Phosphodiesterase 4D contributes to angiotensin II-induced abdominal aortic aneurysm through smooth muscle cell apoptosis

Abdominal aortic aneurysm (AAA) is a permanent expansion of the abdominal aorta that has a high mortality but limited treatment options. Phosphodiesterase (PDE) 4 family members are cAMP-specific hydrolyzing enzymes and have four isoforms (PDE4A-PDE4D). Several pan-PDE4 inhibitors are used clinically. However, the regulation and function of PDE4 in AAA remain largely unknown. Herein, we showed that PDE4D expression is upregulated in human and angiotensin II-induced mouse AAA tissues using RT-PCR, western blotting, and immunohistochemical staining. Furthermore, smooth muscle cell (SMC)-specific Pde4d knockout mice showed significantly reduced vascular destabilization and AAA development in an experimental AAA model. The PDE4 inhibitor rolipram also suppressed vascular pathogenesis and AAA formation in mice. In addition, PDE4D deficiency inhibited caspase 3 cleavage and SMC apoptosis in vivo and in vitro, as shown by bulk RNA-seq, western blotting, flow cytometry and TUNEL staining. Mechanistic studies revealed that PDE4D promotes apoptosis by suppressing the activation of cAMP-activated protein kinase A (PKA) instead of the exchange protein directly activated by cAMP (Epac). Additionally, the phosphorylation of BCL2-antagonist of cell death (Bad) was reversed by PDE4D siRNA in vitro, which indicates that PDE4D regulates SMC apoptosis via the cAMP-PKA-pBad axis. Overall, these findings indicate that PDE4D upregulation in SMCs plays a causative role in AAA development and suggest that pharmacological inhibition of PDE4 may represent a potential therapeutic strategy.


INTRODUCTION
Abdominal aortic aneurysm (AAA) is a chronic vessel wall degenerative disease characterized by the irreversible progressive dilatation of the abdominal aorta over 3.0 cm (1.5-fold of normal abdominal aorta) 1 . AAA is a multifactorial disease that predominantly affects elderly males 65 years or older 2 . AAA rupture is a lifethreatening medical emergency, with mortality rates > 81% 3 . At present, there are no effective clinical medicines to prevent, delay, or reverse the growth or rupture of AAA, except surgical repair 4 . Therefore, understanding the molecular mechanisms and identifying regulators underlying the pathogenesis of AAA are essential for developing potential therapeutic strategies. Smooth muscle cells (SMCs) are the major cell type in the medial layer of large arteries. SMCs regulate vascular contractility in response to pulsatile blood flow and pressure in aortas. SMCs are the major contributors to elastin fibers that maintain the elasticity of aortas. Increasing evidence has indicated that the loss of vascular SMCs contributes to aortic wall weakening, dilation, and aneurysm formation 5,6 .
The second messenger cyclic nucleotides 3′,5′-cyclic adenosine monophosphate (3′,5′-cAMP) and 3′,5′-cyclic guanosine monophosphate (3′,5′-cGMP) are important in regulating smooth muscle contractile function and vessel wall structure integrity. Abnormal cyclic nucleotide homeostasis contributes to a variety of cardiovascular diseases [7][8][9] . Cyclic nucleotide phosphodiesterases (PDEs) are a superfamily of enzymes responsible for hydrolyzing cyclic nucleotides and thus play crucial roles in regulating the duration, magnitude, and compartmentalization of cyclic nucleotide responses 10 . There are 11 families within the PDE superfamily (PDE1 to PDE11), and each family contains multiple subtypes. PDE dysregulation has been associated with numerous diseases 11 . Over the past decades, PDEs have been proven to be ideal and feasible drug targets. Several family-specific PDE inhibitors are currently used or are currently under clinical trials for the treatment of a variety of diseases [12][13][14] .
A few reports have suggested potential links between cAMP signaling and AAA. For example, ablation of SMC-specific Gsα, a G-protein responsible for cAMP production, exaggerated AAA formation in mice 15 . Cilostazol, an inhibitor of cAMP-hydrolyzing PDE3 and adenosine uptake 16 , suppressed AAA in rats and mice 17 . The mechanistic action of cilostazol in AAA remains to be clarified: cAMP versus adenosine. Recently, we reported that PDE1C, a Ca 2+ /calmodulin-stimulated PDE, contributed to AAA by promoting SMC senescence by suppressing cAMP-mediated activation of SIRT1 18 . These previous studies showed a protective effect of cAMP on AAA prevention. However, the roles of different cAMP signaling pathways regulated by distinct cAMP-PDEs in AAA remain unclear. PDE4 family members are cAMP-specific hydrolyzing enzymes encoded by four isoforms: PDE4A, 4B, 4C, and 4D. Several PDE4 family-selective inhibitors-such as rolipram, roflumilast, apremilast, and crisaborole-have been used to clinically treat inflammatory diseases, including inflammatory airway diseases, psoriatic arthritis, and atopic dermatitis 19 . Understanding the roles of PDE4 isozymes and the effects of PDE4 inhibitors in AAA is critical for developing potential therapeutic strategies.
