Overexpression of the neuronal human (pro)renin receptor mediates angiotensin II-independent blood pressure regulation in the central nervous system

Abstract

Despite advances in antihypertensive therapeutics, at least 15–20% of hypertensive patients have resistant hypertension through mechanisms that remain poorly understood. In this study, we provide a new mechanism for the regulation of blood pressure (BP) in the central nervous system (CNS) by the (pro)renin receptor (PRR), a recently identified component of the renin-angiotensin system that mediates ANG II formation in the CNS. Although PRR also mediates ANG II-independent signaling, the importance of these pathways in BP regulation is unknown. Here, we developed a unique transgenic mouse model overexpressing human PRR (hPRR) specifically in neurons (Syn-hPRR). Intracerebroventricular infusion of human prorenin caused increased BP in Syn-hPRR mice. This BP response was attenuated by a NADPH oxidase (NOX) inhibitor but not by antihypertensive agents that target the renin-angiotensin system. Using a brain-targeted genetic knockdown approach, we found that NOX4 was the key isoform responsible for the prorenin-induced elevation of BP in Syn-hPRR mice. Moreover, inhibition of ERK significantly attenuated the increase in NOX activity and BP induced by human prorenin. Collectively, our findings indicate that an ANG II-independent, PRR-mediated signaling pathway regulates BP in the CNS by a PRR-ERK-NOX4 mechanism.

NEW & NOTEWORTHY This study characterizes a new transgenic mouse model with overexpression of the human (pro)renin receptor in neurons and demonstrated a novel angiotensin II-independent mechanism mediated by human prorenin and the (pro)renin receptor in the central regulation of blood pressure.

INTRODUCTION

Hypertension is one of the major risk factors for cardiovascular diseases. Despite advances in antihypertensive therapeutics, at least 15–20% of patients with high blood pressure (BP) have resistant hypertension (7, 21). The mechanisms underlying this resistance remain poorly understood, but elevation of sympathetic activity is one of its major characteristics (23, 60). The (pro)renin receptor (PRR) is a single-pass transmembrane receptor that has been identified in various tissues, including the brain, heart, placenta, liver, pancreas, lung, and kidney (39). We recently reported that PRR is present in several cardiovascular regulatory regions in the murine brain (35). Binding of prorenin to PRR induces a conformational change in prorenin that exposes the enzymatic site, leading to increased ANG II formation both in vitro and in vivo (26, 36, 49). In addition to promoting ANG II formation, binding of prorenin to PRR activates ANG II-independent intracellular signaling pathways that result in an increase in the production of profibrotic and proinflammatory factors (24, 28) as well as reactive oxygen species (ROS) (38, 44, 59). ANG II-dependent prorenin/PRR signaling has been shown to play significant physiological and pathophysiological roles in the central regulation of BP in animal models (14, 68). However, the importance of ANG II-independent prorenin/PRR signaling in the neural regulation of BP remains undefined. The goal of our study was to examine whether activation of an ANG II-independent prorenin/PRR signaling pathway regulates BP and to understand the mechanisms involved in vivo.

To achieve our goal, we generated a unique transgenic mouse model that expresses human PRR (hPRR) under the control of the neuron-specific synapsin 1 promoter (Syn-hPRR). This experimental design is innovative because human prorenin cannot cleave mouse angiotensinogen. Consequently, administration of human prorenin to Syn-hPRR mice cannot activate the ANG II-dependent pathway and instead selectively activates ANG II-independent PRR signaling in neurons. This provides a novel model for investigating the importance of these pathways in the neural control of BP in vivo. We found that intracerebroventricular infusion of human prorenin induced an elevation of BP in Syn-hPRR mice that was independent of ANG II formation but required phosphorylation of ERK1/2 and activation of NADPH oxidase 4 (NOX4) in the central nervous system (CNS).

METHODS

Animals

The Syn-hPRR cDNA transgene (NCBI Reference ID: NM_005765) was cloned into the pSYN-hPRR plasmid using SpeI and XhoI restriction enzymes, sequenced, and amplified. SpeI and XhoI doubled digestion yielded a 6,033-bp fragment containing the rat synapsin 1 promoter sequence and the hPRR cDNA transgene (Fig. 1A). The correct fragment was purified and microinjected in fertilized C57BL/6J × SJL/J (B6SJLF2) mouse embryos at the University of Iowa Genome Editing Facility. All mice were provided standard mouse chow and water ad libitum. All procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by Institutional Animal Care and Use Committees of the University of Iowa and University of Nevada-Reno. All mice in this study were used after at least seven generations of backcrossed breeding with C57BL/6J mice from the Jackson Laboratory (Bar Harbor, ME). The colony was maintained by breeding hemizygous transgenic mice with wild-type C57BL/6J mice. Both male and female Syn-hPRR mice and nontransgenic (NT) littermates (8–10 wk old) were used in this study.

Fig. 1.

Generation and characterization of transgenic mice expressing the human (pro)renin receptor (hPRR) transgene under control of the neuron-specific rat synapsin 1 promoter (Syn-hPRR). A: schematic of the Syn-hPRR construct used to develop transgenic mice. B: representative RT-PCR gel showing transgene expression of hPRR and β-actin in various tissues of Syn-hPRR mice (founder line 57994-1). C: representative RT-PCR gel showing transgene expression of hPRR and β-actin in various tissues in nontransgenic (NT) littermates. B, brain; H, heart; K, kidney; Lu, lung; Li, liver; S, spleen; M, skeletal muscle; V, vessel; F, white adipose tissue (fat); P, pancreas; +, positive control (Syn-hPRR construct used as a template); −, negative control (no-template PCR). D: hPRR protein in the cortex, hypothalamus, and brain stem of Syn-hPRR mice and NT littermates (n = 3–4/group). E: hPRR mRNA levels in brain homogenates of Syn-hPRR mice and NT littermates (n = 3–4/group) determined by quantitative real-time RT-PCR). F: mouse PRR mRNA levels in brain homogenates of Syn-hPRR mice and NT littermates (n = 3/group) determined by quantitative RT-PCR. G: PRR protein immunofluorescence in the subfornical organ (SFO), paraventricular nucleus (PVN), nucleus of tractus solitarius (NTS), area postrema (AP), and rostral ventrolateral medulla (RVLM) in Syn-hPRR mice and NT littermates (n = 3/group). Note that images of gels in B and C (top) were from two different agarose gels.

