Mapping single-nephron filtration in the isolated, perfused rat kidney using magnetic resonance imaging

Abstract

The kidney has an extraordinary ability to maintain glomerular filtration despite natural fluctuations in blood pressure and nephron loss. This is partly due to local coordination between single-nephron filtration and vascular perfusion. An improved understanding of the three-dimensional (3-D) functional coordination between nephrons and the vasculature may provide a new perspective of the heterogeneity of kidney function and could inform targeted therapies and timed interventions to slow or prevent the progression of kidney disease. Here, we developed magnetic resonance imaging (MRI) tools to visualize single-nephron function in 3-D throughout the isolated perfused rat kidney. We used an intravenous slow perfusion of a glomerulus-targeted imaging tracer [cationized ferritin (CF)] to map macromolecular dynamics and to identify glomeruli in 3-D, followed by a bolus of a freely filtered tracer (gadolinium diethylenetriamine penta-acetic acid) to map filtration kinetics. There was a wide intrakidney distribution of CF binding rates and estimated single-nephron glomerular filtration rate (eSNGFR) between nephrons. eSNGFR and CF uptake rates did not vary significantly by distance from the kidney surface. eSNGFR varied from ∼10 to ∼100 nL/min throughout the kidney. Whole single-kidney GFR was similar across all kidneys, despite differences in the distributions eSNGFR of and glomerular number, indicating a robust adaptive regulation of individual nephrons to maintain constant single-kidney GFR in the presence of a natural variation in nephron number. This work provides a framework for future studies of single-nephron function in the whole isolated perfused kidney and experiments of single-nephron function in vivo using MRI.

NEW & NOTEWORTHY We report MRI tools to measure and map single-nephron function in the isolated, perfused rat kidney. We used imaging tracers to identify nephrons throughout the kidney and to measure the delivery and filtration of the tracers at the location of the glomeruli. With this technique, we directly measured physiological parameters including estimated single-nephron glomerular filtration rate throughout the kidney. This work provides a foundation for new studies to simultaneously map the function of large numbers of nephrons.

INTRODUCTION

Kidney function relies on local and systemic control of vascular perfusion and tubular reabsorption to regulate filtration, osmotic pressure, and blood pressure. The kidney has an extraordinary ability to maintain glomerular filtration despite natural fluctuations in blood pressure or loss of nephrons (1). Kidney function is regulated in part by the relationships between local single-nephron function and vascular perfusion. Single-nephron function has been primarily examined through studies using micropuncture (2, 3) and intravital microscopy (4–7). Both have provided much new understanding of individual nephron function in small animals. However, the kidney is spatially heterogenous, and the coordination of nephrons is still poorly understood, despite evidence of functional connectivity (8).

The importance of directly measuring single-nephron function has been highlighted by a recent study showing population-level differences in estimated average single-nephron glomerular filtration rates (SNGFRs) in humans (9). A low nephron number is likely a significant risk factor for lifetime risk of kidney disease (10), and a reduced nephron number is likely an indicator of pathology in the kidney (11). An improved understanding of the functional coordination between nephrons and the vasculature may provide a new basis for targeted therapies and timed interventions to slow or prevent the progression of chronic kidney disease.

Magnetic resonance imaging (MRI) techniques using a glomerulus-targeted tracer, cationic ferritin (CF), have been developed to map individual glomeruli in animal models and in human kidneys (12–17). CF-enhanced MRI (CFE-MRI) can be used to map functioning, CF-labeled glomeruli throughout the kidney in three dimensions (3-D), ex vivo and in vivo (18, 19). Here, we build on this technology to investigate the function of individual nephrons in the isolated, perfused rat kidney, as a research tool. This is a critical step toward the direct measurement of single-nephron function in vivo. Advantages of the model system of the isolated, perfused kidney include the ability to control physiological parameters such as perfusion rates during the experiment, while avoiding confounding motion, recirculation of tracers, or systemic contributions to regulation of arterial pressures and flow typically encountered in vivo (20–23). MRI is advantageous because it provides 3-D maps of the organ at high resolution without ionizing radiation. It also provides flexible contrast, allowing for investigations of different physiological and soft-tissue features within a single study.

