APOE4 negates the effects of ovarian hormones on cerebrovascular endothelial and mitochondrial function Authors: Mackenzie N Kehmeier, Alexandra Famiano, Abigail E Cullen, Thomas Leonhardt, Skylyn J Ferguson, Madeleine Snyder, Carrie E McCurdy, Daniel J Tyrrell, Nabil J Alkayed, Ashley E Walker Published: 2026 ABSTRACT AbstractAbstractThe APOEε4 allele and oestrogen deficiency independently predispose females to an increased risk of vascular and metabolic impairments, but their cerebrovascular effects are less understood. The purpose of this study was to determine the interaction between APOE genotype and oestrogen on cerebrovascular endothelial and mitochondrial function. We studied young female homozygous APOEε3 and APOEε4 mice (n = 19–20/group; ∼6 months old) that were fed a high‐fat diet and were ovariectomized (OVX), OVX and supplemented with 17β‐oestradiol, or left intact. In APOEε3 mice, OVX was associated with impaired posterior cerebral artery endothelium‐dependent dilatation, which was rescued by 17β‐oestradiol. However, in APOEε4 mice, there was no effect of OVX or 17β‐oestradiol on cerebral artery endothelial function. Carotid artery passive stiffness was greater with OVX and lower with 17β‐oestradiol treatment in APOEε3 mice, but there was no impact of OVX or 17β‐oestradiol in the APOEε4 mice. In cerebral arteries and arterioles, 17β‐oestradiol led to higher mitochondrial complex I respiration in APOEε3 but not APOEε4 mice. These functional differences were concomitant with group differences in mitochondrial DNA copy number, antioxidant enzymes and pro‐inflammatory factors. Overall these results indicate that the APOE genotype modulates the impact of OVX and oestradiol on the cerebral vasculature. We found that 17β‐oestradiol enhances cerebrovascular endothelial and mitochondrial function in OVX APOEε3 mice but not in APOEε4 mice. This suggests that 17β‐oestradiol supplementation may have more cerebrovascular benefits for APOEε4 non‐carriers. Key points Females have twice the risk of Alzheimer's disease than males, and the APOEε4 genetic variant has a greater risk for Alzheimer's disease than the APOEε3 variant.The risk for Alzheimer's disease increases after menopause in females, suggesting that the loss of female sex hormones may play a role.There are highly inconsistent results among past studies examining the interaction between APOE genotype and oestrogens on brain outcomes, and their impact on the vasculature has not been studied.We aimed to determine the impact of APOEε4 genotype on the cerebrovascular response to ovariectomy and oestradiol.We found that oestradiol improved cerebral artery endothelial function and mitochondrial respiration in ovariectomized APOEε3 mice following ovariectomy.In contrast APOEε4 mice were resistant to the beneficial effects of ovarian hormones on cerebrovascular and mitochondrial function.This research suggests that APOE genotype may be a consideration when weighing the risks and benefits of prescribing hormone replacement therapy to postmenopausal females. Keywords: APOE genotype, cerebral artery, endothelium‐dependent vasodilatation, mitochondrial respiration, oestrogen Background Apolipoprotein E (APOE) genotype is the greatest genetic risk factor for late‐onset Alzheimer's disease (LOAD). The APOE polymorphic alleles APOEε2 (ε2), APOEε3 (ε3) and APOEε4 (ε4) have a worldwide frequency of 8%, 78% and 14%, respectively (Farrer et al., 1997), and the ε4 allele is associated with a 3‐ to 10‐fold greater risk for LOAD compared to the ε3 allele (Riedel et al., 2016). Two‐thirds of LOAD cases are females, with postmenopausal females carrying an ε4 allele having higher rates of LOAD compared to age‐matched males with an ε4 allele (Farrer et al., 1997), suggesting an interaction between ε4 and oestrogen deficiency. Hormone replacement therapy has benefits for postmenopausal females; however conflicting reports exist as to how APOE genotype modulates the effects of oestrogen on LOAD risk. Previous studies indicate that oestrogen has greater benefits for ε4 carriers than non‐ε4 carriers (Kantarci et al., 2016; Saleh et al., 2023; Yue et al., 2007), or similar effects between ε4 carriers and non‐carriers (Rippon et al., 2006) for brain‐related outcomes such as brain volumes, amyloid‐β accumulation and memory. Other reports suggest that hormone replacement has no effect or even negative effects on cognitive function in ε4 carriers, while having positive effects in non‐carriers (Burkhardt et al., 2004; Valen‐Sendstad et al., 2010; Yaffe et al., 2000). In addition to these contradictions in the literature, previous studies have not investigated the interactive effect of ε4 and oestrogen on cerebrovascular dysfunction, a key pathophysiological factor in LOAD. One of the first biomarkers of LOAD is altered cerebral blood flow, closely followed by metabolic changes in the brain (Iturria‐Medina et al., 2016). Disruption in blood flow is linked to endothelial dysfunction, resulting from reduced nitric oxide (NO) bioavailability, excessive reactive oxygen species (ROS) and an overall pro‐inflammatory milieu (Katusic et al., 2023; Zlokovic, 2011). As females progress through the menopause transition endothelial function declines and can be restored by treatment with oestrogen (Moreau et al., 2020). Oestrogen not only enhances NO bioavailability but also supports an antioxidant, anti‐inflammatory state and significantly influences mitochondrial function, a primary source of ROS (Moreau & Hildreth, 2014). In parallel metabolic dysfunction is a hallmark of the LOAD brain, stemming from impaired glucose delivery and utilization. This reduced glucose metabolism, detectable in the early stages of LOAD via 18F‐FDG‐PET imaging (Caminiti et al., 2018; Iturria‐Medina et al., 2016), is particularly evident in perimenopausal females carrying the ε4 allele, who exhibit a lower metabolic rate than their ε3 counterparts (Mishra et al., 2022; Mosconi et al., 2017, 2018). Additionally the ε4 allele is associated with altered mitochondrial dynamics and dysfunction in astrocytes (Schmukler et al., 2020) and neurons (Chang et al., 2005). Mitochondria influence vascular tone and health (Busija et al., 2016), possibly linking their dysfunction to the altered cerebral haemodynamics observed in ε4 carriers. Older mice (>18 months) have impaired cerebrovascular mitochondrial function compared to young mice (Sakamuri et al., 2022; Tyrrell et al., 2020). However the impact of APOE genotype on cerebral vascular mitochondrial function has yet to be investigated. Furthermore understanding the interaction between APOE genotype and oestrogen on cerebrovascular mitochondria will provide novel insight into the mechanisms underlying vascular dysregulation associated with menopause and LOAD. The purpose of this study was to investigate the interaction between APOE genotype and oestrogen status on cerebral vascular function and