Syn-hPRR Mice Genotyping

Snipped tail samples were obtained on day 7–10 after birth, and genomic DNA was isolated using a Red Extract-N-Amp Tissue PCR kit (Sigma-Aldrich, St. Louis, MO). Genotyping was performed by PCR as described by the manufacturer using 4 μl of tail extract and the primer pair 5′-TCCAGACCCTACGGACAAGA-3′ (forward) and 5′-CCTCACGACCAACTTCTGCA-3′ (reverse), yielding a 560-bp product.

RNA Isolation, Conventional PCR, and Real-Time Quantitative PCR

Total RNA was isolated from various tissues of Syn-hPRR mice and NT littermates as previously described (35, 36). hPRR expression in the mouse brain, heart, kidney, lung, liver, spleen, muscle, vessels, white adipose tissue (fat), and pancreas was detected by conventional PCR using primers specific for hPRR and mouse β-actin designed using PrimerQuest Software (Integrated DNA Technologies, Coralville, IA). PCR products were analyzed by agarose gel electrophoresis with ethidium bromide staining, and a low-range DNA ladder (Invitrogen) was used as a marker for PCR product size. Images were captured using a UVP Bio-imaging System. For quantitative PCR experiments, total RNA was isolated from homogenates of the brain hypothalamus from a freshly euthanized mouse using an RNeasy Mini Kit (Qiagen, Germantown, MD) and reverse transcribed using a QuantiTect RT Kit (Qiagen). Quantitative PCR was performed using a QuantiTect SYBR Green kit (Qiagen). Relative hPRR, mouse PRR (mPRR), and mouse NOX2 and NOX4 mRNA levels, normalized to those of mouse GAPDH, were determined using the ΔΔC_T relative quantification method (where C_T is threshold cycle), where the relative expression ratio was calculated as  ${{2(}^{\mathrm{\Delta }{\text{C}}_{\text{T,Target}}-\mathrm{\Delta }{\text{C}}_{\text{T,GAPDH}}}}^{})$. The list of primers used is shown in Table 1.

Table 1.

Quantitative RT-PCR primers used

Primer Name	Forward Primer Sequence	Reverse Primer Sequence
Human PRR	5′-ACTCGCAGTGGGTAACCTGTTTCA-3′	5′-AACTCTACCACTGCATTCCCACCA-3′
Mouse PRR	5′-TCTCTCCGAACTGCAAGTGCTACA-3′	5′-CCAAACCTGCCAGCTCCAATGAAT-3′
Mouse NOX2	5′-CCCTTTGGTACAGCCAGTGAAGAT-3′	5′-CAATCCCGGGCTCCCACTAACATCA-3′
Mouse NOX4	5′-TGAACTACAGTGAGATTTCCTTGAAC-3′	5′-GACACCCGTCAGACCAGGAA-3′
Mouse β-actin	5′-CGTGAAAAGATGACCCAGATCA-3′	5′-TGGTACGACCAGAGGCATACAG-3′
Mouse GAPDH	5′-AATGTGTCCGTCGTGGATCTGA-3′	5′-GATGCCTGCTTCACCACCTTCT-3′

PRR, (pro)renin receptor; NOX, NADPH oxidase.

Western Blot Analysis

Tissue lysates from the brain cortex, hypothalamus, and brain stem were collected for SDS-PAGE and Western blot detection of hPRR, as previously described (44). Briefly, targeted tissues were harvested and homogenized in lysis buffer (Thermo Scientific, Waltham, MA) containing protease inhibitor cocktail (Sigma-Aldrich). The lysate was microcentrifuged at 13,000 rpm for 10 min, and the supernatant was transferred to a clean tube. Protein concentration was measured using a BCA assay kit (Thermo Scientific). Lysate samples containing equal amounts of protein (25 μg) were mixed with SDS-PAGE sample buffer (Invitrogen, Carlsbad, CA), heated at 70°C for 10 min, and electrophoresed on a 4–20% Tris-glycine gradient gel (catalog no. EC2026BOX, Invitrogen). After proteins had been transferred to nitrocellulose membranes using an iBlot Gel Transfer Device (Invitrogen), membranes were blocked by incubation with 5% nonfat milk in PBS containing 0.1% Tween 20 for 2 h at room temperature. Membranes were then incubated overnight at 4°C with primary antibody against rabbit anti-human PRR (37 kDa) diluted at 1:200 (Novus, Littleton, CO) and mouse anti-β-actin (42 kDa) diluted at 1:1,000 (Cell Signaling, Danvers, MA). After being washed, membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit and goat anti-mouse secondary antibodies for 1 h at room temperature. Membranes were then washed, and immunoreactive proteins were detected using SuperSignal ECL reagents (Invitrogen) and a UVP Bioimaging System.

For total and phosphorylated (p-)ERK measurements, the hypothalamus was collected from Syn-hPRR mice and NT littermates at the end of the 10-min intracerebroventricular infusion of either artificial cerebrospinal fluid (aCSF) or human prorenin. Lysates containing 25 µg protein, determined by BCA assay, were combined with 4× SDS buffer, heated to 70°C for 10 min, and resolved by SDS-PAGE on a 4–20% Tris-glycine gel. After proteins had been transferred to a nylon membrane and blocked with 5% BSA, membranes were incubated with rabbit anti-total ERK1/2 (1:1,000, Cell Signaling) or rabbit anti-p-ERK1/2 (1:2,000, Cell Signaling) primary antibodies overnight and then with horseradish peroxidase-conjugated goat anti-rabbit antibody (1:10,000) for 2 h at room temperature on a rocker. Membranes were then washed and imaged using SuperSignal ECL reagents (Invitrogen). p-ERK1/2 levels were normalized to those of total ERK1/2 protein.

Immunofluorescence Labeling

Mouse brains were collected and immediately fixed in 10% formaldehyde. Brains were embedded in paraffin and cut into 5-μm-thick sections with a microtome. Antigen retrieval was performed by placing slides in citrate buffer (3 g sodium citrate in 1 liter distilled water and 150 μl HCl, pH 6) and steaming for 45 min. After lid removal and cooling for 15 min, sections were rinsed with PBS and blocked by incubation in PBS + 0.3% Triton X-100 + 10% goat serum for 60 min at room temperature. Sections were then incubated overnight at 4°C with rabbit anti-PRR antibody (1:400) generated in our laboratory (35) or rabbit IgG as a negative control. Slides were washed three times and then incubated with Alexa 594-conjugated goat anti-rabbit antibody (1:1,000, Invitrogen) for 2 h at room temperature. Slides were washed, mounted, and then imaged with a confocal microscope. It should be noted that this antibody recognizes both mouse and human PRR, as previously characterized by our laboratory (35, 36, 44).

