Stimulation of diuresis and natriuresis by renomedullary infusion of a dual inhibitor of fatty acid amide hydrolase and monoacylglycerol lipase

Abstract

The renal medulla, considered critical for the regulation of salt and water balance and long-term blood pressure control, is enriched in anandamide and two of its major metabolizing enzymes, cyclooxygenase-2 (COX-2) and fatty acid amide hydrolase (FAAH). Infusion of anandamide (15, 30, and 60 nmol·min⁻¹·kg⁻¹) into the renal medulla of C57BL/6J mice stimulated diuresis and salt excretion in a COX-2- but not COX-1-dependent manner. To determine whether endogenous endocannabinoids in the renal medulla can elicit similar effects, the effects of intramedullary isopropyl dodecyl fluorophosphate (IDFP), which inhibits the two major endocannabinoid hydrolases, were studied. IDFP treatment increased the urine formation rate and sodium excretion in a COX-2- but not COX-1-dependent manner. Neither anandamide nor IDFP affected the glomerular filtration rate. Neither systemic (0.625 mg·kg⁻¹·30 min⁻¹ iv) nor intramedullary (15 nmol·min⁻¹·kg⁻¹·30 min⁻¹) IDFP pretreatment before intramedullary anandamide (15–30 nmol·min⁻¹·kg⁻¹) strictly blocked effects of anandamide, suggesting that hydrolysis of anandamide was not necessary for its diuretic effect. Intramedullary IDFP had no effect on renal blood flow but stimulated renal medullary blood flow. The effects of IDFP on urine flow rate and medullary blood flow were FAAH-dependent as demonstrated using FAAH knockout mice. Analysis of mouse urinary PGE₂ concentrations by HPLC-electrospray ionization tandem mass spectrometry showed that IDFP treatment decreased urinary PGE₂. These data are consistent with a role of FAAH and endogenous anandamide acting through a COX-2-dependent metabolite to regulate diuresis and salt excretion in the mouse kidney.

one of the most important functions of the kidneys is the regulation of salt and water balance, which is considered critical for the long-term control of blood pressure. Indeed, the concept that underlying renal dysfunction is a contributory factor in essential hypertension is gaining increased acceptance (12). There are many examples of mechanisms by which the kidneys regulate salt and water balance, including pressure diuresis (11), hemodynamic effects of renal medullary blood flow (MBF; Ref. 27), and modulation of sympathetic nervous system afferents to the kidneys (14). The actions and mechanisms of the renin-angiotensin-aldosterone system as a positive regulator of blood pressure are well-established, but the nature of negative regulators as antihypertensive factors in the kidney is less well understood.

N-arachidonoylethanolamine (anandamide), the N-acylethanolamide of the C20:4(n-6) fatty acid arachidonic acid, was the first identified endogenous cannabinoid (8). Anandamide levels are high in several tissues, including the brain and liver, where it is proposed to act as a modulator of sensory perception and energy consumption and use. The kidneys represent another organ containing high content of anandamide and its primary hydrolyzing enzyme, fatty acid amide hydrolase (FAAH; Ref. 24). However, few studies have addressed the role of anandamide or FAAH in the kidney. Anandamide affects glomerular filtration via its influence on afferent and efferent arteriolar tone, supporting a role in the renal cortex (7, 15). Endocannabinoids also play a role in the renal medulla (20, 32). We have reported that infusion of anandamide into the mouse renal medulla stimulates urine formation as well as sodium and potassium excretion. Interestingly, the selective cyclooxygenase-2 (COX-2) inhibitor, celecoxib (CEL), blocked effects on renal excretory function. This finding suggests that a COX-2 metabolite of anandamide (Fig. 1, pathway 2), rather than anandamide itself (Fig. 1, pathway 1), mediates the effects of medullary anandamide on diuresis and tubular sodium and potassium excretion. The observation that a similar infusion of prostamide E₂, the COX-2-PGE₂ synthase product of anandamide metabolism, elicits effects similar to anandamide supports a role of prostamide E₂ as a mediator in the mechanism of anandamide. Prostamide E₂ is the major prostamide formed by mouse renal tissue homogenates, especially from renal medulla (19, 32).

Fig. 1.

