Enzymes of the UDP-glucuronosyltransferase (UGT) superfamily catalyze the covalent linkage of glucuronic acid, derived from UDP-glucuronic acid (UDPGA), to typically lipophilic substrates containing a carboxylic acid, hydroxyl or amine functional group (Miners and Mackenzie, 1991). Thus, glucuronidation is an important elimination mechanism for numerous drugs, environmental chemicals, and endogenous compounds (e.g., bilirubin, fatty acids, eicosanoids, and hydroxysteroids) in humans (Miners and Mackenzie, 1991; Tsoutsikos et al., 2004). UGTs have been classified into two families, UGT1 and UGT2, and of the human UGT proteins identified to date, 16 have the capacity to catalyze the glucuronidation of endogenous compounds and/or xenobiotics: UGT 1A1, 1A3, 1A4, 1A5, 1A6, 1A7, 1A8, 1A9, 1A10, 2A1, 2B4, 2B7, 2B10, 2B15, 2B17, and 2B28 (Miners et al., 2004). Individual UGTs exhibit distinct but overlapping substrate selectivities and differ in terms of regulation of expression. Age, diet, disease states, induction, and inhibition by chemicals, ethnicity, genetic polymorphism, and hormonal factors are known to influence UGT activity (Miners and Mackenzie, 1991; Miners et al., 2004). Tissue-specific expression is also a feature of UGTs, and although most UGTs are expressed in liver, only UGT 1A3, 1A6, 1A9, 2B4, 2B7, 2B10, 2B11, 2B15, and 2B17 have been identified in kidney (Tukey and Strassburg, 2000).

The structural diversity of the human kidney underlies its multiple functions. These include excretion of polar chemicals and metabolites, and the regulation of fluid, electrolytes, blood pressure, and hormone secretion. Subdivided anatomically into the cortex and medulla, the functional unit of the kidney is the nephron, which consists of the renal corpuscle (glomerulus and Bowman's capsule), proximal convoluted tubule (PCT), loop of Henle (LOH), and the distal convoluted tubule (DCT). Although incorrect in a strict anatomic sense, the term “nephron” commonly also includes the collecting duct (CD). This heterogeneity has made defining the cellular localization of UGT in human kidney difficult; consequently, studies are limited. UGTs have been reported as occurring exclusively in epithelial cells of the PCT (Murray and Burke, 1995), but they are absent in the glomerulus, DCT, vasculature, and collecting tubules of the renal cortex (Peters et al., 1987).

In addition to endogenous synthetic and metabolic capability, the human kidney metabolizes a variety of drugs, including NSAIDs (McGurk et al., 1998; Bowalgaha and Miners, 2001; Soars et al., 2001). Metabolism is the principal route of NSAID elimination in vivo, and only a small fraction of the administered dose is excreted unchanged in urine (Murray and Brater, 1993). Like other carboxylic acids, NSAIDs are metabolized largely as glucuronides, either of the parent drug or oxidative metabolites. Of the UGTs identified in the kidney, UGT1A9, UGT2B7, and, to a lesser extent, UGT1A3 are the predominant NSAID-glucuronidating forms (Sakaguchi et al., 2004; Kuehl et al., 2005). NSAID acyl-glucuronides are electrophilic, and they bind covalently to plasma and tissue proteins both in vitro and in vivo (Spahn-Langguth and Benet, 1992).

It is generally assumed that the metabolic clearance of drugs is determined principally, if not exclusively, by hepatic metabolism and that the kidneys, with comparatively smaller organ mass, contribute minimally to systemic clearance. Interestingly, the intrinsic clearances (CL_int) of the glucuronidated substrates 4-methylumbelliferone and mycophenolic acid by the human kidney are comparable with those of human liver (Bowalgaha and Miners, 2001; Olyaei et al., 2003; Tsoutsikos et al., 2004). In addition, glucuronidation of the anesthetic propofol is 4-fold greater by human kidney than human liver (McGurk et al., 1998). However, there is little information on the localization of UGTs in human renal cortex and medulla, and the contribution of individual renal UGTs to drug (including NSAID) glucuronidation remains poorly defined. This lack of knowledge precludes a complete understanding of NSAID glucuronidation and the relationship between renal drug glucuronidation and target organ toxicity.

