Empagliflozin Attenuates Pulmonary Arterial Remodeling Through Peroxisome Proliferator-Activated Receptor Gamma Activation

Abstract

The loss of peroxisome proliferator-activated receptor gamma (PPARγ) exacerbates pulmonary arterial hypertension (PAH), while its upregulation reduces cell proliferation and vascular remodeling, thereby decreasing PAH severity. SGLT2 inhibitors, developed for type 2 diabetes, might also affect signal transduction in addition to modulating sodium-glucose cotransporters. Pulmonary arterial smooth muscle cells (PASMCs) isolated from patients with idiopathic pulmonary arterial hypertension (IPAH) were treated with three SGLT2 inhibitors, canagliflozin (Cana), dapagliflozin (Dapa), and empagliflozin (Empa), to investigate their antiproliferative effects. To assess the impact of Empa on PPARγ, luciferase reporter assays and siRNA-mediated PPARγ knockdown were employed to examine regulation of the γ-secretase complex and its downstream target Notch3. Therapy involving daily administration of Empa was initiated 21 days after inducing hypoxia-induced PAH in mice. Empa exhibited significant antiproliferative effects on fast-growing IPAH PASMCs. Empa activated PPARγ to prevent formation of the γ-secretase complex, with specific impacts on presenilin enhancer 2 (PEN2), which plays a crucial role in maintaining γ-secretase complex stability, thereby inhibiting Notch3. Similar results were obtained in lung tissue of chronically hypoxic mice. Empa attenuated pulmonary arterial remodeling and right ventricle hypertrophy in a hypoxic PAH mouse model. Moreover, PPARγ expression was significantly decreased and PEN2, and Notch3 levels were increased in lung tissue from PAH patients compared with non-PAH lung tissue. Empa reverses vascular remodeling by activating PPARγ to suppress the γ-secretase-Notch3 axis. We propose Empa as a PPARγ activator and potential therapeutic for PAH.

Brief Commentary

Background

The study addresses the role of peroxisome proliferator-activated receptor gamma (PPARγ) in pulmonary arterial hypertension (PAH) and the potential therapeutic impact of sodium-glucose cotransporter 2 (SGLT2) inhibitors. PPARγ downregulation worsens PAH, while its upregulation attenuates PAH. SGLT2 inhibitors have shown cardioprotective effects in type 2 diabetes. This study explores whether SGLT2 inhibitors impact signal transduction beyond glycemic control.

Translational Significance

These findings suggest, for the first time, that the SGLT2 inhibitor empagliflozin is a potential therapy for PAH based on its ability to activate PPARγ and inhibit presenilin enhancer 2 (PEN2), which plays a crucial role in γ-secretase complex stability and its downstream target Notch3, thereby reducing vascular remodeling. These results underscore the significance of PPARγ in regulating the γ-secretase–Notch3 axis, offering a new target for PAH treatment. These findings broaden the therapeutic approach to PAH, highlighting the novel mechanism of clinical cardiovascular protection provided by empagliflozin.

Pulmonary hypertension (PH) is diagnosed through right heart catheterization and requires a mean pulmonary artery pressure (PAP) of over 20 mmHg at rest.^1,2 Pulmonary arterial hypertension (PAH) involves structural changes in the pulmonary arteries, such as thickening of the medial and intimal layers, fibrotic obliteration, and the development of plexiform lesions. These changes result in right ventricular dysfunction and ultimately failure.^3,4 These lesions are complex, proliferative structures of pulmonary arterial endothelial and smooth muscle cells.^4,5 Current therapies (nitric oxide and prostacyclin mimetics, calcium channel blockers, and endothelin receptor antagonists) manage vasomotor tone and cell proliferation^6,7 but do not reverse vascular remodeling, making PAH incurable. Novel therapeutic approaches are required for treating PAH.

Sodium–glucose cotransporter 2 (SGLT2) inhibitors, a novel group of antihyperglycemic medications, offer clinical cardioprotection for patients with type 2 diabetes.⁸⁻¹⁰ Empagliflozin (Empa) reduces heart failure (HF) hospitalizations and adverse cardiovascular events in those with coronary artery disease.⁹ In 2020, Chowdhury B. and colleagues found that Empa reduced mortality and prevented pathological vascular changes and RV fibrosis in a PAH rat model,¹¹ though the mechanism was unclear. HF is the primary cause of death in PH patients, emphasizing the need for new treatments. SGLT2 inhibitors lower blood sugar, possess cardioprotective properties,¹²⁻¹⁴ and may interfere with HF progression through various mechanisms

Notch3, a member of the NOTCH family of receptors, has been implicated in regulating arterial smooth muscle cell (SMC) identity and proliferative capacity to evoke vascular remodeling in the development of PAH.^15,16 Cleavage of the Notch receptor by the γ-secretase complex releases the Notch3 intracellular domain (ICD), whose translocation to the nucleus is thought to be required for the transcription of target genes, such as Hes family BHLH transcription factor 5 (HES5).¹⁷ The proteolytic γ-secretase complex in the cellular membrane consists of presenilin 1 and 2 (PSEN1 and 2), nicastrin, anterior pharynx-defective 1 (APH1), and presenilin enhancer 2 (PEN2).¹⁸ PEN2 plays a crucial role in γ-secretase complex stability.¹⁹ In experimental PAH, overexpression of PEN2 was shown to promote PASMC motility and growth via γ-secretase-mediated Notch3 signaling.²⁰ In addition, the transcription factor peroxisome proliferator-activated receptor gamma (PPARγ) regulates various biological processes, such as cell growth, apoptosis, inflammation, and insulin sensitivity.^21,22 PPARs and retinoid X receptors (RXRs) are ligand-activated transcription factors that act as coactivators or corepressors to coordinately regulate gene expression.²³ Once activated, the PPARγ–RXR complex binds DNA at PPAR response elements (PPREs) in the promoter region of target genes to activate or repress transcription.^24,25 Recent evidence indicates that the loss of PPARγ causes severe PAH.²⁶⁻²⁹ PPARγ upregulation suppresses cell proliferation and vascular remodeling, thereby attenuating PH.^30,31 Mice with PH that were treated with rosiglitazone (a PPARγ activator in the thiazolidinedione (TZD) class) showed a decrease in numerous characteristics of PH, including RV hypertrophy, vascular SMC proliferation, and collagen and elastin deposition.^21,32 PPARγ activators are suggested as potential therapeutics for PH.^33,34

The role of SGLT2 inhibitors in pulmonary vascular remodeling in PAH is unclear. We hypothesized they affect signal transduction. This study used PASMCs from humans with IPAH and a hypoxia-induced PAH mouse model to explore SGLT2 inhibitors’ impact on artery thickening and vascular biology, examining cellular and molecular changes.

