, Introduction
Diabetic nephropathy (DN) is a common microvascular complication of diabetes, being the leading cause of end-stage kidney disease worldwide [1,2]. Despite the enormity of the problem on a global scale, our current therapeutic options for DN are limited and the underlying mechanisms remain unclear.
Pericytes are a heterogeneous group of extensively branched cells located in microvessels. They express a wide range of surface markers, none of which is specific in isolation. Moreover, expression of cell surface markers is dynamic, changing with the developmental state and in pathological conditions. Proteoglycan nerve/glial antigen 2 (NG2) is recognized as an activated pericyte marker.
Among other functions, pericytes stabilize blood vessels, regulate vascular tone [3], synthesize matrix, participate in repair, and serve as progenitor cells [4,5]. Pericyte–endothelial cell crosstalk is essential in both remodeling and quiescent vasculature; this complex interaction often being disrupted by disease states [6].
Activated pericytes, which have been found to accumulate after renal injury, are associated with reduction of pericyte vascular coverage or detachment from endothelial cells linked to microvascular abnormalities, leading to enhanced insights into vascular regulation in renal disease. There is mounting evidence that pericyte dysfunction plays a critical role in many microvascular diseases [7], including diabetes-related pathologies [8].
DN is characterized by various pathological alterations within glomeruli and the tubular system, leading to disruption of filtration barriers and proteinuria [9]. The pathological features include glomerulosclerosis, mesangial matrix expansion, glomerular basement membrane thickening, podocyte loss, and interstitial fibrosis [9–11]. Understanding the mechanisms of fibrosis in the kidney is paramount to development of new therapeutics to counteract progression of diabetes-related kidney disease.
Recent studies [12–14] have highlighted the role of pericytes in fibrosis. Chronic injury triggers pericyte proliferation and differentiation into collagen-secretory, contractile myofibroblasts that migrate away from vessels, causing microvascular rarefaction. In nondiabetic murine kidney injury models, myofibroblasts that have originated from pericytes have been shown to add substantially to extracellular matrix formation, vessel destabilization, and fibrogenesis [3].
Nowadays, it is recognized that changes in pericyte biology are directly associated with biochemical changes in diabetes, changes that result in diffuse microvascular complications. However, the properties of pericytes in diabetes-related kidney disease have been poorly investigated. Here, we focused on whether activated pericytes and the subsequent peritubular capillary dysfunction and pericyte–myofibroblast transition is associated with renal fibrosis in DN.
Methods
Patients
The study cohort comprised patients (n = 13) who had been diagnosed with DN by biopsy in Beijing Friendship Hospital from July to December 2019. The control group comprised 12 patients diagnosed with minor glomerular lesions without foot fusion of podocytes. These patients’ baseline clinical characteristics are shown in Supplementary Table 1 (available online).
This study was approved by the Bioethics Committee of Beijing Friendship Hospital, Capital Medical University (No. 2023-P2-017-01). We obtained written informed consent to inclusion in the study from all participants.
Animals
Eight-week-old db/db mice (male, 34–37 g) and their nondiabetic db/m littermates (8-week-old, male, 18–22 g) were purchased from Changzhou Cavens Laboratory Animals (license key: SCXK [Su] 2021-0013). The db/db mice were hyperglycemic when tested after arrival in our laboratory at 8 weeks of age (mean blood glucose concentration, 23.92 ± 0.87 mM/L).
All mice were kept in a specific pathogen-free environment and fed adaptively (ambient temperature, 20–24 °C; relative humidity, 50%–55%; light cycle, 12 hour–12 hour; free food and water). The mice were euthanized at week 24 and samples (serum and kidney) were obtained for subsequent investigation.
The animal protocol conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health and was approved by the Animal Experimentation Ethics Committee of Beijing Friendship Hospital, Capital Medical University (No. 21-2023).