In this work, we acquired time series of the MRI signal across the kidney after injection of a glomerulus-targeted imaging tracer (CF) to map macromolecular binding and to identify glomeruli in 3-D, followed by a bolus of a freely filtered tracer [gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA)] to map filtration kinetics. Time series from MRI during the Gd-DTPA bolus are consistent with the timing of glomerular filtration, concentration at the distal convoluted tubule, and urinary excretion (24, 25). Single-nephron function has been previously detected using injectable tracers in vivo in small regions of the cortex (26, 27). Measurements of single-nephron function could eventually be directly colocalized to anatomic structures such as the vascular networks (7, 28). This work provides a tool to examine the function of individual nephrons throughout the kidney.

METHODS

Radiofrequency Coil Assembly and Custom Kidney Holder

We built an MRI-compatible frame for the isolated, perfused rat kidney to minimize motion of the kidney during imaging, support the radiofrequency (RF) coil, secure the perfusion tubing, and continuously remove perfusate from the renal vein and ureter. We constructed a custom, multiloop transmit/receive surface RF coil (19, 29) with 22-mm-diameter loops using 14-gauge magnet-grade wire (MR200, Remington Industries). Tuning and matching were performed using variable capacitors (Johanson Technology), and power balance was achieved using a fixed capacitor. We used a 50-Ω RG400 double-shielded coaxial cable to connect the coil to the MRI system.

Kidney Preparation

All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Virginia. Male Sprague-Dawley rats (n = 4) were anesthetized with ketamine and xylazine at 100 mg/kg and 11.4 mg/kg, respectively. We adapted the technique of Czogalla et al. (23) to isolate the right kidney for perfusion. Ligatures were placed on the cranial and caudal sides of the right renal artery. Additional ligatures were placed on the left renal artery, superior mesenteric artery, and caudal side of the renal vein. The aorta was cannulated with the cannula end set at the junction of the right renal artery and aorta. A ligature was tied over the cannula to prevent back flow, and the cannula hub was secured by gluing it to the artery. Immediately after the cannula was implanted, the kidney was perfused with 50 mL of Lifor organ preservation solution (Detraxi) at 3 mL/min. Lifor is a commercial organ preservative solution (19), similar to other clinically used solutions (30). The cannula was temporarily capped to prevent back flow. We then isolated the kidney with the attached cannula and submerged it in Lifor solution in a closed container and placed the container on ice for transport to the MRI.

Perfusion Apparatus

The kidney was placed in the custom imaging frame with 1% (wt/vol) agar in 0.9% NaCl surrounding the kidney to minimize motion and limit dehydration throughout all experiments. The coil was placed around the kidney in the custom frame (Fig. 1A). An MRI-compatible temperature probe (Small Animal Instrument) was placed in the agar to continuously monitor temperature. The temperature was maintained at 37°C using a controllable halogen lamp. We perfused the kidney with physiological medium containing Krebs–Ringer solution with fraction V BSA (5.5 g/100 mL), select amino acids, and 95% O₂-5% CO₂ by bubbling to mimic physiological conditions (21, 22, 31, 32). We used a single-channel peristaltic pump (Watson–Marlow) with thermoplastic tubing for perfusion. We performed this initial study with a constant flow design to carefully control the perfusion rate of the contrast agents. The perfusion rate was maintained at 6 mL/min (22, 33) without recirculation. Line pressure was measured following the peristaltic pump at the beginning and the end of each experiment using a research-grade blood pressure transducer (Harvard Apparatus).

Figure 1.

