cycloxygenase-2 is expressed in vasculature of normal and...

Cycloxygenase-2 Is Expressed in Vasculature of Normal andIschemic Adult Human Kidney and Is Colocalized withVascular prostaglandin E2 EP4 Receptors

KARINA L. THERLAND,* JANE STUBBE,* HELLE C. THIESSON,*PETER D. OTTOSEN,‡ STEEN WALTER,§ GRITH L. SØRENSEN,† OLE SKØTT,*and BOYE L. JENSEN**Department of Physiology and Pharmacology, †Department of Immunology and Microbiology, University ofSouthern Denmark, and ‡Institute of Pathology, §Department of Urology, Odense University Hospital,Odense, Denmark.

Abstract. The study was performed to elucidate the distributionand cellular localization of cyclooxygenase (COX)-2 in humankidney and to address localization of downstream targets forCOX-derived prostanoids. Cortex and outer and inner medullatissue were obtained from control kidneys (cancer specimens),kidneys with arterial stenosis, and kidneys of patients whoreceived angiotensin II inhibition or acetylsalicylic acid. Ribo-nuclease protection assay and Western blot test revealed thatCOX-1 and -2 mRNA and protein were expressed in all regionsof human kidney (mRNA ratio, cortex:outer medulla:innermedulla COX-1 1:3:20 and COX-2 1:1:3). In adult kidney,immunohistochemical labeling for COX-2 was associated withsmooth muscle cells in pre- and postglomerular vessels andwith endothelium, particularly in vasa recta and medullarycapillaries. Western blot test confirmed COX-2 expression in

renal artery. COX-2 had a similar localization in fetal kidneyand was additionally observed in Henle’s loop and maculadensa. Human tissue arrays displayed COX-2 labeling of vas-cular smooth muscle in multiple extrarenal tissues. VascularCOX-2 expression was significantly increased in kidneys witharterial stenosis. COX-1 was colocalized with microsomalprostaglandin E2 synthase (PGES) in collecting ducts, andPGES was also detected in macula densa cells. VascularCOX-2 was colocalized with prostaglandin E2 EP4 receptorsbut not with EP2 receptors. Thus, renovascular COX-2 expres-sion was a constitutive feature encountered in human kidneysat all ages, whereas COX-2 was seen in macula densa only infetal kidney. Vascular COX-2 activity in human kidney andextrarenal tissues may support blood flow and affect vascularwall-blood interaction.

A rate-limiting step in prostaglandin formation is catalyzed bycyclooxygenase (COX). Two isoforms are recognized: consti-tutive COX-1 and inducible COX-2. Selective COX-2 antag-onists have recently been developed and are in widespreadclinical use. In contrast to expectations, COX-2 selectiveblockers exhibit adverse effects related to kidney and cardio-vascular function; they have antinatriuretic properties; theylower GFR and RBF and can lead to acute renal failure; andthey aggravate preexisting hypertension and may lead to ad-verse cardiovascular outcomes (1–6). Thus, the data indicateimportant physiologic roles for COX-2 in human cardiovascu-lar and renal homeostasis, but the renal correlates of theseclinical observations remain poorly understood. Both COXgenes are expressed in human kidney (7,8), but the quantitative

relation of COX-2 to COX-1 expression is not known. Whetherthere are differences in expression level over time or in kidneyregions is also not known. Data on the cellular localization ofCOX-2 in human kidney are inconsistent.

COX-2 has been observed in thick ascending limb of Hen-le’s loop (cTAL), including the macula densa in fetal humankidney (8–10) and in the kidneys of children with Bartter’ssyndrome (11), a location first described in rat kidney (12).Several studies did not detect COX-2 in the cTAL or maculadensa of adult human kidneys (7,8,10), whereas other reportsshowed an age- or disease-dependent expression of COX-2 inmacula densa (11,13,14) and a functional role of COX-2 forstimulation of renin secretion in humans (15,16). COX-2 hasalso been observed in renal vascular tissue (7–9), and obser-vations indicate that intact prostaglandin synthesis is importantfor maintenance of renal function during renovascular hyper-tension with renal arterial stenosis (17–19). It is not knownwhether this dependence involves COX-2 activity.

Three major objectives of the study presented here were todetermine the quantitative regional distribution and cellularlocalization of COX isozymes in normal adult human kidneyand compare to fetal kidney; to assess COX-2 in human kid-neys with arterial stenosis and after pharmacologic interven-tions; and to elucidate colocalization of COX enzymes with

Received December 10, 2003. Accepted February 5, 2004.Correspondence to Dr. Boye L. Jensen, Department of Physiology and Phar-macology, University of Southern Denmark, Winsløwparken 21, 3, DK-5000,Odense C, Denmark. Phone: �45-65-50-37-96; Fax �45-66-13-34-79; [email protected]

1046-6673/1504-1189Journal of the American Society of NephrologyCopyright © 2004 by the American Society of Nephrology

DOI: 10.1097/01.ASN.0000124673.79934.24

J Am Soc Nephrol 15: 1189–1198, 2004

downstream target molecules. We used quantitative mRNAand protein assays in combination with immunohistochemicalanalysis to address these issues in human nephrectomysamples.

