Novel Treatment of Hypertension by Specifically Targeting E2F for Restoration of Endothelial Dihydrofolate Reductase and eNOS Function Under Oxidative Stress

Abstract

We have shown that hydrogen peroxide (H₂O₂) downregulates tetrahydrobiopterin (H₄B) salvage enzyme dihydrofolate reductase (DHFR) to result in eNOS uncoupling and elevated blood pressure. Here we aimed to delineate molecular mechanisms underlying H₂O₂ downregulation of endothelial DHFR by examining transcriptional pathways hypothesized to modulate DHFR expression, and effects on blood pressure regulation of targeting these novel mechanisms. H₂O₂ dose- and time-dependently attenuated DHFR mRNA and protein expression, and enzymatic activity in endothelial cells. Deletion of E2F binding sites, but not those of Sp1, abolished H₂O₂ attenuation of DHFR promoter activity. Overexpression of E2F1/2/3a activated DHFR promoter at baseline, and alleviated the inhibitory effect of H₂O₂ on DHFR promoter activity. H₂O₂ treatment diminished mRNA and protein expression of E2F1/2/3a, while overexpression of E2F isoforms increased DHFR protein levels. Chromatin immunoprecipitation (ChIP) assay indicated direct binding of E2F1/2/3a to the DHFR promoter, which was weakened by H₂O₂. E2F1 RNA interference attenuated DHFR protein levels, while its overexpression elevated H₄B levels and H₄B/H₂B ratios in vitro and in vivo. In Ang II-infused mice, adenovirus-mediated overexpression of E2F1 markedly abrogated blood pressure to control levels, by restoring endothelial DHFR function to improve NO bioavailability and vasorelaxation. Bioinformatic analyses confirmed a positive correlation between E2F1 and DHFR in human endothelial cells and arteries, and downregulation of both by OxPAPC. In summary, endothelial DHFR is downregulated by H₂O₂ transcriptionally via an E2F-dependent mechanism, and that specifically targeting E2F1/2/3a to restore DHFR and eNOS function may serve as a novel therapeutic option for the treatment of hypertension.

Summary

These new findings indicate that specifically targeting E2F may serve as a novel therapeutic option for the treatment of hypertension.

Introduction

According to the 2017 report of the American Heart Association, hypertension affected 85.7 million adults (≥20 years of age) in the US between 2011 and 2014¹. Hypertensive patients display phenotype of endothelial dysfunction, which is characterized by decreased nitric oxide (NO) bioavailability and uncoupling of endothelial nitric oxide synthase (eNOS)^{2, 3}. While the molecular mechanisms of hypertension remain to be fully elucidated, a number of studies from our group and others have revealed a central role of eNOS uncoupling in the pathogenesis of hypertension^4–7. In an Angiotensin II (Ang II)–induced hypertensive mouse model, we found that Ang II induces deficiency of dihydrofolate reductase (DHFR) specifically in endothelial cells, which in turn leads to eNOS uncoupling and elevated blood pressure^{5, 6}. DHFR is the rate limiting salvage enzyme for the eNOS cofactor tetrahydrobiopterin (H₄B), deficiency of which is responsible for the uncoupling switch of eNOS^{4–6, 8, 9}. Furthermore, we investigated the mechanisms of Ang II-induced endothelial DHFR deficiency. We have demonstrated that Ang II downregulates DHFR expression through hydrogen peroxide (H₂O₂)⁸. In addition, we have shown that Ang II-induces eNOS uncoupling via H₂O₂ dependent DHFR deficiency and consequent reduction in H₄B bioavailability, in cultured endothelial cells, wild type mice, hph-1 mice, and apoE null mice ^{5, 6, 9}, ¹⁰. In these mouse models, modestly uncoupled eNOS causes hypertension while severely uncoupled eNOS promotes formation of abdominal aortic aneurysms^{5, 9, 10}.

Increasing evidence has demonstrated that oxidative stress plays an important role in the development of various vascular diseases^11–19. Our recent studies have shown that restoration of DHFR function with either folic acid (FA) diet or DHFR overexpression recoupled eNOS to attenuate Ang II-induced oxidative stress, NO deficiency, and elevated blood pressure^{4, 5, 7}. Therefore, identification of DHFR modulating mechanisms is of great importance in revealing novel therapeutics for hypertension via restoration of eNOS coupling activity. Several regulatory mechanisms for DHFR have been demonstrated in systems other than endothelial cells. For example, at transcriptional level, the transcriptional factor Sp1 has been shown to bind to the mouse DHFR promoter to upregulate its expression ²⁰. At posttranscriptional level, miR-24 has been shown to bind to DHFR mRNA to result in reduced DHFR protein expression²¹. In addition, the DHFR inhibitor methotrexate (MTX) can upregulate DHFR protein expression by attenuating the self-binding of the DHFR protein to its mRNA to allow increased translation of DHFR mRNA^{22, 23}. Nonetheless, molecular mechanisms underlying regulation of DHFR by Ang II-induced oxidative stress have remained unknown.

