Growth Factor Midkine Aggravates Pulmonary Arterial Hypertension via Surface Nucleolin

Pulmonary arterial hypertension (PAH) is a progressive fatal disease caused by pulmonary arterial remodeling. Midkine regulates cell proliferation and migration, and it is induced by hypoxia, but its roles in pulmonary arterial remodeling remain unclear. Serum midkine levels were significantly increased in PAH patients compared with control patients. Midkine expression was increased in lungs and sera of hypoxia-induced PAH mice. Hypoxia-induced pulmonary arterial remodeling and right ventricular hypertrophy were attenuated in midkine-knockout mice. Midkine-induced proliferation and migration of pulmonary arterial smooth muscle cells (PASMC) and epidermal growth factor receptor (EGFR) signaling were significantly increased under hypoxia, which also induced cell-surface translocation of nucleolin. Nucleolin siRNA treatment suppressed midkine-induced EGFR activation in vitro, and nucleolin inhibitor AS1411 suppressed proliferation and migration of PASMC induced by midkine. Furthermore, AS1411 significantly prevented the development of PAH in Sugen hypoxia rat model. Midkine plays a crucial role in PAH development through interaction with surface nucleolin. These data define a role for midkine in PAH development and suggest midkine-nucleolin-EGFR axis as a novel therapeutic target for PAH.

www.nature.com/scientificreports www.nature.com/scientificreports/ Increased serum midkine levels were shown to be independently associated with unfavorable outcomes in patients with heart failure 16 . Moreover, cardiac dysfunction in pressure overload mouse models was reported to induce midkine expression in the lung 17 , and the expression of this gene in respiratory epithelium was demonstrated to be involved in the pathogenesis of pulmonary arterial remodeling 13 . We previously reported that midkine promotes EGFR activation 15 . However, a precise mechanism underlying this process remains unclear.
Nucleolin, reported to be a candidate receptor of midkine 18 , is abundantly expressed in nucleus, mediating ribosome biogenesis and RNA metabolism 19 . Recent studies have demonstrated that it is expressed as well at cell surface and mediates tumorigenesis and cancer-related neoangiogenesis 20,21 . Furthermore, nucleolin was recognized as a EGFR binding partner 22 , stabilizing it and preventing degradation 23,24 . Although the level of surface nucleolin expression in PASMCs remains to be determined, hypoxic exposure induces PASMC phenotypic switch, inducing the proliferation of these cells 25 . Therefore, we hypothesized that hypoxia induces the upregulation of surface nucleolin, which facilitates midkine binding to surface nucleolin and EGFR activation, leading to the development of PAH.

Results
Serum midkine levels are increased in pAH patients and mice with hypoxia-induced pAH. We measured serum midkine levels in patients with PAH or those with chronic heart failure. Baseline patient characteristics are presented in Table 1. Serum midkine levels were significantly increased in patients with PAH in comparison with those with heart failure without increased PVR (Fig. 1a). Next, we investigated midkine expression levels in hypoxia-induced PAH mice model. Midkine mRNA levels were shown to be significantly upregulated in the lungs of these mice (Fig. 1b). Midkine expression levels in the lung during hypoxia were shown to increase in a time-dependent manner (Fig. 1c). Midkine expression was observed in respiratory epithelium, arteriole, and vascular adventitia in hypoxia-induced PAH mice lung (Fig. 1d,e). Serum midkine levels were increased in PAH mice compared with those in the sham control as well (Fig. 1f).
