Expression and function of β-site amyloid precursor protein cleaving enzyme 2 in vascular endothelium

Abstract

The physiological function of β-site amyloid precursor protein-cleaving enzyme 2 (BACE2) in vascular endothelium of systemic arteries is unknown. In the present study we generated conditional tamoxifen-inducible endothelial BACE2 deficient mice (eBACE2^−/− mice). Electron-microscopic and western blot analyses revealed that BACE2 protein is mainly present in endothelial cells of aorta. Genetic deletion of BACE2 in endothelial cells significantly impaired endothelium-dependent relaxations to Ca²⁺-ionophore A23187 in eBACE2^−/− aortas as compared to tamoxifen treated control mice irrespective of sex. Blockade of nitric oxide synthase (NOS) with N^ω-nitro-L-arginine methyl ester abolished relaxations to A23187. In contrast, endothelium-independent relaxations to nitric oxide donor diethylamine-NONOate were unchanged. Expression of endothelial NOS protein and levels of cyclic nucleotides were also unaffected in eBACE2^−/− mice. Further analysis of the mechanisms underlying impaired endothelial function demonstrated that treatment with thromboxane A₂ receptor antagonist SQ29548 ameliorated relaxations to A23187 in the aorta of male and female eBACE2^−/− mice. Furthermore, mRNA and protein expressions of cyclooxygenase-2 as well as production of thromboxane A₂ and prostaglandin F_2α were significantly increased in the aorta of eBACE2^−/− mice. In contrast, production of 6-keto prostaglandin F_1α and prostaglandin E₂ were not affected. In addition, ex-vivo treatment of wild-type aortas with proinflammatory cytokines decreased protein expression of BACE2. The results of our study suggest that increased production of vasoconstrictor prostanoids are responsible for impairment of endothelium-dependent relaxations to A23187 in the aorta of eBACE2^−/− mice. We report previously unrecognized role of BACE2 in control of endothelial arachidonic acid metabolism and vasomotor function.

NEW & NOTEWORTHY

The exact physiological role of BACE2 in endothelial function of systemic arteries is unknown. Our study shows that thromboxane A₂ and prostaglandin F_2α are responsible for impairment of endothelium-dependent relaxations in endothelium-specific BACE2-deficient mice. The data support important role of BACE2 in control of endothelial arachidonic acid metabolism and vasomotor function.

INTRODUCTION

Endothelium is an important regulator of vasomotor function in the cardiovascular system. Endothelial cells continuously produce and release nitric oxide (NO) which promotes vasodilation under physiological conditions (1). Recently, we and others have shown that β-amyloid (Aβ) precursor protein (APP) as well as β-site APP-cleaving enzyme 1 (BACE1) and its homologue β-site APP-cleaving enzyme 2 (BACE2) are expressed in cerebral and systemic arteries (2–9). BACE2 cleaves wild-type APP within the Aβ region and hence exerts α-secretase like activity and can limit Aβ production (9–11). While BACE1 is mainly expressed in brain neurons, a growing body of evidence indicated that BACE2 is highly expressed in non-neuronal cells such as pancreas, kidney, stomach, and liver (8, 12). Furthermore, genetic inactivation of BACE2 causes hyperinsulinemia and insulin resistance in high fat fed mutant mice laking enzymatic activity of BACE2 (13, 14). We and others have recently reported that BACE2 is present in vascular endothelial cells (15–18) and in vascular smooth muscle cells (19–21). Existing evidence suggests that BACE2 is vascular protective protein (16). This concept is supported by studies demonstrating that, deficiency of BACE2 impairs expression and function of endothelial nitric oxide synthase (eNOS) in cultured human brain microvessels endothelial cells (16). However, the function of BACE2 in endothelial cells of systemic arteries is not known. In the present study, we generated conditional knockout mice laking BACE2 in endothelial cells to test the hypothesis that endothelial BACE2 plays a role in the control of vasomotor function in systemic arteries.

MATERIALS AND METHODS

Generation of eBACE2^−/− Mice

Experimental protocols were approved by the Institutional Animal Care and Use Committee of Mayo Clinic, and they comply with the National Institute of Health and the ARRIVE guidelines. B6NCrl;B6N-A^tm1Brd Bace2^{tm1a(EUCOMM)Wtsi}/Orl mice (RRID:IMSR_EM:10882) were obtained from INFRAFRONTIER/European Mutant Mouse Archive (EMMA) (22). The mice were crossed with Flp deleter mice to generate flox/+ Bace2^tm1c (BACE2^flox/+) mice at the TAAM-CNRS facility in Orléans, France (23). Floxed BACE2 allele (BACE2^flox/flox) mice were generated by crossing of BACE2^flox/+ mice in our laboratory and were identified by the PCR method and primers provided by TAAM-CNRS (23). Female BACE2^flox/flox mice were mated to male Cdh5(PAC)-CreERT2 mice expressing inducible cre deleter under the control of the vascular endothelial cadherin (Cdh5(PAC)) promoter (Cdh5-Cre⁺; RRID:IMSR_TAC:13073; Taconic Biosciences, Germantown, NY (24)) to generate BACE2^flox/+;Cdh5-Cre⁻ and BACE2^flox/+;Cdh5-Cre⁺ mice. Male BACE2^flox/+;Cdh5-Cre⁺ mice were then crossed with female BACE2^flox/flox mice to generate BACE2^flox/flox;Cdh5-Cre⁻ (wild-type [WT] littermates) mice and BACE2^flox/flox;Cdh5-Cre⁺ (eBACE2^−/−) mice. DNA was extracted using Purelink genomic DNA isolation kit (Cat# K1820–02; Invitrogen (RRID:SCR_008817), Carlsbad, CA), and the genotyping was performed using Cre primers provided by Taconic (Figure 1A). All mice were maintained on standard chow with free access to drinking water. WT littermates and eBACE2^−/− mice were selected from same or different breeder pairs by randomization and studied in parallel across the study.