In this study, we aimed to explore the regulation, function, and mechanistic action of the PDE4 isozyme in AAA pathogenesis and development. Because our pilot study showed that PDE4D is most prominently upregulated in SMCs of human and mouse AAA tissues, we focused on SMC PDE4D in the current study. Furthermore, PDE4D deficiency improved vascular pathogenesis and AAA development in SMC-specific Pde4d knockout mice. The PDE4 inhibitor rolipram protected against AAA in mice. In addition, we demonstrated that PDE4D promotes SMC apoptosis via the cAMP-PKA-pBad axis. Step Mouse Genotyping Kit (Vazyme Biotech Co., Ltd., PD101-01, Nanjing, China) was used to test tail biopsies of mice by PCR. Primers targeting the wild-type allele, the floxed allele, and cre recombinase are listed in Supplementary Table 2. After genotyping analysis, Apoe −/− Pde4d flox/flox and Apoe −/− Pde4d SMC−/− littermates were used for study. Only male mice were used, as AAA primarily affects elderly males in humans 2 .
AAA was defined as 50% dilation in the diameter of the external abdominal aorta compared with the normal abdominal aorta of the control group. The aortas were excised, and the maximal external diameter of the abdominal aorta was measured by two different investigators using a stereoscope at the time of necropsy. The animals were raised in a specific pathogen-free (SPF) facility.

Human aortic tissues
Human specimens were obtained under protocols approved by the Ethics Committee of Beijing Anzhen Hospital, Capital Medical University. Human samples from the anterior region of the aneurysmal aortic wall were obtained from nine AAA patients who had undergone open aortic aneurysm surgery repair (Project Number: 2017-051X). The patients were diagnosed with AAA by repeated ultrasonography or CT angiography. Specimens from patients with collagen vascular disease or severe chronic kidney disease (estimated glomerular filtration rate < 30 mL [min·1.73 m 2 ] −1 ) were excluded. Samples were snap-frozen in liquid nitrogen directly after surgery. Corresponding adjacent human aortic non-AAA tissue specimens used for controls were collected from the bodies of six deceased donors with no detectable vascular disease (Volunteer Corpse Donation Reception Station). Samples in the freezers were snap-frozen in liquid nitrogen after dissection. Informed consent was obtained from all study participants. The clinical information associated with the samples is shown in Supplementary Table 1.

RNA sequencing analysis
The filtered and trimmed reads of each sample were aligned to the rat reference genome (Rnor_6.0) using HISAT2 software with default parameters 24,25 . The read counts of genes were calculated using the R packages Genomic Features and Genomic Alignments 26 . The FPKM (fragments per kilobase million) of genes was determined using the DESeq2 package 27 . Differentially expressed genes were identified using a paired t test with p value < 0.1. Pathway enrichment analysis was performed using the MetaCore online tool (https://portal.genego.com/). An FDR cutoff of < 0.05 was set to select for significant pathways. The relationship between pathways and genes was visualized using the R package GOplot (v1.0.2) 28 .

Histological, immunohistochemical, and immunofluorescent analysis
Human aortic tissue samples were vertically embedded in optimal cutting temperature compound (O.C.T, Thermo Fisher, US) and stored at −80°C. Mouse AAA aortas were dissected at the maximal suprarenal outer aortic diameter and embedded vertically in O.C.T compound for histological evaluation. Corresponding adjacent normal aortic vascular tissues were collected from non-Ang II-infused control mice. At least ten-fifteen serial frozen sections, six μm thick, were sectioned from the AAA portion, covering the maximal dilated aorta and control tissues using a freezing microtome (Leica CM1860, Germany). Immunohistochemical staining was performed as previously described 29,30 . Tissue sections were incubated with rabbit anti-PDE4D (1:100, Abcam, ab14613, Britain) antibody overnight at 4°C, followed by incubation with anti-rabbit IgG-peroxidase conjugate (Beijing Zsbio Biotechnology, China) for thirty mins at room temperature and developed by adding the substrate 3-amino-9-ethylcarbazole (AEC). Images were taken with a microscope (Nikon, digital sight DS-Fi2, Germany). For quantitative analysis, the positive staining intensity was semiquantified with Image-Pro Plus Software (Media Cybernetics, Bethesda, MD). For each animal, a total of five fields were randomly selected, quantified, and averaged. For negative controls, sections were incubated with goat anti-rabbit secondary antibody using the primary antibody omitted ( Supplementary Fig. 1a-c). In addition, mouse aortic sections were stained for elastin staining (Modified Verhoeff Van Gieson Elastic Stain Kit, Sigma, Cat#: HT25A, US). Elastin staining is indicated by the darkest color, and elastin degradation was rated according to the key provided by the manufacturer 31 . Mouse heart tissues were stained with hematoxylin and eosin (H&E) (Solarbio, Cat#: G1120, China).

Measurement of blood pressure by tail-cuff plethysmography
Systolic and diastolic blood pressure were measured via the noninvasive tailcuff method (CODA, Kent Scientific, US) before osmotic pump implantation and later sacrifice as described 32 . Briefly, mice were warmed to~37°C in restraint tubes on the heating pad with their tails restrained in the occlusion tail cuff. Mice were trained for~3-15 min until a stable blood pressure was recorded. The average of at least five successful measurements was considered the blood pressure of the mice on that recording day.

Flow cytometry
SMC apoptosis was investigated via flow cytometry. Briefly, the digested SMC suspension was passed through a 40 μm Cell Strainer (Biologix Group, Ltd., Jiangsu, China) and stained according to the protocol of the Annexin V-FITC/PI Apoptosis Detection Kit (Dojindo, Japan, AD10). All groups of cells in the mixture were used as the gating control samples stained with only Annexin V-FITC or PI or without any dye. The data were acquired and analyzed by BD Accuri C6 flow cytometry (BD Biosciences, US). The gating strategy was to divide positive and negative events in negative, Annexin V-FITC, and PI gating control samples (Supplementary Fig. 12a-d).