Apparatus, procedure, and typical experiments using magnetic resonance (MR) imaging (MRI) to measure single-nephron function in the isolated perfused rat kidney. A: illustration of a custom MRI-compatible frame for the kidney, radiofrequency (RF) coil, and attached perfusion tubing. B: timeline and description of contrast agent injections during the experiments. Cationized ferritin (CF) and gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA) were injected at separate times. C: three-dimensional (3-D) visualization from MRI of a perfused kidney and representative image slices. MR images before and after CF infusion exhibited a slow appearance of dark spots due to the accumulation of CF in glomeruli. Ex vivo MRI confirmed the specific binding of CF to kidney glomeruli. A bolus of Gd-DTPA increased the T1-weighted (T1-w) MRI signal throughout the kidney immediately after injection. D: typical time series in each voxel in regions corresponding to the green squares in C are shown in squares in a grid, mapped to a two-dimensional (2-D) image slice, during injection of CF (left) or Gd-DTPA bolus (right). The central square is a voxel containing a glomerulus. E: time series of a single voxel containing a glomerulus after CF accumulation, with a slow decrease in the signal (S) magnitude in the glomerulus. The delay time and rates of this decrease varied between glomeruli. F: time series of the same voxel after injection of a Gd-DTPA bolus showing an initial increase and peak in the MRI signal magnitude that decreased as the unrecirculated Gd-DTPA was removed from the vasculature. Several minutes later, a second peak was observed only in and around the glomeruli, consistent with filtration and concentration in the distal convoluted tubule. au, arbitrary units; ROI, region of interest; T2*-w, T2*-weighted. Scale bar = 1 mm.

Tracer Injections and MRI

CF transiently binds to anionic proteoglycans in the glomerular basement membrane (GBM) (13, 14, 16). We infused CF at a low concentration (Sigma-Aldrich) to observe its rate of binding to the GBM in individual glomeruli using CFE-MRI. Physiological medium without CF was perfused for 5 min to establish the baseline signal, followed by 0.033 mg/mL CF in the medium for 10 min (0.2 mg/min). This concentration of CF was low enough to only be detected in the kidney cortex and in the glomeruli after accumulation, but not in the vasculature during perfusion. We next injected Gd-DTPA (BioPal), a freely filtered imaging tracer (24). We perfused the kidney with the physiological medium without Gd-DTPA for 5 min to reestablish the baseline signal. We then injected a bolus of 12.5 µmol Gd-DTPA over 20 s. The perfusate was not recirculated, ensuring that any signal enhancement at the glomerulus after the initial bolus had passed would reflect only filtered Gd-DTPA in the tubule. After injection of the Gd-DTPA bolus, the kidney was perfused with physiological medium without Gd-DTPA for 10 min.

Imaging was performed during perfusion using a Bruker ClinScan 7 T/30 MRI (Bruker, Berillica, MA) with Siemens Syngo (Siemens) software and a 3-D gradient recalled echo (GRE) protocol with the following parameters: field of view = 20.000 × 13.125 × 14.080 mm, matrix size = 192 × 71 × 64, and resolution = 104.2 × 184.9 × 220.0 µm³. For dynamic CFE-MRI, we used T2*-weighted GRE with an echo time (TE)/repetition time (TR) = 12.6/45, flip angle = 45°, and whole kidney image acquisition time = 2.83 min. To image Gd-DTPA kinetics, we used T1-weighted GRE with TE/TR = 3.23/22 ms, flip angle = 45°, and whole kidney image acquisition time = 1.38 min.

After the experiments, the kidneys were perfused with PBS followed by 10% formalin for fixation and stored in 10% formalin for high-resolution MRI.

Image Processing

The following software was used for 3-D coregistration, segmentation of glomeruli, and analyses: AFNI (National Institutes of Health), MiPAR (MIPAR Image Analysis), and MATLAB (The MathWorks). In this initial study, we performed the analysis on only half of the kidney (from the renal artery to the pole). This portion of the kidney was located entirely within the rings of the surface coil where there was minimal loss of RF power. Images were resampled by a factor of 2 in every direction using Lanczos interpolation in Matlab. An iterated and linearized weighted least squares algorithm was used (3dvolreg, AFNI) for 3-D coregistration of the images. Glomeruli were segmented in the images by first segmenting the cortex from the whole kidney and then using an adaptive threshold function in MiPAR (Box Median, window size = 7 voxels, percent = 67%). We aimed to minimize partial volume effects by limiting our analysis in the remaining kidney half to glomeruli with a signal magnitude at least 30% lower than background, ensuring that the glomeruli were approximately in the middle of the voxel. We then selected a random sampling of 3,000 glomeruli identified to be near centered in the voxel to compare distributions across all animals. High-resolution ex vivo analysis after perfusion (not shown) indicated that 90% of all CF-labeled glomeruli had signal magnitude reduced by at least 30% from background.