Materials and MethodsHuman Tissue Samples

All patients gave written informed consent to participate in thestudy. The study was approved by the local ethics committee. None ofthe cancer patients had received chemotherapy or radiation therapybefore nephrectomy. Samples of normal kidney tissue were obtainedfrom unaffected areas in nephrectomy specimens with tumors. Fourkidneys with artery stenosis were obtained during the 3-yr samplingperiod. Kidneys were extirpated at the Department of Urology,Odense University Hospital, and immediately transported to the In-stitute of Pathology. Each kidney, with or without tumor, was dividedinto cortex, outer medulla, and papilla and frozen in liquid nitrogen.Tissue blocks were fixed in 4% paraformaldehyde overnight andembedded in paraffin. Other sections were embedded and frozen inCO2-cooled isopentane and kept at �80°C until use. In three kidneys,it was possible to dissect a piece of the renal artery, which wasimmediately frozen. Paraffin-embedded human tissue arrays and fetalkidneys were obtained from the archive of the Department of Pathol-ogy, Odense University Hospital.

Human Umbilical Artery Endothelial Cell CulturePrimary cultures of human umbilical artery endothelial cells were

established by collagenase treatment of umbilical cords. The cordswere obtained from the Department of Obstetrics and Gynecology,Odense University Hospital, Denmark, after informed consent wasobtained. Human umbilical artery endothelial cells were seeded ongelatin-coated tissue-culture plastic, and the cultures were maintainedin EGM-2 medium (Clonetics, San Diego, CA) and kept at 37°C, 95%O2, 5% CO2. Cells were detached with trypsin-EDTA solution(Sigma, Rødovre, Denmark) and used until passage 7. Lipopolysac-charide (10 �g/ml, Sigma) was added to culture flasks for 5 h, and cellprotein was isolated.

Molecular ProtocolsTotal RNA was isolated from tissue by columns (RNeasy Midi Kit,

Qiagen, Albertslund, Denmark). RNA was quantified by measuringoptical density at 260 nM. RT-PCR was used to amplify sequencesspecific for human COX-1, COX-2, renin, and �-actin (20). Allprimers were synthesized with 5' restriction sites for BamHI (sense)and EcoRI (antisense) (Invitrogen, Scotland) to allow for directionalcloning (vector pSP73, Promega-Ramcon, Birkerød, Denmark), se-quencing, and in vitro transcription (20). Primers were as follows: forCOX-1, sense: 5'-ATG TCA TCA GGG AGT CTC-3', antisense:5'-AAG CAG TCC AGG GTA GAA-3' (accession numberxm011834, bases 1339 to 1514, 176 bp); and COX-2, sense: 5'-GTGAAA CCA TGG TAG AAG-3', antisense: 5'-AGT AGT ACT GTGGGA TTG-3' (accession number xm00173416, bases 1648 to 1924,277 bp, with the 3' half of the amplified sequence not found inCOX-1); renin, sense: 5'-ATG AAG AGG CTG ACA CTT-3', anti-sense: 5'-GAG AAA GCC ACT GAC TGT-3' (285 bp spanning twointrons) (21); �-actin, sense: 5'-CCA AAG TTC ACA ATG TGG-3',antisense: 5'-CAC GAA AGC AAT GCT ATC-3' (accession numberxm004814, bases 1392 to 1579, 188 bp).

Solution Hybridization and Ribonuclease ProtectionAssays

mRNA levels were estimated by solution hybridization followed byA/T1 ribonuclease protection assay as described (20). Protectedprobes were excised from dry gels and radioactivity was quantified ina � counter. COX-1 and -2 probes had the same specific radioactivity,which permitted direct comparison of isoform mRNA levels.

Immunohistochemical and ImmunofluorescenceAnalysis of Kidney Sections

Processing of tissue for immunohistochemical analysis was essen-tially as previously described (22). Two COX-2 antibodies were used;rabbit anti-human directed against amino-terminal amino acids 50 to111 (Santa Cruz, AH Diagnostics, Aarhus, Denmark), and rabbitanti-human against the carboxy-terminal amino acids 567 to 599,unique to COX-2 (Cayman Chemicals Co., AH Diagnostics, Aarhus,Denmark). Both antibodies have been used previously on humantissue (7,9,11,23). COX-1 antibody was from Santa Cruz (goat-antihuman). Antibodies directed against microsomal prostaglandin E syn-thase (PGES; rabbit anti-human), EP4 receptor (rabbit anti-human)and EP2 receptor (rabbit anti-rat) were from Cayman Chemicals.Antigen retrieval was carried out by microwave cooking in citratebuffer (Dako, Glostrup, Denmark) for 20 min or in a pressure cookerfor 5 min. The sections were blocked with 5% dry milk in 0.05%Tween–Tris-buffered saline (TTBS), and then incubated with dilutedprimary antibodies overnight at 4°C. Next, the sections were incu-bated with horseradish peroxidase (HRP)-conjugated secondary anti-body directed against the relevant species (Dako). The Santa CruzCOX-2 antibody was visualized according to the Power Vision polyHRP amplification kit (Immuno-vision Technology Co., Brisbane,CA). Signals were visualized by incubation for 2 to 20 min with0.01% diaminobenzidine (Dako) and 0.02% H2O2. For preabsorptioncontrols, the antibody and 10 �g/ml peptide was incubated in 5%milk-TTBS for 2 h. Immunolabeling for EP4 receptor was carried outon unfixed cryosections that were incubated in 0.5% Triton-X–100PBS, and next were blocked in 5% dry milk–PBS. The sections wereincubated overnight at 4°C with EP4 primary antibody (1:50), andsignals were visualized as described above. Double-immunofluores-cence labeling for COX-1 and COX-2 was performed after blockingwith 5% milk-TTBS. Diluted primary antibodies were incubatedovernight at 4°C, and the sections were then incubated with Alexa-fluor 594 conjugated donkey–anti-goat antibody (1:250, MolecularProbes) followed by incubation with Alexa 488 conjugated goat–anti-rabbit antibody (1:200, Molecular Probes). Sections were inspectedwith an epifluorescence microscope (Olympus BX51).

