Extracellular glutathione peroxidase (eGPx) is a secreted selenoenzyme with GPx activity. eGPx protein and activity are found in blood plasma and other extracellular fluids. eGPx in plasma is predominantly derived from the proximal tubules of kidneys in humans. Two types of human proximal tubule cells were cultured on semipermeable polycarbonate membranes to determine whether these cells secrete eGPx in a polarized direction. Immortalized human proximal tubule HK-2 cells and primary human proximal tubule cells formed confluent monolayers when cultured on these membrane inserts in culture dishes, as evidenced by transepithelial resistance. Both cell lines also constituted a barrier to diffusion of a fluoresceinated dextran of 75 kDa, a size similar to eGPx homotetramers. In both cell lines, 6- to 12-fold more35S-methionine-labeled eGPx was immunoprecipitated from the basolateral media than from the apical media, indicating basolateral secretion of eGPx. eGPx was immunolocalized to the extracellular fluid at the basolateral surface of proximal tubules in human kidney. These data support the conclusion that eGPx is secreted through the basolateral membrane of human kidney proximal tubule cells into the extracellular fluid of the kidney, and from there enters blood plasma.
- antioxidant enzymes
the family of selenium-dependent glutathione peroxidases (GPx) consists of four known members. The similarities among the different GPx include1) a dependency on selenium for enzymatic activity (29); 2) the presence of a selenocysteine residue inserted during protein translation that confers enzymatic activity (17); 3) a ping-pong enzymatic reaction in which the selenium in selenocysteine reduces peroxides (7); and 4) oxidation of selenium, followed by glutathione-mediated reduction. There are significant differences between these members of the GPx family. The first described GPx is a cytosolic enzyme thought to be present in all tissues, exists as a homotetramer with an apparent subunit size of 22 kDa, and is called cGPx (or GPx1). cGPx can reduce both hydrogen peroxide and organic peroxides but apparently has a distinct preference for glutathione as the reductant (7). Phospolipid GPx (PLGPx or GPx2) is so named because of its ability to reduce phospholipid hydroperoxides and is distributed in both the membrane and the cytosolic fractions as a monomer (13, 34). Plasma GPx was originally described as distinct from cGPx based on enzymatic and immunological properties and is known as extracellular GPx (eGPx or GPx3) (5, 15, 25, 32,33). In studies in humans and mice, eGPx has been found in blood plasma (32), breast milk (5), in fluid lavaged from lung (1), and in amniotic and exocoelomic fluid surrounding the developing mouse embryo (22). eGPx exists as a homotetramer with a subunit size of 23 kDa on SDS-PAGE and a predicted subunit size of 25.3 kDa based on its cDNA sequence (31). The difference between these experimental and calculated subunit sizes is probably due to the cleavage of the consensus NH2-terminal 24-amino acid signal sequence characteristic of secreted proteins, as predicted using the SignalP V2.0 program (28). eGPx in vitro reduces organic hydroperoxides, phospholipid hydroperoxides, and hydrogen peroxide (32, 37). The fourth member of the family is an additional cytosolic GPx that is found predominantly in the epithelial cells of the gastrointestinal tract (giGPx or GPx4) and appears to have enzymatic properties similar to cGPx (10, 16).
eGPx is synthesized and secreted by some cell lines, including the Caki-2 kidney cell line and the Caco2 colon carcinoma cell line (3). Transcripts for eGPx are found at high levels in the S1 and S2 segments of the proximal tubules of the kidney in humans and mice (4, 22, 26), and in somewhat lesser amounts in lung, heart, and intestine (4, 11, 22, 41). In addition, mRNA for eGPx is found at high levels in a spatially restricted pattern in mouse uterus and deciduum during pregnancy and in the fetus during development (22). The eGPx in plasma is derived predominantly from the kidney, because anephric individuals have very low plasma GPx activity and plasma eGPx protein. Plasma eGPx activity increases after kidney transplantation in these patients (4, 35,40).