To investigate the kinetics of CF binding, we evaluated the time series of the MRI signal in each voxel identified with a glomerulus before, during, and after CF infusion using a biexponential pharmacokinetic model and nonlinear least squares fitting as follows:

$${\text{S}}_{\text{i}}\left(t\right)=A+Bt+k[e^{-{\alpha }_{\text{1}}(t-t_0)}-e^{-{\alpha }_{\text{2}}(t-t_0)}]$$ (1)

Here, S_i(t) is the normalized signal (in %) at time t, A is the y-intercept (in %), B is a linear factor (in min⁻¹), k is the scale constant (in %), t₀ is the delay time (in min), α₁ is the rate after CF (in min⁻¹), and α₂ is the CF uptake rate (in min⁻¹). Average R² from fitting was 0.77. Signal magnitude in each voxel time series was normalized to average magnitude of the first two time points before the arrival of the tracer. We also calculated the area of the curve (AUC) in each time series between the time of initial decrease in the signal until it approached its lower asymptote, as a measure of total CF uptake.

To model the filtration of Gd-DTPA, we calculated the AUCs of each of two peaks observed in the time series of each voxel identified to contain a glomerulus. Signal magnitude was normalized to the average magnitude of the first three time points before the bolus. An average time series measured from the renal artery was used to calibrate the model and express Gd-DTPA in each voxel in units of nanomoles. AUC calculated from the first peak (AUC_{Peak 1}) was a measure of total Gd-DTPA through the voxel. AUC of the second peak (AUC_{Peak 2}), observed between 4 and 9 min after the first peak, was a measure of Gd-DTPA reappearing through the same voxel, likely Gd-DTPA traversing the distal convoluted tubule after filtration. We measured the effective filtration fraction (eFF) as eFF = AUC_{Peak 2}/AUC_{Peak 1}. The following monoexponential model was fitted between the first time point to the first peak to measure the perfusion rate in each glomerulus (Q_glom): S_i(t) = Ce^−α3t, C is a scaling factor (in %), and α₃ is the rate of signal increase to the Gd-DTPA peak (in min⁻¹). Average R² from fitting was 0.69. We then estimated SNGFR in each nephron (eSNGFR) using Eq. 2 (28) as follows:

$$\mathrm{eSNGFR}=Q_{\mathrm{glom}}·\mathrm{eFF}$$ (2)

To confirm that AUC measured from the second peak at the distal convoluted tubule, near the glomerulus, reflects total filtered Gd-DTPA, we compared it with the AUC measured from the second peak in neighboring voxels. The AUC in neighboring voxels should reflect total filtered Gd-DTPA in other segments of the tubule. We randomly selected glomeruli (n = 45) throughout the kidney in each animal. We analyzed neighboring voxels in a region of interest of 3 × 3 × 3 voxels (Supplemental Fig. S1). We then analyzed the Gd-DTPA time course in each voxel to determine the timing of the second peak and calculate AUC_{Peak 2}. We used AUC_{Peak 1} and glomerulus perfusion from the corresponding central glomerulus voxel to calculate eFF and eSNGFR in these neighboring voxels.

High-Resolution MRI of the Fixed Kidney and Image Processing

We measured glomerular number (N_glom) in three kidneys using high-resolution ex vivo MRI as an independent assessment of the natural variation in nephron number. Fixed kidneys were washed in PBS for 24 h and imaged using high-resolution ex vivo MRI to confirm CF labeling of glomeruli and to measure N_glom. Imaging was performed using an Agilent 11.74-T DirectDrive MRI (Agilent). A T2*-weighted 3-D GRE protocol was used with TE/TR = 15/100, flip angle = 30°, field of view = 27.0 × 27.0 × 30.0 mm, matrix size = 512 × 512 × 512, and resolution = 52.7 × 52.7 × 58.6 µm³. Custom software (19) was used to measure N_glom from 3-D CFE-MRI.

