Cerebrospinal fluid (CSF) and brain tissue was collected at necropsy from rhesus macaques (Macaca mulatta) that had been involved in various unrelated Institutional Animal Care and Use Committee approved research projects. Previously, the animals had been maintained on photoperiods comprising 12 h light and 12 h of darkness per day, and cared for in accordance with National Research Council’s Guide for the Care and Use of Laboratory Animals. Euthanasia was performed following an established protocol recommended by the American Veterinary Medical Association Guidelines for the Euthanasia of Animals, and post-mortem tissues, including CSF, were subsequently obtained through the ONPRC Tissue Distribution Program.

In Experiment 1, CSF was collected at necropsy from the cisterna magna, from 44 animals (males and females), aged 8–31 years. The samples were snap frozen in liquid nitrogen, stored at −80°C, and subsequently assayed for Aβ₄₀ and Aβ₄₂ by Myriad RBM (Austin, TX, USA) using Luminex technology and Human CustomMAP (HMPC109). Both Aβ₄₀ and Aβ₄₂, concentrations were measured as single determinations. The least detectable doses (LDD) were 0.018 and 0.055 ng/ml, respectively, and the intra-assay coefficients of variation were 6.5 and 4.0%.

At necropsy, brains were flushed with 0.9% saline, and the left temporal lobes dissected and immersion fixed in 4% paraformaldehyde for 5–9 days at 4°C. Next, the tissue blocks were rinsed in 0.1 M phosphate buffer and immersed in cryoprotectant (0.1M phosphate buffer with 10% glycerol and 2% DMSO) for 1–2 days, and then immersed in cryoprotectant containing a higher of glycerol (0.1M phosphate buffer with 20% glycerol and 2% DMSO) for 3–4 days. The tissue blocks were then flash frozen in isopentane at −75°C, and stored at −20°C. Frozen coronal sections (30 μm) were subsequently cut through the amygdala and at least 4 sections (spaced 900 μm apart) from each animal were immunohistochemically stained for Aβ, to serve as biological replicates.

The immunohistochemistry was performed on free-floating amygdala sections using a standard avidin-biotin-peroxidase procedure (VECTASTAIN ABC kit; Vector Laboratories, Burlingame, CA, USA) and 3,3’-diaminobenzadine tetrahydrochloride (Sigma-Aldrich, St. Louis, MO, USA) as the chromogen. Importantly, the procedure involved incubation of sections with one of two widely used primary mouse monoclonal antibodies against Aβ, either 10D5 (Creative Biolabs, Shirley, NY, USA) or 4G8 (Biolegend, San Diego, CA, USA), at a concentration of 1:5,000 for ∼48 h at 4°C. In both cases, goat anti-mouse IgG-biotin was used as the secondary antibody (Jackson ImmunoResearch, West Grove, PA, USA).

In Experiment 2, Aβ antibody 10D5 was used to immunohistochemically stain amygdala sections from 16 adult males and females, with 4 representatives from each of the following age groups: (1) < 9 years, (2) 10–19 years, (3) 20–29 years, and (4) > 30 years.

In Experiment 3, Aβ antibody 4G8 was used to immunohistochemically stain amygdala sections from 6 young (8–15 years) and 6 old (23–28 years) males that had previously served as controls in an unrelated study (Urbanski, 2017).

In Experiment 4, Aβ antibody 4G8 was used to immunohistochemically stain amygdala sections from 12 ovariectomized females, aged 20–28 years, that had been maintained on a regular monkey chow diet. As previously described (Kohama et al., 2016), all of the animals had been ovariectomized approximately 48 months prior to necropsy and 5 of them received immediate estradiol hormone replacement therapy (HRT) in the form of subcutaneous elastomer implants; the implants were replaced every few months so as to maintain serum estradiol concentrations at levels slightly above those typically observed during the late follicular phase of the menstrual cycle (Downs and Urbanski, 2006; Sorwell and Urbanski, 2013).

In Experiment 5, Aβ antibody 4G8 was used to immunohistochemically stain amygdala sections from 14 ovariectomized females, aged 20–23 years, that had been maintained on a high-fat, high-sugar Western-style diet (WSD), starting at the time of surgery and continuing until time of necropsy approximately 30 months later (Urbanski et al., 2017; Purnell et al., 2019). Five of these animals began receiving estradiol HRT immediately in the form of subcutaneous elastomer implants that were replaced every few months, so as to maintain plasma estradiol concentrations at a mid- to late late-follicular level; in 4 other animals, the estradiol HRT was delayed for 12 months after ovariectomy as part of the original experimental design (Purnell et al., 2019).

