INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

STUDIES ELUCIDATING the role of prostanoids in the kidney have identified apparently conflicting effects. Prostaglandins (PGs) inhibit tubular transport and promote the excretion of NaCl and water; for example, prostaglandin E₂ (PGE₂) inhibits the absorption of NaCl along medullary thick ascending limb and collecting tubule (3, 17) and antagonizes the hydrosmotic effect of vasopressin (2). These functional changes suggest participation of PGs in the adjustment of renal function during extracellular volume expansion. In contrast, the circulatory effects of PGs, to promote renin secretion and to protect renal blood flow and glomerular filtration rate, appear to be active only in volume depletion (4, 10). These two opposing functional roles can be reconciled if PG formation is compartmentalized and subject to differential regulation. The present study was undertaken to examine whether expression of cyclooxygenase, one of the potential rate-limiting steps in PG formation, is subject to differential regulation in the kidney.

Two isoforms of cyclooxygenase (COX) or prostaglandin H synthase have been identified by molecular approaches, a constitutive form (COX-1) and an inducible form (COX-2) (6, 9, 27, 29). The two forms share similar enzymatic properties but differ markedly with respect to cellular expression pattern and regulation (5, 14). COX-1 is expressed constitutively in a wide variety of tissues. This is the isoform that has generally been predicted to regulate renal H₂O and Na transport, as well as to have other "housekeeping" functions, such as control of platelet function and gastroprotection (5). COX-2 is much more restricted in its expression; in certain cell types, it can be dramatically induced by inflammatory stimuli or growth factors (16, 25). It has a role in inflammatory responses and may act to diminish activation of the apoptotic pathway (39). One of the few cell types where constitutive expression of COX-2 has been described is the macula densa in the renal cortex (13), an epithelial cell type that is part of the juxtaglomerular apparatus. There is substantial evidence that these cells play a key role in regulation of renin release and local control of vascular tone (3).

The present studies examined the effect of altering dietary salt intake on renal COX expression. We found that COX-1 expression did not change with altered Na intake. In contrast, expression of COX-2 showed cell-specific regulation. In confirmation of earlier observations (13), a low-Na diet increased COX-2 in the renal cortex, particularly the macula densa cells. Unexpectedly, however, a high-Na diet increased COX-2 in the renal medulla. These results are consistent with the concept that cortical prostaglandin production is upregulated during volume depletion, thus diminishing locally the impact of vasoconstrictor hormones, whereas medullary prostaglandin production increases in volume expansion, thus promoting the excretion of NaCl and water.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Animals and dissection. Experiments were done in Sprague-Dawley rats (body wt range, 175-200 g) fed a control, low-salt, or high-salt diet. Rats in the low-salt group (n = 10) received an intraperitoneal injection of 2 mg/kg furosemide on day 1 and were kept on a diet containing 0.03% NaCl for 7 days with drinking water ad libitum. Rats in the high-salt group (n = 10) received a diet containing 3% NaCl and 0.45% NaCl as drinking fluid for 7 days. The control group (n = 3) was fed a standard diet containing 0.2% NaCl and had free access to water.

