Biomechanical regulation of cyclooxygenase-2 in the renal collecting duct

Abstract

High-dietary sodium (Na), a feature of the Western diet, requires the kidney to excrete ample Na to maintain homeostasis and prevent hypertension. High urinary flow rate, presumably, leads to an increase in fluid shear stress (FSS) and FSS-mediated release of prostaglandin E₂ (PGE₂) by the cortical collecting duct (CCD) that enhances renal Na excretion. The pathways by which tubular flow biomechanically regulates PGE₂ release and cyclooxygenase-2 (COX-2) expression are limited. We hypothesized that FSS, through stimulation of neutral-sphingomyelinase (N-SM) activity, enhances COX-2 expression to boost Na excretion. To test this, inner medullary CD3 cells were exposed to FSS in vitro and mice were injected with isotonic saline in vivo to induce high tubular flow. In vitro, FSS induced N-SM activity and COX-2 protein expression in cells while inhibition of N-SM activity repressed FSS-induced COX-2 protein abundance. Moreover, the murine CCD expresses N-SM protein and, when mice are injected with isotonic saline to induce high tubular flow, renal immunodetectable COX-2 is induced. Urinary PGE₂ (445 ± 91 vs. 205 ± 14 pg/ml; P < 0.05) and microdissected CCDs (135.8 ± 21.7 vs. 65.8 ± 11.0 pg·ml⁻¹·mm⁻¹ CCD; P < 0.05) from saline-injected mice generate more PGE₂ than sham-injected controls, respectively. Incubation of CCDs with arachidonic acid and subsequent measurement of secreted PGE₂ are a reflection of the PGE₂ generating potential of the epithelia. CCDs isolated from polyuric mice doubled their PGE₂ generating potential and this was due to induction of COX-2 activity/protein. Thus, high tubular flow and FSS induce COX-2 protein/activity to enhance PGE₂ release and, presumably, effectuate Na excretion.

the western diet, high in sodium (Na) content, has been implicated in the epidemic of hypertension in the United States. Maintenance of renal Na homeostasis is critical to prevent systemic hypertension; and therefore, renal Na excretion, in the face of dietary Na challenges, is critical. Several normal physiologic mechanisms function to augment Na excretion during dietary Na loading including suppression of the renin-angiotensin-aldosterone axis, pressure natriuresis, and release of atrial natriuretic peptide. In addition to hormonal regulation, autocrine and paracrine induction of prostanoids (8), nucleotides (28), and endothelin-1 (16) in the collecting duct (CD) inhibits Na reabsorption in the face of high-dietary Na. The redundancy of paracrine inhibitors of epithelial Na channel-mediated Na transport secreted by the CD suggests they work in concert to augment Na excretion. In fact, in endothelin-1 knockout mice, urinary prostaglandin E₂ (PGE₂) excretion is increased under normal- and high-dietary Na conditions (11) to stimulate Na excretion. Thus, these CD-derived molecules play an important role in renal Na homeostasis.

Excretion of urinary prostanoids, specifically PGE₂, is increased in Na-fed mice, leading to a naturiesis. Some investigators have demonstrated that increases in medullary hypertonicity induce renal tubular cyclooxygenase (COX)-2 mRNA and protein and urinary PGE₂ excretion (32–35). On the other hand, data from our lab suggest that increases in urine flow rate and fluid shear stress (FSS) stimulate PGE₂ release and COX-2 mRNA expression in CD epithelia (8). We hypothesize that tubular flow rate and FSS stimulate a coordinated response to enhance PGE₂ release: 1) acutely, to activate cPLA2 and release arachidonic acid (AA), the substrate of COX-1/2, and 2) chronically, to induce COX-2 gene and protein expression so as to augment the capacity of cyclooxygenases to metabolize AA into intermediate endoperoxides and, finally, into PGE₂. These acute and chronic processes, we believe, are coordinated to ensure appropriate amounts of paracrine PGE₂ are synthesized to excrete Na and maintain Na homeostasis. In this study, we sought to extend our initial findings on the effect of tubular flow and FSS on PGE₂ expression by evaluating the in vivo and in vitro effects of tubular flow and FSS, respectively, on the regulation of COX-2 protein expression through a sphingomyelinase-dependent mechanism. Sphingomyelinases are enzymes that convert sphingomyelin to ceramide and this compound, in turn, acts as an intracellular signaling molecule. Sphingomyelinases are principally composed of two types: 1) acidic and 2) neutral sphingomyelinase (N-SM). N-SM enzymatic activity has been reported by others to be stimulated by FSS in endothelial cells and to induce COX-2 protein expression (4, 5, 27). Acid and N-SM activity has been reported in whole kidney, renal cortex, and, more specifically, proximal tubule (12, 37). N-SM is also observed in the mechanosensitive mesangial cell (9); however, the expression of N-SM in the CD has not been described.

MATERIALS AND METHODS

Cell culture.

