Glucagon-like peptide-1 (GLP-1) mediates the protective effects of dipeptidyl peptidase IV inhibition on pulmonary hypertension

Pulmonary hypertension (PH) is a progressive disease leading to death ultimately. Our recently published data demonstrated that inhibiting dipeptidyl peptidase IV (DPP-4) alleviated pulmonary vascular remodeling in experimental PH. However, whether glucagon-like peptide-1 (GLP-1) mediated the protective effect of DPP-4 inhibition (DPP-4i) on PH is unclear. In the present study, GLP-1 receptor antagonist (exendin-3) abolished the protective effects of DPP-4 inhibitor (sitagliptin) on right ventricular systolic pressure (RVSP) and pulmonary vascular remodeling (PVR) in monocrotaline (MCT, 60 mg/kg)-induced PH in rat. Notably, activation of GLP-1 receptor by GLP-1 analogue liraglutide directly attenuated RVSP and PVR in MCT-induced PH, as well as bleomycin- and chronic hypoxia-induced PH. Moreover, liraglutide potently inhibited MCT-induced inflammation and suppressed MCT-induced down-regulation of vascular endothelial marker (VE-cadherin and vWF) in lung. In vitro studies showed liraglutide reversed TGF-β1 (5 ng/ml) combining IL-1β (5 ng/ml) induced endothelial-mesenchymal transition (EndMT) in human umbilical vein endothelial cells (HUVECs), which could be abolished by GLP-1 receptor antagonist (exendin-3). Furtermore, liraglutide suppressed TGF-β1-IL-1β-induced phosphorylation of both Smad3 and ERK1/2. Our data suggest that GLP-1 mediated the protective effects of DPP-4i on pulmonary vascular and RV remodeling in experimental PH, which may be attributed to the inhibitory effect on EndMT.


Background
Pulmonary hypertension (PH), characterized by pulmonary vasoconstriction and vascular remodeling, is a progressive disease leading to right ventricular (RV) failure and finally death [1,2]. The hyperproliferative responses of pulmonary vascular medial smooth muscle cells contribute to the thickening of large elastic vessels and the muscularization of small arteries, thereafter leading to increased pulmonary vascular resistance [3,4]. Besides, accumulation of inflammatory cells around pulmonary arteries, including macrophages and mast cells, is another obvious profile of pulmonary vascular remodeling (PRV) [5]. Recently, endothelial-mesenchymal transition (EndMT) has been noted in the pulmonary vascular intima both in patients with PH [6,7] and in animal model of PH [6,8]. EndMT results in endothelial dysfunction and pro-inflammatory response [9]. Nowadays, inhibiting EndMT has been suggested as a promising therapeutic strategy for the treatment of PH [10].
Our previous study demonstrated that dipeptidyl peptidase IV (DPP-4) inhibition (DPP-4i) alleviated pulmonary vascular remodeling in monocrotaline (MCT)-induced PH rats [11]. DPP-4 is a serine protease which selectively cleaves off N-terminal dipeptides from its substrates such as glucagon-like peptide-1 (GLP-1) [12]. GLP-1, an incretin hormone derived from enteroendocrine cells, can be rapidly degraded by DPP-4, resulting in a short half-life time about 2 min [13]. Accumulating data from both pre-clinical and clinical studies suggest that GLP-1 or GLP-1 analogues are beneficial for cardiovascular diseases (CVDs). For instance, infusion of recombinant GLP-1 could improve LV function in animals with advanced dilated cardiomyopathy [14]. Activating GLP-1 receptor (GLP-1R) with GLP-1 analogue could improve the survival of human aortic endothelial cells after ischemia-reperfusion injury [15]. Notably, GLP-1 could induce an endothelial-dependent relaxation of norepinephrine-challenged constriction of pulmonary artery rings [16]. These results suggest activation of GLP-1/GLP-1R axis may be beneficial for pulmonary circulation under specific stress conditions. However, whether GLP-1/GLP-1R axis could mediate the therapeutic effect of DPP-4i on PH or could exert anti-PH activity itself needs to be further validated. In the present study, sitagliptin (a selective DPP-4 inhibitor), exendin-3 (a GLP-1R antagonist), and liraglutide (a GLP-1 analogue) were used to investigate the effect of GLP-1/GLP-1R axis on PRV in rat model of MCT-induced PH. In addition, the effects of GLP-1R activation on endothelial-mesenchymal transition (EndMT) of human umbilical vein endothelial cells (HUVECs) were investigated.