The E2Fs are a family of transcription factors, initially identified as a cellular activator that binds to the adenoviral E2 gene promoter²⁴. Members of E2F family have been reported to play opposite roles in biological processes such as transcriptional activation and repression, cell proliferation and apoptosis, and tumor inhibition and carcinogenesis²⁵. The E2F family contains several evolutionally conserved domains, including a DNA-binding domain, a dimerization domain, a transcriptional activation domain, and a pocket protein binding domain^{26, 27}. In the DHFR promoter region, there are four reiterated Sp1 binding sites, and two overlapping and inverted E2F binding sites²⁸. Therefore, we set out to examine the hypothesis that transcriptional mechanisms involving E2F family members are involved in the H₂O₂ attenuation of DHFR and subsequent regulation of blood pressure.

In this study, we investigated molecular mechanisms underlying oxidative stress downregulation of DHFR by specifically examining transcriptional pathways that are hypothesized to regulate endothelial DHFR expression, especially the E2F family transcriptional factors. We found that H₂O₂ transcriptionally decreased mRNA expression of E2F1/2/3a, which resulted in attenuated protein abundance and their binding to DHFR promoter as assessed by chromatin immunoprecipitation (ChIP) assays. This was accompanied by decreased DHFR mRNA and protein expression. Based on this novel molecular mechanism, adenovirus-mediated in vivo delivery of E2F1 reduced blood pressure via restoration of endothelial DHFR expression, recoupling of eNOS, augmented NO bioavailability, as well as improved endothelium-dependent vasorelaxation. In addition, bioinformatic analyses confirmed a positive correlation between E2F1 and DHFR in human endothelial cells and arteries, and downregulation of both by oxidized phospoholipids (OxPAPC). Taken together, our data indicate that specifically targeting E2F to improve DHFR and eNOS function under oxidative stress may serve as a novel therapeutic option for the treatment of hypertension.

Materials and Methods

Note: data, analytic methods, and study materials will be made available to other researchers upon publication of this research manuscript. These will not be stored online or publicly but can be shared electronically or physically upon request.

Most of the Methods see online supplemental materials

Construction of DHFR reporter plasmids and luciferase assay

Wild-type DHFR promoter (DHFR-1, −771 to +9) was amplified from HEK293T genomic DNA and sub-cloned into pGL3 basic luciferase reporter plasmid, which is a generous gift from Dr. Xiangming Ding (UCLA, USA). The deletions of transcriptional factor binding sites were generated by PCR. As shown in Fig 1, DHFR-2 is deleted of two SP1 binding sites from −553 to −516. DHFR-3 is deleted of two SP1 binding sites from −121 to −77. DHFR-4 is deleted of two E2F binding sites from −67 to −56. All mutants were confirmed by sequencing. For luciferase activity assay, bovine aortic endothelial cells (BAECs) were seeded in 48 well plates, and plasmids of pBH-E2F1/2/3a/7/8 (150 ng), pCMV-E2F4/5/6 (100 ng), and equal amount of corresponding empty vectors were co-transfected with wild-type (WT) DHFR promoter or different deletions (200 ng/well), as well as pRLrenilla luciferase reporter plasmid (5 ng/well), which was a generous gift from Dr. Yibin Wang (UCLA, USA). After incubation for 5 hrs, BAECs were starved in medium 199 with 5% FBS overnight and subjected to 100 μmol/L H₂O₂ stimulation for 12 hrs. Luciferase activity was determined by the dual-luciferase reporter assay system (Promega, Fitchburg, WI, USA). All experiments were performed in triplicates from independent repeats using different passages of cells.

Fig. 1: E2F1, E2F2 and E2F3a mediate hydrogen peroxide inhibition of DHFR promoter activity.

