.01), which are proportional to osmotic water permeability. c Glycerol permeability measured by 14C-glycerol uptake at 90 s (SE, asterisk, p < 0.01). d In vitro wound healing assay. Left Light micrographs of wounded cell monolayers showing impaired wound closure at 24 h in AQP3 knock-down (RNAi) human keratinocytes. Right Speed of wound closure in control and AQP3 knock-down keratinocytes (SE, n = 6, asterisk, p < 0.01)

In vitro assays of wound healing were done in confluent NHK monolayers. Cells were scraped to produce an ~500-μm wide defect in the cell layer. Figure 2d shows significantly slowed wound closure in AQP3 knock-down keratinocytes at 24 h after creation of the linear wounds. The wounded area is repopulated by adjacent cells through the combined processes of cell migration and proliferation [5, 23]. Therefore, we next measured cell migration and proliferation, separately, to determine the role(s) of AQP3 on wound repair.

Reduced cell migration in AQP3-deficient keratinocytes

The transwell migration assay was done to determine the role of AQP3 in keratinocyte cell migration. Cell migration towards FBS (chemotactic stimulus) was assayed using a modified Boyden chamber. Migrated cells on the bottom surface of the transwell membrane were stained and the migration quantified by optical absorbance. Migration of AQP3 knock-down keratinocytes toward 10% FBS was reduced significantly compared to control cells (Fig. 3a). No difference in migration was seen in the absence of a serum gradient. F-actin staining with fluorescent phallotoxin revealed cell protrusions (lamellipodia) at the wound edge during wound closure (Fig. 3b, left). The area (expressed per cell length) of the protrusions was significantly lower in the AQP3 knock-down keratinocytes (Fig. 3b, right). F-actin staining was similar in areas of wild-type and AQP3 null cells not including lamella, suggesting that delayed wound healing in AQP3-deficient cells was not due to a cytoskeletal abnormality. Cell size and adherence did not differ significantly in control and AQP3 knock-down human keratinocytes (Figs. 3c,d).
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig3_HTML.gif
Fig. 3

Reduced cell migration in AQP3-deficient human keratinocytes. a Transwell cell migration assay. Migrated human keratinocytes (quantified by absorbance after crystal violet staining) at 6 h after cell plating (SE, n = 6, asterisk, p < 0.01). The bottom chamber contained 10% or 1% FBS. bLeft Fluorescence micrographs of cells stained with phallotoxins showing protrusions at the leading edge (arrowheads). Bar 20 μm. Right Area of protrusions per length of cells at the wound edge (SE, ~100 cells analyzed, asterisk, p < 0.01). c Adherence of human keratinocytes to type I collagen-coated or non-coated wells, quantified by crystal violet absorbance (SE, n = 6). d Cell diameter of freshly trypsinized keratinocytes (SE, ~100 cells measured per genotype). Differences in c and d not significant

AQP3-dependent cell migration was also examined in keratinocyte cell cultures from epidermis of wild-type and AQP3 null mice. We previously reported evidence in multiple systems for involvement of water-selective AQPs, such as AQP1 and AQP4, in cell migration but not in cell proliferation [1114]. To determine whether AQP3-facilitated water and/or glycerol transport is responsible for the impaired wound healing in AQP3 deficiency, AQP3 null keratinocytes were infected with adenoviruses encoding GFP, AQP1, or AQP3 (GFP-Ad, AQP1-Ad, and AQP3-Ad). Figure 4a shows that these cells are indistinguishable in their appearance by phase-contrast light microscopy. Adenoviral infection with AQP1- or AQP3-Ad resulted in expression of the respective proteins with expected molecular size of ~28 kDa (Fig. 4b). Water permeability (for AQP1 and AQP3 infections) and glycerol permeability (for AQP3 infection) were restored by adenovirus infection (Fig. 4c,d). Figure 4e shows significantly impaired migration towards 10% FBS in keratinocyte cultures from AQP3 null mice. Cell size and adherence did not differ significantly in wild-type vs AQP3 null keratinocytes (not shown). AQP1 or AQP3 replacement by adenovirus infection corrected the migration defect in AQP3 null keratinocytes (Fig. 4f). Because AQP1 transports water but not glycerol, these data support the conclusion that absence of AQP3-facilitated water transport in the AQP3 null keratinocytes is responsible for their migration defect.
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig4_HTML.gif
Fig. 4