Ethical approval
All procedures in the present study were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (publication no. 85-23, revised 1996) and approved by the Institutional Animal Care and Use Committee of Nanjing Medical University (NJMU/IACUC-1601196).

Experimental animals and design
For MCT model of PH, 64 male Sprague-Dawley (SD) rats (200 ± 10 g) (Shanghai Bikai Laboratory Animal Company, Shanghai, China) were randomly assigned into two groups after raised under standard laboratory conditions for 1 week. One group received an intraperitoneal injection of 0.8-mL saline (n = 24), when the other group was injected intraperitoneally with MCT (60 mg/ kg, Sigma-Aldrich, MO, USA). Once rats were challenged with MCT, 16 of them were administrated with sitagliptin (SG) daily (80 mg/kg, gavage, Januvia, Merck Sharp, UK) with/without exendin-3 injection (Ex-3, 40 μg/kg, intraperitoneal, Santa Cruz Biotechnology, USA). Eight MCT-treated rats and eight saline-treated rats were randomly picked up for a daily subcutaneous injection with liraglutide (Li) at the doses of 0.2 mg/kg (NovoNordiskA/S, Denmark). The body weights of the rats were measured every other 2 days for dose adjustment during the following 4 weeks.
For bleomycin (BLM) model of PH, 32 rats were assigned into two groups randomly. One group (n = 16) was treated with 50-μL 0.9% saline intratracheally, when the other group (n = 16) was given an intratracheal administration of BLM (4 U/kg, Nippon Kayaku, Tokyo, Japan). Thereafter on the same day, half of the rats in two groups were daily injected with liraglutide (0.2 mg/ kg, subcutaneous) for 4 weeks.
For chronic hypoxia model of PH, 16 rats were housed under normal oxygen, when the other 16 rats were raised in a hypoxia chamber (Biospherix Ltd., USA) for 4 weeks in which the fraction of oxygen was maintained at 10%. Eight rats out of these two groups (normoxia and hypoxia) were injected with liraglutide daily (0.2 mg/kg, subcutaneous), when the rest rats were given 0.2 mL saline daily.

Hemodynamic measurement and specimen collection
On day 28, after rats were anesthetized by 1% pentobarbital sodium (60 mg/kg, intraperitoneal), RV systolic pressure (RVSP) was determined using right-heart catheterization via right jugular vein according to the protocol described previously [17]. Thereafter, rats were sacrificed for specimen collection. The lung and RV tissues were harvested for morphological evaluation, quantitative real-time reverse transcriptase polymerase chain reaction (RT-qPCR), and Western blot analysis. The weights of RV as well as left ventricle + interventricular septum (LV + S) were measured respectively.

Assessment of pulmonary morphology
Collected rat lung tissues and RV tissues were immediately fixed in 4% polyformaldehyde (pH 7.4) for 24 h, embedded in paraffin, and then sliced into 4 μm sections. Then lung sections were stained with hematoxylin and eosin (H&E), toluidine blue (TB), picrosirius red (PSR) staining, Masson trichrome staining (MTS), and primary antibody against CD68 (Abcam, MA, USA) according to the protocols described previously [17]. Pulmonary arterial wall thickness (PAWT) was determined according to the ratio of (external diameter − internal diameter) to external diameter based on H&E staining. The right ventricular hypertrophy was determined by the cross section area (CSA) of cardiomyocytes using H&E staining. Graphs were observed under a Leica 2500 microscope (Leica Microsystems, Wetzlar, Germany).