Western Blot TestProtein was isolated and quantitated as described (22). Twenty-

microgram protein samples were used for COX-2 assays and 60-�gsamples for COX-1 assays. Proteins were separated by 4–12% SDS-PAGE and electroblotted (Bio-Rad Laboratories, Copenhagen, Den-mark) onto PVDF Immobilon membranes (Millipore, Glostrup, Den-mark). Membranes were blocked and incubated with anti–COX-2antibodies (1:1000) or COX-1 antibody (1:1000) in 5% dry milk–TTBS overnight. Membranes were incubated with HRP-coupled sec-ondary antibody (1:2000) for 1 h. Bound secondary antibody wasdetected by enhanced chemiluminescence kit (ECL plus WesternBlotting Detection System, Amersham Biosciences, Horsholm, Den-mark) and exposed to x-ray film. Proteins for PGES detection wereseparated on a 10% SDS-PAGE gel and reacted with primary anti-body (1:1000).

1190 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 1189–1198, 2004

Statistical AnalysesAll values are given as mean � SEM. Unpaired t test was used to

determine statistical difference when two groups of data were com-pared. P � 0.05 was considered statistically significant.

ResultsDistribution of COX mRNA and Proteins in AdultHuman Kidney Regions

By ribonuclease protection, both COX transcripts were de-tected in all regions of 11 adult human kidney samples thatwere selected from patients who did not receive any chronicmedication before nephrectomy. COX-1 and -2 were expressedat equal levels in kidney cortex, but COX-1 was more abundantthan COX-2 in outer and inner medulla (Figure 1, A and B).There was a significant cortical-medullary gradient for bothisoforms. Compared with cortex, COX-1 was expressed at 3and 20 times higher levels in outer and inner medulla, whereasCOX-2 was modestly but significantly more highly expressedin inner medulla compared with cortex and outer medulla(Figure 1B). Renin mRNA was detected in cortex only, whichconfirmed correct separation of tissue (Figure 1A). The COX-2antibodies, recognizing amino- and carboxy-terminal parts ofhuman COX-2, reacted with a protein with the expected size of72 kDa on Western blot tests of kidney cortex and innermedulla. Figure 1C shows the results with the COOH-terminalCOX-2 antibody. There was no significant differencein COX-2 protein level between cortex and inner medulla. TheCOX-1 antibody detected a single protein with the anticipatedmolecular size of 70 kDa. COX-1 protein level was signifi-cantly higher in the inner medulla compared with cortex,similar to the data on mRNA distribution.

Immunohistochemical Localization of COX Isoforms inAdult and Fetal Human Kidney Sections

In adult kidney, COX-2 immunoreactivity was associatedwith preglomerular, glomerular, and postglomerular vessels(Figure 2, a, b, d, f, and j, k). Preglomerular vessels of allcalibers from arteries to arterioles were labeled in the smoothmuscle cell layer (Figure 2, b, d, and k; Figures 4A and 5B). Inglomeruli, labeling was observed in parietal epithelial cells(Figure 2b) and in few intraglomerular cells, probably podo-cytes (Figure 4A) (8). In the postglomerular vasculature, la-beling was associated with outer medullary vasa recta, as seenin the compound picture of a longitudinal section and in crosssections of outer medulla (Figure 2, a, f, and j).

The inner medullary capillary network was labeled forCOX-2, which implies that COX-2 is expressed in endothelium(Figure 2, f and j). The immunopositive vessels were identifiedon the basis of the presence of erythrocytes in their lumen. Inadult kidney, no COX-2 immunoreactivity was associated withcollecting ducts, proximal tubules, loops of Henle, distal tu-bules, or medullary interstitium (Figure 2, a, f, and j). Notably,we did not observe immunolabeling in macula densa in any ofthe samples analyzed (Figure 2b). A second, COX-2–specificantibody directed against the NH2-terminal of human COX-2was applied. The same pattern of labeling was obtained withthis antibody (Figure 2, j and k). In two human fetal kidneys

(gestational weeks 29 and 30), COX-2 immunoreactivity wasassociated with the parietal epithelial cells of Bowman’s cap-sule and with endothelium and vascular smooth muscle (Figure

Figure 1. Distribution of cyclooxygenase (COX)-1 and COX-2 mRNAand protein in human kidney regions. (A) Autoradiographs depicting thedistribution of COX-1, COX-2, and renin mRNAs in dissected humankidney regions from three patients. Messenger RNA determinations wereperformed in duplicate RNA samples with two amounts of total RNAfrom each human kidney region (20 and 40 �g from cortex and outermedulla and 10 and 20 �g from inner medulla). (B) Messenger RNA-cRNA hybrids were cut out of the dry gels and protected radioactivity inthe probes was quantitated in a � counter. Each column represents theaverage counts per microgram of RNA normalized with �-actin mRNAlevel � SEM from 11 separate kidneys. (C) Upper panels display theresult of Western blot tests for COX enzymes with protein isolated fromcortex and inner medulla of human kidneys. Twenty micrograms ofprotein from the cortex and inner medulla was used for the COX-2 assay,and 60 �g was used to detect COX-1. Band densities were evaluated; thelower panels display average densitometry units per square millimeter �SEM (n � 3). OM, outer medulla; IM, inner medulla; * P � 0.05.