metabolism. We hypothesized that ε4 female mice would have worse cerebral artery endothelial function (pressure myography), worse cerebrovascular mitochondrial function (Oroboros respirometry), greater large and cerebral artery stiffness (aortic pulse wave velocity (PWV) and passive stiffness) and worse cognitive function (Morris water maze, nest building and accelerating rotarod) than ε3 female mice. We further hypothesized that ovariectomy (OVX) would cause greater deficits in these outcomes in ε4 mice compared to ε3 mice, and that these deficits would be mitigated in 17β‐oestradiol‐treated mice. However, in contrast to our hypotheses, we found that ε4 conferred resistance to the effects of ovarian hormones on cerebral endothelial and mitochondrial function. Methods Animal characteristics All animal procedures conformed to the Guide for the Care and Use of Laboratory Animals (eighth edition, revised 2011), ARRIVE guidelines 2.0, and were approved by the Institutional Animal Care and Use Committee at the University of Oregon. We purchased ε3 (JAX#02 9018) and ε4 (JAX#02 7894) mice from the Jackson Laboratories and established breeding colonies. We studied homozygous ε3 and ε4 female mice at 6 months of age. All mice were on a standard chow diet (Lab Diet, PicoLab Rodent Diet 20, 5053) until they received OVX or sham surgery at 4 months of age, after which they were fed a high‐fat, high‐cholesterol diet for ∼2 months until they were killed (Atherogenic Rodent Diet, Teklad, TD.02028; 42.6% kcal from fat). As ApoE's primary function is lipid transport, the effects of ε4 may be dependent on dietary lipids, and as such we chose to use the high‐fat, high‐cholesterol diet. Food and water were provided ad libitum, and mice were housed on a 12/12‐h light–dark cycle at 24°C. Mice were killed by exsanguination under 5% isoflurane, followed by cardiac puncture immediately prior to vascular studies. See Table 1 for animal characteristics. Table 1Animal characteristics E3 sham E3 OVX E3 OVX + oestradiol E4 shamE4 OVXE4 OVX + oestradiol n 20 1919 191919Age (months) 6.3±0.86.3 ± 0.76.0 ± 0.6 6.4 ± 0.56.6 ± 0.56.2 ± 0.6Body mass (g) b 23.7 ± 2.925.6 ± 4.824.4 ± 2.027.0 ± 4.9 * 29.0 ± 5.2 ‡ 25.8 ± 4.9Heart mass (mg) 0.12 ± 0.010.13 ± 0.020.12 ± 0.020.13 ± 0.020.11 ± 0.020.12 ± 0.01Percentage heart: body mass 0.5 ± 0.070.5 ± 0.10.5 ± 0.080.49 ± 0.080.42 ± 0.060.47 ± 0.1Liver mass (g) a 2.1 ± 0.32.3 ± 0.42.4 ± 0.4 * 2.1 ± 0.42.3 ± 0.42.4 ± 0.4Percentage liver: Body mass a , b 8.9 ± 1.59.3 ± 2.210.1 ± 1.97.7 ± 1.1 ‡ 8.1 ± 1.5 ‡ 9.7 ± 1.9Spleen mass (g) a 0.13 ± 0.020.14 ± 0.030.17 ± 0.10.11 ± 0.010.13 ± 0.020.15 ± 0.05 § , || Percentage spleen: body mass 0.57 ± 0.20.6 ± 0.20.60 ± 0.10.42 ± 0.07 * 0.5 ± 0.20.64 ± 0.3WAT mass (g) a , b 0.46 ± 0.30.84 ± 0.60.36 ± 0.20.85 ± 0.7 * 1.3 ± 0.8 ‡ 0.62 ± 0.6Percentage WAT: body mass a , b 1.9 ± 0.93.0 ± 1.71.5 ± 0.7 † 3.0 ± 1.94.1 ± 2.2 ‡ 2.1 ± 1.6 || Gastroc mass (g) b 0.14 ± 0.030.14 ± 0.030.13 ± 0.040.16 ± 0.030.15 ± 0.04 † 0.14 ± 0.04Percentage gastroc: body mass 0.59 ± 0.10.54 ± 0.10.54 ± 0.10.6 ± 0.10.54 ± 0.10.55 ± 0.1Soleus mass (g) 0.01 ± 0.0020.01 ± 0.0050.009 ± 0.0020.02 ± 0.030.02 ± 0.030.01 ± 0.004Percentage soleus: body mass 0.04 ± 0.0070.05 ± 0.020.03 ± 0.0080.07 ± 0.080.05 ± 0.070.04 ± 0.02Uterus mass (g) a 0.10 ± 0.040.04 ± 0.050.09 ± 0.080.08 ± 0.040.03 ± 0.020.07 ± 0.1 || Percentage uterus mass: body weight a 0.4 ± 0.20.18 ± 0.20.4 ± 0.40.3 ± 0.10.12 ± 0.080.31 ± 0.317β‐oestradiol (ng/mL) a , c 0.0040 ± 0.00200.0009 ± 0.0003 * 0.0007 ± 0.0007 * 0.0015 ± 0.00200.0006 ± 0.0003 * 0.0025 ± 0.0020Oestrone (ng/mL)0.0010 ± 0.00030.0009 ± 0.00030.0008 ± 0.00020.0007 ± 0.00020.0010 ± 0.00200.0020 ± 0.0010E2/E1 ratio3.6 ± 1.51.4 ± 0.61.6 ± 0.83.3 ± 0.71.1 ± 0.054.4 ± 3.3Progesterone (ng/mL) a 2.0 ± 1.91.6 ± 0.81.2 ± 0.52.9 ± 1.31.2 ± 0.51.3 ± 0.8Total cholesterol (µM) a 3897 ± 8144063 ± 6155233 ± 11034442 ± 11525027 ± 15245685 ± 1317 Note: Data are presented as mean ± SD.Abbreviation: WAT, white adipose tissue. a P < 0.05 main effect of oestrogen. b P < 0.05 main effect of genotype. c P < 0.05 interaction effect.* P < 0.05 vs. E3 sham. † P < 0.05 vs. E3 OVX. ‡ P < 0.05 vs. E3 oestradiol. § P < 0.05 vs. E4 sham. || P < 0.05 vs. E4 OVX. Animal characteristics Note: Data are presented as mean ± SD. Abbreviation: WAT, white adipose tissue. P < 0.05 main effect of oestrogen. P < 0.05 main effect of genotype. P < 0.05 interaction effect. P < 0.05 vs. E3 sham. P < 0.05 vs. E3 OVX. P < 0.05 vs. E3 oestradiol. P < 0.05 vs. E4 sham. P < 0.05 vs. E4 OVX. Ovariectomy and oestradiol implants Female mice were randomly assigned to sham (Sham), OVX or OVX supplemented with 17β‐oestradiol (OVX + oestradiol). OVX procedures were performed at 4 months of age. Before the OVX procedure mice were anaesthetized with 2% isoflurane via nose cone. Heart rate, breathing rate and body temperature were monitored throughout the surgery. A small incision (∼3 mm) was made on the posterior cervical area of the mouse, and the ovary was excised. The incision was closed, and the mice were monitored until a veterinarian approved their return to their home cage (Kehmeier et al., 2024). One week post‐OVX a 60 day slow‐release 0.36 mg 17β‐oestradiol pellet (Innovative Research of America, Sarasota, FL, USA) was inserted subcutaneously. Plasma hormones and total cholesterol Blood samples were collected via cardiac puncture with a heparin‐coated syringe. The blood was centrifuged, then plasma was collected and frozen in liquid nitrogen and stored at −80°C. Plasma hormone concentrations for 17β‐oestradiol (E2), oestrone (E1) and progesterone were analysed using liquid chromatography–mass spectrometry (LC‐MS). Total cholesterol was measured using enzyme‐linked immunosorbent assay (ELISA). All plasma analyses were performed by the Endocrine Technologies Core at Oregon National Primate Research Centre at Oregon Health & Science University. Hormone levels that were undetected were entered as 0, and the ratio of E2 to E1 was calculated and reported (see Table 1). Glucose tolerance test Mice were fasted for ∼4 h prior to the glucose tolerance test. Blood was obtained from a tail cut and assessed for baseline glucose levels using an iPet pro glucometer (Ultimed Incorporated, Excelsior, MN, USA). The mice then received 2 mg/g of 0.15 mg/µL dextrose solution via intraperitoneal injection. At 15, 30, 45, 60, 90 and 120 min postinjection blood glucose was measured from the already punctured tail cut. Glucose curves were created, and the area under the curve (AUC) was calculated. Cerebrovascular reactivity Endothelium‐dependent vasodilatation was assessed ex vivo in isolated pressurized posterior cerebral arteries (PCAs), as previously described (Cole et al., 2022; Kehmeier et al., 2022). Arteries were excised and cannulated onto glass micropipettes in a myograph chamber (Danish MyoTechnology Inc., Denmark) filled with a physiological salt solution. All arteries