Measuring Spatial Patterns in the Kidney

Spatial patterns of glomerular volumes might correlate with spatial patterns of CF uptake and eSNGFR. To assess this, we mapped glomerular volumes and eSNGFR. All glomeruli were categorized into one of 10 groups based on distance from the kidney surface, which were spaced by 170 µm: 0–170 µm, 170–340 µm, etc. The deepest group consisted of all glomeruli >1,530 µm from the kidney surface.

Previous studies (12, 19) have shown that large glomeruli represent a small fraction of glomeruli compared with the total number of glomeruli in this strain of rats. To determine whether eSNGFR is associated with glomerular volume, we examined glomeruli in groups based on size. Glomerulus size was determined based on signal analysis of CF-labeled glomeruli, similar to previous work (16, 18, 19). CF-labeled glomeruli were randomly chosen throughout the kidney and categorized into groups based on glomerulus size determined by a threshold of signal magnitude and half width of full profile from CF-labeled glomeruli (16, 18, 19): 1) small, <84.5 µm; 2) medium, 84.5–169.7 µm; and 3) large, >169.7 µm.

The cortex was segmented based on distance from kidney surface: 1) defined as the superficial cortex, 0–850 µm; and 2) defined as the deep cortex, >850 µm from the kidney surface.

Statistics

Student’s two-sample t test and Wilcoxon rank sum functions in Matlab were used to compare group means and medians and to test the null hypothesis that group distribution means or medians were equal (P < 0.05).

RESULTS

We performed dynamic MRI of the isolated perfused rat kidney using a custom MRI-compatible frame (Fig. 1A). We first examined the binding kinetics of CF in glomeruli in 3-D MRI and then measured filtration kinetics of Gd-DTPA in the same glomeruli (Fig. 1, B and C). During CF infusion, dark punctate spots slowly appeared in the MR images throughout the cortex over a period of ∼10 min (Fig. 1C). The distribution of spots was consistent with the distribution of glomeruli observed in ex vivo MRI (Fig. 1C) and consistent with previous work (10–12). The MR signal stopped decreasing within 10 min, by the end of CF infusion (Fig. 1E). We identified glomeruli in the 3-D MR images based on signal magnitude compared with local tissue background. The kidney from one rat (rat 1) had some leakage of perfusate at the cannula insertion during perfusion, causing reduced changes in signal magnitude after injection of the tracers. Because our measurements of each time series were normalized to the concentration of the delivered dose, we retained this kidney in all analyses and achieved consistent results. Median perfusion pressure in all kidneys was 173 ± 18 mmHg at the beginning of perfusion and 209 ± 38 mmHg at the end of perfusion.

Directly after injection of the bolus of Gd-DTPA, signal magnitude increased throughout the kidney (Fig. 1C). The time series exhibited two distinct peaks in voxels throughout the kidney cortex (Fig. 1F): the large first peak (peak 1), appearing shortly after bolus injection, and a second peak (peak 2) detectable several minutes later. Only voxels in and around glomeruli exhibited both peaks. The first peak was primarily caused by the passage of the injected bolus through the vasculature, as indicated by its onset directly after the injection. The second peak was ∼50% lower in magnitude than the first peak, and its timing indicated that it originated from Gd-DTPA that had returned to the vicinity of the glomeruli in the distal convoluted tubule after filtration and concentration.

We measured total CF uptake in each voxel containing a glomerulus using the time series of the MRI signal during CF infusion. We grouped glomeruli into two groups based on differences in the time course of CF uptake. The onset time of signal decrease was one significant difference between glomeruli. Thus, we analyzed glomeruli in “early” and “late” groups using the fitted delay time (t₀; Fig. 2A): t_{0, Early} ≤ 3.2 min, t_{0, Late} > 3.2 min. Glomeruli in the early group comprised 54% of the glomeruli across all animals. Early group glomeruli had significantly greater (P < 0.05) total CF uptake during the experiment compared with glomeruli in the late group (Fig. 2C) and had significantly slower uptake (P < 0.05; Fig. 2D). We did not detect a difference in the spatial distributions of early and late group glomeruli.

Figure 2.