Here, we report the immunohistochemical localization of UGT1A proteins and UGT2B7 in cortical and medullary sections of normal human kidney, and characterization of the kinetics of S-naproxen glucuronidation by human kidney cortical microsomes (HKCM) and human kidney medullary microsomes (HKMM) to ascertain the relative contribution of each to renal metabolism. S-Naproxen was chosen as the probe substrate since glucuronidation is the principal route of elimination in humans, and it is estimated to be responsible for 60% of in vivo clearance (Upton et al., 1980; Vree et al., 1993); it is a substrate for both UGT2B7 and UGT1A enzymes (Bowalgaha et al., 2005); it forms an acyl glucuronide; and it has been implicated in the development of renal toxicity (Whelton et al., 2003).

Materials and Methods

Materials

Polyclonal antipeptide antibodies, raised in rabbits, to human UGT1A proteins and UGT2B7, and human UGT1A1 and UGT2B7 “standards” were obtained from BD Biosciences (Sydney, Australia). The antibodies (20-μl aliquots) were treated initially with sodium azide [0.1% (v/v)], and they were stored at –70°C. Normal rabbit immunoglobulin (NRIgG) and the EnVision+ peroxidase (goat anti-rabbit) detection system were purchased from Dako Denmark A/S (Glostrup, Denmark), and S-naproxen and normal goat serum (NGS) were from Sigma-Aldrich (Sydney, Australia). All other chemicals and reagents were of the highest analytical grade available.

Human Kidney Tissue Collection and Processing

Human kidney cortical and medullary tissue from 10 male subjects (K2–K11) undergoing radical nephrectomy for malignant disease was obtained from the joint Flinders Medical Centre/Repatriation General Hospital Tissue Bank (Adelaide, South Australia). Approvals for the collection, use of the kidney tissue, and kidney donor details were provided by the Research and Ethics Committee of the Repatriation General Hospital and the Flinders Clinical Research Ethics Committee of Flinders Medical Centre. Renal tissue distant to the primary tumor was collected immediately following surgery, and it was sectioned into cortex and medulla before either 1) fixation overnight in 10% neutral phosphate-buffered formalin (1:10 dilution of 40% formaldehyde in aqueous 0.03 M sodium dihydrogen orthophosphate and 0.045 M disodium hydrogen orthophosphate, pH 7.0), 2) immediate use for subcellular fractionation and isolation of microsomal protein, or 3) storage at –70°C for later use. Representative formalin-fixed and paraffin-embedded tissue samples were stained with hematoxylin and eosin and examined by a specialist histopathologist (A. Thomas). Typical for the age of the donors (43–83 years), only age-related benign nephrosclerosis and hypertensive-like vascular changes were observed. All sections were confirmed as solely cortex or medulla.

HKCM (n = 6; K6–K11) and HKMM (n = 5; K6, K7, K9–K11) were prepared using a standard differential ultracentrifugation technique as reported by Tsoutsikos et al. (2004). Microsomal protein was resuspended in Na₂HPO₄, pH 7.4, 20% (w/v) glycerol, analyzed for protein concentration (Lowry et al., 1951), and stored at –70°C before use.