Figure 1.

A) Original PCR representing genotyping of wild-type littermates (WT) and endothelium-specific BACE2-deficient (eBACE2^−/−) mice. Mice carrying LoxP sites on BACE2 gene (floxed BACE2) were crossed with C57BL/6-Tg(Cdh5-cre/ERT2)1Rha mice (Cdh5-cre⁺-Tg). The PCR products were 461 bp for floxed allele, 325 bp for wild-type allele, and 720 bp for Cre recombinase (Cre⁺). B) Quantitative RT-PCR analysis of BACE2 mRNA in endothelial cells isolated from the aorta of tamoxifen treated wild-type littermates (WT) and endothelium-specific BACE2-deficient (eBACE2^−/−) mice (*p=0.0022, Mann-Whitney U-test; n=6 mice per group). C) Representative Western blot for BACE2 protein expression in male WT littermates and eBACE2^−/− mice aortas. β-Actin is shown as a loading control. D) Quantitative Western blot analysis of BACE2 protein expression in the aorta of tamoxifen treated male WT littermates and endothelium-specific BACE2-deficient (eBACE2^−/−) mice. (*p= 0.0006; unpaired Student’s t-test; n=10 mice per group). All results are represented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range. E) Quantitative RT-PCR analysis of BACE2 mRNA expression in aorta and brain cortex of wild-type mice (*p=0.0286, Mann-Whitney U-test; n=4 mice per group). All results are presented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range.

Tamoxifen Treatment

To induce Cre recombination, young male and female WT littermates and eBACE2^−/− mice (2.5–4 months old) were fed chow diet containing tamoxifen (Cat# TD.130859; 400 mg tamoxifen citrate/kg diet, 2016; Envigo, Madison, WI) for 10 weeks (25). The body weight of mice and weight of chow consumption per cage were recorded weekly. At the end of tamoxifen treatment all mice were anaesthetized with overdose of pentobarbital (200–250 mg/kg BW, i.p.) followed by exsanguination via cardiac puncture for blood collection. The aortas were harvested and placed in cold (4°C) physiological Krebs solution (pH 7.4) composed in mmol/L: NaCl 118.6, KCl 4.7, CaCl₂ 2.5, MgSO₄ 1.2, KH₂PO₄ 1.2, NaHCO₃ 25.1, glucose 10.1, and EDTA 0.026. Aortas were dissected free from perivascular fat and blood was rinsed out under a light microscope.

Quantitative Real-Time PCR

To establish endothelium-specific deletion of BACE2, dissected aortas were rinsed in phosphate-buffered saline solution (PBS; Cat# 14190–144; GIBCO^® Thermo Fischer Scientific Inc., Waltham, MA), and endothelial cells were flushed out of aorta with lysis buffer from the kit (Purelink RNA Mini kit; Cat# 12183018A; Invitrogen). Isolation and purification of mRNA was performed according to the manufacturer instruction as previously described (26). Total mRNA was quantified using NanoDrop One (RRID:SCR_023005, Thermo Scientific Biosciences, Hanover Park, IL). Equal amount of mRNA was reversed to cDNA by using SuperScript III First-Strand Synthesis System kit (Cat# 18080–051; Invitrogen). Quantification was performed using CFX Connect Real-Time System (RRID:SCR_026760, Bio-Rad, Hercules, CA) and SsoAdvanced Universal SYBR Green Supermix (Cat# 1725270; Bio-Rad). The following primers spanning the deletion site exon 3 of the BACE2 gene (NCBI Reference Sequence: NC_000082.7) forward: 5’-GCAGATCACAAACGTGGTGG-3’ and reverse: 5’-AGGAACTTGCACTTGTCGGG-3’ were used. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as reference control as previously described (5). The PCR conditions were for activation at 95°C for 10 seconds, and for amplification at 95°C for 10 seconds and 60°C for 40 seconds (39 cycles). The amount of gene of interest was calculated as 2^−ΔCT (where ΔC_T = C_T of targeted BACE2 gene - C_T of internal control GAPDH gene, and C_T represents the cycle threshold) (27).

In separate studies, aortas from WT littermates and eBACE2^−/− mice as well as brain cortex and aorta from wild-type (BACE2^+/+) mice with endothelium were homogenized in lysis buffer, and cDNA synthesis was performed as described above. PrimePCR SYBR Green Assay mouse PTGS2 (Cat# qMmuCED0003742), BACE2 (Cat# qMmuCID0012853), and GAPDH (Cat# qMmuCED0027497) primers (Bio-Rad, RRID:SCR_008426) were used for amplification. The PCR conditions were for activation at 95°C for 30 seconds, and for amplification at 95°C for 10 seconds and 60°C for 30 seconds (39 cycles).

Glucose, Lipid Profile, and Aβ_1–40

Blood samples were transferred to a tube containing EDTA. Blood glucose levels were measured in whole blood immediately with Accu-Chek (Roche Diagnostics, Indianapolis, IN). Thereafter blood samples were centrifuged at 2000 rpm at 4°C for 10 minutes and the supernatants stored at −80°C until assayed. Plasma levels of cholesterol, HDL, and triglyceride were determined in the Targeted Biomarker Core Laboratory of Mayo Clinic using Hitachi 912 chemistry analyzer (Roche Diagnostics, Indianapolis, IN). Mouse-specific Aβ_1–40 ELISA kit was used to perform measurements of circulating levels of Aβ_1–40 (Cat# KMB3481; Invitrogen, Camarillo, CA).

