Gonadal hormones contribute to sex differences in behavior, pathology and epigenetic modifications in the 3×Tg-AD mouse model of Alzheimer’s disease

Breeding pairs of homozygous, triple-transgenic AD mice possessing PSen1_M146V, APP_swe, and tau_P301L transgenes [83] and wild type (WT) B6.129SF2/J mice were purchased at 6 weeks of age from the Jackson Laboratory (Bar Harbor, ME). All mice were bred and maintained in our vivarium, where they were group-housed (for breeding pairs, 2–3 mice/cage; for cohorts, 4 mice/cage) and kept under standard laboratory conditions and 12-hour reverse light cycle: light phase 7 P.M.−7 A.M., room temperature ∼22 °C, humidity ∼62%, low fat rodent chow and tap water available ad libitum, and Beta Chip bedding changed twice per week. All animal-related protocols were performed in accordance with the rules and regulations of the Canadian Council of Animal Care and approved by the local Animal Research Ethics Board.

Transgene constructs: Human APP (695 isoform) cDNA harboring the Swedish double mutation (KM670/671NL) and human MAPT carrying a mutation (P301L) were expressed under control of the mouse Thy1.2 expression cassette, resulting in overexpression of APP and tau [83]. For Psen1, the human PS1 M146V knock-in mutation of the mouse Psen1 gene results in expression of a presenilin-1 protein with the human familial Alzheimer’s disease-linked mutation PS1M146V at normal physiological levels (https://www.jax.org/strain/004193↗; [46]).

At 3 months of age, the mice were gonadectomised under isoflurane gas anesthesia to remove either bilateral testicles or ovaries to deplete endogenous gonadal hormones. Non-gonadectomised controls underwent sham surgeries. All mice were individually housed post-surgery until endpoint. Body weight was measured biweekly post-surgery as an important monitoring parameter.

After acclimatization and 5 days of habituation to experimenters, mice underwent a battery of behavioral tests pre- and post-surgery at 2.5 and 6 months of age, respectively. The following measures and paradigms were used sequentially: basic neurological reflexes, basket test, Rotarod test (these were used to screen for non-memory-related neurological deficits), step-down test, and Morris water maze (MWM). The last two tests were used only for post-operative animals, considering their potential impact on memory training.

Reflexes: To assess basic neurological function, mice were tested for clasping reflex, visual placing reflex, and geotaxis. All mice with surgeries, including all genotypes and sexes, showed good reflexes comparable to mice without surgeries.

Basket test: The test assesses muscle strength, motor coordination and sensorimotor deficits. A mouse was placed in the center of a short basket box (8” high, made of wire mesh) which was then slowly inverted. The time it took for the mouse to fall was recorded by stopwatch.

Rotarod test: This test uses a rotating bar (Med Associates, Fairfax, Vermont), to examine muscle strength and sensorimotor coordination. Three 5-min trials daily were given over three days, with latency and speed at fall recorded. The duration (seconds) on the rod was automatically recorded via computer.

Step-down test: The step-down test is designed to assess the readiness of a mouse to escape from an elevated platform by descending onto a firm, dark surface (usually black cardboard) in a brightly-lit room [5, 99]. This task assesses acrophobia and, as such, it is proposed to evoke an anxiety-like response to height and/or bright light [70]. A small platform (10 × 8 cm) made of wire mesh was placed on top of a 7 cm-high rectangular Pyrex glass bowl. The mouse was placed on the wire mesh, and the latency to step down with all four paws was recorded during a 10-min trial.