Materials and Methods

Patient Characteristics

Human residual lung tissues were collected from eight patients undergoing lung surgery or transplantation at National Taiwan University Hospital (Table 1); three patients did not have PAH, and five patients had IPAH. Samples were deidentified, and informed consent was obtained from all participants. The study protocol was approved by the Human Research Ethics Committees at National Taiwan University Hospital (Institutional Review Board approvals 201409069RINA and 202303018RINC) and Chang Gung Memorial Hospital (Chang Gung Medical Foundation Institutional Review Board approval 104–0287B) and conducted in accordance with the principles of the Declaration of Helsinki.

Table 1. Characteristics of Patients Lung Tissue^a.

study ID	age-gender	diagnosis
A. non-PAH patients
1	41-F	lung cancer
2	53-M	SAH
3	69-F	lung cancer

study ID	age-gender	PAP (s/d/m)	PVR (WU)	6MWD (m)	PAH medications
B. IPAH patients
1	24-F	147/56/54	13.94	157	Iloprost
					Riociguat
					Macitentan
					Bosentan
2	28-F	81/40/55	19.8	93	macitentan
3	28-M	128/48/80	20.55	447	Remodulin
					Sildenafil
					Macitentan
4	52-F	75/34/50	11.23	410	Macitentan
					Sildenafil
5	55-M	102/45/69	23.9	360	Remodulin

^aIPAH, idiopathic pulmonary arterial hypertension; SAH, subarachnoid hemorrhage; PAP, pulmonary artery pressure (mmHg); s, systolic; d, diastolic; m, mean; PVR, pulmonary vascular resistance (dynes/s com^–5); 6MWD, distance (m), 6 min walking distance.

Culture of Human Pulmonary Arterial Smooth Muscle Cells

Normal human PASMCs for the non-PAH groups were obtained from a commercial source (Lonza) (Table 2), grown in cell growth medium (Lonza) and used between passages 4 and 7. These cells were starved in serum-free culture medium for 24 h before experiments. PASMCs isolated from pulmonary vessels of patients with IPAH were grown in smooth muscle growth media-2 (Lonza).³⁵ Primary IPAH PASMCs were subcultured at a 1:4 ratio in culture dishes and used at passages 4–6.

Table 2. Characteristics of Human Non-PAH Vascular Cells^a.

human PASMC Lot. no	age, yr	sex	race	supplier
21TL347546	23	F	C	Lonza
7F3602	47	F	C	Lonza
419239	35	M	H	Lonza
559495	64	M	C	Lonza

^aM, male; F, female; C, Caucasian; H, Hispanic.

SGLT2 Inhibitor Cytotoxicity

MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assays were carried out to determine SGLT2 inhibitor cytotoxicity and the effective dose in pulmonary arterial cells. Human PASMCs were obtained from Lonza or isolated from IPAH patients and plated into 96-well plates at a density of 10,000 cells/cm². The cells were treated for 24 h in serum-supplemented culture medium with various SGLT2 inhibitors at concentrations from 0.1 to 100 μM for 24 h. Following treatment, the medium was replaced with 100 μL of MTT solution (diluted 1/10 from the 5 mg/mL stock solution) in phenol red-free culture medium. After 3 h, the medium was removed, and insoluble formazan crystals were dissolved in 100 μL/well DMSO and measured spectrophotometrically at an optical density of 570 nm. The cytotoxicity of SGLT2 inhibitors was compared to that of 10% FBS only.

Cell Proliferation Assay

The effects of the three SGLT2 inhibitors [canagliflozin (Cana), dapagliflozin (Dapa), and empagliflozin (Empa)] at concentrations of 0.1, 1, and 10 μM were measured by an ELISA-based 5-bromo-2-deoxyuridine (BrdU) incorporation assay kit (Roche Diagnostics Co.) according to the manufacturer’s instructions. In brief, non-PAH and IPAH PASMCs were seeded at 2 × 10⁴ cells/well in 24-well plates with culture medium (Lonza) and allowed to adhere overnight. Cells were treated with or without SGLT2 inhibitors for 24 h, and proliferative activity was assessed.

Immunocytochemical Analysis

Immunocytochemical analysis of PASMCs was performed using primary antibodies against Ki67 (MA5-14520 Invitrogen). At the end of the experiments, cells were rinsed with PBS and fixed with a 3% paraformaldehyde solution for 15 min at room temperature. The paraformaldehyde solution was then removed, and the cells were washed twice with 1× PBS, blocked with 1% goat serum/1% BSA in PBS for 30 min, and incubated with primary antibodies for 1 h. The staining protocol was performed according to the DakoCytomation LSAB2 System-HRP (K0675, Dako) manufacturer’s instructions to observe the locations of elastin in the PASMCs. Nuclei were visualized by hematoxylin staining (Gibco, Invitrogen).

Western Blotting

Tissue samples or cell samples were homogenized in lysis buffer containing 50 mM HEPES (pH 7.0), 250 mM NaCl, 5 mM EDTA, 0.1% NP40, protease inhibitor cocktail III (Calbiochem 539134; 1:100) and phosphatase inhibitor cocktail IV (Abcam ab201115; 1:100). Immunoblotting was performed using primary antibodies for PPARγ (57 kDa) (Abcam ab59256; 1:1000), PEN2 (13 kDa) (Cell Signaling 5451; 1:1000), nicastrin (110, 120 kDa) (Cell Signaling 5665; 1:1000), APH1α (30 kDa) (Abcam ab12104; 1:500), Presenilin 1 (PSEN1) (22 kDa) (Cell Signaling 5643; 1:1000), Presenilin 2 (PSEN2) (23 kDa) (Cell Signaling 2192; 1:1000), Notch3 (human: 97 kDa; mouse: 72 kDa) (Cell Signaling 5276; 1:3000) and HES5 (18 kDa) (Abcam ab194111; 1:1000). Peroxidase-conjugated antimouse IgG (Jackson 115-035-003; 1:10,000) or antirabbit IgG (Jackson 111–035–003; 1:10000) secondary antibodies were used as needed. The blots were visualized using an enhanced chemiluminescence detection system (Millipore WBULS0500), and bands quantified by densitometry were normalized to those of GAPDH (Santa Cruz sc-32233; 1:10000).