Vascular Reactivity Studies

Mouse thoracic aortas were cut into 4-mm rings and mounted in organ chambers filled with Krebs solution (37°C, 94%O₂/6%CO₂) for isometric force recording (PowerLab, AD Instruments, RRID:SCR_001620, Colorado Springs, CO). Aortic rings of WT littermates and eBACE2^−/− mice were stretched to optimal force of 1.5 grams, after which the aortas were contracted two-times with 80 mmol/L KCl and washed out. Concentration-dependent response curves to prostaglandin F_2α (PGF_2α; 10⁻⁸–3×10⁻⁵ mol/L; Cat# 16020; Cayman (RRID:SCR_008945), Ann Arbor, MI) and L-phenylephrine (Phe; 10⁻⁹–10⁻⁵ mol/L; Cat# P6126; Sigma-Aldrich (RRID:SCR_008988), St. Louis, MO) were cumulatively obtained. After washing out, endothelium-dependent relaxations in response to A23187 (10⁻⁹–10⁻⁶ mol/L; Cat# C7522; Sigma) were recorded in the absence and in the presence of NOS inhibitor N^ω-nitro-L-arginine methyl ester (L-NAME; 3×10⁻⁴ mol/L, 30 minutes; Cat# N5751; Sigma) or SQ29548 (1 μmol/L, 30 minutes; Cat# 19025; Cayman Chemical, Ann Arbor, MI). In addition, endothelium-dependent relaxations in response to acetylcholine (10⁻⁹–10⁻⁵ mol/L; Cat# A6625; Sigma) and endothelium-independent relaxations in response to NO donor diethylammonium (Z)-1-(N,N-diethylamino) diazen-1-ium-1,2-diolate (DEA-NONOate; 10⁻⁹ – 10⁻⁵ mol/L; Cat# 82100; Cayman) were cumulatively obtained. Various concentrations of PGF_2α (3×10⁻⁶–8×10⁻⁶ mol/L) and phenylephrine (10⁻⁷–10⁻⁶ mol/L) were used to achieve similar submaximal contractions in WT littermates and eBACE2^−/− mice, and each relaxation response is expressed as percentage of submaximal contractions. Each contraction induced by PGF_2α and phenylephrine is expressed as percentage of second KCl contraction which was obtained at the beginning of the experiment.

Cyclic Nucleotides

Ten millimeters long aortas were opened longitudinally and were incubated in minimal essential medium (MEM; Cat# 11095–080; GIBCO^® Thermo Fischer Scientific, RRID:SCR_008452, Waltham, MA) composing 0.1% BSA (Cat# 820451; Millipore, Burlington, MA), 100 U/mL penicillin/100 μg/mL streptomycin (Cat# 15140–148; GIBCO^®), and 100 μmol/L 3-isobutyl-1-methylxanthine (Cat# I5879; Sigma) for 1 hour at 37°C in CO₂ incubator. Some of the aortas were incubated with 10 μmol/L indomethacin (Cat# I7378; Sigma) for additional 30 minutes, after which 0.1 μmol/L A23187 (Cat# C7522; Sigma) was added and all samples were incubated for additional 10 minutes. Aortas were frozen immediately in liquid N₂ and stored at −80°C. Aortas were homogenized in lysis buffer containing 50mM Tris-HCl (pH 6.0; Cat# BP153–1; Thermo Fisher), and 2% Triton X-100 (Cat# T9284; Sigma) according to the manufacturer instruction. Samples were centrifuged at 5000 rpm and the supernatants were subjected to cyclic guanosine monophosphate (cGMP; Cat# ADI-900–164; Enzo Life Sciences (RRID:SCR_003900), Farmingdale, NY) and cyclic adenosine monophosphate (cAMP; Cat# STA-500, Thermo Fisher Scientific) measurements. The protein was determined using colorimetric kit (Cat# 5000116; BioRad), and the obtained results were used to normalize cGMP and cAMP levels.

Release of Prostanoids

Ten millimeters long thoracic aortas were opened longitudinally and were incubated in MEM of the same composition for 100 minutes as described above. Subsequently, the incubation medium was collected, quickly frozen in liquid N₂, and stored at −80°C until assayed. Colorimetric ELISA kits (Cayman, Ann Arbor, MI) were used for measurements of prostaglandin F_2α (PGF_2α; Cat# 516011–96), thromboxane B₂ (TXB₂, a stable metabolite of thromboxane A₂ (TXA₂); Cat# 501020–96), prostaglandin E₂ (PGE₂; Cat# 514010–96), and 6-keto prostaglandin F_1α (6-keto PGF_1α, a stable metabolite of prostaglandin I₂ (PGI₂; Cat# 515211–96). All results were normalized against protein levels.