In Experiments 2 and 3, four immunohistochemically-labeled coronal amygdala sections were analyzed from each animal, while in Experiments 4 and 5, five sections were analyzed; the selected section were spaced at 900-μm intervals, centered around the central amygdala. The slide-mounted stained sections were scanned using a Leica Aperio AT2 Slide Scanner (Leica Microsystems, Buffalo Grove, IL, USA). Next, the region of interest (ROI) was selected, using QuPath 0.3.2, and exported to NIH Image J 1.53k using the Extensions feature in QuPath. In Image J, the amygdala was carefully outlined using the Polygon Tool and transformed to an 8-bit, black and white image; the threshold of this image was adjusted until the amyloid plaques became visible. Analysis of the particles was then performed using a particle size threshold set above 250 pixels; this enabled the detection of Aβ plaque density without interference of cellular membrane-bound amyloid precursor protein. The pixel area of each plaque in the amygdala region was summed to calculate the total area covered by plaques; this result was then multiplied by 100 and divided by the total area of the amygdala to obtain the percentage of the amygdala area covered by Aβ plaques. This was then averaged across the amygdala sections from each animal.

Results were analyzed using GB-STAT software (Dynamic Microsysyems, Silverspring, MD, USA). Spearman’s correlation was used to analyze relationships between the following: (1) Aβ₄₀ and Aβ₄₂ concentrations; (2) age and Aβ₄₀ concentrations; (3) age and Aβ₄₂ concentrations; and (4) age and Aβ₄₂ to Aβ₄₀ concentration ratios. Mean concentrations of Aβ₄₀ and Aβ₄₂ in the CSF were compared using paired Student’s t-test, while group means in experiments involving immunohistochemistry were compared using unpaired Student’s t-test.

Concentrations of Aβ₄₀ and Aβ₄₂ in the CSF were significantly (p < 0.001) correlated but showed increased divergence at higher concentrations (Figure 1A). Although no obvious age-related increase or decrease in CSF concentrations was detected in either form of Aβ (Figures 1B, C), the mean concentration of Aβ₄₀ was significantly (p < 0.001) higher than that of Aβ₄₂. The CSF concentration ratio of Aβ₄₂ to Aβ₄₀ showed only a marginally significant (p = 0.049), age-related increase (Figure 1D).

Aβ plaques are usually undetectable in the brains of rhesus macaques before the age of ∼20 years (Uno, 1993). In the present study, no animals younger than 20 years had Aβ plaques that exceed 0.1% of the amygdala area, whereas all of the animals older than 30 years did (Figure 2). Consequently, we considered 0.1% to represent a significant threshold in Aβ plaque development. Moreover, because animals aged 20–29 years showed variable levels of Aβ expression, this age group served as the focus of the subsequent experiments involving manipulation of diet and the sex-steroid environment.

Firstly, we wanted to corroborate the age-related increase in Aβ plaque density, and so we examined Aβ plaque density in another cohort of animals, comprising 6 young (8–15 years) and 6 old (23–28 years) males, taking advantage of available archived brain tissue from an unrelated study. The immunohistochemistry in this and all subsequent experiments utilized an alternate widely used Aβ primary antibody (4G8), instead of 10D5. As expected, none of the young males showed a significant number of Aβ plaques in the amygdala. In contrast, 50% of the older animals showed some plaques but in only one animal did the plaque density occupy > 0.1% of the amygdala area (Figures 3A, B).

Examples of positive Aβ immunohistological staining in the amygdala of aged ovariectomized female rhesus macaques are shown in Figure 4. Regardless of dietary manipulation (i.e., standard monkey chow or WSD), there was a clear difference in Aβ plaque density observed in ovariectomized (Ovx) animals compared to those receiving estradiol HRT (Ovx + E) (Figures 4A, C vs. Figures 4B, D). More than half of the Ovx animals in both dietary treatment groups showed significant amygdala expression of Aβ plaques, whereas only one Ovx + E animal did (Figures 5B, C), similar to the old males (Figure 5A). Because the Ovx animals from the two dietary studies (i.e., regular diet and WSD) had significantly different post-ovariectomy intervals (i.e., ∼48 months versus ∼30 months) it was not possible to determine with any certainty if the amygdala plaque density was significantly enhanced by a WSD. Overall, however, there appeared to be no obvious effect of diet, and so the data from the two dietary studies were pooled. Despite similarity in mean age between the Ovx and Ovx + E animals (Figure 6A), the density of Aβ plaques was significantly (p < 0.001) lower in the latter group and more similar to that observed in the age-matched gonad-intact males (Figure 6B).

Alzheimer’s disease (AD) is the most common form of dementia, affecting about 6.7 million people in the USA aged over 65 years; it also remains the fifth-leading cause of death among this age group (2023). And yet, there is still no effective cure for AD. Some pharmaceutical interventions, such as donepezil, provide temporary relief from AD symptoms, by blocking the breakdown of acetylcholine (Marucci et al., 2021; Rong et al., 2021). While others, such as memantine, regulate the activity of the excitatory neurotransmitter glutamate (McKeage, 2009; Kishi et al., 2017; Guo et al., 2020). However, none of these interventions has been shown to effectively retard the progression of the disease. More recently, aducanumab and lecanemab, two monoclonal antibody-based therapies designed to decrease the amount of Aβ plaques in the brain have become available (Sevigny et al., 2016; Shi et al., 2022; Mead and Fox, 2023; van Dyck et al., 2023). However, their effectiveness at improving cognitive function remains unclear and there is concern about potential negative side effects (Tian Hui Kwan et al., 2020; Avgerinos et al., 2021; Decourt et al., 2021).