Immunohistochemistry Studies

Tissue Processing and Epitope Retrieval. Formalin-fixed kidney tissue (n = 6; K2–K7) was embedded in paraffin following a standard overnight tissue processing schedule (Leica TP 1050; Leica Microsystems, Wetzlar, Germany). Serial, parallel sections (4 μm) were cut, mounted onto positively charged Superfrost Plus microscope slides (Menzel-Glaser, Braunschweig, Germany), stored at 37°C, and used within 30 days. Epitope retrieval was performed on deparaffinized and rehydrated cortical and medullary sections by boiling (4 min) and then simmering (10 min) in 0.01 M sodium citrate buffer, pH 6.0, followed by cooling to room temperature (15 min). The sections were rinsed in water and placed in Tris-buffered saline (0.1 M Tris-HCl, pH 7.6, in isotonic NaCl) before incubation with either of the primary UGT antibodies or with the negative control media.

Antibody Adsorption. Aliquots (20 μl) of either the UGT2B7 or UGT1A antibody were thawed, diluted in normal goat serum [2% (v/v) in Tris-buffered saline containing 0.1% sodium azide (v/v)], and stored at 4°C. The UGT1A and UGT2B7 primary antibodies (diluted 1:1500) were incubated with human UGT1A1 protein (1 and 10 μg) and UGT2B7 protein (1, 10, and 25 μg), respectively. Incubations (final volume 1.5 ml) were carried out sequentially at three temperatures: 37°C (2 h), room temperature (overnight), and 4°C (2 h), followed by centrifugation at 10,410g (10 min). The supernatant fraction was applied to human kidney cortical and medullary sections in place of the primary antibody (see following section).

Immunostaining and Quantification. Preliminary immunoblotting experiments confirmed binding of the UGT1A and UGT2B7 antibodies to HKCM and HKMM, and a lack of binding of NRIgG to UGT1A1 and UGT2B7. In addition to the supplier's specification of lack of cross-reactivity between the UGT2B7 antibody and UGT1A subfamily enzymes and UGT2B15, we further established no cross-reactivity with UGT2B4 and UGT2B17 (data not shown). The UGT1A antibody recognizes all UGT1A proteins, but not UGT2B4, 2B7, 2B15, and 2B17 (supplier's specification; Uchaipichat et al., 2004).

Sections were incubated overnight with either negative control media (NGS or NRIgG diluted 1:2250 in NGS) or UGT1A or UGT2B7 primary antibodies (diluted 1:1500 in NGS). Endogenous peroxidase activity was quenched using hydrogen peroxide [1% (v/v); 10 min]. Sections were incubated (1 h) using the EnVision+ peroxidase detection system (goat anti-rabbit) horseradish peroxidase-labeled antibody, followed by incubation with the substrate solution made up of the chromogen 3,3′-diaminobenzidine tetrahydrochloride and hydrogen peroxide [0.1% (v/v); 5 min]. The color reaction was terminated by rinsing with water. Sections were counterstained in Harris' hematoxylin, briefly differentiated in acidified alcohol [0.5% HCl in 70% (v/v) ethanol], and “blued” in a saturated solution of lithium carbonate. Sections were dehydrated in absolute ethanol, cleared in xylol, mounted using DePeX mounting medium (BDH Laboratory Supplies, Poole, UK), and dried overnight (60°C).

Labeled UGT enzymes were visualized by the presence of an insoluble reddish brown precipitate. The intensity of staining, which was readily observed by light microscopy, was scored by two investigators (A. Thomas and P. Gaganis) according to a semiquantitative method (Okonogi et al., 2001; Saarikoski et al., 2005): –, absence of positive staining; 1+, weak; 2+, moderate; and 3+, intense staining.

Image Capture. Sections were photographed using the Video Pro 32 image analysis system (Leading Edge P/L, Adelaide, Australia) made up of an Olympus BX40 microscope (Olympus, Tokyo, Japan) with a 2.5× camera eyepiece and a continuous interference filter monochromator for the enhancement of contrast between hematoxylin (blue nuclear counterstain) and 3,3′-diaminobenzidine tetrahydrochloride. Images were captured with a color digital charge-coupled device video camera (Panasonic model GP-KR222) with the automatic gain disabled. Correction of nonuniform illumination was standardized using a bright field image of the background illumination. Images were photographed using either a 10× or 20× objective.