Identification of EndMT
After cells were challenged with TGF-β1-IL-1β for 72 h, cells were observed under a phase contrast microscope (Nikon, Tokyo, Japan), then fixed in 4% polyformaldehyde for 30 min, blocked with 3% FBS, incubated with primary antibodies against vascular endothelial cadherin (VE-cadherin) and vimentin overnight at 4°C, and then labeled with second antibody (donkey anti-mouse IgG, Alexa Fluor 555; and donkey anti-rat IgG, Alexa Fluor 488) at room temperature for 1 h. The morphologies of HUVECs were then identified under a Leica 2500 microscope. Besides, cells were also collected for western blot analysis to determine the expression of endothelial markers [VE-cadherin, zonula occludens-1 (ZO-1)] and mesenchymal markers [alpha smooth muscle Actin (α-SMA) and Vimentin].
Total RNA was extracted from lung tissues using Trizol reagent (Gibco BRL, Grand Island, NY) according to the manufacturer's instructions. Reverse transcription was conducted with SYBR Premix Ex Taq™ kit (TaKaRa Bio Inc., Dalian, China). Quantitative RCR was performed using an ABI Prism 7500 FAST apparatus (Applied Biosystems, Foster City, CA, USA). The expression for each gene was determined as the relative expression of the housekeeping gene β-actin. PCR primers are listed in Table 1.

Statistical analysis
All values are presented as means ± S.E.M.. Comparisons among groups were performed using one-way ANOVA followed by least significant difference (LSD) post hoc test. P < 0.05 was considered statistically significant.

GLP-1R antagonist abolished the protective effects of DPP-4i on PH
As presented in Fig. 1a, the expression of GLP-1 was almost undetectable in lungs of control rats, while it was significantly expressed in lungs of MCT-challenged rats. Daily treated with DPP-4i (sitagliptin, SG, 80 mg/kg) significantly inhibited MCT-induced up-regulation of GLP-1 in lung tissues. Nevertheless, this effect of SG was completely blocked by the GLP-1R antagonist (Ex-3, 40 μg/kg, daily). Similarly, SG alleviated MCT challenge resulted loss of weight gain (Δweight) in rats, which was alleviated by Ex-3 (Fig. 1b). In addition, the inhibitory effects of SG on MCT-induced increment of RVSP (Fig. 1c), hypertrophy of pulmonary vascular medial layer (Fig. 1d), hypertrophy of RV [the ratio of RV to  . 1e), and hypertrophy of cardiomyocytes (CSA) (Fig. 1f ) were abolished by Ex-3 ( Fig. 1c-f ). These results indicate that antagonism of GLP-1R could abolish the protective effects of DPP-4i on PH.
These results suggest that GLP-1 analogue could prevent MCT-induced PH in rats directly.

Liraglutide suppressed MCT-induced inflammation and EndMT in pulmonary arteries
Inflammatory cell infiltration around pulmonary vascular contributes to PRV during the development of PH [5]. The accumulation of CD68 positive macrophages around pulmonary arteries was noted in MCT-challenged rats, which was attenuated by the treatment of liraglutide (Fig. 3a).
Similarly, liraglutide treatment significantly inhibited the accumulation of TB staining positive mast cells around pulmonary arteries (Fig. 3b). Besides, MCT treatment significantly up-regulated mRNA levels of TGF-β1, IL-1β, IL-6 and TNF-α in lung tissue (Fig. 3c). Consistent with the results from inflammatory cell infiltration in lung, liraglutide completely reversed up-regulation of TGF-β1, IL-1β, TNF-α, and partially reversed up-regulation of IL-6 mRNA in lung induced by MCT treatment.
In addition, immunofluorence analysis showed liraglutide significantly reduced the number of α-SMA positive endothelial cells (labeled by CD31) (Fig. 3d), white arrow). RT-qPCR detection indicated that liraglutide partly reversed MCT-induced down-regulation of endothelial marker mRNA (VE-cadherin and vWF) and up-regulation of mesenchymal marker mRNA (vimentin and fibronectin) in lung (Fig. 3e). In addition, endothelial marker VE-cadherin along with mesenchymal marker fibronectin were selected for further examination by western blot. As indicated in Fig. 3f, liraglutide significantly reversed MCT-induced increase in fibronectin and recovered VE-cadherin partially.