Sample collection
The body weight of each mouse was recorded weekly. Blood was obtained by tail venipuncture and blood glucose was measured with One Touch Ultra Test Strips (Yuyue). During Week 24, 24-hour urine samples were collected using metabolic cages and used to calculate the albumin-to-creatinine ratio. Concentrations of blood urea nitrogen, plasma and urinary creatinine, and urine microalbumin were measured using an automatic analyzer (AU5800; Meikangbaitai).
Histology
Kidney tissues were fixed in 10% formaldehyde. Paraffin-embedded kidney tissues were cut into 4-µm-thick sections and then stained with hematoxylin and eosin (H&E), periodic acid-Schiff (PAS), and Masson’s trichrome for light microscopy (Olympus).
Measurements of mesangial expansion, calculated as a ratio of mesangial to glomerular area, and percentages of Masson’s trichrome-positive areas, an indicator of the severity of renal fibrotic lesions, were obtained by using Image J software. For each sample, we analyzed five randomly selected nonoverlapping fields. All sections were examined by two pathologists in a blinded manner.
Kidney tissues were fixed with 2.5% glutaraldehyde and transmission electron microscopy (JEM-1400 plus; Jeol) was used to identify detachment of pericytes from endothelial cells.
Immunohistochemical staining
Kidney tissue was fixed with 4% paraformaldehyde. Next, 3-µm-thick sections cut from paraffin-embedded tissue were deparaffinized, rehydrated, and antigen retrieved, then incubated with primary antibodies (CD34 ab81289 and platelet-derived growth factor receptor beta [PDGFR-β) ab91066; both Abcam) overnight at 4 °C. The secondary antibodies tested were horse-radish peroxidase-conjugated goat anti-rabbit antibodies (ImmunoReagents). All sections were examined using an Eclipse 80i microscope (Nikon). Immunohistochemistry staining was analyzed using Image J software to calculate the percentages of positive areas. Scoring was evaluated by a “blinded” investigator on coded slides. At least eight fields were selected randomly for photo documentation.
Multiplex fluorescent immunohistochemistry and multispectral imaging
Multiplex immunofluorescence staining was performed using a PANO 7-plex Kit (0004100100; Panovue). The following primary antibodies were applied sequentially: NG2 (ab259324; Abcam), cluster of differentiation 31 (CD31) (CST77699; Cell Signaling Technology), alpha-smooth muscle actin (α-SMA) (ab5694; Abcam), collagen I (CST72026S; Cell Signaling Technology), and PDGFR-β (ab32570; Abcam), followed by horse-radish peroxidase-conjugated secondary antibody incubation and tyramide signal amplification. The slides were microwave-heat-treated after each application of trichostatin A. Nuclei were stained with 4′-6′-diamidino-2-phenylindole (DAPI) (D9542; Sigma-Aldrich) after all the antigens had been labeled. The stained slides were scanned using an PDone VS200 slide scanner (Panovue) and analyzed with OlyVIA software (Olympus). Whole-slide scanning of multispectral fluorescent images was performed using an Olympus VS200 MTL (Olympus), in conjunction with an Olympus UPLXAPO20X objective lens. Whole-slide bright field and epifluorescence images were analyzed using QuPath software (Queen’s University of Belfast).
The positive-stained cells were quantified using 200× the total magnification. In each sample, we calculated at least five randomly selected nonoverlapping fields in the cortex and medulla.
Statistical analysis
Data are presented as the mean ± standard error of the mean or the median with interquartile range. IBM SPSS version 23 (IBM Corp.) was used for statistical analysis, and GraphPad Prism 9.0 (GraphPad Software, Inc.) to generate graphs. Data with a normal distribution were compared between groups using the Student t test. Non-normally distributed data were analyzed using a nonparametric test. Statistical significance was set at p < 0.05.
Results
Basic physiological variables and findings on stained sections of kidney tissues from db/db and db/m mice
The db/db mice had higher blood glucose concentrations and were heavier than db/m mice (Table 1). The db/db mice showed a poor renal outcome, as shown by a significantly higher urinary albumin-to-creatinine ratio, plasma creatinine and blood urea nitrogen concentrations than in db/m mice (Table 1).