Small Interfering RNA (siRNA)

siRNA specific for PPARγ and scrambled siRNA (as a control) were purchased from Dharmacon (Dharmacon/Thermo Fisher Scientific). siRNAs were then transfected into PASMCs with RNAi MAX (Invitrogen) in accordance with the manufacturer’s instructions.

Cloning of the PEN2 Promoter and Construction of Chimeric Luciferase Reporter Plasmids to Assess PPARγ Activity

For the luciferase reporter experiments, a fragment of the human PEN2 promoter (nucleotides −889 to −290) was amplified by PCR from human genomic DNA (forward: 5′-ATAGGCGGGACATGGTGACGAC-3′; reverse: 5′-CGCGAGACGCCCCAGGTTTC-3′). The PCR product was ligated into the pGL3Basic vector (Promega, Madison, WI) at the MluI and XhoI sites. Putative PPARγ–RXRα binding elements (−737 to −725) were found in the PEN2 promoter. Site-directed mutagenesis of the PPARγ binding elements was performed by PCR using the Q5 Site-Directed Mutagenesis kit (NEW ENGLAND Bio Laboratories).

Transfection and Luciferase Assay

Cells were seeded into 6-well plates, grown to approximately 80% confluence and subsequently transfected with 2 μg of plasmid DNA using Lipofectamine 2000 (Invitrogen 11668019) following the manufacturer’s instructions. After 24 h, the cells were treated with Empa with or without GW9662 for 1.5 h and then lysed. Luciferase activity was measured with a luciferase assay kit (BioThema 484-001) and detected by a Varioskan LUX instrument (Thermo Scientific).

Animal Experimental Design

Mice (8-week-old C57BL/6) were exposed to chronic hypoxia (10% O₂) to induce hypoxic pulmonary vascular remodeling, as described previously.³⁰ The chronic effects of Empa were assessed in mice exposed to hypoxia for 28 days to stimulate PAH progression. After 7 days, mice were administered saline or the SGLT2 inhibitor (Empa) at a dosage of 10, 25, or 100 mg/kg/day³⁶ for 3 weeks by gavage. The animal experimental protocols were approved by the Chang Gung University IACUC (permit number: CGU110–014), and the study adhered to the National Institutes of Health recommendations in the Guide for the Care and Use of Laboratory Animals. Mice were housed and maintained at Chang Gung University, provided with a standard chow diet, and given free access to water. Surgeries were performed under IP anesthesia with Zoletil 20 mg/kg (Virbac) and 5 mg/kg Rompun (Bayer) to minimize suffering, and animal tissues and blood samples were collected after sacrifice.

Hemodynamic Measurements and Cardiovascular Evaluation

The mice were anesthetized by IP injection of 20 mg/kg Zoletil (Virbac) and 5 mg/kg Rompun (Bayer) to minimize suffering, mean arterial blood pressure (MBP) data were analyzed using the BP-98A tail-cuff method (Softron, Japan), and the right ventricular systolic pressure (RVSP) was measured using a 1F catheter with a pressure-tipped transducer (Millar) through the right jugular vein. After sacrifice, the weight of RV wall was separated from the left ventricular (LV) wall and ventricular septum. The weights of the RV wall, LV wall, and ventricular septum were determined. RV hypertrophy was expressed as the ratio of RV wall weight to the combined weight of the free LV wall and ventricular septum.

Paraffin Embedding and Microscopy

The lungs were preserved in a 10% paraformaldehyde solution and divided by lobe into tissue blocks for paraffin embedding. The resulting tissue blocks were sliced into 5 μm sections to observe the type of muscular PA remodeling. Vascular medial hypertrophy was characterized using α-SMA staining (Sigma–Aldrich, MS-113-P; 1:1000) to determine the percent medial wall thickness (MWT). The MWT was measured using NIS Elements Imaging software from Nikon at 200× magnification, and the diameter was calculated and defined as follows: (longest diameter + shortest diameter)/2.

Immunohistochemical Analysis

Immunohistochemical analysis of human or mouse lung tissues was performed with primary antibodies against α-smooth muscle-actin (α-SMA) (Sigma–Aldrich, MS-113-P; 1:1000), PPARγ (Cell Signaling, 2435; 1:100), PEN2 (Cell Signaling, 5451; 1:100) and Notch3 (Abcam, ab26878; 1:500). Staining of α-SMA was used to indicate the medial layer of small PAs. To analyze PPARγ, PEN2, and Notch3 expression, lung tissue sections were incubated with rabbit antibodies against these proteins and mouse antibodies against α-SMA for 1 h. Afterward, a secondary antibody was applied for 30 min [Alexa-488-conjugated secondary antibody for α-SMA (green, Abcam, ab150113; 1:500) and Alexa-594-conjugated secondary antibody for PPARγ, PEN2, and Notch3 (red, Abcam, ab150080; 1:1000)], and DAPI (Abcam, ab104139) staining was used to visualize nuclei. Fluorescence imaging was performed using a confocal microscope (Confocal TCS SP8XL; Leica) at the Microscope Core Laboratory of Chang-Gung Memorial Hospital.

Data Analysis

All data are presented as the mean ± SEM. The 2-tailed Student’s t test was used to evaluate differences between two groups, while one-way ANOVA with Dunnett’s or Bonferroni post hoc correction was employed to compare data among multiple groups. Statistical significance was considered at P < 0.05.

Results

Effect of Empa on IPAH SMC Proliferation and Migration

To determine the cytotoxicity of SGLT2 inhibitors in pulmonary arterial cells, an MTT assay was performed using human PASMCs obtained from Lonza as controls (non-PAH) or isolated from IPAH patients. After 24 h, the cells were treated for 24 h in serum-supplemented culture medium with various SGLT2 inhibitors (Figure 1A) at concentrations from 0.1 to 100 μM. The results showed cytotoxicity of the SGLT2 inhibitors at 100 μM in PASMCs (Figure 1B,C). Therefore, a reasonable dose range for SGLT2 inhibitors was determined to be 0.1–10 μM. To further characterize the effects of SGLT2 inhibitors, including Cana, Dapa, and Empa, non-PAH or IPAH PASMCs were induced to proliferate in culture medium containing 10% FBS and then treated with SGLT2 inhibitors (Figure 1D,E); the results showed that IPAH PASMC proliferation was inhibited by Empa starting at 0.1 μM but not by Cana or Dapa (Figure 1D,E). The proliferation of IPAH PASMCs induced by 10% FBS was inhibited by Empa in a dose-dependent manner (0.1, 1, and 10 μM) (Figure 1D,E). Empa inhibited significantly the proliferation of IPAH PASMCs but not of non-PAH PASMCs as evidenced by Ki67 immunofluorescence staining (Figure 1F,G). Treatment of IPAH PASMCs with Empa led to a significant decrease in serum-induced proliferation compared with vehicle treatment. Thus, fast-growing IPAH PASMCs may be more sensitive to Empa.