Potential pathways responsible for the metabolism and biological activity of anandamide following its infusion into the renal medulla. Anandamide may act directly on its cognate receptors (1), or it may act indirectly through its biotransformation by COX-2 to an active metabolite (2). Alternatively, it may undergo hydrolysis by FAAH to release free arachidonic acid before undergoing COX-2 metabolism to PGE₂ (3). TRPV4, transient receptor potential vanilloid 4.

The purpose of this study was to test the hypothesis that the intramedullary infusion of a FAAH inhibitor will mimic the effects of exogenous anandamide, thereby supporting a role of endogenous endocannabinoids as modulators of renal medullary function. For this purpose, we utilized the organophosphate derivative isopropyl dodecylfluorophosphonate (IDFP), which inhibits both major endocannabinoid hydrolases, FAAH and monoacylglycerol lipase (MAGL; Ref. 30). The requirement for COX-1 vs. COX-2 activity in the mechanism of anandamide and IDFP was evaluated using selective inhibitors of these enzymes, SC-560 and CEL, respectively. In addition, the question of whether FAAH plays a necessary role in the effects of IDFP was addressed by testing IDFP in homozygous FAAH knockout mice.

MATERIALS AND METHODS

Anandamide, SC-560, IDFP, PGE₂, and PGE₂-d₄ were from Cayman Chemical (Ann Arbor, MI). CEL was from Sigma Chemical (St. Louis, MO). All other reagents for in vivo use were of the highest grade available.

Animals.

This study used 2- to 4-mo-old male C57BL/6 mice purchased from Charles River and homozygous null FAAH^−/− knockout mice (6) from a colony maintained at Virginia Commonwealth University on a C57BL/6 background. All experiments were conducted under an animal use protocol approved by the Virginia Commonwealth University Institutional Animal Care and Use Committee.

Surgical preparation of mice.

Mice were anesthetized and prepared surgically for each of the following protocols as described previously (23). Anesthesia was induced with ketamine (100 mg/kg; Ketathesia; Harry Schein Animal Health, Dublin, OH) and thiobutabarbital (75 mg/kg; Inactin; Sigma Chemical) injected intraperitoneally. The mice were placed on a thermostatically controlled warming table to maintain body temperature at 37°C. After tracheotomy, catheters were placed in the jugular vein and carotid artery for intravenous infusions and measurement of mean arterial pressure (MAP), respectively. The catheter placed in the carotid artery was attached to a pressure transducer connected to a data acquisition system (WinDaq; DATAQ Instruments, Akron, OH).

Renal intramedullary interstitial infusion studies.

The technique for the construction and placement of a catheter for renal medullary infusion (26) is established in our laboratory (23, 32, 37). In brief, a catheter with a tapered tip (2.5 mm) was implanted into the renal outer medulla vertically from the dorsal surface and anchored into place on the kidney surface with VetBond tissue adhesive (3M, Minneapolis, MN). The medullary catheter was infused with solution containing 90% phosphate-buffered saline [205 mM NaCl, 40.5 mM Na₂HPO₄, and 9.5 mM NaH₂PO₄ (pH 7.4, 610 mOsm)] and 10% of ethanol at a rate of 2 μl/min to maintain the patency of the interstitial infusion line. For the collection of urine from the right kidney, the left ureter was ligated and then cut to keep open on the side proximal to the kidney, and the bladder was catheterized to collect urine. The urine formation rate (UV) was determined gravimetrically, and urinary sodium and potassium concentrations were measured by flame photometry. The UV and the sodium (U_Na) and potassium (U_K) urinary excretion rates were factored per gram of kidney weight. Mice received a continuous intravenous infusion of 0.9% NaCl solution containing 2% albumin at a rate of 5 μl·min⁻¹·25 g body wt⁻¹ to replace fluid loss and maintain a constant hematocrit (~42%). The lack of effect on hematocrit was confirmed in our study. Medullary blood flow (MBF) was measured using a laser Doppler flow probe (OxyFlo Pro; MNP 100XP; Oxford Optronix, Oxford, United Kingdom) placed on the lower lateral surface of the kidney. Renal blood flow (RBF) was measured with a transonic flow probe around the right renal artery (21). After a 1-h equilibration period and 2 10-min control sample collection periods during infusion of the vehicle alone, anandamide or IDFP were infused for 30-min periods (1 10-min dead space-removing period and 2 sequential 10-min sample collection periods) at each dose level (15, 30, and 60 nmol·min⁻¹·kg⁻¹; 15, 30, and 60 IDFP dose groups). Anandamide and IDFP were first dissolved in ethanol before adding phosphate-buffered solution for a final osmolarity of 550 mOsm.