Murine immortalized inner medullary CD3 (IMCD3) cells were grown in DMEM/F12 (with 10% FBS) on 25 × 75-mm slides and studied when they reached confluence between 3 to 7 days. We only used cells up to passage 10 due to the risk of genetic drift.

Induction of FSS.

Cells grown on slides were placed in laminar flow chambers (Glycotech manufactured chamber), maintained at 37°C, and subject to shear of 0.4 dyn/cm² using phenol red-free, serum-free DMEM/F12 containing penicillin/streptomycin for varying durations. FSS was calculated based on Poiseulle's law; τ = μγ = 6μQ/a²b where τ = wall stress (dyn/cm²), γ = shear rate (per s), μ = apparent viscosity of the fluid (media at 37°C = 0.76 cP), a = channel height (cm), b = channel width (cm), and Q = volumetric rate (ml/s). The DMEM/F12 was not recirculated into the perfusion chamber. Static control cells were exposed to the same solution and duration as sheared cells, but without exposure to FSS. One milliliter of serum- and phenol red-free DMEM/F12 was incubated with either static or sheared cells for 1 h for measurement of PGE₂ secretion (8, 17). Cells from the Glycotech chamber were then collected for total protein or for measurement of sphingomyelinase activity.

Sphingomyelinase activity.

IMCD3 cells were processed and assayed according to the directions specified in the Sphingomyelinase Fluorometric Assay Kit (Cayman Chemical). In short, static or sheared cells were collected and washed in PBS; cells were then gently lysed in SMase buffer with a glass dounce and centrifuged. The supernatant and pellet were individually collected and frozen at −80°C for analysis. A standard curve was generated using phosphocholine (0 to 50 μM), a positive control well containing sphingomyelinase, and background well containing SMase buffer solution alone. The samples derived from cells were then plated into wells. Developer (100 μl) was pipetted into each well and 20 μl of sphingomyelin substrate were added (except not into the background wells). The plate was incubated for 30 min at 37°C and then fluorescence was measured in a SpectraMax M2 fluorescence plate reader (535-nm excitation and 590-nm emission; Molecular Devices).

Western blotting.

Western blot analysis was performed as previously described (7). Protein lysates were generated from kidney or IMCD3 cells by incubating and homogenizing tissue in lysis buffer (10 mM Tris, pH 7.2, 1 mM EGTA, 1 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1.0% Triton X-100) and protease inhibitors (1 mM PMSF, 10 μg/ml leupeptin, 5 μM pepstatin A, 1 mM benzamidine, 30 mM sodium fluoride, 2 mM sodium orthovanadate, and 1 μg/ml aparotinin) on ice. The supernatant was collected and assayed for protein content using the BCA protein assay (Thermoscientific). Thirty to one hundred micrograms (depending on the abundance of the signal) of protein lysate were resolved electrophoretically and transferred to Immobilon filters (Millipore, Billerica, MA). Filters were blocked in 5% nonfat dried milk and 0.05% Tween and immunoblotted with a primary antibody (see Reagents). After being washed, blots were incubated with a horseradish peroxidase-conjugated secondary antibody (Sigma, St. Louis, MO) and bands were visualized by the West Pico enhanced chemiluminescence kit (Pierce, Rockford, IL). After the membrane was stripped and blocked, the blot was incubated with an anti-actin or anti-GAPDH-specific antibody and visualized using the same methods as the primary antibody.

Reagents.

Inhibitors were as follows: 10 μM CAY10404 (Cayman Chemical), 20 μM GW4869 (Cayman Chemical), and 100 μM AA (Cayman Chemical). Antibodies were as follows: rabbit anti-COX-2 (1:1,000; Cayman Chemical), rabbit anti-COX-1 (1:1,000; Cayman Chemical), rabbit anti-cPLA2 (1:500; Cell Signaling), rabbit anti-N-SM (1:500; for Western) or (1:100; for immunofluorescence; Abcam), mouse anti-actin (1:1,000; Cell Signaling), mouse anti-GAPDH (Santa Cruz Biotechnology) antibody and goat anti-rabbit conjugated to horseradish peroxidase or goat anti-mouse conjugated to horseradish peroxidase (1:5,000; Sigma).

Animals.

Six- to eight-week-old FVB male mice were obtained from Charles River Labs (Wilmington, MA) and housed at the Icahn School of Medicine Center for Comparative Medicine and Surgery. To induce a high urine flow rate, mice were subcutaneously injected with isotonic saline (volume = 20% of body wt) while the control mouse was sham injected (no fluid given). The amount of saline to safely induce a natriuretic/polyuric response was based on studies by Galla and Luke (10) who injected up to 20% of saline into rodents. The injected, volume-expanded (VE), and sham-injected, control mice were then placed in separate metabolic cages for 6 h to collect urine. At the end of the collection period, kidneys were either extracted to generate protein lysate from cortex and medulla for Western blotting or cortical CDs (CCDs) were microdissected by hand (6) to measure PGE₂ release. All protocols were approved by the Institutional Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai. Animals were euthanized in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Immunofluorescence.