(A) Schematic representation of the DHFR promoter and its deletion constructs. DHFR-1 is the wild-type (WT) DHFR promoter (−771 to +9) inserted into the pGL3 luciferase reporter vector. DHFR-2/3/4 were deletion constructs containing DHFR promoters missing SP1 or E2F binding sites (missing −553 to −516, −121 to −77 or −67 to −56). (B) Luciferase reporter constructs of wild-type DHFR promoter and deletion variants were co-transfected with pRLrenilla luciferase plasmid into BAECs. Twenty-four hours later, cells were exposed to H₂O₂ (100 μM) for another 12 h in M199 medium containing 5% FBS. Luciferase activity was measured by the dual-luciferase reporter assay system and normalized to the renilla luciferase activity. Data were presented as Mean±SEM. *p<0.05, ***p<0.001 vs. ctrl, n=3. (C) pBH-vector, pBH-E2F1, pBH-E2F2, pBH-E2F3a, pBH-E2F7 or pBH-E2F8 plasmid was co-transfected with wild-type DHFR luciferase reporter as well as renilla luciferase reporter. Twenty-four hours later, cells were exposed to H₂O₂ (100 μM) for another 12 h in M199 medium containing 5% FBS. Luciferase activity was measured as described above. The group of empty pBH-vector without H₂O₂ treatment was used as control, and all other groups were shown as the fold differences compared to it. (D) The data of (C) were re-analyzed by normalizing H₂O₂ treated pBH-E2F groups to their corresponding groups of pBH-E2F plasmid without H₂O₂ treatment. (E) pCMV-vector, pCMV-E2F4, pCMV-E2F5 or pCMV-E2F6 plasmid was co-transfected with wild-type DHFR luciferase reporter as well as renilla luciferase reporter. Then the same treatment was performed as in (C). The group of empty pCMV-vector without H₂O₂ treatment was used as control, and all other groups were shown as the fold differences compared to it. (F) The data of (E) were re-analyzed by normalizing H₂O₂ treated pCMV-E2F groups to their corresponding groups of pCMV-E2F plasmid without H₂O₂ treatment. All data are presented as Mean±SEM. **p<0.01, ***p<0.001 vs. empty vector (C and E) or corresponding control group (D and F), n=3.

Real-time telemetry blood pressure measurement

Male C57BL/6 mice (12 weeks old) were purchased from Charles River Laboratories (Wilmington, MA, USA). Wireless blood pressure transmitters were implanted into the animals as we previously published^{5, 7}. The catheter of the blood pressure probe was inserted into the left carotid artery, whereas the body of the probe was inserted into the right flank. Animals were given one week to recover from the surgery. Then Ad-E2F1 or Ad-Luc (10⁹ IFU) in 50 μl sterilized PBS was injected directly into the left ventricle of the heart. After three days’ recovery, mice were subcutaneously implanted with osmotic pumps (DurectCorp, Cupertino, CA, USA) containing Ang II (0.7 mg/kg/day, Sigma-Aldrich, St. Louis, MO, USA). Blood pressure was monitored for the infusion period of 14 days. Measurements were made daily from 11:00 AM to 4:00 PM at a 100-Hz sampling rate. Average blood pressure was calculated daily as the average of the entire recording period. Two weeks after Ang II infusion, mouse aortas were used for measurement of NO bioavailability using ESR, or isolation of aortic endothelial cells as previously described ^{6, 7, 9, 10}. The endothelial cell fraction and the endothelial cell-denuded aortas were lysed for Western blot analysis. The use of animals and experimental procedures were carried out based on protocols approved by the UCLA Institutional Animal Care and Use Committee (IACUC).

Determination of NO bioavailability using ESR

Aortic NO bioavailability was determined using ESR as previously described^{5–7, 9, 10}. Briefly, freshly isolated aortas were cut into 2mm rings, and then incubated in freshly prepared NO specific spin trap Fe²⁺(DETC)₂ colloid (0.5 mmol/L) in nitrogen bubbled, modified Krebs/HEPES buffer at 37°C for 60 min, in the presence of calcium ionophore A23187 (10 μmol/L). The aortic rings were then snap frozen in liquid nitrogen and loaded into a finger Dewar for measurement with ESR. The instrument settings were as the followings: center field, 3440; sweep width, 100 G; microwave frequency, 9.796 GHz; microwave power 13.26mW; modulation amplitude, 9.82 G; 512 points of resolution; and receiver gain 356.

Determination of eNOS uncoupling using ESR

eNOS uncoupling was assessed by ESR analysis as we previously described^{5, 7–10, 29, 30}. In brief, freshly isolated aortas were homogenized in lysis buffer supplemented with protease inhibitor cocktail³¹. Then the tissue lysate was incubated with superoxide specific spin trap 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (CMH, 500 μmol/L) in the presence or absence of L-NAME (10 μmol/L) and/or PEG-SOD (20 U/mL). Next, the mixture was loaded into a capillary and analyzed for superoxide production by ESR. The SOD inhibitable superoxide production (normalized by protein concentration) is shown in Fig. 5E. The difference in superoxide production in the presence or absence of L-NAME was indicative of eNOS coupling/uncoupling activity. In controls where eNOS is coupled, L-NAME increases measured superoxide production by removing the buffering effects of NO on superoxide. When uncoupled, eNOS produces superoxide instead of NO, and inhibition of eNOS by L-NAME results in a decrease in measured superoxide production.

