Abstract
The incidence and the mortality of septic acute kidney injury are high, partly because the pathogenesis of sepsis-induced renal dysfunction is not clear. The objective of this study was to investigate the upregulation of renal inducible nitric oxide synthase (iNOS) in human endotoxemia and sepsis and the effect of NO on tubular integrity. Septic patients and endotoxemia that was induced by a bolus injection of 2 ng/kg Escherichia coli LPS in human volunteers were studied. In addition, the effect of co-administration of the selective iNOS inhibitor aminoguanidine was evaluated. The urinary excretion of the cytosolic glutathione-S-transferase-A1 (GSTA1-1) and GSTP1-1, markers for proximal and distal tubule damage, respectively, was determined. In septic patients, an almost 40-fold induction of iNOS mRNA in cells that were isolated from urine was found accompanied by a significant increase in NO metabolites in blood. The mRNA expression of iNOS was induced 34-fold after endotoxin administration. LPS-treated healthy volunteers showed a higher urinary excretion of NO metabolites compared with control subjects. Urinary NO metabolite excretion correlated with urinary GSTA1-1 excretion, indicating proximal tubule damage, whereas no distal tubular damage was observed. Co-administration of aminoguanidine reduced the upregulation of iNOS mRNA, urinary NO metabolite, and GSTA1-1 excretion, indicating that upregulation of iNOS and subsequent NO production may be responsible for renal proximal tubule damage observed.
Septic shock is the leading cause of death in intensive care units, with an overall mortality rate of approximately 50% (1). The incidence rate of severe sepsis lies between 50 and 100 cases per 100,000 individuals in industrialized nations (2). In addition, the incidence of acute kidney injury (AKI) in sepsis and severe sepsis is 19 and 23%, respectively (3), supporting the need for further understanding of the pathophysiology of sepsis-induced renal dysfunction. Apart from renal hypoperfusion as a result of hemodynamic instability of septic patients (4), other mechanisms, such as local effects of cytokines or excessive nitric oxide (NO) release, may play a role in the development of renal failure.
Sepsis is characterized by the release of proinflammatory cytokines, such as TNF-α and IL-1β, which are associated with endothelial and tissue injury seen in AKI (5). Besides systemic cytokine production, kidney glomerular mesangial cells produce TNF-α after LPS exposure (6). Different animal sepsis models showed a decrease in both mortality and the incidence of renal failure with anti-TNF therapies (7–9). Unfortunately, anticytokine therapies failed to show a survival benefit in patients with sepsis (10,11).
During the past decade, NO has been considered an important contributory factor to the pathogenesis of septic shock. Various inflammatory stimuli, such as LPS and proinflammatory cytokines, stimulate the endogenous NO production by activation of an inducible NO synthase (iNOS) (12). Among various other effects, excessive production of NO causes systemic vasodilation, which in turn leads to renal vasoconstriction with sodium and water retention (3). Furthermore, a local induction of iNOS, which is constitutively expressed in the kidney, in particular in the medulla and proximal tubules, may be the cause of peroxynitrite-related tubular injury as a result of local formation of reactive oxygen and nitrogen species during systemic inflammation (13).
In contrast to the vast amount of in vitro and in vivo animal experiments concerning the role of iNOS during sepsis, there is far less evidence of increased iNOS induction during systemic inflammation and sepsis in humans. To the best of our knowledge, nonselective iNOS inhibitors have been used exclusively in septic patients (14), and, until now, no human clinical studies with selective inhibitors are available.
Our study was designed to investigate whether systemic inflammation during sepsis and experimental human endotoxemia result in an upregulation of renal iNOS expression. Moreover, the effect of NO on tubular integrity was studied. For this, we examined the urinary excretion of the cytosolic enzymes glutathione-S-transferases (GST) that are present in the proximal tubule (GSTA1-1) and distal tubule (GSTP1-1) as biomarkers for AKI (15). In addition, we examined the effect of the selective iNOS inhibitor aminoguanidine to determine whether the damage to the proximal tubules is caused directly by an overproduction of NO or indirectly by circulating cytokines.
