Introduction

Pulmonary arterial hypertension (PAH) is a poor prognostic disease characterized by remodeling of the pulmonary arteries and right heart failure. In 2000, heterozygous germline mutations in the bone morphogenic protein receptor type 2 gene (BMPR2) were identified1, and about 30% of patients with idiopathic/heritable PAH were reported to have the BMPR2 pathogenic variants2,3. Although several pathogenic or susceptible genes have been identified, more than half of patients with idiopathic PAH have unknown genetic backgrounds.

The RNF213 p.Arg4810Lys variant was observed in approximately 10% of Japanese patients with idiopathic PAH4, and associated with poor reactivity to a combination of recently available vasodilators and poor clinical outcomes5. The RNF213 p.Arg4810Lys variant was initially identified as the susceptible gene variant of moyamoya disease in Japan6,7, and the RNF213 pathogenic variants has been reported as an independent risk allele in Asian and Caucasian patients with moyamoya disease8,9. Moreover, the RNF213 p.Arg4810Lys variant was recently identified as a risk allele for several vasculopathies, including intracranial artery stenosis10, and peripheral pulmonary stenosis11. Moyamoya diseases sometimes coexist with adults-onset pulmonary artery stenosis, suggesting the association of RNF213 variants12. These rare vasculopathies may be linked with RNF213 variants under the concept of “RNF213-associated vasculopathy”. Despite its clinical importance, the exact pathophysiological role of RNF213 in these vasculopathies remains unclear13.

The molecular mechanisms of RNF213 have been investigated previously. Human endothelial cells overexpressing RNF213 p.Arg4810Lys developed abnormal tubular formation and proliferation in vitro, but cells with RNF213 silencing did not14,15. These studies implied that the RNF213 p.Arg4810Lys variant may have gain-of-function activity. In addition, RNF213 knockout mice and transgenic mice with the endothelial-specific overexpression of RNF213 p.Arg4828Lys, which corresponds to the human p.Arg4810Lys variant, did not develop the phenotypes of moyamoya disease or PAH in a normoxic environment14,15,16. A hypoxic environment induced PAH in mice with the endothelial-specific overexpression of RNF213 p.Arg4828Lys variant16, but whether this transgenic animal model represents the features of patients with PAH is unclear.

The aim of this study was to investigate the underlying mechanisms involved in the development of RNF213-associated vasculopathy using a mouse model that recapitulates the genetic status of patients with RNF213-associated vasculopathy, as a germline mutation, and to identify a novel therapeutic target related to these underlying mechanisms.

Results

Hemodynamics

The sequences of the wild-type allele, mutant allele, and single-strand oligodeoxynucleotides are shown in Fig. 1a. Mice were divided into four groups by RNF213 genotype (RNF213+/+ or RNF213+/p.Arg4828Lys) and housing environment (normoxia or hypoxia) (Fig. 1b). Body weights were significantly lower in the hypoxic environment groups (Fig. 1c). Pulmonary hypertension was not observed in RNF213+/+ and RNF213+/p.Arg4828Lys mice in the normoxic environment. RNF213+/p.Arg4828Lys mice exposed to a hypoxic environment for 12 weeks developed significant pulmonary hypertension compared with RNF213+/+ mice, as shown by the right ventricular systolic pressure (RVSP) [40.4 (interquartile range (IQR), 39.3–41.6) mmHg vs. 35.6 (IQR, 33.0–36.9) mmHg, p = 0.006] (Fig. 1d) and the weight ratio of right ventricle and left ventricle with septum (RV/(LV + S)) [0.38 (IQR, 0.36–0.40) vs. 0.33 (IQR, 0.30–0.35), p = 0.005] (Fig. 1e). Lung specimens from RNF213+/p.Arg4828Lys mice analyzed by Elastica van Gieson staining showed a significantly higher percent medial wall thickness (%MWT) compared with RNF213+/+ mice [37.7 (IQR, 36.3–39.7) % vs. 32.5 (IQR, 29.9–35.3) %, p < 0.001] (Fig. 1f and g).