Lucigenin Luminescence Assay of NOX Activity

Animals were euthanized at baseline or immediately after a 10-min intracerebroventricular infusion (0.3 μl/min) of aCSF, human prorenin (100 ng/μl), or human prorenin plus the MEK1/2 inhibitor U-0126 (3 pmol). Fresh tissues were homogenized in homogenization buffer (20 mM HEPES and 1 mM EDTA) containing protease inhibitor cocktail (Sigma). The lysate was microcentrifuged at 4,000 rpm for 15 min at 4°C, and the supernatant was transferred to a clean tube. Protein concentration was measured using a BCA assay kit (Pierce, Rockford, IL). NOX activity was measured at 37°C using a lucigenin-based luminescence assay in a mixture containing 100 µg protein/well, 200 µM NAD(P)H, and 10 µM lucigenin in 200 µl reaction buffer [containing (in mM) 20 HEPES (pH 7.4), 9.9 NaCl, 4.7 KCl, 1.2 MgSO₄, 2 KH₂PO₄, 1.9 CaCl₂, 25 NaHCO₃, and 11.1 glucose]. Luminescence was measured using a microplate luminometer (Synergy H4, BioTek). In each sample, NOX activity was also measured in the presence of the flavoprotein inhibitor diphenyleneiodonium (DPI; 10 µM). Chemiluminescence readings were obtained at 60-s intervals for an overall measurement time of 60 min, a period over which maximal chemiluminescence was achieved. The final luminescence reading was obtained by subtracting the total intensity from the intensity in the presence of DPI in each sample to ensure that the measured activity was attributable to NOX. A buffer blank (<2% of the homogenate signal) was subtracted from each reading. Relative luminescence units (RLUs) were normalized to controls (44).

ANG II Measurement Using Liquid Chromatography Tandem-Mass Spectrometry

Syn-hPRR and their NT littermates were euthanized immediately after the 10-min intracerebroventricular infusion (0.3 μl/min) or either aCSF or human prorenin (100 ng/μl), and the hypothalamus was isolated by blunt dissection, snap frozen in liquid nitrogen, and stored at −80°C until analysis. Frozen hypothalamic tissues were sent to Attoquant Diagnostics (Vienna, Austria) for liquid chromatography-tandem mass spectrometry analysis for ANG II levels as previously described (46, 63).

Physiological Recordings

Implantation of telemetry probes.

Syn-hPRR mice and their NT littermates were initially anesthetized using 4–5% isoflurane in 100% O₂ and flushed at 1 l/min for 90 s, after which anesthesia was maintained by adjusting isoflurane to 0.75–1.5%. The neck was shaved and then sterilized with alcohol swabs. An incision (~1 cm) was made to separate the oblique muscle and tracheal muscle and expose the right carotid artery. The catheter of a radiotelemetry transmitter (PA-C10, Data Science) was surgically implanted in the right carotid artery and secured. The body of the transmitter was embedded in a subcutaneous skin pocket under the right arm. After a 14-day recovery period, baseline BP, heart rate (HR), and locomotor activity were continuously recorded in conscious mice for 48 h (33–35).

Intracerebroventricular cannulation.

Mice were initially anesthetized using 4–5% isoflurane in 100% O₂ and flushed at 1 l/min for 90 s; anesthesia was subsequently maintained by adjusting isoflurane to 0.75–1.5%. The hair on the top of the head was shaved, and mice were placed on a stereotaxic apparatus by adjusting ear bars to fit just above the ear canal. The top of the skull was cleaned with cotton buds, and 3% H₂O₂ was then applied. A hole was drilled, and the cannula was inserted using a digital stereotaxic apparatus (Stoelting, Wood Dale, IL) with intracerebroventricular coordinates (0.3 mm posterior to the bregma, 1.0 mm to the right or left of the midline, 3.2 mm from the top of the brain) and mounted on the clean dry skull using Super Glue Ultra Gel Control Adhesive (Loctite, Westlake, OH). Cannulas were prepared by cutting 9-mm segments of 25-gauge needles (BD Medical, Franklin Lakes, NJ) and blunting the ends. A plastic mounting disk from a brain infusion kit (catalog no. 8851, Alzet, Cupertino, CA) was prepared to fit each of the 25-gauge cannulas to be mounted on the skull. Cannulas were sterilized in 75% alcohol overnight before use.

Intracerebroventricular infusion and BP recording in conscious freely moving mice.

Each mouse was housed in a single cage on top of a telemetry receiver platform; the telemetry transmitter was turned on, and baseline BP was recorded for at least 30 min before intracerebroventricular infusion. An inner infusion cannula prepared from 32-gauge tubing (Microgroup, Medway, MA) was connected to an ~30-cm-long P-10 catheter. With the use of a NE-1002X syringe pump (New Era Pump Systems), mice were intracerebroventricularly infused (0.3 µl/min) for 10 min with the following reagents/reagent combinations: aCSF, human prorenin (300 ng), human prorenin + losartan (30 pmol), human prorenin + captopril (30 pmol), human prorenin + PRO20 (1, 3, 10, 30, 100, 300 µM), ANG II (300 ng), ANG II + losartan (30 pmol), carbochol (300 ng), human prorenin + DPI (30, 300 pmol), human prorenin + U-0126 (3 pmol), human prorenin + GCD-0994 (0.3 pmol, MedChem Express, Monmouth Junction, NJ), or human prorenin + polyethylene glycol-linked catalase (PEG-catalase; 0.67 U/µl). BP and HR were continuously recorded during infusion and a 30- to 60-min postinfusion period (37) using the telemetry system. Data are presented as absolute BP and HR traces or changes in mean arterial blood pressure (ΔMAP).

NOX2 or NOX4 knockdown using adenovirus-delivered shRNA.

