FGF21 alleviates pulmonary hypertension by inhibiting mTORC1/EIF4EBP1 pathway via H19

Abstract Long non‐coding RNAs (lncRNAs) play a significant role in pulmonary hypertension (PH). Our preliminary data showed that hypoxia‐induced PH is attenuated by fibroblast growth factor 21 (FGF21) administration. Therefore, we further investigated the regulatory role of long non‐coding RNAs in PH treated with FGF21. RNA sequencing analysis and real‐time PCR identified a significantly up‐regulation of the H19 after FGF21 administration. Moreover, gain‐ and loss‐of‐function assays demonstrated that FGF21 suppressed hypoxia‐induced proliferation of pulmonary artery smooth muscle cells partially through upregulation of H19. In addition, FGF21 deficiency markedly exacerbated hypoxia‐induced increases of pulmonary artery pressure and pulmonary vascular remodelling. In addition, AAV‐mediated H19 overexpression reversed the malignant phenotype of FGF21 knockout mice under hypoxia expose. Further investigation uncovered that H19 also acted as an orchestra conductor that inhibited the function of mechanistic target of rapamycin complex 1 (mTORC1) by disrupting the interaction of mTORC1 with eukaryotic translation initiation factor 4E–binding protein 1 (EIF4EBP1). Our work highlights the important role of H19 in PH treated with FGF21 and suggests a mechanism involving mTORC1/EIF4EBP1 inhibition, which may provide a fundamental for clinical application of FGF21 in PH.

FGF21 administration. Moreover, gain-and loss-of-function assays demonstrated that FGF21 suppressed hypoxia-induced proliferation of pulmonary artery smooth muscle cells partially through upregulation of H19. In addition, FGF21 deficiency markedly exacerbated hypoxia-induced increases of pulmonary artery pressure and pulmonary vascular remodelling. In addition, AAV-mediated H19 overexpression reversed the malignant phenotype of FGF21 knockout mice under hypoxia expose. Further investigation uncovered that H19 also acted as an orchestra conductor that inhibited the function of mechanistic target of rapamycin complex 1 (mTORC1) by disrupting the interaction of mTORC1 with eukaryotic translation initiation factor 4E-binding protein 1 (EIF4EBP1). Our work highlights the important role of H19 in PH treated with FGF21 and suggests a mechanism involving mTORC1/EIF4EBP1 inhibition, which may provide a fundamental for clinical application of FGF21 in PH.

K E Y W O R D S
fibroblast growth factor 21, long non-coding RNAs, molecular mechanism, pulmonary hypertension, pulmonary vascular diseases glucose, lipid metabolism and insulin resistance were found in early studies. 3−5 Recently, mounting studies have reported the protective role of FGF21 in cardiovascular diseases. For example, FGF21 was found to ameliorate diabetes-induced endothelial dysfunction in mouse aorta 6 and prevent angiotensin II-induced hypertension in mice. 7 However, the role of FGF21 in PH is unclear. Our preexperimental research indicated that FGF21 protein expression levels were downregulated in pulmonary arterioles, serum and lung tissues in rats with hypoxia-induced PH (HPH). 8 Exogenous administration of FGF21 relieved PH and right ventricular hypertrophy 9 . Therefore, we put forward the following hypothesis: The FGF21 gene expression is reduced when the body responds to hypoxic stimulation and may act as a protective factor in the pathological process of HPH. In this study, FGF21 knockout mice were used to further study its function in vivo. In addition, the underlying mechanism in PH treated with FGF21 is thought to be complex and has most certainly attracted our great interest.
Recently, numerous studies on non-coding RNAs have involved them in the pathogenic mechanisms of many diseases. 10−12 Long non-coding RNAs (LncRNAs) are a class of diverse non-coding RNA transcripts with a length of more than 200 nucleotides (nt), and no protein coding potential. 13 The role of lncRNA in PH has attracted widespread interest, which has led to the extensive use of highthroughput sequencing to characterize the transcriptome profile of lncRNAs associated with PH. 14,15 Numerous studies have shown that various lncRNAs, such as lncRNA MEG3 and lncRNA MANTIS, play a key role in the occurrence and development of PH. 16,17 In the present study, changes in lncRNA expression profiles after treatment with FGF21 in PH were detected by RNA-sequencing. Among the top upregulated lncRNA transcripts after treatment with FGF21, lncRNA H19 (H19) was validated and selected for further study. H19 is a highly conserved imprinted gene located in the 11p15·5 region of chromosome. 18 Studies by Li et al. 19 and Maegdefessel et al. 20 revealed that H19 is implicated in the regulation of pathophysiological processes, such as hypoxia and apoptosis in cardiomyocytes and abdominal aortic smooth muscle cells. Thus, we hypothesized that H19 may play an important role in PH treated with FGF21.
However, little is known about the regulatory role and mechanisms of H19 in PH treated with FGF21. In this study, we found that FGF21 promotes H19 expression, and then, H19 acts as a key protective factor, involved in the treatment of PH by FGF21. We also investigated the effect of H19 in PH treated with FGF21 through functional verification, and identification of downstream mechanisms involved. were purchased from the Animal Center of the Chinese Academy of Sciences (Shanghai, China). All animals were allowed to acclimatize to the specific pathogen-free animal facility for at least one week before use. All procedures were conducted in strict compliance with relevant regulations approved by the Animal Ethics Committee of Wenzhou Medical University.