Materials and Methods
Septic Patients and Human Endotoxemia
The ethical committee of the Radboud University Medical Centre Nijmegen approved the studies, and informed consent was obtained from legal representatives of the patients and from all 33 healthy volunteers. Patients who were admitted to the intensive care unit entered the study when they were between 18 and 80 yr of age; had a proven or suspected infection; had two of four systemic inflammatory response syndrome (SIRS) criteria (16) that existed for <24 h; and had acute onset of end-organ dysfunction in the preceding 12 h, including sustained hypotension that required vasopressor therapy. We excluded patients with an Acute Physiology and Chronic Health Evaluation II (APACHE II [17]) >28 or expected survival of <24 h, known confirmed Gram-positive or fungal sepsis, or chronic renal failure that required hemodialysis or peritoneal dialysis or those who were treated with immunosuppressants or high doses of glucocorticosteroids (equivalent to prednisone >1 mg/kg per d). Throughout the study, all patients received standard conventional therapy, and antibiotic aminoglycosides were not used as antibiotic treatment. Arterial blood and catheterized urine were collected at several time points between 0 and 24 h after inclusion.
Study design of the human endotoxemia experiments and healthy volunteers have been described in detail (18). The healthy volunteers were asked to fill in a questionnaire of the dietary intake of nitrate for 24 h before and during the experiment. A single intravenous bolus injection of 2 ng/kg body wt LPS (E. coli O:113; United States Pharmacopeia Convention, Rockville, MD) was given to eight healthy volunteers. Eleven healthy volunteers served as control subjects and received the vehicle (0.9% NaCl) only (protocol 1). In protocol 2, 14 healthy volunteers received 2 ng/kg body wt endotoxin, seven of whom additionally received an intravenous loading dose of 5 mM aminoguanidine (Clinalfa, Laufelfingen, Switzerland) after 1 h followed by a maintenance dose of 1.5 mM/h during 4 h. The calculated dose was based on the pharmacokinetics of aminoguanidine, described by Foote et al. (19), and taking the IC50 of aminoguanidine (31 μM, ([20]) into consideration. In contrast to protocol 1, healthy volunteers received an intravenous infusion of a glucose/saline solution of 1500 ml before LPS treatment and 150 ml/h during the experiment. This change in fluid management was introduced after two patients had developed a severe vagal response in other studies (21). There were no significant differences among the four experimental groups with respect to age, gender, body mass index, and vital signs (Table 1).
Demographic characteristics of the septic patients and healthy volunteersa
Arterial blood and urine were collected at several time points. To prevent bacterial growth and RNA degradation, we washed urine containers with 2% chlorhexidine (Genfarma, Zaandam, The Netherlands) and 0.5 M NaOH, respectively. Urine volumes were recorded, and samples for the determination of NO metabolites were frozen at −80°C until assayed. C-reactive protein (CRP) and leukocyte counts were determined by routine clinical chemistry. The concentration of different cytokines in blood was measured using simultaneous Luminex Assay (R&D Systems, Minneapolis, MN), as described previously (22).
Chemical Assays
The total amount of the stable NO metabolites, nitrate and nitrite, were determined as a measure for the production of NO radical, using the Griess reaction, according to Moshage et al. (23). Heparinized plasma and urine samples were diluted four-fold and 40-fold with distilled water, respectively. The amounts of GSTA1-1 and GSTP1-1 in urine were determined to differentiate between proximal and distal tubular cell injury (24). GSTA1-1 and GSTP1-1 were assayed by ELISA as described previously (25,26). Each sample was determined in triplicate.
Determination of iNOS mRNA Expression
Urine samples were centrifuged at 700 × g for 10 min at 4°C. RNA was isolated from the cells in the pellet by using 1 μg of polyA-carrier RNA and TRIzol reagent (Invitrogen, Breda, The Netherlands) according to the manufacturer’s instructions. Total RNA was reverse-transcribed into cDNA using Moloney murine leukemia virus reverse transcriptase and a pd(N)6 sodium salt primer (Invitrogen). Subsequently, quantitative real-time PCR (RQ-PCR) on cDNA was performed according to the TaqMan protocol in optical tubes (Applied Biosystems, Zwijndrecht, The Netherlands). Human iNOS and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were amplified with a predeveloped Gene Expression Assay provided by Applied Biosystems (iNOS; Hs00167248_m1, GAPDH; Hs99999905_m1). For this assay, 12.5 μl of PCR master mix, 5 μl of cDNA sample, 18 μM for each primer, and 5 μM TaqMan with a FAM reporter dye at the 5′ end and a nonfluorescent quencher at the 3′ end were used. The amplification was performed after an initial warm-up phase of 2 min at 50°C for optimal PCR Master Mix activity and 10 min at 95°C, which served as denaturing step. Forty amplification cycles were completed at 95°C for 15 s and 60°C for 1 min. Finally, the cDNA was subjected to RQ-PCR using the ABI PRISM 7700 single reporter Sequence Detection System (Applied Biosystems). All experiments were performed in triplicate. Sample quantities were normalized to the expression of the housekeeping gene, GAPDH.
