COX2 activity promotes organic osmolyte accumulation and adaptation of renal medullary interstitial cells to hypertonic stress.

The mechanism by which COX2 inhibition decreases renal cell survival is poorly understood. In the present study we examined the effect of COX2 activity on organic osmolyte accumulation in renal medulla and in cultured mouse renal medullary interstitial cells (MMICs) and its role in facilitating cell survival. Hypertonicity increased accumulation of the organic osmolytes inositol, sorbitol, and betaine in cultured mouse medullary interstitial cells. Pretreatment of MMICs with a COX2-specific inhibitor (SC58236, 10 micromol/liter) dramatically reduced osmolyte accumulation (by 79 +/- 9, 57 +/- 12, and 96 +/- 10% for inositol, sorbitol, and betaine respectively, p < 0.05). Similarly, 24 h of dehydration increased inner medullary inositol, sorbitol, and betaine concentrations in vivo by 85 +/- 10, 197 +/- 28, and 190 +/- 24 pmol/microg of protein, respectively, but this increase was also blunted (by 100 +/- 5, 66 +/- 15, and 81 +/- 9% for inositol, sorbitol, and betaine, respectively, p < 0.05) by pretreatment with an oral COX2 inhibitor. Dehydrated COX2-/- mice also exhibited an impressive defect in sorbitol accumulation (88 +/- 9% less than wild type, p < 0.05) after dehydration. COX2 inhibition (COX2 inhibitor-treated or COX2-/- MMICs) dramatically reduced the expression of organic osmolyte uptake mechanisms including betaine (BGT1) and sodium-myo-inositol transporter and aldose reductase mRNA expression under hypertonic conditions. Importantly, preincubation of COX2 inhibitor-treated MMICs with organic osmolytes restored their ability to survive hypertonic stress. In conclusion, osmolyte accumulation in the kidney inner medulla is dependent on COX2 activity, and providing exogenous osmolytes reverses COX2-induced cell death. These findings may have implications for the pathogenesis of analgesic nephropathy.

The maintenance of intracellular solute concentrations and cell volume is a prerequisite for optimal cell function. Changes in extracellular tonicity typically result in corresponding changes in intracellular solute concentration that allow cells to regain their normal size (1). In renomedullary cells, osmotic equilibrium with elevated extracellular NaCl concentrations is achieved primarily by intracellular accumulation of small organic solutes ("organic osmolytes") (2,3). In contrast to inorganic electrolytes, organic osmolytes do not severely impair cell function even at high concentrations and are thus considered "compatible osmolytes" (3). Reduced accumulation of compatible osmolytes may cause decreased cell viability in the renal medulla (4) as well as inhibition of growth and, ultimately, cell death (5).
Prostanoids comprise a diverse family of biologically active lipids derived from the enzymatic metabolism of arachidonic acid by prostaglandin G 2 /H 2 synthase (cyclo-oxygenase (COX)). 1 Although prostanoid synthesis occurs in all cells and tissues, the kidney is a particularly rich source for prostanoids. Cyclo-oxygenase is a key enzyme regulating the formation of prostaglandins from arachidonic acid, and two isoforms of COX have been identified, a constitutively expressed isoform (COX1) and an inducible isoform (COX2) (6 -8). These two proteins are encoded by different genes, which are ϳ60% homologous and are both inhibited by non-selective nonsteroidal anti-inflammatory drugs (9). In contrast to these similarities, the expression patterns of COX1 and COX2 genes are quite different. COX1 is constitutively expressed in several tissues and is thought to participate in housekeeping functions, including the maintenance of gastric epithelial integrity. COX2 expression can be dramatically induced in inflammatory cells by cytokines as well as by mitogenic factors (10). COX2 is expressed in several tissues, including those of the kidney, lung, and brain (11)(12)(13)(14). In the kidney, COX2 is expressed in the cortical thick ascending limb and MICs, whereas COX1 mRNA appears to be mainly expressed in the collecting duct (12,15). Recent studies have documented tonicity-dependent regulation of COX2, but not COX1, expression in MICs (16) in vivo and in vitro (17). These studies suggest that COX2, a major target for nonsteroidal anti-inflammatory drug action, might serve an osmoprotective function in MICs, a site of early injury in analgesic nephropathy (18). This study explored the possible relationship between COX2 activity and organic osmolyte-dependent adaptation of medullary interstitial cells to hyperosmotic stress.