Midkine deficiency ameliorates rate pulmonary arterial remodeling in PAH mice. We exposed WT and MK-KO mice to chronic hypoxia for 4 weeks to investigate the impact of midkine on the pulmonary arterial remodeling. There were no significant differences in the pulmonary arterial findings between WT and MK-KO mice under normoxic conditions. The ratio of fully-muscularized vessels and medial wall thickness were shown to increase following the exposure to hypoxia in the WT mice. Midkine deficiency was shown to significantly suppress pulmonary arterial remodeling rate (Fig. 2a-c). The increased ratio of RV/LV + S and the decreased ratio of pulmonary artery acceleration time to pulmonary artery ejection time, determined by echocardiography, were also decreased in MK-KO mice, in comparison with those in the WT mice (Figs. 2d,e). Hypoxia-induced phosphorylation of EGFR and extracellular signal-regulated kinase (ERK) 1/2 was shown to be suppressed in MK-KO mice, compared with that in the WT mice (Fig. 2f). The expression of cell proliferation markers, Cyclin B1 and proliferation cell nuclear antigen (PCNA), was significantly increased in the hypoxic WT mice, but considerably attenuated in the MK-KO mice (Fig. 2g).
Hypoxic conditions enhance midkine-induced pASMcs migration and proliferation. Next, we performed scratch assay to evaluate the effects of midkine on the migration of PASMCs, and showed that the www.nature.com/scientificreports www.nature.com/scientificreports/ treatment with recombinant midkine significantly induces the migration of these cells, compared with that of the vehicle control cells. Interestingly, midkine-induced PASMC migration was significantly enhanced by hypoxic conditions, compared with the effects of normoxia (Fig. 3a,b). The incorporation of 5-bromo-2-deoxyuridine (BrdU) revealed that midkine significantly induces PASMC proliferation, compared with that of the control cells, which was further enhanced by hypoxic conditions (Fig. 3c). hypoxia-induced PAH mice. (a) Serum MK levels in patients who underwent right-heart catheterization to determine pulmonary vascular resistance (PVR). (b) Relative MK mRNA expression in mouse lungs 4 weeks after the exposure to normoxia or hypoxia. (c) MK protein expression in mouse lungs during the exposure to hypoxic conditions. Expression levels were normalized to those of β-tubulin. Full-length blots are presented in Supplementary Figure 1 (d) Mouse lungs 4 weeks after the exposure to normoxia or hypoxia were stained with antibodies against MK (red), nuclear marker (DAPI, blue), and cell type markers (green) for vascular smooth muscle cell (α-smooth muscle actin (SMA)), stromal cell (vimentin), and endothelial cell (platelet endothelial cell adhesion molecule (PECAM)). Arrowheads in each image show both MK and the cell type marker positive cells. (e) Fluorescent intensity was quantified and normalized to the background intensity by using ImageJ software (version 1.42; https://imagej.nih.gov/ij/). (f) Serum MK concentrations in mice exposed to normoxia or hypoxia were evaluated. Data are expressed as mean ± SE (n = 6-10 mice per group). *ANOVA post-hoc Tukey's honest significant difference, P < 0.05 vs. the indicated control.
www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ nucleolin and eGfR mediate midkine-induced alterations. EGFR and ERK1/2 were shown to be phosphorylated at 3 min and 10 min, respectively, following the treatment of PASMCs with midkine. This EGFR signaling activation was significantly promoted by the exposure to hypoxia (Fig. 3d). We used nucleolin-specific small interfering RNA (siRNA) to silence nucleolin expression, and stimulated PASMCs using the recombinant midkine, which demonstrated that the silencing of this gene leads to a significant suppression of midkine-induced EGFR and ERK1/2 phosphorylation (Fig. 3e). Midkine induced-PCNA expressions were similarly increased by the exposure to hypoxia, and suppressed by nucleolin depletion (Fig. 3d,e). Additionally, we showed that EGFR-specific siRNA treatment significantly suppressed midkine-induced ERK1/2 phosphorylation and PCNA expression (Fig. 3f).
