Radial arteries in the presence or absence of calcification from individuals with kidney failure who experienced arterial venous fistular surgery (n = 3, respectively) were collected from Shuguang Hospital of Shanghai University of Traditional Chinese Medicine from September 2019 to January 2021. All procedures were conducted as per the Declaration of Helsinki and approved by the IRB of Shuguang Hospital affiliated with Shanghai University of TCM (approval number: 2018-575-04-01). Tissues were fixed in 4% paraformaldehyde-fixed and paraffin-embedded tissues (4 µm thick) and use them for further analysis.

Ten-week-old male C57BL/6 mice (SPF grade) were bought from Shanghai SLAC Laboratory Animal Co., Ltd. The Animal Experimentation Ethics Committee of Shanghai University of Traditional Chinese Medicine (PZSHUTCM190920010) gave its approval for the protocol.

The adenine-induced CKD model was established over an eight-week period, following a previously described protocol with minor modifications (Supplementary Fig. 1). After a one-week acclimatization, ten-week-old male mice were fed with either the experimental diet or a standard control diet. The adenine diet was formulated with 10% casein to improve palatability and 1.2% phosphate to exacerbate VC [26, 27]. Seventeen mice were classified into two groups in a random manner: chow diet group (n = 8), and the adenine diet group (n = 9). Some mice were treated with 3-DZNeP (A8182, CAS: 120964-45-6, 98%, Ape Bio, Houston, TX, USA), a carbocyclic analog of adenosine used to block EZH2 activity (1 mg/kg/every 3 days) during the eight-week diet program. Eighteen mice were classified into three groups in a random manner: chow diet + vehicle (DMSO) group (n = 6), adenine diet + vehicle (DMSO) group (n = 6), adenine diet + 3-DZNep (1 mg/kg/per 3 days) group (n = 6).

Mice were euthanized by intraperitoneal injection of an overdose of pentobarbital sodium (100 mg/kg body weight; drug concentration: 8 mg/mL). Ca deposition was measured in the dissected thoracoabdominal aortas of mice. Plasma levels of blood urea nitrogen (BUN), creatinine (Cre), Calcium (Ca), alkaline phosphate (ALP), and Pi were estimated by automatic biochemical analyser (AU680, Beckman Coulter, Brea, CA, USA).

Thoracic aortas (from the arch to the iliac bifurcation) were dissected from male C57 mice weighing between 22–24 g or male Sprague-Dawley rats weighing between 180–200 g. Rat were euthanized by intraperitoneal injection of an overdose of pentobarbital sodium (150 mg/kg body weight; drug concentration: 50 mg/mL). After the removal the adventitia, vessels were sliced into 1–2 mm rings from mice or 3–5 mm rings from rats, which were cultured in HG-DMEM (11965092, Gbico, USA) medium containing 15% foetal bovine serum (FBS, 04-001-1ACS, Biological Industries, Israel) and 0.5% penicillin/streptomycin (P/S, C0222, Beyotime Biotech, Shanghai, China) at 37 °C, 5% CO₂ atmosphere. Arota rings were incubated with a calcifying medium (pH 7.4) containing 2.6 mmol/L inorganic Pi, which is sodium phosphate (NaH₂PO₄/Na₂HPO₄) to induce calcification. Following incubation for indicated time points (2, 3, 5 and 7 days), the samples were harvested and analyzed. Each ex vivo experiment was conducted with at least three biological replicates.

The rat smooth muscle embryonic thoracic aorta cell line A7r5 was bought from the National Collection of Authenticated Cell Cultures, Chinese Academy of Medical Sciences (Cat. GNR 7). Cells were kept at 37 °C, 5% CO₂ atmosphere using high-glucose DMEM (HG-DMEM; 11885084, Gibco, Waltham, MA, USA) enriched with 10% FBS and 0.5% P/S. The cell line was authenticated by the Chinese Academy of Medical Sciences and tested mycoplasma negative. Calcification was induced by 2.6 mmol/L high Pi (Pi, NaH₂PO₄/Na₂HPO₄, PH = 7.4) in DMEM medium or 10 mmol/L $\beta$-glycerophosphate ($\beta$-GP, CAS:154804-51-0, 99%, Sigma, St. Louis, MO, USA) and 1.4 mmol/L CaCl₂ in DMEM medium. Following incubation for indicated time points (1, 2 and 7 days), the samples were harvested for further analysis. Each in vitro experiment was conducted with at least three biological replicates.