Cell Culture and Animals
Mouse medullary interstitial cells were prepared as described previously (19). C57BL mice, obtained from The Jackson Laboratory (Bar Harbor, ME), were sacrificed, both kidneys were rapidly removed under sterile conditions, and the inner medullary regions were excised, minced, and suspended in Dulbecco's minimum essential medium (DMEM, Invitrogen) containing 10% fetal calf serum. The suspension was then injected intracutaneously under sterile conditions at two or three different locations of the ventral abdominal wall of an isogeneic recipient mouse. After 3-4 days the recipient animal was sacrificed, and the firm yellow nodules were removed aseptically. The nodules were minced under sterile conditions, and cells were then trypsinized in 0.05% trypsin-EDTA at 37°C for 15 min, washed, and pelleted at 300 ϫ g. The pellet was resuspended in DMEM containing 10% fetal calf serum and 50 g/ml each penicillin and streptomycin (Sigma). Cells were plated in 75-ml flasks and grown until confluent. Mouse medullary interstitial cells (MMICs) were studied after three or four passages and for experiments moved to DMEM containing 2% fetal calf serum. MMIC showed characteristic features of inner medulla fibroblasts, including cytoplasmic Oil Red O staining (20,21). COX2-deficient mouse primary medullary interstitial cell cultures were generated from COX2Ϫ/Ϫ animals as described above. COX2Ϫ/Ϫ mice were genotyped (C57BL6/129) as described previously and are an established colony in our laboratory (22).

Experimental Protocols
In Vitro-The accumulation of organic osmolytes in cultured MMICs was quantified using an isocratic HPLC system as described previously (23). Briefly, cells were grown to confluency and then adapted to hyperosmolar conditions in 100 mosmol/liter increments (exposure to each increment for 24 h, final osmolarity of 600 mosmol/liter, produced with NaCl in DMEM and 2% fetal calf serum). MMICs were kept at 600 mosmol/liter for 24 h before harvest. In some experiments SC58236 (16), a COX2-specific inhibitor (Monsanto Searle, St. Louis, MO), was added to all incubation media at a concentration of 10 mol/liter. At the end of the incubation period cells were harvested, washed with hyperosmotic medium (600 mosmol/liter), and pelleted using 5 min of centrifugation at 200 ϫ g. Osmolytes were quantitated in the neutralized perchloric acid extracts of the pellets as described previously (24).
In Vivo-To examine the effect of COX2 inhibition on mouse renal medullary organic osmolyte content, C57BL mice were treated with (n ϭ 6) or without (n ϭ 6) the COX2-specific inhibitor SC58236. COX2 inhibitor was diluted to a final concentration of 0.6 mg/liter in H 2 O with 0.01% Tween and 0.2% polyethylene glycol 200 as solvents and given to mice via drinking water 3-4 days before a 24-h dehydration (25). Inner medullary osmolyte concentrations were quantified in kidneys of C57BL WT (ϩSC58236) and COX2Ϫ/Ϫ mice after a 24-h dehydration. Neutralized perchloric acid extracts from renal inner medullary tissue sections were prepared as described previously in detail (26), and cellular organic osmolyte concentrations were quantified by an isocratic HPLC system. Intracellular osmolyte concentrations in both in vivo and in vitro studies were normalized to tissue protein concentrations in each of the samples. Protein concentrations were determined using the method described by Bradford (27).