Morris Water Maze (MWM): The MWM is commonly used to assess visual-spatial learning and memory in rodents [27]. An 8-day MWM protocol was used to habituate the animals to the apparatus, assess their sensorimotor capacity, and examine their ability to form a spatial map [59]. The polyethylene swimming pool (dia. = 105 cm, H = ~ 60 cm, San Diego Instruments, San Diego, CA), filled with tap water at ∼24 °C, had a circular platform made of clear Plexiglas (dia. 15 cm) positioned in the northwest quadrant. Mice were trained in four cue trials (Day 1), with the platform above the water surface and a blue cylinder on the top. Starting positions (north, south, east, or west quadrant) were randomly chosen on each day. The cue trial was considered completed when the mouse found the platform or if a 2 min trial expired, in which case the mouse was placed on the platform. In both cases the mouse remained on the platform for 45–50 s to form a spatial map. On Days 2–5, the platform was hidden, and blocks of acquisition trials (4 trials/day) were given over four days. To examine whether a spatial learning strategy was employed in solving the task, a 2-min probe trial was given on Day 6, and time spent in the northwest quadrant was measured. The cued reversal trial consisted of placing the platform in the quadrant opposite to its original location (southeast quadrant), and then two acquisition trials were performed with the platform visible and two with the platform submerged. The following measures were recorded for cued and acquisition trials: latency to find the platform (either visible or submerged), distance traversed, and swimming speed.

The protocol was adapted from The Jackson Laboratory (Bar Harbor, Maine) to detect the inserted, mutated human tau gene. Crude extraction of DNA from 2 to 3 mm mouse tail was performed with REDExtract-N-Amp™ Tissue PCR Kit (Sigma-Aldrich, Oakville, ON, Canada) according to manufacturer’s instructions. Amplifications were performed in a total volume of 15 µl containing 7.5 µl of REDExtract-N-Amp PCR mix (Sigma), 0.5 µM of each primer, and 3 µl of mouse tail extract as template. The primer sequences were from The Jackson Laboratory and ordered from IDT (Coralville, IA, USA). The expected polymerase chain reaction (PCR) product of the transgene (fragment of P301L) was 254 bp long, and the internal positive control (IPC), a fragment of the mouse interleukin 2 gene, was 324 bp long. Primer information is listed in Table 1. The PCR protocol consisted of 10 cycles of touchdown (1 °C drop every 2 cycles/min of the annealing temperature from 65 °C to 60 °C [44], followed by 28 cycles of 30 s at 94 °C, 1 min at 60 °C, and 1 min at 72 °C. The final extension cycle was 10 min at 72 °C. The size of the amplification products was determined by agarose gel electrophoresis.

Mice were anesthetized with an intraperitoneal injection of a ketamine/xylazine mixture at 7 months of age. Blood (∼1 ml) was collected from the chest cavity within 15 s of severing the vena cava into a 1.5 ml Eppendorf tube. The blood was left to clot on ice for 20 min, and then serum was isolated by centrifugation at 12,298 ×g for 5 min at room temperature. Blood vessels were intracardially perfused with ∼120 ml of phosphate buffered saline (PBS) over 5 min. Extracted brains were wet weighed, dissected, and frozen until use.

The levels of testosterone and estradiol were measured in undiluted serum samples. For males, the Calbiotech Mouse/Rat Testosterone ELISA (Calbiotech, El Cajon, CA) was used according to the manufacturer’s instructions, with the standard curve extended to include testosterone levels from 0.05 to 12 ng/ml. Estradiol levels in serum of females was assayed using the Double Antibody RIA 17β-Estradiol kit (MP Biomedicals, Irvine, CA USA). The manufacturer’s protocol was followed with 2 changes: (1) volume of all reagents and samples were halved because of limited serum volume, (2) 5 pg/ml was added to the estradiol standard curve as the lowest detectable concentration.