Murine kidneys were snap-frozen in isopentane precooled with liquid nitrogen for immunohistochemistry. Kidneys were sectioned and preserved at −80°C until kidneys were to be stained. Tissue sections were washed in PBS and permeabilized with 0.3% Triton X-100 in PBS. Tissue auto-fluorescence was reduced with a solution of 0.1% glycine and 1% BSA in PBS and blocked with a solution of 1% BSA, 10% FBS in PBS. Next, the N-SM primary antibody, dissolved in a 0.1% Triton X-100, 0.1% BSA, 1% FBS solution of PBS, was incubated overnight at 4°C. The tissue was washed, incubated with Alexa 488-labeled goat anti-rabbit (1:500; Molecular Probes) for 1 h, and then washed. Rhodamine-labeled dolichos biflorus agglutinin (1:200; Vector Laboratories) was placed on the tissue for 15 min and washed. Imaging was performed on a confocal Leica TCS SP5 DM microscope.

PGE₂ secretion by microdissected CCDs and COX-2-specific cyclooxygenase activity.

Approximately 1 to 3 mm of CCDs (1–4 tubules/sample) were microdissected by hand (6) in cold Ringer lactate (RL) from sham-injected controls and saline-injected VE mice and transferred to 60 μl of RL and incubated at 37°C for 30 min. The microfuge tube was centrifuged for 5 min to pellet the CCD, and the supernatant was collected and stored at −80°C for measurement of PGE₂ (13). The tubule was also collected and frozen at −80°C to measure PGE₂. To compute total or COX-2-specific PGE₂ synthesizing activity in individual CCDs, the CCDs were affixed to poly-l-lysine-coated 0.5 × 0.2-cm coverslips and the coverslips were transferred between different solutions for incubation at 37°C. First, the CCDs were incubated in RL containing 100 μM AA (the substrate for COXs) for 30 min at 37°C, and PGE₂ was measured in the supernatant (13, 29). The CCD was then preincubated in a RL solution containing 10 μM CAY10404, a COX-2 inhibitor, for up to 30 min at 37°C. Next, the tubule was transferred to a RL solution with 100 μM AA and 10 μM CAY10404 for 30 min, to evaluate COX-2-specific activity (13, 29).

PGE₂ was measured in supernatants and CCDs using the PGE₂ enzyme immunoassay (EIA) kit (Cayman Chemical) (8). Supernatants were rapidly frozen to −80°C and kept at this temperature before analysis. CCDs were rapidly cooled at −80°C, to lyse and freeze the tubule, and PGE₂ was measured utilizing the PGE₂ EIA kit.

Statistics.

Data are given as means ± SE (n = number of slides or CCDs). Statistical analyses were performed using unpaired t-tests (Sigmaplot version 11.0) for cell culture experiments and between treated and untreated CCDs. Paired t-tests were used for Western blotting experiments.

RESULTS

FSS, in vitro, and high tubular flow rates, in vivo, induce PGE₂ synthetic enzymes.

High-dietary NaCl (8%) ingestion in rodents induces COX-2 mRNA expression in the renal medulla (3, 34), which we speculate may, in part, be related to high urine and tubular flow rates. Our laboratory demonstrated that a physiologic level (0.4 dyn/cm²) (14) of FSS for 2 h induces COX-2 mRNA and tended to raise COX-2 protein expression in IMCD3 cells, but this did not reach statistical significance (8). This response to FSS, we believe, was due to the fact that protein synthesis requires more time before a change in protein expression can be detected by immunoblotting. To test this, IMCD3 cells were exposed for 4 h to 0.4 dyn/cm² of FSS or no FSS, and immunoblotting was performed on protein lysates. Immunodetectable COX-2 protein was increased in cells subject to 4 h of FSS compared with static cells (Fig. 1), suggesting FSS induces immunodetectable COX-2 protein expression.

Fig. 1.

Fluid shear stress (FSS) induces cyclooxygenase (COX)-2 protein expression, but inhibition of neutral-sphingomyelinase (N-SM) activity by GW4869 suppresses this effect. COX-2 protein expression was measured under 3 conditions in inner medullary collecting duct (IMCD)3 cells: 1) static, 2) FSS (0.4 dyn/cm²), and 3) FSS plus treatment with GW4869 (20 μM). FSS induced COX-2 protein abundance compared with static cells while GW4869 suppressed the effect of FSS on COX-2 expression. The number in parentheses above each lane represents the densitometric ratio of COX-2 to actin.