myo-[ 3 H]Inositol Uptake Study
Mouse medullary interstitial cells were grown to confluency in 96well plates (Costar 3598, Corning Inc., Corning, NY), stepwise adapted to a final medium osmolarity of 600 mosmol/liter using NaCl, and then incubated for 60 min with different concentrations of unlabeled and 3 H-labeled inositol (1000-125 mol/liter unlabeled inositol mixed with 1 nmol of [ 3 H]inositol). [ 3 H]Inositol was purchased from ICN, Boston, MA, with a specific activity of 10 -20 Ci/mmol/liter. Isotonic controls were kept at 295 mosmol/liter and then incubated for 60 min with identical inositol solutions as in hypertonic experiments. Selected hyperosmotic and isotonic groups were incubated with 10 mol/liter SC58236 for 120 min before inositol incubation. Cell count and viability studies were performed before and immediately after SC58236 incubation. At the end of the inositol incubation, cells were washed twice with ice-cold isotonic and hypertonic DMEM, respectively, and lysed in 1% SDS. [ 3 H]Inositol counts were measured using a Beckman scintillation counter.

RNA Extraction and Generation of BGT1 and Aldose Reductase Probes for Northern Blot
Total RNA was extracted from mouse renal inner medulla using the guanidinium thiocyanate extraction method (28). RNA yield and purity were estimated spectrophotometrically by determining absorbance at 260 and 280 nm. The integrity of the RNA was checked on a nondenaturing agarose gel. Mouse medullary RNA was reverse transcribed using the Ready To Go T-primed first-strand kit (Amersham Biosciences). The betaine transporter (BGT1) cDNA probe was generated using reverse transcriptase-PCR as described previously (24). A 2-l aliquot of the reverse transcriptase reaction was used as a template for PCR reaction to obtain a betaine-␥-aminobutyric acid transporter probe by PCR (94°C denaturation, 60°C annealing, 72°C extension, 35 cycles) using specific primers (sense, 5Ј-CGG GGC GGA TCG TTT CTA T-3Ј; antisense, 5Ј-GTG GCC CTG ACC TAA GTT CCT G-3Ј, NCI GenBank TM accession number ACU26342) as described in Ref. 20. Probe analysis on 2% agarose gel showed a product length of 471 bp. The aldose reductase probe was generated using reverse transcriptase-PCR (94°C denaturation, 53°C annealing, 70°C extension, 35 cycles) using specific primers (sense, 5Ј-GCA TGG TGA AAG GGG CCT GCC-3Ј; antisense, 5Ј-GCT GGT GTC ACA GAC TTG G-3Ј, NCI GenBank TM accession number NM-012498) After PCR, an aliquot was separated on a 2% agarose gel showing a product length of 545 bp. The cDNA probe for sodium myo-inositol transporter (SMIT) was a generous gift from Dr. Moo Kwon, Johns Hopkins University, Baltimore, MD. A glyceraldehyde-3-phosphate dehydrogenase probe was amplified by PCR as described previously (20). All probes were 32 P-labeled using the Prime-It RII random primer labeling kit (Stratagene, La Jolla, CA). Northern blot assays were carried out as described previously (24).

Cell Viability
To determine the COX2 inhibitor-mediated effect on cell viability, MMICs were grown to confluency in 96-well plates and either adapted to 600 mosmol/liter as described under ''Experimental Protocols'' or kept under isotonic conditions (295 mosmol/liter) with and without 10 mol/liter SC58236 in the incubation medium. In further experiments we evaluated COX2-mediated cell death in cultured MMICs under hypertonic stress with and without preincubation with individual osmolytes. MMICs were grown either under increasing medium osmolarities up to a final osmolarity of 600 mosmol/liter with and without 20 mol/liter SC58236 or preincubated for 24 h at mildly elevated hyperosmolarity (430 mosmol/liter) with medium containing 1.0 mmol/liter of individual osmolytes (inositol, sorbitol, or betaine), before the addition of COX2 inhibitor and raising of the osmolarity of the medium to 600 mosmol/liter. Cell viability was determined using MTT assay as described previously (29). Briefly, 40 l of MTT solution (1 mg/ml) were added to each well, and plates were incubated at 37°C for 4 h. At the end of the incubation, cells were dissolved in 100 l of acid isopropyl alcohol solution under agitation, and absorbance was measured at 595 nm using a microtiter plate reader (Spectra Max 190, Molecular Devices, Sunnyvale, CA). Percent viability was calculated as (OD of drug-treated sample/OD of control sample) ϫ 100.