Extraction of soluble tau and Aβ species was based on published methods [72, 98]. In brief, frozen cortical samples (approx. 100 mg) were sonicated in ∼1 ml (1:10 volume) Tris-buffered saline (TBS) pH 7.4 with protease inhibitors (Halt™ Protease Inhibitor Cocktail, Thermo Fisher Scientific, Waltham, MA), 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF, Bio Basic, New York, NY), and phosphatase inhibitors (PhosSTOP EASYpack™ phosphatase inhibitor tablets, Roche, Mississauga, ON), and kept on ice for 5–10 min. For soluble Aβ analysis, TBS homogenates were centrifuged for 20 min at 14,000 ×g at 4 °C, and supernatants (S1) were collected, aliquoted, and frozen at − 80 °C until use. For analysis of soluble tau, TBS homogenates were centrifuged for 20 min at 27,000 ×g at 4 °C, and supernatants (S1) were collected and frozen for subsequent use. Protein concentrations in each fraction were measured using a detergent-compatible protein assay (Bio-Rad Laboratories, Mississauga, ON, Canada). Transgenic (human) Aβ42 protein levels in TBS-soluble S1 fractions were measured using the Amyloid beta 42 Human ELISA Kit (Invitrogen, San Diego, CA) per manufacturer’s instructions. Transgenic (human) total tau and phospho-tau 181 were assessed with the Human Total Tau ELISA kit and the Human Tau (Phospho) [pTau181] ELISA Kit (Invitrogen, San Diego, CA). Concentrations were acquired with a MultiskanGO and SkanIt software (Thermo Scientific, Nepean, ON, Canada) at 450 nm. These ELISA kits recognize only transgenic human amyloid-beta and tau, and mouse (endogenous) gene products were not detected. Values are presented as pg of the human gene product in 1 µg or 1 mg of total protein, depending on transgene production.

ChIP and PCR were performed as previously described [79]. Briefly, cortical samples were incubated in 1% formaldehyde for 10 min at 37 °C, and 1.25 M glycine was added to quench the reaction. Next, samples were washed with PBS, and SDS lysis buffer was added to all samples prior to sonication at 40% power, 6× for 10 s with 50 s rest using a Bioruptor^® Plus sonication device (Diagenode, Denville, NJ) with a model CL-18 probe (Fisher Scientific, Ottawa, ON, Canada). Samples were centrifuged at 17,000 ×g for 10 min, aliquoted, and diluted with ChIP dilution buffer (MilliporeSigma, Oakville, ON, Canada). Samples were treated with 20 µl of Millipore Protein G magnetic beads and 1 µl of anti-macroH2A1 (Cell Signalling Technology, Danvers, MA, USA) or anti-histone H3 (MilliporeSigma) antibody overnight at 4 °C. The next day, samples were washed sequentially with low-salt (20 mM Tris-HCl pH 8.0, 2 mM EDTA pH 8.0, 0.1% SDS, 1% Triton X-100, 150 mM NaCl) and high-salt (20 mM Tris-HCl pH 8.0, 2 mM EDTA pH 8.0, 0.1% SDS, 1% Triton X-100, 500 mM NaCl) buffers, LiCl (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.25 M LiCl, 1% NP-40, 1% Sodium Deoxycholate), and then Tris-EDTA (TE) buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and incubated with rotation for 5 min between washes. Immune complexes were extracted from both ChIP and input samples using proteinase K (0.625 mg/ml in TE buffer) and heated at 65 °C for 2 h followed by 95 °C for 10 min before purification with a PCR Purification Kit (Bio Basic Inc., Markham, ON, Canada). Primers (Table 1) were designed to detect specific sequences in the promoter regions of (1) mouse Arc, Fos, Mapt, App, and (2) mouse Thy1.2 expression cassette for human App_sweP301L [83], and (3) mouse Psen1 promoter and Psen1_M146V. ChIP data were calculated as % input, then normalized against the (histone H3) control.

RNA was extracted from mouse cortical samples in TRIzol using the EZ-10 spin column total RNA extraction kit (BioBasic, Markham, ON, Canada). Complementary DNA was synthesized using a high-capacity cDNA Reverse Transcription Kit (Applied BioSystems, Waltham, MA, USA). Primers are shown in Table 1. Data were normalized to the geometric mean of β-actin and GAPDH [60, 79]. Data were then normalized to the same sex WT Sham, defined as “relative expression”.

