Aquaporin-1 Facilitates Epithelial Cell Migration in Kidney Proximal Tubule

AQP1 Null Mice

AQP1 null mice were generated by targeted gene disruption as described (2). Protocols were approved by the UCSF Committee on Animal Research.

Primary Culture of Kidney Proximal Tubule Epithelial Cells

Proximal tubule cells were cultured from 1- to 2-mo-old wild-type and AQP1 null mice (in a CD1 genetic background) as described (14). Briefly, the renal cortices were minced and incubated for 30 min in Hanks’ solution that contained 140 U/ml collagenase (Worthington, Lakewood, NJ) and 0.75 mg/ml trypsin inhibitor (Sigma, St. Louis, MO). After removal of undigested fragments that settled by gravity, the suspension was mixed with an equal volume of Hanks’ solution that contained 10% FBS and centrifuged for 7 min at 50 × g (4°C). Cortical tubules were purified by gradient centrifugation in DMEM/F-12 that contained 40% Percoll for 20 min at 36,000 × g (4°C). The band that contained proximal tubules was collected and washed twice with DMEM/F-12. Cells were cultured in DMEM/F-12 that contained 2 mM glutamine, 15 mM HEPES, 5 μg/ml transferrin, 5 μg/ml insulin, and 50 nM hydrocortisone. These cells were previously shown to be predominantly of proximal tubule origin (14), and this was confirmed here by immunostaining using an anti-megalin antibody (Santa Cruz Biotechnology, Santa Cruz, CA).

Adenovirus-Mediated AQP1 Expression

A recombinant adenovirus vector encoding rat AQP1 (AQP1-Ad) was generated as described (15). Proximal tubule cells were infected with AQP1-Ad at a concentration of 100 PFU/cell. The medium was replaced after 1 h of incubation with virus at 37°C. Experiments were done at 48 to 72 h after viral infection.

Immunofluorescence and Immunoblot Analysis

Immunofluorescence staining was done on paraformaldehyde-fixed cells with anti-AQP1 primary antibody (Chemicon International, Temecula, CA) and anti-rabbit Cy3-conjugated IgG (Sigma) secondary antibody. For immunoblot analysis, AQP1 antibody and horseradish peroxidase–conjugated secondary anti-rabbit IgG antibody (DAKO, Carpinteria, CA) were used for detection by enhanced chemiluminescence (Amersham Pharmacia, Piscataway, NJ).

Osmotic Water Permeability

Cells that were grown on glass coverslips were incubated with 20 μM calcein acetoxymethyl ester (Molecular Probes, Eugene, OR) for 30 min at room temperature and then mounted in a perfusion chamber designed for rapid solution exchange (<100 ms). The time course of cytoplasmic calcein fluorescence, which changes reciprocally with cell volume (16), was monitored in response to exchanging perfusate between PBS (300 mOsm) and hypertonic PBS (600 mOsm) that contained 300 mM mannitol.

In Vitro Migration and Wound Healing

Migration was assayed using a modified Boyden chamber (Corning Costar, Cambridge, MA) that contained a polycarbonate transwell membrane filter (6.5 mm diameter, 8 μm pore size) (17) coated with 10 μg/ml fibronectin. Cells (10⁴) were plated on the upper chamber in DMEM/F-12 that contained 1% FBS. The lower chamber contained DMEM/F-12 with 10% or 1% (control) FBS. Cells were incubated for 6 h at 37°C in 5% CO₂/95% air. Adherent cells were counted after staining with Coomassie blue, and migrated cells that remained on the bottom surface were counted after nonmigrated cells were scraped from the upper surface of the membrane with a cotton swab.

In vitro wound healing was assayed in confluent cell monolayers. Cells were scraped in an approximately 400-μm-wide strip of the cells using a standard 200-μl pipette tip (10). The wounded monolayers were washed twice to remove nonadherent cells. Wound healing was quantified as the average linear speed of the wound edges over 24 h. Cells were stained with phallotoxins conjugated to Alexa Fluor 488 (Molecular Probes) to count lamellipodia and to visualize the F-actin cytoskeleton. In some experiments, time-lapse photography of the wound edges was performed by phase-contrast microscopy (600 frames over 10 h).