Fig. 5: Restoration of NO bioavailability and reduction in blood pressure by adenovirus mediated E2F1 overexpression in vivo.

Wild-type mice were injected with Ad-luc or Ad-E2F1 (10⁹ IFU) directly into the left ventricle of the heart three days before being infused with Ang II (0.7 mg/kg/day) for 14 days. (A) Representative Western blots of eNOS, E2F1, DHFR, and β-actin in primarily isolated aortic endothelial cells from Ang II-infused animals. (B) Blood pressure responses as monitored by an intracarotid telemetry method. (C) and (D) The aortic NO bioavailability as determined by ESR. (C) Representative ESR NO spectra. (D) Quantitative grouped data of bioavailable NO. Mean±SEM, n=3–5, ***p<0.001. (E) eNOS uncoupling activity as determined by ESR: L-NAME-dependent superoxide production. Mean±SEM, n=5, **p<0.01. (F) Endothelium-dependent vasorelaxation in response to acetylcholine in pre-contracted aortic rings. Mean±SEM, n=3–4, *p<0.05, Ad-Luc vs. Ad-Luc+Ang II. ##p<0.01, Ad-E2F1+Ang II vs. Ad-Luc+Ang II.

Bioinformatic analysis of E2F1 correlation to DHFR

We collected four large scale human transcriptomic data related to the vasculature, including human aortic endothelial cells³², aortic artery, coronary artery, and tibial artery from the Gene by Tissue Expression (GTEx) study³³, to examine the correlation between E2F1 and DHFR. The aortic endothelial cell dataset was retrieved from GSE30169, which contains 629 microarrays of human primary aortic endothelial cells with and without treatment with oxidized phospoholipids (OxPAPC), a pro-inflammatory factor involved in atherosclerotic plaque formation³². There were 307 control arrays and 322 OxPAPC treated arrays. Generally, duplicate arrays on 158 cultures for both control and OxPAPC conditions were generated; in 12 cases, expression per condition was based on a single array, and 1 culture had triplicate OxPAPC arrays and 4 cultures had quadruplicate OxPAPC arrays. The expression intensity values were normalized with the robust multi-array average (RMA) normalization method using the justRMA function of R package “affy”^{34, 35}. The gene-level expression was then obtained by the average expression of probes for the same gene. From the GTEx project v7³³, 53 tissue-specific RNASeq expression data are available (https://gtexportal.org/home/datasets). We collected the expression data on three artery related tissues: aortic artery, coronary artery and tibial artery, where the sample size of these three tissues are 299, 173 and 441, respectively. The normalized counts per gene, TPMs (transcripts per million)³⁶ were used for correlation analysis.

Pearson correlation was calculated between E2F1 and DHFR in each datasets and various conditions. Differences in the expression levels of E2F1 and DHFR between control and OxPAPC-treated aortic endothelial cells were assessed using two-sided Wilcoxon rank sum test ³⁷.

Results

DHFR expression and activity are downregulated by hydrogen peroxide in BAECs.

Our previous study has shown that H₂O₂ attenuates DHFR protein expression in endothelial cells ⁸, which mediates Ang II-induced eNOS uncoupling to result in development of hypertension in wild type mice, and formation of abdominal aortic aneurysm (AAA) in genetic strains ^5–10. In order to explore detailed molecular mechanisms underlying H₂O₂ regulation of DHFR, BAECs were treated with different doses of H₂O₂ for 24 hrs, or 100 μmol/L H₂O₂ at indicated time points³⁸. As shown in Fig. S1A and S1B, H₂O₂ dose- and time-dependently attenuated protein abundance of DHFR. Furthermore, DHFR activity determined by HPLC mirrored this response (Fig. S1C and S1D). To further investigate whether H₂O₂-induced changes in protein abundance of DHFR is a result of transcriptional regulation, DHFR mRNA levels were determined by qRT-PCR. As demonstrated in Fig. S1E and S1F, H₂O₂ downregulated DHFR mRNA levels in a dose- and time-dependent manner.

H₂O₂ inhibition of DHFR promoter activity requires E2F binding sites.