Raw data were analyzed using SPSS 20 software package (SPSS Inc., Chicago, IL) and Prism-GraphPad (GraphPad Software, San Diego, CA). Analysis of variance (ANOVA) was used in the overall analysis, with Genotype (3×Tg-AD vs. WT), Sex (male vs. female) and Treatment (gonadectomy vs. sham) as between-group factors, and Trial or Age as within-group factors, where applicable. To reveal modulatory effects of gonadectomy on behavior, 3×Tg-AD pathology, and epigenetic factors (histone variants), the focus of this statistical analysis was on the significant interaction between Genotype and Treatment (represented as Genotype × Treatment in the following text), or interaction between Genotype and Sex and Treatment (represented as Genotype × Sex × Treatment in the following text). Given our previous extensive behavioral, neuropathological and immunological characterization of this mouse model [60, 61, 70], other significant main effects were less important for the current research question. The significance level was set at p ≤ 0.05. For statistically significant ANOVAs, post hoc tests were carried out, corrected by controlling the false discovery rate (FDR) (two-stage step-up method of Benjamini, Krieger and Yekutieli). Graphs indicate mean values ± SEM, with significant differences of q ≤ 0.05, q < 0.01, q < 0.001, and q < 0.0001 shown as *, **, *** and ****, respectively.

Body mass is an important parameter to estimate malnutrition and post-surgery complications, especially infection [65]. As shown in Fig. 1A and B, mice either gained weight or did not lose weight over time [Time: F(9, 670) = 118.7, p < 0.0001], indicating no post-operative complications, malnutrition or infection. Gonadectomised (GDX) male mice were lighter than sham-treated mice, and male AD mice were lighter than male WT mice [Fig. 1A, Time × Treatment: F(9, 423) = 19.70, p < 0.0001; Time × Genotype: F(9, 423) = 12.25, p < 0.0001, q < 0.0001 between WT Sham and GDX or AD Sham and GDX groups], indicating that this genetic model displayed growth retardation and that removal of male gonadal hormones indiscriminately affected both the 3×Tg-AD and WT control animals. In females, no significant differences were found between any groups (Fig. 1B). GDX did not cause any changes in brain weight in either males or females (data not shown).

Gonadectomy (GDX) eliminated serum testosterone in male WT and AD mice compared to the animals with sham surgeries [Fig. 1C, Treatment: F(1, 21) = 21.42, p = 0.0001, Genotype: F(1, 23) = 2.95, p = 0.099, Genotype × Treatment: F(1, 21) = 2.998, p = 0.098, q = 0.0003 between male WT Sham and GDX groups, q = 0.026 between male AD Sham and GDX groups]. GDX also significantly reduced serum estradiol in female WT and AD animals compared to those with sham surgeries (Fig. 1D, Treatment: F(1, 20) = 11.42, p = 0.003, Genotype: F(1, 28) = 28.87, p < 0.0001, Genotype × Treatment: F(1, 20) = 0.12, p = 0.735, post hoc: q = 0.03 between female WT Sham and GDX groups, q = 0.03 between female AD Sham and GDX groups). Levels of serum testosterone in Sham male AD mice and serum estradiol in Sham female AD mice were lower than in Sham WT animals (Fig. 1C, D and q = 0.02 for testosterone between AD Sham and WT Sham males, q < 0.0001 for estradiol between AD Sham and WT Sham females).

Our hormone assays show the efficacy of our surgical procedures. Testosterone was reduced to zero and estradiol was significantly reduced after removal of gonads. This difference in the degree of reduction between the sexes could be attributed to the different distributions of such hormones in male and female. In males, testosterone is primarily synthesized in the Leydig cells of the testes [32], whereas in premenopausal females, although estrogens are produced primarily in the ovaries, corpus luteum, and placenta, a small but significant amount of estrogens can also be produced by nongonadal organs such as the liver, heart, skin, and brain [26]. Furthermore, estrous cycles in female mice can significantly affect the levels of serum estradiol [81], and no synchronization of estrous cycle was carried out before endpoint in this study. This could result in a large variance in the levels of estradiol in the sham-operated animals. Nevertheless, we demonstrate a statistically significant reduction of estradiol levels in GDX compared to sham groups.

Our data suggest that gonadal hormones may possess some common functions beyond sex or sex-dependent roles. Some of our behavioral tests revealed sex-specific functions of gonadal hormones, while others did not. For instance, GDX significantly reduced body weight in males, but not females. In contrast, GDX increased anxiety (acrophobia in the step-down test) in both male and female 3×Tg-AD animals.