Figure 1.

Effects of SGLT2 inhibitors on serum induced proliferation. (A) Structures of compounds: (1S)-1,5-anhydro-1-C-[3-[[5-(4-fluorophenyl)-2-thienyl]methyl]-4-methylphenyl]-d-glucitol (Canagliflozin) (Cana); (1S)-1,5-anhydro-1-C-[4-chloro-3-[(4-ethoxyphenyl)methyl]phenyl]-d-glucitol (Dapagliflozin) (Dapa); (1S)-1,5-anhydro-1-C-[4-chloro-3-[[4-[[(3S)-tetrahydro-3-furanyl]oxy]phenyl]methyl]phenyl]-d-glucitol (Empagliflozin) (Empa). (B, C) SGLT2 inhibitors inhibited the viability of PASMCs. PASMCs were treated with SGLT2 inhibitors (0.01, 0.1, 1, 10, and 100 μM). Data are presented as the mean ± SEM for each group (n = 4 per group). Statistical significance was evaluated by one-way ANOVA with Dunnett’s post hoc test. *P <0.05 and ***P < 0.001 compared with the untreated group. (D, E) Empa significantly inhibited the proliferation of IPAH PASMCs only. (F, G) Ki67 immunofluorescence staining. Data are presented as the mean ± SEM (n = 4). **P < 0.01 and ***P < 0.001 compared to the untreated group; one-way ANOVA with Dunnett’s post hoc test.

Empa Stimulates PPARγ Expression and Reduces the γ-Secretase-Notch3 Axis in PASMCs

IPAH PASMCs exhibit lower PPARγ expression^34,37 and increased activity of the γ-secretase–Notch3 signaling axis, which is associated with cell proliferation and has been implicated in the pathogenesis of PAH;^15,20 however, the effect of Empa on this critical signaling pathway in PAH has not been evaluated. We first evaluated whether Empa, an SGLT2 inhibitor that offers clinical cardioprotection to patients with type 2 diabetes,⁸ is associated with PPARγ or the γ-secretase–Notch3 signaling axis in human non-PAH or IPAH pulmonary arterial cells. Western blot analysis of whole-cell lysates showed that Empa upregulated PPARγ and downregulated the γ-secretase subunit PEN2 and its downstream targets Notch3 and HES5 in a time- and dose-dependent manner. In the time course experiment, PPARγ levels were increased at 1–24 h, and the levels of the γ-secretase subunit PEN2 and the downstream targets Notch3 and HES5 were decreased from 1 to 24 h in Empa-treated human IPAH PASMCs (Figure. 2). In the dose–response experiment, IPAH PASMCs were treated with 0.1, 1, or 10 μM Empa (Figure 3). Western blot analysis of whole-cell lysates showed that Empa treatment increased PPARγ levels and decreased the levels of the γ-secretase subunit PEN2 and the downstream targets Notch3 and HES5 in a dose-dependent manner. These results indicate that Empa attenuates PASMCs by activating PPARγ and reducing signaling through the γ-secretase complex, specifically targeting PEN2 and the downstream targets Notch3 and HES5.

Figure 2.

Empagliflozin induces PPARγ protein expression and downregulates γ-secretase/Notch3 signaling. (A) Western blots showing protein levels in non-PAH and IPAH PASMCs treated with 10 μM empagliflozin (Empa) for up to 24 h. GAPDH was used as the loading control. (B) Densitometric quantification of protein expression. Data are shown as the mean ± SEM (n = 3). *P < 0.05, **P < 0.01, and ***P < 0.001 versus the untreated group; one-way ANOVA with Dunnett’s post hoc test.

Figure 3.

Effect of empagliflozin on PPARγ-mediated inhibition of γ-secretase signaling. (A) Western blots showing protein levels in IPAH PASMCs treated with 0, 0.1, 1, or 10 μM empagliflozin (Empa) with GAPDH as the loading control. (B) Densitometric quantification of protein expression. Data are shown as the mean ± SEM (n = 3). *P < 0.05, **P < 0.01, and ***P < 0.001 versus the untreated group; one-way ANOVA with Dunnett’s post hoc test.