Kinetics of S-Naproxen Acyl Glucuronidation

S-Naproxen acyl glucuronide formation by HKCM (n = 6) and HKMM (n = 5) was measured using a modification of the method of Bowalgaha et al. (2005). Preliminary studies established linearity of product formation with respect to incubation time and microsomal protein concentration, and less than 10% consumption of substrate occurred during the course of incubations. The incubation mixture (0.2 ml for HKCM or 0.1 ml for HKMM) contained MgCl₂ (4 mM), UDPGA (5 mM), S-naproxen (5–2000 μM), and HKCM (100 μg) or HKMM (50 μg) in phosphate buffer (0.1 M; pH 7.4). Incubations, which were performed in duplicate, were conducted for 30 min (HKCM) or 60 min (HKMM) at 37°C. Reactions were terminated by the addition of an equal volume of glacial acetic acid in methanol [4% (v/v)] and cooling on ice. Following centrifugation (5000g for 10 min at 10°C), an aliquot of the supernatant fraction (30 μl) was injected onto the HPLC column. The HPLC system was made up of an Agilent 1100 series instrument with a solvent delivery system, auto injector and variable wavelength UV-VIS detector (Agilent Technologies, Sydney, Australia) fitted with a Novapak C18 analytical column (4-μm particle size, 3.9 × 150 mm; Waters, Sydney, Australia). The analytes were detected by UV absorption at 230 nm. The mobile phase consisted of water/acetonitrile/glacial acetic acid [700:300:1.2 (v/v)], delivered at a flow rate of 1.5 ml/min. S-Naproxen acyl glucuronide and S-naproxen eluted at 2.4 and 11.4 min, respectively. The S-naproxen acyl glucuronide peak was confirmed by reference to an authentic standard and by hydrolysis with β-glucuronidase. However, standard curves were prepared using S-naproxen, at 0.25 to 5 μM, since S-naproxen acyl glucuronide is hygroscopic and unsuitable for use as the prime standard (Bowalgaha et al., 2005). Thus V_max (and CL_int) values should be considered “apparent”. S-Naproxen standard curves were linear over the range 0.25 to 5 μM(r² > 0.99). The lower limit of quantification of the assay was 0.02 μM, and the overall assay interday coefficient of variation was 1.9% (n = 7). Incubation samples were analyzed within 12 h; serial injection of incubation samples (containing high, medium, and low substrate concentrations) treated with glacial acetic acid in methanol [4% (v/v)] demonstrated that the S-naproxen acyl glucuronide product was stable (<5% decrease in peak height) to at least 12 h.

Fluconazole has previously been demonstrated to be a selective inhibitor of UGT2B7 (Uchaipichat et al., 2006). Using the incubation and chromatography conditions described above, inhibition of the high- and low-affinity components of HKCM- and HKMM-catalyzed S-naproxen acyl glucuronidation (see Results) by fluconazole (2.5 mM) was investigated to assess the contribution of UGT2B7 to these reactions. Given the limited availability of renal tissue, inhibition experiments were performed with microsomes from three kidneys only.

Data Analysis

Data points represent the mean of duplicate determinations, except where indicated. The kinetic parameters K_m and V_max were derived from fitting untransformed data to the single- and two-enzyme Michaelis-Menten (MM) equations using an extended least-squares modeling program (EnzFitter version 2.0.18.0; Biosoft, Cambridge, UK). The goodness of fit was determined by comparison of statistical parameters (95% confidence intervals for the curve fit and F-statistic). Kinetic data are reported as the mean ± S.D. In vitro intrinsic clearance was calculated as V_max/K_m.