Immunoblot Analysis

Aortas were collected and stored at −80°C until assayed. Aortas were homogenized in lysis buffer composed in mmol/L: 50 Na₄P₂O₇ (Cat# 221368; Sigma), 50 NaF (Cat# S6521; Sigma), 50 NaCl (Cat# S271–3; Fisher), 5 EDTA (Cat# E6635; Sigma), 5 EGTA (Cat# E4378; Sigma), 0.1 Na₃VO₄ (Cat# S6508), 10 HEPES (Cat# H7523; Sigma), 1% Triton X-100 (Cat# T9284; Sigma), 0.5 PMSF (Cat# P7626; Sigma), and protease inhibitor cocktail (Cat# P8340, Sigma). Samples were centrifuged and supernatants were subjected to protein assay. Samples were boiled in 6x reducing Laemmli SDS buffer at 100°C for 10 minutes. Equal amount of proteins were separated on 7.5% or 10% mini-protean TGX gels (Cat# 4561024 or Cat# 4561034, respectively; Bio-Rad) and were transferred to nitrocellulose membrane (Cat# 16201120; Bio-Rad). Membranes were blocked by using 5% blotting-grade Blocker (Cat# 1706404; Bio-Rad) and 0.1% Tween 20 (Cat# P1379; Sigma) in Tris-buffered salt (TBS; pH 7.4) solution overnight. The membranes were probed with primary antibodies against endothelial nitric oxide synthase (eNOS; 1:250, Cat# 610297, RRID:AB_397691, BD Biosciences, San Jose, CA), inducible nitric oxide synthase (iNOS; 1:250, Cat# 610333, RRID:AB_397723, BD Biosciences), BACE2 (1:250, Cat# SC-271212, RRID:AB_10609490, Santa Cruz, Dallas, TX), cyclooxygenase-1 (COX-1; 1:500, Cat# 35–8100, RRID:AB_2533223, Invitrogen), cyclooxygenase-2 (COX-2; 1:250, Cat# 610204, RRID:AB_397603, BD Biosciences), and prostaglandin I₂ synthase (PGI₂S; 1:20,000, Cat# 100023, RRID:AB_10078998, Cayman) for 4 hours. Membranes were washed twice with TBS containing 0.1% Tween 20 and probed with secondary HRP monoclonal or polyclonal antibodies (NA931: RRID:AB_772210, or NA934: RRID:AB_772206, respectively; 1:2000; GE Healthcare Life Sciences, Pittsburgh, PA) for one hour. As loading control, blots were rehybridized with monoclonal anti-β-actin (1:50,000; Cat# A5316, RRID:AB_47674, Sigma). The bands were visualized using Odyssey Fc imaging system (RRID:SCR_023227, Li-Cor, Lincoln, NE) and were quantified by Image Studio^™ Software (RRID:SCR_015795, version 5.0; Li-Cor). Protein expression was normalized to β-actin. The selectivity of primary antibodies was validated in previous study (28). Furthermore, the antibody selectivity of BACE2 was tested in aortas obtained from non-induced WT littermates and eBACE2^−/− mice and by using different secondary light chain specific HRP-conjugated monoclonal anti-mouse IgG (1;2000; Cat# MCA152P; Bio-Rad; Supplemental Figure S1).

Electron Microscopy

Isolated aortic ring segments from wild-type (BACE2^+/+) mouse were fixed in 10% buffered Formalin (Cat# C4320–101; Cardinal Health, Waukegan, IL) overnight, dehydrated in a series of ethanol from 40% to absolute while progressively lowering the temperature to −20°C, embedded in LR white resin (Cat# 18181; Ted Pella, Inc., Redding, CA) and polymerized using a 15W UV cooled polymerization chamber (Model 6202, Ted Pella, Inc., Redding, CA). Thin sections (200 nm) were mounted on nickel grids and labeled with BACE2 antibody (1:100 dilution; Cat# SC-271212; Santa Cruz) overnight in a humid chamber at 4°C. Grids were rinsed in 0.1 mol/L phosphate buffer and incubated in a goat anti-mouse secondary antibody conjugated to 10 nm gold beads (1:50 dilution; Cat# 25129; Electron Microscopy Sciences, Hatfield, PA) for 1 hour at room temperature. Grids were then rinsed again in 0.1 mol/L phosphate buffer and fixed in 4% paraformaldehyde + 1% glutaraldehyde, followed by a final rinse in water. When labeling was completed, the sections were stained with lead citrate (Cat# 16707235; Fisher Scientific, Hampton, NH) prior to transmission electron microscopy. Images were acquired using a JEOL1400+ TEM (RRID:SCR_020179, JEOL, Peabody, MA) operating at 80kV with an AMT Nanosprint 12 camera (AMT Imaging, Woburn, MA).

Ex-vivo Cytokines Treatment

Ten millimeters long aortas obtained from male wild-type (BACE2^+/+) mice were opened longitudinally and were incubated in MEM (composing 0.1% BSA, 100 U/mL penicillin/100 μg/mL streptomycin) with PBS or cocktail of inflammatory cytokines containing 20 ng/mL recombinant mouse tumor necrosis factor-α (TNFα; Cat# 410-MT; R&D Systems, Minneapolis, MN; RRID:SCR_006140), 1 ng/mL recombinant mouse interleukin-1β (IL1β; Cat# 401-ML; R&D Systems), and 50 ng/mL recombinant mouse interferon-γ (INFγ; Cat# 485-MI; R&D Systems) for 24 hours at 37°C in CO₂ incubator. After incubation the aortas were subjected to Western blot analysis.

Statistical Analysis

The collection of data was not blinded with respect to the tamoxifen treatment. GraphPad Prism software version 10.2 (RRID:SCR_002798; GraphPad Prism, Boston, MA) was used to perform graphical design and statistical analyses. In the present study, no data were excluded, and n indicates the number of mice used per one experiment. Sample sizes were determined based on previous studies (28). Continuous data were expressed as mean ± standard error of the mean (SEM) or median and interquartile range. The data were assessed for normality with Shapiro-Wilk test. Unpaired, two-tailed Student’s t-test was performed to compare two groups of WT littermates and eBACE2^−/− mice. One-way analysis of variance (ANOVA) followed by the Tukey’s multiple comparisons test was used for multiple comparison. Non-parametric Mann-Whitney U-test was used to compare the means of two groups that were not normally distributed. For analysis of vascular reactivity studies, concentration-response curves were analyzed by two-way ANOVA followed by Sidak’s multiple comparisons. A p-value of less than 0.05 was considered statistically significant.