| Animals and HPH models
The construction of HPH mouse model was described previously. 21 Briefly, C57BL/6 mice were randomly divided into seven groups (n = 6 per group), including N, N+F21 KO, H, H+F21, H+F21 KO, H+AAV-H19 and H+F21 KO+AAV-H19. FGF21 was obtained from Prospec (CYT-281, Prospec, Rehovot, Israel). Briefly, FGF21 (1 mg) was dissolved in 10 ml 0·1% bovine serum albumin (BSA) solution. Then, the medicine was injected intraperitoneally into mice at the prescription dose of 200 μg/kg/day before entering the hypoxia chamber. The AAV virus was constructed and injected into the mouse through the tail vein at a dose of 1 × 10 11 viral genome (vg)/mouse at the first day, and repeated at the 14th day. All the hypoxia groups were housed in a normobaric hypoxic chamber (9%-11% oxygen concentration) all days for 4 weeks, and all the normoxia groups were exposed to room air. All mice were provided access to food and water freely.

| Invasive hemodynamic and RV hypertrophy measurements
Invasive haemodynamics measurements were measured as described. 9 To assess the mean right ventricular pressure (mRVP) and the mean carotid arterial pressure (mCAP), a Bio Amp, and a PowerLab 8/35 multichannel biological signal recording system (AD Instruments, AUS) were used. To assess the RV hypertrophy, the right ventricle was dissected from the left ventricle and interventricular septum, and then, these dissected samples were weighed to obtain the ratio of right ventricle weight to left ventricle plus interventricular septum weight (RV/LV+S) as well as the ratio of right ventricle weight to body weight (RV/BW).

| Tissue staining
10μm-thick paraffin-embedded lung tissue blocks were sectioned as described previously. 9 Haematoxylin-Eosin (HE) staining was performed according to the routine protocols. The pulmonary arteries were selected randomly (external diameters of 25-100 mm) with a Nikon inverted microscope (Nikon, Tokyo, Japan). We calculated the ratios of the pulmonary artery wall area to the total area (WA/TA) and the ratios of the wall thickness to the total thickness (WT/TT) to reflect pulmonary arterial remodelling by Image-Pro Plus 6·0 (Media Cybernetics, USA). Masson staining was performed according to the manufacturer's instructions (Masson's Trichrome Stain Kit, Solarbio, Beijing, China). The ratio of the collagen area to the total area of the recording area was calculated to reflect the degree of collagen deposition. Six mice were selected and scored for WA/TA, WT/TT and collagen deposition.
Images were acquired using a fluorescence microscope (Olympus, Tokyo, Japan). The ratio of Ki67 + cells over total rPASMCs of the recording area was calculated and analysed.