The adenovirus for EZH2 (Ad-EZH2) (packed in pADM-FH-GFP vector) were acquired from WZ Biotech company (Shangdong, China). Cells cultured at $\approx$80% confluence were adenovirus-infected (multiplicity of infection was 10) for 48 h.

Rat-Scramble small interfering RNA (siRNA), and Rat-BMP2 siRNA were synthesized by GenePharma (Suzhou, China). Supplementary Table 1 displays the siRNA sequences. Cells were cultured in six-well plates and were transfected with 20 nmol/L siRNA per well via Lipofectamine 2000 Reagent (TL201, Vazyme, Nanjing, China).

VSMCs and aortic rings were treated for seven days in calcified medium with 2.6 mmol/L high Pi (NaH₂PO₄/Na₂HPO₄, PH = 7.4) to induce calcification. Following dissection from CKD animals, aortic tissues were homogenized, and VSMCs were isolated. These samples were then incubated in 0.6 mol/L HCl to facilitate Ca extraction. The Ca content in the HCl supernatant was analyzed via a commercial Kit (o-Cresolphthaleine complexone Assay) (20162400906, BioSino, Beijing, China) and normalized to protein concentration, which was detected via the Bradford assay (PA102, Tiangen, Beijing, China).

Following fixation with 4% paraformaldehyde, kidney samples, thoracic aortas from CKD models, and cultured aortic rings were embedded in paraffin. Each sample was sliced with a thickness of 4 µm. Masson’s trichrome staining was conducted on kidney slices according to the protocol of Trichrome stain kit (D026, Nanjing Jiancheng, Nanjing, China). IHC staining was performed on sections of thoracic aortas from CKD models and cultured aortic rings. The slices’ incubation was conducted with respective primary antibodies against Runx2 (1:200, 8486, CST, Danvers, MA, USA), EZH2 (1:200, 5246, CST, Danvers, MA, USA) and BMP2 (1:200, ER80602, HUABIO, Hangzhou, China). A microscope (80i, Nikon, Tokyo, Japan) was utilized to capture images.

For cell culture, VSMCs were grown in growth or calcifying mediums in six-well plates for 10 days, then the cells were PBS-rinsed three times and followed by fixation in 10% paraformaldehyde for 10 min. After another three times wash, cells were incubated with 1% Alizarin Red S for 30 min and PBS-rinsed three times again. Positively stained cells appeared red.

Von Kossa staining was performed on sections of thoracic aortas from CKD models and cultured aortic rings. The incubation of slices was first conducted in 5% silver nitrate (AgNO₃) solution under ultraviolet light for 30 min and subsequently washed with 5% sodium thiosulfate (Na₂S₂O₃). Hematoxylin was utilized to counterstain nuclei. Upon staining, the calcified nodules changed from brown to black. A microscope (80i, Nikon, Japan) was utilized to obtain images.

Protein lysates from rat primary VSMCs or cultured aortic rings were extracted via RIPA lysis buffer (P0013B, Beyotime Biotech, Shanghai, China). Immunoblotting was conducted as per previously published guidelines [28]. Primary antibodies are Runx2 (1:1000, 8486, CST, USA), OPN (1:1000, BS1264, Bioworld, Dublin, OH, USA), MSX2 (Msh homeobox 2) (1:1000, ab69058, abcam, Cambridge, UK), BMP2 (1:1000, NBP1-19751, Novus, Littleton, CO, USA), a-SMA (1:10,000, ET1607-53, HUABIO, Hangzhou, China), EZH2 (1:1000, 5246, CST, Danvers, MA, USA), H3k27me3 (Histone H3 Lysine 27 Trimethylation) (1:1000, 9733, CST, Danvers, MA, USA), SM22 (1:10,000, A6760, abclonal, Wuhan, China), Calponin (1:10,000, ET1606-17, HUABIO, Hangzhou, China), $\alpha$-tubulin (1:2500, AT819, Beyotime, Shanghai, China). The immunoreactive bands were detected with an ECL substrate (Tanon, Shanghai, China, 180-501) after probing with the HRP-conjugated secondary antibodies (Beyotime Biotech, Shanghai, China, A0208 or A0216). Band intensities were subsequently analyzed through densitometry via Quantity One software 4.6.8 (Basic) (Bio-Rad, Hercules, CA, USA).