Results

Antibody Adsorption. Nonspecific binding of the primary UGT antibodies to non-UGT epitopes was excluded by antibody preadsorption. Compared with the control (no UGT1A1 protein added, Fig. 1A), UGT1A antibody labeling was reduced by preadsorption with 1 μg of UGT1A1 (Fig. 1B), and it was prevented by 10 μg of UGT1A1 protein (Fig. 1C). Similarly, compared with the control (Fig. 1D, no UGT2B7 protein added), UGT2B7 immunoreactivity was decreased by preadsorption with 10 μg of UGT2B7 microsomal protein (Fig. 1E), and it was completely prevented by 25 μg of UGT2B7 (Fig. 1F).

Negative Controls. Two negative control sections using either normal goat serum or a commercial preparation of preimmune serum of NRIgG in place of the primary antibody were included routinely. When using normal goat serum, there was no evidence of positive staining in the cortical (Fig. 2A) or medullary sections (Fig. 2E). However, incubation with NRIgG resulted in a slight pink “blush” indicative of nonspecific protein binding homogeneously distributed in the interstitium, glomeruli, and tubules of the renal cortex (Fig. 2B) and the interstitium, loops, and collecting ducts of the renal medulla (Fig. 2F). Subsequent grading of UGT1A and UGT2B7 immunoreactivity was determined relative to the nonspecific background staining observed using NRIgG.

Immunolocalization of UGT1A Proteins and UGT2B7 in the Renal Cortex. There was no evidence of UGT1A and UGT2B7 immunoreactivity in the glomeruli and Bowman's capsule and the associated cortical vasculature, including glomerular capillaries, arteries, and afferent/efferent arterioles and veins (Fig. 2, C and D). In contrast, the epithelial cells of the PCT exhibited strong localized “pockets” of intense staining (3+; n = 6) with the UGT1A antibody (Fig. 2C), whereas UGT2B7 immunoreactivity was highly variable, ranging from weak to moderate staining (1+ to 2+; n = 6) (Fig. 2D; Table 1). Similar variability in staining was observed in the DCT for UGT1A and UGT2B7 (Table 1; Fig. 2, C and D). Although the pattern of immunoreactivity for UGT1A and UGT2B7 in the DCT was similar, the intensity of staining for UGT1A proteins and UGT2B7 was generally more pronounced on the apical or luminal side of the cytoplasmic membrane.

The macula densa, identified as columnar cells adjacent to the glomerulus, exhibited equivalent staining to that of the associated DCT, with weak immunoreactivity (1+) for UGT1A proteins (Fig. 3A) and moderate staining (2+) for UGT2B7 (Fig. 3B). UGT immunoreactivity was not detected in the interstitial cells of the juxtaglomerular apparatus. There was positive staining of two of the identifiable components comprising medullary rays; the thick-ascending loop of Henle (TALH) and the cortical collecting ducts. The third component, the proximal straight tubule, was difficult to distinguish from PCT, and none were identified with certainty. Variable immunoreactivity was observed across the TALH and CD with both the UGT1A and UGT2B7 antibodies (Table 1).

Immunolocalization of UGT1A Proteins and UGT2B7 in the Renal Medulla. There was absence of positive immunoreactivity for UGT1A and UGT2B7 in medullary veins and arteries. The CDs were grouped as either small or large CD. UGT1A and UGT2B7 immunoreactivity in the small CDs ranged from weak (1+) to moderate (2+) in five of the six kidneys studied (Fig. 2G) with the exception of K7, which exhibited intense (3+) staining with the UGT1A antibody. Intensity of staining was similar in the large CDs for UGT1A proteins and UGT2B7, ranging from weak (1+) to moderate (2+) (Fig. 2, G and H), respectively. Variable immunoreactivity (1+ to 2+) was observed in epithelial cells of the thin LOH and the TALH for both UGT1A and UGT2B7 (Fig. 2, G and H). An exception was K4 where the TALH stained intensely (3+) for UGT1A proteins. Interestingly, intense staining was observed for UGT1A proteins on the luminal side of the TALH in a proportion of the UGT-positive cells.