were preconstricted with phenylephrine (10–60 µM) to obtain 20%–40% preconstriction of luminal diameter. Increases in luminal diameter in response to increasing concentrations of endothelium‐dependent dilators acetylcholine (ACh: 1×10−9 to 1×10−4 M) and insulin (1–10,000 ng/mL) were determined. ACh and insulin responses were repeated in presence of Nω‐nitro‐l‐arginine methyl ester hydrochloride (L‐NAME, 30 min, 0.1 mm). Endothelium‐independent dilatation was measured by increases in luminal diameter to sodium nitroprusside (SNP, 1×10−10 to 1×10−4 M). See Table 2 for artery characteristics. Table 2Posterior cerebral artery characteristicsVariableE3 shamE3 OVXE3 OVX + oestradiolE4 shamE4 OVXE4 OVX + oestradiolMaximal diameter (µm)154 ± 8.0140 ± 20.0161 ± 21.7138 ± 16.1144 ± 10.1169 ± 16.8EC50 ACh, log M−7.2 ± 0.5−7.2 ± 0.4−7.5 ± 0.5−6.9 ± 0.4−7.0 ± 0.5−7.3 ± 0.5Insulin, mg/mL0.75 ± 0.50.75 ± 0.61.3 ± 0.51.1 ± 0.51.1 ± 0.51.1 ± 0.6SNP, log M−7.4 ± 0.6−7.1 ± 0.6−6.5 ± 0.6−7.7 ± 0.7−7.6 ± 0.7−7.6 ± 0.5Preconstriction (%)ACh27.8 ± 6.534.0 ±7.229.9 ± 7.731.3 ± 5.930.4 ± 5.243.4 ± 4.6Insulin28.9 ± 7.230.1 ± 13.832.7 ± 9.928.3 ± 10.230.2 ± 6.838.9 ± 10.5SNP36.5 ± 10.130.3 ± 5.134.9 ± 8.439.2 ± 15.134.1 ± 12.638.6 ± 11.5 Note: Data are presented as mean ± SD.Abbreviations: ACh, acetylcholine; SNP, sodium nitroprusside. Posterior cerebral artery characteristics Note: Data are presented as mean ± SD. Abbreviations: ACh, acetylcholine; SNP, sodium nitroprusside. Cerebrovascular mitochondrial function Ex vivo cerebrovascular mitochondrial function was measured by high‐resolution respirometry using O2k instruments (Oroboros, Innsbruck, Austria) (Tyrrell et al. 2020). Brain arteries and arterioles were carefully dissected from the brain and placed in ice‐cold BIOPS (10 mM Ca‐EGTA buffer, 0.1 µM calcium, 20 mM imidazole, 20 mm taurine, 5 mm K‐MES, 0.5 DTT, 6.56 mm MgCl2, 5.77 mm ATP, 15 mm phosphocreatine, pH 7.1), followed by permeabilization with 30 µg/mL saponin in BIOPS for 30 min and a 30 min wash in MIRO5. The arteries were then placed in the respirometry chamber. All data were collected at 37°C in a superoxygenated environment (200–400 µmol/L O2). Mitochondrial integrity was confirmed by measurement of respiratory responses to cytochrome C (inclusion threshold <15%). Each chamber received the same titration protocol probing with carbohydrate‐based substrates with the following sequential additions: 5 mm pyruvate, 2 mm malate, 10 mm glutamate, 5 mm ADP, 10 mm cytochrome c, 10 mm succinate, 1 mm FCCP, 10 µM rotenone, 10 µM antimycin A and 10 µM myxothiazol. This provides an in‐depth evaluation of cerebral artery respiratory capacity and substrate control during oxidative phosphorylation, as well as the extent of non‐phosphorylating respiratory leak and non‐coupled enzymatic capacity of the electron transport system. Once respiratory measures were completed the tissue was removed and stored at −80°C for further analysis. DNA content and mRNA expression Cerebral arteries and arterioles from respirometry were used to measure mitochondrial DNA (mtDNA) content, and an additional set of cerebral arteries was used to measure mRNA expression. Cerebral arteries from respirometry were thawed, MIR05 was removed and 200 µL of Quizol was added to the samples, and DNA isolation was performed according Quiros et al. (2017). DNA content was quantified using nanodrop (Nanodrop 2000 Spectrophotometer, ThermoFisher Scientific, Waltham, MA, USA). The DNA samples underwent real‐time PCR (Quantstudio 5, ThermoFisher Scientific) for mitochondrial DNA content using 16S and ND1, and for nuclear DNA using HK2. The ΔΔCt was calculated: ΔCt = Ct(mtDNA gene) − Ct(nDNA gene); ΔΔCt = ΔCt(sample of interest) − ΔCt(control sample), with the control being the ε3 sham. In an additional cohort of mice we conducted gene expression measurements to identify the components of our samples. Using a similar technique as the sample collection for respirometry arteries and arterioles underwent RNA isolation, reverse transcription and quantitative PCR (qPCR) and were compared to hippocampus samples that underwent the same preparation. Gene expression was performed for the following genes: Pecam1 (endothelial cells), Acat2 (smooth muscle cells), Gfap (astrocytes), Rbfox3 (neurons) and A1f1 (microglial). The ΔΔCt, using 18S as the control gene, was normalized to the hippocampus. Expression of additional mitochondria and inflammation genes were also measured in cerebral arteries, using 18S as the control gene and normalized to the ε3 sham group. Primer sequences are available upon request. Protein expression The hippocampus was carefully dissected from the other brain tissue, and the aorta was dissected and cleaned of perivascular fat. Both were flash‐frozen in liquid nitrogen. Samples were homogenized in lysis buffer and protease inhibitor cocktail, then internally sonicated thrice at 70% power for 10 s intervals. Lysate protein concentration was assessed using a commercially available kit (Pierce BCA protein assay kit; Thermo Fisher). Bio‐Rad stain‐free TGX gels were used for protein separation. The stain was activated to ensure protein separation and then transferred onto a PVDF membrane. Membranes were imaged after transfer to verify the total protein transferred, and then blocked with 1.5% milk in TBS‐T for 20 min, followed by overnight incubation with primary antibody in PBS containing 3% bovine serum albumin (BSA). After incubation membranes were washed and incubated with a species‐appropriate HRP‐conjugated secondary antibody. After washes membranes were incubated in Azure chemiluminescence (Azure Biosystems, Dublin, CA, USA) and imaged using a ChemiDoc Imaging system (ChemiDoc, BioRad, Hercules, CA, USA). Image J was used to determine protein band density, and relative protein expression was calculated as protein density normalized to GAPDH density. Primary antibody concentrations were 1:1000 and were purchased from Cell Signaling, unless otherwise noted, and include: Gapdh (2118S) (1:2000); LRP1 (64099S); Phosopho‐Akt (Ser 473)(9271S); Akt (9272S); Citrate Synthase (14039T); NDUSF1 (70246T); NFkB1 p105 & p50 (1422‐1‐AP, Proteintech); NFkB p65 (10745‐1‐AP, Proteintech); ApoE (ab52607, Abcam). ApoE was normalized to total protein, as it is similar in molecular weight to GAPDH. Values were normalized to the ε3 sham group. The following antibodies were quantified on the same blot against the same GAPDH: citrate synthase and LRP1; NDUFS1 and NFκB p65; and NFκB p105 and p50. Arterial stiffness Aortic stiffness was measured in vivo by PWV (Kehmeier et al., 2022). Mice were anaesthetized using isoflurane (2% isoflurane in 100% oxygen) and placed supine on a temperature‐controlled heating pad (37°C). Electrodes were placed on distal portions of limbs to record ECG, whereas Doppler transducers were positioned on the aortic arch and abdominal aorta. The distance between the transducers was