Quantitative Micro-Computed Tomography Imaging of Vascular Dysfunction in Progressive Kidney Diseases

Peritubular Capillary Rarefaction Correlates with Fibrosis in Chronic Kidney Disease Models

First, we analyzed mice with I/R-induced injury and monitored the entire time course from acute ischemic injury (day 1) to late-stage renal fibrosis (day 56).^14,21 Histologically, a continuous increase of collagen I, fibronectin, and α-smooth muscle actin (αSMA)-positive fibroblasts confirmed the development of progressive fibrosis and was paralleled by a continuous decrease of Meca-32-positive capillaries (Figure 2, A and B). Significant correlations between these parameters demonstrated an association between progressive fibrosis and capillary loss (Figure 2, B and C). Importantly, the prominent early reduction of Meca-32 positive peritubular capillaries in I/R injury, found already on day 1, preceded the onset of fibrosis.

In order to exclude model-specific effects related to I/R-induced progressive kidney injury, the correlation between progressive fibrosis and continuous vessel rarefaction was histologically confirmed in UUO (days 1–10) and Alport mice (6–8 weeks). Comparable to the time-course findings obtained in the I/R model, a prominent early reduction of Meca-32 positive cortical peritubular capillaries on day 3 also preceded the onset of significant fibrosis, as found on day 5 in the UUO model (Figure 3A). Analogously, also in Alport mice, progressive loss of Meca-32 positive capillaries was paralleled by fibrosis development (Figure 3B).

Functional In Vivo µCT Imaging Allows Accurate Assessment of Reduced Vessel Functionality in Chronic Kidney Disease Models

To monitor renal blood vessels during the progression of renal disease, we employed a contrast-enhanced µCT approach for in vivo imaging using an iodine-based contrast agent optimized for blood-pool imaging.¹⁸ This approach enabled the noninvasive visualization of renal vessels with a spatial resolution of 35 µm voxel side length, as exemplified by cross-sectional images and 3D renderings (Figure 4A). Longitudinal in vivo µCT scans of mice with I/R-induced kidney injury confirmed the significant loss of functional blood vessels already at early time points after I/R injury and an even more pronounced rarefaction in late-stage fibrosis (Figure 4B, Supplemental Videos 1 and 2). Accordingly, renal rBV values decreased already early (−20% on I/R day 1) and progressively (up to −43% on I/R day 56, compared with baseline renal rBV values quantified in sham controls), demonstrating that fibrosis progression was paralleled by a gradual loss of vessel functionality (P<0.001; Figure 4C). Similar to the histologic findings, the reduction in functional vessels assessed by in vivo µCT was observed before the onset of interstitial fibrosis (Figure 2B).^7,22–24

We confirmed the continuous reduction of vessel functionality in Alport and UUO mice. For both models, in vivo µCT showed significantly reduced rBV values in diseased versus control kidneys (Figure 5, A and B for UUO; Figure 5, D and E for Alport mice, respectively). Strikingly, in the UUO model, capillary rarefaction and the gradual loss of vessel functionality were also observed very early, before the onset of overt fibrosis, i.e., on day 3 (Figures 3A and 5B).

To assess the accuracy of in vivo µCT for quantifying vascular dysfunction in kidney fibrosis, rBV values were correlated with histologically reconstructed equivalents (Supplemental Figure 1).¹⁸ As exemplified by the I/R model, a significant correlation between rBV values determined by in vivo µCT and immunohistochemistry was observed (R²=0.98, P<0.001, Figure 4D). This high congruency was also found in the UUO model and in Alport mice (Figure 5, C and F, respectively). Within all the experiments, quantified renal rBV values were extremely consistent and showed very low variability in both healthy and diseased animals, further corroborating the high accuracy and robustness of our methodology.

Renal rBV in contralateral (right) kidneys of mice with I/R or UUO supported the progressive course of kidney damage in the injured (left) kidneys. Renal rBV was significantly increased in the contralateral kidneys from day 14 in I/R and day 7 in UUO mice onwards (+17% at I/R day 56 and +21% at UUO day 10; Supplemental Figure 2, A and B), suggesting a compensatory hypertrophy of these kidneys. The total kidney volume of contralateral kidneys correspondingly increased during disease progression (Supplemental Figure 3, A and B). The injured left kidneys in I/R mice also showed an initial increase in organ volume, most likely due to post-ischemic swelling (I/R day 1–3), followed by continuous organ shrinkage from day 5 onwards. In UUO, as expected due to the progression of hydronephrosis, kidney volumes continuously increased (Supplemental Figure 3B).