Hypoxic conditions induce nucleolin translocation to the cell surface. We observed that the exposure to hypoxic conditions did not alter the expression of nucleolin, EGFR, and other midkine receptor candidates, including low density lipoprotein related protein 1 (LRP1) and integrin β1 in PASMCs (Fig. 4a). However, the expression of nucleolin in the membrane fraction of these cells was significantly increased after hypoxia exposure, while nuclear nucleolin levels were significantly decreased. The expression of LRP1 and integrin β1 in both membrane and nuclear fractions did not change (Fig. 4b). Next, we observed the localization of green fluorescent protein (GFP)-tagged nucleolin during cell exposure to hypoxia using time-lapse imaging. GFP-tagged nucleolin was observed to be predominantly expressed in the cell nuclei under the normoxic conditions. However, following the exposure of cells to hypoxic conditions, GFP-nucleolin expression decreased in nuclei and increased in the extranuclear lesions, in a time-dependent manner (Fig. 4c). Confocal laser microscopy revealed that GFP-tagged nucleolin co-localized with cell surface protein Na/K-ATPase during hypoxia (Fig. 4d). Next, we examined the impact of hypoxic exposure on midkine-nucleolin interactions. Immunoprecipitation assay revealed that midkine interacts with nucleolin, and this interaction was enhanced by hypoxic exposure (Fig. 4e). Further, we observed that nucleolin interacts with EGFR as well, and although this interaction was shown to be weak even under hypoxic conditions, midkine treatment led to its increase (Fig. 4f).

AS1411 inhibits midkine-induced cellular processes.
Immunoprecipitation assay results demonstrated that the nucleolin inhibitor AS1411 inhibits the interactions between nucleolin and biotinylated recombinant midkine (Fig. 5a). AS1411 significantly inhibited midkine-induced EGFR and ERK1/2 phosphorylation, compared with that in the control cells (Fig. 5b). Midkine-induced migration and proliferation of PAMSCs under the hypoxic conditions were significantly suppressed following the pretreatment with AS1411 (Fig. 5c,d).

AS1411 ameliorates the development of Sugen/hypoxia (SuHx)-induced PAH. We investigated
whether AS1411 can suppress the pulmonary arterial remodeling and development of PAH in SuHx-triggered PAH rat model, because this model is characterized by severe PAH, and has properties similar to the pathological features of human PAH, which is irreversible and progressive under normoxic conditions 26 . Rats were divided into three groups, as shown in Fig. 5e. SuHx rats were treated with AS1411 or vehicle after the establishment of PAH as previously described 27 . Western blot analysis demonstrated that midkine expression increases in the lungs of SuHx rats, compared with that in the normoxic control (Fig. 5f). AS1411 treatment led to a significant decrease in the RV systolic pressure in SuHx rats, compared with that in the controls (Fig. 5g). The decreased ratio of pulmonary artery acceleration time/pulmonary artery ejection time determined was shown to be attenuated as well after the treatment with AS1411, compared with that of the placebo control (Fig. 5h). Enlarged RV chamber area and RV hypertrophy were significantly suppressed in AS1411-treated rats (Fig. 5i,j). Histological analysis revealed that AS1411 led to a significant decrease in the occlusive vessel ratio and PCNA expression (Fig. 5k,l).

Discussion
Here, we demonstrated that midkine was shown to induce the proliferation and migration of PASMCs and midkine deficiency attenuated pulmonary arterial remodeling in hypoxia-induced PAH mice. Hypoxia induced the translocation of nucleolin to the cell surface, and enhanced midkine-induced activation of EGFR signaling. Silencing of nucleolin expression by siRNAs and AS1411 suppressed midkine-induced EGFR signaling and PASMC proliferation, and this inhibitor was shown to attenuate the development of PAH by suppressing pulmonary arterial remodeling.
The pathogenesis of PAH involves a combination of factors, including gene mutation, inflammation, pulmonary endothelial dysfunction, and pulmonary vasculature cell proliferation. PASMCs obtained from patients with PAH are more sensitive to growth factors than those from the control subjects, and sustained growth signaling causes the excessive proliferation of PASMCs 7 . EGFR signaling is a major contributor of PASMC proliferation, migration, and survival, and an EGFR inhibitor was shown to ameliorate pulmonary arterial remodeling in monocrotaline-induced PAH rat 28 . Although circulating midkine was reported to exacerbate cardiac remodeling via EGFR signaling 15 , the relationship between serum midkine levels and pulmonary arterial remodeling has not been elucidated. Here, we demonstrated that serum midkine levels in patients with PAH are considerably higher than in the patients with heart failure without pulmonary arterial remodeling. The upregulation of serum midkine levels and EGFR signaling activation were observed in hypoxia-induced PAH mice, and midkine deficiency attenuated EGFR activation and pulmonary arterial remodeling. These results suggest a crucial role of midkine in the development of PAH.
