Nintedanib ameliorates experimental pulmonary arterial hypertension via inhibition of endothelial mesenchymal transition and smooth muscle cell proliferation

Neointimal lesion and medial wall thickness of pulmonary arteries (PAs) are common pathological findings in pulmonary arterial hypertension (PAH). Platelet-derived growth factor (PDGF) and fibroblast growth factor (FGF) signaling contribute to intimal and medial vascular remodeling in PAH. Nintedanib is a tyrosine kinase inhibitor whose targets include PDGF and FGF receptors. Although the beneficial effects of nintedanib were demonstrated for human idiopathic pulmonary fibrosis, its efficacy for PAH is still unclear. Thus, we hypothesized that nintedanib is a novel treatment for PAH to inhibit the progression of vascular remodeling in PAs. We evaluated the inhibitory effects of nintedanib both in endothelial mesenchymal transition (EndMT)-induced human pulmonary microvascular endothelial cells (HPMVECs) and human pulmonary arterial smooth muscle cells (HPASMCs) stimulated by growth factors. We also tested the effect of chronic nintedanib administration on a PAH rat model induced by Sugen5416 (a VEGF receptor inhibitor) combined with chronic hypoxia. Nintedanib was administered from weeks 3 to 5 after Sugen5416 injection, and we evaluated pulmonary hemodynamics and PAs pathology. Nintedanib attenuated the expression of mesenchymal markers in EndMT-induced HPMVECs and HPASMCs proliferation. Phosphorylation of PDGF and FGF receptors was augmented in both intimal and medial lesions of PAs. Nintedanib blocked these phosphorylation, improved hemodynamics and reduced vascular remodeling involving neointimal lesions and medial wall thickening in PAs. Additionally, expressions Twist1, transcription factors associated with EndMT, in lung tissue was significantly reduced by nintedanib. These results suggest that nintedanib may be a novel treatment for PAH with anti-vascular remodeling effects.


EndMT of HPMVECs
HPMVECs were cultured in Microvascular Endothelial Cell Growth Medium-2 (Lonza). TGF-β2 (2.5 ng/mL), tumor necrosis factor (TNF)-α (2 ng/mL), and interleukin (IL)-1β (4 ng/ mL) were used for the induction of EndMT as previously described [17]. The cytokines were added to the medium with or without a 3-h preincubation with nintedanib (1 μM). We had confirmed this concentration of nintedanib to have maximum inhibitory effects in preliminary experiments. The expressions of von Willebrand factor (vWF) and CD31 proteins, which are endothelial markers, as well as fibronectin and collagen 1 proteins, which are mesenchymal markers, were analyzed at 48 h after stimulation by western blotting to confirm EndMT of the HPMVECs.

Proliferation assay of HPASMCs
HPASMCs were cultured in Smooth Muscle Growth Medium-2 (Lonza). We evaluated the proliferation of HPASMCs using both the Cell Counting Kit-8 (CCK-8) and Bromodeoxyuridine (BrdU) ELISA kit as previously described, respectively [18,19]. HPASMCs were treated with multiple growth factors, specifically 5% fetal calf serum, PDGF-BB (30 ng/mL), FGF2 (2 ng/mL), epidermal growth factor (EGF) (0.5 ng/mL), and insulin-like growth factor-1 (IGF-1) (0.5 μg/mL) with or without imatinib (3 μM) or nintedanib (0.3 μM). The viability of HPASMCs was evaluated time-dependently after stimulation. In advance of this assay, we examined the concentration-dependent inhibitory effects of nintedanib and imatinib, the latter of which is another TKI for PDGF signaling, on the proliferation of HPASMCs induced by PDGF-BB. Based on these preliminary results, we selected 0.3 μM and 3 μM of nintedanib and imatinib, respectively, as concentrations that produced maximum inhibitory effects. We also performed LDH assay to assess the cell toxicity of nintedanib and imatinib as previously described [20]. Value of LDH in the medium with nintedanib and imatinib were normalized by that in the control medium without TKI. The protein expression of extracellular signal-regulated kinase (ERK)1/2 and AKT with or without phosphorylation, which are downstream effectors of PDGF and FGF signaling, was also evaluated by western blotting at 2 h after stimulation of the HPASMCs.