Within epithelial cells of the PCT, DCT and TALH continuous granular staining was observed in the nuclear envelope, and focal granular staining within nuclei for UGT1A proteins. A similar pattern of immunoreactivity was observed for UGT2B7. Positive staining for UGT1A proteins and UGT2B7 was not observed in the nuclear envelope and nuclei of epithelial cells of the CDs and thin LOH.

Kinetics ofS-Naproxen Acyl Glucuronidation by HKCM and HKMM. The glucuronidation kinetics of S-naproxen were investigated using HKCM and HKMM from six kidneys over the substrate concentration range 5 to 2000 μM. S-Naproxen acyl glucuronide formation by HKCM and HKMM exhibited “biphasic” kinetics (Fig. 4, A and B). Data were well fitted to the two-enzyme Michaelis-Menten equation. Derived kinetic constants (mean ± S.D.) for the high- and low-affinity components using HKCM as the enzyme source were K_m1 = 34 ± 14 μM, V_max1 = 134 ± 98pmol/min · mg protein, K_m2 = 627 ± 541 μM, and V_max2 = 133 ± 71pmol/min · mg protein. CL_int values, calculated as V_max/K_m, for the high- and low-affinity components of S-naproxen acyl glucuronidation by HKCM were 5.1 ± 4.2 and 0.73 ± 0.73 μl/min · mg, respectively. Derived kinetic constants for the high- and low-affinity components of S-naproxen acyl glucuronidation by HKMM were K_m1 = 45 ± 14 μM, V_max1 = 48 ± 19 pmol/min · mg protein, K_m2 = 654 ± 488 μM, and V_max2 = 45 ± 7 pmol/min · mg protein. CL_int for the respective high- and low-affinity components were 1.2 ± 0.6 and 0.09 ± 0.03 μl/min · mg. At S-naproxen concentrations of 10 and 1000 μM, fluconazole inhibited cortical S-naproxen acyl glucuronidation by 73 and 8%, respectively, and medullary S-naproxen acyl glucuronidation by 68 and 12%, respectively.

Discussion

This study is the first to report the cellular localization and semiquantitative expression of UGT1A proteins and UGT2B7 in cortical and medullary segments of the human nephron. Moreover, we provide the first evidence of UGT1A and UGT2B7 expression in the macula densa, loops of Henle, and collecting ducts. Exclusion of antibody cross-reactivity, inclusion of negative controls using NRIgG and normal goat serum, and attenuation of labeling by antibody preadsorption supports the conclusion that the renal tissue antigens detected were UGTs. Although a comprehensive understanding of the role of UGTs in the kidney is lacking, their presence throughout the nephron suggests a role for UGT enzymes in renal metabolic elimination processes.

Comparatively little variability with respect to UGT expression was observed in tissue from the six kidneys. Similar to other studies (Peters et al., 1987; Girard et al., 2003) UGT1A and UGT2B7 immunoreactivity was not demonstrated in Bowman's capsule or the glomerulus. Moderate-to-intense staining was evident for UGT2B7 and UGT1A proteins in epithelial cells of the PCT. Previous studies in animals have reported that UGT activity is primarily localized to the cortical region (Rush and Hook, 1984) and that rates of glucuronidation are highest in cells of the PCT (Lohr et al., 1998; Schaaf et al., 2001). It is not surprising that UGTs are expressed in the PCT in humans, which is located adjacent to the glomerulus, and it is the initial site of exposure to the glomerular ultrafiltrate. The capacity of the PCT for active transport of organic compounds further increases the propensity for exposure to high concentrations of potentially nephrotoxic xenobiotics (Schaaf et al., 2001). In addition, localization in epithelial cells of the PCT is consistent with a role of UGTs in limiting the biological activity of mineralocorticoids and glucocorticoids (Girard et al., 2003).