Determination of Kidney Function
Creatinine, urea, and sodium were determined in blood and urine by routine clinical chemistry. The amount and the pattern of proteins in urine samples were determined by gel electrophoresis. Samples were diluted to an equal amount of total creatinine, fractionated, and loaded on a 10% SDS polyacrylamide gel. Various amounts of BSA were used to determine the amount of protein semiquantitatively. After electrophoresis, the gels were rinsed three times for 5 min with ultrapure water and subsequently stained for 1 h with Gelcode blue stain reagent (Pierce, Rockford, IL). Finally, for enhancement of stain sensitivity, the gels were destained for at least 1 h with ultrapure water.
Western Blot Analysis
Western blotting was performed to determine the expression of the proximal tubular protein aquaporin-1 (27) in urine samples. Protein was isolated from centrifuged urine (700 × g, 10 min, 4°C) and homogenized in a buffer (pH 7.4) that contained 10 mM Tris-HCl, 250 mM sucrose, 1 mM PMSF, 1 μg/ml aprotinin, 5 μg/ml leupeptin, 1 μg/ml pepstatin, and 10 μM E64. Protein concentration was determined with the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA) using BSA as a standard. Samples were separated on a 12% SDS polyacrylamide gel and transferred to Hybond-C pure nitrocellulose membrane (Amersham, Buckinghamshire, UK). For detection of aquaporin-1, 1:200 diluted mAb (provided by Dr. P.M.T. Deen, Department of Physiology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands) was used.
Immunocytochemistry
Lectin and megalin staining was used to identify proximal tubule cells in urine samples. The cytopreparation of the urinary sediments and immunocytochemistry were determined as described previously (28). The combined binding of the lectins Erythrina cristagalli agglutinin (FITC coupled) and Sophora japonica agglutinin (rhodamine coupled; Vector Laboratories, Burlingame, CA) indicates that cells are of proximal tubule origin. DNA was stained with 4′6-diamidino-2-phenylindole (Sigma, Zwijndrecht, The Netherlands). For detection of megalin, an endocytic receptor that is highly expressed on the proximal tubule (29), 1:50 diluted mouse mAb 1H2 (provided by Dr. Argraves, Medical University of South Carolina, Charleston, SC [30]), was used.
Statistical Analyses
Values are given as mean ± SE. Data were analyzed using GraphPad Prism 4.0 for Windows (GraphPad Software Inc., San Diego, CA). Changes of parameters in time were tested using one-way ANOVA with Bonferroni correction. For iNOS mRNA expression, a Mann-Whitney test was used. Differences between experimental groups were tested by ANOVA for repeated measures using SPSS 12.0.1 for Windows (SPSS Inc., Chicago, IL). A two-sided P < 0.05 was considered significant.
Results
Septic Patients
Ten patients with septic shock (Table 1) entered the study during an 8-mo period. The APACHE II score of the septic patients was 20.2 ± 1.4. Six patients had a Gram-negative bacterial infection, most frequently E. coli. Kidney function of all patients was impaired as indicated by an average creatinine clearance of 62 ± 14 ml/min and a urinary protein excretion of 430 ± 110 mg/d. Two patients had a fractional sodium excretion of >2% (2.11 ± 0.01), indicating acute tubular necrosis, and eight patients had a fractional sodium excretion of <1% (0.15 ± 0.03), indicating impaired renal perfusion.
The concentration of NO metabolites in plasma was increased two-fold, as shown in Figure 1A. The amount of NO metabolites in urine also was increased but not significantly different from that of control subjects, most likely as a result of the severely impaired renal function (Figure 1B). In addition, the mRNA expression of iNOS in cells that were isolated from urine was increased almost 40-fold (Figure 1C) in septic patients compared with healthy volunteers. The relative expression of iNOS mRNA in control healthy volunteers was normalized for the average cycle threshold (CT) value of the housekeeping gene, GAPDH (CT = 23.8 ± 0.3, ΔCT = 12.0 ± 0.1), and set to 1.