Empa Suppresses PEN2 Expression via PPARγ Activation

PEN2 plays a crucial role in maintaining γ-secretase complex stability and trafficking.¹⁹ Silencing of PEN2 through RNA interference reduces PSEN1 and PSEN2 levels, impairs nicastrin maturation, and leads to deficient γ-secretase complex formation.³⁸ Furthermore, PEN2 overexpression has been shown in experimental PAH to promote the motility and growth of SMCs via γ-secretase-mediated Notch3 signaling.²⁰ Therefore, we hypothesized that Empa may activate PPARγ to decrease the expression of γ-secretase, with specific effects on PEN2 and the downstream Notch3 signaling pathway, which together mediate the development of PAH in PASMCs. Western blot analysis of whole-cell lysates showed that 10 μM Empa treatment increased PPARγ levels and decreased the levels of the γ-secretase subunit PEN2 and its downstream targets Notch3 and HES5, but these changes could be reversed by 10 μM GW9662 (a PPARγ antagonist), thus reflecting the crucial role of PPARγ in the effects of Empa (Figure 4A,B). Moreover, PPARγ is a nuclear transcription factor that regulates target genes by recognizing PPREs in promoter regions.²³ Here, PPARγ–RXRα binding elements (−737 to −725) were found in the PEN2 promoter. To further confirm the critical role of PPARγ nuclear translocation and binding to the PEN2 promoter in the biological activity of Empa, we evaluated the ability of Empa to activate PPARγ and thus directly affect the downregulation of PEN2 by PPARγ. We cloned a plasmid to monitor the Empa-driven transactivation of a PEN2-specific reporter (MLP-luc). Bioinformatic analysis identified one putative PPARγ–RXRα binding site located at −737 to −725 bp in the PEN2 promoter of PASMCs. The luciferase reporter analyses confirmed that PEN2 transcription was reduced by Empa only when the promoter construct contained the PPARγ binding site, and this effect was reversed by the PPARγ antagonist GW9662 (Figure 4C). Mutational analyses confirmed that PPARγ reduced PEN2 transcription only when the promoter construct contained the PPRE (Figure 4C). These results demonstrate that Empa activates PPARγ to suppress transcription of the γ-secretase component PEN2, which may be involved in PASMC proliferation and pulmonary arterial remodeling. To determine whether PPARγ regulates γ-secretase–Notch3 signaling in PASMCs, we then knocked down PPARγ expression in human PASMCs (Lonza) with siRNA (Dharmacon/Thermo Fisher Scientific). siRNA-mediated knockdown of PPARγ significantly increased the expression of the γ-secretase subunits PEN2, nicastrin, and APH1α and the downstream targets Notch3 and HES5 (Figure 4D,E). PEN2 stabilizes γ-secretase, and PEN2 silencing impairs complex formation and nicastrin maturation and decreases PSEN1/2 levels.^19,38 Our findings suggest the significance of PPARγ in reducing levels of the γ-secretase subunit PEN2 along with its downstream targets Notch3 and HES5 in PASMCs. We present the initial discovery of Empa as a PPARγ activator and the identification of a new signaling pathway known as the PPARγ–γ-secretase–Notch3 axis in PAH PASMCs.

Figure 4.

Impact of empagliflozin on PPARγ activation to suppress the γ-secretase–Notch3 axis via transcriptional regulation of PEN2. (A) Western blots showing protein levels in IPAH PASMCs treated with empagliflozin (Empa) (10 μM) and then treated with or without GW9662 (10 μM). GAPDH was used as the loading control. (B) Densitometric quantification of protein levels. Data are shown as the mean ± SEM (n = 3). *P < 0.05, **P < 0.01, and ***P < 0.001 compared with the untreated group; #P < 0.05, ##P < 0.01 and ###P < 0.001 compared with the empagliflozin (Empa) group; one-way ANOVA with the Bonferroni post hoc correction. (C) Left: schematic of the putative PPARγ–RXRα binding site in the PEN2 gene promoter with the mutated constructs. Right: Graph of the luciferase assay results. PASMCs were transiently transfected with wild-type (WT) or mutant promoter constructs and treated with or without 10 μM empagliflozin (Empa). Data are shown as the mean ± SEM (n = 4). ***P < 0.001 versus the control group; ###P < 0.001 versus the empagliflozin group; one-way ANOVA with the Bonferroni post hoc correction. (D) PASMCs were transfected with PPARγ siRNA or scramble RNA. Representative immunoblots and (E) densitometric quantification of protein expression (n = 3). *P < 0.05, **P < 0.01 compared with the scramble siRNA group; 2-tailed Student’s t test.

Effect of Empa on Hemodynamic and Structural Changes in Chronically Hypoxic Mice

We showed that Empa inhibits IPAH pulmonary arterial cell proliferation and stimulates PPARγ expression to reduce γ-secretase–Notch3 signaling in PASMCs in vitro. We aimed to evaluate whether Empa treatment can effectively prevent the development of PAH in mice exposed to hypoxia (10% O₂). Mouse body weight was tracked weekly from 8 to 12 weeks of age. Chronically hypoxic mice were treated with Empa at doses of 10, 25, and 100 mg/kg/day by oral gavage for 4 weeks (Figure 5A). Empa treatment significantly decreased body weight (Figure 5B), which may be related to its action on renal effects.³⁹ In 12-week-old mice, the body weight changes were potentially due to the effects of the SGLT2 inhibitor Empa, which was administered for 3 weeks (Figure 5B). No effect on heart rate (Figure 5C) or mean arterial pressure (MAP) was observed (Figure 5D). Consequently, chronically hypoxic mice developed severe PH within 28 days, characterized by a significant increase in RV systolic pressure (RVSP) compared with that in normoxic mice. Empa treatment (10, 25, or 100 mg/kg/day) significantly reduced RVSP to the level in the normoxic mice (Figure 5E). Reduced RVSP was associated with reduced RV weight (Figure 5F). However, in the 100 mg/kg/day treatment group, Empa had a weaker effect on RV weight than in the 10 mg or 25 mg groups, which may be due to a potential dose–response relationship, wherein higher doses saturate the medication’s effects or the effects plateau. The 100 mg/kg/day dosage might have exceeded the optimal therapeutic range for reducing RV weight, potentially triggering physiological responses or side effects that counteracted its cardioprotective effects. Additionally, genetic or physiological factors in the high-dose group might have made mice less responsive to the benefits of Empa. We further quantified the degree of muscularization of PAs with a diameter ranging from 25 to 100 μm by α-SMA staining and measuring the MWT. As expected, the medial wall of PAs was significantly thicker in the hypoxia group than in the control group (Figure 5G,H). In chronically hypoxic mice, Empa significantly reduced the MWT of PAs compared to no treatment. Our findings demonstrate that Empa effectively prevents PAH development in mice exposed to hypoxia.

Figure 5.

Impact of empagliflozin on pulmonary hemodynamics and pulmonary arterial structural changes in hypoxia-induced pulmonary hypertension. (A) Design and optimization of the appropriate treatment strategy. (B) After 1 week of adaptive feeding, normoxic mice and hypoxic mice were exposed to normoxia or 10% O₂, respectively, for 4 weeks, and some mice were also treated for 3 weeks with empagliflozin (Empa) (10, 25, 100 mg/day) by oral gavage. (n = 10 normoxia group; n = 5–10 hypoxia/empagliflozin (Empa) groups; and n = 20 hypoxia group). Body weight changes in mice after 3 weeks of empagliflozin intervention. Data are presented as the mean ± SEM (C) Heart rate (beats per minute; bpm) and (D) mean arterial pressure (MAP) as determined by mouse tail cuff. (E) Right ventricular systolic pressure (RVSP). (F) Ratios of RV to LV plus the septum weight (RV/LV+S) in 5 different groups. (G, H) Vascular medial wall thickness (MWT) of PAs (25–100 μm) were identified by α-SMA staining. Pulmonary vascular thickness (scale bar = 25 μm) was identified by α-SMA staining (brown). Data are presented as the mean ± SEM for each group. Statistical significance was evaluated with one-way ANOVA and the Bonferroni post hoc correction. *P < 0.05; **P < 0.01, ***P < 0.001 compared with the normoxia group; #P < 0.05, ##P < 0.001, ###P < 0.001 compared with the hypoxia group.