Preparation of pulmonary hypertensive rats
The PAH rat model was established using Sugen 5416 (VEGF receptor-1,-2 inhibitor) and chronic hypoxic exposure as previously described [14,21] Briefly, adult male Sprague Dawley rats (150-180 g) were injected with Sugen 5416 (20 mg/kg; Cayman Chemical Co, MI) subcutaneously and exposed to hypobaric hypoxia (360 mmHg, 10% O 2 ) for 3 weeks. After returning to normoxic conditions (760 mmHg, 21% O 2 ), the rats were analyzed immediately [SuHx (3W) group] or were chronically treated with either vehicle (0.5% hydroxyethylcellulose) [SuHx(5W) group] or nintedanib [50 mg/kg/day, SuHx(5W) + Nin group] by oral gavage for 2 weeks. We determined the dose of nintedanib based on previously published experiments [15]. Control rats received a single vehicle injection and were exposed to normoxic conditions for 5 weeks, with vehicle or nintedanib (50 mg/kg/day) gavage from 3 to 5 weeks [Nx(5W)] and [Nx(5W) + Nin] groups, respectively. Hemodynamic measurements were performed at 3 or 5 weeks after vehicle or Sugen 5416 injection. mg/kg intraperitoneal). A polyvinyl catheter was inserted into the right ventricle (RV) via the right jugular vein for the measurement of RV systolic pressure (RVSP) with the PowerLab data acquisition system (AD Instruments, CO). Systemic systolic arterial pressure (SAP) and heart rate (HR) were continuously monitored, and we excluded the rats with an HR of consistently less than 300 beats/min. After the hemodynamic measurements, all rats were euthanized by an overdose of pentobarbital sodium, and their hearts and lungs were collected for RV / left ventricle (LV) + septum weight ratio (RV/LV+S) measurements to show RV hypertrophy and for histological evaluations. The right lungs were stored for protein measurement, and the left lungs were inflated with 10% buffered formalin at a constant pressure of 20 cm H 2 O for histological analyses. Fixation was allowed to proceed overnight.

Morphological analysis of PAs
The inflated and fixed left lungs were embedded in paraffin. All sections were cut at 5-μm thickness and were stained with elastic Van Gieson stain. A quantitative analysis of PA luminal obstruction was performed as described previously with minor modifications [14]. We counted between 100 and 200 small PAs [outer diameter (OD) < 200 μm] per whole left lobe at ×400 magnification using an image analysis system (KS400; Carl Zeiss Imaging Solutions, Germany) in a blind manner. We measured the ratio of medial wall area to luminal area to analysis of medial wall thickening for the arteries with an OD between 50 and 200 μm as described previously with minor modifications [23]. Vessels with an OD < 50 μm were used for the assessment of occlusive neointimal lesions and were scored as follows: no evidence of occlusive neointimal formation (grade 0), partial luminal occlusion (� 50%; grade 1), and severe luminal occlusion (> 50%; grade 2).

Expression of Twist1 in rat lung tissues
The expression of Twist1 protein, which is an EndMT-related transcription factor [11], was evaluated by western blotting in lung tissues from Nx(5W), SuHx(5W), and SuHx(5W) + Nin groups to assess the contribution of EndMT in the vascular remodeling of PAs in rat PAH and to assess the anti-EndMT effect of nintedanib.

Statistical analysis
Data are presented as means ± SE. Statistical analysis was performed using one-way ANOVA (Prism 6; GraphPad Software, La Jolla, CA, USA). Differences were considered significant at P < 0.05.

Inhibitory effect of nintedanib on EndMT of HPMVECs
We confirmed decreased expression of vWF and CD31 proteins and increased expression of fibronectin and collagen 1 proteins by western blotting. Nintedanib attenuated the upregulation of mesenchymal markers in the stimulated HPMVECs, but did not prevent the downregulation of endothelial markers (Fig 1A and 1B). After stimulation with TGF-β2, TNF-α, and IL-1β, the morphology of HPMVECs changed from a cobblestone to spindle-shaped morphology, and nintedanib tended to prevent its change. (Fig 1C).

