Endothelin Receptor Blockade Improves Cerebral Blood Flow-Mediated Dilation in a Mouse Model of Alzheimer’s Disease Free

Cerebral blood flow (CBF) is severely diminished in patients with Alzheimer’s disease (AD) [1]. In elderly persons with a high risk of AD, reduced CBF precedes the onset of cognitive decline and amyloid deposition [2]. Reduced CBF has also been observed in early stages of the disease in mouse models [3, 4]. In AD patients, the greater the reduction in CBF, the more notable the impact on cognitive performance. Converging observations have highlighted the critical relationship between neurovasculature and brain health. However, the mechanisms and pathogenic relevance of this link are yet to be fully understood, with the following question in particular still needing clarification: is cerebrovascular dysfunction a consequence of the neurodegenerative process or a pathogenic contributor [5]?

Endothelial function, evaluated by measuring flow-mediated dilation (FMD) in the brachial artery, is impaired in patients with AD compared to age-matched individuals; the amplitude of FMD has been shown to be inversely correlated with the disease stage [6]. A recent study has also shown the correlation between FMD and the extent of the beta-amyloid burden in cognitively normal older adults [7]. Moreover, the increase in middle cerebral artery blood flow velocity induced by exercise is reduced with beta-amyloid burden in these patients (cognitively normal older adults) [8].

We previously demonstrated the presence of endothelial dysfunction in a mouse model of AD (APP-PS1) prior to the onset of clinical and histological disease [9]. Blocking endothelial production of nitric oxide accelerates the development of cognitive disorders and brain damage independently of blood pressure levels, thus suggesting that endothelial dysfunction has a more deleterious effect on the disease process than blood pressure levels and is an early marker of AD [9].

Resistance arteries are small blood vessels located immediately upstream of capillaries. Under physiological conditions, the basal tone of resistance arteries regulates local blood flow and capillary pressure. Any changes in the structure and/or function of resistance arteries will impair their essential role, i.e., inverse changes in radius and hemodynamic resistance during arterial blood pressure variations in order to maintain capillary pressure and microvascular perfusion. A rise in capillary pressure with arterial hypertension can lead to damage and structural loss of microvessels and thus relative tissue ischemia [10]. In the absence of arteriolar vasodilation during reductions in blood pressure, tissue perfusion becomes inadequate, and ischemia is aggravated [11]. Vasomotor tone results from the interaction between pressure-induced smooth muscle contraction (myogenic tone) and FMD, which, in turn, is mediated via the activation of endothelial cells by shear stress and the neurovascular unit [12]. When vasodilation/vasoconstriction within the cerebral vascular networks is absent or impaired, organ damage and disorders associated with cardio- and cerebrovascular risk factors are exacerbated. FMD in human large arteries is essentially dependent on acute production of nitric oxide by endothelial cells in response to an acute increase in shear stress [13‒15], and reduced FMD is a hallmark of endothelium dysfunction [15‒17].

Endothelin-1 (ET-1) is a potent endothelium-derived contracting factor that counters endothelium-dependent dilation [18]. ET-1 is implicated not only in the reduction of CBF in several pathological conditions [19] but also in the attenuated FMD in hypertension, diabetes, hypercholesterolemia, and obesity [20‒22].

We previously reported that APP-PS1 mice exhibit temporal order memory deficits in the episodic-like memory task and that this cognitive impairment is associated with a greater cortical amyloid burden [23]. The present study, therefore, sought to gain a clearer understanding of early vascular impairment in this mouse model of AD by investigating the cerebrovascular reactivity (CVR) deficits that occur in response to hypercapnia after a carbon dioxide rebreathing challenge, known to predict cognitive decline in patients [24]. We investigated CVR in vivo, in APP-PS1 mice, and in wild-type (WT) littermate controls under normoxic hypercapnic conditions. Ex vivo FMD was measured in isolated posterior cerebral arteries (PCAs) from APP-PS1 mice and WT littermate controls. We also studied in vivo and ex vivo CVR after endothelin receptor blockade. In APP-PS1 mice, endothelium-dependent cerebral vasodilation was impaired but restored after endothelin receptor blockade.

Experiments were conducted on a total of 38 male hemizygous APP-PS1 mice and 36 nontransgenic (WT) male control littermates [9]. Mice were 5-6 month-old. All animal procedures were approved by the Local Committee on Animal Research and Ethics (protocol number APAFIS#2413-2015102313463997v2, APAFIS# 2018051412442187v3, and APAFIS#03560.02).

