Microvascular remodeling and altered angiogenic signaling in human kidneys distal to occlusive atherosclerotic renal artery stenosis

ABSTRACT

Background

Renal artery stenosis (RAS) is an important cause of chronic kidney disease and secondary hypertension. In animal models, renal ischemia leads to downregulation of growth factor expression and loss of intrarenal microcirculation. However, little is known about the sequelae of large-vessel occlusive disease on the microcirculation within human kidneys.

Method

This study included five patients who underwent nephrectomy due to renovascular occlusion and seven nonstenotic discarded donor kidneys (four deceased donors). Micro-computed tomography was performed to assess microvascular spatial densities and tortuosity, an index of microvascular immaturity. Renal protein expression, gene expression and histology were studied in vitro using immunoblotting, polymerase chain reaction and staining.

Results

RAS demonstrated a loss of medium-sized vessels (0.2–0.3 mm) compared with donor kidneys (P = 0.037) and increased microvascular tortuosity. RAS kidneys had greater protein expression of angiopoietin-1, hypoxia-inducible factor-1α and thrombospondin-1 but lower protein expression of vascular endothelial growth factor (VEGF) than donor kidneys. Renal fibrosis, loss of peritubular capillaries (PTCs) and pericyte detachment were greater in RAS, yet they had more newly formed PTCs than donor kidneys. Therefore, our study quantified significant microvascular remodeling in the poststenotic human kidney. RAS induced renal microvascular loss, vascular remodeling and fibrosis. Despite downregulated VEGF, stenotic kidneys upregulated compensatory angiogenic pathways related to angiopoietin-1.

Conclusions

These observations underscore the nature of human RAS as a microvascular disease distal to main vessel stenosis and support therapeutic strategies directly targeting the poststenotic kidney microcirculation in patients with RAS.

Graphical Abstract

Graphical Abstract.

KEY LEARNING POINTS.

What is already known about this subject?

• Studies of stenotic kidneys in animal models have demonstrated a loss of microvascular structure and impaired angiogenic signaling.

• However, the study of these alterations in human poststenotic nephrectomized kidneys is lacking.

What this study adds?

• This is the first study to quantify and demonstrate microvascular rarefaction via three-dimensional micro-computed tomography images in poststenotic human kidneys.

• There is an overall loss of peritubular capillaries (PTCs) in poststenotic kidneys despite upregulation of newly formed PTCs in poststenotic kidneys, possibly due to an overwhelming hypoxic stress.

What impact this may have on practice or policy?

• This study affirms the role of microvascular alterations in stenotic kidneys, which may explain the difficulty in reversing kidney injuries in poststenotic kidneys.

• These observations support therapeutic strategies directly targeting the poststenotic kidney microcirculation in patients with RAS.

INTRODUCTION

Renovascular hypertension is an important cause of secondary hypertension, with atherosclerotic renal artery stenosis (RAS) constituting the most common etiology [1]. RAS leads not only to hypertension, but also to myriad physiologic and pathologic alterations in the stenotic kidneys. Reduced renal blood flow and increased tubular solute transport activity generate a progressively hypoxic milieu, which contributes to kidney injury [2–4]. Activation of the renin–angiotensin–aldosterone system (RAAS) results in vasoconstriction, increased oxidative stress [5] and release of inflammatory modulators from the stenotic kidneys [6].

Atherosclerosis further augments renal injury and aggravates tubulointerstitial fibrosis and glomerulosclerosis [7]. Animal studies have revealed prominent renal microvascular loss [3], which eventually become irreversible, suggesting persistent microvascular damage [8]. However, the extent of poststenotic microvascular alterations, angiogenic modulators and cellular changes in poststenotic human kidneys remains obscure. Few studies have directly examined tissue and microvascular changes comprehensively in human kidneys secondary to RAS, partly due to lack of techniques capable of reconstructing the intra renal microcirulation [9, 10]. Micro-computed tomography (CT) is a high-resolution imaging tool capable of three-dimensional (3D) visualization of microvascular structures in animal models [2] as well as assessment of treatment [11, 12]. However, given the challenges of procuring, perfusing and scanning dissected kidney samples in vitro, this tool has not yet been applied in poststenotic human kidneys, limiting characterization of their microvascular architecture.