Kidneys for nephron segment dissection were obtained from animals anesthetized by an intraperitoneal injection of 110 mg/kg thiobutabarbital (Inactin; Byk Gulden, Constance, Germany). Pieces of tissue from cortex, outer medulla, and inner medulla were cut from coronal slices, and tissue samples were transferred immediately into microcentrifuge tubes containing 300 µl TRI Reagent (Molecular Research Center, Cincinnati, OH). Tubes were snap frozen in liquid nitrogen and stored at 80°C for later RNA isolation and cDNA synthesis. For microdissections, the aorta was cannulated below the level of the kidneys and ligated proximal to the origin of the left renal artery. After the renal vein was severed, the left kidney was perfused with 30 ml of cold saline, followed by perfusion with 30 ml cold Dulbecco's modified Eagle's medium (DME; Sigma Chemical, St. Louis, MO), containing 1 mg/ml collagenase. The kidney was removed, cut into coronal slices, and incubated in the collagenase solution at 37°C for 18-22 min. Slices were rinsed with ice-cold phosphate-buffered saline [containing (in mM) 140 NaCl, 3 KCl, 1.5 KH₂PO₄, and 8 NaH₂PO₄ · H₂O], placed into DME medium containing 1% fetal calf serum, and maintained at 4°C during dissection. Segments of proximal convoluted tubule (PCT), proximal straight tubule (PST), medullary (MTAL) and cortical thick ascending limbs (CTAL) of the loop of Henle, distal convoluted tubule (DCT), cortical collecting duct (CCD), outer medullary collecting duct (OMCD), and inner medullary collecting duct (IMCD) were dissected and identified by intrarenal location and appearance. To obtain the macula densa-containing segment (MDCS), glomeruli were dissected together with attached portions of the distal nephron. After removal of the ascending limb and distal tubule, the adherent tubular segment containing the macula densa cells was carefully detached from the glomerulus. Length of these specimens was typically 50-75 µm. Specimens dissected in this manner include some adjacent thick ascending limb cells but have previously been shown to be highly enriched in macula densa markers (41). For comparison, glomeruli were dissected free of the adherent macula densa. Lengths of the dissected segments were measured with a calibrated eyepiece micrometer. In general, 5-10 glomeruli, 10 MDCS (0.5-0.7 mm), or 2-10 mm of other tubule segments were pooled to constitute one sample. Samples were transferred in 10 µl dissection medium into 100 µl guanidine isothiocyanate (GITC) buffer (4 M GITC, 25 mM sodium acetate, and 0.8% -mercaptoethanol), snap frozen in liquid nitrogen, and stored at 80°C for later RNA extraction and cDNA synthesis. To detect background contamination, 10 µl of dissection medium were obtained at the termination of dissection and run as a blank through the RNA isolation, cDNA synthesis, and polymerase chain reaction (PCR) process.

Cultured cells. RNA was isolated from cultured primary rat mesangial cells, prepared by standard methods (33), and from three cells lines: renal medullary interstitial cells (18), from mIMCD-K2 cells (30), and from M-1 cells (38).

Reverse transcription-PCR. To isolate RNA, tubular and glomerular samples were thawed in an ice slurry bath and sonicated for 15 s. After 20 µg Escherichia coli ribosomal RNA (Boehringer-Mannheim, Indianapolis, IN) was added as carrier, the lysate was layered onto a discontinuous gradient of CsCl (100 µl 97% CsCl and 20 µl 40% CsCl) in a 220-µl polycarbonate ultracentrifuge tube. Specimens were centrifuged at 16°C for 2 h at 300,000 g in a TLA 100 ultracentrifuge with fixed angle rotor (Beckman Instruments, Fullerton, CA). The RNA pellet was redissolved in 0.3 M sodium acetate and precipitated with ethanol.

cDNA was synthesized by incubating RNA at 42°C for 1 h in a total volume of 20 µl of the manufacturer's buffer, containing 100 units of human Moloney murine leukemia virus reverse transcriptase (Superscript, GIBCO-BRL; Life Technologies, Gaithersburg, MD), 0.5 µg oligo(dT) (Pharmacia, Piscataway, NJ) as primer, 20 units of RNasin (Promega Biotech, Madison, WI), 10 mM dithiothreitol, 0.5 mM dNTP (Pharmacia), and 1% bovine serum albumin (Boehringer-Mannheim). cDNA was precipitated with 5% linear acrylamide and resuspended in tris(hydroxymethyl)aminomethane-EDTA (Tris-EDTA) buffer at a dilution adjusted so that 5 µl of cDNA corresponded to 1 mm of tubule or 1.4 glomeruli, except for the MDCS, which was resuspended at a ratio of 5 µl of cDNA/0.1 mm tubule.