To examine the effects of pretreatment with the COX-1-selective (SC-560) or the COX-2-selective inhibitor (CEL), these drugs were administered at 0.2 μmol·min⁻¹·kg⁻¹ in the intravenous vehicle (2% bovine serum albumin in normal saline) for 30 min before the control infusion periods. The endocannabinoid hydrolysis inhibitor, IDFP, was infused as described below. This surgical preparation and protocol procedure have been widely used in the studies of renal physiology, and infusion of vehicle alone did not produce any significant change in renal function over the period of testing. A sham control study was performed using mice treated identically to the IDFP-infused mice (i.e., identical surgeries and instrumentation) except that vehicles only were infused the entire experimental period. At the end of the experiment, blood was drawn by intracardiac puncture for determination of hematocrit, and the right and left kidneys were removed and weighed. The position of the medullary catheter was confirmed after sagittal sectioning of the right kidney.

Determination of glomerular filtration rate.

Glomerular filtration rate (GFR) was estimated using the FITC-inulin-based method (25) as described previously by our laboratory (22). A 0.5-ml bolus of FITC-inulin (4 mg/ml) was given, followed by a steady intravenous infusion (4.0 mg/ml), which continued throughout the experiment. After each experiment, FITC-inulin levels in blood and urine samples were measured with excitation and emission wavelengths of 480 and 530 nm, respectively, using an automated fluorescence microplate reader (FLx800; BioTek Instruments, Winooski, VT) and Gen5 data analysis software.

HPLC-electrospray ionization tandem mass spectrometry analysis of PGE₂.

Urinary PGE₂ was measured in urine by a method adapted from Di Marzo et al. (9). Briefly, to 40 µl of urine, 50 ng of PGE₂-d₄ internal standard was added. The sample was then extracted using 2:1 (vol/vol) chloroform/methanol (1 ml) and 0.73% sodium chloride (200 µl). After centrifuging, the organic phase was collected, and the aqueous phase was extracted twice more with chloroform (1 ml). The organic phases were combined and evaporated to dryness under nitrogen, reconstituted in 50 µl of methanol, and placed in autosampler vials for high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (HPLC-ESI MS/MS) analysis. PGE₂ was separated using a Shimadzu HPLC system (Kyoto, Japan) with a Discovery HS C18 Column 15 cm × 2.1 mm, 3 µm (Supelco, Bellefonte, PA) kept at 25°C and detected using a Sciex QTRAP 6500 system with an IonDrive Turbo V source for TurboIonSpray run in multiple reaction monitoring mode. The mobile phase was 60% water and 40% acetonitrile with 1 g/l ammonium acetate and 0.1% formic acid. The flow rate was 0.5 ml/min, and the injection volume was 5 µl. Total run time was 5.0 min. The transition ions monitored for PGE₂ and PGE₂-d4 are shown in Table 1. A calibration curve was constructed for the assay by linear regression using the peak area ratios of the calibrators and internal standard. The standard curve ranged from 1 to 500 ng with a correlation (r²) of 0.9982. Urine creatinine concentration was determined by a Jaffe reaction assay kit according to the manufacturer's protocol (Cayman Chemical) for expression of the PGE₂ concentration in units of nanograms per milligram creatinine.

Table 1.

Multiple reaction monitoring parameters for urinary PGE₂ analysis by HPLC-ESI MS/MS

Compound	Transition Ions, m/z	DP, V	CE, V
PGE2	351 > 271	−40	−23
	351 > 315	−40	−15
PGE2-d4	355 > 275	−40	−22
	355 > 319	−40	−15

m/z, Mass-to-charge ratio; DP, deprotonation energy; CE, collision energy.

Statistical analyses.