Cell Size, Adhesion, and Proliferation

Cells were suspended in medium after trypsinization for measurement of cell diameter by transmission light microscopy at high magnification (×250). For measurements of cell adhesion, 96-well plates were coated at 4°C overnight with collagen type I (10 μg/ml; BD Biosciences, Bedford, MA), laminin (10 μg/ml, natural mouse laminin; Invitrogen, San Diego, CA), or human fibronectin (10 μg/ml; Roche Diagnostics, Indianapolis, IN). After rinsing, uncoated surfaces were blocked with PBS that contained 2% heat-denatured BSA for 1 h at 37°C. Cells (2 × 10⁴ cells) were added to each well and incubated for 1 h at 37°C. After cells were washed twice with PBS to remove unattached cells, adhered cells were stained with Hoechst 33342 for 1.5 h, and the fluorescence of each well was measured (excitation 360 nm, emission 530 nm) as described (18). Cell proliferation was assayed using a commercial BrdU Cell Proliferation Assay Kit (Calbiochem, La Jolla, CA).

Ischemia-Reperfusion Model of Acute Renal Injury

Mice were anesthetized with ketamine and xylazine, injected intraperitoneally, and maintained at 37°C using a warming pad. The left renal pedicle was identified and clamped for 30 min using a vascular clamp. The clamp was removed, and reperfusion was confirmed visually. The incision was closed by suture, and the mice were allowed to recover. Kidneys were harvested at the specified time points and fixed in formalin for paraffin sectioning or 4% paraformaldehyde for frozen sections. For measurement of proliferation, in some experiments mice received an intraperitoneal injection of 5-bromo-2′-deoxyuridine (BrdU) (100 μg/g body wt; Sigma) in saline 2 h before being killed.

Paraffin-embedded sections were stained with periodic acid-Schiff reagent (Sigma) and hematoxylin. Tubular injury was evaluated in kidney sections at 5 d after ischemia-reperfusion by two investigators who were blinded to genotype information. A minimum of 100 proximal tubules from five mice were examined, and the number of tubules with >5 μm inner diameter (dilated tubules) were counted, based on established procedures (19,20). The percentage of dilated tubules was computed. Cell proliferation was measured in paraffin-embedded sections by staining with BrdU antibody (Abcam, Cambridge, MA) followed by biotinylated anti-mouse IgG and horseradish peroxidase–conjugated ABC reagent (Vector Laboratories, Burlingame, CA). F-actin in frozen sections was stained by Alexa Fluor 488–conjugated phalloidin (Molecular Probes).

Characterization of Proximal Tubule Cell Cultures

Proximal tubule epithelial cells were cultured from kidney cortex of adult wild-type and AQP1 null mice. The primary cell cultures were indistinguishable in their appearance by phase-contrast light microscopy (Figure 1A, left). Megalin immunostaining indicated that >98% of cells in the cultures were of proximal tubule origin (Figure 1A, right). AQP1 immunostaining in cells that were cultured from kidneys of wild-type mice showed AQP1 protein expression in a diffuse intracellular vesicular and plasma membrane pattern, with no staining in cells from AQP1-deficient kidneys (Figure 1B). Infection of AQP1-deficient cells by AQP1-Ad at a multiplicity of infection of 100 PFU/cell produced AQP1 protein expression in approximately 90% of the cells. Immunoblot analysis showed AQP1 protein expression with the expected molecular size of approximately 28 kD in wild-type and AQP1-Ad infected cells (Figure 1C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Characterization of proximal tubule cell cultures. (A, left) Phase-contrast micrographs of primary cultures of proximal tubule cells from wild-type (+/+) and AQP1 null (−/−) mice. (A, right) Immunostaining of cells from wild-type mouse for the proximal tubule cell marker megalin (green, with DAPI-stained blue nuclei). Bar = 20 μm. (B) Aquaporin 1 (AQP1) immunostaining (red). Where indicated, cells were infected with adenovirus encoding AQP1 (AQP1-Ad) as described in Materials and Methods. Bar = 10 μm. (C) AQP1 immunoblot of cell homogenates (10 μg protein/lane). (D) Osmotic water permeability of proximal tubule cells measured by calcein fluorescence quenching. (Left) Representative time course data showing responses to rapid changes in perfusate osmolality between 300 and 600 mOsm. (Right) Reciprocal exponential time constants (τ⁻¹) in five separate sets of measurements (SE, *P < 0.01), which are proportional to osmotic water permeability.