Involvement of AQP3 water permeability in keratinocyte migration. a Phase-contrast micrographs of primary cultures of keratinocyte from wild-type and AQP3 null epidermis. Bar 20 μm. b Immunoblot analysis of AQP1 and AQP3 in keratinocyte cultures from mice of indicated genotype and after indicated adenoviral infections. c Osmotic water permeability measured by calcein fluorescence quenching in wild-type and AQP3 null keratinocytes and in AQP3 null keratinocytes after infection with adenoviruses encoding GFP, AQP1, or AQP3 (GFP-Ad, AQP1-Ad, and AQP3-Ad). Reciprocal exponential time constants (τ−1) in six separate sets of measurements (SE, double asterisks, p < 0.01; asterisk, p < 0.05). d Glycerol permeability measured by 3 min 14C-glycerol uptake (SE, n = 5, asterisk, p < 0.01). e Transwell cell migration assay of wild-type and AQP3 null mouse keratinocytes (SE, n = 6, asterisk, p < 0.01). The bottom chamber contained 10 or 1% FBS. f Restoration of defective migration of AQP3 null keratinocytes after infection with adenoviruses encoding AQP1 or AQP3 (SE, n = 6, asterisk, p < 0.01)

Impaired cell proliferation in AQP3-deficient keratinocytes

To investigate a possible role of AQP3 in cell proliferation during wound healing, we measured BrdU incorporation into keratinocytes during wound closure both in living mice and cell cultures. Numerous BrdU-positive cells were seen in epidermis of wild-type mice compared to that seen in AQP3 null mice (Fig. 5a, left). Figure 5a (right) shows a significantly lower percentage of BrdU-positive cells in AQP3-deficient epidermis at 3 days. In the in vitro wound healing assay, the number of BrdU-positive cells (green, proliferating cells) and DAPI-stained cells (blue, nuclear staining) were counted within 250 μm of the wound edge at 10 h after wounding and in confluent cells. Figure 5b (left) shows many BrdU-positive control cells but remarkably fewer BrdU-positive cells in the AQP3 knock-down monolayer. Averaged data are summarized in Fig. 5b (right). Few BrdU-positive cells were seen in confluent, non-wounded cell monolayers. These observations suggested the involvement of AQP3 in not only cell migration during wound repair but also in cell proliferation.
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig5_HTML.gif
Fig. 5

Impaired cell proliferation in AQP3-deficient keratinocytes during wound healing. a BrDU staining before wounding (top) and at 3 days after wounding (bottom). Bar 50 μm. Right Percentage of BrdU-positive cells in epidermal basal layer within 500 μm from the wound margin (SE, n = 4, asterisk, p < 0.01). bLeft BrDU staining during wound closure in control and RNAi knock-down human keratinocyte cultures. Bar 50 μm. Right Percentage of BrdU-positive cells within 250 μm of the wound margin (SE, n = 4, asterisk, p < 0.01 for control vs RNAi). c DNA synthesis in control and RNAi knock-down human keratinocyte cultures. EGF was added after overnight incubation in medium lacking growth factors (‘starved’). 3H-thymidine was added 18 h after EGF addition and incubated for 2 h (SE, n = 4–5, asterisk, p < 0.01). d In vitro wound healing assay in human keratinocytes in medium with or without EGF (starved). Speed of wound closure in control and AQP3 knock-down keratinocytes (SE, n = 5–6, asterisk, p < 0.05, double asterisks, p < 0.01)