| RNA isolation and qRT-PCR
Total RNA was extracted from the lung tissue, and rPASMCs using Trizol (Invitrogen), and chloroform. Relative cDNA was synthesized from 1 mg of RNA using the iScript cDNA Synthesis Kit (Bio-Rad, USA). The lncRNAs and gene expression were measured using qRT-PCR (CFX96 Real-Time System, Bio-Rad, USA). β-actin or GAPDH served as endogenous controls. The relative fold change was calculated using the 2 -△△CT method. The primer sequences were shown in Table 1.

| Cell culture and transfection
Cell isolation, cultures and hypoxia exposure methods were as previously described. 21 On reaching 80%-90% confluence, rPASMCs were passaged and subsequently used for experiments at passages 4-6. RPASMCs were 60%-70% confluent at the time of treatment. The siRNAs and plasmids were synthesized by RiboBio Co.  Table S1.

| Cell proliferation assay
A cell counting kit-8 (CCK-8) assay (Dojindo, Kumamoto, Japan) was used as previously described. 19 Briefly, the rPASMCs were subjected to different treatments with or without siRNA, or plasmid, and then counted, seeded into 96-well culture plates, at a density of 8-10 ×   Experiments for each condition were performed in triplicate.

| Construction of H19 overexpression adenoassociated virus vectors
Recombinant adeno-associated virus vectors of serotype 9 (rAAV9) that expresses mouse H19 was prepared by Taitool Bioscience Co., Ltd (Shanghai, China). Mouse H19 DNA was amplified by polymerase chain reaction from mouse splenocyte complementary DNA, using the primers 5′-ACCGGGTGTGGGAGGGGGGTG-3′, and 5′-AT GACTGTAACTGTATTTATTGATGGACCCA-30, and then, DNA segment was inserted into the pAAV-CMV_bGI-PA plasmid to construct pAAV-CMV_bGI-mH19 PA vector. The AAV vector was prepared according to the previously described protocol for transfection with three plasmids, and no adenovirus, with minor changes to use the active deflation system. 22 In short, 60% fused human embryonic

| Statistical analysis
All statistical analyses were performed with GraphPad Prism 6·0 (GraphPad Software, San Diego, CA, USA). All data were expressed as mean ± standard error mean (SEM). All the results presented were represented from at least three independent experiments.
Comparisons between two groups were analysed by unpaired twotailed Student's t-test, and multiple comparisons were analysed by were considered statistically significant.

| FGF21 relieves HPH in vivo
Given that FGF21 shows beneficial effects in cardiovascular disease, 23−25 we first examined the expression level of FGF21 in HPH mice. HPH mice showed decreased hepatic FGF21 expression and plasma FGF21 levels, and exogenous administration of FGF21 increases hepatic and plasma FGF21 levels ( Figure S1A-B, Supplementary material and method). Compared with heathy controls (HCs), circulating FGF21 levels were notably decreased in subjects with high-altitude PH (HAPH) (Supporting Information Figure   S1C). These data demonstrate that FGF21 is downregulated in HAPH patients and plays a role in PH disease.
We next investigate the effect of FGF21 on HPH. The increased mRVP and right ventricular hypertrophy index (reflected by RV/(S+LV) and RV/WT) in hypoxia-exposed mice compared with normoxia-  Figure S2 and Table 2). The expression changes of the top10 co-downregulated lncRNAs (N > H & H+F21 > H) were validated by qRT-PCR in the tested lung tissue samples ( Figure 2D).
Among them, H19 (NR_130974.1) was abundantly expressed in the H+F21 group ( Table 2). The phylogenetic analysis revealed that H19 is highly conserved in humans, mice and rats ( Figure 2E) and is mainly located in the cytoplasm according to lncLocator ( Figure 2F).
Moreover, H19 has been reported to be a multifunctional lncRNA that regulates cell growth. 20,26 Thus, we hypothesize that H19 may play an important regulatory role in HPH.

| FGF21 partially reduces hypoxia-induced rPASMC proliferation through H19 in vitro
To further study the functional role of H19 in HPH, we additionally examined H19 expression in rPASMCs. The qRT-PCR analysis results showed that H19 was upregulated by FGF21 in a dose-dependent manner in the range from 12.5 to 100 ng/ml after exposure to hy- Overall, these results indicate that FGF21 partially reduces hypoxiainduced rPASMC proliferation through H19 in vitro.