A histopathological study in nephrectomized atherosclerotic RAS revealed predominant tubulointerstitial atrophy and relative glomerular sparing [10]. Upregulation of transforming growth factor-β (TGF-β) and macrophages in biopsies obtained from poststenotic kidneys [9] may partly explain these alterations.

Several systems might modulate microvascular density and vitality in the poststenotic kidney. The angiopoietin (Ang)-tyrosine kinase receptor (Tie) system includes proangiogenic mediators that primarily stabilize and promote vascular integrity. Ang-1 exerts different effects depending on the vascular beds and mechanism of injury; its upregulation promotes kidney regeneration in fatty acid–induced kidney injury [13]. Thrombospondin-1 (TSP-1), an anti angiogenic factor, participates in kidney diseases including glomerulonephritis, diabetic nephropathy and ischemia-reperfusion injury [14]. TSP-1 also binds and activates TGF-β, culminating in fibrosis [14]. The role of pericytes has been studied mainly in acute kidney injury models. Pericytes normally provide vascular structural support but might detach from vessels, transform into myofibroblast upon migration into the interstitial space and thence induce interstitial fibrosis [15]. Growth factors including vascular endothelial growth factor (VEGF) and Ang-1 favor sustenance of the microcirculation. β3-integrin regulates vascular development and response to VEGF [16] and is often expressed in newly formed or sprouting vessels [17]. Nonetheless, the effects of RAS on these angiogenic mediators are unknown, particularly in human poststenotic kidneys.

This study was designed to test the hypothesis that human kidneys distal to occlusive renovascular disease develop microvascular rarefaction and remodeling compared with relatively healthy kidneys. We applied micro-CT for 3D visualization and quantification of changes within the intra-renal microcirculation, and assessed the expression of major angiogenic cytokines that may be involved in these alterations.

MATERIALS AND METHODS

Patient selection

Kidneys were harvested between 2016 and 2019 from patients with severe occlusive RAS referred for unilateral nephrectomy and from deceased donors whose kidneys were deemed to be unsuitable for transplantation.

RAS patients were included in the study if they had severe or labile hypertension secondary to severe RAS and failed or were unable to tolerate medical or interventional therapy. The severity of RAS was confirmed by renal artery Doppler ultrasound and the function of each kidney by technetium-99m mercaptoacetyltriglyceride renal scan. The RAAS activity was assessed by plasma renin and aldosterone levels. Written informed written consent was obtained and the study was approved by Mayo Clinic Institutional Review Board (16-009485).

RAS patients underwent laparoscopic nephrectomy, whereas open nephrectomy was used in deceased donors. The surgeon attempted to preserve the renal artery with the sample to allow in vitro cannulation to inject contrast for the microvascular study. The surgically removed kidney was placed into a phosphate saline buffer to prevent cell death and tissue alterations. Discarded kidneys of otherwise heathy deceased potential donors were graciously provided by the nonprofit organ procurement organization, Lifesource (Minneapolis, MN, USA). These kidneys were processed similar to nephrectomized RAS kidneys and delivered to the lab on ice within 24 hours of harvest.

A slice of kidney containing cortex and medulla was placed in 10% buffered formalin and another stored at −80^○C.

Microvascular architecture

For micro-CT, renal segments were flushed with heparinized saline and then perfused with an intravascular contrast agent (Microfil, Flow-Tech, Carver, MA, USA) through a branch of each renal artery. Samples were then prepared and scanned as described [3, 8, 12]. Total, outer and inner cortical microvascular densities based on size (small, 20–200 µm;  medium, 200–300 µm;  large, 300–500 µm) and transcortical microvascular tortuosity were assessed using Analyze (Biodynamic Research Unit, Mayo Clinic, Rochester, MN, USA). The tortuosity of the entire length of transcortical tomographically isolated microvessels was calculated by the ratio of their path length and straight-line distance.