Nitric oxide (NO) metabolites in plasma and urine and inducible NO synthase (iNOS) expression in sepsis. NO metabolite levels (NOx) in plasma (A) and urine (B) and iNOS mRNA expression (C) are given for control healthy volunteers (□; n = 11) and septic patients (▪; n = 10 for A and B, n = 6 for C). Each parameter was determined three times at various time points between 0 to 24 h after inclusion for each septic patient. Urinary excretion of NOx was corrected for creatinine excretion. The relative expression of iNOS mRNA in control healthy volunteers was normalized for the average cycle threshold (CT) value of the housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (CT = 23.8 ± 0.3, ΔCT = 12.0 ± 0.1), and set to 1. Data are expressed as mean ± SEM (**significantly different compared with the controls, P < 0.01).
Human Endotoxemia
Demographic Data and Signs of Inflammation.
All healthy volunteers in the LPS groups experienced the expected influenza-like symptoms, such as headache, nausea, and chills, starting 60 to 120 min after the endotoxin challenge. These symptoms were only mild, and all volunteers were symptom-free within 8 h after LPS administration. As shown in Table 2 for protocol 1, maximum body temperature was significantly elevated after 4 h and CRP increased from ≤5 mg/L at baseline and reached its maximum at t = 24 h. Furthermore, administration of LPS resulted in the expected increase of pro- and anti-inflammatory cytokines (TNF-α and IL-10, respectively) and leucopenia (t = 1 h) followed by leukocytosis (t = 12 h). In control healthy volunteers, no changes in temperature, CRP, cytokines, and white blood cells were observed (Table 2). LPS administration altered baseline hemodynamics. Mean heart rate increased, and the mean arterial pressure decreased significantly at 5 h compared with the control group (Table 2).
Inflammatory and hemodynamic response to LPSa
There were no significant differences in signs of inflammation between the healthy volunteers who received LPS in the absence or presence of aminoguanidine (see Table 2, protocol 2). Nevertheless, healthy volunteers from protocol 2 had significantly lower peak levels for both body temperature and proinflammatory cytokines compared with the LPS healthy volunteers from protocol 1. However, this effect cannot be explained by dilution because we demonstrated before that the LPS-induced proinflammatory response is attenuated, whereas the IL-10 release demonstrated a trend to be increased (31).
Amount of NO Metabolites Increases in Urine after LPS.
NO metabolite levels (NOx) in plasma and urine were measured at various time intervals after LPS treatment. Blood samples show a slight, although NS, increase over the period of 3 to 12 h after the administration of endotoxin (Figure 2A). Co-administration of aminoguanidine resulted in a decline in NOx plasma concentration (NS; Figure 2B), compared with the LPS group. Conversely, the cumulative amount of NOx in urine increased significantly during the period 3 to 24 h (Figure 3A), which was partially prevented by aminoguanidine (Figure 3B). Finally, there were no significant differences in both plasma and urinary NOx levels between the two protocols for the LPS healthy volunteers.
NO metabolites in plasma. NOx levels in plasma were measured at various times after LPS treatment. NOx levels in control subjects (□; n = 11) and after LPS treatment (▪; n = 8) are given for protocol 1 (A) and after LPS (▪; n = 7) together with aminoguanidine (□ ; n = 7) for protocol 2 (B). Blood samples did not show a significant change in NOx concentration after the administration of endotoxin compared with control subjects and after the administration of aminoguanidine. Data are expressed as mean ± SEM.
NO metabolites in urine. NOx levels in urine were measured at various times after LPS treatment. NOx levels in controls (▪; n = 11) and after LPS treatment (▵; n = 8) are given for protocol 1 (A) and after LPS (▵; n = 7) together with aminoguanidine (•; n = 7) for protocol 2 (B). The cumulative amount of NO metabolites in urine was significantly more compared with the control group during the period 3 to 24 h after LPS treatment. Aminoguanidine partially prevented the increase in NOx excretion by LPS. Data are expressed as mean ± SEM and analyzed by ANOVA with repeated measures over the complete curve. Significantly different compared with the control group of protocol 1, **P < 0.01, or compared with the LPS group of protocol 2, ##P < 0.01.
iNOS mRNA Expression in Isolated Cells of the Urine Is Enhanced after LPS Administration.