recorded. Using the Doppler signal processing workstation programme (DSPW; Indus Industries, Webster, TX, USA) the time for the pulse to travel from the arch to the abdominal aorta was analysed, and the time was then divided by the distance between the transducers. Two researchers analysed the files independently, and the average was taken. Estrous stage was noted for the sham mice at the time of PWV measurement. Passive arterial stiffness was measured ex vivo in the carotid artery and PCA by changes in lumen diameter and medial wall thickness during increasing intraluminal pressure after a 60 min incubation in a calcium‐free solution (Walker et al., 2015). From the stress–strain curves β‐stiffness and elastic modulus at low pressure (EMLP) (between 7 and 30 mmHg) and high pressure (EMHP) (between 50 and 75 mmHg) were calculated, as previously described (Reeve et al., 2023). Indirect calorimetry Before ex vivo studies mice were singly housed in an environmental cabinet for 5 days. Using metabolic cages (Promethion, Sable Systems, Las Vegas, NV, USA) we measured food and water intake, body weight, oxygen consumption (VO2), carbon dioxide production (VCO2) and water vapor during the 12:12 h light–dark cycles. Metascreen software (Sable Systems Inc.) was used for data collection. Data were then processed using ExpeData and Macro Interpreter (Sable Systems Inc.) and analysed on an hourly basis. VO2, VCO2 production and expenditure (kcal/h) were normalized to body weight (kg). The first 24 h was excluded from data analysis for an acclimatization period. Cognitive function Learning was assessed using the Morris water maze. Mice underwent a 3 day training period, where they performed three trials every session, with two sessions per day. All sessions were analysed and recorded using EthoVision XT12 software (Noldus, Wageningen, the Netherlands) (Duvoisin et al., 2010). Anxiety was measured using open‐field testing. Mice were placed in a 40 cm2 square arena with a white matte floor and walls. Mice were in the arena for 10 min, and their exploration was recorded using Ethovision XT12 software. The centre of the 20×20 cm square was marked in the software, and exploration criteria were defined as the amount of time the centre of the body was inside the centre square. Mice that spend more time exploring (more time in the centre of the arena) have lower levels of anxiety (Duvoisin et al., 2010). Nest building is a test of instinctual behaviour and was performed over a 12 h period. Mice were singly housed overnight in a fresh cage containing a condensed cotton nestlet. The next morning the researcher scored the nests. The same researcher scored all nests. Nests were scored on a scale of 0–5 with the following criteria: 0: nestlet is untouched; 1: nestlet has been moved but no significant tearing; 2: significant tearing, but no formation of a nest; 3: significant tearing, indication of nest or nest is in the corner of the cage; 4: significant tearing, formation of nest in the corner; 5: significant tearing, formation of nest in the corner and the nest walls are high, making a bowl‐like nest (Deacon, 2012). Motor co‐ordination testing was performed using a rotarod (47,650 Rota‐Rod NG, Ugo Basile, Gemonio, Italy) over a 2 day period. The first day was a training session, and mice were required to stay on the rod rotating at 4 rpm for 90 s. The second day consisted of three trials, with a 10 min rest between trials, where the rod accelerated during each trial from 4 to 40 rpm with a cut‐off time at 5 min. The time was recorded when the mouse fell from the rod or rotated around the rod two consecutive times (Schreihofer et al., 2010). Statistical analysis Statistical analyses were performed using GraphPad Prism 10. A two‐way ANOVA was used to determine the interaction effects and independent effects of genotype and oestrogen status. In the case of a significant F‐value post hoc analyses were performed using a Tukey correction for preplanned comparisons. A repeated‐measures ANOVA was used to determine group differences in dose–response curves. Significance was set at P < 0.05, and values are represented as mean ± SD. Outliers were identified as z score >2 and were removed from the dataset. Results Animal characteristics ε4 mice had a higher body mass than ε3 mice (genotype effect P = 0.001; Table 1). Perigonadal white adipose tissue mass had a main effect of oestrogen and genotype (oestrogen effect, P = 0.004, genotype effect, P = 0.0005) and was highest in the ε4 OVX group (Table 1). Confirming successful OVX surgeries uterus mass and plasma 17β‐oestradiol were lower in OVX mice (oestrogen effect, P = 0.002, P = 0.02). Oestradiol treatment was associated with higher liver mass (oestrogen effect, P = 0.007), spleen mass (oestrogen effect, P = 0.0002) and total cholesterol (oestrogen effect, P = 0.04). Plasma progesterone was higher in the sham groups (oestrogen effect, P = 0.04; Table 1). Cerebral artery endothelial function is impaired using ovariectomy in ε3 mice but not ε4 mice For the PCA response to ACh, ε3 OVX mice had 20% less vasodilatation than the ε3 sham (maximum: P = 0.005, dose–response: P = 0.0002) and 18% less vasodilatation than the ε3 OVX + oestradiol treatment mice (maximum: P = 0.002, dose–response: P < 0.0001; Fig. 1A,C ). Among the ε4 mice PCA vasodilatation to ACh did not differ with OVX or oestradiol status (all P > 0.05; Fig. 1B,C ). For the maximal vasodilatation to ACh there was a significant effect of oestrogen status (P = 0.008) but no effect of APOE genotype (Fig. 1C ). For the PCA vasodilatation in response to insulin there was a main effect of genotype, with the ε4 mice showing lower vasodilatation than the ε3 mice (P = 0.04); however there was only a trend for oestrogen status (P = 0.09; Fig. 1D–F ). In the presence of L‐NAME the vasodilatory response to ACh and insulin was nearly absent and did not differ between groups (dose–response and maximal responses, all P > 0.05; Fig. 1A,B,D,E ), suggesting that the group differences in vasodilatation are due to differences in NO bioavailability. Endothelium‐independent dilatation in response to SNP was not different between groups (dose–response and maximal responses, all P > 0.05), indicating that the differences in ACh‐mediated vasodilatation were due to endothelial dysfunction (Fig. 1G–I ). Sensitivity (EC50) to vasodilators and the amount of preconstriction also did not differ between groups (all P > 0.05; Table 2). Endothelium‐dependent dilatation of the posterior cerebral artery (PCA) is impaired after ovariectomy and rescued by oestradiol replacement in APOEε3 but not APOEε4 females Female homozygous APOEε3 (E3) and APOEε4 (E4) mice were studied at 6 months after receiving a sham, ovariectomy (OVX) or OVX + oestradiol. In isolated PCAs endothelium‐dependent dilatation was measured to (A–C) increasing doses of acetylcholine (ACh) and (D–F) insulin, in the presence and