PCR reactions were performed in a total volume of 50 µl in the presence of (in mM) 0.1 dNTP, 10 dithiothreitol, 50 KCl, 1.5 MgCl₂, 10 Tris (pH 8.3), 0.001% gelatin, 0.5 pmol of each primer, 1.25 U Taq DNA polymerase (Perkin-Elmer Cetus, Norwalk, CT), 1.5 µCi [³²P]dCTP (Amersham, Arlington Heights, IL), and 1-6 µl of tissue cDNA or control template. Mineral oil was layered on top of each sample to prevent evaporation. After an initial denaturation at 94°C for 3.5 min, PCR amplification was performed for 30 cycles at 94°C (denaturation), 54°C (annealing), and 72°C (extension) at cycle durations of 1.5 min for denaturation, annealing, and extension. Samples were incubated for an additional 8 min at 72°C before completion.

Initial PCR determinations used 2 µl of cDNA (equivalent to a 0.4 mm/tubule or 0.3 glomerulus) to assess for presence or absence of product. When product was detected, a second series of PCR determinations was performed for quantitation. The quantitative assays used serial 10-fold dilutions of cDNA of each experimental sample to ensure that all PCR determinations fell within the linear amplification range. Thus it was possible to ensure that quantitative estimates were derived only from determinations in the range where product abundance was proportionate to the amount of initial template. Water, dissection medium blanks, and samples carried through RNA isolation and cDNA synthesis without adding reverse transcriptase were run as controls. Dissection medium and reverse transcription (RT)-negative samples consistently yielded no product.

Primer selection. Primers for both COX-1 and COX-2 were derived from published sequences (9). Primers were chosen in areas of minimal cross homology between COX-1 and COX-2 and high species conservation and positioned to span at least two intron-exon junctions, to distinguish cDNA from genomic DNA. The sequence of the oligonucleotide primers and their location in the published cDNA sequence are as follows: sense COX-1, 5' CTG CTG AGA AGG GAG TTC CAT 3' (bp 602-621); antisense COX-1, 5' GTC ACA CAC ACG GTT ATG CT (bp 981-1000); sense COX-2, 5' ACA CTC TAT CAC TGG CAT CC 3' (bp 1229-1248); and antisense COX-2, 5' GAA GGG ACA CCC TTT CAC AT 3' (bp 1794-1813). COX-1 and COX-2 primers are predicted to amplify products of 398 and 584 bp, respectively. Concurrent RT-PCR amplification of -actin as a housekeeping gene was used to control for variations in the efficiencies of RNA isolation and RT, as previously described (41).

Product identity was confirmed by restriction enzyme digestion of PCR products, using commercially available restriction enzymes (New England Biolabs or Boehringer-Mannheim) and incubation conditions suggested by the manufacturer. Sequencing of plasmid DNA was performed by the dideoxy chain termination method, using sequence-specific primers and Sequenase 2.0 (U.S. Biochemical, Cleveland OH).

Western blotting. Slices of cortex, outer medulla, and inner medulla were dissected and homogenized in 30 mM Tris · HCl, pH 7.5, and 100 µM phenylmethylsulfonyl fluoride. Proteins from whole cell lysate were separated on sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis gel and transferred onto nitrocellulose membrane. The blots were blocked overnight with 5% nonfat dry milk in Tris-buffered saline (TBS), followed by incubation for 1 h with rabbit anti-murine polyclonal antiserum to COX-2 (Cayman Chemical, Ann Arbor, MI) at 1:500. After washing with TBS, blots were incubated with goat anti-rabbit horseradish peroxidase-conjugated secondary antibody and visualized with enhanced chemiluminescence ECL kits (Amersham). For immunoblotting of COX-1, the blot was stripped in 100 mM 2-mercaptoethanol, 2% SDS, and 62.5 Tris · HCl, pH 6.7, incubated at 50°C for 30 min. Immunodetection was performed as described, except that murine monoclonal antibody to COX-1 (Cayman Chemical) at 1:2,000 was used for the primary antibody incubation.