Adenoviruses (Ad) expressing shRNA targeting NOX2 (Ad-NOX2-shRNA), NOX4 (Ad-NOX4-shRNA), or scrambled shRNA (Ad-sc-shRNA), developed and characterized by Dr. Robin Davisson (Cornell University) (45), were purchased from the University of Iowa Carver College of Medicine Viral Vector Core Facility. For in vivo knockdown of NOX2 or NOX4 in Syn-hPRR mice, Ad-shRNAs were delivered by intracerebroventricular injection (1 × 10⁸ plaque-forming units in 100 nl) using a stereotaxic apparatus (coordinates: 0.3 mm posterior to the bregma, 1.0 mm to the right of the midline, and 3.3 ventrally) (35) through intracerebroventricular cannulas. Adenovirus knockdown efficiency in the hypothalamic tissues was validated by real-time quantitative PCR for NOX2 and NOX4 5 days after intracerebroventricular injection. Baseline BP was recorded 5 days after virus injection in telemetry transmitter-implanted mice that received Ad-sc-shRNA, Ad-NOX2-shRNA, or Ad-NOX4-shRNA. A schematic depiction of the protocol is shown in Fig. 7C. Mice were intracerebroventricularly infused with human prorenin, as described above in the intracerebroventricular cannulation and infusion protocol.

Fig. 7.

NADPH oxidase (NOX)4 knockdown prevents the human prorenin-induced blood pressure (BP) elevation in mice overexpressing human (pro)renin receptor specifically in neurons (Syn-hPRR). A: relative NOX2 mRNA levels normalized to adenoviruse (Ad)-scrambled-shRNA (Ad-sc-shRNA) in hypothalamic tissues of Syn-hPRR mice (n = 3–5/group). B: relative NOX4 mRNA levels normalized to Ad-sc-shRNA in hypothalamic tissues of Syn-hPRR mice (n = 3–5/group). C: schematic showing the time frame of intracerebroventricular administration for the adenovirus protocol. D: change in mean arterial pressure (ΔMAP) over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of human prorenin (300 ng) in Syn-hPRR mice previously administered Ad-sc-shRNA (n = 3), Ad-NOX2-shRNA (n = 4), Ad-NOX4-shRNA (n = 4), or Ad-NOX2-shRNA + Ad-NOX4-shRNA (n = 4). E: ΔMAP over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of polyethylene glycol (PEG)-catalase (2.01 units) with or without coinfusion of human prorenin in Syn-hPRR mice (n = 3/group). *P ≤ 0.05 vs. Ad-sc-shRNA (A and B) vs. Ad-sc-shRNA and Ad-NOX2-shRNA (D) or vs. PEG-catalase (E); #P ≤ 0.05 vs. human prorenin. All BP recordings were obtained by telemetry from nonanesthetized, conscious, freely moving mice.

Statistical Analysis

Data are expressed as means ± SE and were analyzed by one-way (one variant) or two-way (two variants) ANOVA for multiple groups followed by a Tukey post hoc test to compare replicate means as appropriate. Statistical comparisons were performed using Prism7 software (GraphPad, La Jolla, CA). P values of ≤0.05 were considered statistically significant. StatMate 2 (GraphPad) were used for power analysis, and a power of 95% was used to determine the sample size.

RESULTS

Development and Characterization of Syn-hPRR Mice

We generated eight founder mice expressing the hPRR transgene under the control of the neuron-specific rat synapsin 1 promoter (Syn-hPRR; Fig. 1A). All founders were successfully bred to establish transgenic lines. Among these founders (Fig. 1B), line 57994-1 exhibited high levels of hPRR mRNA in the brain but not in other tissues, including the heart, kidneys, lung, liver, spleen, muscles, blood vessels, white adipose tissue, and pancreas (Fig. 1C). To determine in this line whether Syn-hPRR transgene mRNA expression was reflected at the protein level, we analyzed lysates of brains from Syn-hPRR and wild-type (NT) littermate control mice by Western blot analysis using a selective anti-hPRR antibody (Fig. 1D). hPRR protein was detected in the cerebral cortex, hypothalamus, and brain stem of Syn-hPRR mice but was not found in NT littermates. Real-time quantitative RT-PCR confirmed that hPRR mRNA was abundant in Syn-hPRR mice and absent from NT littermates (Fig. 1E), whereas endogenous mPRR mRNA levels were similar between Syn-hPRR and NT mice (Fig. 1F). Immunolabeling of Syn-hPRR mouse brain sections showed that hPRR was expressed throughout the brain but was notable for its high expression levels in many cardiovascular control regions, including the subfornical organ, paraventricular nucleus, nucleus of the solitary tract, area postrema, and rostral ventrolateral medulla (Fig. 1G). The anti-PRR antibody used for immunolabeling reacts with both mPRR and hPRR; however, the fluorescence intensity was much higher in Syn-hPRR mice than in NT littermates.

Human Prorenin Induces an ANG II-Independent Elevation in BP in Syn-hPRR Mice

Transgenic overexpression of hPRR in neurons did not alter BP, HR, or cardiac circadian rhythms in mice (Fig. 2A and B) and had no effect on locomotor activity (Fig. 2C). As shown in a representative tracing in Fig. 3, baseline BP and HR before intracerebroventricular infusion were similar in all groups of mice, whereas intracerebroventricular infusion of human prorenin induced an elevation of BP in Syn-hPRR mice that persisted for ~45–60 min but had no effect on HR. Summarized graph data (Fig. 4) showed that intracerebroventricular infusion of human prorenin rapidly increased MAP (ΔMAP: 24.7 ± 2.4 mmHg) compared with intracerebroventricularly infused aCSF in Syn-hPRR mice (ΔMAP: 0.5 ± 2.6 mmHg, P = 0.0002) and NT littermates (ΔMAP: −1.06 ± 1.12 mmHg, P < 0.0001). Although BP trended lower after coinfusion of prorenin and the ANG II type 1 receptor (AT₁R) antagonist losartan (ΔMAP: 18.7 ± 5.2 mmHg, P = 0.380) or angiotensin converting enzyme inhibitor captopril (ΔMAP: 19.0 ± 1.1 mmHg, P = 0.441) compared with intracerebroventricularly infused prorenin alone, the differences did not reach statistical significance. More importantly, the increase in BP induced by human prorenin remained significantly elevated in the presence of either losartan (P = 0.0011) or captopril (P = 0.0009) compared with aCSF infusion in Syn-hPRR mice. These data suggest that human prorenin mediates BP elevation largely independent of ANG II/AT₁R signaling in Syn-hPRR mice. Intracerebroventricular infusion of aCSF or human prorenin, with or without losartan or captopril, did not alter BP in NT mice. Furthermore, there was no significant difference in the levels of ANG II among the hypothalamus of either NT or Syn-hPRR mice after intracerebroventricular infusion of either aCSF or human prorenin (Fig. 4C). To further support our finding that the specific BP effect is mediated by direct prorenin/PRR signaling, we intracerebroventricularly infused different doses of PRO20, a peptide PRR antagonist (37), together with human prorenin in Syn-hPRR mice. We showed that PRO20 caused a dose-dependent attenuation of the human prorenin-induced BP response in Syn-hPRR mice, with an IC₅₀ value of 7.2 μM (Fig. 4, D and E).