Many studies demonstrated that midkine plays important roles in cell migration and proliferation [29][30][31] . However, the mechanism underlying these processes has not been clarified. Although midkine upregulation has been demonstrated as a predictive marker for poor outcomes in patients with cancer and chronic heart failure, the cell proliferative effects of midkine does not always promote pulmonary arterial remodeling in these patients 16,32,33 . Therefore, we hypothesized that the increased expression or distribution change of midkine www.nature.com/scientificreports www.nature.com/scientificreports/ (e) PASMCs were treated with biotinylated midkine (MK) or vehicle, and exposed to normoxia or hypoxia, and midkine-nucleolin binding was analyzed. Full-length blots are presented in Supplementary  Figure 9. (f) NCL-EGFR interactions in PASMCs treated with MK or vehicle, and exposed to hypoxia. Strep-HRP, horseradish peroxidase conjugate streptavidin. Full-length blots are presented in Supplementary Figure 10. www.nature.com/scientificreports www.nature.com/scientificreports/ receptors was involved in the development of PAH. We examined the expression levels and distribution of candidates receptors for midkine, such as LRP1, integrin β1, and surface nucleolin 15 . Casttelano et al. reported that exposure to hypoxia induces the expression of LRP1 in human coronary arterial smooth muscle cells (hCASMC) 34 . However, we did not observe similar changes in this study, which may be explained by the different cell types and hypoxic conditions we used, since the effects of hypoxia on PASMCs were reported to differ from those observed using hCASMC 35 . However, we observed that hypoxia leads to the alterations in the subcellular distribution of nucleolin, which facilitates midkine-induced activation of EGFR signaling. To the best of our knowledge, this study is a first report demonstrating the role of surface nucleolin in midkine-induced EGFR activation and subsequent PASMC proliferation and PAH development. Several mechanisms of abnormal EGFR activation in different cells have been observed, such as due to autocrine/paracrine ligand loops, gene mutation, overexpression and impaired degradation of EGFR, and the heterodimerization of EGFR with ErbB2 36 . Similar to that observed in a previous report 37 , there was no change in EGFR expression in PASMCs grown under hypoxic www.nature.com/scientificreports www.nature.com/scientificreports/ conditions. Moreover, as observed in the previous studies, using cancer cell lines, we observed a weak interaction between nucleolin and EGFR 23 , although nucleolin was expressed on the surface of PASMCs under hypoxic conditions. This aberrant interaction required midkine stimulation. Furthermore, the silencing of nucleolin and the inhibition of midkine binding to nucleolin attenuated midkine-induced EGFR activation, indicating that midkine may facilitate EGFR activation via nucleolin. Guanine adenine-rich domain at C terminal portion of nucleolin was demonstrated to be a lesion responsible for the interaction with both midkine and EGFR 18,22 . We observed that EGFR activation had a trend toward an increase by silencing nucleolin expression, although the role of nucleolin in EGFR activation remains to be elucidated. Wolfson et al. demonstrates that overexpressed nucleolin promotes ligand-independent EGFR activation 38 . In contrast, Reyes-Reyes el al reported that nucleolin depletion also activates EGFR via Rac1 activation 39 . It is possible that the difference in expression levels of surface nucleolin cause this discrepancy. Although our data revealed the involvement of nucleolin in midkine-induced EGFR activation, further study is needed to investigate the role of midkine in the interaction between nucleolin and EGFR activation.