Measurement of cAMP
A cAMP Direct Fluorometric Immunoassays Kit (Abcam, ab138880) was used to measure the cAMP level according to the manufacturer's instructions. For in vitro cAMP quantification, cells plated in sterile 96well plates (1 × 10 5 cells/well) were lysed with 100 μL/well of Cell Lysis Buffer. For in vivo cAMP quantification, aortic tissues were lysed in 20 μL/ mg of Cell Lysis Buffer. Briefly, all samples and standards (75 μL each) mixed with 25 μL of 1x HRP-cAMP were incubated in plates at room temperature for 2 h on a plate shaker. After four washes, the plate was incubated with 100 μL/well of AbRed Working Solution for 45 min. The fluorescence was measured at Ex/Em = 540/590 nm using a Biotek Synergy™ H1 microplate reader. The standard curve was determined by regression analysis using a logistic curve-fit and prepared for every experiment independently.
Cell proliferation assay SMC proliferation was assayed using BeyoClick™ EdU-488 Cell Proliferation Assay Kit (Beyotime, Beijing, China) following the manufacturer's instructions. Images were taken by a fluorescence microscope (Nikon Eclipse Ti2, Japan), and the signals were counted in three random visual fields for each sample.

Statistical analysis
Data were statistically analyzed using GraphPad Prism 9 (GraphPad Software, LLC, San Diego, CA). Data are expressed as the mean ± SEM. The Shapiro-Wilk test was used to test for normal distribution. The Brown-Forsythe test or F test was used to test for equality of variances. Normally distributed datasets with equal variance were analyzed with the parametric unpaired Student's t test for 2 independent groups and the parametric one-way ANOVA or two-way ANOVA followed by the Holm-Sidak's post hoc test for ≥3 groups. Normally distributed datasets without equal variance were analyzed with a parametric Welch's t test for 2 independent groups and the parametric Welch ANOVA with Dunnett's T3 post hoc test for ≥3 groups. Where a normal distribution could not be confirmed, the nonparametric Mann-Whitney test was used for 2 independent groups. All tests were two sided. A p value < 0.05 was considered statistically significant. Detailed statistical analyses and plotting methods are listed in Supplementary Table 3.

RESULTS
PDE4D expression is upregulated in human and mouse AAA tissues To explore PDE4 expression during AAA formation, we assessed the mRNA levels of individual PDE4 family isoforms (PDE4A-D) in both human and mouse AAA tissues. Human AAA tissues were collected from AAA patients via surgery, and control non-AAA tissues were corresponding tissues collected from the bodies of deceased donors with no detectable vascular disease (Supplementary Table  1). Mouse AAA was induced by a chronic infusion of Ang II (1000 ng kg −1 min −1 ) for twenty-eight days via osmotic minipumps in Apoe −/− male mice fed a high-fat diet (HFD): a well-established murine AAA model 20 . Control mice received vehicle infusion. Mouse AAA tissues and corresponding normal aortic tissues were dissected from the suprarenal aortic regions of the mice with AAA and the control mice, respectively. Among the four PDE4 isozymes, only PDE4D expression levels were significantly increased in human AAA compared to non-AAA tissues (Fig. 1a). Similar observations were obtained for mouse AAA tissues (Fig. 1b). Henceforth, we focused on PDE4D in this study. Consistently, we found that the PDE4D protein levels were significantly higher in human (Fig. 1c, d) and mouse AAA tissues (Fig. 1e, f). Immunohistochemical staining further confirmed the PDE4D protein increase in human (Fig. 1g, h) and mouse AAA lesions (Fig. 1i, j). PDE4D antibody specificity was supported by negative controls performed in mouse and human , and PDE4D mRNA levels (fold change versus one of the non-AAA subjects) in human tissues. ***p < 0.001, Welch's t test for PDE4A, Mann-Whitney test for PDE4B/PDE4C, unpaired Student's t test for PDE4D, mean ± SEM, non-AAA (n = 6), AAA (n = 9). Human AAA tissues were collected from AAA patients during surgery, and non-AAA tissues were collected from corresponding tissues of deceased donors with no detectable vascular disease. b RT-PCR quantification of Pde4a, Pde4b, Pde4c, and Pde4d mRNA (fold change versus one of the control subjects) in mouse AAA tissues from the Apoe −/− mice treated with 1000 ng kg −1 min −1 angiotensin II (Ang II) and a high-fat diet (HFD) and control mouse abdominal aortas from the Apoe −/− mice treated with saline. *p < 0.05, unpaired Student's t test, mean ± SEM, controls (n = 4), AAA (n = 8). Mouse AAA aortas were dissected at the maximal suprarenal outer aortic diameter, and controls were collected from the corresponding suprarenal abdominal aortas of control mice. (c, d). Representative western blots and quantification of PDE4D protein levels in human AAA and non-AAA tissues normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) protein (fold change versus non-AAA subjects). **p < 0.01, Mann-Whitney test, mean ± SEM, non-AAA (n = 6), AAA (n = 9). (e, f) Representative western blots and quantification of PDE4D protein levels in mouse AAA and control tissues normalized to GAPDH protein (fold change versus control subjects). ***p < 0.001, unpaired Student's t test, mean ± SEM, controls (n = 4), AAA (n = 8). (g, h) Representative images of immunohistochemistry staining of PDE4D (g) and quantification of PDE4D staining intensity per medial area (h) in human tissues. A total of five random images per human in each group were selected for statistical analysis of the ratio of positive areas to vascular areas. L lumen. **p < 0.01, unpaired Student's t test, mean ± SEM, non-AAA (n = 6), AAA (n = 9). (i, j) Representative images of immunohistochemistry analysis of PDE4D (i) and quantification of PDE4D staining intensity per medial area (j) in mouse tissues. A total of five random images per mouse in each group were selected for statistical analysis of the ratio of positive areas to vascular areas. L lumen. *p < 0.05, Welch's t test, mean ± SEM, controls (n = 4), AAA (n = 8). Fig. 1a-g). These results demonstrated an upregulation of PDE4D expression in AAA.