RQ-PCR was used to determine the levels of iNOS mRNA in cells that were isolated from urine samples. Overnight urine samples until 24 h after LPS administration were used for RQ-PCR. The relative expression of iNOS in control healthy volunteers (see the Septic Patients section) were normalized for GAPDH expression and set to 1. Analysis of three separate experiments revealed a strong induction (34 ± 15; Figure 4) in iNOS levels after LPS treatment compared with the control subjects. Treatment with both LPS and aminoguanidine reversed this induction in iNOS mRNA but not completely (induction fold 6 ± 2; Figure 4) compared with LPS alone.
Induction of iNOS mRNA. iNOS mRNA expression 24 h after LPS treatment in isolated cells of urine samples was determined with quantitative real-time PCR without (▪; n = 4) or with co-administration of aminoguanidine (□ ; n = 4). ΔCT values for the control subjects, as presented in Figure 1, were set to 1. Data are expressed as mean ± SEM. Significantly different compared with the control group, *P < 0.05, or compared with the LPS group, #P < 0.05.
Kidney Function Is Unchanged but Urinary Protein Excretion Increases after LPS.
Kidney function was not altered after LPS administration; creatinine and urea clearance and the fraction sodium excreted did not differ between the two groups in both protocols. In addition, the total volume of urine that was excreted during 24 h was not different for the control and LPS healthy volunteers in protocol 1, compared with basal levels (Table 3). As expected, the total urine volume produced increased in the healthy volunteers from the prehydration experimental protocol 2.
Kidney functiona
The amount and the pattern of excreted proteins were different after LPS treatment, compared with the control subjects (Figure 5A). Twelve and 24 h after LPS administration, low molecular weight proteins clearly were visible in urine samples, which may indicate proximal tubule damage. These proteins were not found in control samples. Co-administration of aminoguanidine resulted in a decrease in excreted proteins after 6 and 12 h compared with the LPS group from protocol 2 (Figure 5B).
Protein pattern of urine samples. (A and B) Representative images of urine samples separated on a 10% polyacrylamide gel. Samples were diluted to an equal amount of total creatinine. Lanes 1 and 2 show different amounts of BSA, 1.3 and 0.65 μg, respectively. (A) Lanes 3 through 5 show urine samples of one participant 6, 12, and 24 h after LPS treatment (protocol 1). Urine samples from the same time points of a control subject are visible in lanes 6 through 8. Lane 9 shows a urine sample of a septic patient. After LPS administration, low molecular weight proteins (LMWP) were visible in urine samples. (B) Lanes 3 through 5 show urine samples of the same time points after treatment with both LPS and aminoguanidine. Lanes 6 and 7 show urine samples of a prehydrated participant who was treated with LPS in the absence of aminoguanidine (protocol 2). Lane 9 shows a urine sample of another septic patient. (C and D) Representative Western blots of aquaporin-1 present in isolated cells from urine samples 12 h after LPS treatment. (C, protocol 1) After LPS administration, aquaporin-1 is upregulated, compared with control urine. (D, protocol 2) Aminoguanidine prevents this upregulation. (E and F) Identification of proximal tubule cells in urinary sediments of both healthy volunteers after LPS treatment and of septic patients (data not shown) by lectins (E) and megalin antibody (F). Proximal tubule cells were positive for both Erythrina cristagalli agglutinin (FITC coupled) and Sophora japonica agglutinin (rhodamine coupled) and stain positive for megalin. DNA was stained with 4′6-diamidino-2-phenylindole.
Western blot analysis showed an increase in aquaporin-1 in urine of LPS-treated healthy volunteers, compared with control subjects (Figure 5C), which was prevented by aminoguanidine (Figure 5D).
Immunocytochemistry showed the presence of proximal tubule cells in urinary sediments of both healthy volunteers (Figure 5) and septic patients (data not shown). The number of total cells and proximal tubule cells increased 24 h after LPS treatment, compared with baseline urine. Besides cells that were positive for both Sophora japonica agglutinin and Erythrina cristagalli agglutinin (Figure 5E, yellow), positive cells for megalin (Figure 5F) were visible in these urine samples.
Renal Proximal Tubule Injury after LPS.