To date there has not been a direct demonstration of the synthesis and secretion of eGPx by the proximal tubule cells of the kidney. This report presents data to show that proximal tubule cells make and secrete eGPx. These cells can be grown in Transwell cultures in a system allowing study of the polarized secretion of newly synthesized eGPx, which in turn allows us to determine that eGPx is predominantly secreted basolaterally. In addition, this report demonstrates the extracellular location of eGPx in the interstitial space of the kidney, consistent with the Transwell data.
MATERIALS AND METHODS
Cells and media.
Two types of cells derived from proximal tubules of the kidneys were utilized in these studies. The HK-2 cell line was obtained from Dr. R. Zager at the University of Washington, who derived the cells from normal human proximal tubules and cultured and immortalized them by transduction with human papilloma virus genes E6/E7 (30). The cells were originally cultured in DMEM-based defined media containing epidermal growth factor. For these studies, cells were routinely cultured in DME/F-12 medium containing 15% fetal bovine serum (FBS). Primary human renal proximal tubule (RPT) cells were obtained from Clonetics (San Diego, CA). RPT cells were cultured in defined medium containing 1.0 μg/ml hydrocortisone, 20 ng/ml human epidermal growth factor, 1.0 μg/ml epinephrine, 13 ng/ml triiodothyronine, 20 μg/ml transferrin, 10 μg/ml insulin, and 0.5% FBS obtained from the vendor. The Madin-Darby canine kidney (MDCK) cell line (catalog no. CCL-34) was obtained from the American Type Culture Collection (Rockville, MD) and was cultured in MEM containing 10% FBS.
Growth of cells on semipermeable Transwells.
Cells that had been cultured in tissue culture flasks were trypsinized, enumerated, and suspended in a modified MEM containing 5 μM calcium (low-calcium medium) and dialyzed FBS (dFBS) (36). Cells were added to the insert of polycarbonate Transwells obtained from Costar/Corning at a concentration of 3 × 105cells/cm2 membrane area of the insert. For a 6-well Transwell plate, this corresponded to a volume of 1.5 ml containing 2 × 106 cells in the insert, whereas 2.0 ml of medium were added to the well. After 4 h at 37°C in a humidified incubator with 5% CO2 atmosphere, the low-calcium MEM/dFBS was exchanged for MEM/FBS containing normal concentrations of calcium (1.26 mM).
Biotinylation of surface proteins of HK-2 cells.
HK-2 and MDCK cells that had been cultured on Transwells for 1 wk were rinsed with PBS containing calcium and magnesium (PBS/CM). Sulfosuccinimidyl-6-(biotinamido)hexanoate (sulfo- NHS-LC-biotin, Pierce, Rockford, IL) was dissolved in DMSO at 200 mg/ml. Apical proteins were labeled by adding 0.25 ml sulfo-NHS-LC-biotin (1.0 mg/ml in cold PBS/CM) to the Transwell insert and transferring the insert to the Transwell plate already containing 1.5 ml ice-cold PBS/CM. Basolateral proteins were labeled by adding 1.0 ml ice-cold PBS/CM to an insert and placing the insert on a 0.25-ml drop of sulfo-NHS-biotin/PBS/CM on parafilm. After 30 min of labeling at 4°C, the filters were washed four times with PBS/CM, excised, and extracted with PBS containing 1 mM EDTA and 1% Triton X-100. The protein concentration of this lysate was determined using a modified Lowry assay kit from Bio-Rad. The proteins in the lysate were precipitated with 5 vol of cold acetone and solubilized in SDS sample buffer. Samples (10 μg protein) were separated on 10% acrylamide-SDS gels (24) and transferred to a polyvinylidene difluoride membrane. The blot was blocked with PBS containing 2% BSA (wt/vol), incubated with streptavidin-horseradish peroxidase (1 μg/ml), and visualized with chemiluminescent substrate (Pico-Western, Pierce).
Transepithelial resistance of cells cultured on Transwells.