TUNEL Assay
For in situ detection of apoptotic cells, TUNEL assay was performed using ApopTag fluorescein in situ apoptosis detection kit (Intergen, Purchase, NY). MMICs were adapted either to increasing medium osmolarities up to a final osmolarity of 600 mosmol/liter with and without 20 mol/liter SC58236 or preincubated for 24 h with mildly elevated hyperosmolarity (430 mosmol/liter) with medium containing 1.0 mmol/liter of the individual osmolyte (inositol, sorbitol, or betaine), before the addition of COX2 inhibitor and the raising of medium osmolarity. At the end of the incubation period, cells were fixed in 2% paraformaldehyde/phosphate-buffered saline, pH 7.4, on silanized glass slides, incubated with terminal deoxynucleotidyltransferase enzyme in a humidified chamber at 37°C for 1 h, and immersed in stop/wash buffer for 10 min at room temperature. The cells were then incubated with fluorescein-conjugated anti-digoxigenin antibody for 30 min at room temperature, washed, and counterstained with 1.0 g/ml propidium iodide. Fluorescence-positive cells were counted using a Zeiss fluorescence microscope (Axioscope, Germany). The percentage of apoptotic cells was estimated as (fluorescein-positive/propidium iodidepositive) ϫ 100.

DNA Fragmentation Assay
For detection of apoptotic DNA cleavage, DNA fragmentation assay was performed using Wizard genomic DNA purification kit (Promega, Madison, WI). MMICs were either adapted to a final osmolarity of 600 mosmol/liter with and without 20 mol/liter SC58236 or preincubated for 24 h with mildly elevated hyperosmolarity (430 mosmol/liter) with medium containing 0.5 or 1.0 mmol/liter of an osmolyte mixture (inositol, sorbitol, and betaine together), before the addition of COX2 inhibitor and raising of the osmolarity of the medium to 600 mosmol/liter. DNA fragmentation was visualized by electrophoresis on a 1% agarose gel containing ethidium bromide.

Determination of Caspase-3 Activity
Cells were prepared in a fashion identical to the TUNEL protocol described under "TUNEL Assay." Caspase-3 activity was detected using the Apo-ONE TM homogenous caspase-3 assay kit (Promega, Madison, WI). At the end of the experiment incubation period, 100 l of homogenous caspase-3 reagent was added to each well, and plates were shaken at 300 rpm for 30 s. Plates were incubated at room temperature for 60 min, and then fluorescence was measured at an excitation wavelength of 480 nm and an emission wavelength of 530 nm. Negative controls worked appropriately.

Statistical Analysis
Results are presented as mean Ϯ S.D. of 4 -6 replicates. Differences were considered statistically significant if p Ͻ 0.05, using analysis of variance and Bonferroni t tests.
Effect of COX2-specific Inhibitor on Osmolyte Gene mRNA Expression-To explore the underlying mechanisms of the above observations, we investigated the effect of SC58236 on betaine (BGT1) and SMIT transporter mRNA levels in MMICs. Messenger RNA levels for both the betaine transporter BGT1 and the inositol transporter SMIT were diminished by 44 Ϯ 11 and 23 Ϯ 7%, respectively, after 48 h at 600 mosmol/liter with   COX2 inhibitor (Fig. 4). Furthermore, cultured MMICs from COX2Ϫ/Ϫ mice also exhibited dramatically reduced medullary sorbitol accumulation and aldose reductase mRNA levels (27 Ϯ 3% at 400 and 63 Ϯ 3% at 500 mosmol/liter medium tonicity, respectively, p Ͻ 0.05, see Fig. 5). Interestingly, under isotonic conditions cultured COX2Ϫ/Ϫ MMICs showed similar aldose reductase messenger RNA levels as compared with MMICs from WT mice.