DNA synthesis was assayed in control and AQP3 knock-down keratinocytes to determine the involvement of AQP3 in growth factor-induced cell proliferation. For these studies, we chose EGF, which is abundantly released at wound sites as one of the key regulators of keratinocyte proliferation [24]. Figure 5c shows reduced cell proliferation in AQP3 deficient keratinocytes. Similar results were seen in mice primary keratinocyte cultures (not shown). Figure 5d shows that EGF accelerated wound repair in NHK, with lesser effect in AQP3 knock-down cells. EGF receptor expression was identical in control and AQP3-deficient keratinocytes by immunoblot analysis (not shown). These data suggest the involvement of AQP3 in cell proliferation-induced stimuli such as wounding or growth factors. Therefore, we hypothesized that AQP3 might be involved in cellular responses to stimuli leading to cell proliferation.

Impairment of MAP kinase cell signaling in AQP3-deficient keratinocytes

Mitogen-activated protein kinase (MAPK) signaling in response to inflammation, stress, stimuli, or mitogens is a key determinant of cell proliferation, differentiation, and apoptosis [25, 26]. To determine the involvement of AQP3 in the cellular response to exogenous stimuli, we measured MAPK activation after treatment with the PKC/MAPK activator phorbol 12-O-tetradecanoylphorbol-13-acetate (TPA). Non-phosphorylated and phosphorylated extracellular signal-regulated kinase (ERK) were measured in mouse epidermis, as well as p38 kinase and c-Jun NH2-terminal kinase (JNK). Figure 6a shows greatly reduced phosphorylated p38 and JNK in epidermis from TPA-treated wild-type vs AQP3 null mice. Similar results were seen in keratinocyte primary cell cultures (Fig. 6b), with the greatest difference seen for phosphorylated p38. We also confirmed that the impaired p38 activation in AQP3 deficiency is not due to a difference in sensitivity to TPA or to delayed p38 phosphorylation (not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig6_HTML.gif
Fig. 6

Impaired mitogen-induced MAPK signaling in AQP3-deficient epidermis and keratinocytes. a Epidermis was excised at 30 min after treatment with vehicle (acetone) or TPA. MAPK signaling assayed by immunoblot analysis using antibodies against phospho-ERK (p-ERK), ERK, phospho-p38, p38, phospho-JNK, JNK, and β-actin. Blots representative of four separate sets of experiments. b MAPK signaling in keratinocyte cell cultures from mice at 30 min after incubation with vehicle (0.001% DMSO) or TPA (10 nM). Blots representative of three separate sets of experiments. c Immunoblot of phospho-p38 and p38 in mouse keratinocytes at 30 min after TPA exposure (10 nM). AQP3 null keratinocytes were infected with indicated adenoviruses encoding GFP, AQP1, or AQP3. d Phospho-p38 ELISA in mouse keratinocytes treated with vehicle (0.01% DMSO), 100 nM TPA, or 20 ng/ml EGF (SE, n = 3–4, asterisk, p < 0.01)

To determine whether AQP3-facilitated water and/or glycerol transport is responsible for the AQP3-dependent MAPK signaling, TPA-induced p38 activation was determined in mouse keratinocytes infected with adenoviruses encoding GFP, AQP1, or AQP3. Figure 6c shows that AQP3 replacement corrected the defect in p38 activation, whereas AQP1 had little effect, suggesting a link between AQP3-facilitated glycerol transport and p38 activation. To further investigate the apparent impairment of p38 activation in AQP3 deficiency, phospho-p38 was assayed in wild-type and AQP3 null keratinocyte in response to EGF and to TPA. Figure 6d shows greatly reduced TPA- and EGF-induced p38 activation in AQP3-deficient keratinocytes.

