Prolyl 4-Hydroxylase Domain Protein 3-Inhibited Smooth-Muscle-Cell Dedifferentiation Improves Cardiac Perivascular Fibrosis Induced by Obstructive Sleep Apnea

Background Intermittent hypoxia (IH) induced by obstructive sleep apnea (OSA) is a leading factor affecting cardiovascular fibrosis. Under IH condition, smooth muscle cells (SMAs) respond by dedifferentiation, which is associated with vascular remodelling. The expression of prolyl 4-hydroxylase domain protein 3 (PHD3) increases under hypoxia. However, the role of PHD3 in OSA-induced SMA dedifferentiation and cardiovascular fibrosis remains uncertain. Methods We explored the mechanism of cardiovascular remodelling in C57BL/6 mice exposed to IH for 3 months and investigated the mechanism of PHD3 in improving the remodelling in vivo and vitro. Results In vivo remodelling showed that IH induced cardiovascular fibrosis via SMC dedifferentiation and that fibrosis improved when PHD3 was overexpressed. In vitro remodelling showed that IH induced SMA dedifferentiation, which secretes much collagen I. PHD3 overexpression in cultured SMCs reversed the dedifferentiation by degrading and inactivating HIF-1α. Conclusion OSA-induced cardiovascular fibrosis was associated with SMC dedifferentiation, and PHD3 overexpression may benefit its prevention by reversing the dedifferentiation. Therefore, PHD3 overexpression has therapeutic potential in disease treatment.


Background
Obstructive sleep apnea (OSA) is a common disorder which is characterized by total or partial collapse of upper airways alternating with normal breathing [1,2]. Obstruction in gas exchange leads to chronic intermittent hypoxia (CIH), oxygen desaturation, hypercapnia, and arousal [3,4]. The Wisconsin Sleep Cohort Study reported that OSA prevalence was 4% in middle-aged men and 2% in middle-aged women (age 30-60 years) [5]. Subsequent studies suggest that prevalence is higher than previously reported in highincome countries (10% in women and 20% in men) [6,7]. In consideration of the obesity pandemic, the number of cases of OSA is likely to increase. A recent study shows that OSA affects 34% of men and 17% of women and is largely undiagnosed [8]. CPAP treatment, the gold standard therapy, exists but is often poorly tolerated for patients with OSA. Thus, the development of new or combinations of treatments are needed.
OSA is associated with increased cardiovascular morbidity and mortality, including hypertension, coronary heart disease, heart failure, pulmonary hypertension, and atrial fibrillation [1,9]. OSA can cause intermittent hypoxia (IH), hypercapnia, and sleep fragmentation. IH is the main injury factor leading to cardiovascular morbidity and mortality [10,11]. Sympathetic overactivation and systemic oxidative stress may be the main mechanisms associated with IH [1]. These abnormalities all contribute to the development of cardiovascular remodelling, including ventricular hypertrophy, endothelial dysfunction, carotid intima-media thickness, and alterations in the coronary microcirculation. The obvious modifications in cardiovascular remodelling are 2 BioMed Research International characterized by the disruption of normal myocardial structure through excessive collagen deposition. Our previous study confirms that OSA can induce cardiac perivascular fibrosis [12]. However, the detailed mechanisms remain unclear. Given the increased risk of cardiac remodelling in OSA patients, an improved understanding of the underlying mechanisms is necessary.
Different from terminally differentiated cells, vascular smooth muscle cells (VSMCs) own a distinctive ability of plasticity to alternate from a differentiated/contractile state, which expresses elevated levels of contractile proteins, such as -smooth muscle actin ( -SMA), towards a dedifferentiated/synthetic state, which expresses increased levels of osteopontin (OPN) at different pathologic conditions [13,14]. VSMC phenotypic switching is widely involved in atherosclerosis and research has proved that inhibiting VSMC phenotypic switching may be beneficial in atherosclerosis [14,15]. Hypoxia increased the involved genes' expression in dedifferentiation and profibrosis of human bladder smooth muscle cells [16]. Hypoxia inducible factor-(HIF-) inhibition can decrease systemic vascular remodelling diseases [17]. However, no research reports show that VSMC phenotypic switching is involved in OSA-induced cardiac perivascular fibrosis, and its mechanisms remain to be further elucidated.
The PHD3, which is located in the cytoplasm [18], has been regarded as a cellular low oxygen sensor [19]. With sufficient oxygen availability, PHD3 can hydroxylate the HIF-via von Hippel-Lindau protein (pVHL); however, it can maintain cell survival or proliferation functions under hypoxia [19]. PHD3 is different from other PHDs. First, PHD3 is abundantly expressed in the heart [20]. Second, PHD3 can retain its activity under prolonged mild hypoxia (2%-5%), so PHD3 has been deemed as the main regulator in prolonged and mild hypoxia [21]. Under hypoxia, PHD3 is found in high-order complexes. It is different from the aggregates which reveal under normoxia [22]. When PHD3 is released from the aggregates, it may be able to freely interact with its cytoplasmic and nuclear targets [23]. As a controversial protein, PHD3 maintains carcinoma cell growth [24], but other researchers also consider that the loss of PHD3 allows tumours to overcome hypoxic growth inhibition and sustain proliferation [25]. However, the function of PHD3 in VSMCs phenotypic switching has not been reported.
Based on these findings, we hypothesized that PHD3 may exhibit a protective effect on IH-induced cardiac microvascular fibrosis via inhibiting VSMCs phenotypic switching. To uncover the mechanisms, we explored its role in vitro and vivo.