The transepithelial resistance of cells cultured on Transwells was measured with a Millicell-ERS instrument from Millipore (Bedford, MA) according to the manufacturer's instructions. Polycarbonate Transwells (2.5-cm diameter, 4.9-cm2 surface area) were treated with rat tail tendon collagen for 1 h at room temperature, after which they were sterilized by exposure to ultraviolet light for 2 h in a tissue culture hood. Cells were seeded as above in low-calcium MEM/dFBS for 4 h, following by a change to MEM/FBS containing 1.26 mM calcium. Resistance was measured every day. Transepithelial resistance of the monolayer was then calculated by subtracting the resistance of Transwells without cells from the resistance of Transwells with cells and multiplying the difference by the surface area of the Transwell, with the units of the results being ohms times square centimeters (14).
Transepithelial diffusion of fluorescent dextran on Transwells.
Fluorescein-labeled dextran (molecular mass 75 kDa) was obtained from Molecular Probes (Eugene, OR). Cells were cultured on Transwells as described above. After 5 days of culture, the Transwell inserts and culture wells were washed three times with Hanks' buffered salt solution (HBSS) containing glucose, calcium, and magnesium. A solution of HBSS containing 1 μg/ml fluorescein-labeled dextran was added to the apical Transwell inserts only, and HBSS was added to the basolateral compartment. After 2 h, the apical and basolateral media were collected. Fluorescence of dilutions (1:5–1:50) of the apical, basolateral, and starting media was determined in a PerkinElmer F 4010 spectrofluorometer (Norwalk, CT), with excitation and emission wavelengths of 480 and 525 nm, respectively.
Metabolic labeling of cells cultured on Transwells.
Cells that had been cultured on collagen-coated Transwells for 5 days, as described above, were metabolically labeled with35S-Express labeling reagent obtained from NEN (Boston, MA). Cells were rinsed twice with HBSS and then incubated in methionine- and cysteine-deficient MEM/dFBS for 30 min at 37°C. At that time, the Transwells were removed from the multiwell culture dish and placed on 0.1 ml methionine- and cysteine-deficient MEM/dFBS containing 0.1 mCi 35S (labeling medium) on a piece of parafilm in a petri dish. The apical side of the cells contained 1.5 ml of the same medium without 35S. The Transwells were incubated in this manner at 37°C. After 30 min, the Transwells were transferred back to the multiwell plates, and the labeling medium on the parafilm was quickly scavenged with two washes of medium and transferred to the basolateral compartment of the Transwell plates, such that the final volume was 2.0 ml. The Transwell plates were then returned to the 37°C incubator. Throughout the experiment, the35S-Express reagent in the basolateral compartment was the only exogenous source of methionine. At various times, Transwells were removed, and the cell-free media from the apical and basolateral compartments were saved for immunoprecipitation.
Immunoprecipitation of 35S-labeled eGPx.