Next, promoter activity assay was performed using WT DHFR promoter and three promoter deletions as shown in Fig. 1A (DHFR-2: deletion of two Sp1 sites from −553 to −516; DHFR-3: deletion of two Sp1 sites from −121 to −77; DHFR-4: deletion of two E2F sites from −67 to −56). Of note, H₂O₂ treatment resulted in a decline in WT DHFR promoter activity by 59% (Fig. 1B). Deletion of E2F binding sites (DHFR-4), but not that of Sp1 sites (DHFR-2 and DHFR-3), completely abolished the inhibitory effects of H₂O₂ on DHFR promoter activity (Fig. 1B). These data indicate that E2F binding sites are required in H₂O₂ regulation of DHFR promoter activity.

Diminished E2F1, E2F2 or E2F3a mediates H₂O₂-induced reduction in DHFR promoter activity.

Since E2F binding sites were found critical for the regulation of DHFR promoter activity by H₂O₂, we next examined which one(s) of the eight members of the E2F family was involved^{27, 39, 40}. Promoter activity assays indicate that without H₂O₂ treatment transfection of E2F1, E2F2 and E2F3a activated DHFR promoter activity (Fig. 1C), while that of E2F4, E2F5, E2F7 and E2F8 inhibited its activity (Fig. 1C and1E). Transfection with E2F6 plasmid had no obvious effect (Fig. 1E). Under the treatment with H₂O_2, E2F1, E2F2 or E2F3a overexpression effectively alleviated H₂O₂-induced attenuation in DHFR promoter activity (Fig. 1D and1F). Moreover, H₂O₂ significantly reduced mRNA levels of E2F1, E2F2 and E2F3a (Fig. 2A-C), which resulted in corresponding decreases in their protein abundances by around 50% (Fig. 2D-F). These data indicate that deficiencies in E2F1/2/3a are specifically involved in H₂O₂ downregulation of DHFR.

Fig. 2: Hydrogen peroxide downregulates E2F1, E2F2 and E2F3a mRNA and protein expression while E2F1/2/3a overexpression upregulates DHFR protein abundance.

(A-C) BAECs were harvested after exposure to H₂O₂ (100 μM, 24 h). E2F1/2/3a mRNA levels were determined by quantitative RT-PCR with the housekeeping gene GAPDH as an endogenous control. Means±SEM, *p<0.05, ***p<0.001 vs. ctrl, n=3. (D-F) BAECs were harvested after exposure to H₂O₂ (100 μM, 24 h). E2F1, E2F2 and E2F3a antibodies were used to detect protein levels of E2F1/2/3a. Means±SEM, **p<0.01 vs. ctrl, n=3–4.

H₂O₂ weakens binding of E2F1/2/3a to DHFR promoter.

To further examine the direct roles of E2F1/2/3a in H₂O₂ regulation of DHFR, ChIP assay was employed to evaluate physical interaction between E2F1/2/3a and DHFR promoter with or without H₂O₂ treatment. Two primers for each E2F isoform were used to amplify targeted DNA sequence harboring E2F1/2/3a. Meanwhile, a negative primer that recognizes 5 kb upstream of E2F binding sites was also designed. As shown in Fig. 3A and3B, compared to the IgG group E2F1 antibody significantly pulled down more E2F binding fragment, implicating that E2F1 physically associates with DHFR promoter. Similar results were found when E2F2 (Fig. 3C and3D) or E2F3a (Fig. 3E and3F) was immunoprecipitated. As is obvious, H₂O₂ significantly inhibited binding to DHFR promoter of all three E2F isoforms.

Fig. 3: ChIP-qPCR assay indicating hydrogen peroxide inhibition of the binding of E2F1, E2F2 and E2F3a to DHFR promoter.

(A-B) BAECs were treated with 100 μM H₂O₂ for 24 h and harvested for ChIP assay. Cross-linked chromatin was immunoprecipitated with normal rabbit IgG or E2F1 antibody and analyzed by qPCR. Two primers (P1 and P2) encompassing the predicted E2F binding site were used for ChIP-qPCR. (C-F) BAECs were transfected with pCMV-E2F2 (C and D) or pCMV-E2F3a (E and F) for 36 h before exposure to H₂O₂ (100 μM, 12 h). Cross-linked chromatin was immunoprecipitated with normal rabbit IgG or HA antibody and analyzed by qPCR using two primers. For all data the level of enrichment was expressed as fold changes relative to IgG immunoprecipitates. Data are presented as Mean±SEM. *p<0.05, **p<0.01, ***p<0.001, n=3–7.

E2F1 increases H₄B bioavailability via upregulation of DHFR in vitro and in vivo.