Fitted time series of the normalized magnetic resonance imaging signal magnitude in individual voxels containing glomeruli during cationized ferritin (CF) infusion and uptake. A: the time series in each voxel containing a glomerulus was fitted to a biexponential pharmacokinetic model. The fitted delay time (t₀) was used to group glomeruli into either early-appearing or late-appearing groups: t_{0, Early} ≤ 3.2 min and t_{0, Late} > 3.2 min. B: the area of the time series between the initial decrease to the final asymptote was used as a measure of total CF uptake. C: mean and standard deviation of total CF uptake between glomeruli in the early and late groups. Glomeruli in the early group had increased total CF uptake compared with glomeruli in the late group. D: the rate of CF uptake was significantly different between early- and late-appearing glomeruli.

We mapped single-nephron filtration in the same voxels using the time series of the MRI signal after Gd-DTPA injection (Fig. 3A, inset). The distributions of AUC calculated for each peak 1 and peak 2 were unimodal. On average, AUC_{Peak 2} was higher in glomeruli for which AUC_{Peak 1} was also higher (mean R² = 0.46 ± 0.2; Fig. 3A). Time between peak 1 and peak 2 varied from 4 to 9 min between glomeruli within individual kidneys. About 10% of glomeruli had a long (>6.8 min) time interval between peaks. Median eFF was at least 15% lower in these glomeruli, and eSNGFR was 12% lower.

Figure 3.

Summary of pharmacokinetic parameters measured from magnetic resonance imaging time series in the isolated rat kidney in voxels containing individual glomeruli after an intravenous bolus of gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA). A, inset: the area under the curve was used to measure total Gd-DTPA passing through a voxel, both in the first peak (Peak-1 Gd-DTPA) and in the second peak (Peak-2 Gd-DTPA). A: Peak-1 Gd-DTPA correlated with Peak-2 Gd-DTPA. B: distributions of perfusion rates in measured glomeruli in each animal. C: estimated filtration fraction (eFF) in glomeruli in each animal. D: distributions of estimated single-nephron glomerular filtration rate (eSNGFR) in glomeruli. E: estimated single-kidney GFR (eSKGFR) based on the cumulative distribution of eSNGFR and total glomerular number (N_glom) measured in each kidney. R1, rat 1; R2, rat 2; R3, rat 3; R4, rat 4; ΔS, change in signal.

Computed perfusion rate, eFF, and eSNGFR in glomeruli were mapped in the reconstructed images, providing distributions of each parameter (Fig. 3, B–D). eSNGFR did not correlate with total CF uptake either within an animal or between animals (average R² < 0.01), indicating that CF binding did not affect glomerular filtration. Mean perfusion rate did not correlate with mean eFF. eFF and eSNGFR were spatially heterogeneous. Median eSNGFR from all animals was 31.4 ± 10.5 nL/min, but eSNGFR varied between glomeruli by an order of magnitude (∼10 to ∼100 nL/min). Estimated single-kidney glomerular filtration rate (GFR), calculated from the distribution of eSNGFR and N_glom, in kidneys from rats 2, 3, and 4 was 1.25, 1.29, and 1.30 mL/min, respectively (Fig. 3E), and was consistent with typical expected single-kidney GFR in rats (34, 35). Median eSNGFR was inversely correlated with N_glom (R² = 0.98) across kidneys.

We correlated glomerulus size with eSNGFR in glomeruli based on size (small, <84.5 µm; medium, 84.5–169.7 µm; and large, >169.7 µm). The fraction of the large glomeruli in the superficial cortex was 10.9 ± 4.6%. In the deep cortex, it was 22.1 ± 5.9%. Median eSNGFRs for small, medium, and large glomeruli were as follows: 32.7 ± 8.3, 35.0 ± 8.9, and 39.0 ± 11.1 nL/min, respectively. Although this difference in eSNGFR between small and large glomeruli was 20%, the difference was not statistically significant across the whole population. eSNGFR in the distal convoluted tubule and eSNGFR in neighboring voxels were loosely correlated (R² = 0.50). Average glomerular volume in all three animals was 10.8 ± 1.4 × 10⁻⁴ mm³.