absence of nitric oxide synthase inhibitor, L‐NAME. G–I, endothelium‐independent dilatation in response to increasing doses of sodium nitroprusside (SNP). A repeated‐measures ANOVA was used for the dose–response curves (* vs. E3 sham, # vs. E3 OVX + oestradiol compared to E3 OVX), and a two‐way ANOVA for the maximal responses was used to determine main and interaction effects (P‐values shown above each graph if significance was reached), followed by Tukey's post hoc analysis. *P < 0.05, **P < 0.01, ***P < 0.001. n = 7–11/group. Data are presented as mean ± SD. Cerebral artery respiration is altered by oestrogen status in ε3 but not ε4 female mice Oestrogen status and APOE genotype interacted to impact CI coupled respiration (interaction: P = 0.03) and CI+CII coupled respiration (P = 0.05) in cerebral vessels (Fig. 2B,C ). This interaction effect for CI was seen when data were normalized to leak (P = 0.047) and nDNA (P = 0.04, data not shown). For CI and CI+CII coupled respiration there was a main effect of oestrogen (CI: P = 0.01, CI+CII: P = 0.003), whereas for CI there was a main effect of genotype (P = 0.02; Fig. 2B,C ). ε3 OVX + oestradiol mice had greater CI and CI+CII respiration than the ε3 sham mice (CI: P = 0.006, CI+CII: P = 0.002) and ε3 OVX mice (CI: P = 0.002, CI+CII: P = 0.002; Fig. 2B,C ). There was also a main effect of oestrogen status in uncoupled respiration, where oestradiol supplementation elevated maximal respiration, primarily driven by the ε3 OVX + oestradiol group (P = 0.01; Fig. 2D ). Interestingly among ε4 mice CI, CI+CII or uncoupled respiration did not differ with oestrogen status (all P > 0.05; Fig. 2A–D ). There were no differences in leak between groups (all P> 0.05; Fig. 2A ). A representative tracing of respirometry can be found in Fig. 2H . After the respiration experiments mtDNA was measured relative to nuclear DNA in the vascular samples. There was an interaction effect of oestrogen status and APOE genotype for mtDNA content (P = 0.03, P = 0.005; Fig. 2E,F ). ε3 OVX + oestradiol mice had more mtDNA content than ε3 shams (16S: P = 0.001, ND1: P = 0.008), ε3 OVX (P = 0.0007, P = 0.0009) and ε4 OVX + oestradiol mice (P = 0.01, P = 0.02); however mtDNA content did not differ in ε4 mice with OVX or oestradiol. Oestrogen status and APOE genotype interact to influence cerebral vascular mitochondrial function Cerebral artery mitochondrial respiration measurements for (A) leak, (B) complex I (CI) respiration, (C) CI+ complex II (CII) respiration and (D) uncoupled maximal respiration after carbonyl cyanide p‐trifluoro‐methoxyphenyl hydrazone. Mitochondrial DNA content measured as the ratio to nuclear DNA content for (E) 16S and (F) ND1. G, gene expression in cerebral arteries (CAs) and hippocampus homogenates for Acat2 (smooth muscle cells), Pecam1 (endothelial cells), Gfap (astrocytes), Rbfox3 (neurons) and A1f1 (microglia) to demonstrate that respiration samples were enriched for the vascular cells. Gene expression was normalized to hippocampal measures. H, representative tracing from the high‐resolution respirometry experiments. A one‐tailed t test was used to determine significance between CA and hippocampal samples. A two‐way ANOVA was used to determine main and interaction effects for other outcomes (P‐values shown above each graph if significance was reached), followed by Tukey's post hoc analysis. *P = 0.05, **P = 0.01. n = 5–7/group. Data are presented as mean ± SD. For the mitochondrial respiration studies our samples of dissected arteries and arterioles were primarily composed of smooth muscle cells and endothelial cells, as indicated by higher expression of Acat2 and Pecam1 in the vascular samples compared to the hippocampus (both P < 0.0001). Our vascular samples contained traces of astrocytes (Gfap), neurons (Rbfox3) and microglia (A1f1), but these were significantly lower than the hippocampus (P = 0.01, P = 0.0008, P = 0.005, respectively; Fig. 2G ). Mitochondrial factors are impacted by oestrogen status and APOE genotype in cerebrovascular tissue There was a main effect of oestrogen status on cerebral artery gene expression of peroxisome proliferator‐activated receptor gamma coactivator 1‐alpha (Pgc1α), such that oestradiol‐treated mice had lower levels (P = 0.02), but there were no effects of genotype (Fig. 3A ). For cerebral artery superoxide dismutase 2 (Sod2) gene expression there was a main effect of oestrogen (P = 0.004) and genotype (P = 0.007; Fig. 3B ). Specifically Sod2 expression was lower in ε3 OVX mice compared to ε3 sham (P = 0.02) and ε3 OVX + oestradiol mice (P = 0.007) but did not differ in ε4 mice with OVX or oestradiol. Moreover Sod2 expression was lower in ε4 sham and OVX + oestradiol mice compared to the ε3 mice of the same treatment (P = 0.007, P = 0.04). Cerebral artery gene expression of oestrogen receptor α (ERα, Ers1) did not differ by APOE genotype or oestrogen status. Protein expression of aortic phosphorylated Akt at serine 473 had a main effect of APOE genotype, being higher in ε4 mice compared to ε3 mice (P = 0.03), but the total Akt or ratio of p‐Akt:total Akt did not differ between groups (P > 0.05; Fig. 3D–F ). Hippocampal protein expression for citrate synthase and mitochondrial CI protein NDUFS1 (NADH:ubiquinone oxidoreductase core subunit S1) did not differ between groups (P > 0.05; Fig. 3G,H ). Mitochondrial‐associated genes and proteins are altered by APOE genotype and oestrogen status Cerebral artery (CA) gene expression for (A) PGC1α (Pgc1a), (B) SOD2 (Sod2) and (C) ERα (Ers1). Aorta protein expression for (D) phosphorylated Akt at serine 473 (pAkt), (E) Akt and (F) pAkt:Akt ratio. Hippocampal protein expression for (G) citrate synthase (CS) and (H) mitochondrial complex‐I protein NDUFS1. I, representative blot images are shown with their corresponding GAPDH images below. Some GAPDH images apply to pairs of proteins quantified on the same blot: CS and LRP1 (Fig. 4); NDUFS1 and NF‐κB p65 (Fig. 4). A two‐way ANOVA was used to determine main and interaction effects (P‐values shown above each graph if significance was reached), followed by Tukey's post hoc analysis. *P = 0.05. n = 3–7/group. Data are presented as mean ± SD. Cerebral artery and hippocampus inflammatory signalling are altered primarily by APOE genotype Cerebral artery gene expression of inflammatory cytokine interleukin 1β (IL‐1β, Il1b) had a main effect of genotype, being higher in ε3 mice than in the ε4 mice regardless of oestrogen status (P = 0.009; Fig. 4A ). Low‐density lipoprotein receptor‐related protein 1 (LRP1) expression was higher in ε4 mice than ε3 mice (genotype effect: P = 0.005; Fig. 4C ). Total nuclear factor κB (NF‐κB) p65 was higher in ε4 mice compared to ε3 mice (genotype effect: P = 0.03; Fig. 4D ). In contrast, NF‐κB p105 expression trended to be influenced by oestrogen status, where OVX mice had greater expression (oestrogen status effect P = 0.06; Fig. 4F ), but