TRIzol reagent (R401-1, Vazyme, China) was utilized to extract total RNA. RNA reverse-transcription to cDNA was conducted via HiScript II Q select RT SuperMix (R233-1, Vazyme, China). Hieff qPCR SYBR Green Master Mix (11203ES08, Yeasen, Shanghai, China) was employed per the manufacturer’s instructions. All amplification reactions were conducted over 40 cycles and were performed in triplicate using Real-Time PCR System (Step one plus, ABI, Carlsbad, CA, USA). Supplementary Table 2 displays the primer sequences.

Data are expressed as mean $\pm$ SD. Among-group variations were evaluated via one-way analysis of variance (ANOVA) with Bonferroni posthoc tests, and two-group comparison were conducted via a t-test. Data analysis was conducted via SPSS 26.0 (IBM Corp., Chicago, IL, USA) and GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA). p 0.05 was deemed significant.

An in vitro model of VC was established by using cell line A7r5 (VSMCs). A significant upregulation of Runx2 and Msx2 was observed in 10 mmol/L $\beta$-GP and 1.4 mmol/L CaCl₂- treated VSMCs as compared to control from 48 h to 72 h (Fig. 1A–C), which were associated with the upregulation of the epigenetic markers EZH2 and H3K27me3 in calcified VSMCs by time (Fig. 1D–F).

Fig. 1.

EZH2 is upregulated in calcified VSMC. Western blot analysis and quantification of osteogenic markers Runx2 and Msx2 (A–C), and EZH2 and H3k27me3 levels (D–F). Data are reported as mean $\pm$ SD, n = 2 per group, from three independent biological trials. *p 0.05, **p 0.01, ***p 0.001 vs. control. EZH2, Enhancer of zeste homologue 2; VSMC, rat vascular smooth muscle cell line; Runx2, runt-related transcription factor 2; Msx2, Msh homeobox 2; H3k27me3, Histone H3 Lysine 27 Trimethylation; SD, standard deviation.

An ex vivo model of VC was established by culturing mouse thoracic aortas in high-phosphate (2.6 mmol/L Pi) HG/DMEM medium for 3, 5, and 7 days. A time-dependent increase in aortic Ca content was observed from day 3 to day 7. This model exhibited a time-dependent increase in Ca deposition (Fig. 2A), a result further verified by Von Kossa staining (Fig. 2B). In parallel, the expression of EZH2 was up-regulated in cultured aorta rings after Pi-stimulation as compared with the control group, as shown by immunostaining (Fig. 2C).

Fig. 2.

EZH2 is upregulated in calcified mouse aortas. Mouse aortas were incubated with medium containing 2.6 mmol/L Pi for 3, 5 and 7 days. (A) Quantification of calcium (Ca) deposition. (B) Von Kossa staining. (C) Immunohistochemical staining of EZH2. Scale bar = 100 µm. Data are mean $\pm$ SD, n = 2 per group, biological triplicate experiments. **p 0.01 vs. control. Pi, phosphate.

We further established an ex vivo model of VC using SD rats. Fig. 3A illustrates that the aortic Ca deposition of rat thoracic aortas was significantly increased upon Pi stimulation from day 5 to day 7. The enhanced calcification of elastic lamellae in rat thoracic aortas in calcifying medium for 7 days was further shown by Von Kossa staining (Fig. 3B). Western blotting analysis demonstrated that Pi stimulation induced elevation of osteogenic markers (Fig. 3C–F), while repressing the expression of VSMC differentiation marker Calponin (CNN), $\alpha$-Smooth muscle actin ($\alpha$-SMA), and Smooth muscle protein 22 (SM22) (Fig. 3G–J). However, a significant up-regulation was observed in the expression of both EZH2 and its downstream target, H3k27me3, after Pi stimulation (Fig. 3K–M).

Fig. 3.

EZH2 is upregulated in calcified rat aortas. Rat aortas were cultivated with calcifying medium with 2.6 mmol/L Pi for 3, 5 and 7 days. (A) Calcium deposition was quantified by calcium assay. (B) Calcified area was shown by Von Kossa staining (Magnification $\times$100, Scale bar = 100 µm). Osteogenic marker expression (Runx2, OPN, Msx2) (C–F) and VSMCs trans-differentiation levels ($\alpha$-SMA, CNN and SM22) (G–J) were analyzed by Western blot analysis and densitometry. EZH2 and H3k27me levels were similarly assessed (K–M). Data are reported as mean $\pm$ SD, n = 2 per group, from three independent biological trials. *p 0.05, **p 0.01, ***p 0.001 vs. control.