Renal protein expression

Frozen kidney samples were homogenized [18] and protein expression was assessed by western blotting. Protein concentration was measured using a BCA Protein Assay Kit (23225, Thermo Fisher Scientific, Waltham, MA, USA), following the manufacturer's instructions. Each protein (100 μg) was then loaded onto gels. Specific antibodies were used against VEGF (1:500, sc7269, Santa Cruz Biotechnology, Dallas, TX, USA), VEGF receptor-1 (FLK-1; 1:1000, sc504 Santa Cruz Biotechnology), Ang-1 (1:500, ab8451, Abcam, Cambridge, UK), Ang-2 (1:1000, ab65835, Abcam), Ang receptor (Tie-2; 1:1000, 610205, Sgma-Aldrich, St. Louis, MO, USA), TSP-1 (1:1000, 1823, Abcam) and hypoxia-inducible factor (HIF)-1α (1:200, 16066, Abcam).

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 1:5000, ab8245, Abcam) was initially evaluated as a loading control. However, RAS kidneys expressed less GAPDH than control kidneys, which could not be reliably corrected despite tissue rehomogenization or carefully equalizing protein loading; α-tubulin displayed a similar pattern. Indeed, ischemic condition can diminish GAPDH protein expression that may be a sub optimal loading control for western blotting [19]. Given the limitations of housekeeping genes that are variably expressed among different tissues and diseases [20], studies have proposed the alternate use of total protein staining for loading control [19, 21, 22]. We, therefore, used Coomassie G-250 (LC6060, Invitrogen, Waltham, MA, USA) stain for total protein as a loading control. Protein expression was quantified with Image-Pro Plus 6.0 software.

Renal gene expression

Frozen human kidney tissue (50 mg) was homogenized, total RNA isolated and its concentrations measured by a NanoDrop Spectrophotometer. Relative quantitative polymerase chain reaction (PCR) was performed using Taqman assays (Thermo Fisher), using primers for VEGF (Hs00900055), FLK-1 (Hs00911700), Ang-1 (Hs00919202), Ang-2 (Hs00169867), Tie-2 (Hs00945150), TSP-1 (Hs00962908) and HIF-1-α (Hs00153153). TATA binding protein (Hs00427620) was used as a reference control. Negative controls were cycled in parallel with each run. Analysis was done on QuantStudio7 Real-Time PCR systems (Applied Biosystems, Waltham, MA, USA). Fold-changes of gene expression in RAS were compared with the control group using the 2^–ΔΔCT method.

Renal histology

We performed immunohistochemistry staining for angiogenic factors to localize their spatial expression in the kidneys. We stained for VEGF (sc-7269, Santa Cruz Biotechnology), Ang-1 (ab8451, Abcam), Ang-2 (ab153934, Abcam), HIF-1α (ab16066, Abcam) and TSP-1 (ab267388, Abcam). Kidney fibrosis was assessed by Masson's trichrome staining in 5-µm sections of each kidney using the MATLAB 2015a program. Vascular endothelial cells were assessed by CD31 immunofluorescence staining (1:100, ab28364, Abcam) and the percentage of CD31⁺ area was quantified by ImageJ (National Institutes of Health, Bethesda, MD, USA). We also confirmed CD31 localization using immunohistochemistry staining (ab28364, Abcam). Because PTCs are below the resolution of micro-CT, they were identified as described [23] and counted per renal tubule (at 100×) on hematoxylin and eosin-stained tissue. PTC neovascularization was assessed as the number/tubule of PTC positive for the angiogenic endothelium marker β3-integrin (1:100, ab75872, Abcam) in immunohistochemistry. Seven to ten random fields of cortex and medulla were assessed and averaged for all staining.

Microvascular remodeling was assessed by α-smooth muscle actin (SMA; 1:100, ab7817, Abcam) immunohistochemistry. We calculated the media:lumen ratio (M:L) in five to seven random vessels per each cortex and medulla. We assessed vascular immaturity by nestin immunohistochemistry staining (33475, Cell Signaling Technology, Danvers, MA, USA). Pericytes were assessed by immunofluorescence and identified via platelet-derived growth factor receptor β(PDGFR-β; 1:100, LS-C338076, LifeSpan Bioscience, Seattle, WA, USA) positivity and myofibroblasts by costaining for α-SMA (1:100, ab7817, Abcam). Due to their close proximity, we used CD31⁺ (1:20, 28364, Abcam) to differentiate endothelial cells from pericytes.

