Abstract
Synthetic agonists of the peroxisomal proliferator-activated receptor subtype γ (PPAR-γ) are highly beneficial in the treatment of type II diabetes. However, they are also associated with fluid retention and edema, potentially serious side effects of unknown origin. These studies were designed to test the hypothesis that rosiglitazone (RGZ, PPAR-γ agonist) may activate sodium- and water-reabsorptive processes in the kidney, possibly in response to a drop in mean arterial blood pressure (MAP), as well as directly through PPAR-γ. Targeted proteomics of the major renal sodium and water transporters and channel proteins was used to identify potentially regulated sites of renal sodium and water reabsorption. RGZ (47 or 94 mg/kg diet) was fed to male, Sprague-Dawley rats (∼270g) for 3 days. MAP, measured by radiotelemetry, was decreased significantly in rats fed either level of RGZ, relative to control rats. Delta MAP from baseline was –3.2 ± 1.2 mm Hg in rats fed high-dose RGZ versus + 3.4 ± 0.8 for rats fed control diet. RGZ did not affect feed or water intake, but rats treated with high-dose RGZ had decreased urine volume (by 22%), sodium excretion (44%), kidney weight (9%), and creatinine clearance (35%). RGZ increased whole kidney protein abundance of the α-1 subunit of Na-K-ATPase, the bumetanide-sensitive Na-K-2Cl cotransporter (NKCC2), the sodium hydrogen exchanger (NHE3), the aquaporins 2 and 3, and endothelial nitric-oxide synthase. We conclude that both increases in renal tubule transporter abundance and a decrease in glomerular filtration rate likely contribute to the RGZ-induced sodium retention.
The role of peroxisomal proliferator-activated receptor subtype γ (PPAR-γ) agonists, thiazolidinediones (TZDs, rosiglitazone and pioglitazone), in the treatment of type 2 diabetes is firmly established. TZDs have been demonstrated to increase insulin sensitivity in diabetic rats (Walker et al., 1999; Ide et al., 2000; Kanoh et al., 2000) and humans (Day, 1999; Fuchtenbusch et al., 2000). The PPAR-γ receptor is a member of the nuclear receptor superfamily of ligand-dependent transcription factors that both positively and negatively regulate gene expression in response to the binding of a number of fatty acid metabolites (Willson et al., 2000). These receptors are found in various tissues, including skeletal muscle, adipose tissue, heart, large and small intestine, colon, and kidney (Willson et al., 2000).
Thiazolidinediones have been demonstrated to lower blood pressure in obese Zucker rats (Walker et al., 1999), Otsuka Long Evans Tokushima Fatty rats (Kosegawa et al., 1999), diabetic mice (Berger et al., 1996), and obese, insulin-resistant humans (Day, 1999; Fuchtenbusch et al., 2000). The effects of TZDs on blood pressure in normotensive rats and humans is less clear. The mechanism by which blood pressure falls is not known, but in insulin-resistant animals, the fall may be at least partly due to increased insulin sensitivity. However, alternative mechanisms, such as peripheral vasodilation due to nitric oxide release, may also have a role. Calnek et al. (2003) reported increased nitric oxide release from cultured endothelial cells in response to PPAR-γ ligands.
Nonetheless, the propensity for these drugs to cause fluid retention and pulmonary and peripheral edema has emerged recently as the most common, serious adverse drug reaction associated with these compounds (Hirsch et al., 1999; Fuchtenbusch et al., 2000; Thomas and Lloyd, 2001; Martens et al., 2002; Idris et al., 2003). Moreover, the pulmonary edema has been associated with congestive heart failure in patients treated with TZDs (Hirsch et al., 1999; Thomas and Lloyd, 2001). The cause(s) of edema and fluid retention with the use of TZDs are not known and are likely multifactorial. However, increased sodium retention at the renal level in all probability, plays a role.
Regulation of NaCl reabsorption in the kidney is accomplished by regulation of key sodium transport proteins that line the renal epithelia (Knepper, 2002). These include the basolateral Na-K-ATPase, as well as the apical proteins: the sodium hydrogen exchanger subtype III (NHE3), the sodium phosphate cotransporter subtype II (NaPi-2), the bumetanide-sensitive Na-K-2Cl cotransporter (NKCC2 or BSC1), the thiazide-sensitive Na-Cl cotransporter (NCC or TSC), and the amiloride-sensitive sodium channel (ENaC).