Inhibitory effect of nintedanib on proliferation of HPASMCs
The proliferation of HPASMCs induced by multiple growth factors (PDGF-BB, FGF2, EGF, and IGF) was significantly greater than that without stimulation in CCK-8 and BrdU assays, and nintedanib significantly inhibited this proliferation at 24 and 48 h after stimulation. Moreover, the inhibitory effect of nintedanib was significantly greater than that of imatinib at 48 h after stimulation in CCK-8 assay (Fig 2A and 2B). Values of LDH in the medium after treatment with imatinib and nintedanib were similar between all groups ( Fig 2C). The phosphorylation of ERK1/2 and AKT was increased in the HPASMCs stimulated with multiple growth factors, and nintedanib remarkably prevented these phosphorylations. This preventive effect of nintedanib was significantly greater than that of imatinib on the phosphorylated AKT ( Fig  2D).
The ratio of medial wall area to luminal area of PAs (50 to 200 μm OD) in [SuHx(5W) +Nin] group was significantly reduced compared to that in SuHx(3W) and SuHx(5W) groups. The grade 1 and grade 2 neointimal occlusive lesions in the small PAs (< 50 μm OD) were also significantly lower in [SuHx(5W) + Nin] than in SuHx(5W). Although we found that treatment with nintedanib reversed the ratio of medial wall area to luminal area and grade 1 neointimal occlusive lesions compared with the SuHx(3W) values, nintedanib treatment lessened, but did not reverse, the development of grade 2 neointimal occlusive lesions (Fig 4A and 4B).  Inhibitory effect of nintedanib (Nin) and imatinib (Ima) on proliferation of human pulmonary arterial smooth muscle cells (HPASMCs) using the CCK-8 assay. Cell viability of HPASMC was assessed at 0, 12, 24, and 48 h after stimulation with multiple growth factors (MGF, platelet derived growth factor-BB + fibroblast growth factor 2 + epidermal growth factor + insulin-like growth factor-1). Con: control HPASMC without MGF stimulation. Values are means ± SE (n = 5). � p<0.0001 MGF vs. Con. § p<0.01 MGF vs. MGF + Ima. # p<0.001 MGF vs. MGF + Nin. ¶ p<0.01 MGF + Ima vs. Con. ϯ p<0.01 MGF + Nin vs. MGF + Ima. (B) Inhibitory effect of Nin and Ima on proliferation of HPASMC using the BrdU assay. Cell viability of HPASMC was assessed at 24 and 48 h after stimulation with MGF. �� p<0.001. Plotted values are means ± SE (n = 6). (C) Cell toxicity of Nin and Ima on HPASMC using the LDH assay. Values of LDH were assessed at 24 and 48 h after treatment of Nin and Ima. Value of LDH in the medium with Nin and Ima were normalized by that in the control medium. Plotted values are means ± SE (n = 6). (D) Expressions of ERK1/2, phosphorylated ERK1/2 (pERK1/2), AKT, and phosphorylated AKT (pAKT) protein with or without Nin or Ima by western blotting. Representative blots are shown in the upper panels. Densitometric signals of the phosphorylated protein were normalized to the corresponding non-phosphorylated protein. Plotted values are means ± SE (n = 6). � p<0.01. �� p<0.001. n.s. not significantly.  Both medial wall thickening and neointimal lesions with the severe PH group tended to be progressive compared to those of the less severe PH group.

Phosphorylation of FGF and PDGF receptors in PAs
The expression of phosphorylated FGFR1 and PDGF receptor-β were significantly increased not only in PAs with thickening medial wall (OD 50-200 μm), but also in PAs with occlusive neointimal lesions (OD < 50 μm), in contrast to the low levels in the PAs of Nx(5W) lungs. The phosphorylation of FGFR1 and PDGF receptor-β in PAs was significantly decreased by 2-week treatment with nintedanib [SuHx(5W) + Nin] (Fig 5A and 5B).

Expression of Twist1 in rat lung tissue
The expression of Twist1 protein by western blotting was significantly augmented in lung tissue from SuHx(5W) rats compared with that from Nx(5W) rats, and nintedanib remarkably reduced this augmented expression (Fig 6).

Discussion
This report assessed the beneficial effect of nintedanib for PAH both in vitro and in vivo. Nintedanib attenuated the progression of EndMT in HPMVECs and inhibited the proliferation of HPASMCs in vitro. In addition, we demonstrated that phosphorylated PDGF and FGF receptors were increased in vascular occlusive neointimal lesions and the thickening medial wall in PAH rats. Two-week treatment of PAH rats with nintedanib significantly ameliorated pulmonary hemodynamics accompanied with improved vascular remodeling of PAs.