RESULTS

Characterization of eBACE2^−/− mice

All eBACE2^−/− mice and their WT littermates tolerated treatment with tamoxifen well, and we observed no mortality. Body weight and intake of tamoxifen was not different between WT littermates and eBACE2^−/− mice (Table 1, male mice, and Supplemental Table S1, female mice) indicating that all mice were treated under same conditions. Blood glucose levels and plasma levels of Aβ_1–40 were not significantly affected in male and female eBACE2^−/− mice (Table 1 and Supplemental Table S1, respectively). Plasma levels of cholesterol and HDL were unchanged in male eBACE2^−/− mice while plasma levels of triglyceride were significantly increased (Table 1). No alterations were observed in female eBACE2^−/− mice with respect to lipid profile (Supplemental Table S1).

Table 1.

Characteristics of male wild-type littermates and eBACE2^−/− mice.

Parameters	Cdh5-Cre−;BACE2flox/flox (WT littermates)	Cdh5-Cre+;BACE2flox/flox (eBACE2−/−)	p-value
BW (g)	26.5±0.4 (21)	26.6±0.4 (21)	0.981
Tamoxifen citrate (mg/kg BW/day)	42.3±1.1 (21)	41.9±0.9 (21)	0.777
Glucose (mg/dL)	187±9 (10)	185±8 (10)	0.873
Aβ1–40 (pg/mL)	230±17 (10)	234±10 (10)	0.849
Total Cholesterol (mg/dL)	45.6±3.7 (10)	49.2±1.3 (10)	0.367
HDL (mg/dL)	32.0±2.7 (10)	34.7±1.0 (10)	0.365
Triglyceride (mg/dL)	74.0±6.9 (10)	91.4±4.5 (10) *	0.0496

Data are means ± SEM, and the numbers of mice are indicated in the parentheses. Wild-type littermates and eBACE2^−/− mice were compared by unpaired Student t-test, and p-values (two-tailed) are shown in this table. BW denotes body weight; Aβ, amyloid-β; WT, wild-type; *, statistically significant.

Expression and Localization of BACE2

Quantitative RT-PCR analysis confirmed that BACE2 mRNA expression was decreased by 94% in endothelial cells isolated from aortas of eBACE2^−/− mice as compared to endothelial cells isolated from WT littermates aortas (p=0.0022; Figure 1B). Endothelium-specific depletion of BACE2 significantly decreased protein expression of BACE2 by 48% in aortas of eBACE2^−/− mice (p=0.0006; Figures 1C and 1D). To further validate effect of tamoxifen treatment, Western blot analysis demonstrated no change of BACE2 expression in non-induced eBACE2^−/− mice, while eBACE2^−/− mice fed with tamoxifen containing chow diet showed lower BACE2 protein expression as compared to tamoxifen-treated WT littermates (Supplemental Figure S1). It is interesting to note that BACE2 mRNA levels were significantly higher in the aorta as compared to brain cortex of wild-type mice (p=0.029; Figure 1E). Electron microscopy analysis of wild-type aorta revealed that BACE2 immunogold particles were localized mainly within the endothelial cell cytoplasm and in the elastica interna (Figure 2). We also observed some BACE2 immunogold particles within the smooth muscle cells (data not shown).

Figure 2.

A) Original electron photomicrographs of wild-type mouse aorta. Immunogold labeling of BACE2 with 10 nm gold particles (arrows) indicated presence of the protein within endothelial cell cytoplasm and into elastica interna. B) Aortic sections were stained with secondary antibody (conjugated to 10 nm gold beads) only. No immunogold particles were detected. Scale bar denotes 400 nm. L indicates lumen; E, endothelium; BM, basement membrane; SM, smooth muscle.

Endothelial Function in eBACE2^−/− mice

Endothelium-specific deletion of BACE2 did not affect the efficacy of endothelium-dependent relaxations to acetylcholine in the aorta of male eBACE2^−/− mice (68±4% vs. 69±4% in WT littermates; n=6; p=0.818). Moreover, the responses to acetylcholine were equipotent between WT littermates and eBACE2^−/− mice (pEC₅₀: 6.75±0.09 and 6.88±0.04, respectively; n=6; p=0.223). However, endothelium-dependent relaxations to Ca²⁺-ionophore A23187 were significantly impaired in the aorta of male eBACE2^−/− mice (p<0.05; Figure 3A). Treatment with NOS inhibitor L-NAME inhibited relaxations to A23187 and abolished the difference between WT littermates and eBACE2^−/− mice (p>0.05; Figure 3B). Endothelium-independent relaxations to NO-donor DEA-NONOate were unaltered in eBACE2^−/− mice as compared to WT littermates (p>0.05; Figure 3C). Contractions to KCl were unchanged in the aorta of male eBACE2^−/− mice (1.85±0.06g; n=6) as compared to male WT littermates (1.91±0.08g; n=6; p=0.818). Moreover, contractions to PGF_2α were not significantly different between WT littermates and eBACE2^−/− mice (p>0.05; Figure 3D).

Figure 3.

A) Endothelium-dependent relaxations to A23187 in isolated aortas from male WT littermates and endothelium-specific BACE2-deficient (eBACE2^−/−) mice. Please note that the responses were significantly different (*p<0.05; n=6 mice per group). B) Effect of L-NAME on endothelium-dependent relaxations to A23187 in WT littermates and eBACE2^−/− mice aortas (p>0.05; n=4 mice per group). C) Endothelium-independent relaxations to NO-donor DEA-NONOate in male WT littermates and eBACE2^−/− mice aortas (p>0.05; n=6 mice per group). D) Concentration-dependent contractions to PGF_2α in isolated aortas from male WT littermates and eBACE2^−/− mice (p>0.05; n=6 mice per group). All results are shown as mean ± SEM and were analyzed by two-way ANOVA followed by Sidak’s multiple comparisons. The contractions were expressed as percentage (%) of KCl response. The relaxations were obtained during contractions induced by PGF_2α and were expressed as percentage (%) change from submaximal contractions to PGF_2α.