tissues (Supplementary
PDE4D expression was largely observed in the medial areas where SMCs resided (Fig. 1g-j), suggesting the expression PDE4D in SMCs. To further confirm the expression of PDE4D in SMCs, we performed double immunofluorescence staining of PDE4D and α-smooth muscle actin (α-SMA), a marker for differentiated contractile and dedifferentiated synthetic SMCs (or myofibroblasts). We observed prominent PDE4D staining in α-SMA-positive cells in human ( Fig. 2a and Supplementary Fig. 2a) and mouse AAA lesions ( Fig. 2b and Supplementary Fig. 2b). Consistently, in cultured rat aorta SMCs, Ang II also increased PDE4D mRNA and protein levels (Fig. 2c-e). Therefore, in this study, we focused on the role of SMC PDE4D in SMC pathogenesis and AAA development.  Supplementary Fig. 3c), and immunostaining ( Supplementary Fig. 1d-g) confirmed PDE4D depletion in mouse aortic SMCs. We also observed no significant alterations in other PDE4 isozymes between Apoe −/− Pde4d flox/flox and Apoe −/− Pde4d SMC−/− aortic tissues ( Supplementary Fig. 3b).
AAA was induced by Ang II infusion (1000 ng kg −1 min −1 ) for twenty-eight days along with an HFD, while the controls were infused with saline (Fig. 3a). The aortas were excised, and the maximal external diameter of the abdominal aorta was measured by two different investigators via stereoscopy at the time of necropsy. AAA was defined as a 50% dilation in the diameter of the external abdominal aorta compared with the normal mouse abdominal aorta. Under Ang II infusion, AAA development was significantly attenuated in the Apoe −/− Pde4d SMC−/− mice compared to the Apoe −/− Pde4d flox/flox mice, as illustrated by the aortic morphology (Fig. 3b). The average external diameter of the abdominal aortas was significantly smaller in the Apoe −/− Pde4d SMC−/− Ang II group than in the Apoe −/− Pde4d flox/flox Ang II group (1.936 ± 0.203 mm vs. 3.073 ± 0.327 mm; Fig. 3c). Supplementary  Fig. 4a includes images of all AAA samples shown in Fig. 3b, c. In addition, the fragmentation and degradation of the aortic elastic lamina-a characteristic feature of AAA-was evaluated. Verhoeff Van Gieson staining for elastin and semiquantitative analysis revealed more profound elastin degradation in the Apoe −/− Pde4d flox/flox Ang II mice than in the Apoe −/− Pde4d SMC−/− Ang II mice (Fig. 3d, e). Mmp2 and Mmp9 mRNA levels were both upregulated in the Apoe −/− Pde4d flox/flox Ang II group, but only Mmp2 expression was rescued in Apoe −/− Pde4d SMC−/− mouse aortas ( Supplementary Fig. 5a, b). In addition, aortic cAMP levels were decreased in the Ang II-induced AAA mice compared with the saline-treated mice, and SMC-specific knockout of Pde4d reversed the cAMP levels in AAA tissues (Supplementary Fig. 6a). These results suggest that PDE4D deficiency in SMCs enhances the stability of the aortic wall and decreases Ang II-induced AAA formation.
Additionally, Ang II infusion increased systolic and diastolic blood pressure (BP) in the Apoe −/− Pde4d flox/flox mice, both of which were significantly lower in the Apoe −/− Pde4d SMC−/− Ang II mice ( Supplementary Fig. 4b, c). Based on a previous report by Daugherty et al., AAA occurred in the Apoe −/− mice infused with Ang II (1000 ng kg −1 min −1 ) for twenty-eight days, accompanied by increased BP. Hydralazine administration (an antihypertensive drug) lowered systolic BP in the Ang II-infused Apoe −/− mice, while hydralazine did not prevent AAA formation 33 . Therefore, it is   4 Effect of rolipram on Ang II-induced AAA in mice. AAAs were induced by Ang II infusion (1000 ng kg −1 min −1 ) and an HFD for 28 days, and controls were infused with saline. Rolipram (3 mg kg −1 d −1 ) was orally administered daily for 28 days. Mouse AAA aortas were dissected at the maximal suprarenal outer aortic diameter, and controls were collected from corresponding suprarenal abdominal aortas of the control mice. a Schematic diagram of the mouse model of Ang II-induced AAA treated with vehicle or rolipram. Vehicle (n = 5) and rolipram (n = 6) with saline infusion, vehicle (n = 14) and rolipram (n = 12) with Ang II infusion. b Representative images of entire aortas of the Apoe −/− mice treated with vehicle or rolipram. c Quantification of the maximal external diameter (mm) of the abdominal aorta measured by two different investigators using a stereoscope. *p < 0.05, ***p < 0.001, Welch's t test between the rolipram (−) control and rolipram (−) Ang II groups, Mann-Whitney test between the rolipram (−) Ang II and rolipram (+) Ang II groups, mean ± SEM, vehicle (n = 5) and rolipram (n = 6) with saline infusion, vehicle (n = 14) and rolipram (n = 12) with Ang II infusion. d Representative images of elastin staining of the mouse arterial wall. L: lumen. e Quantification of the elastin degradation score from the four indicated groups in (d). Elastin staining is indicated by the darkest color. *p < 0.05, **p < 0.01, Mann-Whitney test between the rolipram (−) control and rolipram (−) Ang II groups, unpaired Student's t test between the rolipram (−) Ang II and rolipram (+) Ang II groups, mean ± SEM, vehicle (n = 5) and rolipram (n = 5) with saline infusion, vehicle (n = 5) and rolipram (n = 5) with Ang II infusion.