The schematic illustration of the pathway suggested by the results of this study is presented in Fig. 6. Secreted midkine binds to nucleolin on the cell surface, which results in the activation of EGFR and the downstream ERK1/2. Hypoxia promotes cell surface translocation of nucleolin, facilitating midkine binding to the surface nucleolin, inducing EGFR signaling activation. AS1411 is an aptamer that specifically binds to the surface and cytoplasmic nucleolin molecules, and a promising agent with demonstrated clinical efficacy and low toxicity, and Figure 5. AS1411 suppresses the development of pulmonary arterial hypertension (PAH) by inhibiting midkine (MK)-nucleolin (NCL)-EGFR axis. (a) Pulmonary arterial smooth muscle cells (PASMCs) were treated with vehicle or AS1411, which was followed with the treatment with biotinylated MK. Full-length blots are presented in Supplementary Figure 11. (b) Phosphorylated ERK1/2 and EGFR (p-ERK1/2 and p-EGFR) and total ERK1/2 and EGFR (t-ERK1/2 and t-EGFR) expression levels in PASMCs treated with MK after AS1411 pretreatment. Full-length blots are presented in Supplementary Figure 12. (c) Representative images of MKinduced migration of PASMCs after the pretreatment of cells with placebo and AS1411, and its quantification using ImageJ software (version 1.42; https://imagej.nih.gov/ij/). (d) MK-induced proliferation of PASMCs under hypoxic conditions. (e) Study design. Rats were divided into three groups, and the rats were treated with AS1411 (10 mg/kg/day) or the vehicle for consecutive 5 days after the establishment of PAH. Normoxic control was treated with vehicle and exposed to hypoxic conditions for 3 weeks, which was followed by the reexposure to normoxia.  www.nature.com/scientificreports www.nature.com/scientificreports/ due to these facts, AS1411 was studied in several human clinical anti-cancer trials 40,41 . Since this inhibitor does not affect nucleolin expression and stability 40 , the precise mechanism underlying the observed effects has not been elucidated to date. However, we found that AS1411 strongly inhibits midkine binding to nucleolin, which consequently affects cellular processes.
In this study, we demonstrated that midkine contributes to the development of pulmonary arterial remodeling. Additionally, we elucidated the mechanism underlying midkine-induced proliferation and migration of PASMC via surface nucleolin, followed by EGFR activation. The results of our study demonstrated that midkine plays an important role in the pathogenesis of PAH, and midkine-nucleolin-EGFR axis may represent a novel therapeutic target for PAH.

Limitations
We did not identify what cells were responsible for midkine secretion. Midkine is possibly secreted from various cells such as endothelial cell and alveolar epithelial cell 30,42 , and affects these cells functions. In the current study, we particularly focused the circulating midkine roles in PAH instead of detecting the responsible cells to secrete midkine. We did not clarify the effects of midkine on cells other than PASMC in the development of PAH. Further study is required to detect the cell or tissue to secrete the midkine in PAH models. The number of human PAH samples was small and severity of PAH was relatively mild compared with other studies. We could not measure the pulmonary arterial pressure of mice because of technical difficulty. However, many studies do not perform catheterization to evaluate the pulmonary arterial pressure in PH mice model 43,44 , because pulmonary arterial pressure can be estimated by echocardiographic parameter such as pulmonary acceleration time 45 . Animal models. Midkine knockout (MK-KO) mice, with C57BL/6 background, were established as previously reported 15 . Age-matched 10-week-old male wild-type (WT) mice and MK-KO mice were exposed to hypoxic or normoxic conditions during 4 weeks. Briefly, hypoxic mice were placed in a chamber with gas-mixture of 10% O 2 and 90% N 2 by adsorption-type oxygen concentrator that utilizes the exhaust air (Teijin, Tokyo, Japan). We generated sugen-hypoxia (SuHx) induced PAH rats, which were reported to resemble severe human PAH 46 . Sprague-Dawley rats (200-250 g body weight) received a single subcutaneous injection (s.c) of Sugen 5416 (AdooQ BioScience, CA, USA, 20 mg/kg, catalog No. A12437-50) and were exposed to normobaric hypoxia as previously reported 46 . After 3 weeks of hypoxia, the animals were returned to normoxic conditions and randomly assigned to either the treatment group, which received AS1411 (10 mg/kg) or saline for 5 consecutive days, and housed under normoxic conditions for additional 2 weeks 27 . Five weeks after the treatment with Sugen 5416, animals were anesthetized with intraperitoneal injection of ketamine (80 mg/kg per hour) and xylazine (8 mg/kg per hour) mixture, in order to perform transthoracic echocardiography and right heart catheterization 47 . Right jugular vein was cannulated and optic fiber pressure catheter (420LP, SAMBA Sensors, Sweden) was advanced into right ventricle (RV) to measure right ventricular systolic pressure (RVSP). For the assessment of echocardiography, the ratio of pulmonary artery acceleration time to pulmonary artery ejection time and right ventricular area were assessed using a Vevo2100 (VisualSonics, Toronto, ON, Canada).