To determine whether prostanoids are involved in alterations of vascular responses to A23187, aortas were treated with thromboxane receptor antagonist SQ29548. Of note, phenylephrine was used in this study design as SQ29548 also blocked PGF_2α–induced contractions (data not shown; n=2). Endothelium-dependent relaxations to A23187 were impaired in both male and female eBACE2^−/− mice aortas contracted with phenylephrine (p<0.05; Figure 4 and Supplemental Figure S2, respectively). Treatment with SQ29548 significantly improved relaxations to A23187 in both male and female eBACE2^−/− mice (p<0.05; Figure 4B and Supplemental Figure S2, respectively). In addition, SQ29548 significantly improved relaxations in WT littermates at highest concentration of A23187 (p=0.0155; Figure 4A). Contractions to phenylephrine were not significantly affected by SQ29548 treatment in male WT littermates and eBACE2^−/− mice (40±4% vs. 47±5% without treatment (p=0.505) and 48±5% vs. 63±7% without treatment (p=0.098), respectively; n=8 per group). Moreover, contractions to phenylephrine were not significantly altered by SQ29548 in female WT littermates and eBACE2^−/− mice (36±6% vs. 40±5% without treatment (p=0.588) and 38±3% vs. 46±6% without treatment (p=0.277), respectively; n=6 per group). As there was no sex difference observed with regard to endothelium-dependent relaxations to A23187 (p>0.05; Supplemental Figure S3A), subsequent experiments were performed in male mice only.

Figure 4.

A) Effect of SQ29548 on endothelium-dependent relaxations to A23187 in aortas of male WT littermates (n=8 mice per group; *p=0.016; two-way ANOVA followed by Sidak’s multiple comparisons). B) Effect of SQ29548 on endothelium-dependent relaxations to A23187 in male eBACE2^−/− mice aortas (n=8 mice per group; *p<0.0001; two-way ANOVA followed by Sidak’s multiple comparisons). All results are shown as mean ± SEM and the relaxations were expressed as percentage (%) change from submaximal contractions phenylephrine.

Production of Prostanoids in eBACE2^−/− mice

Under basal conditions, production of TXB₂ was significantly increased in the aorta of male eBACE2^−/− mice (p=0.001; as compared to WT littermates; Figure 5A). Moreover, PGF_2α levels were significantly increased in the aorta of eBACE2^−/− mice (p=0.041; as compared to WT littermates; Figure 5B). In contrast, productions of PGE₂ (Figure 5C; p=0.489) and 6-keto PGF_1α (Figure 5D; p=0.839) were not affected.

Figure 5.

Release of prostanoids from aortas of male WT littermates and eBACE2^−/− mice. Incubation medium was collected and assayed for prostaglandin levels. A) TXB₂ production was significantly elevated in eBACE2^−/− mice (*p=0.001; unpaired Student’s t-test; n=9 mice per group). B) PGF_2α levels were significantly increased in eBACE2^−/− mice (*p=0.041; unpaired Student’s t-test; n=9 mice per group). C) PGE₂ levels were unchanged in eBACE2^−/− mice (p=0.489; Mann-Whitney U-test; n=9 mice per group). D) 6-keto PGF_1α were not affected in eBACE2^−/− mice (p=0.839; unpaired Student’s t-test; n=9 mice per group). All data are presented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range.

Expressions of eNOS and Prostanoid Enzymes in eBACE2^−/− mice

Immunoblot analysis revealed that endothelium-specific deficiency of BACE2 did not affect eNOS protein expression in male eBACE2^−/− mice aortas (p=0.835 as compared to WT littermates; Figure 6A). Interestingly, gene expression of PTGS2 mRNA and protein expression of COX-2 were significantly enhanced in the aorta of eBACE2^−/− mice (p=0.0012 and p=0.024, respectively, as compared to WT littermates; Figures 6B and 6C). In contrast, expressions of COX-1 and PGI₂S were not altered in eBACE2^−/− mice (p=0.983 and p=0.583, respectively; Figures 6D and 6E).

Figure 6.

Representative Western blots and gene expression studies demonstrating protein expression of eNOS (A; p=0.835; unpaired Student’s t-test; n=8 mice per group), gene expression of PTGS2 (B; p=0.0012; unpaired Student’s t-test; n=6 mice per group), protein expression of COX-2 (C; p=0.024; unpaired Student’s t-test; n=10 mice per group), protein expression of COX-1 (D; p=0.983; unpaired Student’s t-test; n=8 mice per group), and protein expression of PGI₂S (E; p=0.583; unpaired Student’s t-test; n=8 mice per group) in aortas of male WT littermates and eBACE2^−/− mice. β-Actin Western blots are shown as a loading control, and the results are expressed as relative densitometry compared with their respective β-actin. All data are presented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range. Statistical significance is marked with an asterisk (*).