Renal oxidative stress was evaluated by in situ production of superoxide anion in dihydroethidium (DHE; D11347, Life Technologies, Waltham, MA, USA) immunofluorescence staining in frozen sections (30 µm). We quantified DHE positivity in cortical and medullary areas compared with positivity of DAPI.

Statistical analysis

Data are displayed as median (range) for continuous variables and number (%) for categorical variables. The normality assumption was tested by the Shapiro–Wilk test. Differences were compared between groups by either unpaired two-sample t-test or Wilcoxon rank-sum test for continuous variables and chi-squared or Fisher's exact test for categorical variables, considering type 1 error (α) ≤ 0.05 to be statistically significant. We utilized the JMP software package (version 14.1.0; SAS, Cary, NC, USA) for statistical analyses.

RESULTS

Demographic data

Between 2016 and 2019 we recruited five RAS patients (five kidneys) who underwent nephrectomy due to occlusive renovascular disease and four deceased donors (seven discarded kidneys) became available. The median age was 57 years (range 52–66) in the donor group and 63 years (range 58–71) in the RAS group (P = 0.10). Most of the patients were female and Caucasian (Table 1). Half of the donors and all RAS subjects were hypertensive, with no statistical differences in the rate of hypertension and diabetes between the two groups. There was no difference in serum creatinine (SCr) between the two groups (P = .807), but it was noted that one deceased potential donor had an SCr of 2.33 mg/dL prior to nephrectomy.

Table 1.

Baseline characteristics of RAS patients and deceased donors

Characteristics	Donors (n = 4)	RAS (n = 5)	P-value
Age (years)	57 (52–66)	63 (58–71)	.100
Male, n (%)	2 (50)	1 (20)	.524
BMI (kg/m2)	26.72 (20.08–27.03)	25.66 (22.23–32.66)	.766
Hypertension, n (%)	2 (50)	5 (100)	.167
Diabetes, n (%)	1 (25)	1 (20)	1.00
Hyperlipidemia, n (%)	0 (0)	5 (100)	<.001
Smoking, n (%)	1 (25)	4 (80)	.21
SCr (mg/dl)	1.30 (0.76–2.33)	1.4 (0.80–1.90)	.807
BUN (mg/dl)	22 (16–25)	23 (10–43)	1.00

Data presented as median (range) unless stated otherwise.

BMI, body mass index; BUN, blood urea nitrogen.

RAS had a median duration of hypertension of 9 years (1–40) (Supplementary data, Table S1). All patients had elevated plasma renin activity ranging between 4.8 and 94 ng/ml/h (normal 0.6–3.0 ng/ml/h) and plasma aldosterone ranging between 8.9 and 21 ng/dl (normal ≤21 ng/dl). After surgery, 2/5 (40%) RAS patients were able to discontinue antihypertensive treatment, 2/5 (40%) had a significant reduction in antihypertensive treatment and 1/5 (20%) had persistent hypertension (Supplementary data, Table S1).

Deceased donor kidneys were discarded due to biopsy-determined fibrosis (four kidneys) and high pressure during pulsatile pump perfusion without (n = 1) or with (n = 2) aortic/arterial plaque. The median kidney donor profile index was 94% (range 62–95). All were donated after cardiac death.

Microvascular architecture

Renal arteries of two donor and one RAS kidneys could not be successfully cannulated for technical reasons. Therefore, five donor and four RAS kidneys were included in the micro-CT analysis. Vascular density was analyzed based on size (small, medium and large) and cortical region (total, outer or inner cortical area) and is shown in Fig. 1A. Individual micro-CT images and movies demonstrating 3D reconstruction of renal microvessels are shown in Supplementary data, Figures S1–9 and Videos S1–2.