The major water channel proteins (aquaporins, AQPs) that have been characterized in the kidney include AQP1 to 4 of which AQP1 and AQP2 function on the apical membrane and AQP3 and AQP4 on the basolateral membrane (Knepper et al., 1996). Recently, the cDNAs for most of these major transporter and channel proteins have been cloned. This knowledge has facilitated the production of specific antibodies against these proteins. These new tools now allow direct investigation of sodium transport regulation at the molecular level.
In these studies, we tested the hypothesis that the sodium and water retention accompanying the use of TZDs may be the result of regulation of sodium and water transport proteins of the kidney. We use targeted proteomics via semi-quantitative immunoblotting to comprehensively evaluate changes in protein abundance of the major sodium and water transport proteins in male, Sprague-Dawley rats in response to short-term dietary treatment with rosiglitazone (RGZ). We purposely chose a short time point to study, i.e., 3 days, so that rats would not yet be in equilibrium with regard to sodium and water balance. We anticipated that most of the changes in water and/or sodium balance would be rapid and adjusted for by pressure natriuresis, as well as diuresis, in these young healthy rats with a longer exposure. Second, we hypothesized that some of these changes might be indirect, in response to changes in blood pressure, possibly in response to increased nitrates plus nitrites (NOx) activity. Therefore, blood pressure, via radiotelemetry, as well as NOx excretion and the renal abundance of endothelial nitric-oxide synthase (eNOS or NOSIII), the most prevalent isoform of NOS in the kidney (Cowley et al., 2003), was also measured.
These studies are the first to provide a comprehensive examination of semichronic effects of PPAR-γ agonists on renal sodium retention and the potential molecular mechanisms underlying them. They provide us with insight as to potential adjuvant treatments that might reduce sodium and water retention and edema in patients treated with PPAR-γ agonists.
Materials and Methods
Animals and Study Design. Male, Sprague-Dawley rats were obtained from Taconic Farms (Gaithersburg, MD) at 260 g. Rats were fed either control diet or diet containing rosiglitazone for 3 days. After a 5-day equilibration period, rats were placed on diets. RGZ, a thiazolidinedione, was incorporated into a chow-based diet (LabDiet Rodent Chow 5001; Purina Mills, St. Louis, MO) at a level of 94 mg/kg diet. The diets were made by melting agar (1% by weight) in water (65%), cooling and adding the drug (0.010%), and ground chow (33.99–34%). The dose selected for RGZ was approximately 4-fold higher than the dose that Walker et al. (1999) had shown reduced blood pressure in obese Zucker rats when fed chronically. We chose this higher dose because our study length was very short and we wanted to observe the maximal effect. Rats were fed and provided additional water ad libitum. Rats were singly housed throughout the study in Nalgene metabolic cages (Harvard Apparatus, Holliston, MA) for collection of urine. Sodium balance was calculated by subtracting urinary sodium excretion from dietary sodium intake and using daily measured feed intake. Feed composition (sodium content) was supplied by the manufacturer. All animals were maintained at all times under conditions and protocols approved by the Georgetown University Animal Care and Use Committee, an American Association for Accreditation of Laboratory Animal Care-approved facility.
Rats were euthanized by decapitation and trunk blood collected into both heparinized- and K3-EDTA-tubes (Vacutainer; BD Biosciences Franklin Lakes, NJ). Both kidneys were rapidly removed and either frozen on dry ice for later processing or immediately homogenized.
Radiotelemetry Monitoring of Blood Pressure. In a second study, blood pressure was assessed on a continual basis by radiotelemetry (Dataquest IV; Data Sciences International, St. Paul, MN). Here, a third treatment group of rats was studied with a lower level of RGZ added to the diet, i.e., 47 mg/kg diet, to assess sensitivity of the rats to the drug with regard to blood pressure changes. Again, rats were treated for 3 days with (n = 6–7/group). Before surgery, rats were anesthetized with isoflurane (IsoFlo; Abbot Laboratories, North Chicago, IL). A flexible catheter with a fluid-filled, pressure-sensitive tip was placed in the abdominal aorta via a femoral artery insertion and the telemetry transmitter was secured in a subcutaneous pocket under the left hind leg. Upon recovery, rats were housed in individual cages, and each cage was placed over a receiver panel with output to a computer. After a 1-week recovery period, rats began dietary treatments for 3 days. Blood pressure was recorded for 10 s every 10 min during the study.
