Renal oxygenation changes during water loading as evaluated by BOLD MRI: Effect of NOS inhibition




To demonstrate a possible role for endogenous release of nitric oxide in determining the response of water loading on intrarenal oxygenation as evaluated by blood oxygenation level-dependent (BOLD) magnetic resonance imaging (MRI).

Materials and Methods:

Twelve Sprague Dawley rats (weight 344.9 ± 40.6 g) were equally divided into two groups, A and B. Water loading was implemented by continuous infusion of hypotonic saline containing glucose (0.25% NaCl, 0.5% glucose). Rats in group A were subject to water loading alone, while group B rats were dosed with N-nitro-L-arginine methyl ester, (L-NAME) (10.0 mg/kg) prior to water loading. T2*-weighted images of the kidneys were obtained on a Siemens 3T Verio MRI scanner using a multiple gradient recalled echo (mGRE) sequence.


Consistent with previous reports, group A exhibited a significant decrease in medullary R2* during water loading (40.64 ± 1.10 s−1 to 34.68 ± 1.49 s−1, P < 0.05). On the other hand, in group B there was no decrease in R2* during water loading (48.11 ± 2.38 s−1 to 51.06 ± 2.18 s−1). The increased prewater loading R2* is due to the pretreatment with L-NAME (40.82 ± 3.23 s−1 to 48.11 ± 2.38 s−1, P < 0.05).


Our data suggest for the first time a role for endogenous nitric oxide in determining the response of renal medullary oxygenation to water loading. J. Magn. Reson. Imaging 2011;33:898–901. © 2011 Wiley-Liss, Inc.

INTRARENAL OXYGENATION is now well accepted as an important determinant of renal physiology and pathophysiology (1–5). While the gold standard to evaluate intrarenal oxygenation remains microelectrode-based measurements (1, 2), their utility is limited to animal models. Blood oxygenation level-dependent (BOLD) magnetic resonance imaging (MRI) has been shown to be efficacious in evaluating intrarenal oxygenation both in humans (6–9) and rodent models (10–13).

Among the several different pharmacological maneuvers investigated to date (6, 8, 14), water diuresis is one of the most interesting owing to its relative noninvasive nature. While the data to date suggest a role for prostaglandins in determining the response to water loading on BOLD MRI (9, 15), it is also suspected that nitric oxide (NO) may also play a role (16). Because there are currently no approved nitric oxide synthase (NOS) inhibitors available for human use, we performed studies in rats.

Measurements of oxygen consumption in the human kidneys have been pursued for more than half a century. Early measurements depended on the whole organ determinations by evaluating arteriovenous differences in oxygen saturation (17). These studies demonstrated very small difference in arteriovenous oxygen saturation across the kidney. Not surprisingly, the study also demonstrated a minimal change in arteriovenous oxygen saturation following water loading. Later measurements involving invasive microelectrodes in rat kidneys illustrated a significant gradient in tissue oxygenation within the kidney (1, 2). In fact, it was suggested that kidney should be considered to be made up of two organs, the cortex and the medulla (18), based on their significant hemodynamic differences. While the cortex is supplied with blood flow far in excess of its metabolic needs, the medulla receives very little supply. Further, active transport of sodium chloride in the medullary thick ascending limbs is associated with high energy demand. Together, the medullary oxygenation is commonly described as being at hypoxic levels (3). BOLD MRI measurements inherently can differentiate the cortex and medulla and facilitate evaluation of regional changes in oxygenation in response to physiologic and/or pharmacologic maneuvers (6–9, 11–14).

In this study we hypothesized that, similar to prostaglandins, bioavailability of NO is increased during water diuresis. To confirm this, similar to previous studies with cyclooxygenase (COX) inhibition (9, 15, 16), we evaluated the effect of NOS inhibition on subsequent water loading.