Liraglutide attenuated TGF-β1 combining IL-1β induced EndMT
Recently, EndMT was reported to be involved in the development of PH and suggested to be a promising therapeutic target for the treatment of PH [10]. TGF-β1 and IL-1β are key cytokines contributing to EndMT. Therefore, TGF-β1 (5 ng/ml) in combination with IL-1β (5 ng/ml) (T&I) were used to induce EndMT in cultured HUVECs. As shown in Fig. 5a, HUVECs with cobblestone morphology transited to spindle-like morphology 72 h after T&I challenge. Morphology assay found that liraglutide inhibited T&I-induced EndMT of HUVECs in a dose-dependent manner (25, 50, 100 nM). VE-cadherin and vimentin are specific markers for endothelial cells and mesenchymal cells respectively in EndMT. After T&I challenge for 72 h, VE-cadherin between neighboring HUVECs were almost disappeared, while mesenchymal marker vimentin became abundant (Fig. 5b). When treated with liraglutide together, the percent of vimentin-positive cells undergoing EndMT decreased dose-dependently.
For further confirmation, the expressions of endothelial markers (VE-cadherin and ZO-1) and mesenchymal markers (α-SMA and vimentin) in HUVECs were detected by West blotting. As shown in Fig. 5c, T&I challenge resulted in a significant decrease in endothelial markers (VE-cadherin and ZO-1), and a remarkable increase in mesenchymal markers (α-SMA and vimentin). However, these alterations on EndMT-related cell phenotypic markers induced by T&I were significantly inhibited by liraglutide.
During T&I-induced EndMT, Smad and non-Smad signaling are the most important signaling underlying. Therefore, the phosphorylations of Smad2/3 and ERK1/2, representative kinases of Smad and non-Smad signaling respectively, were determined by Western blotting. As shown in Fig. 5d, treated with T&I for 2 h induced a significant phosphorylations of Smad2/3 and ERK1/2 in HUVECs, which were inhibited by liraglutide in a dose-dependent manner.