| Hypoxia-induced PH augmentation in FGF21 KO mice
In this study, Fgf21 KO mice on a C57BL/6 background were used to determine the role of FGF21 in the pathogenesis of PH ( Figure S3).
MRVP was further elevated in Fgf21 KO mice compared with that in littermate controls under hypoxia conditions, whereas this effect was not observed under normoxia conditions ( Figure 4A-B). There was no obvious difference in the mCAP of Fgf21 KO mice compared with that in the littermate controls, with or without exposure to hypoxia ( Figure 4C). However, RV/(S+LV) and RV/WT in Fgf21 KO mice exposed to hypoxia were found to be markedly higher than those in littermate controls. These effects were particularly pronounced under hypoxia conditions, but some minor changes were also observed under normoxia conditions ( Figure 4D-E). H&E staining was used to evaluate pulmonary arterial remodelling. The ratios of WA/ TA, and WT/TT in Fgf21 KO mice exposed to hypoxia were significantly increased compared with littermate controls ( Figure 4F-H).
In addition, the Masson's trichrome staining ( Figure 4F, 4I) revealed that the excessive deposition of collagen around the pulmonary arteries is associated with the exposure to hypoxia. Together, these results indicated that the loss of FGF21 markedly exacerbates PH.

| H19 overexpression reverses the malignant phenotype of PH in wild-type (WT) and FGF21 KO mice
We next assessed whether AAV-mediated H19 (AAV-H19) overexpression could reverse hypoxia-induced PH in WT mice and   Figure 6B). The tuberous sclerosis complex (TSC1/2) has been established as the major upstream inhibitory regulator of mTOR. 30,31 We next examined the expression of TSC1 and found FGF21 promoted the protein expression level of TSC1 ( Figure 6C).
Activated mTORC1 directly phosphorylates one of its main downstream substrates, namely the eukaryotic translation initiation factor 4E binding protein 1 (EIF4EBP1) and the ribosomal S6 kinase (S6K), which play a significant part in the physiological control of translation initiation. 32 In the HPH mouse model, FGF21 also inhibited the expression of EIF4EBP1, phospho-EIF4EBP1, S6K and phospho-S6K ( Figure 6D-E). We further verified the interference of FGF21 on the mTORC1/EIF4EBP1 signalling pathway using FGF21 KO mice. As anticipated, FGF21 knockout activated the protein level and phosphorylation level of the mTORC1, EIF4EBP1 and S6K, while decreased the protein expression level of TSC1 ( Figure 6F-I).

| FGF21 inhibits the mTORC1/EIF4EBP1 axis via H19 under hypoxia exposure in vitro
Through gain-and loss-of-function assay, we found that phosphorylated and total mTOR, EIF4EBP1 and S6K expression levels were significantly decreased and TSC1 levels were increased by the overexpression of H19 upon exposure to hypoxia ( Figure 7A). The knockdown of H19 increased mTOR, EIF4EBP1 and S6K expression, and phosphorylation levels, and decreased the TSC1 expression ( Figure 7B).
Moreover, the overexpression of H19 enhanced the effect of FGF21 on mTOR, EIF4EBP1 and S6K ( Figure 7C).
EIF4EBP1 binds to EIF4E, thus repressing cap-dependent protein translation at the 5′ cap of the mRNA. In addition, mTORC1 can phosphorylate EIF4EBP1, leading to the separation of EIF4EBP1 from EIF4E, thereby promoting mRNAs translation related to cell survival, proliferation and angiogenesis. 33 Wu et al. reported that H19 disrupted the interaction of EIF4EBP1, and mTORC1, thus inhibiting mTORC1 activity in pituitary tumours. 33 To assess whether H19 plays a similar role in HPH, we used a double luciferase reporter gene assay. The firefly luciferase reporter gene was driven by the CMV promoter (CMV-LUC) to assess the effect of mRNAs translation ( Figure 7D). The transcriptional activity of the luciferase reporters containing CMV-LUC was markedly stronger than that of reporters without CMV-LUC, thus confirming the efficiency of the LUC system ( Figure 7D). In this study, co-transfection of a plasmid containing H19 significantly reduced the transcriptional activity of CMV-LUC, suggesting that H19 might interact with the EIF4EBP1 to inhibit mRNA translation ( Figure 7D). . Data are presented as mean ± SEM, unpaired two-tailed Student's t-test, compared with H group, *p < 0·05, **p < 0·01, ***p < 0·001, ****p < 0·0001 Taken together, these results showed that FGF21 inhibits the mTORC1/EIF4EBP1 axis via H19 under hypoxic conditions.