Fig. 2.

Blood pressure, heart rate (HR), and locomotor activity in mice overexpressing the human (pro)renin receptor specifically in neurons (Syn-hPRR) and nontransgenic (NT) littermates. A–C: summary graphs of mean arterial pressure (MAP; A), HR (B), and locomotor activity (C) in Syn-hPRR mice and NT littermates over a 48-h period as measured by radiotelemetry (n = 4/group). Gray, nighttime period; white, daytime period.

Fig. 3.

Representative traces of the blood pressure (BP) and heart rate (HR) response to artificial cerebrospinal fluid (aCSF) and human (h) prorenin in nontransgenice (NT) littermates and mice overexpressing human (pro)renin receptor (hPRR) specifically in neurons (Syn-hPRR). A and C: representative traces of BP and HR in a NT mouse or Syn-hPRR mouse during intracerebroventricular infusion of aCSF. B and D: representative traces of BP and HR in a NT mouse or Syn-hPRR mouse during intracerebroventricular infusion of human prorenin (hProrenin).

Fig. 4.

Human prorenin induces an elevation of blood pressure (BP) in mice overexpressing human (pro)renin receptor (hPRR) specifically in neurons (Syn-hPRR). A: changes in mean arterial pressure (ΔMAP) over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of artificial cerebrospinal fluid (aCSF; n = 4) or human prorenin (300 ng, n = 9) with or without intracerebroventricular coinfusion of losartan (30 pmol, n = 6) or captopril (30 pmol, n = 3) in nontransgenic (NT) mice. B: ΔMAP over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of aCSF (n = 4) or human prorenin (n = 14) with or without intracerebroventricular coinfusion of losartan (30 pmol, n = 8) or captopril (n = 8) in Syn-hPRR mice. C: ANG II levels in the hypothalamus after intracerebroventricular infusion of either aCSF or human prorenin in NT or Syn-hPRR mice (n = 4/group). D: ΔMAP over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of human prorenin (300 ng) with different doses of PRO20 (1, 3, 10, 30, 100, and 300 μM) in Syn-hPRR mice (n = 4/group). E: dose-inhibition curve presented as a percentage of the maximal response. B: *P ≤ 0.05 vs. Syn-hPRR + aCSF; all BP recordings were obtained by telemetry from nonanesthetized, conscious, freely moving mice.

To examine whether there is an alteration of AT₁R activity in Syn-hPRR mice and further confirm that the doses of losartan used in our experiments were sufficient, we intracerebroventricularly infused ANG II into both NT and Syn-hPRR mice with or without coinfusion of losartan. The elevation of BP was similar between NT and Syn-hPRR mice after intracerebroventricular ANG II infusion (Fig. 5A). Intracerebroventricular coinfusion of losartan completely blocked the ANG II-induced BP elevation in both groups of mice, indicating that there was no alternation of AT₁R activity in BP regulation in mouse transgenic expression of human PRR. Intracerebroventricular infusion of carbachol, a cholinergic agonist, was also performed to examine whether there was an altered response to other stimulus with transgenic expression of hPRR. We observed no difference in BP responses to intracerebroventricular carbachol between NT and Syn-hPRR mice (Fig. 5B).

Fig. 5.

Mice overexpressing human (pro)renin receptor specifically in neurons (Syn-hPRR) have a normal blood pressure (BP) response to ANG II or carbachol. A: change in mean arterial pressure (ΔMAP) over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of ANG II (300 ng, n = 3–5) with or without intracerebroventricular coinfusion of losartan (30 pmol, n = 5) in nontransgenic (NT) and Syn-hPRR mice. B: ΔMAP over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of artificial cerebrospinal fluid (aCSF; n = 4) or carbochol (300 ng, n = 3–4) in NT and Syn-hPRR mice. *P ≤ 0.05 vs. NT + ANG II or Syn-hPRR + ANG II (A) or vs. NT + aCSF or Syn-hPRR + aCSF (B).

Human Prorenin Induces an Increase in NOX Activity in Syn-hPRR Mice

The NOX family constitutes a major category of enzymes that generate ROS, including superoxide and H₂O₂, and it has been shown that members of this family are important for BP regulation in the CNS (27). To determine whether NOX is involved in prorenin-induced hypertension, we measured NOX activity using a lucigenin-based luminescence assay. NOX activity in the hypothalamus, expressed in RLUs, was similar in Syn-hPRR mice (RLU fold change: 1.25 ± 0.14) and NT littermates (RLU fold change: 1.02 ± 0.05, P = 0.407) after intracerebroventricular infusion of aCSF (control; Fig. 6A). Whereas intracerebroventricular infusion of human prorenin did not alter NOX activity in NT mice compared with aCSF infusion (RLU fold change: 1.16 ± 0.06, P = 0.705), it significantly increased NOX activity in Syn-hPRR mice (RLU fold change: 1.54 ± 0.07, P = 0.007; Fig. 6A). Moreover, the increase in NOX activity was not reduced by the cointracerebroventricular infusion of losartan with human prorenin (RLU fold change: 1.38 ± 0.05, P = 0.227) compared with intracerebroventricular infusion of human prorenin alone in Syn-hPRR mice. It remained significantly higher than that of NT mice (P = 0.019). To determine the functional importance of NOX in human prorenin-induced elevated BP, we intracerebroventricularly infused the flavoenzyme inhibitor DPI to inhibit NOX activity and recorded BP using radiotelemetry in conscious freely moving mice. The elevation in BP induced by human prorenin (ΔMAP: 22.0 ± 5.9 mmHg) was dose dependently attenuated by intracerebroventricular coinfusion of DPI, which reduced this increase by ~35% (ΔMAP: 14.4 ± 3.7 mmHg, P = 0.045) and ~79% (4.7 ± 1.0 mmHg, P < 0.0001) at 10 and 100 μM, respectively; in contrast, intracerebroventricular infusion of human prorenin and/or DPI had no effect on BP in NT littermates (Fig. 6B). Taken together, these data indicate that the human prorenin-induced BP response in Syn-hPRR mice requires flavoenzyme activity, possibly that of NOX.