To test whether high urine flow rates in vivo activate the intrarenal PGE₂ synthetic system, specifically COX-2, mice were injected subcutaneously with isotonic saline at 20% of body weight to augment the urine flow rate and mimic in vitro FSS experiments. This maneuver suppresses serum aldosterone concentration in the absence of effects on serum vasopressin (31) and enhances urine flow rate. Mice were placed in a metabolic cage for 6 h to collect urine, killed, and then kidneys were extracted for analysis. The 6-h time point was chosen to ensure that kidneys experienced high tubular flow for at least 4 h because our in vitro data suggested this length of time was necessary to observe immunodetectable COX-2 expression. The urine volume collected from saline-injected mice (2.9 ± 0.2 ml; n = 12) was approximately sixfold greater than sham-injected control mice (0.5 ± 0.1 ml; n = 12, P < 0.05). COX-1, COX-2, and cPLA2 protein abundance were measured in cortex and medulla of saline-injected, VE, and sham-injected, control mice (Fig. 2A). In paired sets of VE and sham-injected control mice (n = 7), steady-state COX-2 protein abundance was greater by ∼70% in the cortex of kidney from VE mice than from sham-injected mice (Fig. 2B; *P < 0.05). Similarly, COX-2 protein expression in medullary kidney from VE mice was ∼60% greater than from uninjected control mice (Fig. 2C; *P < 0.05). However, cortical and medullary expression of COX-1 protein (Fig. 3A) was similar in VE and control mouse kidneys (n = 3; Fig. 3, B and C, respectively). The absence of significant change in COX-1 protein expression in response to VE is consistent with our previous finding that COX-1 mRNA is unchanged in FSS-exposed IMCD3 cells (8). cPLA2, a key enzyme that releases AA from cell membranes to form the substrate for COXs, is also significantly induced in both cortex (Fig. 4, A and B) and medulla (Fig. 4, C and D) of kidney from VE mice compared with uninjected control mice, suggesting intrarenal activation of the PGE₂ synthetic system by high tubular flow.

Fig. 2.

Volume expansion and subsequent diuresis induce renal COX-2 protein abundance. Six- to eight-week-old mice were either sham-injected or injected subcutaneously with isotonic saline at 20% of body weight. The mice were placed into a metabolic cage for 6 h and urine was collected. Kidneys were then extracted and protein lysate was generated from the cortex and medulla. A: single Western blot of renal cortex and medulla of sham- and saline-injected mice demonstrates an increase in COX-2 protein abundance. The open bar identifies that the image was spliced together from a single immunoblot. Densitometric analysis of immunoblots comparing the COX-2 expression in renal cortex (B) and medulla (C) in sham (n = 7)- vs. saline-injected (n = 7) mice demonstrates an increase in COX-2 protein abundance by >60% in kidneys of the latter (*P < 0.05).

Fig. 3.

Volume expansion and subsequent diuresis do not affect expression of renal COX-1 protein abundance. In identical experiments that are described in Fig. 2, COX-1 protein abundance was evaluated in the renal cortex and medulla of sham- and saline-injected mice. A: in a single Western blot, COX-1 protein abundance did not differ between renal cortex and medulla of sham- and saline-injected mice. The open bar identifies that the image was spliced together from a single immunoblot. B and C: further densitometic analysis of immunoblots (n = 3) confirmed no difference in COX-1 abundance between kidneys from control or volume-expanded mice in cortex (B) or medulla (C).

Fig. 4.

Volume expansion and subsequent diuresis induce renal cPLA2 protein abundance. In experiments that are identical to those described in Fig. 2, a single Western blot of renal cortex of sham- and saline-injected mice demonstrates an increase in cPLA2 protein abundance in the cortex of saline-injected mice (A). B: densitometric analysis of immunoblots evaluating the cPLA2 expression in renal cortex of sham (n = 4)- vs. saline-injected (n = 4) mice demonstrates an increase in cPLA2 protein abundance (*P < 0.05). C: cPLA2 protein abundance also increased in the renal medulla of saline-injected mice vs. sham-injected mice. D: densitometric analysis of immunoblots showed that steady-state cPLA2 abundance increased in renal medulla of saline-injected (n = 6) compared with sham-injected (n = 6; *P < 0.05) mice.

N-SM is expressed in CD and is a FSS-sensitive regulator of COX-2 protein abundance.

FSS activates N-SM activity in endothelial cells to release ceramide that stimulates COX-2 protein abundance (4, 5). We propose that a similar mechanism functions in CD cells to induce COX-2 protein abundance and augment the potential to generate PGE₂. First, we verified that N-SM protein is expressed in IMCD3 cells and murine kidney by immunoblotting and immunofluorescence studies. N-SM was abundantly expressed in IMCD3 cells (Fig. 5) and in cortex and medulla of mouse kidney (Fig. 5). In addition, N-SM was localized to the CCD (Fig. 6A) in vivo. To identify principal cells (PCs), and hence the CCD, murine kidney sections were incubated with rhodamine-labeled dolichos biflorus agglutinin (red) and to label N-SM, a rabbit anti-N-SM antibody was incubated with the tissue (green). PCs (white arrows) and intercalated cells (ICs; yellow arrows) both express N-SM; however, N-SM is primarily observed in the cytoplasm of ICs while PCs express N-SM in both cytoplasm and apically (Fig. 6A).