Blood/Plasma Analyses. Blood glucose levels were monitored daily by a glucometer (TheraSense; Freestyle, Alameda, CA) in blood obtained from a tiny cut made with a razor blade in the tail of the rats. After sacrifice, whole blood was centrifuged at 3000 rpm (Sorvall RT 6000 D; Sorvall, Newton, CT) at 4°C for 20 min to separate plasma. Plasma aldosterone and insulin levels were measured by radioimmunoassays [Coat-A-Count; Diagnostic Products Incorporated, Los Angeles, CA (aldosterone) and RI-13K; Linco Research Incorporated, St. Charles, MO (insulin)]. Plasma was also analyzed for creatinine (Jaffe rate method, Creatinine Analyzer 2; Beckmann Diagnostic Systems Group, Brea, CA).
Urine Analyses. Urine, collected daily, was analyzed for sodium (ion-selective electrode system, EL-ISE Electrolyte System, Beckman Coulter Inc., Brea, CA), creatinine, and NOx (Griess reaction; colorimetric kit no. 780001; Cayman Chemical, Ann Arbor, MI).
Targeted Proteomics. We term the primary method used to assess the regulation of protein abundance of the major renal sodium and water transport proteins “targeted proteomics”. With this approach, we have predetermined a set of target proteins of interest and then acquired specific antibodies to them (Knepper, 2002). Over the past 5 years, we have produced or commercially obtained an array of antibodies against the primary sodium and water transport proteins that are expressed along the renal epithelial. These proteins were selected based on their importance as a primary means of transepithelial water or sodium transport based on previous physiological studies. Here, we use these antibodies to comprehensively evaluate changes in abundance of these proteins in response to RGZ.
Preparation of Samples for Immunoblotting. Whole left kidneys or right kidney inner medullas were homogenized (Tissumizer; Tekmar Company, Cincinnati, OH) in a chilled, buffered, isolation solution containing 250 mM sucrose (Sigma-Aldrich., St. Louis, MO), 10 mM triethanolamine (Sigma), 1 μg/ml leupeptin (Bachem, Torrance, CA) and 0.1 mg/ml phenylmethylsulfonyl fluoride (Sigma-Aldrich) adjusted to pH 7.6. Protein concentrations of homogenates were measured by the BCA protein assay reagent kit (Pierce Chemical, Rockford IL). All samples were diluted with isolation solution to a concentration of between 1 and 3 μg/μl and solubilized at 60°C for 15 min in Laemmli sample buffer.
Immunoblotting. Initially, “loading gels” were done. Five micrograms of protein from each sample was electrophoresed on 12% polyacrylamide gels (Bio-Rad, Hercules, CA) and then stained with Coomassie Blue dye. Selected bands were scanned (Scan Jet IIC; Hewlett Packard, Palo Alto, CA) to determine density (NIH Image software, Bethesda, MD) and relative amounts of protein loaded in each lane. Finally, protein concentrations were corrected to reflect these measurements. For immunoblotting, 5 to 30 μg of protein from each sample was loaded into individual lanes of minigels of 7, 10, or 12% polyacrylamide. After electrophoresis, proteins were transferred to pure nitrocellulose membranes (Bio-Rad). After a 5% milk block, membranes were probed overnight at 4°C with the desired primary antibody. The production, affinity purification, and characterization of the polyclonal antibodies against NHE3, NaPi-2, NKCC2, NCC, α-, β-, and γ-ENaC, as well as aquaporins 1 to 4 have been described previously (Ecelbarger et al., 1995, 1996; Terris et al., 1995; Nielsen et al., 1996; Terris et al., 1996; Kim et al., 1998, 1999; Masilamani et al., 1999). The mouse monoclonal antibody to the α-1 subunit of Na-K-ATPase was obtained from Upstate Biotechnology (Lake Placid, NY). Mouse monoclonal eNOS antibody was obtained from BD Biosciences (San Jose, CA). Blots were probed overnight with primary antibodies (0.05–2 μg/ml), washed, incubated with a horseradish peroxidase-coupled secondary antibody (Kirkegaard and Perry Laboratories, Gaithersburg, MD), and washed again. Sites of antibody-antigen reaction were visualized using luminol-based enhanced chemiluminescence (LumiGLO; Kirkegaard and Perry Laboratories) before exposure to X-ray film. Relative intensities of the resulting immunoblot band densities were determined by laser scanning followed by analysis with NIH Image software.