Figure 1
figure 1

Development of pulmonary arterial hypertension in RNF213+/p.Arg4828Lys mice. (a) A single-strand oligodeoxynucleotide (ssODN) was designed in silico. (b) RNF213+/+ and RNF213+/p.Arg4828Lys mice were housed in normoxic and hypoxic environments. Body weight (c), right ventricular systolic pressure (RVSP) (d), and ratio of right ventricular weight to left ventricle plus septum weight [RV/(LV + S)] (e) were measured in each experimental group. (f) Representative elastica van Gieson staining of lungs from rats in each experimental group. Scale bars: 20 μm. (g) Percent medial wall thickness (%MWT) in small pulmonary arteries (20–60 μm in diameter) was measured (n = 16 arteries from four mice per group). Statistically significant p-values are shown. Hetero, heterozygous; Hx, hypoxia; Mut, mutant; Nx, normoxia; WT, wild-type.

Full size image

Gene expression

Microarray analysis of lungs from RNF213+/p.Arg4828Lys mice showed 127 upregulated genes and 113 downregulated genes (Fig. 2a and b), and the top 15 genes are listed in Fig. 2c. A molecular complex detection algorithm was applied to this network to identify the neighborhoods of proteins. The circadian rhythmic process (red circles) and G alpha i signaling event pathways (blue circles) are highlighted in Fig. 2d. Gene ontology analysis demonstrated a significant upregulation of the rhythmic process, vascular development, including mitogen-activated protein kinase, AMP-activated protein kinase, phosphatidylinositol-3 kinase signaling pathways, and metabolism of lipids (Fig. 2e).

Figure 2
figure 2

Gene expression is regulated by the RNF213+/p.Arg4828Lys variant. The mRNA expressions of RNF213+/p.Arg4828Lys mice fed in hypoxia conditions were compared with those of RNF213+/+ mice fed in hypoxia conditions. (a) Volcano plots. Red dots indicate significantly upregulated genes and green dots indicate significantly downregulated genes in RNF213+/p.Arg4828Lys mice. (b) Heat-map analysis. (c) Top 15 genes that were up or downregulated in RNF213+/p.Arg4828Lys mice. (d) Protein-protein network analysis. (e) Gene ontology analysis.

Full size image

CXCL12 chemokine signaling

We focused on the CXCL12 signaling pathway because it is associated with the circadian rhythmic and G alpha i signaling event pathways17,18. The protein level of CXCL12 was significantly elevated in lungs of RNF213+/p.Arg4828Lys mice fed in hypoxic environment (supplemental fig. s1). CXCL12 is a specific ligand of the CXCR4; thus, AMD3100, a CXCR4 antagonist, was administered intraperitoneally to clarify the association between the CXCL12/CXCR4 axis pathway and the development of PAH (Fig. 3a). Real-time quantitative polymerase chain reaction demonstrated that the mRNA expression of CXCL12 was significantly elevated in the lungs of RNF213+/p.Arg4828Lys mice compared with RNF213+/+ mice. The expression of CXCL12 was significantly increased in RNF213+/p.Arg4828Lys mice injected with AMD3100, probably because of the induction of a positive feedback loop related to the blockade of CXCR4 signaling (Fig. 3b). The mRNA expression of forkhead box M1 (FoxM1), a factor downstream of CXCR419, was significantly elevated in RNF213+/p.Arg4828Lys mice compared with RNF213+/+ mice, and this significance was diminished by the administration of AMD3100, suggesting that AMD3100 inhibited the signal transduction of CXCR4 (Fig. 3c). Significant remodeling of the right ventricular and pulmonary arteries was observed in RNF213+/p.Arg4828Lys mice compared with RNF213+/+ mice; however, when comparing RNF213+/p.Arg4828Lys mice administered AMD3100 and RNF213+/+ mice, there was no significant difference in body weight [22.9 (IQR, 19.3–26.3) vs. 25.2 (IQR, 24.8–27.9) g, p = 0.76] (Fig. 3d), RVSP [36.5 (IQR, 35.6–39.5) vs. 35.6 (IQR, 33.0–36.9) mmHg, p = 0.48] (Fig. 3e), RV/(LV + S) [0.36 (IQR, 0.34–0.38) vs. 0.33 (IQR, 0.30–0.35), p = 0.14] (Fig. 3f), and %MWT [10.2 (IQR, 9.5–11.0) vs. 11.0 (IQR, 9.6–11.8)%, p = 0.36] (Fig. 3g and h) in a hypoxic environment.