Glycerol replacement corrects impaired wound healing and cell proliferation in AQP3-deficient mice

To test the involvement of AQP3-facilitated glycerol transport in wound healing in vivo, we determined whether glycerol supplementation could correct the delayed wound healing in AQP3 null mice. Mice were given glycerol orally for 3 days before wound creation. Figure 7a shows that glycerol administration significantly improved wound healing in AQP3 null mice but did not fully correct wound healing to that in wild-type mice. The reduced number of BrdU-positive cells in AQP3 null epidermis was, however, fully corrected by glycerol administration (Fig. 7b). Decreased cell proliferation in human AQP3 knock-down keratinocytes was also restored after glycerol supplementation (Fig. 7c). Together, these results provide evidence for involvement of AQP3-facilitated glycerol in keratinocyte cell proliferation during wound healing.
https://static-content.springer.com/image/art%3A10.1007%2Fs00109-007-0272-4/MediaObjects/109_2007_272_Fig7_HTML.gif
Fig. 7

Glycerol replacement corrects impaired wound healing in AQP3 deficiency. a Percentage of initial wound area in wild-type and AQP3 null mice without and with glycerol administration (SE, n = 5). Where indicated mice were given glycerol orally for 3 days before wounding and during wound healing. b Percentage of BrdU-positive cells in epidermal basal layer within 500 μm from the wound margin in glycerol-treated mice (SE, n = 4). Differences not significant. c Cell proliferation assayed by BrdU ELISA in control and RNAi knock-down human keratinocytes. Keratinocytes were incubated with 10 mM glycerol in starved media overnight. BrdU was added 18 h after adding EGF. Ratios expressed as BrdU incorporation per cell number (SE, n = 5–6, asterisks, p < 0.01). d Proposed mechanism for AQP3-dependent wound healing involving AQP3-facilitated glycerol transport in epidermal cell proliferation and AQP3-facilitated water transport in epidermal cell migration

Discussion

Healing of cutaneous wounds was slowed in vivo in AQP3-deficient mice. In wild-type mice, the wound edge during the healing phase was hyperproliferative, as shown by hyperplasia and a high percentage of BrdU-positive epidermal keratinocytes. AQP3 expression was strongly elevated in these hyperproliferative cells. In contrast, much less hyperplasia and BrdU-positivity was found during wound healing in AQP3 null mice. We showed previously the selective reduction in glycerol content in stratum corneum and epidermis of AQP3-deficient mice, without differences in glycerol content in dermis and serum [19], which was interpreted in terms of impaired glycerol transport from dermis to more superficial layers of skin through the normally highly glycerol-permeable basal layer of epidermal keratinocytes. Glycerol supplementation, under conditions found previously to normalize glycerol content and hydration in outer layers of skin [20], partially corrected the defect in wound healing in AQP3 null mice. The involvement of AQP3 in cell proliferation may be a general phenomenon of importance to other AQP3-expressing cells in tissues outside of the skin. AQP3 is prominently expressed in epithelial cells at the ocular surface and in regenerating basal enterocytes in colon. We found recently that corneal surface re-epithelialization is impaired in AQP3-deficient mice in vivo [15] and that colonic function and survival were greatly reduced in AQP3-deficient mice in dextran sulfate and acetic acid models of colitis [27]. Although those studies did not establish a mechanism for the phenotypes, in both cases, defects in in vivo cell proliferation were established based on relative hypoplasia and reduced BrdU positivity in AQP3 deficiency.

A recent study showed that AQP3 knock-down reduced EGF-induced migration of human skin fibroblasts in a phagokinetic track motility assay [28]. It was suggested that EGF increased cell migration by increasing AQP3 expression via the EGF receptor and phosphoinositide 3-kinase and ERK signal transduction pathways. Studies here using keratinocyte cell cultures provided strong evidence for distinct defects in cell migration and proliferation in AQP3 deficiency. We used normal human keratinocyte cultures, in which AQP3 expression was greatly reduced by siRNA treatment, as well as keratinocytes cultured from wild-type and AQP3 null mice, and adenovirus-corrected AQP3 null keratinocytes. Defective migration of AQP3 knock-down or knock-out keratinocytes was found using in vitro wound healing and/or transwell assays. Reduction in lamellipodial surface area was seen in migrating AQP3 knock-down keratinocytes at the leading edge of a healing wound in vitro. Impaired migration of AQP3 null keratinocytes was corrected by adenovirus-mediated replacement of AQP3 or of the water-selective transporter AQP1, providing evidence that the water transporting function of AQP3 is responsible for AQP3-facilitated keratinocyte migration (Fig. 7d). These results are in agreement with an increasing body of evidence for involvement of AQPs in migration of many different types of cells, including tumor cells, as mentioned in “Introduction”. We believe that AQP-facilitated water influx in lamellipodia facilitates their dynamic formation and extension, resulting in accelerated cell migration. However, cell AQP expression is not an absolute requirement for migration because of the substantial water permeability of lipid bilayers, such that water permeability in AQP-expressing cells is generally only 5–20-fold greater than that after AQP deletion or inhibition [29].