Immunoprecipitation and analysis of 35S-labeled eGPx from the culture media were performed by modification of previous methods (6). The culture medium was centrifuged at 1,000g for 5 min. The cell-free supernatant was brought to 1 mM iodoacetic acid, 1 mM phenylmethylsulfonyl fluoride (PMSF), 0.1% SDS, and 1 mM N-ethylmaleimide from concentrated stock solutions and stored at −80°C until it was processed further. After thawing, the media were incubated with 0.3 mg of normal rabbit IgG plus 100 ml Pansorbin cells from Calbiochem (San Diego, CA) for 1 h at 4°C. After centrifugation, the cleared supernatants were incubated with specific rabbit anti-human eGPx antibodies (raised against human eGPx purified from plasma), the properties of which have previously been described (32). A 50% slurry of protein A-Sepharose CL4B was prepared and incubated with rabbit anti-eGPx for 2 h at 4°C. After being washed in PBS, an aliquot (0.1 ml of 50% slurry containing 0.3 mg anti-eGPx) of the protein A-anti-eGPx slurry was added to each sample of medium and gently rocked for 1.5 h at 4°C. The resulting immunoprecipitates were washed with high-stringency buffer [(in mM) 15 Tris, 5 EDTA, 2.5 EGTA, 120 NaCl, 25 KCl, and 1 PMSF, pH 7.5, as well as 1% Triton X-100, 1% deoxycholate, and 0.1% SDS] for 1 h at 4°C. The suspensions were then underlaid with 10% sucrose and centrifuged at 4°C for 5 min at 1,000 g. The immunoprecipitates were sequentially washed in high-salt buffer [(in mM) 15 Tris, 5 EDTA, 2.5 EGTA, and 1 PMSF, pH 7.5, as well as 1% Triton X-100, 1% deoxycholate, 0.1% SDS, and 1 M NaCl] and low-salt buffer (15 mM Tris and 5 mM EDTA). The beads with immunoprecipitated proteins were then boiled in SDS-PAGE sample buffer containing 100 mM dithiothreitol and electrophoresed on SDS-polyacrylamide gels (12.5% acrylamide) (24). After electrophoresis, the gels were equilibrated in Amplify autoradiofluorography-enhancing solution (Amersham, Piscataway, NJ), dried, and exposed to Kodak X-Omat AR X-ray film for 2 wk. The films were scanned on a densitometer, and the amount of 35S-eGPx was quantified using Quantity One software from Bio-Rad (Emeryville, CA). Calculations were performed on the raw densitometry data, while the images of the films in the figures were enhanced for brightness and contrast with Photoshop software from Adobe (San Jose, CA).
Normal sections of human kidney removed because of a kidney tumor were fixed in paraformaldehyde and embedded in paraffin by standard techniques. Sections (5 μm) were prepared on polylysine-coated glass slides. After being dewaxed in xylene and hydrated through a series of ethanol washes, the slides were incubated for 30 min with 0.3% H2O2. The slides were then incubated for 1 h with either of the following antibody solutions at a concentration of 10 μg/ml in PBS: 1) normal rabbit IgG (Sigma, St. Louis, MO); 2) two affinity-purified anti-human eGPx IgGs raised against the peptide sequences CGFVSQRGQEDSKMD-amide and CLSYMRRQAALGVKRK-COOH, both conjugated to keyhole limpet hemocyanin; or3) two affinity-purified anti-human cGPx IgGs raised against the peptide sequences QSVYAFSARPLAGGEPVC-amide and DIEPDIEALLSQGPSC-amide, conjugated to keyhole limpet hemocyanin. After several washes with PBS, the slides were then incubated for 1 h with biotin-sp-conjugated, affinity-purified donkey anti-rabbit IgG (cleared by the manufacturer by incubation with human serum) from Accurate Scientific (Westbury, NY) at a 1:2,500 dilution in PBS. After several washes with PBS, the slides were incubated with streptavidin-horseradish peroxidase at a 1:100 dilution (Calbiochem, La Jolla, CA) in PBS. After being washed, the slides were incubated with a nickle-enhanced diaminobenzidine chromagen kit according to the manufacturer's instructions (Biomeda, Foster City, CA) and counterstained with hematoxylin.
Differential labeling of surface proteins of HK-2 cells cultured on Transwells.