Statistical analysis. Data are expressed as means ± SE. All statistical comparisons were made with a paired Student's t-test, with P < 0.05 considered significant.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Confirmation of PCR product identity. PCR product identity was verified by restriction digestion, using glomerular cDNA for COX-1 and macula densa cDNA for COX-2 (Fig. 1). The COX-1 PCR product was 398 bp in length, and digestion with the restriction endonucleases Ava I, Acc I, and Stu I yielded the expected products. The COX-2 PCR product was 584 bp in length, and fragments obtained after digestion with Hinf I and Pst I were of expected sizes. In addition, 140 bases from the 5' end of the COX-2 product were sequenced and were found to be identical to the published cDNA sequence (data not shown).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Verification of polymerase chain reaction (PCR) product identity by restriction endonuclease digestion. A: cyclooxygenase-1 (COX-1) product, derived from glomerulus, was digested with Ava I, Acc I, and Stu I. B: COX-2 product, derived from microdissected macula densa segment, was digested with Hinf I and Pst I.

Distribution of COX-1 and COX-2 expression in kidney regions. The expression of COX isoforms in renal cortex, outer medulla, and inner medulla was determined by RT-PCR and Western blotting. The two COX isoforms exhibited a similar distribution pattern in the different regions of the kidney. By Western blotting, both COX-1 and COX-2 were virtually exclusively localized to inner medulla (Fig. 2). By RT-PCR, COX-1, as well as COX-2, mRNAs were also predominantly detected in inner medulla, but small signals were also obtained in cortex and outer medulla (in the example shown in Fig. 2, the cortical COX-1 mRNA signal is not seen because of insufficient exposure time). The apparent discrepancy in the pattern of outer medullary and cortical COX-1 and COX-2 expression by Western blotting and RT-PCR is presumably because of the difference in sensitivity of the two techniques.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Localization of COX-1 and COX-2 expression in kidney regions by reverse-transcription (RT)-PCR (A) and Western blotting (B). IM and OM, inner and outer medulla, respectively; CO, cortex.

Expression of COX-1 and COX-2 in cultured renal cells. Expression of both COX-1 and COX-2 mRNA was assessed in cultured cells of renal origin. By both RT-PCR (as shown in Fig. 3) and Western blotting (data not shown), both COX-1 and COX-2 were abundant in renal interstitial cells. COX-1 was also abundant in collecting duct-derived cells and in primary mesangial cells, but COX-2 expression was substantially lower in these other cell types.

View larger version (63K): [in this window] [in a new window]   Fig. 3.   Expression of COX-1 and COX-2 in cultured cells of renal origin. Shown is PCR product obtained from cDNA from two collecting inner medullary collecting duct-derived cell lines (NK-2 and M1), from primary mesangial cells, and from renal interstitial cells (RIC).

Localization of COX-1 along the nephron. Abundant amounts of COX-1 mRNA were found in the glomerulus. When the glomerulus was separated from the adherent distal tubule segment, which includes the macula densa, COX-1 product was found to be localized in the glomerulus rather than in the adherent tubule (Fig. 4). COX-1 was also consistently detected in MTAL and all portions of the collecting duct (CCD, OMCD, and IMCD). Low levels of expression (approximately ×2 background) were occasionally detected in PCT (2 of 7), but COX-1 PCR products were not obtained from cDNA derived from PST (0 of 6), CTAL (0 of 7), or MDCS (0 of 6). The relative abundance of COX-1 product (expressed per glomerulus or mm tubule) along the nephron was IMCD = CCD = OMCD > glomerulus > MTAL.

View larger version (52K): [in this window] [in a new window]   Fig. 4.   Gels from representative PCR runs showing localization of COX-1 and COX-2 mRNA in glomeruli (Glom) without macula densa-containing segment (MDCS), MDCS, and cortical thick ascending limb of Henle's loop (CTAL).