Fig. 5.

N-SM is expressed in IMCD3 and murine kidney. Western blot analysis of protein lysate from static IMCD3 cells, renal cortex, and renal medulla demonstrates expression of N-SM, an enzyme implicated in flow-mediated COX-2 protein expression in endothelial cells.

Fig. 6.

N-SM is expressed in principal cells (PCs) and intercalated cells (ICs) of renal cortical collecting duct (CCD). Renal tissue was incubated with rhodamine-labeled dolichos biflorus agglutinin to identify PCs (red) and incubated with rabbit-anti-N-SM to localize N-SM (green). A: PCs (white arrow and cells labeled red) express N-SM apically and in the cytoplasm while ICs (yellow arrow and cells not labeled red in the CCD) express N-SM cytoplasmically. B: negative control lacking the primary anti-N-SM antibody. Immunolocalization was repeated in kidneys from 3 mice.

Next, we tested whether FSS activates sphingomyelinase activity in IMCD3 cells. To this end, we first evaluated, by Western blotting, whether the supernatant (cytosolic) or pellet (noncytosolic) fraction of IMCD3 cells expressed N-SM protein. We found that N-SM protein abundance was much greater in the supernatant than in the pellet (Fig. 7A). The sphingomyelinase activity in the supernatant of FSS-exposed and static IMCD3 cells was assayed and normalized to total protein from that fraction. We chose to expose IMCD3 cells for only 2 h of FSS, rather than 4 h, because we suspect that sphingomyelinase stimulation induces COX-2 expression, so it needs to be activated before the increase in COX-2 protein abundance. The sphingomyelinase activity was ∼10-fold greater in sheared IMCD3 (32.2 ± 7.7 pmol/min·ml⁻¹·μg⁻¹ protein; *P < 0.05; Fig. 7B) than static controls (4.4 ± 1.4 pmol/min·ml⁻¹·μg⁻¹ protein). Next, we tested whether FSS induced N-SM-specific activity by exposing IMCD3 cells to FSS in the absence and presence of a specific N-SM inhibitor, GW4869 (20 μM). Approximately 50% of the shear-sensitive sphingomyelinase activity was due to N-SM (Fig. 7C).

Fig. 7.

N-SM is expressed and sphingomyelinase activity is induced by FSS in supernatants of IMCD3 cells. The supernatant and pellet of IMCD3 cells were isolated according to the procedure outlined in the sphingomyelinase assay kit. A: immunoblot of the supernatant and pellet isolated from IMCD3 cells, not exposed to FSS, demonstrates abundant expression of N-SM in the supernatant. B: to test whether flow activates sphingomyelinase activity, IMCD3 cells were exposed to 0.4 dyn/cm² for 2 h, and cellular supernatant was isolated and assayed for sphingomyelinase enzymatic activity. The total sphingomyelinase activity, normalized to protein content, was greater in sheared (32.2 ± 7.7 pmol/min·ml⁻¹·μg⁻¹; n = 3, *P < 0.05) than in static (4.4 ± 1.4 pmol/min·ml⁻¹·μg⁻¹, n = 3) cells. C: to test whether N-SM-specific activity is induced by FSS, IMCD3 cells were exposed to FSS for 2 h in the absence or presence of 20 μM GW4869, a N-SM-specific inhibitor. FSS induced total sphingomyelinase activity to 31.5 ± 5.7 pmol/min·ml⁻¹·μg⁻¹ while GW4869 repressed sphingomyelinase activity significantly to 16.2 ± 4.7 pmol/min·ml⁻¹·μg⁻¹ (*P < 0.05).

Since FSS induced COX-2 protein and N-SM activity, shear-exposed cells were treated with GW4869 (20 μM) to examine whether inhibition of FSS-induced N-SM activity repressed FSS-induced COX-2. N-SM specific inhibition suppressed COX-2 protein abundance by >50% in sheared cells compared with that observed in untreated cells (Fig. 1), suggesting that shear-stimulated N-SM activity influences COX-2 protein abundance.

Moreover, sphingomyelinase activity generates ceramides that bind kinase suppressor of Ras (KSR) to induce Raf-1 and Ras, and thus, MAPK (4, 5, 38) to stimulate COX-2 protein abundance (27). IMCD3 cells were treated with a ceramide analog (C2-ceramide, an analog of N-acetylsphingosine) for 4 h. C2-ceramide (0–100 μM) induced COX-2 protein abundance (Fig. 8A) at concentrations of 20 and 100 μM compared with untreated controls (Fig. 8B; *P < 0.05), suggesting that the end product of sphingomyelinase activity regulates immunodetectable COX-2 expression in CD cells.

Fig. 8.