Anatomical Ex Vivo µCT Imaging Reveals Macro-to-Microvascular Alterations in Chronic Kidney Disease Models

Upon in vivo imaging, mice were perfused with the vascular casting agent Microfil,¹⁸ and subsequently performed high-resolution µCT scans of excised kidneys resulted in a spatial resolution of approximately 3 µm voxel side length. Previous studies have reported on Microfil perfusion of the renal vasculature in mice and rats.^25–28 Here, we used this high-resolution ex vivo µCT method to analyze vessel morphometrics that can be derived from perfused renal vasculature, such as branching, size and tortuosity, in progressive kidney fibrosis.

After threshold-based vessel tree segmentation, compared with healthy controls, a significant loss of small-caliber arterial vessels was observed in fibrotic kidneys (Figures 6A, 7A and 8A; Supplemental Videos 3 and 4). When analyzing vascular branching in detail, the reduction of the renal rBV in progressive kidney fibrosis (Figure 4C) was morphologically linked to a strong reduction of peripheral branching points: i.e., in I/R day 14 kidneys, the number of 4th (Aa. interlobulares) and 5th order (afferent arterioles) branching points significantly decreased by −38% and −74%, respectively (P<0.001; Figure 6B). The number of peripheral branching points progressively decreased until day 21 with up to −60% and −89% reduction of 4th and 5th order branching points, respectively (P<0.001, Figure 6B). High-resolution µCT imaging revealed similar findings for Alport kidneys (Figure 7, A and B) and an even more pronounced reduction in UUO with up to −68% and −91% reductions for 4th and 5th order branching points, respectively (P<0.001; Figure 8, A and B). Importantly, branching points of large-caliber arterial vessels (Aa. segmentales, interlobares and arcuatae) were not altered. These findings indicate that independently of the underlying disease, kidney injury is not only associated with the well-described rarefaction of peritubular capillaries^1,14,29,30 but also with a pronounced rarefaction of functional small-caliber arteries (from Aa. interlobulares downward).

Surprisingly, when quantifying the size of renal vessels, a yet unrecognized reduction of luminal diameter of all large-to-small-sized arterial vessels during disease progression was observed. As exemplified by I/R day 14, a massive reduction in diameters of Aa. segmentales (−14%), interlobares (−21%), arcuatae (−30%), interlobulares (−36%), and afferent arterioles (−22%) was found compared with sham-treated controls (P<0.001; Figure 6C). Strikingly, vessel diameters were not further reduced beyond day 14, even though fibrosis and vessel rarefaction progressed. A similar reduction of arterial vessel diameters was confirmed in Alport and UUO kidneys (Figures 7C and 8C).

To extend these analyses, we also quantified vessel tortuosity, i.e., the ratio of the vessel length between two nearby branching points and the linear distance between these two (see Figure 1). In line with observed reductions in vessel size and branching points, a clear increase in vessel tortuosity was observed in all three models during fibrosis progression. In I/R day 14, an increase in vessel tortuosity was observed in Aa. interlobares (+4%), arcuatae (+6%), interlobulares (+11%), and afferent arterioles (+16%; P<0.001; Figure 6D). Beyond I/R day 14, tortuosity further increased in Aa. interlobulares and afferent arterioles (+14% and +21%, respectively; P<0.0001). µCT-based determinations of vessel tortuosity in Alport kidneys revealed similar findings (Figure 7D) and again an even stronger effect in UUO kidneys with increased tortuosity by +18% for Aa. arcuatae, +18% for Aa. interlobulares and +28% for afferent arterioles (P<0.001; Figure 8D). Collectively, these findings exemplify that high-resolution µCT is highly suitable for a comprehensive assessment of vascular alterations and revealed significant changes in vessel branching, diameter, and tortuosity during the progression of kidney diseases.

Kidney Injury Models

Chronic kidney injury was surgically induced in 6-week-old C57bl/6 wild-type mice by ischemia (30 min)/reperfusion injury (I/R) or unilateral ureteral obstruction (UUO) of the left kidney, as described previously.^29,55 Sham-operated animals were used as controls. As a third model for progressive renal injury, 6–8-week-old Col4a3^−/− mice bred on a 129/SvJ background, a model of Alport syndrome, were investigated alongside age-matched wild-type littermates. All mice were housed in a specific pathogen-free environment under ethical conditions approved by German legal requirements.