Effect of Empa on PPARγ, γ-Secretase Complex, and Notch3 Expression in Chronically Hypoxic Mice

In this study, Empa inhibited IPAH pulmonary arterial cell proliferation and stimulated PPARγ expression to reduce γ-secretase–Notch3 axis signaling in PASMCs in vitro. Accordingly, we determined whether Empa can effectively activate PPARγ and suppress the expression of γ-secretase subunits, along with their downstream targets Notch3 and HES5, in the lungs of chronically hypoxic mice. Empa treatment significantly increased PPARγ expression levels and reduced the protein levels of PEN2 and Notch3, as detected by immunohistochemistry (PPARγ, PEN2, and Notch3 localization) (Figure 6A–C) and Western blot analysis (quantification of protein expression) (Figures 6 D and S1). These results indicate that Empa can activate PPARγ to downregulate the expression of γ-secretase subunits, specifically PEN2. Inactivated γ-secretase reduces Notch3/HES5 signaling in pulmonary arterial remodeling.

Figure 6.

Effect of empagliflozin on PPARγ, PEN2, and Notch3 expression. (A) Immunohistochemistry analysis of (A) PPARγ, (B) PEN2, and (C) Notch3 protein levels in the distal (scale bar: 25 μm) parts of pulmonary arteries (PAs). These proteins in PAs were identified by α-SMA double staining of vascular SMCs, and nuclear staining with DAPI is shown in blue. (D) Western blot analysis of PPARγ, PEN2, and Notch3 levels in the lung tissue of hypoxia-induced PAH mice treated with empagliflozin (10 mg/day). Data are presented as the mean ± SEM for each group (n = 3 per group). Statistical significance was evaluated with one-way ANOVA and the Bonferroni post hoc correction. *P < 0.05 and ***P < 0.001 compared with the normoxia group; #P < 0.05 and ##P < 0.01 compared with the hypoxia group.

Overexpression of the γ-Secretase Subunits PEN2 and Notch3 and Decreased PPARγ Expression in PAs are Associated with IPAH

Evidence indicates that the loss of PPARγ causes severe PAH.^21,30 Overexpression of γ-secretase and Notch3 has been implicated in regulating arterial SMC identity and proliferative capacity, leading to vascular remodeling and PAH development.^15,20 Here, in lung tissues of patients without or with IPAH, we investigated the protein levels of PPARγ and the γ-secretase subunit PEN2 and the target protein, Notch3 in PASMCs. PPARγ, PEN2 and Notch3 were detected by immunofluorescence in the medial layer of PAs of patients without PAH, and these proteins colocalized with α-SMA, indicating that PPARγ is expressed in the SMCs of PAs. In contrast, PPARγ signals were hardly detectable in PAs of IPAH patients (Figure 7A), but PEN2 (Figure 7B) and Notch3 (Figure 7C) were overexpressed in the lungs of patients with IPAH. Quantification of protein levels by Western blot analysis revealed a significant decrease in PPARγ (57 kDa) and significant increases in PEN2 (13 kDa) and Notch3 (97 kDa) in IPAH PASMCs compared with non-PAH PASMCs (Figure 7D). The findings show that the pulmonary arterial medial layers of patients with IPAH are characterized by the upregulation of Notch3 and PEN2 and the downregulation of PPARγ.

Figure 7.

Change in the expression of PPARγ, PEN2, and Notch3 in PAH. Immunofluorescence analysis of the localization of (A) PPARγ, (B) PEN2, (C) Notch3 and SMA in IPAH PA tissue. (D) Western blot analysis of PPARγ, PEN2 and Notch3 in non-PAH PASMCs (n = 4) and IPAH PASMCs (n = 5). GAPDH was used as the loading control. Densitometry quantification of protein levels is shown as the mean ± SEM *P < 0.05 and ***P < 0.001 compared to non-PAH PASMCs; two-tailed Student’s t test.

Discussion

SGLT2 inhibitors have been shown to possess cardioprotective properties in different clinical trials. This study illustrates a novel mechanism for the SGLT2 inhibitor Empa, which functions as a PPARγ activator in fundamental γ-secretase-induced Notch3 signaling, as a potentially effective treatment in experimental PH. Importantly, our findings demonstrate that Empa-induced PPARγ to suppress the expression of the γ-secretase subunit PEN2, which mediates Notch3 signaling during PH. Moreover, our work indicates that the administration of Empa reversed pulmonary arterial remodeling and reduced RVSP and the RV ratio in chronically hypoxic mice. This finding provides a novel mechanism for Empa and indicates that increasing PPARγ expression to decrease γ-secretase-induced Notch3 signaling may be a novel therapeutic approach for treating PAH, namely, with SGLT2 inhibitors. The schema shows the proposed signaling events in PAH PASMCs and the effect of Empa.

We used isolated IPAH PASMCs to determine whether the three SGLT2 inhibitors Cana, Dapa, and Empa inhibit pulmonary arterial vascular cell proliferation. Our study design permitted the determination that SGLT2 inhibitors are effective in fast-growing IPAH PASMCs but not in controlling pulmonary arterial vascular cells. These three SGLT2 inhibitors,Cana, Dapa, and Empa, have been reported to inhibit vascular cell proliferation. Cana blocks the proliferation and migration of human and rat aortic SMCs⁴⁰ and of human umbilical vein endothelial cells (HUVECs),⁴¹ whereas Empa and Dapa failed to inhibit the proliferation of HUVECs at 1–30 μM.⁴¹ Most previous studies on the antiproliferative effect of SGLT2 inhibitors have utilized normal vascular cells. In PAH, dynamic and progressive pulmonary vascular remodeling is accompanied by the localization of various pulmonary vascular cells in the arterial wall, including PASMCs, PAECs, and adventitial fibroblasts.^4,15,21,42 In PASMCs from patients with IPAH, proliferation is two to three times faster than that in control (non-PAH) cells, and the known abnormalities in the various vascular cell types include the mutation causing loss of bone morphogenetic protein receptor type II (BMPRII) signaling,⁴³ the overexpression of Notch,¹⁵ high expression of γ-secretase²⁰ and a reduction in signaling through PPARγ pathways.^21,30 Therefore, this study is the first to screen and identify SGLT2 inhibitors [canagliflozin (Cana), dapagliflozin (Dapa), and empagliflozin (Empa)] that block the proliferation of PH vascular cells and exhibit different behaviors in IPAH and non-PAH cells. Our results show that IPAH PASMC proliferation was inhibited by Empa starting at 0.1 μM but not by Cana or Dapa. These findings may relate to the increased sensitivity of fast-growing IPAH vascular cells to Empa.