We next created mouse model of VC associated with CKD by feeding mice with adenine and a high-Pi diet (Supplementary Fig. 1). Blood levels of creatinine (CREA), blood urea nitrogen (BUN), uric acid (UA), phosphate (Pi), and alkaline phosphatase (ALP) were elevated in CKD mice (Fig. 4A–E) (Supplementary Tables 3,4). Elevated calcium deposition (Fig. 4F) and intensified Von Kossa staining (Fig. 4G upper) in the aortic tissue confirmed successful VC induction in the adenine-fed CKD mice. Fig. 4G middle and Fig. 4H shows that EZH2 was upregulated in calcified aortas of CKD mice, as shown by IHC staining and qRT-PCR, which was associated with elevated osteogenic marker Runx2 (Fig. 4G lower and Fig. 4I).

Fig. 4.

Vascular EZH2 is upregulated in mice with CKD-associated VC. In eight-week adenine-triggered renal failure model. Plasma levels of renal function indices were measured (A–E). Aortic calcification was assessed by calcium content quantification (F; chow diet, n = 8 and adenine diet, n = 9) and Von Kossa staining, with EZH2 expression (Magnification $\times$200, scale bar =100 µm) and Runx2 expression (Magnification $\times$100, scale bar =100 µm) evaluated by IHC (G). (H,I) Aortic Runx2 and Ezh2 mRNA levels were analyzed by qRT-PCR. Data are mean $\pm$ SD. *p 0.05, **p 0.01 vs. chow diet. qRT-PCR, Quantitative reverse transcription polymerase chain reaction; VC, Vascular calcification.

We further measured the expression of EZH2 in vascular tissues of CKD patients. VC was revealed by Von Kossa assays in radial arteries from individuals with kidney failure who underwent arterial venous fistular operation. A significantly higher expression EZH2 in tunica media of calcified arteries was revealed by IHC, supporting the positive correlation of EZH2 expression and VC in patients with CKD (Fig. 5).

Fig. 5.

EZH2 upregulation in human Calcifying arteries associated with CKD. Von Kossa staining (upper) in radial arteries in the presence of calcification from CKD individual who experienced an arterial venous fistula operation. Immunohistology analysis of EZH2 expression (Magnification $\times$400, Scale bar = 100 µm) (Lower). CKD, Chronic kidney disease.

We next examined the potential function of EZH2 in VC in vitro. VSMC Ca deposition and calcifying nodules were increased in A7r5 cells after incubation with high Pi calcifying medium (2.6 mmol/L Pi) for 3, 5, 7 or 10 days (Supplementary Fig. 2A,B). CCK-8 assessment was conducted to observe the optimal concentration of 3-DZNeP on A7r5 cells (Fig. 6A). Compared to the control group, two days of treatment with 3-DZNeP did not reduce cell viability below 100% at any concentrations from 1 to 20 µmol/L. Dose-dependent treatment with 3-DZNeP (2, 10 and 20 µM) for seven days reduced Ca deposition in the presence of calcifying culture medium (Fig. 6B). VSMCs were treated with 3-DZNeP (20 µmol/L) under $\beta$-GP and CaCl₂ stimulation for 48 h. WB results indicated that 3-DZNep treatment alleviated the up-regulation of osteogenic marker OPN and Runx2 (Fig. 6C–E). 3-DZNep is a compound with an adenosine structure that competitively inhibits S-adenosylhomocysteine hydrolase (SAH), blocking the breakdown of S-adenosylhomocysteine (SAM) and thereby suppressing SAM-dependent EZH2 methyltransferase activity, which may have potential off-target effects [29]. For this reason, we performed genetic gain-of-function experiments. In contrast to the pharmacological data, adenovirus-mediated EZH2 overexpression markedly exacerbated Ca deposition induced by phosphate (Fig. 6J) and elevated EZH2, Runx2 and OPN expression (Fig. 6F–I), which provides more specific evidence for the function of EZH2 in driving VSMCs osteogenic differentiation.

Fig. 6.