Data are presented as the means ± SE. For comparisons of multiple groups of data, one- or two-way analysis of variance (ANOVA) was performed using a Tukey post hoc test when significant differences were found. A P value ≤0.05 was considered statistically significant.

RESULTS

Effects of intramedullary infused anandamide are blocked by a COX-2- but not COX-1-selective inhibitor.

Following a 1-h equilibration period, the mean urine formation rates for two 10-min vehicle control infusion periods were 10.5 and 9.7 μl·min⁻¹·g kidney weight⁻¹, whereas for sodium and potassium excretion, the rates were 1.2 and 1.8 and 0.46 and 0.49 μmol·min⁻¹·g kidney weight⁻¹, respectively (Fig. 2). Infusion of anandamide alone dose-dependently increased the urine formation rate to 14, 15.9, and 19 μl·min⁻¹·g kidney weight⁻¹ at infusion rates of 15, 30, and 60 nmol·min⁻¹·kg⁻¹. As observed previously, the anandamide effect was accompanied by an increase in urinary excretion rates of sodium and potassium, but there was no effect on MAP. Whereas CEL completely blocked the effects of intramedullary-infused anandamide on UV, U_Na, and U_K, SC-560 administered at the same dose rate had no effect on the responses to anandamide. These findings support the assertions that COX-2 metabolites of anandamide mediate the effects of infused anandamide in the renal medulla of mice.

Fig. 2.

Effect of intramedullary interstitial infusion of anandamide on mean arterial pressure, urine formation, and urinary sodium and potassium excretion in the absence or presence of either a COX-1- or COX-2-selective inhibitor. C indicates control infusion periods with vehicle alone; 15, 30, and 60 indicate dose rates for intramedullary anandamide infusion (in units of nanomoles per kilogram per minute); and P indicates posttreatment control infusion periods. Animal groups received a 20-min pretreatment with either vehicle (Vehl; 2% albumin in saline) or vehicle containing SC-560 or celecoxib (CEL) at 0.2 μmol·min⁻¹·kg⁻¹ by intravenous infusion immediately before the 1st control infusion period. Data represent the means ± the standard error of the mean of each group. *Significant difference vs. the C2 control group (P < 0.05) by 2-way ANOVA with Tukey post hoc test (n = 6–7 mice per group). AEA, N-arachidonoylethanolamine.

Stimulation of diuresis and natriuresis using an inhibitor of FAAH and MAGL and selective blockade by COX-2 inhibition.

To determine whether endogenous endocannabinoids have a similar effect, we investigated the effect of IDFP, which inhibits endocannabinoid hydrolases with IC₅₀s in the nanomolar range (0.8 and 3 nM for FAAH and MAGL, respectively) while having low inhibitory potency toward acetylcholinesterase (IC₅₀ >6,000 nM; Ref. 30). Infusion of IDFP into the outer medulla of the mouse kidney had no effect on MAP but dose-dependently increased the urine formation rate, elevating it from 11.1 ± 1.4 to 21.3 ± 1.9 μl·min⁻¹·g kidney weight⁻¹ in the C2 control and 60 IDFP dose groups, respectively (Fig. 3). The sodium excretion rate was increased between these two groups (1.30 ± 0.36 to 2.69 ± 0.57 μmol·min⁻¹·g kidney weight⁻¹, respectively). Potassium excretion was not significantly affected. The IDFP effect on urine formation rate was completely prevented by pretreatment with CEL but not SC-506, suggesting that the effect of the endogenous endocannabinoids is also COX-2-dependent. These results support the hypothesis that endogenous endocannabinoids can influence renal excretory parameters. The similarity in the pattern of effects of IDFP to those of anandamide suggests that inhibition of FAAH leading to accumulation of anandamide is the basis for the effect of IDFP.

Fig. 3.

Effect of intramedullary interstitial infusion of IDFP on mean arterial pressure, urine formation, and renal sodium and potassium excretion in the absence or presence of a COX-1- or COX-2-selective inhibitor. C indicates control infusion periods with vehicle alone; 15, 30, and 60 indicate rates of intramedullary IDFP infusion (in nanomoles per kilogram per minute); and P indicates posttreatment control infusion periods. Animal groups received a 20-min pretreatment with either vehicle (VEH; 2% albumin in saline) or vehicle containing SC-560 or celecoxib (CEL) at 0.2 μmol·min⁻¹·kg⁻¹ by intravenous infusion immediately before the 1st control infusion period. *Significant difference vs. the respective C2 control group (P < 0.05) by 2-way ANOVA with Tukey post hoc test (n = 6–7 per group).