Localization of COX-2 along the nephron. COX-2 PCR products were regularly and abundantly amplified in cDNA from MDCS and CTAL (Fig. 4) and at substantially lower levels in IMCD. COX-2 mRNA was occasionally detected but at low levels in glomeruli without MDCS (4 of 20), PCT (4 of 10), DCT(3 of 6), and OMCD (4 of 10). Product was consistently absent in PST (0 of 10), MTAL (0 of 8), and CCD (0 of 10). The relative yield of COX-2 product, expressed per millimeter tubule, was assessed by limiting dilution in the three segments in which it was consistently expressed: for the three segments, product abundance was in the approximate ratio of 100:10:1 (MDCS:CTAL:IMCD). Representative gels showing expression of COX-1 and COX-2 mRNA in glomeruli without macula densa and in MDCS and surrounding CTAL are presented in Fig. 4.

Effect of dietary salt on COX-1 and COX-2 expression in kidney regions. Treatment of rats with a high-salt diet for 7 days induced a 4.5-fold increase in COX-2 mRNA (Fig. 5) and 9.5-fold increase in COX-2 protein (Fig. 6) levels in the inner medulla, whereas the low-salt dietary pretreatment had no significant effect on COX-2 expression in this kidney region (Figs. 5 and 6). In contrast, rats on a low-salt diet had a 3.3-fold increase in cortical COX-2 mRNA levels, whereas cortical COX-2 mRNA decreased 2.9-fold in response to high salt intake (Fig. 7). Levels of COX-1 mRNA or protein were not altered by varying salt intake in either cortex or inner medulla.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   A: representative RT-PCR gels of COX-2 and -actin from inner medulla of Sprague-Dawley rats fed with low-salt (LS), normal salt (NS), or high-salt (HS) diet. B: graphic analysis of relative abundance of COX-2 PCR product.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   A: Western blotting analysis of COX-1 and COX-2 in inner medulla of each 4 Sprague-Dawley rats fed with low, normal, or high-salt diets. B: densitometric analysis of COX-2 in inner medulla of normal and salt-manipulated rats.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   A: representative phosphoimages of COX-2, COX-1, and -actin RT-PCR products from cortex of Sprague-Dawley rats fed diets of low, normal, and high salt content. B: graphic analysis of relative abundance of renal cortical COX-2 PCR products.

Effect of dietary salt on COX-1 and COX-2 gene expression in microdissected nephron segments. We further examined the regulation of COX-1 and COX-2 expression in the microdissected nephron segments in Sprague-Dawley rats fed a low- or high-salt diet. The animals were handled in pairs, with treatment initiated on the same day and studies completed on the same day; paired samples of each segment were carried through RNA isolation, cDNA synthesis, and PCR quantitation. As in the initial series, PCR product for COX-1 was abundant in glomerulus, MTAL, CCD, OMCD, and IMCD. Changes in dietary salt intake did not significantly affect product abundance from any segment (Table 1). In this series, PCR product for COX-2 was again observed to be very abundant in MDCS and consistently detected in CTAL and IMCD (Fig. 8). Comparison of product yields (determined as ³²P incorporation into PCR product bands/mm tubule) is summarized in Table 1. Expression levels of MDCS and CTAL mRNA showed a 7.7-fold (n = 9, P = 0.004) and 3.3-fold (n = 5, P = 0.019) stimulation, respectively, in low-salt compared with high-salt animals (Fig. 8). PCR products were detected in IMCD of both low- and high-salt animals, with no significant differences in product yields between the two groups. The low levels of COX-2 expression in glomeruli without MDCS, PCT, DCT, and OMCD were not significantly altered by dietary salt.

View larger version (53K): [in this window] [in a new window]   Fig. 8.   RT-PCR gels from representative PCR runs showing COX-2 and -actin PCR product obtained from microdissected nephron segments from rats on high- and low-salt diets. Results are shown from 3 segments in which COX-2 was regularly detected.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Distribution of renal cyclooxygenases. The first aim of the present series of experiments was to examine the distribution of COX isoform expression in the rat kidney. The highest expression levels for both isoforms were observed in the inner medulla. The high basal levels of COX-1 were predicted, since this is the isoform generally assumed to mediate renal salt and water regulation (5). COX-2 was initially considered as inducible form of cyclooxygenase, involved predominantly in inflammation. However, data from both Harris et al. (13) and our present study have demonstrated the high basal levels of COX-2 in the medulla. The inner medulla thus joins the macula densa as one of the unusual tissue sites where COX-2 is constitutively expressed. The results of Western blots, admittedly only roughly quantitative, indicate that amount of COX-2 in the inner medulla is of the same order of magnitude as COX-1. Protein for both isoforms was also detected in cortex but at substantially lower levels.