Methods
The experiments conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health. The study protocol was approved by the Institutional Ethics Committee of Shandong University and Southeast University.
. . Study Design. C57BL/6J wild-type (WT) male 6-weekold mice (21-23 g) were purchased from Yangzhou University (Yangzhou, China). A total of 60 mice were randomly assigned to normoxia or IH exposure for 1 week. After one week, the IH mice were randomly divided into 5 groups: IH, IH+shRNA PHD3 NC (shNC), IH+lentiviral PHD3 NC (LvNC), IH+shRNA PHD3, and IH+lentiviral PHD3 (we followed the method of . . . IH. For 3 months, mice, which need to IH exposure, received 60 hypoxic events/h (20 s at 5% O 2 followed by 40 s of room air) during 8 h/d, corresponding to severe OSA. Control mice with normoxic breathing were placed in an identical device, but the hypoxic gas was replaced by room air [12,26,27]  . . Histology and Immunohistochemistry. The preparation of section and immunohistochemistry analysis was performed as described [12]. . . Cell Culture. VSMCs were purchased from ScienCell Research Laboratories, Inc. (Carlsbad, CA). Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) and 2 mM glutamine was used. Cells were cultured in a humidified 5% CO 2 and 95% air incubator at 37 ∘ C and 3-5 passages were used. Cells were treated with IH for 72 h. Air-phase set-point consisted of a 35minute hypoxic period, followed by 25 minutes of reoxygenation (21% O 2 and 5%CO 2 ) [12, 28, 29] (we followed the method of ). Before IH treatment, the VSMCs were infected with lentivirus at a multiplicity of infection of 15 for 24 h. Cells were treated with siRNA (GenePharma Shanghai, China) in order to inhibit HIF-1 expression. The target sequence for siHIF-1 was 5'-CCAUGUGACCAUGAGGAAATT-3' , and the negative sequence was 5'-UUUCCUCAUGGUCACAUGGTT-3' .
. . Western Blot Analysis. Whole-cell proteins were isolated from freshly dissected mice hearts and cell lysates. Western blot analysis was performed as described [30]. Primary antibodies was used against PHD3 (1:500 Novus), HIF-1 . . RNA Extraction and Quantitative Real-Time PCR. TRI-ZOL (TIANGEN) was used to extract total RNA from tissues or cultured VSMCs. Real-time PCR was performed using SYBR green reagent (TIANGEN) and ABI 7500 system. . . Statistical Analysis. Results are from 3 repeated experiments. Data are reported as the mean ± SD. Results were compared by 2-tailed Student t-test for 2 groups and oneway ANOVA followed by Tukey's t-test (2-tailed) for multiple groups. SPSS (v16.0; SPSS, Inc, Chicago, IL) was used for analysis. Differences were considered statistically significant at p < 0.05.

Result
. . IH Induced Collagen Deposition. To ascertain whether IH can induce cardiac fibrosis, histochemistry stain was used. Picrosirius red staining revealed greater collagen deposit in the region of perivascular and vascular of the IH than control. Quantitative analysis showed a 2.4-fold increase in IH group as compared with the control group (0.57 ± 0.11 versus 0.24 ± 0.07, p<0.05) (Figure 1(a)). Immunohistochemistry showed that IH enhanced collagen I expression when compared with the control group (0.76 ± 0.12 versus 0.32 ± 0.13, p<0.05) (Figure 1(b)). However, IH exposure had no effect on the expression of collagen III (p>0.05) (Figure 1(b)). Interestingly, compared with control, IH exposure did not obviously increase collagen deposition in myocardium (Supplementary Figure 1). In other words, the region of IH-induced collagen deposition only existed in cardiac perivascular and vascular tissues.