Clinical studies have shown that SGLT2 inhibitors provide cardioprotective benefits to patients with type 2 diabetes.^8,9 Among SGLT2 inhibitors, Empa was the first to show substantial reductions in hospitalizations due to HF and major adverse cardiovascular events in patients with type 2 diabetes with coronary artery disease.⁹ In 2021, Michael E and colleagues reported the effect of Empa on PAP in patients with HF in a randomized placebo-controlled trial, in which PAP assessments were used to demonstrate the direct effect of reductions in PAP on decongestion in patients with HF.⁴⁴ These authors suggested that mechanisms beyond diuresis likely contribute to the observed reductions in PAP with Empa.⁴⁴ Our understanding of the mechanisms underlying the off-target effects of SGLT2 inhibitors is increasing and explains the benefits observed in large clinical trials of this class of hypoglycemic drugs. Thus, we analyzed isolated IPAH PASMCs in vitro to further ascertain the unrecognized off-target effects of Empa. In IPAH PASMCs with higher proliferative capacity, the known abnormalities include overexpression of γ-secretase²⁰ and NOTCH¹⁵ and reduced PPARγ signaling.²⁹ Interestingly, Western blot analysis of whole-cell lysates showed that Empa treatment upregulated PPARγ and downregulated the γ-secretase subunit PEN2, Notch3, and HES5 in a time- and dose-dependent manner in human IPAH PASMCs. These results indicate that Empa attenuates PASMCs by activating PPARγ and reducing signaling through the γ-secretase subunit PEN2 and Notch3. These data illustrate a novel mechanism of the SGLT2 inhibitor Empa as a potent and selective PPARγ activator. Empa was the first SGLT2 inhibitor identified to lower blood glucose levels by inducing glucosuria in patients with type 2 diabetes.⁹ Among antidiabetic drugs, TZDs are insulin sensitizers that significantly reduce glucose, lipid, and insulin levels in animal models of NIDDM and obesity.⁴⁵ Moreover, TZDs are selective PPARγ activators with antidiabetic effects.⁴⁶ PPARs have been reported to regulate inflammation, lipoprotein metabolism, and glucose homeostasis and have potential clinical implications as targets for the treatment of obesity, diabetes, atherosclerosis and PAH.^33,47

Furthermore, the upregulation of PPARγ in response to Empa treatment is linked to the downregulation of γ-secretase, as well as its downstream targets Notch3 and HES5. NOTCHs are transmembrane proteins that undergo a series of proteolytic cleavages, resulting in release of the ICD, which forms a nuclear complex with DNA-binding proteins.^48,49 The ICD-containing complex functions as a transcription factor and binds the promotors of hairy/enhancer of split (HES) and related genes.^48,49 Cleavage of the NOTCH ICD to activate the canonical NOTCH pathway is achieved by the γ-secretase complex.^17,50 Increased γ-secretase activity causes the abnormal accumulation of amyloid β (Aβ) peptides that form amyloid plaques, a critical component of the pathogenesis of Alzheimer’s disease,⁵¹ or cleaves the irregular Notch3 ICD to upregulate HES5, an important contributor to the pathogenesis of PAH.²⁰ γ-secretase inhibitors have been developed as potential treatments for multiple diseases, such as Alzheimer’s disease⁵¹ and PAH.^15,20 The γ-secretase complex contains five subunits: PSEN1, PSEN2, nicastrin, APH1α, and PEN2.⁵² Further bioinformatic analysis revealed putative PPARγ–RXRα binding elements in the PEN2 and PSEN1 promoters and a PPARγ binding site in nicastrin promotor. Studies have shown that PPARγ activation can decrease the expression of γ-secretase components, including PSEN1 and nicastrin, leading to reduced γ-secretase activity in Alzheimer’s disease.^53,54 PEN2 is the key factor for γ-secretase complex stability and mediates the endoproteolytic cleavage of full-length PSENs, APH-1 and nicastrin.^19,52 We therefore focused on investigating the PPARγ–RXRα binding site in the PEN2 promotor by luciferase reporter assays. Mutation of the PPARγ–RXRα site in the PEN2 promotor abolished luciferase expression, confirming that PEN2 transcription is regulated by Empa-induced PPARγ only when the promoter contains the PPARγ–RXRα binding site. Transient transfection studies showed decreased PEN2 promoter binding in Empa-treated PASMCs, and this effect was reversed by the PPARγ antagonist GW9662. The PPARγ–RXRα complex can function as a repressor of PEN2 gene expression. Furthermore, in assays testing whether PPARγ functions as a repressor of PEN2 gene expression, siRNA-mediated PPARγ knockdown was found to increase the protein levels of PEN2 and other γ-secretase subunits that lead to increased Notch3 expression, indicating a crucial role for PPARγ-mediated suppression of the γ-secretase/Notch3 signaling pathway in the effects of Empa.