At the cellular level, COX-1 mRNA was found to be widely distributed: it was expressed at substantial levels along the entire collecting duct, in MTAL segments, in glomeruli freed of adherent macula densa, and in cultured renal interstitial cells, mesangial, and collecting duct-derived cell lines. Expression of COX-2 was more restricted: it was expressed at high levels in cultured renal interstitial cells and in the macula densa segment but not in the remainder of the glomerulus, consistent with previous observations in rat (13) in CTAL and, at substantially lower levels, in IMCD. In one recent report (19), COX-2 was not detected in human macula densa, and it is possible that constitutive expression in the macula densa is species dependent. The low levels detected in IMCD would appear not to account for the very high medullary COX-2 abundance, and it is likely that the bulk of inner medullary COX-2 is not localized in epithelial cells but rather in medullary interstitial cells (13) and possibly in vascular cells (19). We cannot exclude the possibility that the COX-2 mRNA detected in IMCD was, in part, the result of contamination with vascular or interstitial cells, since IMCDs are difficult to dissect free of any adherent tissue.

Overall, the medullary predominance of renal COX expression is in agreement with numerous earlier studies demonstrating that the greatest fraction of prostaglandins formed in kidney originate from the medulla (4, 8) and with functional studies that show, even under basal conditions, interfering with prostaglandin synthesis alters medullary transport function (4). In cortical segments, on the other hand, under basal conditions, prostaglandin production appears to be low (8, 40), corresponding to the much lower levels of gene expression. The highly localized expression of COX-2 in a small cell population of the cortex is in agreement with the low COX-2 levels in the cortex as a whole. The presence of COX-2 in macula densa and surrounding CTAL cells suggests that prostanoids generated by macula densa cells play a role in juxtaglomerular apparatus-mediated regulation of vascular tone (tubuloglomerular feedback) and/or renin secretion, a conclusion supported by functional observations (3).

Effect of variations in salt intake on renal COX expression. Variations in salt intake did not alter the expression of COX-1 in cortex or medulla, and no changes were observed in dissected nephron segments. Previous studies in other tissues have shown that COX-1 levels do not vary substantially under a variety of experimental conditions, observations consistent with the categorization of COX-1 as a constitutive gene product. At the molecular level, this is reflected by the absence of a TATA box in the promoter region of the COX-1 gene (20). Our results suggest that the changes in local levels of prostaglandins with salt intake are probably not due to changes in COX-1 gene expression. It is worth noting, however, that COX-1-mediated prostaglandin production is likely to undergo regulatory changes, most notably by control of the release of arachidonic acid from membrane phospholipids. For example, angiotensin II has been shown to produce an immediate increase in prostaglandin formation by activation of a calcium-dependent phospholipase (33, 43).

In contrast to COX-1 expression, COX-2 expression was observed to vary markedly with variations in Na intake. Unexpectedly, changes in salt intake had directionally opposite effects on COX-2 expression in renal cortex and inner medulla. Whereas cortical COX-2 expression was stimulated by treatment with a low-Na diet and inhibited by high salt intake, medullary COX-2 levels increased following treatment with a high-salt diet and tended to be reduced by low salt intake. This differential expression strongly suggests that either COX-2 gene transcription or the stability of COX-2 mRNA is regulated in a cell-specific fashion. The mechanism for this cell-specific regulation is currently unclear. It is possible that the induction of cortical COX-2 gene expression by low salt is mediated by the stimulation of angiotensin II induced by volume depletion and that, since angiotensin II concentrations in the cortical interstitium may be higher than in the medulla, this effect shows regional differences. Angiotensin II has been shown to activate the mitogen-activated protein kinase (MAP kinase) pathway (32), and COX-2 induction can occur through this mechanism (12). On the other hand, expression of medullary COX-2 may be predominantly controlled by local tissue conditions, such as changes in osmolarity and mechanical stretch. Both tonicity and stretch have been shown to activate COX-2 expression in other cell types, in part at least, via MAP kinase pathways (1, 12, 42). Inner medullary interstitial cells, probably an important source of medullary COX-2, are tethered to vasa recta and loops of Henle (21) and thus potentially subjected to mechanical stimulation whenever flow in medullary structures changes. Thus variations in salt intake may be associated with different extracellular signals in the cortex and medulla; these responses could converge on similar or identical intracellular pathways controlling COX-2 gene expression.