Effect of systemic or intramedullary infusion of IDFP on responses to anandamide.

We also addressed the question of whether the responses to anandamide are dependent on its initial hydrolysis by FAAH to release arachidonic acid (Fig. 1, pathway 3). To test this, the effects of IDFP, given either systemically or by intramedullary infusion, on the responses to anandamide were evaluated. For systemic IDFP, a dose of 0.625 mg/kg (~2,100 nmol/kg) administered intravenously over 30 min was selected because higher doses were observed to reduce MAP and UV dramatically (data not shown). Intravenous IDFP at 0.625 mg·kg⁻¹·30 min⁻¹ did not prevent anandamide from inducing urine formation or salt excretion (Fig. 4A).

Fig. 4.

Effects of intramedullary anandamide on blood pressure and renal urinary end points after pretreatment with IDFP. A: after intravenous IDFP (0.625 mg·kg⁻¹·30 min⁻¹). B: after intramedullary IDFP pretreatment (15 nmol·min⁻¹·kg⁻¹). I1 and I2 indicate the control IDFP infusion periods; 15, 30, and 60 indicate rates of intramedullary anandamide infusion (in nanomoles per kilogram per minute); and P indicates posttreatment control infusion periods. *Significant difference vs. the I2 pretreatment control group (P < 0.05) by 1-way ANOVA with Tukey post hoc test (n = 5–7 per group).

The results showing the effect of intramedullary IDFP (15 nmol·min⁻¹·kg⁻¹) before anandamide are shown in Fig. 4B. Anandamide stimulated diuresis and salt excretion, but the dose-response curve was altered. In addition to a leftward shift suggestive of an increase in potency, the overall dose-response curve assumed an inverted-U shape, with its maximum at 7.5 nmol·min⁻¹·kg⁻¹ and decreasing progressively as the anandamide dose rate increased. These differences are consistent with an increase in potency of anandamide as a result of an inhibition of anandamide hydrolysis and a contribution of multiple receptors/mechanisms at high renal medullary anandamide concentrations.

Effect of intramedullary infusion of anandamide or IDFP on GFR.

To determine whether the increased renal function parameters observed after anandamide or IDFP administration were due to changes in GFR, the effects of these agents on GFR were determined, and the data are presented in Fig. 5. Neither anandamide nor IDFP significantly altered the GFR either in the presence or absence of the selective COX-2 inhibitors.

Fig. 5.

Effects of intrarenal medullary infusion of anandamide or IDFP on the glomerular filtration rate (GFR) in the presence or absence of COX-1- or COX-2-selective inhibitors. Effect of intramedullary infusion of anandamide (A) or IDFP (B; at 15, 30, or 60 nmol/kg/min). Animal groups received a 20-min pretreatment with either vehicle (2% albumin in saline) or vehicle containing SC-560 or celecoxib (CEL) at 0.2 μmol·min⁻¹·kg⁻¹ by intravenous infusion immediately before the 1st control infusion period (n = 5 per group). Data were analyzed by 2-way ANOVA.

Effect of intramedullary IDFP on RBF and MBF.

Intramedullary IDFP had no effect on RBF, but MBF increased significantly, from −2% in the C2 pretreatment control to 27% in the highest IDFP treatment group (Fig. 6). This moderate increase in MBF persisted during the posttreatment phase of the experiment. This result suggests that the diuretic and natriuretic effect may be attributable in part to renal hemodynamic effects of IDFP.

Fig. 6.

The effect of intramedullary IDFP infusion on renal blood flow (RBF) and medullary blood flow (MBF) in C57BL/6J mice. C1 and C2 are pretreatment control periods that received intramedullary infusion of vehicle alone; 15, 30, and 60 indicate the dose rates (in nanomoles per minute per kilogram) for the IDFP treatment periods; P1 and P2 are the post-IDFP vehicle infusion periods. *Significant difference vs. the C2 control group (P < 0.05) by 1-way ANOVA with Tukey post hoc test (n = 5).