Fig. 6.

Increased NADPH oxidase (NOX) activity contributes to human prorenin-induced elevated blood pressure (BP) in mice overexpressing human (pro)renin receptor specifically in neurons (Syn-hPRR). A: hypothalamic NOX activity in Syn-hPRR mice and nontransgenic (NT) littermates infused intracerebroventricular (0.3 µl/min) with artificial cerebrospinal fluid (aCSF), human prorenin (300 ng, n = 4–8/group), or human protenin (300 ng) + losartan (30 pmol) (n = 3/group) measured using a lucigenin-based luminescence assay and expressed as fold increases in relative luminescence units (RLUs). B: change in mean arterial pressure (ΔMAP) over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of human prorenin with or without diphenyleneiodonium (DPI; 30 or 300 pmol) in Syn-hPRR mice and NT littermates (n = 3–4/group). *P ≤ 0.05 vs. NT with the same treatment (A) or Syn-hPRR + human prorenin vs. NT + human prorenin (B); #P ≤ 0.05, Syn-hPRR + human prorenin + DPI (300 pmol) vs. Syn-hPRR + human prorenin. All BP recordings were obtained by telemetry from nonanesthetized, conscious, freely moving mice.

NOX4 Knockdown Prevents the Human Prorenin-Induced BP Elevation in Syn-hPRR Mice

To examine the physiological significance of this flavoenzyme requirement and determine which NOX isoforms mediate the direct PRR signaling in central BP regulation, we knocked down NOX4 or NOX2 using Ad-mediated delivery of shRNA and measured prorenin-induced changes in BP in Syn-hPRR mice. Ad-NOX4-shRNA, Ad-NOX2-shRNA, or Ad-sc-shRNA were delivered via intracerebroventricular injection in Syn-hPRR mice, and the knockdown efficacy of these shRNAs was validated by real-time quantitative RT-PCR (Fig. 7, A and B). None of these adenoviral constructs altered baseline BP in Syn-hPRR mice (data not shown). Notably, Ad-NOX4-shRNA significantly reduced the intracerebroventricular prorenin-induced elevation in BP (ΔMAP: 6.5 ± 2.8 mmHg, P < 0.0001) compared with either Ad-NOX2-shRNA (ΔMAP: 20.2 ± 5.2 mmHg) or Ad-sc-shRNA (ΔMAP: 22.9 ± 3.2 mmHg; Fig. 7D). In contrast, knockdown of NOX2 did not prevent the human prorenin-induced elevation in BP compared with Ad-sc-shRNA in Syn-hPRR mice (P = 0.99). Collectively, these observations suggest that NOX4 is responsible for the human prorenin-induced increase in BP. Importantly, administration of Ad-NOX4-shRNA to mice that had previously been administered Ad-NOX2-shRNA, as shown in Fig. 7C, significantly attenuated the prorenin-induced BP elevation (ΔMAP: 3.3 ± 2.4 mmHg, P < 0.0001; Fig. 7D) compared with either Ad-NOX2-shRNA or Ad-sc-shRNA.

Previous studies have suggested that NOX4 produces predominantly H₂O₂ (16, 54, 61). To assess the possible functional role of NOX4-generated H₂O₂ in this event, we intracerebroventricularly coinfused human prorenin and cell-permeable PEG-catalase to catalyze the decomposition of H₂O₂ to water. Interestingly, we found that the elevation of BP (Fig. 7E) induced by intracerebroventricular infusion of human prorenin alone (ΔMAP: 18.1 ± 1.8 mmHg) was significantly reduced by intracerebroventricular coinfusion of human prorenin and PEG-catalase (ΔMAP: 1.5 ± 0.5 mmHg, P < 0.0001). In comparison, intracerebroventricular infusion of PEG-catalase alone (ΔMAP: 4.2 ± 0.5 mmHg) did not affect BP in Syn-hPRR mice, which exhibited a baseline BP similar to that observed before intracerebroventricular infusion (Fig. 7E). These data indicate that NOX4 activity, possibly through production of H₂O₂, plays a significant role in the regulation of BP during direct activation of human prorenin/PRR signaling.

Involvement of ERK Phosphorylation in the Prorenin-Induced Elevation in NOX Activity and BP in Syn-hPRR Mice

The goal of these experiments was to examine whether the ERK signaling pathway is involved in the elevated NOX activity and BP induced by prorenin in vivo, particularly in ANG II-independent hypertension. To this end, we first examined changes in the level of activatedERK (p-ERK) in response to intracerebroventricular infusion of human prorenin in hypothalamic tissue. p-ERK levels after intracerebroventricular infusion of aCSF (control conditions) in Syn-hPRR mice (fold change: 1.79 ± 0.47) tended to be higher but were not significantly different from those in NT littermates (fold change: 1.01 ± 0.01, P = 0.404; Fig. 5, A and B). Notably, intracerebroventricular infusion of human prorenin significantly increased p-ERK levels (fold change: 3.37 ± 0.49, P = 0.032) compared with aCSF infusion in Syn-hPRR transgenic mice (Fig. 8, A and B) but not in NT littermates (fold change: 1.12 ± 0.12, P = 0.995). To determine whether ERK signaling plays a role in the prorenin/PRR-mediated elevation of BP, we intracerebroventricularly coinfused conscious freely moving Syn-hPRR mice with U-0126 (15), an inhibitor of MEK1/2 (a major mediator of ERK1/2 phosphorylation), or GCD-0994 (22, 48), a new-generation specific ERK1/2 inhibitor, together with human prorenin, and recorded BP by telemetry. Intracerebroventricular coinfusion of either U-0126 (ΔMAP: 5.89 ± 0.4 mmHg) or GCD-0994 (ΔMAP: 7.84 ± 5.3 mmHg) significantly attenuated the increase in BP induced by human prorenin in Syn-hPRR mice compared with prorenin infusion alone (21.7 ± 7.2 mmHg, P = 0.001; Fig. 8C). To further determine whether this ERK signaling is associated with NOX activation, we measured NOX activity after intracerebroventricular infusion of U-0126. We found that the increase in NOX activity induced by intracerebroventricular infusion of human prorenin (RLU fold change: 1.62 ± 0.1, P = 0.0008) was significantly attenuated by intracerebroventricular coinfusion of U-0126 (RLU fold change: 1.29 ± 0.1) in Syn-hPRR mice (Fig. 8D). However, U-0126 did not completely normalize NOX activity in Syn-hPRR mice compared with NT littermates administered aCSF (RLU fold change: 0.98 ± 0.1, P = 0.0085), suggesting that ERK activation partly accounts for the increase in NOX activity induced by human prorenin.