J Am Soc Nephrol 15: 1189–1198, 2004 COX-2 in Normal and Ischemic Human Kidney 1191

Figure 2. Immunohistochemicallabeling of human kidney sec-tions for cyclooxygenase(COX)-2 and COX-1. (a) Com-pound picture showing COX-2immunoreactive protein in ves-sels of an outer medullary vasarecta bundle. Scale bar � 200�m. In renal cortex, COX-2 wasobserved in vascular smoothmuscle (b, d, k) and in the pari-etal glomerular epithelium (b),but not in the loop of Henle orthe macula densa (b). (c) COX-1was located in a few mesangialcells in glomerulus. Scale bar �50 �m. Cross section of innermedulla revealed (g) COX-1 im-munoreactivity in collectingducts and (f) COX-2 in vasa rectaand capillaries. Incubation of theprimary antibodies with peptidesused to raise the antibodies pre-vented labeling of tissue sections(h, i). Application of a separateCOX-2 antibody directed againstthe amino terminal of COX-2produced a similar staining resultas with the carboxy-terminal an-tibody used in a, b, d, and f (j, k).Human fetal kidney displayedCOX-2 in loop of Henle andmacula densa (l) and in vascula-ture (m).


2, l and m; Figure 4B). COX-2 was not found in the nephro-genic zone, but was associated with loops of Henle, includingmacula densa cells of mature juxtamedullary glomeruli (Fig-ures 2l and 4B), as previously reported (8–10).

In adult kidney, COX-1 immunoreactivity was associatedwith endothelial cells of arteries, with mesangial cells at theglomerular hilum and with cortical interstitial cells (Figure 2, c,e, and g). COX-1 immunoreactivity was detected in collectingduct principal cells in cortex and medulla (Figure 2, c and e).Omission of primary COX antibodies or preabsorption of an-tibodies with surplus of the respective peptides prevented la-beling (Figure 2, h and i).

Vascular Expression of COX EnzymesVascular expression of COX isoforms was examined by

Western blot tests of proteins isolated from human renal artery(n � 2). Both COX-2 antibodies, directed against amino- andcarboxy-terminal epitopes, reacted with a protein from renalartery that was more abundant compared with renal innermedulla and had the expected size of COX-2 (Figure 3, A andB). Significant COX-1 expression was also detected in renalartery but was less abundant compared with COX-1 in innermedulla (Figure 3C). Cultured human endothelial cells exhib-ited COX-2 signals when analyzed on Western blot test (Figure3D)

Cellular Localization of COX-2 in Human Renal andExtrarenal Vasculature

Human kidney sections were double-labeled by immunoflu-orescence for COX-1 and -2. Figure 4A, a through c, shows anafferent arteriole where COX-2 immunofluorescence was as-sociated with smooth muscle and endothelial cells and withparietal epithelial cells of Bowman’s capsule (a, green). TheCOX-1 antibody labeled endothelial cells in the afferent arte-riole (b) and mesangial cells in the glomerulus (b). Endothelialcells were the only cells that occasionally exhibited bothCOX-1 and -2 signals (c, arrow). In renal outer medulla,COX-2 fluorescence was restricted to vasa recta (d, e, green),whereas COX-1 was uniformly associated with collecting ductepithelium (d, e, red). In the inner medulla, COX isoformswere associated with similar structures as in outer medulla;COX-2 was observed in medullary capillaries (f, green). Infetal human kidney tissue at gestational week 30 (Figure 4B),COX-2 signals in cortex were associated with glomeruli (g),loops of Henle (g), and arteries (h). In the immature medulla,vasa recta were COX-2 positive (i).

Next, it was addressed whether vascular expression ofCOX-2 was specific for the kidney. An array of adult humanorgans was labeled with the NH2-terminal COX-2 antibody(Figure 4C). Significant fluorescence signals were associatedwith submucosal arteries and arterioles in ventricle (m), gall-bladder (n), jejunum (o), and urinary bladder (k). Arterial andarteriolar smooth muscle was labeled in skeletal muscle (j) andspleen (l) and in dermis, lung, and placenta (not shown).

Localization of Downstream Targets for COX Productsin Human Kidney

Microsomal PGES is responsible for conversion of the COXproduct prostaglandin H2 to prostaglandin E2 (PGE2). PGESimmunoreactivity in cortex was associated with the maculadensa and not observed in adjacent epithelial cells (Figure 5A,a through c). PGES labeling was detected in a subset of corticalcollecting duct cells and in all inner medullary collecting ductcells (Figure 5A, d and e). On Western blot test, the PGESantibody detected a single protein with the expected size of 16kDa (Figure 5A, f).

Next, Gs-coupled vasodilator receptors for PGE2, EP2 andEP4, were localized in human kidney. EP4 labeling was de-tected in vascular smooth muscle cells of large (Figure 5B, h)and smaller afferent (Figure 5B, j) glomerular vessels (24). In

Figure 3. Western blot tests for cyclooxygenase (COX)-2 using pro-tein isolated from human kidney inner medulla (IM) and human renalartery. A total of 20 �g was used for the COX-2 assay, and 60 �g wasused for the COX-1 assay. (A) COX-2 antibody directed against anepitope in the carboxy-terminal domain of COX-2. (B) COX-2 anti-body directed against an epitope in the amino-terminal of humanCOX-2. (C) COX-1 antibody. (D) Western blot tests for COX-2 with20 �g protein isolated from cultured human umbilical artery endo-thelial cells that were incubated with or without lipopolysaccharide(LPS, 10 �g/ml) for 5 h. Twenty micrograms of protein from IM wasapplied as positive control.