. . PHD Overexpression Improves IH-Induced Cardiac
Perivascular Fibrosis. Our previous study had shown that IH can induce PHD3 expression in vitro and vivo [12]. To ascertain whether PHD3 could improve cardiac fibrosis induced by IH, in this study, lentivirus was applied to intervene the expression of PHD3 in vivo and vitro. Transfection efficiency was evaluated by GFP, which reached values of up to 70% in vivo and 90% in vitro (Supplementary Figure 2B). Western blot revealed that lentiviral vector can increase or decrease PHD3 expression obviously compared with NC group in vivo and vitro ( Supplementary Figures 2A and 2B).
Picrosirius red staining indicated greater collagen deposit in the area of perivascular and vascular of the IH group than control. And quantitative analysis showed a 2.29-fold increase in IH group as compared with control mice (0.55 ± 0.13 versus 0.25 ± 0.07, p<0.01) (Figures 1(c) and 1(d1)).
. . IH Induces VSMC Phenotypic Switching. VSMC phenotypic switching has been shown to contribute to vascular remodelling. To explore whether IH induced VSMC phenotypic switching, we performed in vitro and in vivo experiments.
In vivo, -SMA, a specificity marker of differentiated/contractile state of VSMCs, was assayed by immunohistochemistry. The area of -SMA expression was restricted to the vascular media in both control and IH group. OPN, a specificity marker of dedifferentiated/synthetic state of VSMCs, was not expressed in vascular media in the control group. However, after IH treatment, the expression of OPN dramatically increased in vascular media compared with the control (5.5 ± 0.5 versus 1, p<0.01) (Figures 2(a) and 2(b)) (red arrow). Furthermore, we used immunofluorescencecolocalized staining to detect VSMC phenotypic switching in vitro. We demonstrated that IH exposure significantly increased the expression of OPN, which is consistent with the study in vivo (Figure 2(c)).