While all three E2F isoforms are involved in the downregulation of DHFR by H₂O₂, H₄B levels were determined following overexpression of E2F1 as a representative isoform of the three. As shown in Fig. 4A, transfection of endothelial cells with siRNA targeting E2F1 decreased its protein levels to 61.0%±8.8% of controls, while it also reduced the protein abundance of DHFR to 64.9%±8.7% of controls. On the other hand, overexpression of E2F1 in endothelial cells significantly increased bioavailable H₄B levels (Fig. 4B, HA-empty vector: 5.0±0.2 pmol/mg protein vs. HA-E2F1: 7.1±0.3 pmol/mg protein, p<0.05). Adenovirus-mediated overexpression of E2F1 in vivo also resulted in elevated aortic H₄B bioavailability (Fig. 4C, Control adenovirus Ad-Luc: 0.98±0.05 pmol/mg protein vs. Ad-E2F1: 1.35±0.03 pmol/mg protein, p<0.01). In addition, both in vitro (Fig. 4D) and in vivo (Fig. 4E) overexpression of E2F1 increased H₄B:H₂B ratio by around 1.7 fold. These data indicate that augmentation of the E2F1-DHFR axis is effective in improving H₄B bioavailability.

Fig. 4: E2F1 increases H₄B bioavailability via upregulation of DHFR in vitro and in vivo.

(A) BAECs were transfected with 100 nM E2F1 siRNA for 48 h, and harvested for western blotting. The protein abundances of E2F1, DHFR, and β-actin were detected. Means±SEM, *p<0.05 vs. ctrl, n=5. (B) and (D) BAECs were transfected with pCMV-HA or pCMV-HA-E2F1 for 48 h. H₄B levels were determined by HPLC, and H₄B:H₂B ratio was calculated. Mean±SEM. *p<0.05, n=3–4. (C) and (E) Wild-type mice were injected with Ad-luc or Ad-E2F1 (10⁹ IFU) directly into the left ventricle of the heart for 14 days. Aortic H₄B levels were determined by HPLC, and H₄B:H₂B ratio was calculated. Mean±SEM. **p<0.01, n=4.

Overexpression of E2F1 attenuates Ang II induced hypertension via restoration of endothelial DHFR function to improve NO bioavailability and endothelium-dependent vasorelaxation.

In view of the intermediate role of E2F1/2/3a deficiency in H₂O₂ attenuation of DHFR protein expression and activity, the effects on blood pressure regulation of E2F1 overexpression in Ang II infused mice were examined. Ad-E2F1 (or Ad-Luc as the control) was directly injected into the left ventricle of the heart to achieve overexpression of E2F1. As shown in Fig. 5A, transfection of Ad-E2F1 in vivo led to increased protein abundance of E2F1 and DHFR (Fig. 5A). This has resulted in markedly abrogated blood pressure in Ang II infused mice (Fig. 5B, systolic BP of Ad-E2F1 group at day 14: 136.0±4.8 mmHg vs. systolic BP of Ad-Luc group at day 14: 172.2±5.4 mmHg, p<0.01, real-time telemetry). Upregulation of endothelial DHFR protein expression is anticipated to improve NO bioavailability^{5, 6, 8, 9}. Indeed, Ad-E2F1 overexpression resulted in markedly restored NO bioavailability in Ang II-infused animals (Fig. 5C and5D). Moreover, Ang II-induced eNOS uncoupling activity was also completely attenuated by E2F1 overexpression (Fig. 5E). Next, we evaluated endothelium-dependent vasorelaxation in these animals. As shown in Fig. 5F, Ad-E2F1 overexpression markedly restored vasorelaxation of aortic rings isolated from Ang II-infused animals.

To demonstrate whether H₂O₂ is involved in Ang II downregulation of E2F1, we pre-treated BAECs with PEG-catalase (PEG-CAT, 100 U/ml, 30 min) prior to exposure of the cells to Ang II (100 nmol/L) for 24 hrs. As shown in Fig. S2A, PEG-CAT reversed Ang II-induced downregulation of E2F1, indicating that Ang II attenuates E2F1 expression through H₂O₂. There is also a small and non statistically significant reduction in baseline E2F1 expression with PEG-CAT treatment, which is however not surprising as low levels of ROS are known to play important roles in cell signaling such as that involved in proliferation and migration⁴¹. We further examined which signaling pathway(s) is(are) required for H₂O₂-induced E2F1 downregulation. Inhibition of p38 MAPK by SB203580 (10 μmol/L) completely prevented H₂O₂ downregulation of E2F1, while inhibition of ERK (U0126, 50 μmol/L), PI3K (Wortmannin, 100nmol/L), or JNK (JNK inhibitor, 10 μmol/L) had no effect (Fig. S2B). Taken together, these data indicate that overexpression of E2F1 is highly effective in attenuating Ang II-induced hypertension via restoration of endothelial DHFR expression to improve NO bioavailability and endothelium-dependent vasorelaxation, and that Ang II induced E2F1 deficiency is mediated by H₂O₂ activation of p38 MAPK.