Figure 4. Immunofluorescence analysis of cyclooxygenase (COX) in human kidney and nonrenal tissues. (A) Human kidney cortex wasdouble-labeled for COX-2 (green, a) and COX-1 (red, b), and colocalization was detected in few endothelial cells in the afferent arteriole(yellow, c). Scale bar � 50 �m. Cross section (d) and longitudinal section (e) of human kidney outer medulla and inner medulla (f) labeledfor COX-1 (red) and COX-2 (green), showing separate localization. Scale bar in d and e � 200 �m; f, 50 �m. (B) Labeling of human fetalkidney tissue at gestational week 30 for COX-2. COX-2 signals were associated with glomeruli (g), loops of Henle (g) arteries and arterioles(h), and vasa recta (i). The glomeruli are small compared with (A). Scale bar � 50 �m. (C) Labeling of a human tissue array for COX-2 (greenfluorochrome) showed a vascular localization: skeletal muscle (j), bladder (k), spleen (l), ventricle (m), gallbladder (n), and small intestine (o).Scale bar � 50 �m.


adjacent serial sections, COX-2 was associated with the samesmooth muscle cells as EP4 (Figure 5B, g and i). EP4 immu-noreactivity was detected in vasa recta bundles (not shown).Immunopositivity for the EP2 receptor was associated withthin descending and ascending limbs of Henle’s loop (Figure5C, l) (25). EP2 immunolabeling was detected in the post–macula densa distal convoluted tubule and in connecting tubule(Figure 5C, k). EP2 receptor labeling was associated predom-inantly with the basolateral aspect of the epithelial cells. EP2immunoreactivity was not observed in any part of the renalvasculature or in the macula densa (Figure 5C, k, arrows).

Effect of Renal Artery Stenosis, Angiotensin II (AngII)Inhibitors, and COX Inhibitors on COX Expression inHuman Kidney

COX expression was analyzed in kidneys with renal arterystenosis; in kidneys from patients who had preoperatively re-ceived angiotensin-converting enzyme (ACE) inhibitors, or AT1receptor antagonists; and in patients who had received acetylsal-icylic acid. COX-2 mRNA abundance was significantly higher incortex and medulla from kidneys with arterial stenosis comparedwith control kidneys (Figure 6, A through C). Renin expressionwas strongly elevated in three of the four kidney cortices witharterial stenosis compared with six controls (Figure 6A). COX-1and COX-2 expression was not changed in kidneys from patientswho had received AT1 receptor blockers, ACE inhibitors, oracetylsalicylic acid (not shown). Kidneys with artery stenosisexhibited tubular atrophy and interstitial fibrosis. The vasculaturedisplayed medial and intimal thickening and some arteriolar hy-alinosis. COX-2 immunoreactivity was associated with vascularsmooth muscle in the thickened media layer. In general, there wasa poor morphologic resolution in the cortical labyrinth because ofthe end-stage nature of the kidneys. Therefore, it was not possibleto identify the macula densa or any other specific tubular segmentin these kidneys.

DiscussionIn the present investigation, the major finding is that cyclo-

oxygenase (COX-1 and COX-2) was expressed constitutivelyin all regions of adult human kidney with distinct cellularlocalization. COX-1 was associated with collecting ducts.COX-2 was associated with media smooth muscle and vascularpericytes of all segments of the pre- and postglomerular vas-culature and was significantly elevated in kidneys from pa-tients with renal artery stenosis. In fetal kidneys, COX-2 wasexpressed in vasculature and glomeruli, but also in loop ofHenle and macula densa cells. Thus, COX-2 was detectedconsistently in tubular epithelium only in fetal kidneys,

Figure 5. Immunolocalization of potential targets for cyclooxygenase(COX)-derived prostanoids in human kidney. (A) Localization ofmicrosomal prostaglandin E2 synthase (PGES) immunoreactive pro-tein in human kidney. In cortex, PGES was consistently associatedwith the macula densa (a through c) and with collecting ducts (d).Scale bar � 50 �m. In medulla, PGES was observed uniformly incollecting duct epithelium (e). Anti-PGES antibody detected a singleprotein with the expected size by Western blot test with protein fromhuman kidney inner medulla (f). (B) Localization of prostaglandin E2

(PGE2) EP4 receptors in human kidney. In adjacent serial sections,COX-2 and EP4 receptors were colocalized in arteries (g, h) and

preglomerular arterioles (i, j). Scale bar in (g) � 50 �m. (C) Local-ization of cAMP-coupled PGE2-EP2 receptor immunoreactive proteinin human kidney. EP2 was associated with distal convoluted tubule(k) and loop of Henle (l). Arrows indicate the macula densa regionwhere the EP2 signal waned (k).


whereas vascular expression was a constant feature of humankidney established during fetal life.

Because of the reported controversies on COX-2 localization

in human kidney, the specificity of the COX-2 antibody was ofmajor concern. Two separate anti-human COX-2 antibodiesdirected against different, specific epitopes yielded a similarpattern of histochemical labeling that was different from that ofCOX-1. Labeling was prevented by preabsorption with thepeptides used for immunization. Both antibodies reacted with aprotein with the expected molecular size of COX-2 on Westernblot test of kidney tissue and renal artery, and the intrarenaldistribution of COX-2 protein corresponded to COX-2 mRNA.Moreover, COX-2 immunopositive vessels were encounteredin all kidney zones similar to COX-2 protein and mRNA.COX-2 has previously been observed in adult human kidneyvasculature with a different COX-2 antibody (8) and also innonrenal human arteries (23) and in vascular endothelium ofseveral other species (26,27).