Ceramides induce COX-2 protein abundance in IMCD3 cells. IMCD3 cells were incubated with increasing concentrations of C2-ceramide (0–100 μM) for 4 h and protein lysate was generated from these cells. Immunoblotting of the protein lysate demonstrated that COX-2 protein abundance (A) was greater in cells treated with 20 and 100 μM C2-ceramide (B; *P < 0.05) than untreated controls.

Incubation with AA, as a method to evaluate PGE₂ generating potential of epithelia.

IMCD3 cells were incubated with a high concentration of AA (100 μM), the substrate for COXs, for 1 h, and PGE₂ was measured, as a surrogate of total PGE₂ synthesizing potential (13, 29). IMCD3 cells were either maintained under static conditions or exposed to FSS (0.4 dyn/cm²) for 2 h and then PGE₂ was measured in media in the absence or presence of AA. PGE₂ secretion under static conditions, in the absence of AA, was very low (1.28 ± 0.37 pg·ml⁻¹·μg⁻¹ protein) but was enhanced in static cells incubated with AA (61.7 ± 9.1 pg·ml⁻¹·μg⁻¹ protein; *P < 0.05; Fig. 9), suggesting that the cells have significant reserve to boost PGE₂ synthesis. Sheared cells (34.6 ± 1.2 pg·ml⁻¹·μg⁻¹ protein) also exhibited an increase in PGE₂ secretion after incubation with AA (92.1 ± 5.5 pg·ml⁻¹·μg⁻¹ protein; Fig. 9; *P < 0.05). Sheared cells exposed to AA released the most PGE₂ ($P < 0.05 vs. static with AA and shear alone) implying that cellular PGE₂ generating potential is increased as early as 2 h after experiencing FSS. Our earlier studies suggested no change in COX-1 mRNA, an increase in COX-2 mRNA, and a nonsignificant (P = 0.064) increase in COX-2 protein. We suspect that AA treatment of cells enhances our sensitivity to measure changes in COX protein/activity and/or the overall potential of an epithelium to generate PGE₂ than our immunodetection methods.

Fig. 9.

FSS induces PGE₂ release, as well as, enhances the potential to release PGE₂ in IMCD3 cells. To determine the total PGE₂ generating potential, IMCD3 cells, exposed to static or FSS conditions, were incubated with 100 μM arachidonic acid (AA), the substrate for COXs, and PGE₂ was measured in the media bathing the cells. Under static conditions, PGE₂ secretion, normalized to cellular protein, is very low (1.28 ± 0.37 pg·ml⁻¹·μg⁻¹ protein); however, incubation with AA stimulates PGE₂ secretion ∼60-fold (61.7 ± 9.1 pg·ml⁻¹·μg⁻¹ protein; *P < 0.05 vs. static untreated control). FSS induces PGE₂ release in IMCD3 cells in the absence of AA (34.6 ± 1.2 pg·ml⁻¹·μg⁻¹ protein) compared with static control (*), but it is less than static cells treated with AA (#P < 0.05). FSS with AA led to the greatest increase in PGE₂ (92.1 ± 5.5 pg·ml⁻¹·μg⁻¹ protein), such that PGE₂ release was greater than static alone (*), static treated with AA (#), and shear alone ($P < 0.05), implying that FSS stimulates enzymes that generate PGE₂.

PGE₂ generation in microdissected CCDs.

To evaluate whether fast urinary and tubular flow rates stimulate PGE₂ synthesis in vivo, CCDs were microdissected from VE and control mice, CCDs were placed in RL for 30 min at 37°C, and PGE₂ was measured in the supernatant and tubule. The urinary PGE₂ concentration in saline injected mice (Fig. 10A; 445 ± 91 pg/ml; #P < 0.05) was approximately twofold greater than that measured in control mice (205 ± 14 pg/ml). As would also be expected, the PGE₂ excretion over the 6 h was also significantly greater in VE (1,463 ± 180 pg of PGE₂; P < 0.05) compared with sham-injected controls (112 ± 28 pg of PGE₂). The concentration of PGE₂ measured in the supernatant bathing microdissected CCDs (Fig. 10B; 135.8 ± 21.7 pg·ml⁻¹·mm⁻¹ CCD; #P < 0.05) and in the CCD itself (Fig. 10C; 75.0 ± 17.2 pg·ml⁻¹·mm⁻¹ CCD; #P < 0.05) was approximately twofold greater than that observed in the supernatant (65.8 ± 11.0 pg·ml⁻¹·mm⁻¹ CCD) and in CCDs (33.3 ± 10.9 pg·ml⁻¹·mm⁻¹ CCD) of control mice, respectively, suggesting that high urinary flow rates activate PGE₂ synthetic machinery in vivo.

Fig. 10.