FIGURE 1:

(A) Micro-CT 3D images of microvascular density and tortuosity in nephrectomized donor and RAS kidneys. (B) Spatial densities of medium-sized vessels (0.2–0.3 mm) were significantly lower in RAS versus donor kidneys and those of large-sized vessels tended to be lower as well (P = 0.065). In contrast, spatial densities of small-sized vessels did not differ between RAS and donor kidneys in the whole (P = 0.539), inner (P = 0.902) or outer cortex (0.110). (C) Vascular tortuosity was greater in RAS kidneys; *P ≤ 0.05 versus donors.

RAS kidneys showed lower vascular densities of medium-sized vessels (200–300 µm) than donor kidneys in the total (0.028 versus 0.165; P = 0.037), inner (0.028 versus 0.130; P = 0.020) and outer (0.028 versus 0.130; P = 0.020) cortex (Fig. 1B). RAS kidneys also tended to have diminished densities of large-sized vessels (300–500 µm) compared with donor kidneys in the total (0.010 versus 0.056; P = 0.065), inner (0.014 versus 0.083; P = 0.065) and outer (0.006 versus 0.030; P = 0.065) cortex. In contrast, there were no statistically significant differences between RAS and donor kidneys in densities of small-sized vessels (20–200 µm) in the entire cortex (P = 0.539), inner (P = 0.902) or outer cortex (P = 0.110) (Fig. 1B).

Furthermore, RAS kidneys also had greater microvascular tortuosity than donor kidneys (1.45 versus 1.11; P = 0.05) (Fig. 1C), suggesting vascular immaturity.

Renal proangiogenic protein and gene expression

Compared with donor kidneys, RAS kidneys showed greater protein expression of Ang-1 (P = 0.021), TSP-1 (P < 0.001) and HIF-1α (P < 0.001) but significantly lower VEGF (P < 0.001) expression (Fig. 2A,B). The immunohistochemistry staining of VEGF, Ang-1, TSP-1 and HIF-1α showed findings similar to kidney protein expression. VEGF was expressed primarily in podocytes of donor kidneys, whereas Ang-1 was mostly expressed in interstitial and fibrotic glomeruli in RAS kidneys. In RAS kidneys, TSP-1 was mainly expressed in peritubular capillaries and HIF-1α in podocytes and interstitial cells (Supplementary data, Figure S8). FLK-1 tended to be lower in RAS kidneys, but this did not reach statistical significance (P = 0.084). In contrast, there was no difference in protein expression of Ang-2 (P = 0.666) or Tie-2 (P = 0.160). Interestingly, Ang-2 showed greater expression in sclerotic glomeruli and tubular cells of RAS than donor kidneys via immunohistochemistry staining (Supplementary data, Figure S8). Average housekeeping proteins assessed by Coomassie were not different between the groups (Fig. 2D).

FIGURE 2:

(A) Expression of angiogenic proteins in donor and RAS kidneys. (B) Poststenotic kidneys had greater Ang-1, TSP-1 and HIF-1α protein expression but lower VEGF protein expression compared with donor kidneys. In contrast, there were no differences in VEGF receptor-2 FLK-1, Tie-2 or Ang-2 protein expression between the groups. (C) Coomassie blue total protein stains of donor and RAS kidneys. (D) Total protein loading (expressed by average relative fluorescence units) did not differ between donor and RAS kidneys. *P ≤ 0.05 versus donors.

FIGURE 3:

Angiogenic gene expression (by PCR) in RAS and donor kidneys. RAS kidneys had greater Ang-2 and TSP-1 gene expression than donor kidneys. In contrast, there was no difference in gene expression of VEGF, FLK-1, Ang-1, HIF-1α or Tie-2 between the groups. *P ≤ 0.05 versus donors.

Gene expressions of VEGF (P = 0.270), FLK-1 (P = 0.540), Ang-1 (P = 0.470), HIF-1α (P = 0.637) and Tie-2 (P = 0.637) did not differ between RAS and donor kidneys, whereas Ang-2 was upregulated in RAS kidneys.