The vascular localization of COX-2 was not restricted tohuman kidney because an array of human organs displayedimmunolabeling associated with media of arteries and arte-rioles. Together, the data indicate that renovascular expressionof COX-2 is a general feature not causally related to thepresence of adjacent malignant cells in the human kidney tissueanalyzed. Moreover, despite different gender (three women,eight men), age (range, 45 to 83 yr, median 63 yr), tumor size,and histopathological tumor type (clear cell, transitional pap-illary tumor), there was very little variation between patients inintrarenal COX mRNA and protein levels. This suggests con-stitutive expression in adult human kidney. It is established thatvascular COX-2 expression is stimulated by the shear stress-nitric oxide pathway, by AngII, and by inflammatory mediatorsand cytokines (28–33). COX-2 couples preferentially withprostacyclin (PGI) synthase and inducible PGE2 synthase (34),and COX-2 is the predominant source of PGI2 in vivo(2,23,35,36). Tonical vascular prostacyclin synthesis is crucialto antagonize thromboxane-mediated, injury-induced prolifer-ation and thrombosis (37). The present observation of wide-spread vascular COX-2 offers an explanation for the inhibitoryeffect of COX-2 antagonists on prostacyclin synthesis in hu-mans (2,23,35,36).

In kidneys with arterial stenosis, vascular COX-2 expressionwas augmented in cortex and medulla. Human kidneys witharterial stenoses release net prostaglandin to the systemic cir-culation and COX blockers cause a drop of GFR (19), plasmarenin, and BP (17–19,38). Thus, vascular COX-2 activity,either from smooth muscle or from endothelium, may contrib-ute to elevated renal prostaglandin synthesis under conditionswith diminished blood flow to the kidney. PGE2 and/or PGI2

may serve to support vessel patency and blood supply to theischemic kidney through interaction with vascular IP or EP4receptors (present data) (24,39). AngII has been shown toinduce COX-2 in smooth muscle (29), but COX-2 mRNA wasnot changed in nephrectomy samples from patients that re-ceived AngII receptor blockers or ACE inhibitors. However,because of the few samples analyzed, we cannot exclude aregulatory role of AngII on COX-2 in the human kidney.

In cortex, immunolabeling for PGES was associated withmacula densa and collecting ducts and was not observed in thevasculature similar to rat (40). We did not observe any COX

Figure 6. Expression of cyclooxygenase (COX) in kidneys with renal arterystenosis. (A, B) Autoradiographs displaying hybrids for COX-1, COX-2, andrenin mRNAs in human kidney regions from separate patients with renalartery stenosis and renovascular hypertension (RVH). tRNA is a negativecontrol where probes hybridized with yeast tRNA, and probe bands displaythe radiolabeled probe with no tRNA. (C) Hybrids were cut from the gels,and protected radioactivity in the probes was assayed in a � counter. Eachcolumn displays average, actin-normalized, COX-2 mRNA level � SEM(RVH, n � 4; control cortex, n � 6, and medulla, n � 5). OM, outer medulla.(D) COX-2 immunostaining of kidney tissue from a patient with renal arterystenosis. Abnormal cortical architecture is seen with sclerotic glomeruli,obliterated vessels, fibrosis, and loss of tubules. COX-2 was detected invessel media and endothelium in cortex. Inner medulla was atrophied in thesekidneys. Scale bar � 50 �g.


isoform associated with the macula densa in the adult kidneysanalyzed. This observation essentially confirms previous re-ports where COX-2 was absent in normal adult human kidneymacula densa (7,8,11) but present in fetal kidney (9). COX-2 isthought to be induced in human macula densa in conditionswith low physiologic, or compromised, transepithelial NaCltransport (9,10,11,13). Dependence of plasma renin on COX-2has been shown in normal volunteers given a low-salt diet orfurosemide (16) and in patients with Bartter syndrome (15).The data suggest that COX-2–mediated prostaglandin synthe-sis contributes to macula densa–mediated signal transfer torenin-producing granular cells in these situations. None of thepatients in the study presented here had overt heart disease orreceived loop diuretics; all were elderly and presumably NaClreplete. Thus, we propose that COX-2 protein is below thedetection limit in macula densa of sodium-replete humans butthat it can be rapidly induced and generate substrate for PGESin states of low NaCl intake, diuretic treatment, or defecttransport proteins.

In summary, the present data show constitutive expressionof both COX isoforms in normal adult human kidney. COX-1was colocalized with PGES in collecting ducts. COX-2 waslocalized in vascular smooth muscle and endothelium togetherwith EP4 receptors both in kidney and extrarenal tissues.COX-2 was expressed in cTAL and macula densa in humanfetal kidney during and after active nephrogenesis.

AcknowledgmentsThis work was supported by grants from the Danish Medical

Research Council (22010122, 22010159), the Novo Nordisk Founda-tion, the Danish Heart Foundation (01123022896), the Danish Med-ical Association Research Fund, A. J. Andersens Foundation, J. C.Christoffersens Foundation, and Johannes Fogs Foundation. The tech-nical assistance of Mette Fredenslund and Inge Andersen is gratefullyacknowledged. We thank Anthony M. Carter for English-languagerevision.