Statistics. Data were evaluated by SigmaStat (SPSS Science Inc., Chicago, IL). Unpaired t test, one-way analysis of variance, or Mann-Whitney rank sum test (when data were not normally distributed) were used to compare treated to untreated rats. p < 0.05 was considered statistically significant for all tests.
Results
Physiological Data. Rats treated with RGZ did not differ in final body weight or feed intake from controls (Table 1). Figure 1A shows daily weight gain in these rats. There was some tendency for RGZ-treated rats to gain weight faster; however, the differences were not significant. Water intake was not different between the groups; however, urine volume (average of the 3 days) was reduced in the RGZ-treated rats (Fig. 1B). Similarly, there were no differences in sodium intake; however, urinary sodium excretion (average of 3 days) was reduced by RGZ treatment (Fig. 1C).
Plasma creatinine levels were not significantly different (Table 1). However, creatinine clearance was significantly reduced by RGZ treatment, indicating possibly some fall in glomerular filtration rate. Due to a concomitant fall in creatinine clearance with sodium clearance, the fractional excretion of sodium was not different between the two groups of rats. However, the urinary potassium-to-sodium ratio was increased by RGZ treatment, indicating relatively selective sodium reabsorption in the kidney relative to potassium. No significant differences were found for urinary NOx excretion or the ratio of NOx to creatinine in the urine, although there was some tendency for this to be increased with RGZ. Likewise, final plasma glucose levels were not affected by treatment. Finally, RGZ treatment did not significantly affect the final plasma levels of insulin or aldosterone (Fig. 1D).
Blood Pressure.Fig. 2A shows absolute mean arterial pressure (MAP) in rats treated with either control, low-dose, or high-dose RGZ. These were acquired by averaging all of the blood pressure readings obtained during the entire 24-h period for each rat (144 readings/day/rat). Treatment with either dose of RGZ tended to cause a fall in blood pressure. The MAP was actually lower in the low-dose-treated rats than in their high-dose-treated counterparts. However, baseline MAP was slightly (although not significantly) higher in the high-dose-treated rats. Thus, change in MAP from baseline was also evaluated (Fig. 2B). Here, we found that rats treated with high-dose RGZ and low-dose RGZ had similar decreases in MAP from their corresponding baseline measurements (mean of MAP for 3 days before beginning treatments). Furthermore, plasma renin activity was measured in these rats and not found to be significantly different among the groups (nanograms per milliliter per hour, mean ± S.E.M.): control, 3.58 ± 0.50; low dose, 3.61 ± 1.15; and high dose, 3.08 ± 0.46.
Renal Sodium Transporter/Channel Subunit and eNOS Profile. To address the mechanism of sodium retention with PPAR-γ agonists, we measured the relative protein abundance of the major renal sodium transport proteins. In Fig. 3, we show example immunoblots loaded with whole kidney homogenates from the first set of animals, probed with antibodies against primary sodium transport proteins. RGZ treatment significantly increased the renal abundance of α-1 Na-K-ATPase, NHE3, NaPi-2, and NKCC2 but did not affect the abundance of the distal sodium transport protein, NCC, or any of the subunits of the ENaC. Densitometry revealed an increase in density of the major bands associated with α-1 Na-K-ATPase, NHE3, NaPi-2, and NKCC2 to 182, 150, 130, and 197% of the mean control band densities, respectively.
In Fig. 4, we show the effects of RGZ treatment on the whole kidney abundance of eNOS. RGZ significantly increased eNOS protein in whole kidney to 133% compared with control.
Aquaporins 1 to 3. In Fig. 5, we show immunoblots of whole kidney homogenates probed with antibodies against the water channel proteins, the aquaporins 1 to 3. RGZ caused a marked rise in the protein abundance of AQP2 and AQP3, the collecting duct aquaporins. AQP1 was not significantly affected by treatment in the whole kidney. Finally, we did not evaluate AQP4 abundance because it is difficult to observe this protein clearly on immunoblots of whole kidney, because it is primarily expressed in the inner medulla.