The HK-2 cell line was developed from primary cultures of normal human proximal tubule cells and had been immortalized by transfection with the human papilloma proteins E6/E7 (30). The HK-2 cells (and MDCK cells, for comparison purposes) were cultured on semipermeable Transwell inserts, with a change of medium daily. The surface proteins accessible to a membrane-impermeable biotinylating reagent added to either the apical compartment (the Transwell insert) or the basolateral compartment (Transwell plate) were visualized after SDS-PAGE (Fig. 1; equal amounts of total protein were loaded onto each lane). The pattern of proteins biotinylated from the apical and basolateral membranes is similar in HK-2 cells, with the more easily seen differences attributable to the amount of individual bands. For example, the proteins at ∼81 and 44 kDa are present in higher amounts in the basolateral membrane in HK-2 cells (open arrowheads), whereas the protein at 53 kDa is present in equal amounts in the apical and basolateral membranes (filled arrowhead). For comparison purposes, the biotinylation pattern of MDCK cells that were also cultured on semipermeable membranes is also presented. In these cells, the differences in the patterns of biotinylated proteins are also subtle. For example, the protein at 69 kDa is present in equal amounts in the apical and basolateral membrane (filled arrowhead). The amounts of protein at 31 and 57 kDa are greater in the apical membrane, whereas the protein at 50 kDa is present in greater amounts in the basolateral membranes (open arrowheads). Thus these differential biotinylation patterns suggest that the apical and basolateral membranes of HK-2 cells cultured on Transwells possess different amounts of some proteins in a manner consistent with formation of a polarized monolayer.
Transepithelial resistance of cells cultured on Transwells.
The transepithelial resistance of HK-2 and MDCK cells cultured on collagen-coated Transwells (2.5-cm diameter) was measured, and the results are shown in Fig. 2. The resistance of Transwells without cells was ∼80 Ω throughout the culture period, whereas the uncorrected resistance of Transwells with HK-2 cells increased to ∼100–110 Ω over 1 wk of culture. After correction for the resistance of Transwells without cells, and correction for the surface area of the Transwell inserts, the transepithelial resistance of HK-2 cells stabilized at ∼100 Ω · cm2 after 1 wk of culture. The MDCK cells (a cell line that has characteristics of distal tubules and collecting ducts) developed a transepithelial resistance that stabilized at ∼400 Ω ohm · cm2. The increase in transepithelial resistance of HK-2 cells over 1 wk of culture on the Transwells is consistent with their development of a polarized monolayer on the inserts.
Transepithelial dextran diffusion in Transwells with cultured cells.
To further assess the integrity of monolayers of cells grown on Transwells, we measured the diffusion of a macromolecule with a molecular mass similar to the molecular weight of the eGPx homotetramer. The molecular mass of the human eGPx homotetramer is ∼88 kDa, and the fluorescein-labeled dextran used in this experiment has a molecular mass of 75 kDa. In this experiment, we measured the diffusion of fluorescein-labeled dextran in control Transwells without cells, Transwells with the HK-2 cell line, those with the primary culture of RPT cells, and those with MDCK cells. All Transwells had been coated with collagen before the start of culture. After 5 days of culture, the media were exchanged for HBSS, and fluorescein-labeled dextran was added to the apical compartment only. As can be seen in Table 1, the initial rates of diffusion of HK-2 and RPT cells were 0.13 and 0.33%/h, respectively. The rates of diffusion for HK-2 and RPT cells are 3.0 and 7.6%, respectively, of the rate of diffusion in Transwells with no cells, indicating a permeability barrier to macromolecules of the size of eGPx. The initial rate of diffusion of fluoresceinated dextran by monolayers of MDCK cells was lower than in the proximal tubule cells (0.5% of the rate of diffusion when no cells were present). Thus human proximal tubule cells can be cultured on Transwells and effectively provide a barrier to the diffusion of macromolecules. We used this culture system to determine whether eGPx is secreted by these cells cultured on Transwells and whether eGPx is secreted in a polarized fashion.
Polarized secretion of eGPx by proximal tubule cells.
The HK-2 cell line was cultured on Transwells for 5 days and then metabolically labeled as described in materials and methods. The cell-free media obtained from the apical and basolateral compartments were quantitatively saved and subjected to immunoprecipitation with rabbit anti-human eGPx IgG. These polyclonal antibodies have previously been defined to be specific for eGPx and do not immunoprecipitate human cGPx (6, 32). Figure3 A shows autoradiofluorographs of immunoprecipitated, radiolabeled eGPx from the apical and basolateral compartments of HK-2 cells at different times of metabolic labeling. It is evident that most of the radiolabeled eGPx is found in the basolateral compartment. The bands were quantified by densitometry, and the data are presented in Fig. 3 B. The amount of35S-eGPx was greater in the basolateral compartment than in the apical compartment at all time points. The basolateral-to-apical ratio of the total amount of eGPx is shown in Fig. 3 C. It can be seen that the ratio declines with time, probably due to some cell-independent diffusion of 35S-eGPx from the basolateral to the apical compartment. By extrapolating back to time 0, a theoretical basolateral-to-apical ratio of eGPx secretion of 12 is obtained.