Effects of intramedullary infusion of IDFP on MAP, UV, and MBF in homozygous FAAH knockout mice.

Renomedullary infusion of IDFP had no effect on MAP at the lowest IDFP dose in FAAH^−/− mice, but MAP decreased significantly (P < 0.05) at the 30 and 60 nmol·min⁻¹·kg⁻¹ IDFP infusion rates and remained elevated during the posttreatment phase (Fig. 7). Neither the UV nor MBF were significantly affected by IDFP treatment in FAAH^−/− animals. These results indicate that the stimulatory effects of intramedullary IDFP on urine formation and MBF are FAAH-dependent.

Fig. 7.

Effects of intramedullary IDFP infusion on mean arterial pressure (MAP), urine formation (UV), and medullary blood flow (MBF) in homozygous FAAH^−/− knockout mice. C1 and C2 represent the control infusion periods with vehicle alone; 15, 30, and 60 indicate IDFP dose treatment periods (in nanomoles per kilogram per minute); and P1 and P2 indicate posttreatment control infusion periods. *Significant difference vs. the C2 control group (P < 0.05) by 1-way ANOVA with Tukey post hoc test (n = 6).

Effect of intramedullary infusion of IDFP on urinary PGE₂ concentrations.

Urinary PGE₂ concentrations were determined to assess the role of PGE₂ in the effect of IDFP (Fig. 8). High variation (10- to 100-fold) was observed in the urinary PGE₂ concentrations of C57BL/6J mice from the vehicle control infusion phase (normalized for urinary creatinine). Analysis of mouse urine samples (control vs. IDFP treatment phase from individual mice) indicated a significant decrease in urinary PGE₂ concentration (P < 0.05, Wilcoxon signed-rank test for paired, nonparametric data) in C57BL/6J mice (Fig. 8A). Although the effect was not significant (P = 0.0625; Fig. 8B), a downward trend was also apparent in the PGE₂ levels in control vs. treated FAAH knockout mice, suggesting that the IDFP-induced decrease in PGE₂ occurs independently of FAAH.

Fig. 8.

Effect of intramedullary IDFP treatment on urinary PGE₂ concentration. A: C57BL/6J mice. B: FAAH^−/− knockout mice. Urine samples from the vehicle only (VEH) and IDFP treatment groups (15, 30, and 60 nmol·min⁻¹·kg⁻¹) were combined and analyzed for PGE₂ by HPLC-ESI MS/MS. The data were normalized by urinary creatinine content. Data for individual C57BL/6J (n = 7) and FAAH^−/− knockout (n = 5) mice are shown. Bold hashes represent the arithmetic means of each group. *Significant difference using a Wilcoxon signed-rank test for nonparametric paired data.

MAP, UV, MBF, and hematocrit data for control mice receiving only vehicle infusions.

No significant changes occurred in MAP, urine formation rate, or MBF in male C57BL/6 mice that received identical surgical preparation but only vehicle infusions throughout the entire experimental period (Table 2).

Table 2.

Mean arterial pressure, urine formation, medullary blood flow, and hematocrits for sham control mice

	C1	C2	V1	V2	V3	P1	P2
MAP, mmHg	113 ± 11	117 ± 9	123 ± 6	127 ± 6	125 ± 6	125 ± 7	115 ± 8
UV, μl·min−1·g kidney weight−1	13 ± 3	13 ± 3	11 ± 3	12 ± 3	12 ± 5	12 ± 4	13 ± 4
MBF, %	6 ± 7	3 ± 2	0 ± 9	9 ± 13	2 ± 9	2 ± 13	1 ± 14
Hematocrit, %	48 ± 2

Data shown are the means ± SE for control mice (n = 5). Male C57BL/6J mice received identical surgical preparation but only vehicle infusion during the entire experimental period. C1/C2, V1/V2/V3, and P1/P2 correspond to the pretesting, testing, and posttesting infusion periods. The hematocrit levels were measured at the conclusion of the experiment for individual mice.