Discussion
In the present study, we consider that PHD3 improves the early cardiovascular remodelling induced by OSA. In the field of experimental OSA, considerable studies have focused on the effects of IH, but few studies have explored a novel therapeutic target for preventing early cardiovascular remodelling in patients with OSA. To our knowledge, this research is the first study showing that VSMC phenotypic switching is the basis for the progression of OSA-induced cardiovascular remodelling. Our research identifies a previously undiscovered function of PHD3 in OSA-induced cardiovascular remodelling. Our novel experiment revealed that PHD3 owns a protective role in IH-induced cardiovascular fibrosis by inhibiting smooth muscle dedifferentiation. OSA induces various pathophysiological triggers, but IH plays the most pivotal role in the development of cardiovascular diseases [31]. Some research has demonstrated beneficial effects in animal models with IH short-term exposure [32]. However, long-term exposure (at least 4 weeks) often causes detrimental effects [33]. In the present study, we evaluated the level of cardiovascular fibrosis resulting from IH for 3 months, which can mimic severe OSA in patients. After IH exposure (3 months), immunohistochemical staining indicates that IH induced perivascular and vascular media fibrosis; however, it had no effect on myocardial interstitial fibrosis, which is in accordance with the previous study [34]. Interestingly, compared with the control, the expression of collagen I was significantly increased after IH exposure, but there is no significant difference in collagen III expression in vivo. Perivascular fibrosis, which can reduce the elasticity of microvessels, increases the vascular wall thickness and plays an important role in the development of heart failure [35]. In the present study, we did not access the cardiac function because our previous study indicated that cardiac function decreased after IH exposure [12]. PHD3, a cellular oxygen sensor, is not an activator of cell apoptosis but is a promoter of cell survival under restricted oxygen [19]. PHD3 is thought to be the most important regulator of HIF-1 under severe and prolonged hypoxia compared with other PHDs. Under hypoxia, besides HIF, PHD3 may possess other hydroxylation targets, such as ATF4, PK-M2, and Pax2 [19]. In our previous study, PHD3 expression was upregulated after IH treatment in vivo and vitro, but its function may be limited. After interference and overexpression, we found that the fibrosis improved when PHD3 was overexpressed. Thus, we concluded that PHD3 improves IH-induced perivascular and vascular media fibrosis. To elucidate its mechanism, we designed the experiment in vivo and vitro. Vascular remodelling is dependent on dynamic interactions between many pathological factors. IH is deemed as an independent factor which promotes vascular remodeling [36]. As we know, VSMCs phenotypic switching plays a    key role in the processes of multiple vascular pathologies [37]. Different from the terminally differentiated cells, VSMCs own a distinctive ability of plasticity that allows the phenotypic shift from a differentiated/contractile state to a dedifferentiated/synthetic state [14]. In contractile state, VSMCs are characterized by low proliferation rates and low rates of protein synthesis [38]. However, in the synthetic state, VSMCs are characterized by relatively low contractile protein expression, re-entry into the cell cycle, high level of proliferation and migration, and high rates of protein synthesis and secretion [38]. VSMCs could easily switch between these two states when remodelling is required. A disruption of the balance, for example, the synthetic phenotype predominates, may be an underlying cause of many vascular diseases. With respect to VSMC phenotypic switching, most of the researchers paid their attention on atherosclerosis and aneurysms, and little is known regarding it on OSA-induced cardiovascular remodelling. Therefore, we first evaluated the switching of VSMCs phenotype during OSA-induced cardiovascular remodelling and found that the expression of OPN, a marker of synthetic state, was remarkably upregulated in microvascular media in vivo. To further demonstrate our observations, we designed experiments in vitro. VSMCs increased the expression of OPN after IH treatment. Interestingly, a subtle paradox was present here: the expression of -SMA showed no change in vivo and decreased in vitro. The disparity in results may reflect differences in vivo, which was affected by multiple factors, and in vitro, which was only affected by the single factor. In vitro, we also found that collagen I expression and not collagen III increased after IH-induced VSMCs phenotypic switching. This finding can explain why IH only induced collagen I to deposit in perivascular and vascular media in vivo.
We found another important finding in our study. PHD3 improved IH-induced cardiac microvascular fibrosis via inhibiting VSMC dedifferentiation in vivo and vitro. In our study, we indicated that VSMC dedifferentiation was involved in IH-induced cardiac microvascular fibrosis. Next, we used lentivirus to change PHD3 expression. We found that PHD3 overexpression can improve IH-induced VSMCs dedifferentiation in vivo and vitro. However, shPHD3 cannot improve it. In our study, we demonstrated that PHD3 may be a novel marker which has the capacity to modulate VSMCs phenotypic switching for the first time.
We further explored the molecular mechanism of PHD3 in vitro. Shan, F. indicated that HIF-1 was involved in the phenotypic modulation in pulmonary artery SMCs during hypoxia [39]. Lambert, C. M. indicated that HIF-1 inhibition could decrease systemic vascular remodelling diseases [17]. In our present study, qPCR revealed no change of HIF-1 mRNA in IH/shPHD3/LVPHD3 compared with the control. WB revealed that HIF-1 expression increased after IH exposure compared to control, but LvPHD3 can improve the increased expression. Immunofluorescence indicated that IH resulted in HIF-1 's relocation from cytoplasm to nucleus, and LvPHD3 can maintain most HIF-1 in the cytoplasm. To further clarify, we used siRNA to inhibit HIF-1 . We found that siHIF-1 can improve IH-induced VSMC phenotypic switching. Thus, we concluded that PHD3 improves VSMC dedifferentiation via degrading and inactivating HIF-1 and by not reducing generation. The axis of PHD3-HIF-1 -OPN/ -SMA may be involved in the improvement of VSMC dedifferentiation. Another study indicated that OPN enabled transcriptional upregulation of HIF-1a expression under both normoxia and hypoxia [40]. A plausible explanation for the controversy between these results might lie in the versatility and diversity of HIF-1 and OPN functions and the intrinsic divergence in the research subjects.
Several limitations are present in our research. (1) OSA results in IH, hypercapnia. All the negative factors could lead to cardiovascular remodelling. We only focused on IH. It is a limitation because the IH model does not represent OSA. However, IH is the most powerful factor in OSA-induced negative factors. Complications, such as cardiac remodelling and heart failure, were reported in our previous research [12]. Furthermore, the model has been used to explore OSA and its complications, such as cardiac remodelling, by international and domestic academicians [27]. (2) Cultured VSMCs can switch phenotype when they were passaged, which can cause potential bias in vitro. However, in each experiment, we have tried our best to use VSMCs from the same passage which might reduce the bias. (3) We used human aortic VSMCs to explore the mechanism in vitro. Mouse coronary micro-VSMCs in vitro are a reasonable option. We finally selected human cells because mouse coronary micro-VSMCs are difficult to extract and cannot be purchased from companies. Furthermore, PHD3 is an ortholog in both humans and mice. Hence, we firmly believe that human aortic VSMCs can replace mouse coronary micro-VSMCs in vitro study. (4) We used lentivirus to change the expression of PHD3, and its effect is not better than knockout or transgenic technology. However, in our study, the lentivirus basically achieved the anticipated effect.

Conclusions
We first found that PHD3 can improve cardiac perivascular fibrosis by inhibiting VSMC phenotypic switching in an OSA mice model. The mechanism may be involved in degrading and inactivating HIF-1 . Although the exact underlying mechanisms are not fully understood, given the cardioprotective effects of PHD3 overexpression, PHD3 may be a potential therapeutic target for OSA-induced heart diseases.

Data Availability
The data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflicts of interest.