FIG. 2. Intracellular accumulation of organic osmolytes inositol, sorbitol, and betaine in mouse medullary interstitial cells with and without COX2-specific inhibitor
To exclude the possibility that SC58236 directly blocked osmotic osmolyte transporters, we investigated the effect of SC58236 on [ 3 H]inositol uptake in hyperosmotically treated MMICs, compared with isotonic controls. Hypertonicity (600 mosmol/liter) produced a 4-fold increase in [ 3 H]inositol uptake as compared with isotonic controls (7778 Ϯ 1400 versus 1288 Ϯ 558 cpm/100 g of protein, respectively) that was independent of SC58236 treatment (10 M, 2 h) (data not shown). Furthermore, preincubation with SC58236 for 2 h did not significantly affect cell count or viability under isotonic or hypertonic conditions.
Medullary Interstitial Cell Survival under Hypertonic Stress Is Osmolyte-dependent-To evaluate the role of organic osmolytes in determining the increased MMIC fragility after hyperosmotic stress, we facilitated cellular osmolyte uptake by adding betaine, inositol, and sorbitol to the extracellular medium. When adapted to 600 mosmol/liter culture media by stepwise increases in media tonicity, MMICs tolerated hyperosmotic conditions well without a significant decrease in viability compared with isotonic controls (Fig. 6). Even in these gradually adapted MMICs, COX2 inhibition (SC58236, 10 mol/liter) significantly reduced cell viability (30 Ϯ 14%, p Ͻ 0.05). Importantly, preincubation with the organic osmolytes betaine and sorbitol dramatically increased MMIC survival in COX2-inhibited cells (65 Ϯ 12% improvement in betainetreated and 27 Ϯ 6% in sorbitol-treated MMICs, p Ͻ 0.05) compared with SC-treated cells without osmolytes. Inositol did not significantly increase cell viability in SC58236-treated MMICs (Fig. 6).
As shown in Fig. 7B, caspase-3 activity is enhanced in an osmolarity-dependent fashion in cultured MMICs. Caspase-3 activity increased 4-fold when medium tonicity was increased to 600 mosmol/liter, and preincubation with both betaine and sorbitol (1 mM/liter) significantly reduced caspase activity (by 41 Ϯ 10 and 59 Ϯ 12%, respectively, p Ͻ 0.05) in MMICs cultured at 600 mosmol/liter. This protective effect was maintained in MMICs treated with the COX2-specific inhibitor (SC58236, 10 M) and subjected to hypertonic stress as shown in Fig. 8, A and B. Fig. 8A shows nuclear fluorescein positivity in MMICs cultured at either isotonic (295 mosmol/liter) or hypertonic (500 mosmol/liter) medium conditions. Treatment with SC58236 (10 M) without osmolytes significantly increased the number of apoptotic cells (TDT) and decreased overall cell numbers (PI). This observation was dramatically ameliorated by preincubation with a mixture of organic osmolytes (inositol, sorbitol, and betaine, 1 mmol/liter each). Fig.  8B depicts apoptotic DNA laddering under hyperosmotic stress (lane B, 600 mosmol/liter), which is enhanced by the addition of SC58236 and reversed by increasing concentrations of organic osmolytes. (30) found that the COX2-specific inhibitor SC58236 reduces COX2 activity and COX2 products in renal medullary interstitial cells. The present study shows that inhibition of COX2 activity is associated with decreased hypertonicity-induced osmolyte accumulation in MMICs (Fig. 2). COX2 inhibition most dramatically inhibited betaine accumulation (96 Ϯ 10%), whereas inositol and sorbitol accumulation were less affected by treatment with SC58236. Decreased accumulation of intracellular osmolyte concentrations was also associated with diminished mRNA levels of genes coding for osmolyte transporters and aldose reductase (Figs. 5 and 6). The reductions in betaine and myoinositol transporter mRNA levels seen with the COX2 inhibitor were proportional to the reduction of cellular concentrations of these compounds. In fact, BGT1 showed a nearly 2-fold greater reduction in mRNA levels compared with SMIT (Fig. 4). SC58236 did not have any direct effect on [ 3 H]inositol uptake in MMIC, suggesting that COX2 activity exerts its effect on osmolyte transporter genes possibly at the transcriptional level, rather than on the cytoplasmic membrane osmolyte transporters. The mechanism of the particular sensitivity of BGT1 to COX2 inhibition might therefore be because of transcriptional regulation, but this remains to be elucidated.