there was no effect of genotype. Protein expression of ApoE and NF‐κB p50 was neither influenced by genotype nor oestrogen status (P > 0.05; Fig. 4B,E ). Cerebral artery inflammation is blunted in APOEε4 females, and hippocampal inflammation and ApoE expression are altered by APOE genotype and oestrogen status A, cerebral artery (CA) gene expression for interleukin 1β (IL‐1β) (Il1b). Hippocampal protein expression for (B) ApoE, (C) LRP1, (D) NF‐κB p65, (E) NF‐κB p50 and (F) NF‐κB p105. G, representative blot images are shown with their corresponding GAPDH images below. GAPDH images apply to pairs of proteins quantified on the same blot: LRP1 and citrate synthase (Fig. 3); NF‐κB p65 and NDUFS1 (Fig. 3); NF‐κB p50 and NF‐κB p105. ApoE was normalized to total protein. A two‐way ANOVA was used to determine main and interaction effects (P‐values shown above each graph if significant), followed by Tukey's post hoc analysis. *P = 0.05. n = 3–7/group. Data are presented as mean ± SD. Large and cerebral artery stiffness is influenced by oestrogen and APOE status In vivo large artery stiffness, measured by aortic PWV, was influenced by the interaction of OVX and APOE genotype (P = 0.046; Fig. 5A ); specifically ε4 OVX mice had great aortic PWV compared to ε3 OVX mice (P = 0.02). We measured carotid artery passive stiffness ex vivo and found that β‐stiffness did not differ between any groups (all P > 0.05; Fig. 5B ). For carotid artery EMLP there was an interaction effect (P = 0.01) and main effect of oestrogen status (P = 0.01; Fig. 5C ). ε3 OVX mice had a higher carotid artery EMLP than the ε3 sham (P = 0.001), ε3 OVX + oestradiol groups (P = 0.01) and ε4 OVX (P = 0.006; Fig. 5C ). For EMHP no group differences were observed (P > 0.05; Fig. 5D ). These data were generated from stress–strain curves displayed in Fig. 5E,F . For PCA passive stiffness, β‐stiffness did not differ between groups (all P > 0.05; Fig. 5G ). There was an interaction between APOE genotype and oestrogen status for PCA EMLP and EMHP (EMLP interaction effect: P = 0.0005; EMHP interaction effect: P = 0.02; Fig. 5H,I ). In contrast to the carotid artery, PCA EMLP was lower in ε4 sham mice compared to ε4 OVX mice (P = 0.003) but was not different from the ε4 OVX + oestradiol mice (P > 0.05). These data were generated from stress–strain curves displayed in Fig. 5J,K . Large artery and posterior cerebral artery stiffness are influenced by APOE genotype and oestrogen status A, aortic stiffness measured by pulse wave velocity (PWV). B, carotid β‐stiffness, elastic modulus at (C) low pressure and (D) high pressure were calculated from (E,F) stress–strain curves. Posterior cerebral artery (PCA) (G) β‐stiffness, elastic modulus at (H) low pressure and (I) high pressure were calculated from (J,K) stress–strain curves. A two‐way ANOVA was used to determine main and interaction effects of oestrogen status and genotype on stiffness parameters (P‐values shown above each graph if significant). *P < 0.05, **P < 0.01. n = 7–18/group. Data are presented as mean ± SD. APOE genotype and oestrogen status did not influence cognitive function or whole‐body metabolism The Morris water maze test was used to assess learning, and no differences were observed between groups for the performance during trials (all P > 0.05; Fig. 6A,B ). There was no effect of APOE genotype or oestrogen status on anxiety behaviour, as indicated by the time spent in the centre of the open‐field arena (all P > 0.05; Fig. 6C ), or instinctual behaviour, measured by nest building (Fig. 6D ). There was no effect of APOE genotype or oestrogen status on motor function as measured by accelerating rotarod (Fig. 6E ), swimming velocity during water maze and/or walking velocity during open field (all P > 0.05, data not shown). Anxiety, learning, instinctual behaviour and motor function were not influenced by APOE genotype and oestrogen status Morris water maze (A) representative heat map traces for training trials. B, time to platform during training trials. C, open‐field time in centre. D, instinctual behaviour determined by the score on a nest building task. E, motor co‐ordination, measured by time spent on an accelerating rotarod. A two‐way ANOVA was used to determine main and interaction effects. *P < 0.05, **P < 0.01. n = 3–8/group. Data are presented as mean ± SD. There was a trend for in vivo oxygen consumption to be lower with OVX (P = 0.09) and for energy expenditure to be higher in ε4 mice (P = 0.10; Fig. 7A–D ). Fasting glucose and glucose tolerance did not differ by genotype or oestrogen status (Fig. 7E–H ). In vivo metabolic measures are not influenced by APOE genotype or oestrogen Measured by indirect calorimetry: (A) VO2 and (B) VCO2 normalized to body weight and (C) energy expenditure and (D) respiratory ratio (RER). (E) Fasting glucose, (F) glucose during a glucose tolerance test and (G) glucose area under the curve (AUC) during a glucose tolerance test in APOE3 mice and (H) APOE4 mice. *P < 0.05. n = 2–9/group. Data are presented as mean ± SD. Discussion Contrary to our original hypotheses these results demonstrate that female ε4 mice have resistance to the beneficial effects of ovarian hormones, particularly in relation to endothelial and mitochondrial function. In ε3 mice, OVX impaired cerebral artery endothelial function and mitochondrial respiration, which were rescued by oestradiol. However neither OVX nor oestradiol supplementation had an effect on ε4 mice for these outcomes. The impacts of APOE genotype and oestradiol on endothelial and mitochondrial function are associated with differences in mitochondrial abundance, antioxidant defences, inflammatory signalling and receptor expression. Oestradiol affected carotid artery stiffness only in ε3 mice, and in contrast to other outcomes oestradiol affected cerebral artery stiffness only in ε4 mice. APOE genotype and oestrogen status did not have strong impacts on in vivo metabolism, cognitive function or motor function. In summary we found that ovarian hormones impact cerebrovascular endothelial function, cerebrovascular mitochondrial function and large artery passive stiffness in ε3, but not ε4, female mice. ε4's resistance to oestrogen: what do we know? Several previous studies have assessed the interactive effects of APOE genotype and oestrogen on cognitive function and other neural outcomes. Oestrogen replacement therapy improves or maintains memory and brain volume in ε4 non‐carriers, but these beneficial effects are not seen in ε4 carriers (Ambikairajah et al., 2024; Burkhardt et al., 2004; Yaffe et al., 2000). In some cases oestrogen replacement therapy results in worse cognitive function and a decline in brain volume in ε4 carriers compared to non‐carriers (de Lange et al., 2020; Kang et al., 2004; Valen‐Sendstad et al., 2010). Similarly, supplementation with oestradiol is associated with better brain‐related outcomes in ε3 mice, but not in ε4 mice (Balu