Renal histopathology

We also assessed renal vascularity by CD31 (immunoreactivity), which was allocated to endothelial cells (immunohistochemistry). RAS kidneys showed significantly lower CD31 immunoreactivity area than donor kidneys in both the medulla (P = 0.018) and cortex (P = 0.006) (Fig. 4A, B). Similarly, RAS kidneys demonstrated significant reductions in the number of PTCs per tubule in both cortex (P = 0.006) and medulla (P = 0.011) (Fig. 4C, D).

FIGURE 4:

(A) Renal CD31 endothelial staining (gray, nickel). (B) The percentage of CD31⁺ staining was lower in RAS compared with donor cortex (upper panel) and medulla (lower panel). (C) Representative images demonstrating PTCs (yellow arrows) in the cortex and medulla of both groups. (D) RAS has fewer cortical and medullary PTCs than donor kidneys. *P ≤ 0.05 versus donors.

Masson's trichrome demonstrated more extensive fibrosis in RAS compared with donor kidneys in both the medulla (P = 0.011) and cortex (P = 0.006) (Fig. 5A, B). Microvascular remodeling was assessed by the M:L ratio using α-SMA staining. RAS had a significantly greater M:L ratio than donor kidneys in both the cortex (P = 0.0057) and medulla (P = 0.0107) (Figure 5C, D), indicating microvascular wall thickening. Moreover, nestin, a vascular immaturity marker, showed higher positivity in small vessels in RAS than in donor kidneys (Supplementary data, Figure S10).

FIGURE 5:

(A) Representative images of renal fibrosis assessed by Masson's trichrome in the cortex and medulla of donor and RAS kidneys. (B) Renal fibrosis was more extensive in RAS versus donor renal cortex (upper) and medulla (lower). (C) Representative images of α-SMA staining for assessment of the M:L ratio in RAS and donor kidneys. (D) RAS had greater microvascular wall thickening compared with donor kidneys. *P ≤ 0.05 versus donors.

Pericytes were assessed by PDGFR-β⁺ immunofluorescence. RAS kidneys had more numerous PDGFR-β⁺ cells than the donor cortex (P = 0.002) and medulla (P = 0.006). However, differences in PDGFR-β⁺/α-SMA⁺ double-positive pericytes between RAS and donor kidneys showed only statistical trends in both the cortex (P = 0.061) and medulla (P = 0.095) (Fig. 6A, B). PTC neovascularization was assessed by the number of β3-integrin⁺ PTCs per tubule. Despite microvascular loss, RAS kidneys had a greater number of newly formed PTCs than donor kidneys both in the cortex (P = 0.030) and medulla (P = 0.047) (Fig. 6C, D).

FIGURE 6:

(A) Representative images of PDGFR-β⁺/CD31^– renal pericytes (white arrows). Some pericytes may transform into myofibroblast that coexpress α-SMA and PDGFR-β. (B) RAS kidneys had a greater number of pericytes than donor kidneys, which were more frequently located in the interstitium. (C) RAS kidneys tended to have a greater number of cortical (P = 0.061) and medullary (P = 0.095) pericyte-derived myofibroblast (PDGFR-β⁺/α-SMA⁺) than donor kidneys, which did not reach statistical significance. (D) Representative images of newly formed β3-integrin⁺ PTCs (yellow arrows). (E) RAS kidneys had a greater proportion of β3-integrin⁺ PTCs/tubules than donor cortex and medulla. *P ≤ 0.05 versus donors.

RAS kidneys also had a markedly elevated DHE:DAPI ratio compared with donor kidneys in both the cortex and medulla (P = 0.006 for both), consistent with increased oxidative stress (Fig. 7A, B).

FIGURE 7:

Oxidative stress assessed by in situ production of superoxide anion. (A) Representative images demonstrating cortical and medullary DHE stain in donor and RAS kidneys. (B) RAS had higher oxidative stress than donor kidneys in both the cortical and medullary zones. *P ≤ 0.05 versus donors.

DISCUSSION

This study demonstrates that microvascular loss and remodeling in human poststenotic kidneys are associated with altered angiogenic signaling, tissue oxidative stress and fibrosis. Evidence of angiogenic activity and new vessel formation observed in poststenotic kidneys was not sufficient to sustain the intra renal microcirculation as assessed by both micro-CT and renal histology. These findings underscore the extent of distal intra renal microvascular disease beyond main vessel stenosis that may potentially account for the limited ability of revascularization strategies to restore poststenotic kidney function.