H&E (Fig. 1A, B) and Masson’s trichrome (Fig. 1C, D) staining of tissue from db/db mice showed a paucity of the normal tubular back-to-back structure with focal interstitial fibrosis at week 24 compared with the findings in db/m controls. The diabetic mice had larger glomeruli with increased mesangial matrix and the glomerular capillary basement membranes appeared thickened on PAS-stained sections (Fig. 1E, F).
Interstitial renal fibrosis in db/db mice
Collagen I staining was performed to visualize fibrosis in the renal interstitium. Occasional collagen I-producing cells were found along peritubular capillaries and in the vasa recta in db/m mice (Fig. 2A, C), whereas greater collagen I-positive areas were observed in the cortical interstitial of db/db mice (Fig. 2B). Additionally, there was abundant deposition of collagen I in the interstitial peritubular capillary spaces in the renal medullas of db/db mice (Fig. 2D). Collagen I protein staining was not detected in glomeruli. Collagen I staining was generally clearly extracellular when DAPI was used to identify nuclei. When quantitated, the area ratio of collagen I was greater in 24-week-old db/db mice than in db/m mice in the cortex and medulla of the kidney (Fig. 2E, F).
Pericytes activated in db/db mouse kidneys
In the present study, we used the following three criteria to identify activated pericytes: 1) cells were stained for NG2+ and PDGFR-β+ in kidney tissues; 2) cells were located in perivascular niches; and 3) cells surrounded CD31+ endothelial cells.
In the cortex of the db/m group, there were fewer pericyte-positive areas, and intensity was lower around peritubular renal capillaries than in the db/db group (Fig. 3A). In contrast, abundant pericytes were activated in the cortex of the kidney in db/db mice and surrounded afferent arterioles (d in Fig. 3B,) and peritubular capillaries (e and f in Fig. 3B). As expected, compared with the control group (Fig. 3C), there were more activated pericytes in the medulla of the kidney in db/db group, where they surrounded peritubular capillaries (Fig. 3D). Moreover, activated pericytes in db/db mice did not completely cover endothelial cells (j–l in Fig. 3D), consistent with a disruption in the pericyte–endothelial cell interaction. Quantitative analysis showed that activated pericytes were more numerous in the cortex (Fig. 3E) and medulla (Fig. 3F) of the kidneys in db/db mice than in db/m mice.
Peritubular capillary dilation in kidneys of diabetic mice
In the cortex of kidneys of db/m mice, CD31+ endothelial cells formed dense networks (Fig. 4A), with several peritubular capillaries wrapped around each tubule, and activated pericytes were occasionally observed (c in Fig. 4A). In contrast, dilated, abnormally shaped, peritubular capillaries were observed in the renal cortex (Fig. 4B). Moreover, higher magnification showed that the accumulation of activated pericytes (d in Fig. 4B) and dilated vessels typically lacked pericyte support and had an abnormal shape (e and f in Fig. 4B). Unlike in db/m mice (Fig. 4C), dilated peritubular capillaries were present in the renal medulla of db/db mice, and accumulated pericytes and decreased pericyte coverage were observed (j–l in Fig. 4D).
Examination by electron microscopy also revealed much more severe detachment of pericytes from endothelial cells in db/db than in db/m mice (Fig. 5).
Peritubular capillary loss with activation of PDGFR-β in diabetic nephropathy patients
Immunohistochemical semiquantitative analysis of kidney biopsy samples showed that compared with expression in patients with minor glomerular lesions (Fig. 6A), expression of CD34, which is a marker of endothelial cells, was weaker in the renal interstitium in DN patients (Fig. 6B). In contrast, compared with the control group (Fig. 6C), the expression of PDGFR-β, which is a pericyte marker, was stronger in the interstitial peritubular capillary spaces in DN patients (Fig. 6D).