Discussion
In previous study, we reported that DPP-4i suppressed MCT-induced PH in rats [11]. In the current study, we showed that blockage of GLP-1R abolished protective effects of DPP-4i in MCT-induced PH. Moreover, GLP-1 analogue liraglutide successfully alleviated RVSP, pulmonary arterial remodeling, and RV remodeling in rats with MCT-PH, as well as bleomycin-and chronic hypoxia-induced PH. In vitro study on HUVECs provides evidence that GLP-1 analogue could inhibit TGF-β1&IL-1β-challenged activation of Smad/non-Smad signaling, and prevent TGF-β1&IL-1β-induced EndMT.
DPP-4, a serine protease, cleaves a large number of chemokine and peptide hormones and makes them lose biological function [12]. Among them, stromal cell-derived factor-1α (SDF-1α) [19] and brain natriuretic peptide (BNP) [20,21] presented a protective role in PH. GLP-1 is another important substrate of DPP-4, being inactivated by cleavage of the N-terminal histidine-alanine dipeptide. In the present study, we found GLP-1R antagonist abolished the inhibitory role of DPP-4i in pulmonary vascular remodeling and PH in rats induced by MCT. These results suggested that GLP-1 might be an important downstream molecule mediating the protective effect of DPP-4i  d and i), as well as reducing the CD68 positive macrophages infiltration (e and j). Data were presented as the mean ± s.e.m. (n = 6-8). *P < 0.05, **P < 0.01, ***P < 0.001. BLM, bleomycin hydrochloride on PH. Subsequent study confirmed that GLP-1 could directly prevented pulmonary vascular remodeling in different models of experimental PH. Based on these results, it is suggested that GLP-1 might be a promising target for PH management.
Mounting evidence indicate that GLP-1 exert a broad range of protective effects on cardiovascular diseases, including hypertension [22], chronic heart failure [23,24], and ischemia-reperfusion injury [25]. Not only in preconstricted systemic circulation arteries, but also in pulmonary arteries, GLP-1 receptor agonist could induce a relaxation in endothelium-dependent manner [16,26]. In current study, GLP-1 analogue liraglutide alleviates increment of RVSP in different models of PH in rat, suggesting the anti-vasoconstrictive effect may contribute to the therapeutic effect of liraglutide in experimental PH. Besides sustained constriction of pulmonary arteries, pulmonary vascular remodeling characterized by pulmonary arterial intimal and medial hypertrophy is another predominant profile of PH [27].
EndMT, the most recently discovered pathological features of PH, is a complex process in which cells detach from the endothelial layer, lose their specific molecular markers, and acquire a mesenchymal phenotype [32]. It was reported that GLP-1 protected against hyperglycemic-induced EndMT of human aortic endothelial cells and improves myocardial dysfunction [33]. In experimental PH models, EndMT can be induced by chronic hypoxia [34] or inflammatory responses [9]. Our in vivo study showed that endothelial specific markers were significantly down-regulated and mesenchymal markers were up-regulated in lungs of MCT-induced PH rats, while these changes were partially reversed by the treatment of GLP-1 analogue liraglutide. These results suggest that GLP-1R activation may inhibit EndMT in MCT-induced PH. Although we found that GLP-1 analogue significantly inhibit MCT-induced inflammation, the effects of GLP-1R activation on inflammatory cytokines-induced EndMT is unclear. Our in vitro study showed that TGF-β1 combining IL-1β induced EndMT of HUVECs as indicated by the morphologic change and alterations of endothelial/mesenchymal-specific markers. As expected, GLP-1 analogue liraglutide not only inhibited expression of mesenchymal markers (vimentin and α-SMA), but also recovered expression of endothelial markers (VE-cadherin, ZO-1). Therefore, our data suggest that inhibiting EndMT could be one of the potential mechanisms for the therapeutic effects of GLP-1 analogue on PH.
GLP-1 analogues have also been reported to prevent and reverse monocrotaline-induced pulmonary arterial hypertension by suppressing ET-1 and enhancing eNOS/ sGC/PKG pathways [40], and to improve hypoxia-induced pulmonary hypertension in mice partly via normalization of reduced ET(B) receptor expression [41]. Besides, the anti-proliferation effect of GLP-1 analogues on different vascular smooth muscle cells, including pulmonary arterial smooth muscle cells (PASMCs) [40] and aortic smooth muscle cells (ASMCs) [39,42], might also contribute to the beneficial effect of GLP-1R agonist on PH.
Increased GLP-1 was noted in lung tissues from rats with PH, which means that the rats could partly protect themself against PH via up-regulating GLP-1expresseion. On the other hand, GLP-1, expressed in the lung tissues, might also indicate the severity of PH since GLP-1 was nearly not expressed in lung tissues from control rats. It seems that DPP-4i decreased GLP-1 expression in lung tissues form rats with PH in the present study, but it is not the fact. We suspect that the decreased expression in GLP-1 was mainly attributed to the DPP-4i-mediated alleviation of PH even though DPP-4i could up-regulate GLP-1 directly. Consistent with DPP-4i treatment, GLP-1 analogue liraglutide treatment also reduced GLP-1 expression in lung tissues from rats with PH, which might also result from the alleviation of PH.

Conclusions
GLP-1 mediated the protective effects of DPP-4i on pulmonary vascular and RV remodeling in experimental PH, which may be attributed to the inhibitory effect on EndMT. Targeted on GLP-1 could be a novel therapeutic option for the treatment of PH.