Cyclic Nucleotide eBACE2^−/− mice

Measurements of cyclic nucleotides in aortas indicated that basal levels of cGMP and cAMP were not different between male WT littermates and eBACE2^−/− mice (p>0.05; Figure 7). Stimulation of isolated aortas with Ca²⁺-ionophore A23187 significantly increased cGMP levels in both WT littermates (p=0.022; Figure 7A) and eBACE2^−/− mice (p<0.0001; Figure 7B). A23187 also significantly augmented cAMP levels in WT littermates (p=0.0005; Figure 7C) and eBACE2^−/− mice (p=0.0006; Figure 7D). Treatment with indomethacin significantly inhibited A23187-induced increase in cAMP levels in both WT littermates (p=0.002; Figure 7C) and eBACE2^−/− mice (p=0.0002; Figure 7D) while A23187-stimulated cGMP levels remained unchanged (p>0.05).

Figure 7.

A) Levels of cGMP were increased in A23187-stimulated aortas of male WT littermates (p=0.022) which was unaffected by indomethacin treatment (p=0.9999; one-way ANOVA; n=9 per treatment). B) Levels of cGMP were increased in A23187-stimulated aortas of male eBACE2^−/− mice (p<0.0001) which was unaffected by indomethacin treatment (p=0.9999; one-way ANOVA; n=9 per treatment). C) Levels of cAMP were increased in A23187-stimulated aortas of male WT littermates (p=0.0005) which was inhibited by indomethacin treatment (p=0.002; one-way ANOVA; n=9 per treatment). D) Levels of cAMP were increased in A23187-stimulated aortas of male eBACE2^−/− mice (p=0.0006) which was inhibited by indomethacin treatment (p=0.0002; one-way ANOVA; n=9 per treatment). All data are presented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range. Statistical significances are marked with an asterisk (*) for comparison to basal levels or with a hashtag (#) for comparison to A23187-stimulated arteries.

Ex-vivo Cytokines Studies

Previous study reported that inflammatory cytokines cause downregulation of BACE2 protein expression in neuronal cells (29), we incubated isolated wild-type (eBACE2^+/+) mice aortas for 24 hours with cytokines cocktail (composing of 20 ng/mL TNFα, 1 ng/mL IL1β, and 50 ng/mL INFγ). Western blot analysis revealed that inflammatory cytokines significantly decreased BACE2 expression as accompanied by an increased iNOS protein expression in the aorta (p=0.0091 and p=0.0286, respectively; Figure 8).

Figure 8.

A) Representative Western blot for protein expressions of iNOS and BACE2 in male wild-type (BACE2^+/+) mice aortas after ex-vivo treatment with cytokines cocktail (composing of 20 ng/mL TNFα, 1 ng/mL IL1β, and 50 ng/mL INFγ). β-Actin is shown as a loading control. B) Quantitative Western blot analysis of BACE2 protein expression (*p=0.0091; unpaired Student’s t-test; n=4 mice per group). C) Quantitative Western blot analysis of iNOS protein expression (*p=0.0286; Mann-Whitney U-test; n=4 mice per group). All results are represented as box plots with whiskers showing the median, the 25^th to 75^th percentiles, and min-max range.

Tamoxifen Treatment in Control Mice

Since recent studies reported that tamoxifen treatment affects cholesterol levels in hypercholesterolemic mice (30, 31), we performed control experiments to determine the effect of tamoxifen on circulating levels of cholesterol. Wild-type (eBACE2^+/+) and heterozygous eBACE2^flox/+ mice not carrying Cdh5(PAC)-CreERT2 allele were fed normal chow diet or chow diet containing tamoxifen for 10 weeks in the same conditions as eBACE2^−/− mice. Measurements of lipid profile revealed that plasma levels of cholesterol, HDL, and LDL were not significantly affected by tamoxifen treatment as compared to control mice fed normal chow diet (Supplemental Table S2). Glucose levels in whole blood were also unchanged (Supplemental Table S2). Furthermore, endothelium-dependent relaxations to A23187 were identical, and eNOS protein expression was not significantly altered in aortas of control mice fed normal chow diet (Supplemental Figure S3B).

DISCUSSION

The aim of the present study was to characterize the effect of endothelium-specific deletion of BACE2 on endothelial function in eBACE2^−/− mice. We present several novel findings. First, BACE2 is expressed in endothelial cells of peripheral arteries. Second, genetic deletion of BACE2 in endothelial cells significantly impaired endothelium-dependent relaxations to Ca²⁺-ionophore A23187 in the aorta. Third, thromboxane receptor antagonist significantly improved relaxations to A23187 in eBACE2^−/− mice. Fourth, protein expression of COX-2 as well as production of TXA₂ and PGF_2α were increased in the aorta of eBACE2^−/− mice. Taken together, these results demonstrate that increased production of vasoconstrictor prostanoids is the mechanism underlying vascular dysfunction in eBACE2^−/− mice.

Consistent with our current study, previous findings established that BACE2 protein expression is present in endothelial cells of systemic arteries (17, 18). Moreover, electron microscopic analysis showed intracellular localization of BACE2 in endothelial cells. While some studies have pointed out that a small amount of BACE2 is present at the plasma membrane, its primary function appears to be found mainly within the intracellular compartments (9, 32). This contrasts with its close relative, BACE1 protein, which is found more commonly at the plasma membrane of endothelial cells (7). Of note, we also detected BACE2 immunogold in smooth muscle cells. However, whether deficiency of BACE2 in smooth muscle cells causes vascular dysfunction remains to be determined.

Previous study revealed that BACE2 can cleave APP within the amyloid-beta region which prevents the production of Aβ peptides (10, 11, 33). However, we did not observe any changes in circulating Aβ_1–40 levels in eBACE2^−/− mice indicating that endothelial cells are not major sources of Aβ peptides. Similar observations were made in endothelium-specific APP mice (34). Thus, observed impairment of endothelium-dependent relaxations to A23187 is not caused by increased circulating levels of Aβ peptides. It has been reported that genetic inactivation of BACE2 can also lead to increased β-cell mass in pancreas and increased insulin secretion in vivo (13, 14). Nonetheless, blood glucose levels were unchanged in eBACE2^−/− mice demonstrating that the endothelial cells are not responsible for BACE2-dependent systemic glucose control.