Figure 3
figure 3

Therapeutic intervention of the CXCL12/CXCR4 signaling pathway in RNF213+/p.Arg4828Lys mice. (a) A CXC chemokine receptor type 4 (CXCR4) antagonist, AMD3100, was injected into RNF213+/p.Arg4828Lys mice fed in a hypoxic environment. The mRNA expressions of CXC motif chemokine ligand 12 (Cxcl12) (b) and forkhead box M1 (FoxM1) (c) were measured by real-time quantitative polymerase chain reaction. Body weight (d), right ventricular systolic pressure (RVSP) (e), and ratio of right ventricular weight to left ventricle plus septum weight [RV/(LV + S)] (f) were measured in each group. (g) Representative elastica van Gieson staining of lungs from rats in each experimental group. Scale bars: 20 μm. (h) Percent medial wall thickness (%MWT) in small pulmonary arteries (20–60 μm in diameter) was measured (n = 16 arteries from four mice per group). Statistically significant p-values are shown. Hx, hypoxia; NS, not significant.

Full size image

Expression of CXCR4 in the lungs of mice and humans

The immunostainings of murine lungs were demonstrated in Fig. 4. Lung specimens of RNF213+/+ mice fed in a hypoxic environment (Fig. 4a and c) had lower CXCR4 expression, whereas lungs from RNF213+/p.Arg4828Lys mice fed in a hypoxic environment had significant CXCR4 expression in the vimentin-positive spindle-shaped cells in the adventitia and interstitial lesions (Fig. 4d and f). Human lung specimens obtained by autopsy from two patients with severe PAH who had the heterozygous RNF213 p.Arg4810Lys variant were demonstrated in Fig. 5. Both lung specimens had high CXCR4 expression in the vimentin-positive spindle-shaped cells in the adventitia and interstitial lesions (Fig. 5a and p). Fluorescent immunohistochemistry images demonstrated that CXCR4 expression was overlapped with vimentin-positive cells (Fig. 5q and t).

Figure 4
figure 4

Histopathological images of murine lung specimens. Lung immunostaining of alpha-smooth muscle actin (αSMA), vimentin, and CXC chemokine receptor type 4 (CXCR4) antibodies in RNF213+/+ mice (panels a, b, and c) and RNF213+/p.Arg4828Lys mice (panels d, e, and f). Yellow triangles indicate vimentin-positive cells, and red triangles indicate CXCR4 positive cells. Scale bars: 20 μm in all panels.

Full size image
Figure 5
figure 5

Histopathological images of lung specimens from patients. Representative images of lung specimens from two patients with the heterozygous RNF213 p.Arg4810Lys variant (panels a-h for patient #1, and panels i-t for patient #2) are demonstrated with the staining of alpha-smooth muscle actin (αSMA), vimentin, and CXC chemokine receptor type 4 (CXCR4) antibodies. Yellow triangles indicate vimentin-positive cells, red triangles indicate CXCR4 positive cells, and white triangles were cells in which vimentin and CXCR4 co-exist. Scale bars: 100 μm in panels a, b, c, d, i, j, k, and l; and 20 μm in panels e, f, g, h, m, n, o, p, q, r, s, and t. HE, hematoxylin and eosin stain.

Full size image

Discussion

Our data demonstrated that carrying the RNF213 variant worsened PAH, although mice are vulnerable to PAH development after hypoxia, and this worsened PAH was ameliorated by the inhibition of the CXCL12/CXCR4 signaling pathway. The expression of CXCR4 was elevated in the adventitia and interstitial lesions among pulmonary vessels in lung specimens of mice and patients harboring the RNF213 p.Arg4810Lys variant. These findings suggest that the CXCL12/CXCR4 signaling pathway might be a novel therapeutic target for patients with RNF213-associated PAH.