CCDs exposed to fast urinary and/or tubular flow rates express more PGE₂ than control CCDs. To generate high urinary and/or tubular flow rates, mice were volume expanded with isotonic saline to induce a diuresis while control mice were sham-injected without any solution. A: PGE₂ concentration was greater in the urine of volume-expanded/diuretic mice (445 ± 91 pg/ml; #P < 0.05) than in sham-injected controls (205 ± 14 pg/ml). CCDs were subsequently microdissected from volume-expanded and control mice, incubated in Ringer lactate, and centrifuged. PGE₂ was measured in the supernatant and the lysed CCDs. B: PGE₂ secreted into the media bathing the tubules was greater in volume-expanded/diuretic (135.8 ± 21.7 pg·ml⁻¹·mm⁻¹ CCD; #P < 0.05) than from control (65.8 ± 11.0 pg·ml⁻¹·mm⁻¹ CCD) animals. C: concomitantly, the intracellular PGE₂ was also greater in the volume-expanded (75.0 ± 17.2 pg·ml⁻¹·mm⁻¹ CCD; #P < 0.05) vs. control (33.3 ± 10.9 pg·ml⁻¹·mm⁻¹ CCD) CCDs.

The total PGE₂ generating activity of CCDs was measured by incubating CCDs in RL containing 100 μM AA and then measuring PGE₂, as we did with IMCD3 cells. PGE₂ concentration in the supernatant was greater in CCDs incubated in AA than untreated CCDs (Figs. 10B vs. 11), implying that CCDs from control and VE mice have greater potential to generate PGE₂. CCDs isolated from uninjected control mice secreted less PGE₂ (328 ± 16 pg·ml⁻¹·mm⁻¹ CCD) after exposure to 100 μM AA than VE CCDs exposed to AA (607 ± 108 pg·ml⁻¹·mm⁻¹ CCD; Fig. 11; #P < 0.05), implying that the capability to synthesize PGE₂ in VE mice is approximately twofold greater than control mice. Incubating CCDs of VE mice with a COX-2-specific inhibitor (CAY10404) and AA suppressed the PGE₂ to 336 ± 34 pg·ml⁻¹·mm⁻¹ CCD, which is not different than CCDs isolated from sham-injected mice. This suggested that the AA-mediated increase in PGE₂ from CCDs isolated from VE mice was largely due to an increase in COX-2 activity/protein.

Fig. 11.

Total PGE₂ synthesizing potential is increased in CCDs isolated from volume-expanded mice primarily due to COX-2 activity. To evaluate whether total COX activity is increased in CCDs by fast tubular flow, microdissected CCDs from control and experimental mice were incubated with 100 μM AA, the substrate for COXs, and PGE₂ was measured. CCDs isolated from volume-expanded mice secreted more PGE₂ (607 ± 108 pg·ml⁻¹·mm⁻¹ CCD; #P < 0.05) in response to AA than CCDs from control mice (328 ± 16 pg·ml⁻¹·mm⁻¹ CCD), implicating that total PGE₂ generating activity is greater in CCDs of diuretic kidneys than controls. Treatment of CCDs from volume-expanded mice with CAY10404, a COX-2-specific inhibitor, in the presence of AA significantly repressed PGE₂ release (336 ± 34 pg·ml⁻¹·mm⁻¹ CCD).

DISCUSSION

Growing evidence points to the critical role that hydrodynamic forces play in transepithelial transport, signaling and morphology of the kidney (6–8, 18–20, 26). This study seeks to contribute to this literature by showing that high tubular flow rates in vivo, as follows isotonic volume expansion, induce CD expression of COX activity and enhance PGE₂ release which we propose facilitates Na excretion. In vitro studies of an IMCD3 cell culture model revealed that FSS-induced COX-2 protein expression is regulated by N-SM activity, a mechanism known to regulate flow-mediated COX-2 expression in endothelia (4, 5, 27).

We demonstrate that high tubular flow in vivo produced by isotonic extracellular volume expansion 1) enhances PGE₂ secretion into urine, 2) induces PGE₂ secretion by isolated CCDs, 3) stimulates cortical and medullary COX-2 protein abundance, and 4) boosts total PGE₂ generating activity, which is largely COX-2 activity in CCDs. FSS, a hydrodynamic force generated by fast tubular flow rates applied to our in vitro CD cell culture model, stimulates COX-2 protein expression through N-SM-dependent activity. Moreover, N-SM, the FSS-sensitive sphingomyelinase isoform expressed in IMCD3 cells, is also expressed in murine CD cells (PCs and ICs), suggesting that this shear-sensitive pathway may exist in a native kidney; however, further studies are required to confirm this mechanism in vivo.