During healing of cutaneous wounds, the hair follicle is thought to contribute to epidermal renewal [30, 31]. Recently, stem cells located in the bulge region of the hair follicle were found to facilitate wound repair [32], in which migration of bulge cells into the healing epidermis was seen during wound healing. We found strong AQP3 expression in the hair follicle, including the bulge region (data not shown). AQP3-facilitated water transport might thus also be involved in migration of bulge cells, accelerating wound repair.

After injury, accelerated cell proliferation, which is sustained by growth factors and integrins and matrix metalloprotinases, is important for re-epithelialization during wound healing [5]. The impaired proliferation in AQP3-deficient keratinocytes in the in vivo and in vitro wound healing assays suggested that defective AQP3-dependent keratinocyte cell proliferation is responsible, at least in part, for the defective wound healing phenotype in AQP3 null mice (Fig. 7d). The in vitro proliferation assay showed remarkable impairment of EGF-induced cell proliferation in AQP3-deficient keratinocytes. Mechanistic analysis showed impaired p38 MAPK activation induced by mitogens, including EGF and TPA, in AQP3-deficient keratinocytes. The impairment in p38 activation was corrected by adenoviral delivery of exogenous AQP3 but not the water-selective transporter AQP1. p38 MAPK is one of the MAPKs thought to be involved in regulating epidermal cell proliferation, differentiation, apoptosis, and response to inflammation and stress [26, 33]. In vitro studies using NHK showed the involvement of p38 activation in wound repair [3436]. Although further experiments are needed to investigate the link between p38 MAPK cell signaling and wound repair, our data support the involvement of AQP3 in the cell proliferative response to exogenous stimuli such as mitogens or wounding.

We determined whether AQP3-facilitated water and/or glycerol transport is responsible for cell proliferation. Oral glycerol administration corrected the impairment in wound repair and cell proliferation in AQP3 null mice. Glycerol supplementation also corrected the decreased cell proliferation in human AQP3 knock-down keratinocytes. The detailed metabolic pathways of glycerol in the epidermis have not been established. It has been reported that about 60% of synthesized glycerol is metabolized to glycerol-3-phosphate, which is one of the key metabolic intermediates for ATP production [37]. Although further experiments are needed to investigate the link between AQP3 and cell proliferation, our data thus support the requirement of AQP3-facilitated glycerol transport in cell proliferation during wound repair.

In conclusion, our data provide evidence that impaired wound healing in AQP3 deficiency in vivo results from distinct defects in epidermal cell migration and proliferation, which are consequent to the water and glycerol transporting functions of AQP3, respectively. Our proposed mechanism in Fig. 7d is supported by direct measurements of cell migration and proliferation in keratinocyte cell cultures. A consequence of AQP3-dependent epidermal cell proliferation is the possible involvement of AQP3 in a number of skin disorders associated with epidermal hyperproliferation, such as skin carcinogenesis, psoriasis, and atopic dermatitis and repair of burn and other wounds. Our findings suggest that pharmacological modulation of AQP3 expression or function may be of benefit in increasing epidermal proliferation during wound healing.

Acknowledgments

We thank Liman Qian for mouse breeding and genotype analysis. This work was supported by NIH grants DK35124, EY13574, EB00415, HL59198, HL73856, and DK72517 and Research Development Program and Drug Discovery grants from the Cystic Fibrosis Foundation.

Copyright information

© Springer-Verlag 2007