Inner Medullary Water Channels and α-1-Na-K-AT-Pase. In the rats in which blood pressure measurements had been made, we evaluated the renal abundance of the aquaporins 1 to 4 (as well as α-1-Na-K-ATPase) in a homogenate made of the inner medulla. This was done to examine the regulation of aquaporin-4, which cannot be analyzed by immunoblotting in whole kidney homogenates and also to assess whether regulation of aquaporins by rosiglitazone was similar in the medulla as what we had observed for the whole kidney. Figure 6, A–E, shows example immunoblots for all proteins analyzed in the inner medulla. Figure 6F is a bar graph summary of the densitometry for these proteins. RGZ treatment increased AQP1, AQP2, and AQP3 abundance but not AQP4 (both the 52 and 32 kDa are specific for AQP4) nor α-1 Na-K-ATPase.
Discussion
TZDs such as rosiglitazone (Avandia) and other PPAR-γ agonists have great promise for the treatment of insulin resistance and diabetes. These drugs also apparently have anti-inflammatory and lipid-lowering actions that may reduce blood pressure and retard the progression of cardiovascular disease independently of their effects on insulin sensitivity (Takano and Komuro, 2002). Hepatoxicity associated with an earlier TZD, troglitazone, does not seem to be a problem with these more recent PPAR-γ agonists. Nevertheless, one side effect that persists is the tendency for these agents to lead to fluid retention and edema in some patients, by an unknown mechanism.
Our results provide novel evidence of potential molecular mechanisms underlying fluid retention and edema associated with these drugs. Our findings suggest that there is indeed rapid sodium and water retention due to ingestion of rosiglitazone. This increase in sodium and water retention is coupled to a fall in blood pressure. Furthermore, these changes were associated with a rise in the renal abundance of α-1 Na-K-ATPase, NKCC2, NaPi-2, NHE3, the aquaporins 2 and 3, and eNOS.
The blood pressure-lowering effects of PPAR-γ agonists have been reported to be accomplished, at least partially, through direct vascular effects of these agents to induce vasodilation (Komers and Vrana, 1998), for example, by blockade of Ca2+ mobilization from intracellular stores and by inhibition of extracellular calcium uptake via L-channels (Komers and Vrana, 1998). Depending on the severity of the vasodilation, this might be expected to transiently reduce glomerular filtration rate, at least until volume is expanded. Although autoregulation of blood pressure should allow for maintenance of GFR, it is possibly that the same mechanisms responsible for the peripheral vasodilation, perhaps nitric oxide production, does not allow for full renal efferent arteriole constriction necessary to maintain GFR at the same rate. In agreement with this, we saw a relative reduction in creatinine clearance, an index of GFR, with 3-day RGZ treatment. Schnackenberg et al. (2001) also reported a decrease in glomerular filtration rate, as well as positive sodium balance, in rats treated with rosiglitazone. Recently, Idris et al. (2003) reported a reversible increase in endothelial cell permeability to albumin in cultured pulmonary arterial cells treated with RGZ. This might also play a role in the edema seen with RGZ therapy.
Furthermore, we also observed a modest, yet significant increase in the whole kidney abundance of endothelial NOS in RGZ-treated rats. Evidence for a role for eNOS in affecting blood pressure is quite strong (Forte et al., 1997; Kelm, 2003). Disruption of eNOS gene in mice leads to hypertension (Forte et al., 1997). It has also been reported that NO production is reduced in patients with essential hypertension compared with normotensive individuals (Klahr, 2001). Thus, in our study, it is clearly possible that increased NO production as a result of up-regulated eNOS activity might contribute to the decreased MAP. Furthermore, recent studies have demonstrated that NO up-regulates both NHE3 and NKCC2 abundance in the kidneys of volume-expanded rats (Turban et al., 2003), thus providing a candidate mechanism for the increase in these two proteins that we observed with RGZ treatment.