Bioinformatic analyses confirms positive correlation between E2F1 and DHFR in human endothelial cells and arteries, and downregulation of both by oxPAPC.

We also employed bioinformatic approaches (details see Methods section) to further evaluate the relationship between E2F1 and DHFR using existing datasets of human endothelial cells and arteries. As shown in Fig. 6A, positive correlation between E2F1 and DHFR was found in human aortic endothelia cells, with (middle panel) or without (right panel) treatment of OxPAPC. Correlation was also significant for data that pooled all endothelial cells (Fig. 6A, left panel). We then examined three transcriptomic datasets of human arteries of aortas, and coronary and tibial arteries. We found that expression of E2F1 positively correlated with expression of DHFR in all these human arteries. Moreover, both E2F1 and DHFR were downregulated by oxPAPC in human aortic endothelial cells (Fig. 6C), which is similar to our findings of downregulation of E2F1/DHFR axis in hypertension. Therefore, these bioinformatics analyses further indicate that a loss of E2F1 to result in DHFR deficiency may be an important pathological process to target for vascular diseases.

Fig. 6: Targeting E2F for the restoration of endothelial dihydrofolate reductase and eNOS function to treat hypertension.

(A-C) Validation of the relationship between E2F1 and DHFR using large human transcriptomic datasets. (A) Correlation plots between E2F1 and DHFR in human aortic endothelial cells from all samples (left panel), controls only (middle panel), and oxPAPC treatment experiments (right panel). (B) Correlation plots between E2F1 and DHFR in aortic artery tissues (left panel), coronary artery tissues (middle panel), and tibial artery tissues (right panel). “r” and “p value” represent the Pearson correlation coefficient and statistical significance of the correlation. (C) Significant downregulation of E2F1 (left panel) and DHFR (right panel) between control and oxPAPC-treated aortic endothelial cells, where p value is computed from two-sided unpaired Wilcoxon test. (D) Schematic illustration of a role of E2F/DHFR deficiency in oxidative stress induced hypertension, and of targeting E2F as a novel therapeutic option for the treatment of hypertension. Left panel: Angiotensin II (Ang II) induces a rapid and transient activation of endothelial NADPH oxidase (NOX), resulting in oxidative stress and hydrogen peroxide-dependent downregulation of E2F1 mRNA expression and protein abundance. Loss of E2F1 binding to DHFR promoter results in reduced DHFR protein expression, uncoupling of eNOS, and development of hypertension. Right Panel: Adenovirus-mediated overexpression of E2F1 in vivo transcriptionally upregulates DHFR protein expression to restore eNOS function, which in turn restores nitric oxide (NO) bioavailability and endothelium-dependent vasodilatation to reduce blood pressure.

Discussion

The most significant findings of the present study include: 1) Downregulation of DHFR expression by H₂O₂ is mediated by decreased mRNA expression of E2F1/2/3a and its protein binding to DHFR promoter; 2) Overexpression of E2F1 to restore DHFR expression, H₄B levels, eNOS coupling activity and NO bioavailability completely attenuated elevated blood pressure in Ang II-infused mice. We have previously shown that H₂O₂ mediates Ang II uncoupling of eNOS to induce hypertension^5–10. In the current study we present further evidence that H₂O₂-induced downregulation of E2F1/2/3a expression and binding to DHFR promoter diminishes DHFR transcription and protein abundance. Intriguingly, adenovirus-mediated overexpression of E2F1 to restore DHFR expression in Ang II-infused mice abolished elevation in blood pressure. These data indicate that specifically targeting E2F1 to restore DHFR function and eNOS coupling activity may serve as a novel therapeutic option for the treatment of hypertension.

Oxidative stress has been implicated in the pathogenesis of hypertension^{4–12, 16, 19, 42}. We have previously shown that NADPH oxidase (NOX)-derived vascular superoxide production causes eNOS uncoupling, NO deficiency and increased blood pressure^4–10. In particular, we have established that H₂O₂ produced from transient activation of endothelial NOX induces subsequent deficiency in DHFR to result in persistent H₄B deficiency and eNOS uncoupling⁸, which in turn mediates development of hypertension in Ang II-infused mice^5–9. Our current study mainly focuses on the mechanistic details of H₂O₂ downregulation of DHFR, and efficacies in reducing blood pressure of strategies targeting these mechanisms. Consistent with previous findings, H₂O₂ downregulated DHFR protein abundance in endothelial cells⁸. The present study further demonstrated reduced mRNA expression and impaired activity of DHFR in response to H₂O₂ exposure. The changes in DHFR mRNA appeared much sooner than that of protein, implicating that DHFR might be transcriptionally regulated by H₂O₂. Indeed, the dual luciferase reporter assay revealed that H₂O₂ exposure induces an evident decrease in DHFR promoter activity. Deletion of the E2F binding sites, but not those of Sp1, completely abolished H₂O₂ attenuation of DHFR promoter activity, indicating that E2F is involved in the transcriptional regulation of DHFR by H₂O₂.