generally believed that Ang II infusion-induced AAA formation is independent of BP elevation. Thus, the protective effect of Pde4d SMC−/− against AAA development is unlikely to result from a BP reduction.
The PDE4 inhibitor rolipram attenuates Ang II-induced AAA formation Next, we sought to determine the pharmacological effect of a PDE4 inhibitor on AAA. We selected the pan-PDE4 inhibitor rolipram, which was used clinically for neuroinflammation in the early 1990s 19 . Male C57BL/6J Apoe −/− mice at the age of eight weeks were infused with Ang II or saline and fed an HFD for four weeks. Rolipram (3 mg kg −1 d −1 ) or vehicle (7% ethyl alcohol) was given daily via gavage for four weeks (Fig. 4a). Rolipram (+) treatment significantly reduced the AAA size compared with rolipram (−) vehicle treatment (Fig. 4b). The average external diameter of the abdominal aorta was smaller in the rolipram (+) Ang II group (1.425 ± 0.192 mm) than in the rolipram (−) Ang II    Fig. 4c). Supplementary Fig. 7a includes images of all AAA samples shown in Fig. 4b, c. Moreover, elastin fragmentation in the aortic wall was significantly reduced in the rolipram (+) Ang II mice compared with the rolipram (−) Ang II mice (Fig. 4d, e). Consistently, rolipram also attenuated Mmp2, but not Mmp9, in aortic tissues induced by Ang II (Supplementary Fig.  5c, d). As expected, cAMP levels in the AAA mice were lower than those in the saline-treated mice, and rolipram reversed cAMP levels in AAA tissues ( Supplementary Fig. 6b). These results demonstrate a significant pharmacological impact of rolipram in preventing AAA development in mice. Similar to the results in Pde4d SMC−/− mice, rolipram also reduced systolic and diastolic BP in AAA mice ( Supplementary Fig. 7b, c).

PDE4D promotes SMC apoptosis in vitro and in vivo
To determine the underlying mechanism by which PDE4D participates in AAA formation, we performed bulk RNA-seq to identify PDE4D-regulated genes in rat aortic SMCs treated with PDE4D-specific or scramble control siRNA. RNA-seq revealed 1,848 genes upregulated in SMCs stimulated with Ang II (100 nM, 24 h; Supplementary Fig. 8a, c, d) and 1,551 genes downregulated in SMCs by PDE4D siRNA (Supplementary Fig. 8b, c) (for the full list, refer to Supplementary Dataset 1). Among these genes, 235 were upregulated by Ang II, and the upregulation was reversed by si-PDE4D ( Supplementary Fig. 8c). We next performed MetaCore pathway enrichment analysis of these 235 genes (for the full list, refer to Supplementary Dataset 1). We focused on the top 15 pathways shared by the two groups (control vs. Ang II and siscramble vs. si-PDE4D) based on their minimum false discovery rate (FDR). Among these pathways, apoptosis-related pathways were predominant, which is consistent with the important role of SMC apoptosis in AAA pathogenesis (Fig. 5a).
We next determined the role of PDE4D in regulating SMC apoptosis using PDE4D siRNA to knockdown PDE4D expression or rolipram to inhibit PDE4 activity in SMCs (Supplementary Fig. 9a-c). We found that Ang II increased caspase-3 cleavage, a key event in apoptosis 34 . PDE4D siRNA substantially attenuated Ang II-induced caspase-3 cleavage (Fig. 5b, c). Similarly, PDE4D siRNA also reduced caspase-3 cleavage induced by H 2 O 2 as a reactive oxidative stress (ROS) mediator known to be important in AAA (Supplementary Fig.  10a, b). We also assessed SMC apoptosis via Annexin V/propidium iodide (PI) staining and flow cytometry (Fig. 5d and Supplementary  Fig. 10c). We found that SMC apoptosis induced by Ang II or H 2 O 2 was significantly suppressed by PDE4D siRNA (Fig. 5e and Supplementary Fig. 10d). As with PDE4D siRNA, rolipram also reduced cleaved caspase-3 levels in SMCs treated with Ang II (Fig. 5f, g) or H 2 O 2 ( Supplementary Fig. 11a, b) and decreased SMC apoptosis induced by Ang II (Fig. 5h, i) or H 2 O 2 ( Supplementary Fig.  11c, d). The gating strategy of flow cytometry was to divide positive and negative events in negative, Annexin V-FITC, and PI gating control samples ( Supplementary Fig. 12a-d). These results support a proapoptotic function of PDE4D in SMCs in vitro.