Thermoregulated mice were anesthetized (1% isoflurane) prior to echo-Doppler scanning (Acuson S 3000, Erlangen, Germany) using a 14-MHz linear transducer (14L5 SP), as previously reported [25, 26]. Briefly, two-dimensional ultrasound imaging was used to visualize a horizontal cross-sectional (B-mode) image of the base of the skull and the brain structures of the animal. Color-coded Doppler was then enabled to display color-coded images of intracranial artery blood flow on the screen. A pulsed Doppler sample was placed on the longitudinal axis of the basilar trunk, and blood flow velocity waveforms were acquired. Heart rate, peak systolic, end-diastolic, and time-averaged mean blood flow velocities (mBFVs) were documented from the pulsed Doppler spectrum measurements. Recordings were repeated (1) under air breathing conditions and (2) 5 min after mice began breathing a gas mixture comprising 16% O₂, 5% CO₂, 79% N₂. This hypercapnic normoxic environment triggers metabolic- and endothelial-mediated arteriolar vasodilation that increases blood flow velocities upstream in the basilar trunk. Cerebral vasoreactivity was then estimated as the percentage increase in mBFV recorded under the gaseous mixture compared to mBFV recorded in air. Cerebral vasoreactivity was measured in both APP-PS1 (n = 8) and WT (n = 6) mice in basal conditions and after acute endothelin receptor blockade (bosentan 50 mg/kg, intraperitoneal injection, 5 min before exposure to a second gaseous mixture). No mice were discarded.

The mice were anesthetized with isoflurane (2.5%) and euthanized with CO₂. The brain and mesentery were rapidly removed and placed in an ice-cold physiological salt solution [27]. Two segments of the PCA and two segments of second-order mesenteric resistance arteries (MRAs) were collected for functional analysis. Six APP-PS1 mice and 12 WT mice were used for this protocol. No mice were discarded.

In another series of experiments, 12 APP-PS1 and 12 WT mice were treated with the endothelin receptor blocker bosentan (50 mg/kg/day for 28 days, oral administration). They were compared to 12 APP-PS1 and 12 WT mice. No mice were discarded.

Arterial segments, with internal diameters of approximately 200 μm, were cannulated at both ends on glass microcannulas and mounted in a video-monitored perfusion monitor (Living System, LSI, Burlington, VT, USA) [28]. Individual artery segments were bathed in a 5 mL organ bath containing a physiological salt solution (pH: 7.4, pO₂: 160 mm Hg, and pCO₂: 37 mm Hg), and each artery was perfused using two peristaltic pumps: one controlling flow rate and the other controlled by a pressure-servo control system. Pressure was set at 75 mm Hg, and flow rate (3–50 μL per min; increments of 3, 6, 9, 12, 15, 30, and 50 μL/min) was generated through the distal pipette with the peristaltic pump. FMD was determined both with and without bosentan in the organ bath (1 μmol/L, 30 min). Finally, arteries were bathed in a Ca2+-free physiological salt solution containing ethylene-bis-(oxyethylenenitrolo) tetra-acetic acid (2 mmol/L) and sodium nitroprusside (10 μmol/L). Pressure was then increased incrementally from 10 to 125 mm Hg in the absence of flow to determine the passive arterial diameter and mechanical properties of the vessel wall. In another segment of the PCA, a cumulative concentration-response curve to acetylcholine (ACh) was determined after precontraction of the artery with phenylephrine (1 μmol/L).