Mechanism of pulmonary arterial remodeling in PAH
Medial wall thickening and occlusive neointimal lesion of PAs are common pathological findings in PAH. In the remodeling process of PAs in PAH, medial wall thickening driven by proliferation and hypertrophy of smooth muscle cells occurs early in disease progression. Moreover, the contribution of functional changes in smooth muscle cells of different phenotypes occurs particularly in the establish phase [25]. On the other hand, a greater contribution of neointimal lesions than of medial wall thickening in the development of PAH was shown in an experimental model [14], and the role of EndMT in the progression of neointimal lesions has been recently suggested [10][11][12][13].
The essential role of EndMT in the pathogenesis of various cardiovascular diseases and PAH has been recently demonstrated [26]. Occlusive neointimal lesions in small PAs, including plexiform lesions, are characteristic in PAH, and the proliferative cells that comprise these PAH-specific neointimal lesions have both endothelial and mesenchymal phenotypes [9,11]. Inducers, such as a hemodynamic stress, mechanical injury, hypoxia, and inflammation, upregulate several growth factors and cytokines, including PDGF, FGF, and TGF-β, which play important roles in the progression of EndMT [27,28]. Endothelial cells lose their intercellular adhesion after exposure to these stimuli, and subsequently change from a cobblestone to spindle-shaped morphology. EndMT-induced endothelial cells express both endothelial and mesenchymal markers, and acquire further capacity to proliferate, migrate, and avoid apoptosis [10]. In a PAH model, both endothelial and mesenchymal markers were seen by immunohistochemistry in neointimal lesions of small PAs [12]. Recent reports also have shown that green fluorescent protein (GFP)-labeled endothelial cells in mice were transformed by Sugen 5416 injection and chronic hypoxic exposure to a mesenchymal phenotype with high proliferative and migratory abilities [13]. In human PAH, Ranchoux et al. demonstrated by electron microscopy that caveolae, which are characteristic of vascular endothelial cells, and dense bodies, which are dominant in vascular smooth muscle cells, were present in the same cells comprising occlusive neointimal lesions. Moreover, the expression of Twist1, a transcription factor associated with EndMT, was higher in human PAH lung than in normal human lung [11]. These data indicate a critical contribution of EndMT in the progression of neointimal lesions in PAH.

Tyrosine kinase inhibitors for PAH
Imatinib is a TKI for Bcr-Abl, c-Kit, c-Abl, and PDGF receptor signaling, and has been used for the treatment of chronic myelogenous leukemia and gastrointestinal stromal tumors [29]. Imatinib was also expected to be a novel anti-vascular treatment for PAH due to its inhibitory effect on PDGF receptor. Imatinib was shown to inhibit the proliferation of HPASMCs from PAs in human PAH [30], and long-term treatment with imatinib prevented the development of chronic hypoxia-and monocrotaline-induced pulmonary hypertension in rats [31,32]. Furthermore, imatinib combined with various vasodilators significantly improved 6-minute walking distance and pulmonary vascular resistance in human patients, and a phase III clinical study of imatinib for PAH was implemented [33]. However, imatinib did not receive final approval for PAH because of its serious adverse effects. Dasatinib, a TKI for Bcr-Abl, c-kit, Src family and PDGF receptor signaling, has been used also to treat chronic myelogenous leukemia. Importantly, dasatinib-induced pulmonary hypertension has been previously reported, suggesting specific toxicity of dasatinib on pulmonary vessels [34,35].
Nintedanib is a TKI for PDGF, FGF, and VEGF receptors. FGFR1, 2, and 3; PDGF receptor-α and -β; and VEGF receptor-1, -2, and -3 are receptors that are targeted by nintedanib [36]. Nintedanib was originally developed for the treatment of malignancies as an agent to inhibit angiogenesis, cell proliferation, and migration [37]. Nintedanib also reduced the mRNA and protein expression of extracellular matrix components in fibroblasts from human IPF [38], indicating the ability of nintedanib to treat IPF. Additionally, a recent phase III clinical study for human IPF showed beneficial effects of nintedanib in improving forced vital capacity and reducing the incidence of acute exacerbations. Diarrhea and liver dysfunction are major adverse effects of nintedanib, but these events were generally well tolerated [16]. Hence, nintedanib has been approved globally for the treatment of IPF. In contrast to dasatinib, the results from the phase III study and postmarketing surveillance revealed no evidence to warrant elevating the risk of pulmonary hypertension by nintedanib. Based on this background, we hypothesized that nintedanib was a novel and tolerable treatment for PAH because of its anti-vascular remodeling via inhibition of PDGF and FGF signaling.