Activated pericytes are profibrotic in diabetic nephropathy, differentiating into myofibroblasts
Pericytes are considered a possible source of fibrosis in chronic renal diseases. To determine whether pericytes differentiate into myofibroblasts in DN, pericyte markers (NG2 and PDGFR-β) were co-stained for α-SMA to mark myofibroblasts. In db/m mice, α-SMA staining was rarely detected in peritubular pericytes and always in vascular smooth cells of renal arterioles (Fig. 7A). Most pericytes did not express α-SMA (Fig. 7B, C). In db/db mice, there were more numerous α-SMA–expressing cells in regions of tubulointerstitial fibrosis (Fig. 7D). Overlapping staining for α-SMA and pericyte markers confirmed tubulointerstitial fibrosis (Fig. 7E, F).
We found that NG2+/PDGFR-β+/α-SMA+ cells were much more numerous in db/db mice than in db/m mice in both the cortex and the medulla of the kidney (Fig. 7G, H). This finding was accompanied by a lower number of NG2+/PDGFR-β+/α-SMA– cells in db/db mice (Fig. 7G, FH), which suggested a transition of these cells towards α-SMA+ myofibroblasts. However, fate-mapping studies will be required to confirm this possibility.
Discussion
In the current study, we demonstrated that pericytes in the kidney are activated, accompanied by greater NG2/PDGFR-β-positivity, in db/db mice. Activated pericytes are linked with impaired pericyte coverage of blood vessels and peritubular capillary dilation in the renal interstitia of both diabetic mice and DN patients. An accumulation of activated pericytes co-expressing α-SMA was also observed, suggesting that differentiation of these cells into myofibroblasts contributes to renal fibrosis in db/db mice with DN.
The precise mechanisms underlying the pathogenesis of DN are unknown. It has been suggested that proteinuria, genetic factors, ischemia, inflammation, and ultimately tubulointerstitial renal fibrosis contribute to kidney injury and disease pathogenesis [15]. Previous studies have implicated pericytes in the development of fibrosis and pathogenesis of DN. Protein kinase C activation is central to pericyte dysfunction during the pathogenesis of DN [16]. Koya et al. [17] further implicated pericyte-like cell dysfunction by showing protein kinase-beta inhibition in a mouse model of type 2 diabetes. Protein kinase-beta activation induces the formation of reactive oxygen species, resulting in the expression of transforming growth factor beta, which promotes excess extracellular matrix deposition and fibrosis.
The absence of a specific marker makes pericytes difficult to identify. Their anatomic locations, morphological and structural features, and the absence of endothelial or hematopoietic markers should be considered when exploring how to robustly identify these cells. In the current study, we used multiplex fluorescent immunohistochemistry and multispectral imaging to resolve this difficulty and accurately identify pericytes in mouse kidneys.
Renal pericytes have been identified around terminal arterioles, capillaries, and postcapillary venules. In our study, we consistently identified pericytes within the renal tubular system and vasa recta capillaries.
Our first major finding was that NG2/PDGFR-β-positive cells (activated pericytes) were significantly more numerous in the interstitial of the kidneys of diabetic mice than in the control group. NG2 has been recognized as a marker of activated pericytes in the kidney [18] and they have been shown to accumulate after renal injury. NG2 not only serves as one of the most reliable markers of pericytes but also plays important roles in the recruitment of pericytes and interaction with endothelial cells during the development of microvessels. Zhang et al. [19] demonstrated that luseogliflozin treatment decreases the area of positivity for NG2, suggesting that luseogliflozin ameliorates ischemia reperfusion-induced renal capillary rarefaction and detachment of pericytes from endothelial cells, both of which have been implicated in the development of interstitial renal fibrosis.
The mechanisms underlying the upregulation of NG2 staining in the diabetic kidney have not been well characterized. We speculate that the possible course of events in diabetes is pericyte detachment secondary to pericyte activation, triggering peritubular capillary dysfunction and subsequent pericyte-to-myofibroblast transition and renal fibrosis.