Effect of Organic Osmolytes on COX2 Inhibitor-mediated Decrease in MMIC Viability-Previous studies
In vivo the effect of COX2 inhibition either with SC58236 or COX2 gene disruption was associated with a more pronounced impairment of osmolyte accumulation than in cultured MMICs (Fig. 3). Nevertheless, as seen in cultured interstitial cells, medullary concentrations of betaine and inositol also showed the greatest sensitivity to COX2 inhibition, with inositol levels in COX2 inhibitor-treated or COX2Ϫ/Ϫ dehydrated mice being below even those of control animals (Fig. 3). Because aldose reductase mRNA levels diminish after increased tonicity in cultured COX2Ϫ/Ϫ MMICs, the existence of a down-regulatory factor that becomes fully active in the absence of COX2 could possibly explain these observations. Elucidation of these regulatory mechanisms are under current investigation.
When MMICs were slowly adapted to medium osmolarities of 600 mosmol/liter (100 mosmol/liter increments/24 h), cell survival was equal to isotonic controls (Fig. 6). However, MMICs were quite sensitive to abrupt increases in medium tonicity of more than 200 mosmol/liter/24 h, which were poorly tolerated (data not shown). This time-dependent adaptation is consistent with insufficient time to transcribe and translate genes encoding for osmolyte transporters and enzymes. We therefore tested the effect of COX2 inhibitors on cell survival only in adapted MMICs. SC58236 reduced MMIC survival by 30% at 600 mosmol/liter, even when adequate time to adapt was provided. Preincubation with 1 mmol/liter betaine or sorbitol rescued MMICs and returned survival rates back to normal levels. Betaine showed the greatest potential to prevent MMIC death, similar to its protective capacity in Madin-Darby canine kidney cells exposed to hypertonic stress (31).
Apoptosis was a major mechanism contributing to cell demise in MMICs exposed to hyperosmotic stress, as shown by TUNEL stain and DNA laddering (Fig. 8, A and B). In contrast, provision of exogenous organic osmolytes largely prevented apoptosis in COX2 inhibitor-treated MMICs after hypertonic stress. Similarly, in cultured cells tonicity-dependent activation of caspase-3 activity was reduced by betaine and sorbitol but not inositol (Fig. 7B). These findings suggest that the inability to adequately accumulate organic osmolytes significantly contributes to COX2 inhibitor-mediated action to decrease MMIC survival after hypertonic stress. Furthermore, the results of our studies argue against a direct toxic effect induced by the COX2-specific inhibitor per se.
COX2-dependent Osmolyte Accumulation in MMIC and Mouse Inner Medulla-The accumulation of organic osmolytes in response to exposure of hypertonic medium has been described previously for rat kidney papillary interstitial cells (20). Similar to this study, our results in mouse medullary interstitial cells show that betaine and inositol are the predominant osmolytes accumulated at 600 mosmol/liter medium tonicity, with a lower increase in sorbitol. Burger-Kentischer et al. (20) also detected low levels of glycerophosphorylcholine accumulation, an observation that we could not confirm. This might be because of lesser amounts of cell protein in our experiments, leaving glycerophosphorylcholine concentrations below detection levels or a particular species difference. Interestingly, the present studies showed a different pattern of osmolyte accumulation in the in vivo setting in which sorbitol accumulation predominated as compared with the in vitro setting in which betaine and inositol predominated. Because inner medullary whole tissue sections contain many additional cell types (i.e. collecting duct cells), this discrepancy could be because of a larger relative increase in sorbitol accumulation in papillary-collecting duct cells as compared with interstitial cells, as shown previously (20).
In conclusion, the present studies demonstrate that accumulation of organic osmolytes in kidney medullary cells exposed to hypertonic stress is dependent on COX2 activity. The diminished ability of MMICs to tolerate hypertonic stress after COX2 inhibition can be reversed by providing exogenous organic osmolytes to these cells. To our knowledge this is the first study to link organic osmolyte accumulation to COX2 activity, which provides a possible molecular mechanism contributing to medullary interstitial cell injury in patients with analgesic nephropathy.