DISCUSSION

This work describes the use of MRI to measure single-nephron function in isolated rat kidneys during controlled perfusion. A key finding was the observation of a second peak of contrast enhancement in voxels containing glomeruli, after the bolus of Gd-DTPA had cleared from the vascular circulation. The timing of the MRI signal is consistent with single-nephron glomerular filtration, similar to the timing of filtered fluorescent dyes detected using intravital optical microscopy (4, 36). The experimental time courses and associated kidney microstructure are shown in Fig. 4.

Figure 4.

Top: illustration of a magnetic resonance imaging (MRI) voxel and experimental time course of the isolated perfused rat kidney, with corresponding glomerular and nephron structure. Voxels contain glomeruli, part of the proximal tubule, and part of the distal convoluted tubule from the same nephron. Bottom left: contrast agent 1. Cationic ferritin (CF) is perfused into the kidney, and the transient binding of CF to the glomerular basement membrane is observed with T2*-weighted MRI. CF-labeled glomeruli are then identified throughout the kidney. Bottom right: contrast agent 2. A freely filtered contrast agent, gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA), is administered as a bolus into the kidney with no recirculation. With T1-weighted MRI, the initial bolus causes a bright signal in voxels containing glomeruli. Minutes later, a second peak is observed in the same voxels due to the filtered Gd-DTPA returning to the voxel through the distal convoluted tubule. Scale bars = 0.15 mm.

We mapped the binding kinetics of CF in individual glomeruli throughout the kidney. Total rates and uptake of CF were spatially heterogeneous. Glomeruli exhibited differences in time courses of CF uptake. The most notable difference was the onset time of decrease in signal magnitude during injection of CF. We categorized glomeruli into one of two groups based on the onset time. However, we did not observe a spatial pattern in these onset times. Glomeruli from both groups appeared evenly distributed in the cortex. There may be a spatial pattern that we did not detect, such as proximity to large vessels. Future work is needed to better understand the difference between these two populations. eFF and eSNGFR of Gd-DTPA were also spatially heterogeneous. Interestingly, median glomerular volume varied little (maximum: 12%) between superficial and deep glomeruli. However, there was a wide distribution of glomerular volumes at every depth. This is likely due to the small fraction of large glomeruli and reflects the heterogeneity of glomerular volumes throughout the kidney. Median eSNGFR correlated with glomerular volume, but there was still a high variation of eSNGFR. Future work is needed to determine whether eSNGFR is correlated with any underlying morphology. CF binding did not correlate with filtration rates of GD-DTPA in individual nephrons, indicating that 1) CF labeling does not significantly alter single-nephron filtration, and 2) mechanisms of regulation of CF binding to the GBM may be distinct from those affecting SNGFR in the healthy kidney. This is consistent with the hypothesized limited role of the GBM in glomerular filtration (37, 38).

We quantified eSNGFR based on the time course of Gd-DTPA in voxels containing glomeruli. We made the following assumptions: 1) the relaxivity curve of Gd-DTPA is linear, 2) the voxel covers the glomerulus and distal convoluted tubule, and 3) Gd-DTPA has negligible reabsorption. Based on these assumptions, the AUC should reflect the total amount Gd-DTPA through the voxel at both time points, because it is averaged over the entire time course. Any T2* effect from the concentration of Gd-DTPA in the tubule was negligible due to the short TE and relatively large voxel size relative to the tubule diameter. MRI-based measurement of median eSNGFR was similar to measurements using other techniques (3). The distribution of eSNGFR varied between healthy animals. This is consistent with the natural variation in nephron number, which is well known in healthy animals, and is usually within 15% (12, 19). Here, the sum total of eSNGFR from all nephrons was consistent with whole kidney GFR in the literature (34, 35), and median eSNGFR was directly proportional to the number of glomeruli. This is also consistent with the concept that SNGFR changes to maintain total GFR, given an animal’s nephron endowment. Glomerular volume also varies naturally in healthy animals (12, 19).

In 90% of measurements, we found two, three, or more neighboring voxels with second peaks. We compared eSNGFR in the distal convoluted tubule, in the voxel containing the glomerulus, with eSNGFR in these neighboring voxels. These measurements were moderately correlated. However, there are limitations to measuring eSNGFR in neighboring voxels. With the spatial and temporal resolutions used here, 1) tubule segments from the same nephron may be visible in more than one neighboring voxel, or 2) tubule segments from other nephrons may be visible in these neighboring voxels. Improving spatial and temporal resolution of this technique may allow us to more completely map the physiology of the whole nephron.