In contrast to the study of Peters et al. (1987), who reported absence of UGT in the DCT, our study has shown that UGT1A proteins and UGT2B7 are present in the epithelial cells of the DCT, and, importantly, in the macula densa. Expression in the macula densa has not been reported previously, and it is consistent with the hypothesis that glucuronidation terminates the biological reactivity of endogenous molecules involved in regulating renal hemodynamics (Knights et al., 2005). The macula densa has a role in regulating intravascular renal tone via the renin-angiotensin-aldosterone pathway. Evidence indicates that UGT2B7 is the predominant enzyme responsible for aldosterone glucuronidation (Girard et al., 2003); hence, UGT2B7 in the macula densa may play a role in regulating aldosterone. It is noteworthy that the majority of aldosterone 18β-glucuronide excreted in urine is thought to be formed in extrahepatic tissues, particularly the kidney (Bledsoe et al., 1966).

Two components of the medullary ray were identified; the thick-ascending loop of Henle and the cortical collecting ducts. In both of these segments of the nephron, UGT1A and UGT2B7 expression was variable, with absence of immunoreactivity for UGT2B7 in one kidney and weak-to-moderate staining with both antibodies in the other five kidneys. UGT protein expression in collecting ducts has been previously reported as absent in human renal tissue (Peters et al., 1987) and “weak” in cortical collecting ducts of monkey kidney (Girard et al., 2003). In the present study, UGT1A proteins and UGT2B7 were identified in the medullary collecting ducts and loops of Henle. The former is not surprising given the role of aldosterone in the DCT and collecting ducts and evidence of aldosterone glucuronidation by UGT2B7 (Girard et al., 2003).

Evidence of expression of UGT1A proteins and UGT2B7 in the loops of Henle is noteworthy. It is conceivable that intrarenal glucuronidation terminates the biological activity of eicosanoids, including prostaglandins, synthesized locally in the loops of Henle. This view is supported by evidence that UGT2B7 and UGT1A enzymes glucuronidate a variety of eicosanoids, including prostaglandins B₁ and E₂ (Little et al., 2004). Prostaglandin E₂ is the major prostaglandin produced along all segments of the nephron (Bonvalet et al., 1987).

Additional observations from our study include expression of UGT1A proteins and UGT2B7 in cell nuclei and the nuclear envelope of epithelial cells within PCT, DCT, and TALH. These data support previous reports of UGT expression in the nuclei of human prostate epithelial cells (Barbier et al., 2000), the nuclear envelope of rat liver hepatocytes and epithelial cells of the jejunum, renal PCT, and adrenal glands (Elmamlouk et al., 1981; Chowdhury et al., 1985) and nuclear membranes of human hepatocytes (Radominska-Pandya et al., 2002). Nuclear expression of UGT1A and UGT2B7 in epithelial cells of the renal tubules may reflect an evolutionary mechanism to control nuclear receptor ligand interactions (Radominska-Pandya et al., 2002).

A substantial body of research has provided evidence for renal xenobiotic metabolism. Using HKCM and HKMM, formation of S-naproxen acyl glucuronide exhibited biphasic kinetics characteristic of the involvement of high- and low-affinity enzymes in this pathway. The mean K_m value for the high-affinity component determined from fitting to the two-enzyme MM equation was 18- and 15-fold lower than the K_m values for the low-affinity reaction in cortex and medulla, respectively. Similarly, the microsomal intrinsic clearances of the high- and low-affinity components of the cortex and medulla differed 7- and 12-fold, respectively. Assuming an unbound fraction in plasma of 0.018 (Vree et al., 1993) and a maximum unbound plasma S-naproxen concentration of 5 to 10 μM (Vree et al., 1993; Runkel et al., 1976), substitution of the mean kinetic data in the two-enzyme MM equation indicates that the low-affinity cortical and medullary UGT(s) would be expected to contribute <25 and <5%, respectively, to naproxen renal glucuronidation in humans taking therapeutic doses (250–1000 mg daily).