GDX also impacted sex-specific differences in spatial learning/memory in 3×Tg-AD animals, with detrimental effects particularly in females. Our MWM data show that removal of gonadal hormones from female 3×Tg-AD mice worsens spatial learning compared to males. These data are consistent with prior studies demonstrating loss of spines in ovariectomized female, but not male, rats [75]. In contrast, GDX had the opposite effect on male 3×Tg-AD animals, reflected in better performance on motor tasks (Rotarod) and spatial memory (MWM) following GDX, and implying different roles for male versus female steroid hormones in the regulation of learning and memory. Other studies have proposed that testosterone, by acting as a non-selective sigma antagonist, may produce a tonic dampening of the function of sigma receptors and consequently a decrease in NMDA receptor function that is known to play an important role in spatial learning and memory acquisition [29, 67]. This is in contrast to the well-recognized beneficial functions of estrogen, which increases the concentration of choline acetyltransferase [101], the enzyme needed for acetylcholine synthesis, and enhances communication between neurons in the hippocampus, resulting in improvements in hippocampal-mediated learning [87] and spatial memory [36] in rodents. Interestingly, one study showed that aged ovariectomized 3×Tg-AD female mice treated with leuprolide acetate, a drug used to block testosterone synthesis, exhibited significantly improved spatial learning, supporting the detrimental role of testosterone [85]. Several reports also indicated that chronic treatment with androgenic compounds impair spatial learning and memory in young and middle-aged animals [37, 43], and humans [42, 48].

Although multiple studies have shown that men and women differ on tasks of spatial cognition [58], as far as we know, our report is the first to find that GDX male and female 3×Tg-AD mice show sex-associated differences in spatial learning and memory as measured by MWM. Previous behavioral studies with GDX 3×Tg-AD mice revealed worse performance in the spontaneous alternation behavior test in both male and female animals [17, 96] and a distinct spatial memory deficit in females indicated by the MWM [32], but lack of male MWM data. In our study, the fact that only spatial learning, but not memory (as measured by the probe trial in MWM), was affected in female GDX 3×Tg-AD animals may be attributed to the independent actions of estrogen on different task systems [63].

Our study is among the first to demonstrate that the loss of gonadal hormones in male mice improves spatial learning. Furthermore, our results are consistent with human studies demonstrating that, in men, low testosterone is associated with better spatial abilities [20], and testosterone impairs long-term spatial memory [34, 43, 78]. Thus, one may speculate that while a sudden drop in estrogen, as occurs in female menopause, is detrimental to spatial learning and may contribute to the development of AD, the gradual loss of testosterone in men may be protective for spatial learning. In addition, sex-specific differences in spatial learning suggest that the increased severity of symptoms in females as compared to males with Alzheimer’s disease may be modulated by gonadal hormones.

Our study documents sex-dependent differences in behavioral deficits, AD-like pathology and epigenetic changes related to AD pathology. This sex-dependent shift in phenotypic traits is not isolated to a single colony, as several independent groups have documented sex differences in 3×Tg-AD behavior [13, 14, 16, 25, 40, 88, 89, 94, 105–107], AD-like neuropathology 8, [18, 50, 86, 88, 89], response to environmental enrichment, diet, or exercise [6, 38, 41, 108], lifespan [8, 40, 92], and immunity [8, 40, 41]. Male 3×Tg-AD mice exhibit altered behavior early in life, but older female mice exhibit more severe symptoms than males. Male mice exhibit a loss or delay in AD-like pathology (https://www.jax.org/strain/004807↗) in comparison to female littermates which do not show the same loss of phenotype. Although there are no tangles at 6 months of age in either sex, there is an increase in soluble, phosphorylated tau. Western blots show increased total soluble tau and pTau-181 at 6 months of age in AD females only [60]. All these findings suggest that gonadal hormones may play distinct roles influencing AD-related pathology and behavior in a sex-dependent manner.