Previous reports demonstrated that the CXCL12/CXCR4 pathway is a key modulator of immune system regulation and is involved in leukocyte migration19, cerebellar development20, tumor progression21,22,23 and angiogenesis24,25. CXCL12, also known as stromal cell-derived factor-1, is a homeostatic chemokine mainly expressed by vascular endothelial cells, stromal fibroblasts, and osteoblasts in various organs23. CXCL12 upregulated by cell damage with inflammation or a hypoxic environment forms a heterocomplex with high morbidity group box-1, and binds to CXCR4, a seven-span transmembrane G-protein-coupled receptor26. Activated CXCR4 promotes a conformational change in the Gα subunit, followed by the activation of several intercellular pathways, including Janus kinase-signal transducer and activator of transcription, phospholipase C, mitogen-activated protein kinase, c-Jun N-terminal kinase, p38, and phosphoinositide 3-kinase/Akt/ mammalian target of rapamycin (PI3K-AKT-mTOR) cascades27,28. These signaling pathways are essential for cell proliferation, migration, and the maintenance of immune homeostasis28.

Recent experimental studies highlighted the importance of the CXCL12/CXCR4 signaling pathway in cardiovascular diseases29,30. Activation of the CXCL12/CXCR4 pathway was also involved in the development of chronic pulmonary hypertension and vascular remodeling in monocrotaline-induced rats and SU5416 plus chronic hypoxia rats30,31,32,33. An elevated serum level of CXCL12 in patients with idiopathic PAH was associated with their poor right ventricular function and prognosis34,35,36,37. Our study demonstrated that CXCR4 blockade by AMD3100 injection ameliorated the severity of PAH in a murine model. Furthermore, a previous study demonstrated that numbers of circulating CXCL12 and CXCR4 positive peripheral cells were increased in patients with moyamoya disease37. These studies suggest that the CXCL12/CXCR4 pathway might be associated with the development of RNF213 vasculopathy.

RNF213 encodes a 590 kDa protein that consists of 5207 amino acids, and it contains a ring finger ___domain with E3 ubiquitin-protein ligase activity and two regions of ATPase-associated domains6,7. Previous reports demonstrated that RNF213 regulated the proliferation and angiogenesis of endothelial cells via activation of the PI3K-AKT pathway38, non-canonical WNT signaling39, and proinflammatory cytokines40. The co-culture of RNF213-knockdown endothelial cells (EC) and intact vascular smooth muscle cells (vSMC) led to significant changes in the proliferation and migration of vSMC, suggesting the RNF213 protein in EC regulated the EC-to-vSMC interactions41. Our study demonstrated that CXCL12 mRNA expression was elevated in lung specimens of RNF213+/p.Arg4828Lys mice fed in a hypoxic environment. Previous studies have demonstrated that elevation of FoxM1 expression, the downstream of CXCL12/CXCR4 signal pathway, promoted the proliferation of smooth muscle cells, resulting in the pulmonary vascular remodeling and pulmonary hypertension42,43,44. In our study, we demonstrated that CXCR4 expression was elevated in vimentin-positive spindle-shaped cells in the interstitial lesions, considered as active fibroblasts. Further study is warranted to elucidate the molecular mechanisms and cell interactions associated with the CXCL12/CXCR4 pathway and development of PAH.

This study has limitations. First, we have not evaluated other molecular pathways which were significant in the microarray analysis in lungs of RNF213+/p.Arg4828Lys mice. Although PAH is a multifactorial disease, our experimental model has demonstrated that interfering the CXCL12/CXCR4 signal pathway might improve the development of PAH in part. Second, we did not perform RNA-sequencing to elucidate the molecular mechanism regarding different cell types including endothelial cells and pulmonary smooth muscle cells. In this study, we identified the CXCL12/CXCR4 signal pathway by microarray analysis, and demonstrated that this pathway was highly expressed in both mice and human lung specimens who had RNF213 p.Arg4810Lys variant in the adventitial active fibroblasts. Since recent study has demonstrated the importance of active fibroblasts in development of PAH45,46, targeting active fibroblasts may have potential benefit on treatment for PAH. Third, we could not evaluate the protein-interactions between RNF213 and CXCL12. Since the function of RNF213 has not been fully understood, further studies including immunoprecipitation-mass spectrometry are needed.