Methods
Pulmonary arterial remodeling was assessed as described previously 48 . The lung was fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 24 h, embedded in paraffin, and cut serially. Elastica-Masson staining was performed to measure the medial wall thickness of distal pulmonary vessels (20-100 μm) along shortest external diameter by using ImageJ software (National Institutes of Health) under a microscope (BX50; Olympus, Tokyo, Japan). Medial wall thickness was expressed as follows: Percent of wall thickness = [(external diameter-internal diameter)/external diameter] × 100. Vessels were categorized as fully-muscularized (>70% medial coat of muscle), partially-muscularized (5-70% medial coat of muscle), and non-muscularized vessels (<5% medial coat of muscle). In SuHx rat model, a quantitative analysis of luminal obstruction was www.nature.com/scientificreports www.nature.com/scientificreports/ performed in small pulmonary arteries (outer diameter <50 µm). Vessels were evaluated for occlusive lesions on Elastica-Masson slides and classified as follows: no evidence of neointimal thickness (Open); partial (<50%) luminal occlusion (Partial); and full (>50%)-luminal occlusion (Occlusive) 48 . For each animal, at least 100 vessels were measured at ×400 magnification in a blind manner. For the assessment of right ventricular hypertrophy, the RV was separated from the left ventricle (LV) and septum (S), and RV/(LV + S) ratio was calculated. pASMc primary culture. Rat pulmonary arterial smooth muscle cells (PASMCs, catalog No. R352-05a) were obtained from Cell Applications, Inc. (San Diego, CA, USA). The cells were stored in −80 °C and cultured according to the provider's instructions. PASMCs at passages 5 to 10, and at 70-90% confluence were used for experiments.
protein extraction and western blotting. Total proteins were extracted from the mouse lung and PASMCs using ice-cold RIPA buffer 48 . The protein concentration of each sample was determined using the BCA protein assay (BioRad Laboratories, Inc., Hercules, CA, USA). Equal amounts of protein were electrophoresed on 6-14% sodium dodecyl sulfate (SDS)-polyacrylamide gels and electrotransferred onto polyvinylidene difluoride membranes. Membranes were blocked by 2% bovine serum albumin (BSA) with TBS-T (20 mM Tris-HCL, pH 7.4, containing 150 mM NaCl, 0.1% Tween) and then probed with primary antibodies diluted in TBS-T. After incubation with horseradish peroxidase (HRP) -conjugated secondary antibodies diluted in TBS-T containing 5% milk, immunoreactive bands were detected using ECL kit (Amersham Biosciences, Piscataway, NJ, USA).

enzyme-linked immunosorbent assay (eLiSA). Serum midkine levels of mice and human specimens
were assessed by ELISA (for mice (catalog No. SEA631Mu), Cloud-Clone Corp. Houston, TX, USA; for humans (catalog No. OK-6149), Assay Biotech, Sunnyvale, CA USA,), according to the manufacturer's 16 instructions. Blood samples were obtained from chronic hypoxia-induced pulmonary arterial hypertension (PAH) mice and consecutive 43 human patients with suspected heart failure due to non-ischemic cardiomyopathy who underwent right catheterization in our hospital from January 2008 to December 2009. We excluded the 12 patients with increased pulmonary precapillary wedge pressure (clinical classification group 2), and patients were divided into two groups according to presence of pulmonary vascular remodeling defined as increased PVR (≧3 wood unit).