| DISCUSS ION
Mounting investigations have reported the cardio-cerebral vascular protective effect of FGF21. Our preliminary data found that exogenous administration of FGF21 effectively relieved HPH in rodent models, which partially fills the gap of FGF21 in PH research. 8,9 Our research showed that the reduction of FGF21 level in the serum of HAPH patients, as well as in the serum and liver of mice HPH model, which suggesting FGF21 may act as a protective factor. We found that exogenous treatment with FGF21 could reverse hypoxia-induced pulmonary artery pressure, aggravates pulmonary vascular remodelling and extravascular collagen deposition, whereas FGF21 deficiency exacerbated these negative effects in HPH mice (Figures 1 and 4). These results show that FGF21 is effective against vascular proliferation, inflammatory and hypoxic damage, indicating the potential therapeutic use  Figure 2). H19 is a highly conserved and imprinted lncRNA with multiple biological functions in different diseases. However, the regulatory role of H19 in HPH remains unknown, and it could be an interesting target to explore further.
A recent study showed that H19 expression is decreased in failing hearts from mice, and humans, and acts as a promising therapeutic target for the treatment of pathological cardiac remodelling. 38  In contrast, Zhang 44 and Schultheiss et al. 45 showed that H19 is downregulated in hepatocellular carcinoma. This discrepancy may be due to different sample size and methods. In this study, the administration of AAV-H19 clarifies the positive role of H19 in PH treated with FGF21 in vivo ( Figure 5).
In addition, pathway enrichment analysis of our RNA sequencing data revealed the involvement of the mTORC1 signalling pathway in mediating the effects of FGF21. MTOR is one of the components of mTORC1, and indirectly reflects the levels of mTORC1. The activation of mTORC1 induces cell proliferation, and growth by promoting protein synthesis, ribosome biogenesis, lipid biogenesis and energy metabolism through its downstream targets EIF4EBP1, p70S6K and its substrate S645. 46−48 In this study, FGF21 promoted the expression of H19, which in turn inhibits the mTORC1, EIF4EBP1 and S6K, suggesting that the anti-proliferation effect of FGF21, and H19 can be attributed to the inhibition of mTORC1 signalling (Figures 6 and 7). The upregulation of FGF21 leads to a decrease of mTOR signalling pathway activity in a mouse model of neurodegeneration, which supports this notion. 49 In addition, H19 inhibits the mTOR signalling pathway in hepatic fibrosis, 50  H19-responsive CMV promoter luciferase reporter was transfected into 293T cells with indicated plasmids (n = 4, ANOVA-Bonferroni posthoc test). Data are presented as mean ± SEM, *p < 0·05, **p < 0·01, ***p < 0·001 the expression of TSC1 (Figures 6 and 7 33 Consistent with reported studies, we found, through the luciferase reporter gene experiments, that H19 appears to directly bind EIF4EBP1, thereby attenuating the transcription, and translation effects of EIF4EBP1 (Figure 7).
This study emphasis that FGF21 promotes expression of H19, which in turn to inhibits the mTORC1/EIF4EBP1 signalling pathway under hypoxia exposure ( Figure S4). This study suggests that FGF21 and H19 could serve as potential targets for the development of a therapy for PH.

ACK N OWLED G EM ENT
We are especially grateful to Dr. Yinuo Lin from the Department of Cardiology, the First Affiliated Hospital of Wenzhou Medical University, for the correction and guidance of the article.

CO N FLI C T O F I NTE R E S T S
The authors have declared that no competing interest exists.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw data presented in this study will be made available by the authors without undue reservation.