PDE4D antagonizes PKA-mediated phosphorylation of Bad and induces cleaved caspase-3 and apoptosis in SMCs PDE4 family members are cAMP-hydrolyzing enzymes. We examined the levels of cAMP in vitro using a cAMP kit. As expected, Ang II reduced SMC cAMP and PDE4D siRNA, and rolipram upregulated cAMP levels in vitro ( Supplementary Fig. 6c, d). cAMP activates cAMP-dependent protein kinase A (PKA) or the exchange protein directly activated by cAMP (Epac) 35 . To identify whether PDE4D regulates apoptosis by relying on the PKA and/or Epac pathways, we used PKA-or Epac-selective inhibitors, including PKI (PKA inhibitor) and ESI-09 (Epac inhibitor). We found that PDE4D siRNA attenuated the apoptosis marker cleaved caspase-3 in SMCs induced by Ang II; however, the inhibitory effect of PDE4D siRNA on cleaved caspase-3 was blocked by PKI (Fig. 7a, b) but not ESI-09 (Fig. 7c, d). These data suggest that PDE4D regulates SMC apoptosis in a cAMP-PKA signaling-dependent manner.
Based on our transcriptome sequencing results, the BCL2 antagonist of cell death (Bad) was highlighted in several apoptosis-related pathways ( Supplementary Fig. 16). Bad is an important protein involved in apoptosis. Bad can be phosphorylated by PKA at Ser155 36 . This Bad phosphorylation (pBad) causes Bad inactivation and thus inhibits apoptosis 37 . We therefore investigated the role of PDE4D in Bad phosphorylation. We found that Ang II reduced Bad phosphorylation, which was reversed by PDE4D siRNA (Fig. 7e-h). Interestingly, the effect of PDE4D siRNA on increasing Bad phosphorylation was largely blocked in the presence of PKI (Fig. 7e-h), suggesting that PDE4D negatively regulates the PKA phosphorylation of Bad.

DISCUSSION
In this study, we reported the role and regulatory mechanism of PDE4D in SMC apoptosis and AAA formation (Fig. 7i). We showed that PDE4D played an essential role in SMC apoptosis, at least partially by attenuating PKA-mediated phosphorylation and inactivation of the proapoptotic molecule Bad. We demonstrated PDE4D upregulation in SMCs from human and mouse AAA lesions through multiple approaches, including RT-PCR, western blotting, and immunostaining. To perform in vivo evaluation of the role of SMC PDE4D in AAA formation, we generated SMC-specific Pde4d knockout mice on an Apoe −/− genetic background and tested c Quantification of cleaved caspase-3 protein expression by immunoblotting in (b) normalized to caspase-3 protein (fold change versus control). **p < 0.01, ***p < 0.001, two-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. d Flow cytometric analysis of annexin V/propidium iodide (PI)-stained SMCs treated with or without Ang II (100 nM, 24 h)/PDE4D siRNA (200 nM, 48 h) as indicated. e Quantification of the total (early and late) apoptosis rates of annexin V/PIstained SMCs treated with or without Ang II (100 nM, 24 h)/PDE4D siRNA (200 nM, 48 h) as indicated. ****p < 0.0001, two-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. f Immunoblot analysis of caspase-3 and cleaved caspase-3 expression in the SMCs treated with or without Ang II (100 nM, 24 h)/rolipram (500 nM, 24.5 h). g Quantification of cleaved caspase-3 protein expression by immunoblotting in (f) normalized to caspase-3 protein (fold change versus control). **p < 0.01, two-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. h Flow cytometric analysis of annexin V/PI-stained SMCs treated with or without Ang II (100 nM, 24 h)/rolipram (500 nM, 24.5 h) as indicated. i Quantification of the total (early and late) apoptosis rates of annexin V/PI-stained SMCs treated with or without Ang II (100 nM, 24 h)/rolipram (500 nM, 24.5 h) as indicated. ****p < 0.0001, two-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments.
them in a well-established AAA mouse model. We demonstrated a causative role of SMC PDE4D in aortic wall degeneration and AAA development. Moreover, we identified a protective effect of the PDE4 inhibitor rolipram against vascular degeneration and AAA development. Rolipram is a pan-PDE4 inhibitor that targets all four PDE4 members 38 . While possessing a history of clinical use, it has been shown to induce nausea and emesis in humans 39 , likely due to its blood-brain barrier penetration and inhibition of PDE4D in the brain chemoreceptor trigger zone of vomiting 19,40,41 . Thus, developing potential PDE4 inhibitors that are peripherally restricted will be beneficial for treating peripheral diseases.