E2F family has been categorized into two subfamilies^{43, 44} of “activators”, which include E2F1, E2F2 and E2F3a, and “repressors”, which include E2F3b and E2F4–8. However, this classification has been debated as the so-called activating E2Fs have been shown to repress equal amount of genes they activate, and that E2F4, E2F5 and E2F6 could actually work as activators^{45, 46} while E2F2 and E2F3 turned out to be repressors under some circumstances⁴⁷. Our data indicated that E2F1/2/3a take part in the regulatory process of H₂O₂ attenuation of DHFR, based on data from the luciferase reporter assay (overexpression with reasonable amount of plasmids) and the ChIP assay. E2F1/2/3a functioned as the “activators” in DHFR transcriptional regulation, which could be inhibited by H₂O₂. Hydrogen peroxide mainly decreased E2F1/2/3a at mRNA levels, which resulted in corresponding reduction in protein expression. This led to attenuated amount of E2F1/2/3a that is available for the physical binding to the DHFR promoter. In addition, we have shown that p38 MAPK is required for H₂O₂-induced downregulation of E2F1, whereas ERK, PI3K and JNK are not involved.

It has been shown that the E2F family members are responsible for controlling the transcription of various genes that are essential for cell cycling, development, differentiation, DNA synthesis and replication, and DNA damage repair⁴⁰. However, little is known about their roles in endothelial cells. In the present study, we found that E2F1/2/3a overexpression increased DHFR protein expression in cultured endothelial cells, and that adenovirus-mediated E2F1 overexpression upregulated endothelial DHFR protein levels in vivo. Our previous studies have shown that infusion of Ang II induces hypertension in wild type mice, in which DHFR deficiency-dependent eNOS uncoupling is critically involved^5–10. Restoration of DHFR by either folic acid diet or DHFR overexpression recoupled eNOS to reduce blood pressure^{4–6, 9}. Based on the novel mechanisms identified in this study that H₂O₂ induces DHFR deficiency via attenuation of E2F1/2/3a expression and binding to DHFR promoter, E2F1 was overexpressed using an adenoviral system in vivo. Overexpression of E2F1 in vivo restored endothelial DHFR protein abundance, eNOS coupling activity and NO bioavailability, resulting in improved endothelium-dependent vasorelaxation and reduced blood pressure.

Perspectives:

As schematically illustrated in Fig. 6D (using E2F1 as a representative E2F isoform of E2F1/2/3a), Ang II induces a rapid and transient activation of endothelial NOX, resulting in oxidative stress and H₂O₂-dependent downregulation of E2F1/2/3a mRNA and protein abundance. Loss of E2F1/2/3a binding to DHFR promoter leads to reduced DHFR protein expression, uncoupling of eNOS, and development of hypertension. Adenovirus-mediated overexpression of E2F1 however completely attenuated elevated blood pressure in Ang II infused mice, via upregulation of DHFR protein expression to restore H₄B and NO bioavailability and coupling activity of eNOS. Furthermore, we observed positive correlation between E2F1 and DHFR using four independent human transcriptomic datasets, and confirmed downregulation of both under oxidative conditions in human aortic endothelial cells. These results strongly support the robustness and translational value of our molecular mechanistic findings. In conclusion, our study indicates that specifically targeting E2F1/2/3a may be used as a novel therapeutic option for the treatment of hypertension.

What Is New? –

• First identification of an intermediate role of E2F deficiency in hydrogen peroxide downregulation of DHFR

• First identification that targeting E2F to restore DHFR and eNOS function is highly effective in reducing blood pressure

What Is Relevant?

• Our data reveal a novel mechanism underlying oxidative stress induced hypertension

• Our data establish a novel therapeutic approach for the treatment hypertension

List of Abbreviations

Ang II

Angiotensin II

BAECs

Bovine aortic endothelial cells

ChIP

chromatin immunoprecipitation

DHFR

dihydrofolate reductase

eNOS

endothelial nitric oxide synthase

FA

folic acid

H₂O₂

hydrogen peroxide

H₄B

tetrahydrobiopterin

MTX

methotrexate

NO

Nitric oxide

NOX

NADPH oxidase