In conclusion, a cause-and-effect relationship was identified between the RNF213 p.Arg4810Lys variant and PAH via the CXCL12/CXCR4 pathway, suggesting that this pathway is a novel therapeutic target for RNF213-associated vasculopathy.

Methods

Animal experiments

All animal studies were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health Guidelines. All experiments were performed in accordance with relevant guidelines and regulations. This study was designed in accordance with the ARRIVE guidelines. The experimental procedures were approved by the Boards for Studies in Experimental Animals of Keio University and Nippon Veterinary and Life Science University. Male C57BL/6 mice (20–30 g) were purchased from Charles River Laboratories (Wilmington, MA, USA). Mice with the heterozygous RNF213 p.Arg4828Lys variant, corresponding to the human p.Arg4810Lys variant, were created using CRISPR-Cas9 technology. Mice were housed in a hypoxic (10% oxygen) or normoxic (21% oxygen) for 3 months in the same room. The CXCR4 antagonist, AMD3100, was purchased from Abcam (Cambridge, UK), dissolved with phosphate-buffered saline, and then 5 mg/kg body weight was administrated via intraperitoneal injection every other day28.

Hemodynamic assessment

Each group contained 6–13 mice. Each mouse was anesthetized with 2–3% (v/v) isoflurane on a heat board with heart rate monitoring by electrocardiography. A micro-tip catheter (Millar Instruments, Inc., Houston, TX, USA) was inserted into the right ventricle via the right jugular vein to measure RVSP. Hemodynamic measurements were analyzed using Lab Chart 8 (ADInstruments Ltd., Oxford, UK). After sacrifice by cervical dislocation under isoflurane anesthesia, 10 ml of phosphate-buffered saline (PBS) was slowly perfused through the right ventricle of the mice to remove blood cells from the heart and lungs. The lungs were then excised, and PBS was administered through the bronchia at a pressure of 10 cmH2O to inflate the lungs. The bronchus was ligated with 4 − 0 nylon and the lungs were fixed with 10% formalin solution for one day. On the following day, the lungs were placed in a PBS solution, and embedded in paraffin at 4-µm thickness. The right ventricle was separated from the left ventricle plus septum for the assessment of right ventricular remodeling by the weight ratio of right ventricle and left ventricle with septum (RV/LV + S), and the lung was separated and stored for further analysis.

Assessment of pulmonary vascular remodeling

Lung specimens were stained with hematoxylin and eosin, and elastica van Gieson. To assess pulmonary vessel remodeling, the percent medial wall thickness was analyzed in four small pulmonary vessels of 20–60-µm diameter from an individual lung. Diagonally cut vessels were measured by the mean value of the maximum short axis and the axis perpendicular to the maximum short axis. Images were captured by light microscopy (BZ-9000; Keyence, Osaka, Japan).

RNA isolation and RT-qPCR

Total RNA was isolated from tissues using an Rneasy Mini Kit (Qiagen, Hilden, Germany), and cDNA was synthesized from 200 ng of total RNA using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA, USA), according to the manufacturer’s protocol.

Quantitative mRNA expression was assessed by real-time quantitative polymerase chain reaction (PCR) using the KAPA SYBR Fast qPCR Kit (KAPA BIOSYSTEMS, Wilmington, MA, USA). Samples were quantified using a hot start at 95°C for 2 minutes, followed by 40 cycles at 94°C for 3 seconds, and 60°C for 20 seconds using the ViiA7 system (Applied Biosystems, Waltham, MA, USA). The data were analyzed by the ΔΔCT method. The following primers were used: mouse glyceraldehyde 3-phosphate dehydrogenase (Gapdh): forward 5′-TGCACCACCAACTGCTTAG-3′, reverse 5′-GGATGCAGGGATGATGTTC-3′; mouse Cxcl12: forward 5′-GTTCCCAAAGGTCTGAAGAG-3′, reverse 5′-CATATGGCACAGATGACATTGG-3′; and mouse FoxM1: forward 5’- CACTTGGATTGAGGACCACTT-3’; reverse 5’- GTCGTTTCTGCTGTGATTCC-3’, according to previous reports19,40,44. All quantitative PCR data were normalized to the housekeeping gene Gapdh.