GSTA1-1 and GSTP1-1 levels in urine were measured at various times after LPS treatment to differentiate between proximal and distal tubular cell injury. Volunteers who received LPS had increased urinary excretion of GSTA1-1 between 6 and 12 h (Figure 6A) compared with the control subjects. Increased cumulative urinary excretion of GSTA1-1 was significantly prevented by aminoguanidine. In contrast to GSTΑ1-1, the cumulative urinary excretion of GSTP1-1 was not affected by LPS alone or in combination with aminoguanidine (Figure 6B).
Excretion of glutathione-S-transferases (GST) in urine. Levels of GST that are present in the proximal tubule (GSTA1-1) and distal tubule (GSTP1-1) in urine were measured at various times for control subjects (▪; n = 11), after LPS treatment (▵; n = 8) and after LPS together with aminoguanidine (•; n = 7). (A) As compared with the control subjects, the volunteers who received LPS had increased urinary excretion of GSTA1-1 between 6 and 12 h after administration. Aminoguanidine prevents this upregulation in urinary GSTA1-1 excretion in part. (B) Urinary excretion of GSTP1-1 remained unchanged for the three groups. Data are expressed as mean ± SEM (significantly different compared with the control group, **P < 0.01, or compared with the LPS group, ##P < 0.01). (C) Correlation between the cumulative urinary excretion of NOx and GSTA1-1 24 h after treatment for control subjects (▪; n = 11) and after LPS (▵; n = 8). A significant correlation between these two parameters was observed in the LPS group (rS = 0.6688, P = 0.0131) but not for the control group (rS = 0.0608, P = 0.4922).
The cumulative urinary excretion of NOx metabolites and GSTA1-1 strongly correlated with each other 24 h after LPS administration (rs = 0.67, P = 0.013; Figure 6C), whereas no correlation was visible for control healthy volunteers (rs = 0.06, P = 0.492; Figure 6C). Further analysis revealed a lack of correlation between the cumulative urinary excretion of NOx and GSTP1-1 for both LPS and control subjects (data not shown). In addition, circulating NOx and cytokines were not correlated to the cumulative urinary excretion of GSTA1-1 or GSTP1-1 (data not shown).
Discussion
In this study, we investigated the effects of human experimental endotoxemia and sepsis on renal iNOS expression and tubular integrity. These data are the first to describe an activation of renal iNOS in humans in vivo during septic shock and experimental endotoxemia. We found an increased iNOS mRNA expression in cells that were isolated from urine of both septic patients and healthy volunteers who were treated with endotoxin. Until now, expression of iNOS in humans has been shown only in alveolar macrophages from patients with acute respiratory distress syndrome (32) and in mesenteric arterial smooth muscle and peripheral blood mononuclear cells (33) after sepsis. Another recent study did not detect iNOS in cells or vessels of the systemic circulation in patients with septic shock that was caused by cellulitis but only at the site of infection (34). Besides an induction of iNOS mRNA, we found an increase in plasma NO metabolites in septic patients. In contrast, the amount of NO metabolites in urine of septic patients was not different from that of control subjects, which may be affected largely by the impaired renal function of the septic patients.
Our healthy volunteers did not show a significant increase in NO metabolites in blood samples after administration of LPS. This finding is in agreement with previous human endotoxemia studies (35,36). Increased NO production in our study with septic patients and previous studies (37,38) may be explained by a much higher degree of inflammation, ongoing LPS release, and impaired renal clearance of NO metabolites during septic shock compared with the mild inflammatory response of a single bolus injection of LPS that does not severely impair renal function during experimental endotoxemia.
Contrary to the lack of a significant increase in circulating NO metabolites, LPS treatment increased the cumulative amount of NO metabolites in urine significantly within 3 to 24 h. Renal NO production was associated with an increased urinary excretion of GSTΑ1-1, whereas the excretion of GSTP1-1 was not influenced, indicating NO-mediated proximal tubular damage. Because iNOS is expressed predominantly and constitutively in the proximal tubules, we propose that a local upregulation of renal iNOS, resulting in an increase in NO radicals, is responsible for the proximal tubular injury observed. The correlation of the cumulative amount of urinary excreted NO metabolites with GSTΑ1-1 excretion and lack of correlation of urinary NO metabolites with GSTP1-1 excretion support this pathophysiologic hypothesis. This hypothesis is in agreement with a previous in vitro study in which a direct toxic effect of NO to proximal tubules was demonstrated (39). Moreover, the lack of correlation between circulating cytokine levels and urinary GST excretion suggests that systemic cytokines are not responsible for the tubular injury observed. Whether other urinary markers for AKI, such as kidney injury molecule 1 (KIM-1) or neutrophil gelatinase–associated lipocalin, are present in urine samples will be defined in the near future.