Osmotic water permeability was measured by cell volume changes in response to osmotic challenge using the calcein fluorescence quenching method. Figure 1D (left) shows the reversible time course of cell volume in response to changing perfusate osmolality between 300 and 600 mOsm. The osmotically induced cell volume changes were slower in the AQP1-deficient cells compared with wild-type cells and AQP1-Ad–infected AQP1 null cells. Fitted reciprocal exponential time constants (proportional to water permeability) were reduced by approximately two-fold in the AQP1-deficient compared with wild-type cells (Figure 1D, right). AQP1-Ad infection in AQP1 null cells restored water permeability to approximately that in uninfected wild-type cells. It is interesting that water permeability in AQP1-Ad–infected wild-type cells was similar to that of uninfected wild-type cells despite greater AQP1 protein expression by immunoblot analysis, which may be caused by AQP1 protein mistargeting in the virus-treated cells.

Proximal Tubule Cell Migration In Vitro

In vitro assays of wound healing and transwell migration were done in the primary cultures of proximal tubule cells from wild-type and AQP1 null mice, as well as in AQP1-Ad adenovirus “corrected” cells from AQP1 null mice. Cell migration toward FBS was assayed using a modified Boyden chamber. Adherent cells (before scraping) and migrated cells on the bottom surface of transwell membrane (after scraping) were quantified (Figure 2A). Migration of AQP1-deficient proximal tubule cells toward 10% serum was reduced significantly compared with wild-type cells, with low migration toward 1% serum in both wild-type and AQP1-deficient cells (Figure 2B). AQP1 replacement by AQP1-Ad infection corrected the migration defect in AQP1-deficient cells to approximately that in wild-type cells. Cell size and proliferation were similar in wild-type and AQP1-deficient cells (Figure 2, C and D), indicating that the slower migration in AQP1-deficient cells was not due to a difference in cell size or proliferation. No differences in cell adhesion to collagen I, fibronectin, or laminin were found, suggesting that reduced migration was also not due to altered integrin-dependent adhesion (Figure 2E) (21,22). In support of this conclusion, immunostaining for integrins α3, β1, and β3 were similar in wild-type and AQP1-deficient cells (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

In vitro Boyden chamber assay of proximal tubule cell migration. (A, top) Schematic of modified Boyden chamber that contained a transwell porous membrane filter. (A, bottom) Photographs of adherent proximal tubule cells on transwell filters before scraping (left) and migrated cells (on the bottom filter surface) after scraping (right). Bar = 100 μm. (B) Percentage of migrated cells at 6 h after cell plating (SE, six sets of experiments, *P < 0.01). The bottom chamber contained 10% or 1% FBS. (C) Cell diameter measured in freshly trypsinized proximal tubule cells (SE, 20 cells measured per genotype). (D) Cell proliferation measured by 5-bromo-2′-deoxyuridine (BrdU) incorporation (absorbance at 520 nm; SE, six sets of experiments). (E) Cell adhesion to type I collagen, fibronectin, and laminin measured by fluorescence (SE, six sets of experiments). See Materials and Methods for details. In C through E, differences are NS.