The same experiment was also performed using primary cultures of human proximal tubule cells grown on Transwells. As can be seen in the autoradiofluorograph shown in Fig.4 A, 35S-eGPx is also predominantly immunoprecipitated from the basolateral compartment. The total amounts of 35S-eGPx in the different media (apical and basolateral) as measured by densitometry were determined. As for HK-2 cells, there was more 35S-eGPx in the basolateral compartment than in the apical compartment at all time points. It can be seen in Fig. 4 B that the basolateral-to-apical ratio of immunoprecipitated 35S-eGPx decreases with time. As was done for HK-2 cells, back extrapolation of the ratios to time 0 yields a theoretical basolateral-to-apical ratio of immunoprecipitated 35S-eGPx of ∼6. Similar results are obtained whether the samples are visualized by autoradiofluorography (shown in Figs. 3 and 4) or by phosphorimaging using a Bio-Rad GS505 PhosphorImager (not shown). Thus, in both the immortalized HK-2 cell line and RPT cells, eGPx is predominantly secreted through the basolateral domain of the cells.
Immunohistochemical detection of eGPx in normal sections of human kidney was performed using affinity-purified rabbit antibodies that were raised against peptides specific for human eGPx. For comparison purposes, similar sections of the same human kidney tissue were also probed for cGPx using affinity-purified rabbit antibodies that were raised against peptides specific for human cGPx. These antibodies were demonstrated to be mutually non-cross-reactive by Western blot analysis of human plasma and red blood cell hemolysate. The rabbit anti-human eGPx antibodies detected the 23-kDa eGPx band in plasma but did not detect a band in the hemolysate (Fig.5 A). In contrast, the rabbit anti-human cGPx antibodies detected the 22-kDa cGPx band in the hemolysate but did not detect a band in plasma.
Figure 5 B shows that the anti-eGPx antibodies lightly stain the cytoplasm of cells in proximal tubules in the kidney in a pattern similar to that found by in situ hybridization of eGPx transcripts in proximal tubules, as reported previously (4, 22). A darker pattern of staining of anti-eGPx antibodies is seen in the interstitial space just outside the basolateral membrane of tubule cells, consistent with its localization in the interstitial fluid. eGPx is not detected in the glomerulus, consistent with the absence of eGPx transcripts in the glomerulus by in situ hybridization.
cGPx is detected by anti-cGPx antibodies predominantly in the cytoplasm of the tubular epithelial cells of human kidney, as seen in Fig.5 C. In contrast to the extracellular localization of eGPx, the extracellular space shows no staining with these anti-cGPx antibodies. In a comparison of Fig. 5, B and C, it is apparent that eGPx is present at low levels and cGPx at high levels in the cytoplasm of the same types of tubular epithelial cells. In contrast, only eGPx is found in the extracellular space, consistent with its presence in interstitial fluid. Like eGPx, cGPx is virtually absent from the glomerulus.
The first data to indicate that eGPx was distinct from cGPx were based on immunological and biochemical characterizations of the purified proteins (32, 33). It is now known that the amino acid sequences of eGPx and cGPx are only 45% identical (31). The predicted amino acid sequence of eGPx based on the cDNA sequence indicates the presence of a signal sequence of 24 amino acids, targeting the enzyme for secretion. The sites of synthesis of eGPx have been hypothesized to include kidneys, lungs, gastrointestinal tract of adult mammals (humans, rats, and mice) (2, 4, 11, 22, 26); the deciduum of pregnant mice; the visceral endoderm surrounding mouse embryos (22); and the developing kidneys, heart, skin, and lungs of mouse fetuses (22), all based on the presence of eGPx mRNA transcripts.