The study protocol was approved by the Institutional Animal Care and Use Committee. The rats were purchased from Charles River (Chicago, IL) and housed at the institutional animal care facility starting at least 3 days before the experiments. The rats had free access to food and water throughout the study. Twelve male Sprague Dawley rats (weight 344.9 ± 40.6 g) were divided into two equal groups randomly. Water loading was implemented by continuous infusion of hypotonic saline containing glucose (0.25% NaCl, 0.5% glucose). Rats in group A rats were subject to water loading alone, while group B rats were dosed with N-nitro-L-arginine methyl ester, (L-NAME) (10.0 mg/kg) prior to water loading. There were no significant weight and age differences between groups.

On the day of experiment, rats were anesthetized with ketamine (60–100 mg/kg intraperitoneally [i.p.]) and thiobutabarbital (Inactin, 10 mg/kg i.p.). A catheter (PE-50) was placed in the femoral vein to induce water diuresis by infusion of hypotonic saline using an infusion pump (Genie Plus, Kent Scientific, Litchfield, CT). In Group A, after obtaining six baseline BOLD MR images, animals were subject to water loading by infusion of hypotonic saline containing glucose for 35 minutes with an infusion rate of 2.0 mL/hour. R2* maps were obtained every 3 minutes.

In Group B, after obtaining six baseline images of the kidneys a slow bolus injection of L-NAME was administered through a catheter and BOLD MR images were obtained at intervals of 3 minutes for 45 minutes followed by infusion of hypotonic saline with the same infusion rate as above to induce the water loading.

Imaging Protocol

MRI acquisitions were performed on a 3.0 T Verio (Siemens Medical Solutions, Erlangen, Germany) scanner using a quadrature extremity coil for signal reception. T2*-weighted images of the kidneys were obtained using multiple gradient recalled echo (mGRE) sequences with 12 echoes. Imaging parameters include: TR = 62 msec, TE = 3.1–44.6 msec, no of slices = 1, NEX = 20, FA = 20, slice thickness = 3.0 mm, FOV = 12.0 × 7.5 cm, matrix 256 × 256, bandwidth 260 Hz/voxel.

Data Analysis

R2* (=1/T2*) measurements were performed by drawing circular regions of interest (ROIs), three each on cortex and medulla, on the T2* map generated by the inline processing on the scanner. Typical ROI size was about 20 pixels and were placed based on the cortex or medulla on the anatomical image and copied to the corresponding T2* map. Statistical significance was determined by paired one-tail Student's t-test. Additionally, R2* color maps were generated offline using MatLab (MathWorks, Natick MA). All the maps were scaled from zero to 75 s−1 representing well to poorly oxygenated. An increase in R2* implies a decrease in oxygenation and vice versa.


Figure 1 illustrates an anatomical and the corresponding R2* maps of a representative rat kidney from each group of animals. All the maps were scaled from 0 to 75 s−1 corresponding to the degree of oxygenation from well-saturated (red) to poorly saturated (blue). While water loading decreased medullary R2* in group A, a similar response was absent in group B. The time course of change in R2* in representative animals is illustrated in Fig. 2. Note the clear decrease in medullary R2* in group A during water loading, while a similar response is completely absent in group B. Note that the pre-water load medullary R2* was significantly higher in this group due to the direct effect of L-NAME (14).

Figure 1.

R2* maps of rat kidneys in one representative animal from each of the groups A and B. The top row illustrates images from group A (water load): (a) anatomical, (b) baseline R2* map, (c) R2* map obtained during water loading. The bottom row similarly shows images from group B: (d) anatomical, (e) baseline R2* map, R2* maps obtained (f) post L-NAME, and (g) during water loading. All the maps were scaled from zero (red) to 75 (blue) representing a range of hemoglobin oxygen saturation levels. M indicates outer medulla and C represents the cortex.

Figure 2.

The time course of cortical and medullary variation in R2* during water loading in one representative animal in each of the groups A (top) and B (bottom). Note the significant decrease in medullary R2* values, approaching the cortical values in group A during water loading. Note the absence of a similar decrease in R2* in group B during water loading. Also note the significant increase in R2* in group B prior to water loading illustrating a decline in medullary oxygenation due to the administration of L-NAME. Time zero represents the start of water load in both groups of animals. The error bars show the standard error estimated over three ROIs in each, cortex and medulla. [Color figure can be viewed in the online issue, which is available at]

Figure 3 is the summary plot showing R2* changes during water loading in renal medulla and cortex in the two groups of animals. The medullary R2* in group A decreased significantly during the water diuresis in group A. In group B, a similar response to water loading was completely abolished.