The SGLT2 inhibitor Empa has been reported to attenuate cardiac fibrosis and improve ventricular hemodynamics associated with renal protection in spontaneously hypertensive rats (SHRs).⁵⁵ In 2020, Chowdhury B. and colleagues showed that Empa reduces mortality and prevents the progression of pathological vascular changes and RV fibrosis in an MCT-induced PAH rat model¹¹ but did not elucidate the detailed mechanism. SGLT2 inhibitors may protect against the development of HF through direct and indirect effects on the heart.^12−14,55 In our current study with chronically hypoxic mice, we found marked reversal of PH in response to Empa treatment, which did not affect SAP, demonstrating the selectivity of this approach for abnormal pulmonary circulation. Empa treatment significantly reduced RVSP to the level in the controls. Reduced RVSP is associated with reduced RV weight. However, 100 mg/kg/day Empa had a weaker effect on RV weight than 10 or 25 mg/kg/day Empa in mice with hypoxia-induced PAH. It is possible that Empa exhibits a dose–response relationship, where at higher doses, its effects become saturated or plateau. The high dosage of 100 mg/kg/day might have exceeded the optimal therapeutic range for reducing RV weight. The high dose of Empa may have triggered certain physiological responses or side effects that counteracted its cardioprotective effects on the right ventricle. In addition, it is possible that the mice in the high-dose group had genetic or physiological factors that made them less responsive to the cardioprotective effects of Empa on the right ventricle compared to those in the lower dose groups. These are potential explanations, and further experiments are needed to determine the exact reasons for the observed differences in RV weight among the dosing groups. Interestingly, weekly body weight tracking from 8 to 12 weeks of age revealed that 3 weeks of Empa treatment significantly decreased body weight, which may be due to the renoprotective effect of this inhibitor. Our findings demonstrate that Empa effectively prevents PAH development in mice exposed to hypoxia. Knowledge of the mechanisms involved underlying the off-target effects of SGLT2 inhibitors is progressively increasing and explains the benefits observed in large clinical trials of this class of hypoglycemic drugs. PPARγ has been suggested as a potential therapeutic target in PH.^30,31,33,34 In our current study, we found a prominent reversal of PH in response to the PPARγ agonist Empa.

To our knowledge, this study is the first to recognize Empa as a PPARγ activator and to show the successful therapeutic use of this SGLT2 inhibitor in sensitive and fast-growing IPAH PASMCs and animal models of PH. We found that treatment with Empa reversed hemodynamic and structural changes provoked by chronic hypoxia (10% O₂). Moreover, this study provides cellular and molecular evidence regarding the role of the γ-secretase–Notch3 axis signaling pathway in disease development and the therapeutic response through PPARγ upregulation. In our research, commercial PASMCs were employed for comparison with PASMCs isolated from PAH patients undergoing lung transplantation. These limitations arise from a scarcity of human lung tissue resources and ethical considerations related to human experimentation, making it impossible for us to acquire normal human specimens. Moreover, mice exposed to chronic hypoxia exhibit only vascular remodeling; they lack crucial features of human PAH. Although there are limitations of experimental studies, we believe that our findings prompt consideration of Empa as a PPARγ agonist in therapeutic approaches for patients with progressive PH, in which PPARγ may be upregulated, but future clinical trials of PAH are necessary to confirm this hypothesis. The first related clinical trial (Emphower PoC) is a phase 2 randomized, double-blind, placebo-controlled study evaluating the safety and efficacy of Empa in patients with PAH. The trial is expected to be completed in November 2024. Much remains unknown regarding how Empa prevents the progression of pathological pulmonary vascular remodeling or reduces RV hypertrophy, and future work is needed to confirm the mechanism. As Empa is already broadly used in patients with type 2 diabetes and has proven to be well tolerated in this indication, its clinical development as a new therapy for PH might therefore be imminent. In addition, PPARγ is a nuclear transcription factor that regulates target gene expression by recognizing PPREs in the promoter regions of such genes.²³ The γ-secretase complex includes the subunits PSEN1, PSEN2, nicastrin, APH1α, and PEN2.⁵² Bioinformatics analysis revealed putative PPARγ-RXRα binding elements within the PEN2 and PSEN1 promoters and a PPARγ binding site in nicastrin. PEN2 is the key component for γ-secretase complex stability.^19,52 The results of this study are limited in that we focused on PPARγ regulation of the PEN2 gene, not on the promotors of PSEN1 and nicastrin.

In conclusion, we used isolated IPAH PASMCs to screen for and identify an effective SGLT2 inhibitor, Empa. This inhibitor activates PPARγ, leading to the downregulation of the γ-secretase-specific gene, PEN2, which mediates Notch3 signaling in PH. The efficacy of Empa in enhancing PPARγ signaling and inhibiting γ-secretase may be valuable for treating PAH. It might also be beneficial for other cardiovascular and noncardiovascular diseases like cancer and diabetes (where PPARγ signaling is diminished), and diseases where γ-secretase inactivation is crucial for therapeutic success.

Glossary

Abbreviations

α-SMA

α-smooth muscle-actin

Aβ

amyloid β-peptides

APH1α

anterior pharynx-defective 1

BMPRII

bone morphogenetic protein receptor type II

Cana

canagliflozin

Dapa

dapagliflozin

BrdU

5-bromo-2-deoxyuridine

Empa

empagliflozin

EC

endothelial cell

γ-secretase

gamma-secretase

HF

heart failure

HES5

Hes family BHLH transcription factor 5

HUVEC

human umbilical vein endothelial cell

IPAH

idiopathic pulmonary arterial hypertension

ICD

intracellular domain

LV

left ventricle

MAP

mean arterial pressure

MWT

medial wall thickness

PPARγ

peroxisome proliferator-activated receptor gamma

PPRE

PPAR response element

PSEN1

presenilin 1

PSEN2

presenilin 2

PEN2

presenilin enhancer protein 2

PAH

pulmonary arterial hypertension

PA

pulmonary artery

PAEC

pulmonary artery endothelial cell

PAP

pulmonary artery pressure

PASMC

pulmonary artery smooth muscle cell

PH

pulmonary hypertension

RXR

retinoid X receptor

RV

right ventricle

RVSP

right ventricular systolic pressure

siRNA

small interfering RNA

SMC

smooth muscle cell

SGLT2

sodium-glucose cotransporter 2

SHR

spontaneous hypertensive rats

SAP

systemic arterial pressure

TZD

thiazolidinedione

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.4c00127.

• Effect of empagliflozin treatment on proteins expression in lung tissue of hypoxia-induced PAH mice (PDF)

Author Contributions

H.H.H. for Clinical work, analysis, and interpretation. Y.J.L. and H.H.H. conceptualized and designed the study. Y.J.L., Y.H.Y., Y.L.H., C.D.A., G.J.C., and W.J.C. experimental work, analysis, and interpretation. Y.J.L. had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Y.J.L. Oversaw the experiments, data acquisition and analysis, and manuscript preparation and editing.

This work was supported by grants from Ministry of Science and Technology [MOST 108-2314-B-182-054-MY3 and, 111-2314-B-182-008-MY3], and Chang Gung Medical Research Program [CMRPD1L0151, CMRPD1M0331-2 and BMRPD05].

The authors declare no competing financial interest.