In addition to PDE4 in SMCs, a number of different cell types in AAA tissues may also express PDE4, including endothelial cells, adventitial fibroblasts, and macrophages. However, these different cell types may primarily express distinct PDE4 isozyme(s), among 4A, 4B, 4C, and 4D. It has been shown previously that PDE4B is highly expressed in inflammatory cells and contributes significantly to inflammation 42 . The role of PDE4B in inflammation leads to the development of the PDE4 inhibitor roflumilast for the treatment of COPD. Previous studies have also shown that PDE4D is the dominant PDE4 isozyme expressed in rodent or human arterial VSMCs 43,44 . Because we found that PDE4D is most significantly upregulated in AAA tissues, particularly in SMCs, we thus focused on the role of PDE4D in SMC and AAA development in the current study using SMC-specific PDE4D deficiency animal models. Although our data demonstrated a critical role of SMCderived PDE4D in AAA, we cannot exclude the possible contributions of PDE4D in other cell types to AAA. We also cannot exclude the possible contribution of PDE4B in immune cells to AAA, particularly in the rolipram treatment model, given an important role of inflammation in Ang II-induced AAA. We indeed found that the pan-PDE4 inhibitor rolipram elicited more profound effects on suppressing AAA than SMC-specific PDE4D deficiency (aortic diameters, 1.936 ± 0.203 mm vs. 1.425 ± 0.192 mm). These results suggest that inhibiting other PDE4 isozymes and/or PDE4D in other cell types may contribute to the protective effects of rolipram. It will be of great future interest to elucidate the contributions of distinct PDE4 isozymes in differing cell types to AAA development, given that AAA is a multifactorial disease involving numerous cell types.
In our current study, we focused on the contribution of SMCs in AAA using SMC (Tagln-Cre)-specific Pde4d KO mice. Although Tagln (also called smooth muscle 22α) is expressed on SMCs in cardiac, smooth, and skeletal muscle, Tagln mediates Cre expression at a high level in vascular SMCs. To exclude the effect of Tagln in cardiac myocytes, we indicated that the structure of heart tissues did not change between the Apoe −/− Pde4d flox/flox and Apoe −/− Pde4d SMC−/− mice ( Supplementary Fig. 3d). Furthermore, comparing the Pde4d SMC−/− mice and the mice treated with the c Quantification of cell apoptosis rates with TUNEL staining in Supplementary Fig. 13. A total of five random images per mouse were selected for TUNEL staining statistical analysis. Apoptotic cells examined by TUNEL staining were quantified as the ratio of apoptotic cells to total cells in five fields per mouse. *p < 0.05, **p < 0.01, Welch ANOVA with Dunnett's T3 post hoc test, mean ± SEM, Apoe −/− Pde4d flox/flox (n = 5) and Apoe −/− Pde4d SMC−/− (n = 5) with saline infusion, Apoe −/− Pde4d flox/flox (n = 5) and Apoe −/− Pde4d SMC−/− (n = 5) with Ang II infusion. (d) Representative immunoblot analysis of caspase-3 and cleaved caspase-3 expression in the Apoe −/− mice with or without Ang II/rolipram treatment. e Quantification of cleaved caspase-3 protein expression by immunoblotting in (d) normalized to caspase-3 protein (fold change versus control). *p < 0.05, **p < 0.01, two-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, vehicle (n = 5) and rolipram (n = 5) with saline infusion, vehicle (n = 5) and rolipram (n = 5) with Ang II infusion. f Quantification of cell apoptosis rates with TUNEL staining in Supplementary Fig. 14. A total of 5 random images per mouse were selected for TUNEL staining statistical analysis. Apoptotic cells examined by TUNEL staining were quantified as the ratio of apoptotic cells to total cells in five fields per mouse. **p < 0.01, Mann-Whitney test, mean ± SEM, vehicle (n = 5) and rolipram (n = 5) with saline infusion, vehicle (n = 5) and rolipram (n = 5) with Ang II infusion.
pan-PDE4 inhibitor rolipram, we found that rolipram elicited more profound effects on suppressing AAA than that in the Pde4d SMC−/− mice (aortic diameters, 1.425 ± 0.192 mm vs. 1.936 ± 0.203 mm). These results suggest that inhibiting other PDE4 isozymes and/or PDE4D in other cell types may contribute to the protective effects of rolipram. It will be of great future interest to elucidate the contributions of distinct PDE4 isozymes in differing cell types to AAA development, given that AAA is a multifactorial disease involving numerous cell types.