Figure 3.

Summary plots illustrating cortical and medullary R2* in rat kidneys during water loading in each of the group A (top) and B (bottom). Shown are R2* values in cortex and medulla before and after 35 minutes of water loading. Note that only the decrease in medullary R2* in group A reach statistical significance following water load with P < 0.05 by paired one-tail Student's t-test. The error bars represent SD over all animals. [Color figure can be viewed in the online issue, which is available at]


The presence of renal medullary hypoxia and its consequences to various renal diseases is well accepted (3, 5). A key corollary to the medullary hypoxia theory is that it may not be the level of hypoxia itself that is a risk factor, but rather the compromised ability to maintain renal oxygenation status that is a risk factor. Among the many hypothetical control mechanisms the role for prostaglandins and NO are better appreciated (2). The possible role of prostaglandins was previously illustrated in humans using BOLD MRI (9, 15). It was shown that the medullary oxygenation was significantly improved during water loading, but not when pretreated with COX inhibitors such as naproxen (15) and ibuprofen (9). While a similar role for NO was suspected, similar experiments could not be performed due to the lack of approved NOS inhibitors for human use. Water loading in rodent models have not been reported extensively. It was recently shown that it is feasible to implement a water loading paradigm in rats and, similar to humans, an increase in medullary oxygenation during water loading was reported (16).

Our observations in this study are consistent with the previous reports in rat (16) and human kidneys (9, 15) in terms of the response to water loading. In the original studies that introduced renal BOLD MRI (8), data were acquired with both gradient echo (GRE) and spin echo (SE) to show that changes due to water load were only apparent on the GRE and not on the SE. Future studies could incorporate techniques to allow for simultaneous measurement of R2 and R2* allowing for estimation of R2′ (the susceptibility component) similar to the one proposed by Ma et al (19). This could facilitate a more straightforward interpretation of BOLD MRI data, eliminating confounding effects such as a change in water content.

In the current study we show for the first time a possible role for NO in determining the response to water loading on BOLD MRI. As illustrated in Figs. 1–3, pretreatment with L-NAME completely abolishes the response to water load. This is very similar to the previous observations with COX inhibition in humans (9, 15). A key difference is the apparent direct effect of NOS inhibition observed in this study. At the doses of COX inhibitors and the route of administration used in the human studies (9, 15), there was no measurable change in the R2* values prior to water loading. Compared to the oral administration of COX inhibition in humans, L-NAME was administered intravenously and the dose was higher than what has been reported for investigational use in humans (14).

The role of NO in the vascular endothelium is well recognized and it is known that its production and hence bioavailability is impaired in several diseases such as hypertension, diabetes, atherosclerosis, aging, etc (20). Results from the present study suggest that we may observe a reduced response to water load in subjects with endothelial dysfunction. Previous studies indicated this in elderly (9) and in early diabetics (21). Alternately, one could observe a reduced response to L-NAME, as shown previously (11). Endothelial dysfunction may also predispose to acute renal failure, as suggested by previous studies on contrast-induced nephropathy (2, 22). The results may also explain mechanistically why hydration is an effective way to mitigate effects of radiocontrast, ie, stimulating endogenous protective mechanisms such as prostaglandins and NO. Assuming that a compromised ability to upregulate prostaglandins and/or NO is a risk factor for acute renal failure such as that associated with radiocontrast administration, BOLD MRI response to water loading could potentially be used as a way of identifying subjects at such risk.

In conclusion, similar to prostaglandins, NO may also play a key role in determining the renal medullary oxygenation status during water loading. This could further support the recent strategy to combine NO donation with COX-inhibitors for better pain management with reduced nephrotoxic risk associated with COX inhibition (16).