Segments of the PCA and second-order MRA were collected to assess gene expression by quantitative polymerase chain reaction after reverse transcription of total RNA (RT-qPCR). Six PCA and MRA from both APP-PS1 and WT mice were harvested and stored at -20°C in RNAlater Stabilization Reagent (Qiagen, Valencia, CA, USA) until use. RNA was extracted using the RNeasy^® Micro Kit (Qiagen, Valencia, CA, USA) following the manufacturer’s instructions. RNA extracted (200 ng) was used to synthesize cDNA using the QuantiTect^® Reverse Transcription Kit (Qiagen, Valencia, CA, USA). RT-qPCR was performed with Sybr^® Select Master Mix (Applied Biosystems Inc., Lincoln, CA, USA) reagent using a LightCycler 480 Real-Time PCR System (Roche, Branchburg, NJ, USA). Primers were validated by testing the efficiency of the PCR standard curve. Mouse Primers (Eurogentec) were: Edn1 (forward: tgc​tgt​tcg​tga​ctt​tcc​aa, reverse: ggg​ctc​tgc​act​cca​ttc​t), Ednra (forward: ggg​cat​cac​cgt​ctt​gaa, reverse: gga​agc​cac​tgc​tct​gta​cc), and Ednrb (forward: aat​ggt​ccc​aat​atc​ttg​atc​g, reverse: tcc​aaa​tgg​cca​gtc​ctc​t). Hprt (forward: aag​aca​ttc​ttt​cca​gtt​aaa​gtt​gag, reverse: aag​aca​ttc​ttt​cca​gtt​aaa​gtt​gag), Gapdh (forward: ccg​ggg​ctg​gca​ttg​ctc​tc, reverse: ggg​gtg​ggt​ggt​cca​ggg​tt), and Gusb (forward: ctc​tgg​tgg​cct​tac​ctg​at, reverse: cag​ttg​ttg​tca​cct​tca​cct​c) were used as housekeeping genes. Analysis was not performed when Cq values exceeded 35. Gene expression was quantified using the comparative Cq method. The results were expressed as: 2^{(Ct target−Ct housekeeping gene)}.

The results are expressed as means ± SEM. The significance of the differences in CVR between groups was determined by ANOVA for repeated measurements and by two-way ANOVA and the Bonferroni test to assess intergroup differences. Statistical analyses were performed on the absolute values of mBFVs and on the delta variations of mBFVs observed in the mice, first without and then with bosentan. The significance of the intergroup differences in FMD was determined by two-way ANOVA for consecutive measurements and the agonist-mediated concentration-response curves followed by the Bonferroni test. A two-tailed Mann-Whitney test (comparing 2 groups) or a Kruskal-Wallis test (more than 2 groups) was used for other comparisons as specified in the figure legends. Probability values of less than 0.05 were considered significant.

In line with the findings of our previous study, although systolic blood pressure was slightly and similarly reduced during isoflurane anesthesia in both groups, CO₂ breathing had no effect on arterial pressure in either group (data not shown) [9]. We previously reported that in both WT and APP-PS1 mice, CO₂ breathing induced significant arterial acidosis and hypercapnia with no changes in PO₂, i.e., normoxic hypercapnia [22, 23]. Under control conditions, before CO₂ breathing, no differences in basilar trunk BFV were observed in either group: WT, 12.3 ± 0.5 cm/s; APP-PS1, 11.6 ± 0.4 cm/s (p = 0.126). CO₂ breathing was followed by significant vasodilation in the basilar trunk in WT mice (+12.3 ± 2.4%; p < 0.0001; n = 6), whereas in APP-PS1 mice, there was significant vasoconstriction (−11.4 ± 1.2%; p < 0.0001; n = 8). The difference in CVR between WT and APP-PS1 mice was highly significant (p < 0.0001) (Fig. 1).

After acute treatment with bosentan, BFV was significantly increased in APP-PS1 mice (+10.2 ± 2.2%; p < 0.0001 vs. hypercapnic conditions alone) but not modified in the WT group (Fig. 1). Thus, endothelin receptor blockade normalized CVR in APP-PS1 mice: there were no significant differences in CVR between control values in WT and bosentan values in APP-PS1 mice.

Compared to WT mice, baseline FMD in APP-PS1 mice was significantly attenuated in isolated PCA but not in MRA (Fig. 2a, b). In contrast, ACh-mediated dilation was similar in APP-PS1 and WT mice in both vessel types (Fig. 2c, d). Phenylephrine-induced (1 μmol/L) contraction in PCA and MRA was also similar in both APP-PS1 and WT mice (Fig. 2e, f).

After acute treatment with bosentan (1 μmol/L, 30 min in organ bath), FMD remained unchanged in the PCA and MRA isolated from both APP-PS1 and WT mice (Fig. 2a, b). Although the difference between APP-PS1 and WT mice was significant (0.0187), this was not the case between APP-PS1 and bosentan-treated APP-PS1 mice (p = 0.1781). Nevertheless, the difference in FMD between APP-PS1 and WT mice was abolished by acute bosentan (p = 0.4478). Acute treatment with bosentan had no significant effect on ACh-mediated dilation (Fig. 2c, d) or phenylephrine (1 μmol/L)-mediated contraction (Fig. 2e, f) in either PCA or MRA.