Effect of nintedanib in vitro
In this study, nintedanib significantly prevented the increased expression of mesenchymal markers in EndMT-induced HPMVECs. A recent report showed that nintedanib reduced TGF-β signaling via inhibition of SMAD3 and p38 MAPK pathways [38]. TGF-β signaling is considered a major regulator of EndMT, suggesting that the inhibition of TGF-β signaling is one of the mechanisms of the anti-EndMT effect by nintedanib. Moreover, previous reports showed that PDGF and FGF enhanced signaling of p38/AKT and PI3K/AKT pathways. Increased p38/PI3K/AKT signaling affected the expression of endothelial markers involving vWF and CD31 via transcription factors involving Twist1 [26]. These data suggest that nintedanib can also regulate EndMT via inhibition of PDGF/FGF receptor tyrosine kinase activity in addition to TGF-β signaling. On the other hand, we showed that treatment with nintedanib did not restore the reduced expression of endothelial markers. HPMVECs were pretreated with nintedanib before induction of EndMT in this study. Blockade of VEGF signaling by nintedanib may disrupt the maintenance of endothelial cell characteristics, because VEGF signaling generally plays an endothelial protective role [39].
In HPASMCs, nintedanib also significantly prevented the proliferation induced by multiple growth factors, and the inhibitory effect of nintedanib was significantly greater than that of imatinib after co-stimulation with multiple growth factors. The LDH assay showed no difference even after treatment with nintedanib, suggesting no toxicity of nintedanib for HPASMCs. ERK1/2 and AKT, which are downstream effectors of PDGF and FGF signaling, play roles in the regulation of vascular smooth muscle cell proliferation and apoptosis. Nintedanib also remarkably reduced the phosphorylation of ERK1/2 and AKT after co-stimulation, and this inhibitory effect of nintedanib was greater than that of imatinib. Moreover, past report indicated that oxidative stress, one of important inducer of vascular remodeling in PAH, promoted the fibrotic change of pulmonary smooth muscle cells in systemic sclerosis [40], which change was caused via activation of PDGF and FGF signaling [41]. These in vitro results suggest a beneficial, anti-remodeling effect of nintedanib in the intima and media of PAs in PAH.