We showed that plasma cholesterol and HDL levels were also unaltered in endothelium-specific BACE2 mice. In contrast, plasma triglyceride levels were slightly elevated in both male and female eBACE2^−/− mice, but the increase was significant only in the male mice indicating a sex-specific difference in the magnitude of the change. The mechanism by which BACE2 affects triglyceride metabolism is unclear and remains to be determined. It is important to note that the observed increase in triglycerides levels was not due to tamoxifen toxicity, because our study also established that circulating levels of lipids were not affected by tamoxifen treatment in control mice. We also wish to point out that genetic inactivation of BACE2 does not affect arterial blood pressure, because we showed in our previous study that arterial blood pressure was normal in global BACE2^−/− mice (16). Thus, it is unlikely that increased production of TXA₂ is the result of elevated blood pressure in endothelium-specific BACE2^−/− mice.

Endothelium-dependent relaxations to the Ca²⁺-ionophore A23187 causes receptor-independent translocation of Ca²⁺ into intracellular space and may simultaneously activate eNOS and arachidonic acid metabolism (35–37). Endothelium-dependent relaxations to A23187 were significantly impaired in the aorta of eBACE2^−/− mice. Increase of intracellular Ca²⁺ activates eNOS thereby increasing production of NO (36). Importantly, L-NAME treatment abolished relaxations to A23187 in both WT littermates and eBACE2^−/− mice. Moreover, we provided evidence that eNOS protein expression was not significantly affected in eBACE2^−/− mice suggesting that impaired endothelium-dependent relaxations are not caused by the decreased expression of eNOS. Furthermore, direct vascular smooth muscle responsiveness to NO was unchanged because endothelium-independent relaxations to NO donor DEA-NONOate were not different between WT littermates and eBACE2^−/− mice. Since NO dilates blood vessels by stimulating soluble guanylyl cyclase and subsequent increase in production of cGMP in smooth muscle cells (38), we also determined cGMP levels. Basal and stimulated cGMP levels were not altered in eBACE2^−/− mice aortas indicating that soluble guanylate cyclase function is not impaired.

We also showed that A23187 significantly increased production of cAMP. Prior studies established that A23187 may increase production of PGI₂ thereby stimulating production of cAMP (39). Despite the fact that 6-keto PGF_1α levels were not increased under basal conditions, exposure to A23187 may increase PGI₂ production thereby providing an explanation for detected indomethacin-sensitive increase in cAMP levels. Importantly, basal and stimulated production of cAMP as well as protein expression of PGI₂S were not affected by genetic inactivation of BACE2. These observations reinforce our conclusion that enhanced production of vasoconstrictor prostanoids (see below) rather than impaired vasodilatory effect of PGI₂ is responsible for observed endothelial dysfunction in aorta of eBACE2^−/− mice.

It is well established that endothelium-derived contracting factors such as prostanoids play a central role in both vascular physiological and pathophysiological processes (40). Moreover, prostanoids are produced in endothelial cells (41, 42) and can overcome NO-mediated relaxations by promoting increased vasoconstrictions in vascular smooth muscle cells (40). Most notably, blockade of TXA₂ receptor by SQ29548 significantly improved endothelium-dependent relaxations to A23187 in both male and female eBACE2^−/− mice. Further analysis of prostanoids profile revealed that vascular production of TXA₂ and PGF_2α were significantly augmented in the aorta of eBACE2^−/− mice. In addition, previous report suggests that PGF_2α can activate TXA₂ receptors (40), and this observation is consistent with our findings demonstrating that SQ29548 blocked PGF_2α-induced contractions. The mechanism by which the production of TXA₂ and PGF_2α is increased is unclear. Notably, protein expression of COX-2 was significantly increased in the aorta of in eBACE2^−/− mice while protein expression of COX-1 remained unchanged. These observations suggest that upregulated COX-2 is the most likely source of increased production of vasoconstrictor prostanoids TXA₂ and PGF_2α (43).

Our results have important implications for understanding of BACE2 role in pathogenesis of endothelial dysfunction. Importantly, risks factors for cardiovascular and cerebrovascular disease can lead to the inhibition and/or downregulation of BACE2. For example, a recent study reported that the expression of BACE2 is downregulated in cerebral endothelial cells of patients with Alzheimer’s disease (21). This finding is further supported by the fact that inflammatory cytokines are capable of downregulating BACE2 protein expression in glial cells (29). Indeed, we presented in our study that ex-vivo treatment with inflammatory cytokines decreased BACE2 protein expression in isolated wild-type mice aortas indicating that BACE2 is susceptible to inflammation. We propose that maintaining normal endothelial function requires normal function of BACE2. Moreover, our results underscore the importance of selective inhibition of BACE1 in patients with Alzheimer’s disease treated with BACE1 inhibitors (44).

In conclusion, the present study demonstrates that endothelium-specific deletion of BACE2 impairs endothelium-dependent relaxations. Increased production of vasoconstrictor prostanoids appears to be major mechanism underlying endothelial dysfunction in aorta of eBACE2^−/− mice. Our results support the concept that endothelial BACE2 is an important vascular protective protein responsible for preservation on normal vasomotor function.

Supplementary Material

Supplemental Figures S1-S3 and Supplemental Tables S1-S2: https://doi.org/10.6084/m9.figshare.29156012.

DATA AVAILABILITY

Data will be made available upon reasonable request.