After chronic treatment with bosentan, FMD was similar in PCA and MRA isolated from APP-PS1 and WT mice with a significant difference between WT and APP-PS1 mice and between APP-PS1 and bosentan-treated APP-PS1 mice (Fig. 3a, b). Similarly, ACh-mediated dilation and phenylephrine (1 μmol/L)-mediated contraction were similar in PCA and MRA in both groups of mice (Fig. 3c, d). Measured over pressures ranging from 10 to 125 mm Hg, arterial diameter remained equivalent in APP-PS1 and WT mice, irrespective of endothelin receptor blockade (Fig. 3g, h).

ET-1 (Edn1), endothelin receptor type A (Ednra), and endothelin receptor type B (Ednrb) gene expressions were determined in PCA and MRA isolated from APP-PS1 and WT mice. Two-way ANOVA analysis showed a significant increase in Edn1 and Ednrb gene expression in PCA and MRA with no increase in Ednra (shown in Fig. 4). Intergroup comparisons (Bonferroni test) showed a significant increase in Ednrb gene expression in MRA from APP-S1 mice compared to WT with no increases in the other groups (shown in Fig. 4).

Our findings provide compelling evidence, both in vivo and ex vivo, that cerebral vasoreactivity is impaired in APP-PS1 mice, as evidenced in vivo by a significant reduction in CO₂-induced increases in basilar trunk blood flow and ex vivo, by attenuated FMD in cerebral arteries. Both parameters improved or were even normalized by endothelin receptor blockade.

As previously reported, CO₂ breathing induced significant arterial acidosis and hypercapnia in both WT and APP-PS1 mice, with no changes in PO₂ or SBP, thus suggesting the occurrence of normoxic hypercapnia [26, 29]. Again, in line with earlier findings, basilar trunk blood flow in APP-PS1 mice did not increase but decreased markedly [9]. This suggests not only reduced endothelium-dependent dilation of cerebral arteries in this model of AD but also paradoxical cerebral arteriolar vasoconstriction downstream from the site of Doppler blood flow measurements. This is supported by our finding that FMD was selectively reduced in the PCA of APPS-PS1 mice. Interestingly, there were no alterations in agonist (ACh)-dependent dilation or contractility, pointing to potential changes in the integrity of the endothelium rather than in the vasodilator machinery, which retains its capacity to respond to external stimuli such as ACh [30, 31]. The reduction in FMD in APP-PS1 mice was significant in PCA but not in peripheral arteries (MRA), suggesting that the endothelium was altered in cerebral but not in mesenteric arteries. FMD plays a key role in controlling local CBF. In cerebral circulation, a network of interconnected surface vessels has the ability to redirect blood flow to extend the area of perfusion in response to metabolic demands [32]. Reduced flow responsiveness may therefore have damaging consequences for cerebral tissue. Indeed, reduced FMD has been evidenced in patients with hypertension, obesity, or diabetes [33].

This attenuation of blood flow velocity in vivo and FMD in vitro was abolished by the endothelin receptor blocker bosentan. The acute effect of bosentan on basilar blood flow in vivo suggests that ET-1 is produced by the endothelium in response to changes in flow induced by the hypercapnic challenge. Although a direct effect of CO₂ on ET-1-mediated contraction has been previously shown [34], our ex vivo experiments, showing that bosentan also improved FMD in isolated mouse cerebral arteries, suggest a direct interaction between ET-1 production and impaired blood flow regulation in APP-PS1 mice.

Given that the effect of bosentan on FMD was not observed in MRA, this ET-1-dependent reduction in FMD likely occurs in cerebral arteries only. It has been suggested that CO₂ affects endothelial production of ET-1 [34]. However, the evidence obtained from isolated PCA is independent of the effect of CO₂ on the endothelium. ET-1 is produced after the cleavage of its precursor by the endothelin-converting enzyme (ECE) 1 and 2. ECE2 is an isoform produced by the neuronal cells in the brain and also an Ab-degrading enzyme that has been shown to be upregulated in AD patients [35]. In physiological conditions, the ECE2/ET-1 axis is downregulated by shear stress generated by blood flow at the surface of the endothelium [36, 37]. Low shear stress or impaired flow occurring in AD may activate ECE2 overexpression, thus leading to excessive ET-1 production in response to increased flow [38]. This is in line with the findings from earlier studies that confirmed both upregulation of ET-1 and elevated ECE1/ECE2 expression in postmortem temporal cortex tissue from AD patients [39]. Similarly, in blood vessels isolated from postmortem brains, ECE activity and ET-1 levels are increased in AD patients [40]. Thus, the involvement of ET-1 in the attenuated FMD observed in the present study is likely the consequence of a dysregulated response of cerebral endothelial cells to blood flow. Indeed, the characteristic inflammation and oxidative stress in the vascular system of AD patients [41] disrupt flow responsiveness. Another explanation may be that locally reduced blood flow in AD also affects the phenotype of endothelial cells [42]. The attenuated FMD in AD is therefore likely due to excessive production of ET-1 in response to altered shear stress.