In Vivo Micro-Computed Tomography

Contrast-enhanced in vivo µCT imaging was performed using a dual-energy flat-panel micro-computed tomography scanner (TomoScope 30s Duo; CT Imaging, Erlangen, Germany). Mice were imaged before and immediately after intravenous injection of 100 µl eXIA^TM160XL (Binitio Biomedical, Ottawa, Canada), a high molecular weight iodine-based contrast agent optimized for in vivo µCT-based blood pool imaging. The contrast agent was injected as a bolus into the lateral tail vein. During the entire in vivo imaging process, mice were anesthetized with 1.5% isoflurane in oxygen-enriched air. Dual-energy scans were performed at 41 and 65 kV (0.5 mA and 1 mA) resulting in 2880 projections of size 1032×1024 acquired over a 6-minute time frame. A Feldkamp-type algorithm (CT Imaging), including ring artifact correction, was applied to reconstruct volumetric data with a voxel size of 35×35×35 µm³. The reconstructed data were processed using Imalytics Preclinical software (Gremse-IT, Aachen, Germany) enabling the 3D visualization of blood vessels and interactive segmentation of volumes of interest (VOIs). Quantifications of the renal relative blood volume (rBV) were performed after defining five spherical (500 µm diameter) VOIs per kidney cortex. For each kidney, the rBV was calculated as the ratio of the contrast agent-induced enhancement (in Hounsfield units) of the renal cortex and the enhancement of pure blood which was measured in a large reference blood vessel (aorta) (Figure 1).¹⁸ To quantify the total organ volume of kidneys, 3D µCT-based organ segmentations were performed.

Ex Vivo Micro-Computed Tomography

After in vivo µCT imaging, mice were perfused with Microfil (Flow Tech, Carver, MA), a lead-containing silicone rubber radiopaque contrast agent, by using constant flows adjusted via a perfusion pump (150 ml/h; catheter diameter: 1.0 cm; catheter length: 75 cm), as described previously in more detail.¹⁸ Microfil, which solidifies approximately 20 min after application within the vascular compartment, enables the high-resolution 3D investigation of the microarchitecture of renal blood vessels. To this end, the excised kidneys were formalin-fixed and scanned using a high-resolution SkyScan 1172 µCT (SkyScan, Kontich, Belgium). In detail, kidneys were positioned on a rotation platform and scanned 180° around the vertical axis in rotation steps of 0.3° at 59 kV and an electric current of 167 µA. This resulted in 640 acquired projections (4000×2096 pixels) acquired over a period of 4–6 hours per kidney. Reconstructions were performed using filtered back-projection (Feldkamp-type) resulting in an isotropic pixel size of 3 µm. After 3D volume rendering of reconstructed high-resolution µCT data sets and threshold-based visualization of blood vessels, vascular branching as well as vessel size and tortuosity were semi-automatically analyzed as described in Figure 1. In detail, vessel size and tortuosity were systematically determined by randomly choosing ten vessels per hierarchical order in the renal vessel tree (Aa. segmentales, interlobares, arcuatae, interlobulares and afferent arterioles). For each blood vessel, the diameter was determined in one of three possible (transversal, coronal or sagittal) planes in which the cross-sectional area was relatively round and tortuosity was calculated as the ratio of the shortest vessel path length between two nearby branching points and the linear distance between these branching points. Normal anatomy of healthy kidneys confirmed our approach of tortuosity determination: compared with straight-running arteries, the more windy arcuate arteries showed the highest values for tortuosity (Figures 6D, 7D and 8D). Vascular branching patterns were analyzed by quantifying the number of branching points in total and per rising branching order per primary blood vessel along its course, from the hilus (Aa. segmentales) to the periphery (afferent arterioles) (see Figure 1).

Histology and Immunohistochemistry

Tissue sections were fixed using Methyl-Carnoy’s solution and paraffin-embedded to assess renal morphology using periodic acid-Schiff staining and to perform immunohistochemical stainings against collagen I, fibronectin, αSMA and Meca-32 using routine protocols.⁵⁵ In brief, sections were stained using primary antibody against murine collagen I (Southern Biotechnology Associates, Birmingham, AL), fibronectin (Chemikon International, Temecula, CA), αSMA (Dako, Glostrup, Denmark), and Meca-32 (Developmental Studies Hybridoma Bank at the University of Iowa, Iowa City, IA). Biotinylated secondary antibodies were purchased from Vector Laboratories (Burlingame, CA). Stained tissue slides were scanned using NanoZoomer digital slide scanner (Hamamatsu Photonics, Hamamatsu, Japan). For quantifying the degree of renal fibrosis, 16 photomicrographs (Original magnification,×400) were randomly taken per kidney cortex of each mouse (n=28 I/R, n=20 UUO, n=8 sham, n=10 Alport, and n=10 untreated wild-type mice). Area fractions of collagen I, fibronectin, αSMA, and filled Meca-32 positive blood vessels (see below) were quantified using open-access image analysis software ImageJ version 1.43u (National Institutes of Health, Bethesda, MD). For reconstructing the equivalent of in vivo quantified rBV values on 2D histologic slices, the lumen of Meca-32 positive vascular structures was computationally filled using three photomicrographs per kidney cortex, and the filled Meca-32 area fraction, which most closely reflects the vascular volume, was determined using a previously described custom macro implemented for ImageJ.¹⁸

Statistical Analysis

All data are presented as mean±SD. For assessing statistical significance, the two-tailed unpaired Student’s t test and Pearson’s correlation were used. P values <0.05 were considered as statistically significant. All statistical analyses were performed using Graph PadPrism 5.0 (San Diego, CA).