The roles of cAMP signaling in vascular SMC apoptosis are controversial. Some studies have shown the antiapoptotic effect of cAMP in vascular SMCs. For example, it has been shown that rat aortic SMC apoptosis is attenuated by the activation of cAMP/PKA signaling through the beta-adrenergic receptor (β-AR) agonist adenosine or the PDE4 inhibitor Ro-201724 45 . Some other studies reported a proapoptotic effect of cAMP in vascular SMCs. For example, human vascular SMC apoptosis was promoted by the nonselective elevation of intracellular cAMP levels through 8bromo-cAMP and forskolin or increasing PDE3-regulated cAMP through cilostazol 46 . In the current study, we provided solid evidence and demonstrated that rat aortic SMC apoptosis is attenuated by stimulating cAMP/PKA signaling through the PDE4 Fig. 7 PDE4D promotes apoptosis through Bad phosphorylation regulated by PKA. a Representative immunoblot analysis of caspase-3 and cleaved caspase-3 expression in the SMCs treated with PKI (PKA inhibitor, 10 μM, 60 min) and/or Ang II (100 nM, 24 h) and/or PDE4D siRNA (200 nM, 48 h) as indicated. b Quantification of cleaved caspase-3 protein expression by immunoblotting in (a) normalized to caspase-3 protein (fold change versus control). **p < 0.01, ***p < 0.001, ****p < 0.0001, one-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. c Immunoblot analysis of caspase-3 and cleaved caspase-3 expression in the SMCs treated with ESI-09 (Epac inhibitor, 100 μM, 60 min) and/or Ang II (100 nM, 24 h) and/or PDE4D siRNA (200 nM, 48 h) as indicated. d Quantification of cleaved caspase-3 protein expression by immunoblotting in (c) normalized to caspase-3 protein (fold change versus control). ****p < 0.0001, ns: no significant difference, one-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. e Representative immunoblot analysis of phospho-Bad (pBad) and Bad expression in the SMCs treated with PKI (PKA inhibitor, 10 μM, 60 min) and/or Ang II (100 nM, 24 h) and/or PDE4D siRNA (200 nM, 48 h) as indicated. f Quantification of pBad expression by immunoblotting in (e) normalized to Bad protein (fold change versus control). **p < 0.01, ****p < 0.0001, one-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. g Quantification of pBad expression by immunoblotting in (e) normalized to GAPDH protein (fold change versus control). **p < 0.01, ***p < 0.001, one-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. h Quantification of Bad expression by immunoblotting in (e) normalized to GAPDH protein (fold change versus control). **p < 0.01, ***p < 0.001, ****p < 0.0001, one-way ANOVA with Holm-Sidak's post hoc test, mean ± SEM, n = 3 separate experiments. i PDE4D expression was significantly increased in Ang II-induced AAA, aggravating SMC apoptosis, which promoted pBad via the cAMP-PKA axis.
inhibitor rolipram or specific depletion of PDE4D isozymes. These lines of experimental evidence from ours and others suggest that SMC viability may be differentially, even oppositely, regulated by different cAMP/PKA signalosomes that are coupled to distinct G-protein coupled receptors (GPCRs) and/or PDEs. Although we demonstrated the roles of PDE4D in SMC apoptosis in this study, there are eleven different PDE4D variants reported to date, and the expression of different PDE4D variants in SMCs has been reported previously. The contractile/quiescent vascular SMCs primarily express PDE4D3, while the synthetic/activated SMCs express PDE4D1 and PDE4D2 47 . PDE4D8 is enriched in the pseudopodia of SMCs and regulates SMC migration 48 . These findings suggest that different PDE4D variants may be involved in distinct SMC functions. The specific PDE4D variant(s) in SMC apoptosis remain to be elucidated in the future.
We also found that Ang II-induced elevation of systolic and diastolic BP, measured by tail cuff, was significantly reduced in the SMC-specific Pde4d knockout mice and in the mice treated with rolipram. This raised the question of whether BP reduction contributes to the effect of PDE4D deficiency/inhibition in AAA. Daugherty et al. found that both NE and Ang II raised arterial pressure. However, Ang II but not NE induced AAA. Moreover, blocking this BP elevation using the antihypertensive drug hydroxyzine did not prevent Ang II-induced AAA occurrence and development 33 . Although previous experimental evidence indicates that Ang II induces AAA formation in mice independent of its effect on BP elevation 33 , BP may affect the outcomes of AAA progression and rupture 49 . Thus, we do not exclude the potential impact of BP reduction by PDE4D deficiency or inhibition in AAA progression or rupture.
In addition to apoptosis, we also determined whether PDE4D was associated with other phenotypes of SMCs, including proliferation, dedifferentiation, deposition of extracellular matrix (ECM), and inflammation, in AAA development. We found that Ang II significantly increased the proliferation of SMCs; however, PDE4D siRNA could not suppress the proliferation induced by Ang II (Supplementary Fig. 17a, b). SMCs can dedifferentiate into macrophage-like SMCs or osteoblast-like SMCs by changing from the contractile type to the synthetic type 50 . Our results showed that Ang II-treated SMCs expressed higher levels of the SMC contractile marker proteins smooth muscle 22α (SM22α) and SM-calponin (CNN) as well as lower levels of the synthetic marker osteopontin (OPN), but PDE4D siRNA did not reverse the dedifferentiation phenotype (Supplementary Fig. 18a-c). In addition, we found that Mmp2 and Mmp9 expression was upregulated by Ang II, but PDE4D deficiency only rescued Mmp2 expression in vitro (Supplementary Fig. 19a, b). Moreover, expression of the ECM genes Collagen 1A1 (Col1a1), Collagen 3A1 (Col3a1) and fibronectin 1 (Fn) 51 was elevated by Ang II stimulation, but only Col1a1 expression was reversed by PDE4D siRNA (Supplementary Fig.  20a-c). Regarding inflammation, Sadhan et al. reported that Ang II promoted CCL2, IL-6 and TNF-α expression in SMCs 52 . Therefore, we found that Ang II stimulation increased all three cytokines; however, suppression of PDE4D did not rescue the upregulation of Ccl2, Il-6, and Tnf-α expression ( Supplementary  Fig. 21a-c). These findings indicate that PDE4D in SMCs may affect ECM deposition by regulating Col1a1 and MMP2 expression in AAA progression.
In conclusion, our study shows that PDE4D in SMCs exacerbates Ang II-induced AAA. Furthermore, the inhibition of PDE4 alleviates vascular pathogenesis and AAA formation. We verified the mechanism by which PDE4D influences SMC apoptosis via in vitro and in vivo experimental models and identified those results using a variety of molecular biological methods. This study illustrates that PDE4D in SMCs plays a pivotal role in AAA and that PDE4 inhibitors may be potential targets for AAA treatment.