The range of K_m values for the high-affinity component of S-naproxen acyl glucuronidation by both cortical (20–63 μM) and medullary (26–62 μM) microsomes is similar to that reported for the high-affinity component of human liver microsomal S-naproxen glucuronidation (14–49 μM) and recombinant UGT2B7 (72 μM) (Bowalgaha et al., 2005). In this study, involvement of UGT2B7 was confirmed by use of the selective UGT2B7 inhibitor fluconazole (Uchaipichat et al., 2006). Fluconazole inhibited the high-affinity component of S-naproxen glucuronidation in cortex and medulla by 73 and 68%, respectively, similar to the 70% inhibition of S-naproxen glucuronidation observed with human liver microsomes (Bowalgaha et al., 2005). Collectively, the results indicate that in both the cortex and medulla, UGT2B7 is the high-affinity enzyme involved in renal S-naproxen glucuronidation in humans. Because the low-affinity component was relatively unaffected by fluconazole (<15% inhibition), it is likely that the low-affinity component comprises several enzymes with naproxen glucuronidation capacity. UGT1A6 and UGT1A9 have been identified in human kidney (Sutherland et al., 1993), and both have K_m or S₅₀ values (855–1036 μM) of similar order to the low-affinity component (∼650 μM) of human renal microsomal S-naproxen glucuronidation. Consistent with other reports describing the renal medulla as having lower UGT activity than cortex (Rush and Hook, 1984; Hjelle et al., 1986; Yue et al., 1988), in this study S-naproxen glucuronidation activity differed 2.7-fold between cortex and medulla.

The kidney receives 20 to 25% of resting cardiac output (Saker, 2000), and perfusion of different regions of the kidney is estimated at 90% to the cortex, 6 to 10% to the medulla, and 1 to 2% to the papillae (Goldstein and Schnellmann, 1996). In “normal” individuals, renal perfusion is largely prostaglandin-independent. However, in circumstances of “prostaglandin-dependent” renal perfusion (e.g., congestive heart failure), NSAIDs can cause acute deterioration of renal function and in some instances renal ischemia. It may be speculated that renal glucuronidation normally plays an important “local” role in modulating intrarenal suppression of cyclooxygenase activity by NSAIDS, including naproxen, thereby buffering perturbation of renal hemodynamics. However, in circumstances where cortical and medullary perfusion is compromised, intrarenal concentrations of NSAIDs and/or NSAID acyl glucuronides may become elevated, increasing the potential for acylation of renal proteins by NSAID acyl glucuronides. Accumulation of reactive NSAID acyl glucuronides coupled with ischemia of the medullary papillae is a possible precipitating factor in the development of NSAID-induced renal papillary necrosis (Whelton et al., 2003). Hence, depending on the pathophysiological circumstances, intrarenal glucuronidation of NSAIDs may be either beneficial or deleterious.

In summary, we report that with the exception of the glomerulus, Bowman's capsule, and renal vasculature, UGT1A proteins and UGT2B7 are expressed throughout cortical and medullary segments of the human nephron. This pattern of expression is particularly interesting in view of the widespread synthesis of prostaglandins along the nephron and evidence of their metabolism by UGTs. Furthermore, it is noteworthy that UGT1A and UGT2B7 were localized in the macula densa, consistent with an important role for UGT in the regulation of the mineralocorticoid aldosterone. Consistent with the immunohistochemistry data, S-naproxen was glucuronidated by both HKCM and HKMM. Of the UGTs present in human kidney, UGT2B7 seems to be the principal enzyme responsible for S-naproxen acyl glucuronide formation in both renal cortex and medulla. Finally, we suggest that the ubiquitous distribution of UGTs in mammalian kidneys may buffer physiological responses to multiple endogenous mediators, but at the same time, specific competitive intrarenal xenobiotic-endobiotic interactions may provide an explanation for the adverse renal effects of some drugs, including NSAIDs.