In humans, all GPx activity in plasma is attributable to eGPx, based on precipitation of all GPx activity by antibodies specific for eGPx. GPx activity in the plasma of anephric humans is only 5–25% of normal, indicating that the kidney is the main source of plasma eGPx activity. The low plasma GPx activity in these anephric individuals is entirely due to eGPx, based on specific precipitation of all remaining enzyme activity. Last, plasma GPx activity rises in anephric humans after kidney transplantation and is again entirely due to eGPx (4, 35, 38). The site of synthesis of eGPx transcripts in the kidney has been localized to the proximal tubules in human kidney (4) and further localized to the S1 and S2 segments in mouse kidney (22).
In this study, we wished to determine whether proximal tubule cells that are cultured on Transwells acquire a polarized phenotype. Furthermore, we wished to determine whether eGPx was secreted in a polarized fashion by these proximal tubule cells. Both the immortalized human proximal tubule cell line HK-2 and primary human proximal tubule RPT cells cultured on Transwells developed a polarized monolayer of cells. For these studies, our criteria for a polarized Transwell culture were 1) an increase in transepithelial resistance over time; and 2) the establishment of a diffusion barrier for a macromolecule of similar molecular size to the eGPx homotetramer. Over a period of 1 wk, the HK-2 cells cultured on 2.5-cm Transwells developed a transepithelial resistance of ∼100 Ω · cm2 compared with a transepithelial resistance of 400 Ω · cm2 for the commonly used MDCK cell line, the characteristics of which closely resemble distal tubule or collecting duct (39). In addition, both HK-2 and RPT cells cultured on Transwells developed a diffusion barrier to a fluoresceinated dextran with a molecular mass of 75 kDa, although neither of these proximal tubule cells provided a barrier as impermeable as MDCK cells. However, by these criteria, HK-2 and RPT cells cultured on Transwells were suitable for determining whether these proximal tubule cells synthesized and secreted eGPx in a polarized manner.
The demonstration of polarized secretion of newly synthesized eGPx into the basolateral compartment now provides a third criterion that both the HK-2 and the RPT cells formed a polarized monolayer when cultured on Transwells. It is not clear from our data whether the small amounts of eGPx found in the apical compartment are due to true apical secretion, to diffusion of eGPx from the basolateral to the apical compartment through imperfectly formed monolayers, or some other phenomenon such as basolateral-to-apical transcytosis of eGPx. These data indicate that newly synthesized eGPx is predominantly found in the basolateral media. Proximal tubule cells have also been shown to secrete interleukin (IL)-8 and monocyte chemoatactic peptide 1 predominantly to the basolateral medium (23), with an apparent basolateral-to-apical ratio of 2:3 at early time points (24 h; estimated by inspection of their data).
Further evidence of basolateral secretion of eGPx was obtained by immunohistochemistry of eGPx in sections of human kidney. The predominant localization of eGPx in kidneys appeared to be in the extracellular space near the basolateral membranes of tubular cells, consistent with a localization in the interstitial fluid. The tubular epithelia had a lighter staining of eGPx, whereas the glomeruli were negative for eGPx. The localization of eGPx contrasted with that of cGPx. The tubular epithelia of the kidney were the predominant site of cGPx staining, with none in the extracellular fluid or interstitial space. As was the case for eGPx, cGPx was not found in the glomeruli. The lack of eGPx protein in the glomeruli is consistent with our previous reports of the lack of eGPx transcripts in the glomeruli (4, 22). Similarly, the lack of cGPx in the glomeruli is consistent with our observation that cGPx transcripts are absent in the glomeruli of mouse kidney (data not shown). In an investigation of antioxidant enzymes in rat kidney, Gwinner et al. (19) have reported that isolated glomeruli have much lower levels of superoxide dismutase, catalase, and GPx activity than isolated proximal tubules. Our immunohistochemical data complement those data and indicate a virtual absence of both eGPx and cGPx in the glomerulus of human kidney.