The main goal of the present study was to investigate the therapeutic potential of endocrine-based interventions at reducing the amount of Aβ plaques in the brain, namely, estradiol-based hormone replacement therapy (HRT). Because of the conflicting evidence from clinical studies regarding the beneficial effects of estrogens at reducing the risk of AD (Tang et al., 1996; Kawas et al., 1997; Yaffe et al., 1998; Seshadri et al., 2001; Zandi et al., 2002; Shumaker et al., 2003; Yaffe, 2003; Shao et al., 2012; Pourhadi et al., 2023), we focused our studies on a pragmatic translational non-human primate model of aging.

A more proximate goal, however, was to establish if development of Aβ plaques within the brain follows a similar time course in both rhesus macaques and humans. To do this, we first examined CSF concentrations of Aβ across the first three decades of the rhesus macaque life. As expected, based on previous non-human primate studies (Gearing et al., 1996), we found the concentration of Aβ₄₀ in the monkey CSF to be significantly higher than that of Aβ₄₂, and agreeing with clinical studies that established Aβ₄₀ to be the predominant isoforms of Aβ in humans (Gu and Guo, 2013; Dumurgier et al., 2015; Lehmann et al., 2018). Unlike AD patients (Fukumoto et al., 1996), however, the CSF concentration of Aβ₄₂ or the concentration ratio of Aβ₄₂ to Aβ₄₀ in the oldest rhesus macaques did not show a marked increase. This finding is consistent with the observation that rhesus macaques do not develop full-blown AD during their normal lifespan, despite clearly showing an age-dependent increase in Aβ plaque density and showing mild age-related cognitive decline (without neuronal loss) (Stonebarger et al., 2020, 2021).

To test whether HRT could offer protection against development of Aβ plaques in an estrogen receptor rich part of the brain (Shughrue et al., 1997; Pau et al., 1998; Gundlah et al., 2000), we used a surgically menopausal animal model. Although female rhesus macaques clearly undergo menopause during old age (Gilardi et al., 1997; Downs and Urbanski, 2006; Sorwell and Urbanski, 2013; Luna et al., 2020), the exact age of when this occurs is quite variable between animals. In contrast, ovariectomized old (i.e., > 20 years) female rhesus macaques reliably have extremely low circulating levels of estradiol, like those observed after menopause, and thus surgically menopausal animals represent a more precise experimental model. Furthermore, we examined archived brains from two independent studies involving different diets. In one study the animals had been ovariectomized and maintained on a regular monkey chow for ∼48 months, while in the other study ovariectomized animals had been maintained for ∼30 months on a high-fat, high-sugar Western-style diet (WSD); the latter being an attempt to mimic a common diet of many post-menopausal women in the USA. In both cases, Aβ plaque density was greater in ovariectomized animals that did not receive estradiol HRT, although it should be emphasized that this Aβ plaque density was much lower than what is typically observed in the brains of AD patients (Stonebarger et al., 2020, 2021). Furthermore, the attenuated plaque density in the Ovx + E animals more closely resembled the density observed in age-matched, gonad-intact males. Because old males do not show a marked attenuation of estrogen concentrations, compared to post-menopausal females (Downs and Urbanski, 2006; Urbanski et al., 2014), this finding gives further support to the view that a lack of estrogen contributes to increased Aβ plaque density within the brain. Overall, the present results are consistent with findings from previous studies using transgenic mouse models of AD, which showed sex-differences in accumulation of Aβ plaques in the brain due to estrogen and showed that administration of estradiol was associated with lower Aβ peptide levels compared to untreated controls (Tschiffely et al., 2016; Hu et al., 2023). On the other hand, the present study was performed using a long-lived primate that more closely resembles humans in terms of its behavior, anatomy, physiology and endocrinology, and therefore represents a more translational animal model of aging, especially post-menopausal women.

It is unclear from the present experiments if the HRT prevented or delayed formation of the Aβ plaques, increased their clearance, or a combination of both. It is also unclear if a similar result would have been observed had the HRT been initiated several years after the ovariectomy. On the one hand, results from the Women’s Health Initiative study suggest that many benefits of HRT may be lost if it is delayed by several years after the onset of menopause (Lobo et al., 2014). On the other hand, it is encouraging that HRT in our WSD study showed beneficial effects even in individuals in which it had been delayed for 12 months after the ovariectomy.

Taken together, the present results demonstrate that estradiol HRT can significantly reduce Aβ plaque load in the amygdala of surgically menopausal females, even when consuming a WSD. Because of concern about adverse peripheral side effects associated with conventional HRT, it is unclear if a hormonal-based therapies would gain acceptance by post-menopausal women (German-Castelan et al., 2023). One possible way to overcome this potential problem, however, could be to rely on brain-selective prodrugs, such as 10β,17β-dihydroxyestra-1,4-dien-3-one (DHED) (Tschiffely et al., 2016), as alternatives to traditionally used estrogens. Additional research would be required to demonstrate their efficacy and safety, but they may ultimately prove to be more acceptable endocrine approaches to treating or delaying the onset of AD in humans.