Our choice of a dual ET_A/ET_B receptor antagonist was founded on the assumption that blockade of ET_B receptors would be advantageous in our attempt to restore CBF. In healthy conditions, the vasoconstrictive activity of ET_A receptors on smooth muscle cells is countered by endothelial ET receptors. In pathologies associated with endothelial dysfunction, the vasodilatory activity of endothelial ET_B receptors is lost, whereas ET_B receptors expressed on smooth muscle cells mediate similar vasoconstrictive activity to that of ET_A receptors. Based on these observations, blockade of both ET_A and ET_B receptors should trigger more notable vasodilation (than an ET_A-selective antagonist alone) in pathological states than in healthy individuals. This has been confirmed in clinics where ET_A-selective blockade has been seen to be more effective in healthy volunteers [43] (the vasodilating ET_B receptor remains unblocked), while the efficacy of dual blockade is greater in patients with hypertension, insulin resistance, or overweight. These observations highlight the contributions of both receptors to vasoconstriction in pathological situations [20‒22]. Although no ETA-selective blocker was tested, our data demonstrate the ability of dual blockade to oppose vascular ET-1-mediated effects in the context of AD. In experimentally induced diabetes, further beneficial effects of dual endothelin receptor blockade on cognitive impairments, such as learning and memory, were also associated with improved endothelial function, suggesting that early endothelial protection may provide cognitive protection [44].

The role of the ET system in the later stages of AD and how it contributes to sustained impairment of vascular tone were not investigated in this study. Nevertheless, it is important to note that elevated plasma ET-1 levels have been reported in elderly patients as well as in most cardiovascular conditions, i.e., risk factors for AD, suggesting a likely contribution of the ET system to the vascular abnormalities observed in patients with AD [19].

Another limitation of this study is that only male mice were included, even though AD is more common among women than men [45]. Estrogen and estrogen receptors, and the estrogen receptor alpha in particular, play a key role in the regulation of vascular tone and exert a protective role against cardiovascular disorders [46]. Interestingly, the membrane-bound estrogen receptor alpha affects FMD in the peripheral resistance arteries of both male and female mice [47]. Future studies focusing specifically on the role of this receptor in FMD in cerebral arteries and the role of ET-1 in female APP-PS1 mice will provide critically needed data.

The present study suggests that cerebral circulation and regulation are impaired, both in vivo and in vitro, in male AAP-PS1 mice. The role of ET-1 is clearly foremost in the cerebrovascular dysfunction that leads to reduced CBF. Excessive production of ET-1 in response to increased blood flow in cerebral arteries is one likely explanation, given that blockade of ET-1 receptors normalized the response of these arteries to changes in blood flow. Although this study was performed in male mice only, we nevertheless suggest that endothelin receptor blockade may be a promising therapeutic strategy to minimize cerebrovascular dysregulation in patients with AD.

The authors thank Moyra Barbier for editorial assistance.

All animal procedures were approved by the Local Committees on Animal Research and Ethics in France: Comité d'Ethique en matière d'expérimentation animale Paris Descartes. CEEA-035 (protocol number 01927.02), Comité d’Ethique en Expérimentation Animale des Pays de la Loire (protocol numbers APAFIS#2413-2015102313463997v2 and APAFIS# 2018051412442187v,3), and Comité d’Ethique en Experimentation Animale Lariboisière Villemin n°9 (protocol numbers APAFIS#03560.01).

Marc Iglarz is a former employee of Actelion Pharmaceuticals Ltd., manufacturer of bosentan. The other authors have no conflicts of interest to declare.

No funding was received for this work.

Bernard Lévy, Philippe Bonnin, Marc Iglarz, and Daniel Henrion all made major contributions to the concept and design of the study. Anne-Laure Guihot, Emilie Vessieres, and Philippe Bonnin conducted the experiments and analyzed the data. Bernard Lévy, Philippe Bonnin, Marc Iglarz, and Daniel Henrion wrote the manuscript. All authors have read and approved the final manuscript, accept public responsibility for the work, and have full confidence in the accuracy and integrity of the work of each of the named authors.

Daniel Henrion and Philippe Bonnin contributed equally to this work.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.