et al., 2024; Brown et al., 2008; Chhibber & Zhao, 2017; Klein et al., 2014; Kunzler et al., 2014; Taxier, Philippi, Fleischer et al., 2022). In some cases oestradiol had adverse effects in the ε4 mice with familial Alzheimer's disease mutations (FAD), such as more Aβ plaques after oestradiol treatment (Kunzler et al., 2014). However it should be noted that there is an inconsistency in the literature, with studies also finding that oestrogen replacement therapy improves brain outcomes in human ε4 carriers (Kantarci et al., 2016; Saleh et al., 2023; Yue et al., 2007) and ε4 mice (Balu et al., 2024; Pontifex et al., 2021; Yun et al., 2007). Other studies find no influence of APOE genotype on cognitive outcomes (Gleason et al., 2015; Jacobs et al., 1998; Lindseth et al., 2022; Ryan et al., 2009). These inconsistencies may be due to differences in age, diet or disease stage in humans or the presence of Aβ in mice. Despite the wealth of studies examining the interactions between APOE and oestrogens, no previous studies have examined vascular outcomes. The mechanisms underlying the effects of ε4 on oestradiol signalling remain unknown. Changes in oestrogen receptor signalling are a likely factor as G protein‐coupled ER (GPER), ERα and ERβ impact vasodilatory function and arterial stiffness (Chambliss & Shaul, 2002; Ogola et al., 2021; Pastore et al., 2016). Previous studies found higher ERα in the hippocampus of ε4 mice (Taxier, Philippi, York et al., 2022); however we found no differences in ERα gene expression between groups, but this does not preclude differences in receptor sensitivity. Furthermore ε4 mice have a resistance to hyperhomocysteinaemia (Seaks et al., 2022) and lipopolysaccharide in cultured microglia (Friday et al., 2025), suggesting that perhaps ε4 has a general resistance to stressors, which may extend to the effects of OVX. Thus several mechanisms may underlie the effects of ε4 in the present study, and further research is needed to test their contributions. Endothelial function Our data suggest that oestradiol supports NO‐mediated cerebral artery vasodilatation in young ε3 female mice but not in ε4 mice. However it is important to note that this pattern was found solely for ACh‐mediated vasodilatation, and not for insulin‐mediated vasodilatation. In this study, oestradiol did not rescue insulin‐mediated vasodilatation for either genotype, whereas we previously found that a high‐soy diet preserves cerebral artery dilatation to insulin after OVX in wild‐type C57BL/6 mice (Kehmeier et al., 2024). The mice in the present study were fed a high‐fat diet, whereas the previous study used a low‐fat diet, suggesting that a high‐fat diet may impact the influence of OVX and oestradiol on insulin responses. Overall our findings suggest that ε4 mice are resistant to the impacts of oestradiol on endothelial function. The impact of cerebrovascular mitochondrial function on cerebral vessel dynamics Mitochondrial health impacts vascular function by regulating the energy source for vasodilatation/constriction and producing ROS (Sure et al., 2026; Tyrrell et al., 2020), particularly in the cerebral vasculature, where brain endothelial cells have higher mitochondrial density than other vascular beds (Caja & Enríquez, 2017). High‐resolution respirometry of cerebral arteries and arterioles has been performed only once previously, to our knowledge (Tyrrell et al., 2020), highlighting the novelty of this technique. Previous studies have also examined cerebral microvessel energetics using seahorse analysis (Merdzo et al., 2019; Sure et al., 2018, 2026). Our findings suggest that oestrogen increased mitochondrial CI and CI+CII respiration, with CI driving the overall effect, and this effect is only present in ε3 mice. We also find that oestradiol supplementation in ε3 mice resulted in higher mtDNA content, suggesting increased mitochondrial number or size, a finding absent in ε4 mice. A recent study by Sure et al. demonstrated that although female brain microvessels have a lower basal metabolic rate and ATP production, they rely more on glycolysis compared to males (Sure et al., 2026). Thus it is plausible that the ATP production is preferred through glycolytic pathways in female vasculature, and changes in this would not be reflected in our high‐resolution respirometry experiments. Somewhat counterintuitively, we observed lower PGC1α gene expression in the cerebral arteries from the oestradiol‐treated groups, possibly due to negative feedback from higher mitochondrial abundance. We also found no differences in mitochondrial CI protein NDUFS1 or citrate synthase expression between groups. Although these are mitochondrial proteins, they are encoded by nDNA rather than mtDNA, which may explain the lack of group differences. Furthermore we measured these proteins in the hippocampus because the cerebrovascular tissue we collected was limited in size. Thus, although we did not find effects of oestradiol or APOE genotype on mitochondrial‐related proteins in hippocampus homogenates, it is possible that APOE genotype and oestrogen have a specific impact on the vasculature. Mitochondria, primarily CI and CIII, are the primary sites of ROS production, and mitochondrial‐derived ROS are known to impair vascular function (Kirkman et al., 2021). Although oestradiol supplementation increased CI respiration in ε3 mice, it may not raise oxidative stress due to concurrent upregulation of antioxidant enzymes (Liu et al., 2014). Mitochondrial‐associated SOD2 has previously been observed to be increased by oestrogen supplementation in mice (Liu et al., 2014), similar to our finding in ε3 mice. Interestingly the ε4 genotype was associated with lower SOD2 gene expression, suggesting reduced resilience to oxidative stress of ε4 mice. Additionally data from Sure et al. revealed that female mice had a lower percentage of OxPhos‐derived ATP compared to males, suggesting reduced flux through the electron transport system and lower proton leak (Sure et al., 2026). The lower OxPhos‐derived ATP could suggest less ROS is produced by the electron tranport system in females compared to males, however direct measures of these in APOE and oestrogen deficiency states warrant additional investigation. Other mechanisms that contribute to vascular function APOE genotype is known to influence the expression of receptors and inflammatory mediators (Jofre‐Monseny et al., 2007). Our findings indicated no difference in ApoE expression in the hippocampus, which is contrary to past studies demonstrating lower ApoE expression in ε4 compared to ε3 (Marottoli et al., 2021; Riddell et al., 2008). ApoE's primary receptor in the brain is LRP1. We found that LRP1 is elevated in the ε4 females and not influenced by oestrogen status. The ε4 variant of apoE binds less efficiently to LRP1 than ε3 (Matsuura et al., 2024), and perhaps this lower binding affinity triggered the increase in LRP1 