References1. Bombardier C, Laine L, Reicin A, Shapiro D, Burgos-Vargas R,

Davis B, Day R, Ferraz MB, Hawkey CJ, Hochberg MC, KvienTK, Schnitzer TJ, VIGOR Study Group: Comparison of uppergastrointestinal toxicity of rofecoxib and naproxen in patientswith rheumatoid arthritis. VIGOR Study Group. N Engl J Med343: 1520–1528, 2000

2. Catella-Lawson F, McAdam B, Morrison BW, Kapoor S, KujubuD, Antes L, Lasseter KC, Quan H, Gertz BJ, FitzGerald GA:Effects of specific inhibition of cyclooxygenase-2 on sodiumbalance, hemodynamics, and vasoactive eicosanoids. J Pharma-col Exp Ther 289: 735–741, 1999

3. Swan SK, Rudy DW, Lasseter KC, Ryan CF, Buechel KL,Lambrecht LJ, Pinto MB, Dilzer SC, Obrda O, Sundblad KJ,Gumbs CP, Ebel DL, Quan H, Larson PJ, Schwartz JI, MuslinerTA, Gertz BJ, Brater DC, Yao SL: Effect of cyclooxygenase-2inhibition on renal function in elderly persons receiving a low-salt diet: A randomized, controlled trial. Ann Intern Med 133:1–9, 2000

4. Rossat J, Maillard M, Nussberger J, Brunner HR, Burnier M:Renal effects of selective cyclooxygenase-2 inhibition in normo-

tensive salt-depleted subjects. Clin Pharmacol Ther 66: 76–84,1999

5. Whelton A, White WB, Bello AE, Puma JA, Fort JG: Effects ofcelecoxib and rofecoxib on blood pressure and edema in patients�65 years of age with systemic hypertension and osteoarthritis.Am J Cardiol 90: 959–963, 2002

6. Morales E, Mucksavage JJ: Cyclooxygenase-2 inhibitor–associ-ated acute renal failure: Case report with rofecoxib and review ofthe literature. Pharmacotherapy 22: 1317–1321, 2002

7. Khan KN, Venturini CM, Bunch RT, Brassard JA, Koki AT,Morris DL, Trump BF, Maziasz TJ, Alden CL: Interspeciesdifferences in renal localization of cyclooxygenase isoforms:Implications in nonsteroidal antiinflammatory drug-related neph-rotoxicity. Toxicol Pathol 26: 612–620, 1998

8. Kömhoff M, Froone HJ, Klein T, Seyberth HW, Nüsing RM:Localization of cyclooxygenase-1 and -2 in adult and fetal hu-man kidney: Implication for renal function. Am J Physiol 272:F460–F468, 1997

9. Khan KN, Stanfield KM, Dannenberg A, Seshan SV, BaergenRN, Baron DA, Soslow RA: Cyclooxygenase-2 expression in thedeveloping human kidney. Pediatr Dev Pathol 4: 461–466, 2001

10. Koki A, Khan NK, Woerner BM, Dannenberg AJ, Olson L,Seibert K, Edwards D, Hardy M, Isakson P, Masferrer JL:Cyclooxygenase-2 in human pathological disease. Adv Exp MedBiol 507: 177–184, 2002

11. Kömhoff M, Jeck ND, Seyberth HW, Grone HJ, Nüsing RM,Breyer MD: Cyclooxygenase-2 expression is associated with therenal macula densa of patients with Bartter-like syndrome. Kid-ney Int 58: 2420–2424, 2000

12. Harris RC, McKanna JA, Akai Y, Jacobson HR, Dubois RN,Breyer MD: Cyclooxygenase-2 is associated with the maculadensa of rat kidney and increases with salt restriction. J ClinInvest 94: 2504–10, 1994

13. Khan KN, Stanfield KM, Harris RK, Baron DA: Expression ofcyclooxygenase-2 in the macula densa of human kidney in hy-pertension, congestive heart failure, and diabetic nephropathy.Ren Fail 23: 321–330, 2001

14. Nantel F, Meadows E, Denis D, Connolly B, Metters KM, GiaidA: Immunolocalization of cyclooxygenase-2 in the macula densaof human elderly. FEBS Lett 457: 475–477, 1999

15. Reinalter SC, Jeck N, Brochhausen C, Watzer B, Nusing RM,Seyberth HW, Komhoff M: Role of cyclooxygenase-2 in hyper-prostaglandin E syndrome/antenatal Bartter syndrome. KidneyInt 62: 253–260, 2002

16. Kammerl MC, Nusing RM, Schweda F, Endemann D, StubanusM, Kees F, Lackner KJ, Fischereder M, Kramer BK: Lowsodium and furosemide-induced stimulation of the renin systemin man is mediated by cyclooxygenase 2. Clin Pharmacol Ther70: 468–74, 2001

17. Milot A, Lambert R, Lebel M, Cusson JR, Larochelle P: Pros-taglandins and renal function in hypertensive patients with uni-lateral renal artery stenosis and patients with essential hyperten-sion. J Hypertens 14: 765–771, 1996

18. Zipser RD, Speckart PF, Zia PK, Hahn JA, Boswell WP, HortonR: Release of immunoassayable prostaglandin E by the humanischemic kidney. J Clin Endocrinol Metab 47: 914–917, 1978

19. Imanishi M, Ohta M, Kawamura M, Akabane S, Matsushima Y,Kuramochi M, Kojima S, Kimura K, Takamiya M, Ito K: Aspirintest for differentiation of unilateral renovascular hypertensionfrom hyperreninemic essential hypertension. Am J Hypertens 4:761–768, 1991