Prior studies by others showed that high-Na diets increased urinary PGE₂ excretion and medullary COX-2 expression, as we did; however, these investigators identified enhanced immunodetectable COX-2 protein in interstitial cells (3, 36), not in CDs. In fact, immunodetectable COX-2 was not identified in CDs in either control or high-Na diet animals (36). Our studies suggest that, in both microdissected CCDs and cultured CD cells, flow and shear augment total PGE₂ generating activity and, specifically, COX-2 activity/protein. In our murine model, an increase in tubular flow induces an approximately twofold rise in urinary PGE₂ concentration and CCD-derived PGE₂ within 6 h, suggesting that urinary PGE₂ reflects CD metabolism of AA. Moreover, renal excretion of PGE₂, as measured in the urine, is >10-fold greater in polyuric mice compared with control mice. We speculate the difference in COX-2 activity and localization between the studies by Ye et al. (36) and our current study may reflect differences in 1) the species examined (rat vs. mouse), 2) immunodetection vs. enzymatic activity to assess COX expression, 3) Na ingestion vs. isotonic saline injection, and 4) chronic (3 days) vs. acute (6 h) study (36).

The data fit well within our conceptual framework that tubular flow and FSS rapidly 1) activate cPLA2 through changes in intracellular calcium concentration and phosphorylation to augment AA generation (8) and 2) induce longer-term adaptive changes in COX protein abundance or activity to boost the capacity to convert AA to its intermediate endoperoxides and, hence, PGE₂. This model emerged from our previous work demonstrating that FSS rapidly induces cPLA2 phosphorylation and PGE₂ generation in IMCD3 cells, while COX-2 protein content was not statistically greater than that observed in static control cells (8). In this paper, 4 h of FSS stimulated N-SM-dependent COX-2 protein expression in IMCD3 cells. Moreover, PGE₂ secreted by microdissected CCDs was greater in polyuric than in control kidneys and, importantly, incubation of CCDs with AA was able to further enhance PGE₂ generation, principally through a COX-2-dependent pathway. This suggested to us that acute flow-mediated PGE₂ release requires generation of AA, principally by cPLA2 activation (8). On the other hand, chronic high tubular flow, which accompanies high-Na diets, stimulates COX-2 activity and/or protein to ensure that the total COX activity is sufficient to rapidly process AA to its endoperoxide intermediates so that PGE₂ can be abundantly released and effectuate Na excretion.

Pathologic renal conditions are also associated with high urine/tubular flow rates and increased levels of urinary PGE₂ excretion and renal PGE₂ generation. Before the molecular mechanism underlying Bartter's syndrome was identified, it was already established that urinary PGE₂ was elevated in Bartter's syndrome and that inhibition of renal PGE₂ synthesis with indomethacin (COX-1 and COX-2 inhibitor) restored Na and K balance (2). Further studies implicated that COX-2 was the principal COX isoform contributing to high levels of urinary PGE₂ and that inhibition of COX-2, specifically, suppressed renal PGE₂ release and polyuria (21–23). In addition, autosomal dominant polycystic kidney disease often presents with polyuria and an inability to maximally concentrate the urine (24). Kidneys of rodent and murine models of PKD show elevated cPLA2 protein and total COX activity compared with control mice, and COX-2-specific inhibition suppresses cystogenesis, fibrosis, and macrophage infiltration (1, 25, 30). We speculate that high rates of urine and tubular flow contribute to enhanced renal COX-2 activity and PGE₂ generation which, depending on the context of PGE₂ release, plays a physiologic (Na and K regulation) or pathophysiolgic (proliferation and cystogenesis) role in the kidney (15).

It should be noted that our study does have limitations. Specifically, utilizing our IMCD3 model, we were able to demonstrate the FSS-induced N-SM activity regulates COX-2 protein expression; however, flow activation of N-SM pathway in the native tubule/kidney was not demonstrated, although CCD expression of N-SM in kidney was observed. In regards to the murine model, isotonic saline injection is a crude method to induce high tubular flow rates since injection of saline affects the neurohormonal axis, such as suppressing aldosterone, renin, and angiotensin II. Because the renin-angiotensin-aldosterone axis is suppressed and vasopressin is unaffected by isotonic saline (31), we suspect that change in tubular flow to be the predominant mechanism of action on the tubule. Atrial natruietic peptide (ANP), a small peptide hormone that enhances Na excretion at the CD, is likely elevated with volume expansion; however, little evidence points to ANP regulating CD expression of PGE₂.

In sum, our studies are the first to demonstrate that FSS regulates COX-2 protein expression in a CD cell model through a sphingomyelinase-dependent mechanism. In addition, we demonstrate that CCDs exposed to high tubular flows secrete greater amounts of PGE₂ which, likely, effectuate enhanced renal Na excretion. Finally, tubular flow augments CCD COX activity through stimulation of COX-2 protein or activity that has not been demonstrated in other murine models.

GRANTS

This work was supported by the Department of Veterans Affairs Merit Review 1I01BX000388 (R. Rohatgi) and the Bronx Veterans Medical Research Foundation.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: Y.L., D.F., and R.C.-G. performed experiments; Y.L. and D.F. analyzed data; Y.L. and D.F. prepared figures; R.R. conception and design of research; R.R. interpreted results of experiments; R.R. drafted manuscript; R.R. edited and revised manuscript; R.R. approved final version of manuscript.