DISCUSSION

Our laboratory is investigating the possible relationship of the endocannabinoid system to the renomedullary neutral antihypertensive lipid proposed by Muirhead (29). The latter lipid, known as medullipin, is proposed to originate from the interstitial cells of the renal medulla and to exert its antihypertensive effects by a combination of mechanisms, including vasodilation, stimulation of sodium excretion and loss of body fluid volume, and sympatholytic effects. When exogenous anandamide was infused into the renal medulla, it elicited diuretic and natriuretic effects that were unaccompanied by significant changes in MAP. The effects on diuresis and natriuresis were blocked by CEL but not SC-560 (Fig. 2), suggesting that a COX-2 metabolite of anandamide mediates this effect (Fig. 1). The present study also addressed whether endogenous anandamide or endocannabinoids in the renal medulla similarly modulates renal excretory function, thereby highlighting their physiological significance. The data show that intramedullary infusion of IDFP, which inhibits hydrolysis of the two major endocannabinoids, anandamide and 2-arachidonoylglycerol (2-AG), stimulated both urine formation and sodium excretion. Similar to anandamide, the effects of IDFP were COX-2- but not COX-1-dependent, suggesting a role for a COX-2 metabolite in both cases.

The marked dependence of the anandamide and IDFP effects on COX-2 but not COX-1 was demonstrated using selective COX inhibitors. Pretreatment with SC-560, a member of the same diaryl heterocyclic class as CEL, with strong inhibitory selectivity for COX-1 vs. COX-2 (IC₅₀ values of 9 nM and 6.3 mM, respectively; Ref. 34), had no significant effects on either the anandamide or IDFP responses, whereas the COX-2-selective inhibitor, CEL, completely inhibited their diuretic responses. The basis for the COX-2 dependence of the diuretic effects of anandamide and IDFP remains unknown at present but may be related to the fact that anandamide is a COX-2- but not COX-1-selective substrate (36). The COX-2 enzyme contains a larger side pocket adjacent to the active site of COX-2 able to accommodate the ethanolamine side chain of anandamide (16). The predominance of COX-2 as the major COX in the renal medulla may also contribute (2, 35). Yu et al. (36) showed that anandamide is a substrate for COX-2-PGE₂ synthase, undergoing metabolism to PGE₂-ethanolamide (i.e., prostamide E₂). The role of prostamide E₂ in the mechanism of intramedullary anandamide or IDFP remains unclear, but it is detected in the mouse renal medulla (G. Li and J. K. Ritter, unpublished observations). Homogenates of renal medullary tissue were more active than renal cortical tissue in forming prostamide E₂ from anandamide (32).

Since IDFP inhibits the hydrolytic enzymes of anandamide and 2-AG, FAAH and MAGL, respectively (30), the data showing the diuretic effect of intramedullary IDFP in mice (Fig. 3) do not strictly exclude a role for 2-AG as a mediator. Similar to anandamide, 2-AG is also a substrate for COX-2 but is reportedly an even more efficient substrate than anandamide, undergoing conversion to its respective prostaglandin derivative, PGE₂-2-O-glycerol (17, 18). In our earlier study (32), 2-AG levels were also measured in the mouse kidney and compared between renal cortex and medulla. 2-AG did not show the same marked gradient across the corticomedullary axis as found for anandamide. The data show that the diuretic effect of intramedullary IDFP was blocked in the FAAH knockout mice (Fig. 7), implying that FAAH inhibition is necessary for the mechanism of IDFP. This inhibition would result in increased levels of tissue fatty acid amides, such as anandamide, oleoylethanolamide, and palmitoylethanolamide. However, as the latter two ethanolamines are not substrates of COX-2, they do not account for the observed COX-2 dependence of the IDFP effects.