We found an increase in NHE3. NHE3 is the major apical route for sodium reabsorption in the proximal tubule where approximately 60% of the filtered load of sodium is reabsorbed. NHE3 abundance has been shown to be increased by acidosis (Kim et al., 1999). We (Ecelbarger and Welbourne, 2002) have recently shown that RGZ-treated rats excrete more ammonium in their urine and have a lower urine pH. Increased sodium reabsorption via NHE3 in the proximal tubule has been implicated in the hypertension associated with the obese Zucker rat (Hussain et al., 2001) and in the spontaneously hypertensive rat strain (Li et al., 2001).
No significant changes in protein abundance were found for any of the post-macula densa sodium transport proteins, including the ENaC subunits and the thiazide-sensitive NaCl cotransporter. These proteins are key in day-to-day variation in sodium reabsorption and respond to aldosterone and vasopressin (ENaC only) primarily due to changes in sodium intake. Our findings suggest that RGZ-mediated sodium retention, in contrast, is due to up-regulation of primarily sodium transport proteins in the proximal tubule and thick ascending limb.
Expression of PPAR-γ receptors in the kidney, as assessed by reverse transcription-polymerase chain reaction of microdissected renal tubules, is restricted primarily to the inner medullary portion of the collecting duct (Yang et al., 1999). Of the proteins we have evaluated, the aquaporins 2 and 3, ENaC, and Na-K-ATPase are also expressed at this site. Therefore, these increases could be the results of direct effects of RGZ on the PPAR-γ receptor. In contrast, changes in NHE3, NaPi-2, and NKCC2 are likely to be the result of indirect influences of RGZ, e.g., through influencing renal hemodynamics, which may influence sodium transporter expression via unknown molecular mechanisms.
Finally, the relationships between insulin, insulin signaling, and PPAR-γ agonists are not known. Insulin, itself is antinatriuretic (DeFronzo et al., 1975) and can activate ENaC (Blazer-Yost et al., 1998). Thus, the possibility exists that PPAR-γ agonists may potentiate the effects of insulin through its own receptor especially in the collecting duct. Furthermore, some of these changes in activation of sodium transport routes or channels, might not be detected by measuring protein abundance. For example, ENaC has been shown to be regulated by trafficking or changes in intracellular distribution, in addition to changes in abundance (Masilamani et al., 1999).
Overall, our findings provide us with some understanding as to how TZDs such as rosiglitazone may lead in sodium retention and edema. Our results suggest that peripheral vasodilation may result in a transient fall in blood pressure and glomerular filtration rate. The changes may be compensated by increased sodium retention by the kidney through up-regulation of sodium transport proteins. Tackling this problem remains difficult. It is likely that selective blockade of eNOS might preclude some of the protective effects of these drugs. The use of transporter/exchanger antagonists, e.g., furosemide, to block NKCC2-mediated sodium transport in the thick ascending limb or the NHE3-selective S3226 (Schwark et al., 1998) to block sodium hydrogen exchange in the proximal tubule might prove beneficial. However, blood pressure and acid-base status would need to be carefully monitored.
Acknowledgments
We thank Brittany R. Goodenow and Melissa S. Doud for technical assistance with animal care and blotting and Min Shi for animal surgery to implant blood pressure telemetry devices.
Footnotes
-
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants KO1 DK02672 (to C.E. and J.S.) and RO1 DK38094 (to J.V. and M.S.), Georgetown University, the National Kidney Foundation George E. Schreiner, M.D. Young Investigator Award (to C.E.), a Research Award from the American Diabetes Association (to C.E. and J.S.), and the intramural budget of the National Heart, Lung, and Blood Institute (to M.K.).
-
DOI: 10.1124/jpet.103.058008.
-
ABBREVIATIONS: PPAR, peroxisomal proliferator-activated receptor; TZD, thiazolidinedione; NHE3, sodium hydrogen exchanger (type III), NaPi-2, sodium phosphate cotransporter (type II), ENaC, epithelial sodium channel; NCC, Na-Cl cotransporter; NKCC2, Na-K-2Cl cotransporter of the thick ascending limb; RGZ, rosiglitazone; NOx, nitrates plus nitrites; eNOS, endothelial nitric-oxide synthase; MAP, mean arterial blood pressure; GFR, glomerular filtration rate; AQP, aquaporin.
- Received July 31, 2003.
- Accepted October 24, 2003.
- The American Society for Pharmacology and Experimental Therapeutics