The prevalence of RAS increases with age and with identified vascular disease [24]. As a result of recent prospective trials that failed to identify improved clinical outcomes after renal revascularization for RAS [25, 26], more individuals with RAS are subjected to reduced renal blood flows from large-vessel occlusion than ever before.

Our study was the first to take advantage of micro-CT to assess the human intra renal microvascular architecture. This technique provides 3D images of the microcirculation, thereby allowing quantification of location- and size-based microvascular density as well as vascular tortuosity [27]. We observed a distinctive decrease in the density of medium-sized (0.2–0.3 mm),  corresponding to small interlobar and large arcuate arteries [28, 29], as well as in small-sized (0.02–0.2 mm) vessels in stenotic compared with control kidneys. These data extend observations from a swine RAS model in which we identified the loss of small-sized vessels, which include arcuate, interlobular, afferent and efferent arterioles [30]. The difference in the smallest vessel findings might reflect the limited sample size in our study but likely also reflected vascular risk factors observed in some of our controls, including aging, hypertension and diabetes, conditions that are recognized to modify the number of small-sized vessels in kidneys. We further found that RAS kidneys were also characterized by a diminished number of PTCs compared with donor kidneys, supported by the loss of CD31⁺ vascular endothelial cells in stenotic kidneys. Furthermore, greater vascular tortuosity in RAS kidneys was consistent with vascular immaturity and remodeling. Vascular remodeling was underscored by a higher M:L ratio (wall thickening) in RAS compared with donor kidneys and vascular immaturity by upregulation of nestin [31] in RAS compared with donor kidneys.

We interpret these findings to likely reflect worsening parenchymal hypoxia beyond a stenotic lesion that interferes with cellular functions. Hypoxic kidneys promote PTC loss, eventually leading to renal fibrosis and ultimately chronic kidney disease [32]. The hypoxic milieu upregulates HIF-1α protein [33], as we found in the stenotic kidneys, which in turn has been linked to tubulointerstitial injury [34], interstitial fibrosis [35] and glomerular sclerosis. Immune cells also accumulate in stenotic kidneys, indicating the role of T-cells and mature dendritic cells in the pathophysiology of RAS [36], whereas a decrease in levels of microRNA-26A promotes cellular differentiation and apoptosis [37].

Angiogenic factors are dysregulated during ischemic events and contribute to functional and anatomical changes associated with acute and chronic kidney injury [13, 24, 38]. We found downregulation of VEGF protein expression in poststenotic kidneys, whereas its gene expression was not different from donor kidneys. Although HIF-1α stimulates VEGF expression under acute hypoxic conditions, its expression is downregulated under chronic ischemia [38], because prolonged oxidative injury can degrade VEGF and increase chronic tissue inflammation in RAS kidneys [39]. This is consistent with previous studies in swine RAS demonstrating downregulated tissue expression of VEGF and VEGF receptor-2, possibly due to oxidative stress [3]. Indeed, we observed increased tissue oxidative stress in RAS kidneys. Moreover, besides endothelial cells, VEGF is typically present in podocyte and tubular epithelial cells [40], which are often lost in RAS and may contribute to the loss of VEGF expression [41–44]. In addition, chronic hypoxia can reduce angiogenic response to VEGF via downregulation of VEGF receptor [45].

The poststenotic kidneys exhibited upregulated protein expression of Ang-1, a chemokine secreted from stromal or endothelial cells that regulates endothelial cell proliferation. It binds to Tie-1 or Tie-2 to promote angiogenesis [46], a pathway particularly important for the later stage of vascular development. Ang-1 upregulation might be secondary to the increase in the number of pericytes, which can produce Ang-1 [47], and promotes recovery of endothelial cell permeability and stability [48]. Elevated Ang-1 levels in poststenotic kidneys might represent a compensatory response to counteract ischemic insults [50], as it preserves PTCs in ischemia-reperfusion acute kidney injury [51]. Indeed, despite capillary loss in poststenotic kidneys, we observed more newly formed (β3-integrin⁺) PTCs in poststenotic kidneys, consistent with active angiogenesis. Nevertheless, ischemic injury and vascular loss remained evident in poststenotic kidneys, possibly due to upregulation of anti angiogenic factors such as TSP-1. TSP-1 is linked to several kidney diseases and is a major activator of TGF-β and pro apoptotic factors in ischemic renal injury [14]. Overall, these findings suggest that angiogenic stimuli in poststenotic kidneys, including Ang-1, fail to overcome other processes that obliterate small vessels, possibly related to downregulation of VEGF, increased TSP-1 and oxidative stress.