The nature of this sex-dependent phenomenon may be epigenetic, as we previously showed that compared to WT males, 3×Tg-AD males show increased expression of macroH2A1 (formerly H2afy) [61], which codes for macroH2A1, a variant of the canonical histone H2A, important in neuroplasticity [23, 102] and often upregulated during neurodegenerative processes [33, 47, 52]). Emerging evidence suggests that histone variants are novel epigenetic regulators of memory [93], as they are critical modulators of neural plasticity [100]. As important epigenetic regulators of gene transcription, histone variant binding activities depend on their incorporation into and eviction from chromatin [97], the status of their special chaperones [71], and transcription of their coding genes. These are highly responsive to environmental stimuli, including immunosuppression [60] and age-related regulation in the brain [110]. Sex-specific effects of histone variants have also been implicated in learning and memory [93].

Following gonadectomy, the expression of the macroH2A1 gene, coding for macroH2A1, was downregulated in male AD animals, whereas there was a trend towards upregulation in AD females. This sex-specific difference indicates that gonadal hormones regulate gene expression of macroH2A1 in a sex-dependent manner in AD animals. However, increased mRNA expression does not necessarily translate into increased protein or increased incorporation into chromatin. Our current data revealed that decreased macroH2A1 expression in GDX males does not alter macroH2A1 binding at the mMapt or mApp promoters.

The presence of macroH2A1 at promoter regions has been shown to inhibit gene expression [4, 15, 21, 30]. In our study, an inverse link exists between the binding activity of repressive macroH2A1 at mMapt and expression of mMapt in female, but not male, 3×Tg-AD mice after GDX. This suggests that female gonadal hormones may suppress the expression of the endogenous tau gene through the enhanced incorporation of macroH2A1 into chromatin at the Mapt gene body. Estrogen has been indicated to play protective roles against tauopathy through increasing tau dephosphorylation in human neuroblastoma cells and primary rat cortical neurons [2, 77], and in both hippocampus and cerebellum of mouse brains [91]. MacroH2A1 can regulate target gene expression in part by recruiting the transcriptional coregulator PELP1 to promote estrogen receptor-dependent transcription [54]. Estrogen is known to alleviate neurotoxicity in the brain and protect neurons [62]. So, there is reason to speculate that, in female GDX 3×Tg-AD mice, removal of estrogen may unbalance this mechanism, resulting in elevated mMapt expression, contributing to the deterioration in spatial learning.

In contrast, there were no changes in macroH2A1 binding at the mApp gene in female GDX 3×Tg-AD mice. Nevertheless, the expression of endogenous App was significantly elevated in female GDX 3×Tg-AD mice compared to sham-operated counterparts. Contrary to female AD mice, removal of male gonadal steroids significantly reduced the expression of endogenous App, again without a change in macroH2A1 binding at App. Our data suggest that female steroid hormones reduce endogenous App and tau expression (the latter by targeting macroH2A1 as a crucial regulator of the tau gene), whereas male steroid hormones play an opposing role by increasing endogenous App, independent of macroH2A1 regulation. On the other hand, the opposing male/female difference in App expression may imply that a common regulatory mechanism for App pathology, independent of macroH2A1 binding, exists in both male and female AD brain. One possibility is aromatase, which converts androgens into estrogen in the brain, independent of gonadal sources. Our data are consistent with findings of Overk et al. [84], who demonstrated an inverse relationship between aromatase and Aβ deposition in female 3xTg-AD mice. Thus, GDX in female mice may not alter brain Aβ levels in our study due to the presence of aromatase [84]. Other studies indicate the importance of female gonadal hormones in limiting the spread of neuropathology in AD, as GDX reduces the activity of α-secretase and neprilysin and increases the activity of β-secretase [45]. Although estrogen reportedly upregulates expression of γ-secretase in cultured human neuronal and glial cells [82], no significant effect of GDX on γ-secretase protein expression was observed in sheep brain [9], consistent with our observations of mPsen1 expression in the 3×Tg-AD mouse model.

A previous study has indicated that in 3×Tg-AD animals, murine tau is hyperphosphorylated and promotes neurofibrillary tangle formation by co-aggregating with transgenic tau [7]. An in vitro study also revealed that wild type tau could alter the transcription and alternative polyadenylation profiles of numerous nuclear precursor mRNAs, which then translate to form proteins involved in chromatin remodeling and splicing [76]. Therefore, upregulated mMapt gene expression, possibly followed by a high level of endogenous tau protein expression, may contribute to maintaining a high level of tau pathology after removal of female steroid hormones. On the other hand, the upregulated mApp following GDX in females may also aggravate amyloid pathology, as a study comparing APP/PS1 mice with and without the endogenous mouse App gene indicated that the additional mouse Aβ increased soluble amyloid accumulation [57].