Immunofluorescence. Lung sections were treated with a blocking agent before the incubation with a primary antibody against midkine, and the sections were incubated with primary antibody overnight. This was followed by the treatment with Alexa-555 anti-rabbit secondary antibody (Invitrogen, catalog No. 43957 A) and fluorescein isothiocyanate (FITC) conjugated anti-smooth muscle actin antibody (catalog No. F3777) for 1 h at room temperature, which was accompanied by the incubation with 4′, 6-diamidino-2-phenylindole (DAPI (catalog No. 135-1303); Lonza, Bazel Switzerland) before mounting. The slides were observed under an immunofluorescence microscope (DP-70, Olympus) 15 . Fluorescent intensity was quantified and normalized to the background intensity using ImageJ software (version 1.42; National Institutes of Health, Bethesda, MD; https://imagej. nih.gov/ij/). immunoprecipitation. Cells were lysed in the modified RIPA buffer. Protein extracts were precleaned with protein A/G beads (Invitrogen, catalog No. 1969807) for 30 min. After the centrifugation of protein extracts at 4500 ×g for 2 min, supernatants were incubated with 2 µL of anti-nucleolin antibody overnight at 4 °C. After the incubation, supernatants were incubated with protein A/G beads for 1 h. Pellets were washed four times with RIPA buffer, resuspended in it, and analyzed using 10% SDS-polyacrylamide gel electrophoresis (PAGE).
Separation of subcellular fractionation. Cells were lysed with a fractionation buffer containing 250 mM sucrose, 20 mM HEPES (pH 7.4), 10 mM potassium chloride, 1.5 mM magnesium chloride, 1 mM ethylenediaminetetraacetic acid, 1 mM dithiothreitol, and protease inhibitor cocktail. Nuclear pellets were obtained by centrifugation at 720 ×g. Supernatant were collected and centrifuged again at 10,000 ×g to remove organelle components. To obtain the cell membrane fraction, supernatants were centrifuged at 100,000 ×g for 1 h. Nuclear pellets and membrane pellets were dissolved in RIPA buffer 50 . proliferation assay. Cell proliferation was measured by BrdU incorporation by using a proliferation assay kit (catalog No. 1164-7229001; Roche, Basel, Switzerland) according to the manufacturer's instructions. PASMCs were plated in 96-well plates (5 × 10 3 cells/well). After 24 h, PASMCs were starved for 48 h in the serum-free medium, and afterward, this medium was replaced with Dulbecco's modified eagle medium (DMEM) containing 1% fetal bovine serum (FBS). Cells were treated with 10 µM AS1411 or PBS 2 h before the stimulation with midkine (100 ng/mL) or vehicle, and they were exposed to normoxic or hypoxic conditions (FiO 2 0.03). BrdU label solution was added to each well 16 h before the analysis 51 . Denaturing solution was added to each well and cells were incubated for 30 min at room temperature. Afterward, HRP-conjugated anti-BrdU antibody was added www.nature.com/scientificreports www.nature.com/scientificreports/ to each well and cells were incubated for 1 h at room temperature. The absorbance was read at 450-655 nm on a Benchmark microplate reader (Bio-Rad) and normalized to vehicle control levels.
Migration assay. PASMCs migration was evaluated using the scratch assay 48 . PASMCs were seeded into 24-well collagen coated dish (1.5 × 10 5 /well). Twenty-four hours after the incubation, PASMCs were starved in serum-free medium for 48 h, a linear wound was created using p1000 pipette tips, and medium was replaced with DMEM containing 0.1% FBS. AS1411 (10 µM) was added 2 h before midkine stimulation and cells were exposed to either normoxia or hypoxia (FiO 2 0.03) for 24 h. The scratched area covered by the cells was measured using ImageJ software and normalized to the level of control.
Statistical analysis. All values are expressed as mean ± standard error (SE), except in Table 1, where the data are presented as mean ± standard deviation (SD). P values for pairwise comparisons of groups were calculated from the Student t distribution. Multiple comparisons between groups were analyzed by ANOVA followed by Tukey's Honest Significant Difference test. P-values <0.05 were considered statistically significant. Statistical analyses were performed using R version 3.0.1.