expression. ApoE/LRP1 signalling leads to NF‐κB inhibition, and the reduced efficiency of ε4 ApoE binding to LRP1 potentially explains elevations in NF‐κB in ε4 mice in our study. Although our results for NF‐κB in the hippocampus align with previous findings, our findings of lower expression of pro‐inflammatory cytokine IL‐1β in cerebral arteries from ε4 mice are contradictory. Our findings suggest that the impact of APOE genotype on the inflammatory status of the vasculature may differ from that of the surrounding brain tissue. Arterial stiffness Greater arterial stiffness is associated with cerebrovascular dysfunction and cognitive decline (Cooper & Mitchell, 2016). We find that OVX is associated with greater carotid artery passive stiffness in ε3 mice, similar to previous studies in wild‐type mice (Kilanowski‐Doroh et al., 2024). These changes in carotid artery stiffness could influence cerebral artery endothelial function, as it was recently shown that age‐related changes in carotid compliance have a greater influence on cerebrovascular function than changes in aortic stiffness in females (Lefferts et al., 2023). In contrast, cerebral artery stiffness was affected by oestrogen status in ε4 mice but not in ε3 mice. Overall artery stiffness was the only outcome where ε4 mice were more sensitive to oestrogen status. These findings suggest that the molecular mechanisms underlying changes in arterial stiffness, at least in the context of APOE and oestradiol, are different between the carotid and cerebral arteries. Cognitive and motor function Although we found no differences in cognitive function between groups in this study, previous studies found that ε4 mice have cognitive deficits in multiple behavioural and learning tasks (van Heuvelen et al., 2024). When examining the impact of APOE genotype and oestrogen on cognitive function, a previous study found that oestrogen was associated with better performance on a Novel Object Placement test in ε4 mice, but not when mice were on a high‐fat diet, and oestrogen had no impact on ε3 mice (Christensen et al., 2024). Thus, the high‐fat diet fed to the mice in our studies may explain our lack of findings. Limitations First, although we normalized respirometry data to tissue mass, measurement error is possible due to the small size of cerebral vessels. However similar group differences were observed when normalizing to nDNA and leak (data not shown). Second, ε4 sham and OVX mice had higher body and adipose tissue mass than their ε3 comparators, which may have confounded our results. Although greater body mass is associated with vascular risk factors (Powell‐Wiley et al., 2021), adipose tissue also contains aromatase, leading to oestradiol synthesis (Lee & Den Hartigh, 2025). The ε4 + OVX mice may have had higher adipose‐derived oestrogen than the ε3 + OVX mice; however the results for circulating 17‐β oestradiol and uterus mass suggest that this was not the case. Third, due to ApoE's role in lipid transport we believe it was important to study these mice with a diet more similar in fat content to the average US adult diet, thereby using a ‘high‐fat’ diet. Thus our results are specific to this dietary context, as it is unknown whether dietary fat and cholesterol intake modulate ε4’s effects on the cerebral vasculature. Fourth, mice lack cholesterol ester transfer protein (CETP), which exchanges cholesterol and triglycerides between lipoproteins, and the absence of this protein in mice results in them carrying the majority of their lipid cargo in high‐density lipoprotein (HDL) rather than atherogenic low‐density lipoprotein (LDL) particles. Future studies should consider investigating CETP to better mirror human lipid metabolism in the context of APOE genotype. Fifth, changes in blood pressure were not measured, and changes in blood pressure could potentially explain some of our findings. Sixth, we saw a lack of an OVX effect on insulin response. Lastly, the goal of our study was to compare cerebrovascular effects of oestradiol between the human ε3 and ε4 alleles, as human ε4 is associated with higher vascular and AD risk, whereas the human ε3 is neutral to disease risk. Because murine Apoe was outside the scope of this study design, additional research is needed to understand differences between murine Apoe and human APOE. Conclusion Overall, our results indicate that APOE genotype modulates the impact of ovarian hormones on the cerebral vasculature. We found that OVX impairs and 17β‐oestradiol enhances cerebrovascular function and mitochondrial function in ε3 mice but not ε4 mice. The results suggest that ovarian hormone supplementation with oestradiol may provide greater therapeutic benefit for ε4 non‐carriers. Further research is needed to understand the mechanisms by which ε4 is resistant to ovarian hormones, which will improve our understanding of disease progression and aid in the development of more effective therapeutics for LOAD. Additional information Competing interests None declared. Author contributions M.N.K. and A.E.W. conceived and designed the research. M.N.K., A.F., A.E.C., T.L., S.J.F., M.S., D.J.T., C.M., N.J.A. and A.E.W. collected and analysed the data. M.N.K. and A.E.W. interpreted the results; M.N.K. prepared figures; M.N.K. and A.E.W. drafted the manuscript; M.N.K., A.F., A.E.C., T.L., S.J.F., M.S., C.M., D.J.T., N.J.A. and A.E.W. edited and revised the manuscript and approved the final version. Funding This work was supported by the John L. Luvaas Family Fund (A.E.W. and N.J.A.), NIH R01AG064016 (A.E.W.) and AHA 23PRE1023169 (M.N.K.). The Endocrine Technologies Core (ETC) is supported by NIH grants P51OD011092 (for operation of the Oregon National Primate Research Center) and S10OD026701. Supporting information Peer Review History Biography Mackenzie N. Kehmeier completed her PhD in human physiology and the University of Oregon and is now a postdoctoral fellow at the University of Colorado Anschutz Medical Campus in the Department of Geriatrics. During her PhD she explored the role of the sex steroid hormone oestrogen and phytoestrogens (from soybeans) on cerebral vascular function, while considering their interaction with the APOE genotype. Her current research at Anschutz Medical campus aims to improve the mechanistic understanding of oestrogen loss and the subsequent increase in adiposity on cerebrovascular and systemic micro‐ and macrovascular function in clinical and preclinical models. Handling Editors: Bettina Mittendorfer & Ming Zheng The peer review history is available in the Supporting Information section of this article (https://doi.org/10.1113/JP290852#support‐information‐section). Data availability statement All new data are reported in this paper within the text, figures and tables. References Associated Data Supplementary Materials Peer Review History Data Availability Statement All new data are reported in this paper within the text, figures and tables.