In contrast to the endogenous AD-related genes, binding of macroH2A1 to the Thy1 expression cassette that is upstream of human Mapt and App genes did not result in any significant differences in expression of these two genes in GDX female AD animals compared to sham-treated AD females. Binding of macroH2A1 to the Thy1 promoter was greater in males than in females, however, which may account for the reduced expression of the hApp transgene in AD males compared to females. In contrast, GDX male AD mice did exhibit (1) reduced expression of the App transgene compared to sham-operated mice without altered macroH2A1 binding at the Thy1 gene and (2) a significant increase of macroH2A1 binding at the Psen1 gene without a change in Psen1 gene expression. The fact that reduced App transgene expression occurred in male AD GDX mice in the absence of a difference in macroH2A1 binding at the Thy1 expression cassette, and that the increased macroH2A1 binding to Psen1 did not alter Psen1 gene expression in male AD GDX mice compared to sham-operated animals, implies the existence of additional regulatory factors other than macroH2A1.

Taken together, our data further confirm the differential effects of male and female gonadal steroids on AD-related pathology: male hormones mainly target endogenous and transgenic App independent of macroH2A1 regulation, whereas female hormones target both endogenous macroH2A1-dependent mMapt regulation and macroH2A1-independent mApp regulation, but do not influence transgene expression. Supporting our observations, one study observed that after GDX of aged mice with elevated levels of App mRNA, App was downregulated in males, but upregulated in females [103]. In the same study, supplementation with estradiol decreased App levels in ovariectomized females, whereas testosterone supplementation increased App mRNA levels in castrated males. Thus, gonadal steroids influence sex-dependent differences in AD-like pathology in 3×Tg-AD mice.

GDX in mouse models of AD may approximate a critical aspect of human reproductive aging in males and females by depletion of gonadal hormones in the adult. In females, acute, surgical depletion of gonadal hormones approximates the sudden drop accompanying menopause. However, in males, acute GDX poorly mimics the timing of the slow and progressive reduction in male hormones with age. Whether the rate of hormone decline influences AD-related outcomes is an unknown factor. Compensatory neuroendocrine responses of increased luteinizing hormone (LH) and follicle stimulating hormone (FSH) following removal of gonads should be also considered [31]. These compensatory changes occur in both simulated and true reproductive aging, and therefore in this aspect, GDX in mouse models remains relevant to the human condition [31]. However, in GDX rodents, the levels of serum FSH and LH are rapidly and significantly elevated post surgery [95], mimicking the rapid menopausal changes in women better than the more gradual changes in aging men [51].

Here, depletion of hormones in young adults (at 3 months of age in this study) rather than in aged animals is an accurate model for oophorectomy for ovarian cancer or endometriosis and for prostatectomy for prostate cancer. A series of studies during the last 2 decades have shown unequivocally that bilateral oophorectomy at 45 years of age or earlier is associated with higher risk of cognitive impairment and dementia [39]. Unlike oophorectomy however, it remains unclear whether androgen deprivation therapy is associated with subsequent dementia risk in patients with prostate cancer [53]. Two studies reported no relationship of anti-androgens to Alzheimer’s disease or cognitive disorder [66], whereas another study showed that chemical castration doubles the risk of Alzheimer’s disease in patients with prostate cancer [80]. A recent study utilizing a large cohort of patients with prostate cancer proposed that antiandrogen monotherapy, but not orchiectomy or gonadotropin-releasing hormone agonists, was associated with increased dementia risk, implying different effects of surgical vs. chemical castration [53].

All animal-related protocols were performed in accordance with the rules and regulations of the Canadian Council of Animal Care and were approved by the McMaster University Animal Research Ethics Board.

Not applicable.

The authors declare no competing interests.

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.