This work has several limitations. These experiments were performed using perfusion with a constant flow rate to control the timing of the injected contrast agents. Experiments using constant pressure likely mimic in vivo physiology more accurately. Other tools cannot measure the function of large populations of nephrons throughout the kidney. However, the distributions of eSNGFR here are consistent with optical (4, 5, 39, 40) and micropuncture (2, 3, 41) studies. Our custom RF probe was intended to maximize signal over a region, but we limited our investigations to half of the kidney to avoid possible artifacts from spatially variable RF power. We also limited our analysis to glomeruli identified to be centrally located within voxels. This reduced our observations to 3,000 glomeruli per animal in this initial study. Future work is required to determine whether RF artifacts and partial volume effects can affect the measurements of single-nephron function. The hardware can be further refined to provide coverage of the whole kidney. Future work will also focus on improving spatial and temporal resolution of single-nephron MRI. Finally, we examined only the rat kidney here, but it is possible to perform similar experiments in multipapillary kidneys from other species using hardware designed for the species of interest.

This work builds on previous studies of individual nephron function (2, 4, 6, 21, 32, 40, 42–44) but expands our capability to simultaneously examine large numbers of nephrons. The use of MRI to dynamically map tracer kinetics at the level of the individual nephron provides a new opportunity to study heterogeneous single-nephron physiology and spatially coordinated nephron function (Fig. 4). MRI could also be used to investigate the same kidney in vivo and in vitro to identify the role of systemic regulation in single-nephron function under various conditions. There have been few studies measuring SNGFR repeatedly to assess how SNGFR varies over time. Autoregulatory signaling varies spatially and over time (7, 45) and may drive similar variability in glomerular filtration. A recent study has also demonstrated synchronous autoregulation between nephrons (8). MRI may reveal this larger scale synchronization of nephron filtration, local perfusion, and coupling between nephron and vascular physiology. MRI may provide new perspectives to study the local filtration and perfusion heterogeneity of adjacent nephrons under normal and abnormal states and their impact on kidney function. Repeated measurements of SNGFR could shed light on this unexplored area. Exposing the kidney to different interventions could also reveal variations of SNGFR in response to therapies. This MRI technique complements optical techniques. In addition to CF, other MRI-visible macromolecules could be used as targeted or passive contrast agents to measure changes in the GBM or other kidney structures associated with pharmacological manipulation or disease.

Perspectives and Significance

We have described the use of MRI to investigate single-nephron physiology in the excised, perfused kidney. MRI of the isolated kidney provides a new perspective from which to investigate nephron function across the whole organ. This work provides a foundation for studies to simultaneously map the function of large numbers of nephrons.

SUPPLEMENTAL DATA

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.20496948.v2.

GRANTS

This work was supported by National Institutes of Health (NIH) Grants DK11186102 (to K.M.B. and J.R.C.) and DK11062204 (to K.M.B and J.R.C.). E.J.B. is supported by NIH Grant TL1TR002344. This work used the Bruker ClinScan MRI in the Molecular Imaging Core, which was supported by the University of Virginia School of Medicine.

DISCLOSURES

E.J.B. and K.M.B. own XN Biotechnologies, LLC. K.M.B. and J.R.C. are co-owners of Sindri Technologies, LLC. K.M.B. has a research agreement with Janssen Pharmaceutical, LLC. J.R.C. is a consultant for Medtronics. K.M.B. is co-owner of Nephrodiagnostics, LLC.

AUTHOR CONTRIBUTIONS

E.J.B., J.R.C., and K.M.B. conceived and designed research; E.J.B. performed experiments; E.J.B. analyzed data; E.J.B., J.R.C., and K.M.B. interpreted results of experiments; E.J.B. prepared figures; E.J.B. and K.M.B. drafted manuscript; E.J.B., J.R.C., and K.M.B. edited and revised manuscript; E.J.B., J.R.C., and K.M.B. approved final version of manuscript.