Localization of eGPx protein in the interstitial fluid suggests that the eGPx is secreted through the basolateral membrane of the proximal tubule epithelial cells in the kidney, in agreement with the Transwell experiments. Because most plasma eGPx is derived from the kidney, the eGPx in the interstitial fluid of the kidney is probably the source of plasma eGPx. If release of eGPx from proximal tubules did not occur in a polarized manner, the process of eGPx synthesis and delivery to plasma would be difficult to explain. The alternative scenario of delivery of equal amounts of eGPx into the tubule lumen and into the peritubular space would seem inefficient. An even more inefficient scenario of polarized release into the tubule lumen with subsequent reabsorption is not consistent with these data. Furthermore, GPx activity is not found in human urine, nor does human urine contain inhibitors of plasma GPx activity (4). The most likely conclusion to be drawn from these results is that eGPx is secreted predominantly (if not exclusively) through the basolateral membrane of proximal tubule cells.
Although eGPx activity from different extracellular fluids is easily measured in vitro, it has been more challenging to hypothesize how eGPx could function in these fluids. The concentration of glutathione in the plasma is low compared with that in cells (1–5 μM in plasma vs. 1–10 mM in cells), and the Michaelis-Menten constant for GSH is 5.8 mM (32). The concentration of glutathione in the renal interstitial fluid is not known, but it is 200–400 μM in the lung epithelial lining fluid (12), another site of eGPx activity. In addition, although eGPx is presumed to work as part of a traditional glutathione cycle consisting of glutathione, NADPH, plus glutathione reductase, recent studies have suggested that a reductant other than glutathione may be involved with eGPx cycling. Bjornstedt et al. (8, 9) have shown that thioredoxin, in conjunction with thioredoxin reductase plus NADPH, is capable of cycling eGPx as an antioxidant enzyme (8, 9). However, the concentration of both of these molecules is very low in blood plasma. The physiological activities of eGPx in extracellular fluids will probably be determined by genetic methods. Studies in mice that overexpress eGPx suggest a protective extracellular antioxidant activity for eGPx. Mice that overexpress eGPx (with 2-fold more plasma GPx activity) are protected against both renal ischemia-reperfusion injury (an injury to the kidney) and acetaminophen toxicity (an injury to the liver) (21, 27).
Polarized epithelial cells have been demonstrated to secrete proteins into both the basolateral and apical compartments. Human retinal pigment cells secrete IL-6 and IL-8 through the basolateral membrane (20). Human lung epithelial cells also secrete fibrinogen through the basolateral domain (18). Human endometrial epithelial cells secreted vascular endothelial growth factor through the apical domain, under conditions in which fibronectin was not secreted in a polarized direction. Thus the directional secretion of some proteins depends on the cell type. For example, whereas human endometrial cells secrete vascular endothelial growth factor through the apical membrane, retinal pigmented epithelial cells secrete the same protein only through the basolateral membrane. Thus there are cell-specific mechanisms of targeting the same proteins for secretion through different membranes. Further study of the secretion of the basolateral secretion of eGPx seems warranted because this appears to be the first demonstration of the secretion of an enzyme through the basolateral membrane. Such polarized delivery to the interstitial fluid of the kidney prompts interest in its possible activity in the interstitial space of the kidney. It would also be important to determine whether the same directionality occurs in lung or intestinal epithelial cells, other sites of eGPx synthesis and secretion.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant 2RO1-DK-33231.
Address for reprint requests and other correspondence: J. C. Whitin, Pediatrics/Rm. S-308, Stanford University, 300 Pasteur Dr., Stanford, CA 94305-5208 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published February 19, 2002;10.1152/ajprenal.00014.2001
- Copyright © 2002 the American Physiological Society