Effect of nintedanib in vivo
Classical animal models of pulmonary hypertension that are induced by chronic hypoxic exposure and monocrotaline have no neointimal lesions in the small PAs. Hence, these models were considered inappropriate for the study of PAH. In the present study, we used SuHx-PAH rats, which have neointimal lesions similar to those in human PAH [42], to evaluate the effect of nintedanib. Toba et al. provided a detailed evaluation of the time course progression of vascular remodeling in the SuHx-PAH rat [14]. They observed thickened medial walls and neointimal lesions in PAs from 3 to 5 weeks, whereas fibrotic vasculopathy involving plexiform lesions progressed 8 weeks following injection of Sugen 5416. Although the medial wall thickening decreased to a normal range during the late phase of PAH even with an elevated RVSP, the increased density of severely occlusive PAs was obviously correlated with RVSP elevation. Thus, the authors concluded that the contribution of medial wall thickening was transient in the early phase of PAH, and the role of neointimal lesions in the small PAs seemed to be greater in the progression of disease. Whereas neointimal lesions are an important therapeutic target for the treatment of PAH, vascular lesions with substantial fibrosis are generally considered refractory to treatment [43]. Therefore, we tested the anti-remodeling effect of nintedanib from 3 to 5 weeks in this study, targeting non-plexiform cellular neointimal vascular lesions. The phosphorylation of PDGF and FGF receptors significantly increased in the proliferative neointimal lesions and medial walls in the PAH rat, indicating that these PAH-specific neointimal lesions could be a major therapeutic target of nintedanib. In fact, nintedanib treatment from 3 to 5 weeks significantly reduced the expression of those receptors and prevented the progression of neointimal lesions, resulting in improved pulmonary hemodynamics in the PAH rats. Furthermore, we showed increased expression of Twist1 protein in the lung tissue from SuHx-PAH rats. Twist1 enhances the expression of TGF-β receptor and phosphorylation of SMAD2, and can lead to EndMT in PA endothelial cells [44]. Chronic nintedanib significantly reduced Twist1 protein expression in this study. These findings indicate not only the contribution of EndMT in the development of proliferative occlusive neointimal lesions in PAH rats, but also the anti-EndMT effects of nintedanib. Additionally, it has been reported that BIBF1000, a structural precursor of nintedanib, did not disturb RV function in rats subjected to mechanical pressure overload by pulmonary artery-banding [45]. Another recent study also showed that less dilatation, decreased fibrosis and hypertrophy of right ventricle after chronic treatment with nintedanib in Su/Hx-PAH rat by echocardiography and histological analysis [46], suggesting no cardiac toxicity of nintedanib.
A recently published study demonstrated that nintedanib had no therapeutic effect on rat and human PAH [47], which contradicted our results. Several factors could explain the discrepancy between the studies. For instance, rat strain that were used for experimental PAH were different. A previous report has shown that the severity of the response to Sugen5416 with hypoxic exposure was remarkably different between rat strains [48]. Actually, the RVSP of SuHx PAH in Wistar-Kyoto rats was obviously lower in a recent study than in this study. Therefore, the response to nintedanib also may be dissimilar between strains. As another factor, muscularization of the small PAs was evaluated to assess the effect of nintedanib in a recent study. Such measurement is common for hypoxic or monocrotalineinduced pulmonary hypertension models that have no neointimal lesions, but it is not appropriate for the SuHx PAH rat which has neointimal lesions in the small PAs. In addition, pathological images of small PAs were not shown in that study; thus, it is unclear which parts of the small PAs were evaluated. Furthermore, the delivery route of nintedanib to the rat was not mentioned. Thus, the possibility remain that differences in the administration route of nintedanib may be the cause of the contradictory results. In contrast, Huang et al. reported that chronic treatment with nintedanib reduced pulmonary vascular remodeling in the Fos-related antigen-2 mouse model of systemic sclerosis [49], which was supportive of our results. On the other hand, another more recent study showed that chronic treatment with nintedanib of established late-phase PAH, from 8 to 11 weeks, did not improved pulmonary hemodynamics and vascular remodeling in SuHx rats [46]. The results of these recent studies of established animal and human PAH suggest that the therapeutic power of nintedanib may be limited at least in advanced stages of PAH. Although the efficacy of nintedanib for mild to moderate PAH is still uncertain, the therapeutic benefit of nintedanib may rest with its ability to augment the effect of other vasodilators or to prevent the progression of PAH.

Limitations
Our study has a few limitations. First, we have not yet assessed the phenotype of HPASMCs after stimulation; investigations into the effect of nintedanib on the functional changes in HPASMCs may bring further beneficial information. Second, while HPASMC used in this study is mainly isolated from the proximal PA, contribution of small distal PA is greater than that of large proximal PA in the disease progression of PAH. Thus, data would be more useful by using HPASMC isolated from only small distal PA. Third, the mechanism of the inhibitory effect of nintedanib on EndMT progression is still unclear. We induced EndMT in HPMVECs by using three different inducers simultaneously; nintedanib could have inhibited all three signaling pathways. The inhibitory mechanism was probably very complex and beyond the scope of the current study. Fourth, higher concentrations or longer regimens of nintedanib treatment for PAH in the rat have not yet been evaluated. Further experiments are necessary in the future to evaluate these limitations of our study.

Conclusion
In conclusion, we have shown the beneficial effects of nintedanib for PAH in vitro and in vivo. Chronic treatment with nintedanib reversed the elevated pulmonary arterial pressure in the PAH rat via anti-EndMT effects in HPMVECs and anti-proliferative effects in HPASMCs. Moreover, an expanded indication of nintedanib for the treatment of PAH would be advantageous, since nintedanib has been approved already for the treatment of other human diseases with good tolerability. Although further investigations are necessary, the results of this study indicate that nintedanib may be a novel additional option for the treatment of human PAH, with an anti-vascular remodeling effect. Funding acquisition: Tetsutaro Nagaoka.