Whatever the precise mechanism, the divergent response of cortical and medullary COX-2 activity may explain discrepancies in previous studies: renal PG production and urinary excretion have been reported to correlate directly, inversely, or not at all with salt intake (Ref. 4 and see below). Because the inner medulla is the dominant site of production of renal PGs, our findings would lead to the prediction that a high-salt diet should stimulate overall renal PG biosynthesis. This inference is supported by both short- and long-term experiments from a majority of previous studies. Several studies have shown that volume expansion causes an increase in urinary excretion or renal synthesis of PGs (34): chronic salt loading has been found to stimulate urinary excretion of PGE₂, PGI₂, and PGF₂ (4, 7, 10). Observations by Limas et al. (23, 24) in several strains of rats showed that chronic salt loading consistently stimulated renal medullary PGE₂ synthesis. Furthermore, microsomal fractions prepared from medullary tissue of rats on a high-NaCl diet had higher PG production than cortical microsomes (40). These findings were, however, inconsistent with a small numbers of studies in which low-salt diet or both low- and high-salt diet stimulated medullary PGE₂ synthesis (31, 36). This discrepancy may be related to individual salt manipulations, species, or technical difference. In our recent observation with enzyme immunoassay technique, we found that chronic sodium loading significantly increased the levels of PGE₂, as well as other PGs in both urine and renal medulla (unpublished results). The stimulatory effect of a low-salt diet on PG synthesis has been observed in the cortex (4, 40). Stimulation of cortical COX expression by a low-Na diet is consistent with several studies showing marked effects of COX inhibitors on glomerular filtration rate and renal vascular resistance but only in Na-depleted animals (4).

Modification of dietary salt intake is also associated with directionally opposite changes in the expression of cortical and medullary neuronal nitric oxide synthase (NOS), another enzyme expressed in macula densa cells (35) and in the inner medulla (26). Recent studies demonstrate that a low-salt diet stimulates brain-type NOS (bNOS) mRNA and protein in macula densa (35), whereas a high-salt diet stimulates bNOS, as well as endothelial and inducible NOS, in the inner medulla (26). There is evidence that NO may be involved in macula densa-mediated renin release, a response dominant in salt-depleted states (3, 10, 11). On the other hand, there is a large body of literature demonstrating vasodepressant and natriuretic effects mediated by the renal medulla (i.e., see refs. in Ref. 28); both medullary NO and COX products may participate in these responses, coordinately regulated by dietary salt.

In summary, the present study examined COX-1 and COX-2 expression in the kidney and their regulation by dietary salt. Expression of both COX-1 and COX-2 was predominately localized to inner medulla, but the two isoforms were markedly different in cortical localization, with COX-1 detected in glomerulus and COX-2 detected in the macula densa-containing segment and CTAL. We demonstrated that low- and high-salt diet enhanced COX-2 expression in macula densa and inner medulla, respectively, whereas COX-1 expression was unaffected by salt intake. Divergent regulation of COX-2 in macula densa and renal medulla by dietary salt suggests that PGs in different kidney regions serve different functions. Medullary production of prostaglandins may play a role in promoting the excretion of salt and water and may thereby protect against volume overload, while cortical prostaglandins may protect glomerular circulation from the impact of vasoconstrictor agents formed as part of the volume depletion response.