Figure 3A shows delayed wound closure in AQP1-deficient proximal tubule cells at 24 h after creation of linear wounds in confluent cell monolayers. The closure speed of the wound edge was significantly slowed in AQP1-deficient compared with wild-type cells (Figure 3B). Time-lapse phase-contrast microscopy of wound closed showed different migratory behaviors of wild-type versus AQP1-deficient proximal tubule cells (supplemental videos are available online only). During migration the wild-type cells formed broad lamellipodia with persistent movements toward the wound area and more rapid migration compared with the AQP1-deficient cells. The slowed speed of wound closure in the AQP1 null cells was corrected by infection with AQP1-Ad under conditions that in Figure 1D normalized their osmotic water permeability.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

In vitro wound closure model to study proximal tubule cell migration. (A) Light micrographs of wounded cell monolayers showing delayed wound closure at 24 h in AQP1-deficient proximal tubule cells. Bar = 100 μm. (B) Speed of wound edge in wild-type and AQP1 null cells without and with AQP1-Ad infection (SE, six sets of experiments, *P < 0.01). (C, left) Fluorescence micrographs of cells stained with phallotoxins showing protrusions at the leading edge (top, white arrowheads) and the F-actin cytoskeleton (bottom). Bar = 20 μm. (C, right) Area of protrusions per length of cells at the wound edge (SE, 100 cells analyzed, *P < 0.01).

F-actin staining with fluorescence phallotoxins showed cell protrusions (lamella) in many proximal tubule cells during wound closure (Figure 3C, top left), with a smaller number of AQP1-deficient cells having protrusions. Quantitative analysis of lamella, expressed as the ratio of lamella area per length of cell (at the leading edge), showed a significant reduction in lamella area in the AQP1-deficient cells (Figure 3C, right). F-actin staining was similar in areas of wild-type and AQP1 null cells not including lamella, suggesting that delay in wound healing in AQP1-deficient cells was not due to a cytoskeletal abnormality (Figure 3C, bottom left) (21,22).

In Vivo Response to Acute Renal Tubular Injury

A well-established ischemia-reperfusion injury model of acute tubular injury was used to investigate the potential in vivo relevance of impaired proximal tubule cell migration in AQP1 deficiency. Renal architecture was indistinguishable in wild-type versus AQP1 null control mice (Figure 4A, left). Unilateral renal ischemia-reperfusion (30 min) produced marked pathologic changes in the kidney, although there were no clinical signs of disease or electrolyte/renal function abnormalities (because contralateral kidney was not damaged).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Ischemia-reperfusion model of acute renal tubular injury. (A) Periodic acid-Schiff and hematoxylin staining of kidney sections from wild-type and AQP1 null control mice (left) and 5 d after (right) ischemic/reperfusion (I/P). Bar = 20 μm. Arrowhead, dilated tubule; arrow, degenerated tubule. (B) Percentage of proximal tubules with inner diameter >5 μm (SE, >100 proximal tubules examined per mouse, five mice per group, *P < 0.05). (C) F-actin staining kidneys of wild-type and AQP1 null control mice (left) and just after (middle) and 3 d after (right) I/P. Bar = 20 μm. (D) BrdU staining at 5 d after I/P. Arrow, BrdU-positive staining. Bar = 50 μm.

Figure 4A (right) showed remarkably greater tubule dilation and degeneration in AQP1 null versus wild-type kidneys at 5 d after ischemia-reperfusion. The percentage of proximal tubules with inner diameter >5 μm, which has been taken as a semiquantitative index of proximal tubule damage (19,20), was remarkably greater in AQP1 null mice (Figure 4B). Immunostaining showed F-actin expression concentrated at the brush border in control kidneys from wild-type and AQP1 null mice (Figure 4C, left). Just after ischemia-reperfusion, nearly all tubules in both types of mice were thinned and dilated, with disappearance of the brush border and a discontinuous F-actin distribution. Although the F-actin component of the brush border was largely normalized in kidneys of wild-type mice at 3 d after ischemia-reperfusion, kidneys of AQP1 null mice showed poor F-actin organization. At 5 d after ischemia-reperfusion, the percentage of proximal tubules with at least one BrdU-positive cell was similar in wild-type and AQP1 null mice (4.5 ± 1.6 versus 4.7 ± 0.9%; n = 5 slides; Figure 4D), indicating a similar proliferation rate.