The results of our study support the idea that anandamide-induced diuresis occurs without FAAH-mediated hydrolysis to release arachidonic acid, the precursor to PGE₂, the major renal prostaglandin. In the kidneys, PGE₂ acts as a paracrine hormone to inhibit sodium reabsorption, promotes diuresis, and acts as a vasodilator in the renal vasculature (13). The diuretic effect of IDFP, an inhibitor of FAAH, was accompanied by decreased urinary PGE₂, providing direct support for the PGE₂ independence of this effect. This effect was not due to decreased COX-2 protein as Western blots showed no effect of acute IDFP treatment (data not shown). A plausible explanation for this PGE₂-lowering action is that the elevation of tissue anandamide/2-AG levels occurring secondary to hydrolysis inhibition serves to inhibit COX-2 metabolism of arachidonic acid competitively. In addition, the interaction experiments (Fig. 4) demonstrated that IDFP did not inhibit anandamide-induced diuresis. In fact, the opposite appeared true, that IDFP enhances the potency of intramedullary anandamide (Fig. 4B). These results together with the observations using FAAH knockout mice lead to the conclusion that the diuretic effect of IDFP occurs by a mechanism independent of pathway 3 (involving PGE₂ synthesis) in Fig. 1.

Paronis and colleagues (4, 5, 31) recently reported that various cannabinoid type 1 (CB₁) receptor agonists share the capacity to stimulate diuresis (defined as an increased loss of urine due to increased formation and/or increased micturition) in mice and rats. These findings suggest that the diuretic actions of anandamide or its COX-2 metabolites formed on intramedullary infusion of anandamide or IDFP could be CB₁-mediated (Fig. 1, pathway 1 or 2), which has not yet been addressed in our studies of the mechanism. CB₁ receptor expression is detectable in the renal medulla by RT-PCR and Western blotting, although its levels appear low (J. K. Ritter and P.-L. Li, unpublished observations). It is interesting, however, to note the different shapes of the dose-response curves described for different physiological end points of CB₁ receptor activation (5, 31). In contrast to the conventional dose-response curves observed for antinociceptive effects of CB₁ agonists, the diuresis dose-response curves exhibited a leftward shift to higher potency and an inverted-U shape (5), similar to the curve in Fig. 4B. This may suggest an underlying novelty in the diuretic mechanism of CB₁ agonists.

In this study, intramedullary IDFP was observed to stimulate MBF in the absence of any change in RBF in C57BL/6 mice (Fig. 6), whereas its effect on MBF in FAAH knockout mice was nonsignificant. This observation suggests that the increased diuresis and natriuresis by IDFP is due in part to increased MBF, which is thought to cause washout of the salt gradient in the inner medulla necessary for the ability to concentrate the urine. MBF is thought to be regulated by the net balance of local paracrine or autocrine hormones that vasoconstrict or vasodilate the outer medullary descending vasa recta (10, 27). Substances known to increase MBF with a resulting increase in diuresis and natriuresis include nitric oxide (28), PGE₂ (3), and bradykinin (1, 33). However, it should be noted that our data do not exclude the possibility that COX-2-dependent metabolites of anandamide also stimulate diuresis and natriuresis by direct actions on tubular epithelia leading to inhibition of sodium and water reabsorption. Further studies are required to delineate between these possibilities.

An unexpected finding in our study was the observation that FAAH knockout mice show reduced MAP in response to intramedullary IDFP infusion. This effect occurred in the FAAH knockout but not C57BL/6 mice (background of FAAH knockout), and it was prolonged, lasting even through the following posttreatment control phase. The mechanism of this effect remains to be investigated, but it may be due to inhibition of other serine hydrolase targets in the renal medulla, such as MAGL.

In summary, the current data support a role of renomedullary anandamide acting through a COX-2 metabolite as an endogenous stimulator of diuresis, an action consistent with the reported properties of medullipin. Studies are ongoing with more highly selective FAAH and MAGL inhibitors and to assess the actions and roles of endogenous endocannabinoids as a possible renomedullary system important the long-term blood pressure regulation.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-102539 and National Institute on Drug Abuse Grant P30-DA-033934.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

N.L., J.L.P., P.-L.L., and J.K.R. conceived and designed research; A.A., Z.D., G.L., S.K.D., and J.L.P. performed experiments; A.A., Z.D., G.L., S.K.D., A.L., and P.-L.L. analyzed data; A.A., Z.D., G.L., S.K.D., N.L., J.L.P., A.L., and P.-L.L. interpreted results of experiments; A.A., Z.D., and J.L.P. prepared figures; J.L.P. and J.K.R. drafted manuscript; A.A., Z.D., G.L., S.K.D., N.L., A.L., P.-L.L., and J.K.R. edited and revised manuscript; J.K.R. approved final version of manuscript.