Interestingly, we found abundant pericytes in poststenotic compared with donor kidneys. Pericytes are located adjacent to vascular endothelial cells and support the development and stabilization of the renal vasculature. They express Ang-1 but not Ang-2, Tie-1 or Tie-2, suggesting a role in vessel maturation [47]. In contrast, pericytes can foster renal fibrosis by migrating into the interstitium and transdifferentiating into myofibroblasts to exert profibrotic and proinflammatory effects [52], as demonstrated in murine RAS [53]. Ultimately these lead to cortical and medullary microvascular rarefaction [3, 54, 55]. However, a trend for an increase in α-SMA⁺ pericytes in our study did not reach statistical significance. While this might be due to our small sample size, it might indicate that occlusive RAS fosters primarily pericyte detachment from microvessels, again consistent with vascular injury, rather than their transformation to myofibroblasts.

Our study has some limitations, including a small sample size, yet represents a comprehensive nonautopsy study of the renal microcirculation in nephrectomized human poststenotic kidneys. It should be noted that, RAS patients uncommonly undergo nephrectomy to control blood pressure, given the availability of effective medical and interventional strategies [1]. Likely, less severely affected RAS kidneys exhibit milder changes. While some histological features in the nephrectomized kidneys may theoretically result from systemic hypertension, the stenotic kidney is in fact relatively protected from its effect by virtue of the renal arterial occlusion. Furthermore, murine unilateral RAS stenotic and contralateral kidneys demonstrate histological and angiogenic differences despite the same systemic hypertension [56]. In addition, despite strict screening, deceased donor kidneys may not be ideal controls, as some donors exhibited risk factors that might induce small-vessel loss. Furthermore, identifying vessels by size is an approximation. Tumor nephrectomy samples often do not include an artery for adequate cannulation and Microfil injection. Moreover, they also demonstrate chronic changes and thus represent suboptimal controls [57]. Nevertheless, we successfully identified important features that characterize and distinguish poststenotic from control kidneys. Indeed, our study presents a number of important translational findings that extend previous observations from animal models to human kidneys [3, 58].

In conclusion, our observations underscore the nature of RAS as a small-vessel disease distal to the main vessel occlusion and may support therapeutic strategies directly targeting the poststenotic kidney microcirculation [59]. For example, inducing angiogenesis via stem cell [60] or VEGF [38] delivery improves renal hemodynamics and outcomes even without relieving the main vessel obstruction. Our work provided important insights into human stenotic kidneys that constitute a translational bridge for preclinical data. Future studies are needed to assess the ability of such interventions to restore the microcirculation distal to RAS.

FUNDING

This work was partly supported by the National Institutes of Health (grants DK120292, DK122734 and AG062104).

AUTHORS’ CONTRIBUTIONS

N.K., A.S.Pawar, A.S.Puranik, S.C.T. and L.O.L. conceptualized the study. A.S.Pawar, A.S.Puranik, X.Y.Z., S.C. and S.R.K. collected and processed samples. C.M.F. performed micro-CT and analyzed and interpreted the results. H.T., K.J. and I.S. performed the experiments. N.K. collected data, interpreted results and performed the statistical analysis. N.K., A.L., J.P.G., S.C.T., A.E.M. and L.O.L. drafted and revised the manuscript.

CONFLICT OF INTEREST STATEMENT

L.O.L. is an advisor to AstraZeneca, Butterfly Biosciences and CureSpec. No author has a relevant financial conflict of interest.