Peritubular capillary instability induced by activated pericytes is a feature of vascular pathologies [20]. In diabetic retinopathy, which features the same microvascular findings as DN, pericyte dysfunction and the subsequent microvascular disorder have been widely reported to contribute to microvascular abnormalities [21–23].
In this study, there was a noteworthy correlation between activated pericytes and peritubular capillary dilation. Pericytes are obligatory constituents of microvessels and serve as regulators of vascular development, stabilization, maturation, and remodeling. They regulate not only capillary integrity but also the size of microvessels via regulation of vessel constriction and relaxation [24,25]. It has become clear that activated pericytes are likely to induce disease progression by causing vasoconstriction and basement membrane thickening. It is interesting to note that when pericytes are disrupted during nephrogenesis by targeting Foxd1 transcription factor, this also results in abnormal vascular patterning and dilated capillaries [26]. During angiogenesis, vessel maturation, and stabilization, endothelial cell and pericyte crosstalk are essential to forming and maintaining a functional vasculature [27,28]. Several factors affect the interactions between endothelial cells and pericytes in pathological states, including changes in advanced glycation end products, increased adhesion, and inflammatory cytokines [29,30].
Activated pericytes are believed to be an early hallmark of diabetes-associated microvascular disease, causing abnormal and inadequate neovascularization in diverse vascular beds. Chronic high blood glucose concentrations particularly affect the state of renal pericytes. Activated pericytes contribute to altered interactions with endothelial cells, causing the migration of peritubular pericytes away from capillaries into the interstitial space. This destabilizes microvessels, resulting in peritubular capillary dysfunction. Pericyte detachment from endothelial cells can also be accompanied by their differentiation into fibroblasts and myofibroblasts, a state that favors matrix production, leading to the pathology of DN. Preservation of peritubular capillaries is of great importance in kidney disease: normalization of vasculature and restoration of pericytes can prevent fibrosis.
The other major finding of this study was that activated pericytes that co-express α-SMA are substantially more numerous in diabetic kidneys. We suggested that activated pericytes accompanied by pericyte–myofibroblast transition may contribute to renal fibrosis in db/db mice.
Activated pericytes have been identified as the progenitors of fibroblasts that contribute to the deposition of extracellular matrix [13,31]. Pericyte activation and detachment from capillaries contribute to renal fibrosis and thus to kidney diseases [32–35] because pericytes are largely responsible for the laying down of pathological fibrotic matrix in the interstitial spaces between capillaries and tubules. Several studies [33,36] have found that once kidneys have been damaged, pericytes no longer adhere to the endothelium, thereafter migrating, proliferating, and finally differentiating into myofibroblasts [33]. These studies have demonstrated that pericytes can differentiate into myofibroblasts, thereby contributing to fibrosis. Blocking of PDGFR-β signaling in kidney fibrosis results in a reduction in pericyte activation and subsequent differentiation into myofibroblasts [37,38].
These data give rise to a compelling hypothesis: that peritubular capillary dysfunction and pericyte–myofibroblast transition induced by activated pericytes may be causal factors underlying renal fibrosis in individuals with diabetes. They provide new insights into the association between activated pericytes and the pathophysiology of DN, which prompted us to examine the relationship further.
The current study has some limitations. First, there was a lack of in vitro experiments to support our hypothesis. However, it should be noted that even primary pericytes derived from intact tissue tend to be activated and have characteristics of myofibroblasts when they are cultured under normal conditions without special stimuli. Second, although db/db mice exhibited more severe albuminuria and mesangial matrix expansion, their renal fibrosis was not as severe as we expected. Other animal models for investigating modulation of the behavior of pericytes in the interstitial renal fibrosis of DN are required. Finally, gaining or attenuation of the function of pericytes should be added in further studies to confirm that activated pericytes contribute to the progression of fibrosis in DN. The findings of such studies could lead to improvement in outcomes of complications of diabetes.
In summary, our findings suggest that activated pericytes are associated with interstitial fibrosis in DN accompanied by peritubular capillary dysfunction, and highlight the role of pericytes as the potential of myofibroblast precursor cells in DN.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