20. Andreasen D, Jensen BL, Hansen PB, Kwon TH, Nielsen S, andSkott O: The alpha(1G)-subunit of a voltage-dependent Ca(2�)channel is localized in rat distal nephron and collecting duct.Am J Physiol Renal Physiol 279: F997–1005, 2000

21. Hobart PM, Fogliano M, O’Connor BA, Schaefer IM, ChirgwinJM: Human renin gene: Structure and sequence analysis. ProcNatl Acad Sci U S A 81: 5026–5030, 1984

22. Stubbe J, Jensen BL, Bachmann S, Morsing P, Skott O: Cyclo-oxygenase-2 contributes to elevated renin in the early postnatalperiod in rats. Am J Physiol Regul Integr Comp Physiol 284:R1179–R1189, 2003

23. Belton O, Byrne D, Kearney D, Leahy A, Fitzgerald DJ: Cyclo-oxygenase-1 and -2 dependent prostacyclin formation in patientswith atherosclerosis. Circulation 102: 840–845, 2000

24. Morath R, Klein T, Seyberth HW, Nusing RM: Immunolocal-ization of the four prostaglandin E2 receptor proteins EP1, EP2,EP3, and EP4 in human kidney. J Am Soc Nephrol 10: 1851–1860, 1999

25. Jensen BL, Stubbe J, Hansen PB, Andreasen D, Skott O: Local-ization of prostaglandin E(2) EP2 and EP4 receptors in the ratkidney. Am J Physiol Renal Physiol 280: F1001–1009, 2001

26. Stanfield KM, Khan KN, Gralinski MR: Localization of cyclo-oxygenase isozymes in cardiovascular tissues of dogs treatedwith naproxen. Vet Immunol Immunopathol 80: 309–314, 2001

27. Heymes C, Habib A, Yang D, Mathieu E, Marotte F, Samuel J,Boulanger CM: Cyclo-oxygenase-1 and -2 contribution to endothe-lial dysfunction in ageing. Br J Pharmacol 131: 804–810, 2000

28. Eligini S, Habib A, Lebret M, Creminon C, Levy-Toledano S,Maclouf J: Induction of cyclo-oxygenase-2 in human endothelialcells by SIN-1 in the absence of prostaglandin production. Br JPharmacol 133: 1163–1171, 2001

29. Ohnaka K, Numaguchi K, Yamakawa T, Inagami T: Induction ofcyclooxygenase-2 by angiotensin II in cultured rat vascularsmooth muscle cells. Hypertension 35: 68–75, 2000

30. Pritchard KA Jr, O’Banion MK, Miano JM, Vlasic N, BhatiaUG, Young DA, Stemerman MB: Induction of cyclooxygenase-2in rat vascular smooth muscle cells in vitro and in vivo. J BiolChem 269: 8504–8509, 1994

31. Miralpeix M, Camacho M, Lopez-Belmonte J, Canalias F, BeletaJ, Palacios JM, Vila L: Selective induction of cyclo-oxygenase-2

activity in the permanent human endothelial cell line HUV-EC-C: Biochemical and pharmacological characterization. Br JPharmacol 121: 171–180, 1997

32. Inoue H, Taba Y, Miwa Y, Yokota C, Miyagi M, Sasaguri T:Transcriptional and posttranscriptional regulation of cyclooxy-genase-2 expression by fluid shear stress in vascular endothelialcells. Arterioscler Thromb Vasc Biol 22: 1415–1420, 2002

33. Topper JN, Cai J, Falb D, Gimbrone MA Jr: Identification ofvascular endothelial genes differentially responsive to fluid me-chanical stimuli: Cyclooxygenase-2, manganese superoxide dis-mutase, and endothelial cell nitric oxide synthase are selectivelyup-regulated by steady laminar shear stress. Proc Natl Acad SciU S A 93: 10417–10422, 1996

34. Ueno N, Murakami M, Tanioka T, Fujimori K, Tanabe T, UradeY, Kudo I: Coupling between cyclooxygenase, terminal prosta-noid synthase, and phospholipase A2. J Biol Chem 276: 34918–34927, 2001

35. Cullen L, Kelly L, Connor SO, Fitzgerald DJ: Selective cyclo-oxygenase-2 inhibition by nimesulide in man. J Pharmacol ExpTher 287: 578–582, 1998

36. McAdam BF, Catella-Lawson F, Mardini IA, Kapoor S, LawsonJA, FitzGerald GA: Systemic biosynthesis of prostacyclin bycyclooxygenase (COX)-2: The human pharmacology of a selec-tive inhibitor of COX-2. Proc Natl Acad Sci U S A 96: 272–277,1999

37. Cheng Y, Austin SC, Rocca B, Koller BH, Coffman TM, GrosserT, Lawson JA, FitzGerald GA: Role of prostacyclin in thecardiovascular response to thromboxane A2. Science 296: 539–541, 2002

38. Imanishi M, Tsuji T, Nakamura S, Takamiya M: ProstaglandinI(2)/E(2) ratios in unilateral renovascular hypertension of differ-ent severities. Hypertension 38: 23–29, 2001

39. Kömhoff M, Lesener B, Nakao K, Seyberth HW, Nusing RM:Localization of the prostacyclin receptor in human kidney. Kid-ney Int 54: 1899–908, 1998

40. Campean V, Theilig F, Paliege A, Breyer M, Bachmann S: Keyenzymes for renal prostaglandin synthesis: Site-specific expres-sion in rodent kidney (rat, mouse). Am J Physiol Renal Physiol285: F19–F32, 2003


cycloxygenase-2 is expressed in vasculature of normal and...

Documents