We co-administered the selective iNOS inhibitor aminoguanidine with LPS to examine whether the damage to the proximal tubule cells is caused directly by an overproduction of NO or indirectly by the cytokines released. Although there are more selective iNOS inhibitors, such as L-N6-iminoethyl-lysine and 1400W (20), we selected aminoguanidine because of its previous use in other human studies (40,41). Up to now, only nonspecific inhibitors have been used for examining the role of iNOS during inflammation in humans. A randomized, double-blind, placebo-controlled phase III clinical study of patients with septic shock detected an increase in overall mortality associated with treatment using the nonselective NOS inhibitor l-arginine analogue N(G)-monomethyl-l-arginine (L-NMMA [14]). This detrimental effect could be explained by the fact that L-NMMA blocks both iNOS and the physiologically important endothelial NOS and by the high dosage used in that clinical trial. Different animal studies have demonstrated that selective iNOS inhibition attenuates the sepsis-induced renal dysfunction (42–44). Moreover, in a septic rat model, selective iNOS inhibition was associated with improved survival, whereas an increase in mortality was found during nonselective NOS inhibition (45).
In our study, aminoguanidine significantly inhibited the LPS-induced increase in renal iNOS mRNA expression and urinary NOx excretion. Apparently, apart from inactivation of iNOS through covalent modification of the iNOS protein (46), iNOS mRNA expression also was suppressed by aminoguanidine, an effect that has been described before (47). In addition, aminoguanidine had no effect on the systemic inflammatory response, as symptoms such as fever and cytokine levels were not significantly reduced compared with the LPS-treated healthy volunteers who did not receive aminoguanidine.
Our finding that circulating NOx was not significantly increased during human endotoxemia, whereas the urinary NOx excretion was, suggests that NO is produced mainly in the kidney or in the urinary tract. Different in vitro and in vivo studies have demonstrated that iNOS is present in the bladder as well. Furthermore, iNOS is upregulated in human bladder smooth muscle cells after treatment with LPS (48). In addition, Olsson et al. (49) showed an induction of iNOS mRNA in both the rat bladder and kidney after intraperitoneal injections of LPS. During the human endotoxemia studies, the volunteers were allowed to void, suggesting that the urine remained in the bladder for a certain period of time. This is in contrast with the catheterized urine sampling of septic patients. Therefore, one may argue that the presence of epithelial cells of the bladder in our urine samples may have influenced our findings. Nevertheless, we have reasons to assume that they were mainly of proximal tubular origin. First, we showed the presence of proximal tubule cells in urinary sediment with immunochemistry. Second, we showed that urinary NO metabolites correlated significantly with the urinary excretion of GSTΑ1-1, and it was shown that the levels of this enzyme were below the detection limit in 92% of normal bladder mucosa samples (50). Third, the presence of the protein aquaporin-1, which is located exclusively in proximal tubules (27), was increased after LPS treatment. Circulating immune cells also may contribute to the urinary cell constitution, but we did not find an increased expression of CD14 (marker of monocytes and macrophages [51]) mRNA in cells that were isolated from septic urine as compared with that from the control subjects (data not shown).
Apart from inhibition of iNOS and the release of NO metabolites, aminoguanidine reversed both urinary GSTA1-1 excretion and the excretion of low molecular weight proteins and aquaporin-1, suggesting that selective inhibition of renal iNOS has potential for the treatment of septic acute renal failure.
Conclusion
Induction of renal iNOS during human endotoxemia and sepsis is associated with proximal tubule injury. Furthermore, during human endotoxemia, proximal tubule injury could be prevented through selective inhibition of iNOS.
Acknowledgments
S.H. was supported by a grant from The Netherlands Organisation for Scientific Research. P.P. is a recipient of a Clinical Fellowship grant of The Netherlands Organisation for Scientific Research (ZonMw).
We thank L.T. van Eijk and M.J. Dorresteijn, medical students, for help during the human endotoxemia studies and our research nurses for help during the sepsis study. Furthermore, we thank H.M.J. Roelofs for help with the GST determinations and M.J.G. Wilmer for help with the lectin staining procedures.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received February 9, 2006.
- Accepted April 24, 2006.
- Copyright © 2006 by the American Society of Nephrology