Fig. 8.

ERK phosphorylation mediates the prorenin-induced elevation of NADPH oxidase (NOX) activity and blood pressure (BP) in mice overexpressing human (pro)renin receptor specifically in neurons (Syn-hPRR). A and B: representative Western blots and summary data showing the ratio of phosphorylated (p-)ERK to total (t) ERK protein in Syn-hPRR mice and nontransgenic (NT) littermates (n = 4/group) infused intracerebroventricularly (0.3 µl/min) with artificial cerebrospinal fluid (aCSF) or human prorenin (hPro; 300 ng) for 10 min. C: change in mean arterial pressure (ΔMAP) over the course of a 10-min intracerebroventricular infusion (0.3 µl/min) of human prorenin with or without coinfusion of U-0126 (3 pmol) or GDC-0994 (0.3 pmol) in Syn-hPRR mice and NT littermates (n = 3–4/group). D: hypothalamic NOX activity, measured using a lucigenin luminescence assay, indicating the fold increase in relative light units (RLUs) for NT mice + intracerebroventricular aCSF (control) and Syn-hPRR mice infused intracerebroventricularly with human prorenin (300 ng) or coinfused with human prorenin and the ERK inhibitor U-0126 (3 pmol, n = 4). B: *P ≤ 0.05 vs. Syn-hPRR + aCSF; #P ≤ 0.05 vs. NT + human prorenin. C: *P ≤ 0.05 vs. NT + human prorenin; #P ≤ 0.05 vs. Syn-hPRR + human prorenin. D: *P ≤ 0.05 vs. NT + aCSF; #P ≤ 0.05 vs. Syn-hPRR + human prorenin.

DISCUSSION

Local renin-angiotensin systems (RASs) are present within various tissues and have important physiological and pathophysiological roles in cardiovascular regulation (5, 43). We have previously reported that PRR overexpression mediates both ANG II-dependent and -independent ROS formation and NOX activation in cultured murine neurons (44). Evidence for PRR-mediated ANG II-independent signaling in the CNS has been previously reported (50); however, whether ANG II-independent PRR signaling in the CNS is involved in BP regulation has remained unknown. Our study provides new insights that fill this knowledge gap. The major findings of this study can be summarized as follows:1) a new transgenic mouse model that overexpresses hPRR specifically in neurons has been developed and characterized; 2) human prorenin activates hPRR and mediates the ANG II-independent BP elevation in Syn-hPRR transgenic mice; 3) NOX activity, particularly NOX4, may be responsible for the BP elevation induced by prorenin; and 5) ERK1/2 activation downstream of prorenin/PRR signaling is a key pathway underlying increased NOX activity, and thus ANG II-independent hypertension, in Syn-hPRR transgenic mice.

In addition to increasing the catalytic efficiency of renin and activating prorenin, the binding of prorenin or renin to PRR also directly induces intracellular signaling (17, 24). Transgenic rats overexpressing hPRR in vascular smooth muscle cells exhibit small increases in BP after 6 mo of age, possibly related to the elevation of aldosterone (6). However, a separate report has shown that rats with ubiquitous transgenic expression of the hPRR have normal BP but exhibit kidney damage, as evidenced by proteinuria and glomerulosclerosis (28). The discrepancy in the phenotypes of these two models has yet to be explained; however, it is evident that ANG II-independent PRR effects are involved in the pathology in both animal models (28). Patients develop resistant hypertension to RAS inhibitors, such as losartan and captopril, during long-term antihypertensive treatment (53) through mechanisms that are not fully understood (20, 53). Evidence for elevation of hypothalamic prorenin and/or PRR levels has been shown previously in hypertensive animals (36). We speculate that, in some forms of human resistant hypertension with increased sympathetic activity, an elevation in hypothalamic prorenin and PRR and activation of prorenin/PRR and downstream ANG II-independent signaling pathways may represent one of the underlying mechanisms that leads to this resistance. To explore the possible involvement of this mechanism in the central regulation of BP, we developed a new transgenic mouse model overexpressing hPRR in neurons, the major cell type that expresses PRR in adult rodents and humans (36, 51, 62). The enzymatic reaction between renin/prorenin and angiotensinogen is highly species specific, and human renin/prorenin is unable to cleave mouse angiotensinogen to generate angiotensin peptides (56, 57, 66). Taking advantage of this unique feature, we used human prorenin to stimulate overexpressed hPRR in vivo and thereby isolate the ANG II-independent component of human prorenin/PRR-mediated BP regulation. We showed that activation of hPRR by human prorenin in the CNS results in an elevation in BP that is largely independent of ANG II, a conclusion confirmed by the observation that BP remained significantly elevated in Syn-hPRR mice that received a coinfusion of losartan or captopril with human prorenin compared with control infusion of aCSF. We observed a tendency of losartan or captopril to attenuate BP induced by human prorenin in Syn-hPRR mice, indicating a potential minor effect of ANG II/AT₁R in this model. Consistent with this, the PRR-specific antagonist PRO20 (37) significantly blocked the human prorenin-induced elevation in BP in Syn-hPRR mice in a dose-dependent manner. We were not surprised that intracerebroventricular infusion of human prorenin did not alter BP in conscious freely moving NT mice. Although multiple in vitro studies have indicated that human prorenin acts on the rat PRR and induces activation of downstream signaling pathways (25, 52, 69), an in vivo study (69) showed that the injection of human renin in the nucleus tractus solitarius did not alter BP in normotensive nonanesthetized rats. A following study (25) published by the same group reported that administration of human prorenin in the paraventricular nucleus of the hypothalamus increased sympathetic activity in anesthetized Sprague-Dawley rats; however, it did not alter BP. The authors speculated that the absence of changes in BP might be because of anesthesia. However, to our knowledge, no followup studies have been conducted to date to reinforce this interpretation. Our finding that human prorenin has no effect on BP in NT normotensive animals is consistent with some of these previous reports.