EZH2 suppression reduces VC in vitro and in vivo. (A) Cytotoxicity of the EZH2 inhibitor 3-DZNeP (1–200 µmol/L) was assessed in VSMCs (n = 6 per group). (B) Dose dependent effects of 3-DZNeP (2, 10, 20 µmol/L) on calcium deposition in VSMC calcification model for 7 days were evaluated (n = 2 per group). (C–E) Under $\beta$-GP and CaCl₂ stimulation, VSMCs treated with 3-DZNeP (20 µmol/L, 48 h) were analyzed for Runx2 and OPN expression by Western blotting (n = 2 per group). Conversely, EZH2 overexpression via adenovirus in stimulated VSMCs was used to assess EZH2, Runx2, and OPN protein levels (F–I) and calcium deposition (n = 2 per group), biological three independent experiments (J). In vivo, mice were fed chow, adenine diet, or adenine diet plus 3-DZNeP (1 mg/kg) for eight weeks (n = 6/group). Aortic calcium deposition (K) and plasma phosphorus levels (L) were quantified. Renal injury and fibrosis were estimated by H&E and Masson’s trichrome staining (M; magnification $\times$200, Scale bar = 50 µm), and aortic calcification by Von Kossa staining (magnification $\times$100, Scale bar = 50 µm). Data are represented as mean $\pm$ SD from biological triplicates. Significance is denoted as: **p 0.01, ***p 0.001 vs. control; ^&p 0.05, ^&&p 0.01, ^&⁣&&p 0.001 vs. Pi; ^△p 0.05, ^△△p 0.01, ^△⁣△△p 0.01 vs. Pi+Adv-GFP vs. Pi+Adv-GFP.

Next, we assessed the 3-DZNeP effect on disease progression in the mouse model of VC associated with CKD. CKD mice demonstrated a progressive interstitial fibrosis and adenine crystal deposition in the kidney as shown by Masson and HE staining, which were attenuated by 3-DZNeP (Fig. 6M upper and middle). The level of serum Pi (Fig. 6K) was enhanced in CKD mice and reduced by 3-DZNeP treatment. Notably, Von Kossa staining (Fig. 6M lower) and calcium content assay (Fig. 6L) revealed an increased Ca deposition in aorta tissue in CKD mice, which was significantly reduced by treatment with 3-DZNeP (Supplementary Table 4).

To examine the mechanism of EZH2 underlying its role in VC, the expression of BMP2, a key inducer of osteoblast differentiation was examined. BMP2 protein level was significantly upregulated in high Pi-stimulated VSMCs in a time dependent manner (Fig. 7A,B). Moreover, overexpression of EZH2 by adenovirus increased the expression of BMP2 (Fig. 7C–E). Furthermore, 3-DZNeP treatment diminished the overexpression of BMP2 in high Pi medium stimulated VSMCs, which was associated with down-expression of EZH2 and H3K27me3 (Supplementary Fig. 3A–D). IHC staining result showed that BMP2 was upregulated in calcified cultured aorta rings of mouse (Fig. 7F). The expression of EZH2 and BMP2 was increased, as shown by IHC staining (Fig. 7G), which was in paralleled with the up-regulation of Ezh2 mRNA expression in aorta tissues of CKD mice (Fig. 7H). Mechanistically, BMP2 silencing counteracted the aggravated Ca deposition induced by EZH2-overexpressing adenovirus in calcifying culture medium (Fig. 7I). Moreover, knockdown of BMP2 (Fig. 7J–L) antagonized the upregulation of Runx2 caused by EZH2 overexpression (Fig. 7M,N). Taken together, the data suggest that the progression of VC driven by EZH2 is related to the up-regulation of BMP2.

Fig. 7.

EZH2 may promote VC through up-regulation of BMP2. (A,B) BMP2 expression in VSMCs stimulated with $\beta$-GP and CaCl₂ for 1–3 days was estimated by Western blotting (n = 2 per group). (C–E) Under the same stimulation, EZH2 overexpression via adenovirus was used to evaluate EZH2 and BMP2 protein levels in VSMCs (n = 1 per group). BMP2 expression in calcified mouse aortas was examined by IHC after seven days in calcifying medium (F) and in aortas from CKD-associated VC mice (G; Magnification $\times$200, Scale bar = 100 µm). (H) Aortic of Ezh2 mRNA levels in aortas from adenine-induced renal failure mice were measured by qRT-PCR. Functional interactions were investigated by EZH2 overexpression and BMP2 knockdown in VSMCs, with calcium deposition quantified (I) and EZH2, BMP2 protein levels analyzed by Western blotting (J–L). Runx2 expression under these conditions was also assessed (M,N) (n = 1 per group). Data are mean $\pm$ SD from biological triplicate experiments. *p 0.05, **p 0.01, ***p 0.001 vs. control; ^△p 0.05, ^△△p 0.01, ^△⁣△△p 0.01 vs. Pi+Adv-GFP; ^#⁢#p 0.01, ^#⁢#⁢#p 0.001 vs. Pi+SiRNA-scramble.