Gene expression analysis

Total RNA from lung specimens was subjected to Affymetrix GeneChip Clariom D microarray (Applied Biosystems, Waltham, MA, USA). The gene expression data were analyzed with GeneSpring™ software (Agilent Technologies, Saint Clara, CA, USA). The 1.5-fold threshold and p < 0.05 threshold were defined to classify upregulated or downregulated genes. We used Metascape software for gene set enrichment analysis47,48.

Enzyme-linked immunosorbent assay

Lung specimens were frozen with liquid nitrogen, and stored at -80℃. Protein was extracted by adding the mixed 1xRIPA buffer (10xRIPA buffer 50 µl, SDS solution 50 µl, double distilled water 400 µl, protease inhibitor 5 µl, phosphatase inhibitor 5 µl, 100µM MG132 0.5 µl, and 0.1 M DTT 5 µl (Nacalai Tesque, Inc., Kyoto, Japan)) to each lung on ice. Samples were homogenized uniformly, centrifuge 15,000 rpm for 20 min on 4℃. Supernatant was collected, and the expression of CXCL12 protein was analyzed by Mouse SDF-1 alpha/CXCL12 Quantikine ELISA Kit (R&D Systems, Inc. Minneapolis, MN) according to the manufactural protocol. Absorbance was measured with 450 nm wave.

Human samples from patients with PAH

Genetic tests were approved by the Ethics Committee of Keio University Hospital (approval number: 20140203) and Kyorin University Hospital (approval number: H26-125), and performed with informed consent from the patients after genetic counseling. All research was performed in accordance with relevant guidelines/regulations. Research involving human research participants must have been performed in accordance with the Declaration of Helsinki. DNA samples were collected from the peripheral blood of patients with PAH. Whole-exome sequencing was performed using a HiSeq 2500 platform (Illumina, San Diego, CA) and SureSelectXT Human All Exon Kit (Agilent Technologies) for hybridization capture. Lung specimens from two patients with the heterozygous RNF213 p.Arg4810Lys variant were obtained by autopsy. No organs/tissues were procured from prisoners. Specimens were fixed in 10% formalin immediately after resection, embedded in paraffin, and cut into thin slices.

Pathological analysis

Hematoxylin and eosin staining (Carrazzi’s hematoxylin solution and 1% eosin Y) were performed in mice and human lung specimens. Images were captured using light microscopy (BZ-9000; Keyence, Osaka, Japan). Immunohistochemistry was performed using a Bond-Max automated immunohistochemical staining machine (Leica Microsystems, Milton Keynes, UK), according to the manufacturer’s instructions. The primary antibodies were rabbit anti-CXCR4 monoclonal antibody (#NB100-56437, 1:200; Novus Biologicals, Centennial, CO, US) and rabbit anti-alpha smooth muscle actin polyclonal antibody (1:200; Novus Biologicals, Littleton, Colorado, US), mouse anti-vimentin antibody (#ab8978, 1:200, Abcam, Cambridge, UK), diluted with BOND primary antibody diluent (Leica Microsystems, Milton Keynes, UK). Fluorescent immunohistochemistry images were observed under an LSM 710 confocal laser scanning microscope (Carl Zeiss Meditec AG, Jena, Germany). All representative slides were scanned using a digital slide scanner (NanoZoomer S210; Hamamatsu Photonics, Hamamatsu, Shizuoka, Japan), and the measurements were evaluated using viewing software (NDP.view2; Hamamatsu Photonics).

Statistical analysis

All data were expressed as the median with interquartile range (IQR). Comparisons between two groups were performed by Mann–Whitney U-tests and comparisons among multiple groups by Kruskal–Wallis tests followed by Holm’s method (post-hoc analysis). A p-value < 0.05 was considered statistically significant. Analyses were performed using R software.