An overwhelming body of evidence accumulated from the past several decades shows the existence of endogenous brain RAS activity and their importance in the regulation of autonomic function (11, 42, 55). However, a major question that remains is whether the source of brain prorenin and endogenous levels of prorenin are high enough to have meaningful interaction with PRR. Evidence for the expression of prorenin endogenously in the brain has been elegantly reported (32, 35, 58). In the circulation, there is a significant amount of prorenin, although the majority is not active without association with PRR (8). In fact, the level of prorenin is ~10-fold higher than that of the active renin in healthy individuals (10, 41) but can be as much as 100-fold higher when active renin is suppressed (9). Because of the existence of circumventricular organs in the brain, possible active transport mechanisms, and the leakage of the blood-brain barrier during pathophysiological conditions, we postulate that both endogenous and circulating prorenin may be the source of the brain prorenin. Future studies are warranted to test this hypothesis. With the use of a transgenic mouse model, the present study provides functional evidence for the role of human prorenin/PRR activation in acute neural regulation of BP in an ANG II-independent manner. Further investigations to understand this mechanism in chronic hypertension are needed.

NOX2 and NOX4 are two major NOX isoforms in the CNS (45, 65). Previous studies have elegantly described the mechanisms by which NOX2 and NOX4 are activated in neurons and have demonstrated their importance in the neural regulation of BP (67, 68). However, the links between prorenin/PRR signaling and NOX activation in the regulation of BP, and the identity of the NOX isoforms involved, have remained elusive. Using Ad-mediated delivery of NOX2 and/or NOX4 shRNA, we found that NOX4 activation is responsible for the elevation in BP induced by human prorenin in Syn-hPRR mice. Unlike other NOX isoforms, NOX4 is usually located on intracellular membranes, such as those of mitochondria or the endoplasmic reticulum (2, 30). Controversial reports in earlier studies indicated that NOX4 produces both H₂O₂ and superoxide in cardiomyocytes, vascular endothelial cells, and neurons (3, 12, 31), but more recent studies have suggested that NOX4 produces primarily H₂O₂ (4, 40). We found that intracerebroventricular infusion of the H₂O₂ scavenger catalase (conjugated to PEG to facilitate intracellular access) attenuated the human prorenin-induced elevation in BP in Syn-hPRR transgenic mice, suggesting a possible role for NOX4 in this event. Our findings are consistent with previous in vitro studies that reported that prorenin activates NOX4 in cultured human embryonic kidney cells (13) and kidney collecting duct cells (38) through binding to PRR. Notably, baseline BP in our study was not altered by hPRR overexpression, NOX2, and/or NOX4 knockdown, or intracerebroventricular infusion of PEG-catalase, DPI, or RAS blockers (losartan and captopril). This is not surprising given the many previous reports showing that overexpression (18, 19) or blockade/knockdown (1, 29, 37) of RAS components or redox signaling pathway molecules, including NOX2 and/or NOX4 (45), specifically in the CNS, does not alter baseline BP in normotensive animals. These observations suggest that these signaling pathways are not the primary mechanisms by which BP is regulated in the CNS under normotensive conditions or, alternatively, that this pathway may be buffered by redundant BP regulatory mechanisms.

ERK phosphorylation is one of many important signaling events that have been established downstream of renin/prorenin PRR signaling (64), and evidence for activation of NOX by MAPK/ERK has been reported by several laboratories (44, 47). Here, we showed that hypothalamic ERK phosphorylation is important for intracerebroventricular prorenin-induced elevated BP in Syn-hPRR mice. We further found that prorenin-induced NOX activation in Syn-hPRR mice is attenuated by the MEK1/2 inhibitors, indicating that ERK may act, at least in part, through downstream NOX. One limitation of this study is that the synapsin promoter is a general neuronal promoter; thus, the contribution of hPRR to BP could possibly come from many brain regions. Examining the specific brain nucleus that is responsible for the ANG II-independent PRR action is our future study of interest. In addition, intracerebroventricularly infused proteins or chemical reagents diffuse throughout the brain, achieving higher levels in hypothalamic brain regions and lower levels in the cerebral cortex and medulla (36). We have focused our experiments on the hypothalamus of the brain and limited our interpretations to this region. However, we acknowledge that prorenin/PRR-mediated signaling events also occur in other brain regions, including the brain stem cardiovascular regulatory center (51).

In summary, we have developed a transgenic mouse model overexpressing hPRR. We found that human prorenin, acting via hPRR, mediates an ANG II-independent increase in BP through a mechanism involving ERK phosphorylation and NOX4 activation in the CNS. We conclude that activation of hPRR signaling may be among the pathways responsible for BP regulation.

GRANTS

This work was supported in part by National Heart, Lung, and Blood Institute Grants R01-HL-122770 (to Y. Feng), R01-HL-084207 (to C. D. Sigmund), and R01-HL-091985 (to S. Earley) and American Heart Association National Center Grants 11SDG4880005 and 17IRG33370128 (to Y. Feng).

DISCLOSURES

The content is solely the responsibility of the authors and does not necessarily represent the official views of the granting agencies. No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.P., D.D.J., W.L., and Y.F. conceived and designed research; H.P., D.D.J., W.L., M.N.S., S.A.B., C.J.W., S.G.C., S.Z., and Y.F. performed experiments; H.P., D.D.J., W.L., M.N.S., S.A.B., C.J.W., S.G.C., S.Z., and Y.F. analyzed data; H.P., D.D.J., W.L., M.N.S., S.A.B., S.G.C., S.Z., S.E., C.D.S., and Y.F. interpreted results of experiments; H.P., D.D.J., M.N.S., S.A.B., C.J.W., S.G.C., and Y.F. prepared figures; H.P., D.D.J., and W.L. drafted manuscript; H.P., D.D.J., W.L., M.N.S., S.A.B., C.J.W., S.G.C., S.Z., S.E., C.D.S., and Y.F. edited and revised manuscript; H.P., D.D.J., W.L., M.N.S., S.A.B., C.J.W., S.G.C., S.Z., S.E., C.D.S., and Y.F. approved final version of manuscript.