Estradiol Replacement at the Critical Period Protects Hippocampal Neural Stem Cells to Improve Cognition in APP/PS1 Mice

It has been suggested that there is a critical window for estrogen replacement therapy (ERT) in postmenopausal women with Alzheimer’s disease (AD); however, supporting evidence is lacking. To address this issue, we investigated the effective period for estradiol (E2) treatment using a mouse model of AD. Four-month-old female APPswe/PSEN1dE9 (APP/PS1) mice were ovariectomized (OVX) and treated with E2 for 2 months starting at the age of 4 months (early period), 6 months (mid-period), or 8 months (late period). We then evaluated hippocampal neurogenesis, β-amyloid (Aβ) accumulation, telomerase activity, and hippocampal-dependent behavior. Compared to age-matched wild type mice, APP/PS1 mice with intact ovaries showed increased proliferation of hippocampal neural stem cells (NSCs) at 8 months of age and decreased proliferation of NSCs at 10 months of age; meanwhile, Aβ accumulation progressively increased with age, paralleling the reduced survival of immature neurons. OVX-induced depletion of E2 in APP/PS1 mice resulted in elevated Aβ levels accompanied by elevated p75 neurotrophin receptor (p75NTR) expression and increased NSC proliferation at 6 months of age, which subsequently declined; accelerated reduction of immature neurons starting from 6 months of age, and reduced telomerase activity and worsened memory performance at 10 months of age. Treatment with E2 in the early period post-OVX, rather than in the mid or late period, abrogated these effects, and p75NTR inhibition reduced the overproliferation of NSCs in 6-month-old OVX-APP/PS1 mice. Thus, E2 deficiency in young APP/PS1 mice exacerbates cognitive deficits and depletes the hippocampal NSC pool in later life; this can be alleviated by E2 treatment in the early period following OVX, which prevents Aβ/p75NTR-induced NSC overproliferation and preserves telomerase activity.


INTRODUCTION
The risk for Alzheimer's disease (AD) is associated with age-related loss of ovarian hormones in women (Marongiu, 2019). Prospective and case-control studies have demonstrated that estrogen replacement therapy (ERT) can effectively reduce cognitive deficits in some but not all postmenopausal women (Espeland et al., 2004;Vedder et al., 2014). Growing evidence suggests that the effectiveness of ERT depends on its application during a critical period and a specific treatment duration (Espeland et al., 2004;Zandi et al., 2002). Experiments in aged nonhuman primates and rodents have revealed that 17β-estradiol (E2) treatment following ovariectomy can improve cognitive function (Baxter et al., 2018;Pollard et al., 2018) and alleviate cognitive decline in aged animals with long-term E2 deficiency (Luine, 2014). This critical treatment window may be related to changes in brain responsiveness to E2.
The generation and progression of β-amyloid (Aβ) accumulation in the brain is a pathologic hallmark of AD (Benilova et al., 2012). Many experiments have demonstrated the ability of Aβ to cause synaptic dysfunction and mature neuron death (He et al., 2019;Xu et al., 2012). Neurogenesis in the adult hippocampus involves the proliferation of NSCs and their terminal differentiation into neurons, which is important for preserving cognitive function (Ager et al., 2015). Improving hippocampal neurogenesis can ameliorate cognition in mouse models of AD (Xu et al., 2012), which exhibit a pathologic Aβ plaque burden accompanied by decreased hippocampal neurogenesis and impaired hippocampal-dependent spatial memory (Moon et al., 2014). A previous study has reported that Aβ impairs the proliferation of NSCs in the adult subventricular zone via p75 NTR (Sotthibundhu et al., 2009). Also, both in vitro and in vivo experiments have proven that the aggregated form of Aβ can inhibit telomerase activity (Wang et al., 2015), which has been proposed to preserve stem cell function by the promotion of resting stem cell proliferation and maintaining in telomere length (Boccardi and Paolisso, 2014). Postmenopausal women who received long-term estrogen therapy had longer telomeres than those who did not receive this therapy (Lee et al., 2005). There is increasing evidence that E2 can alter hippocampal structure and function (Frick et al., 2015), with long-term administration of E2 at physiologic doses reducing Aβ accumulation in 3xTg-AD mice (Carroll et al., 2007).
To explore the effective time window of E2 treatment for alleviating AD-associated cognitive impairment and the hippocampal mechanisms, we used female transgenic mice carrying mutant human APPswe and PS1dE9 genes (APP/PS1 mice), in which the production of Aβ appears at approximately 4-5 months of age and dramatically accelerates during 6-8 months of age (Garcia-Alloza et al., 2006). In this study, 4-month-old female APP/PS1 mice were ovariectomized (OVX) to induce an acute decline in E2(surgical menopause), and then we evaluated the effects of E2 administered at different time points postsurgery on hippocampal neurogenesis, Aβ accumulation, and the causal link between the neuroprotective effect of E2 and its hippocampaldependent memory performance.

Animals
APP/PS1 transgenic mice (Stock No. 34829) with a C57BL/6J × C3H hybrid background (Stock No. 004462) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). APP/PS1 mice were maintained by crossing the hemizygotes with noncarrier siblings [later referred to as wild-type (WT) mice], and genotyped by polymerase chain reaction (PCR) analysis of genomic DNA from tail biopsies according to the instructions and our previous publication (Xu et al., 2012). Female hemizygous APP/PS1 mice aged 4-10 months and their WT littermates were used in this study. The present study was approved by the Institutional Animal Care and Ethical Committee of Nanjing Medical University (No. 2014-153), and conducted in accordance with the guidelines of the Institute for Laboratory Animal Research of the Nanjing Medical University. Mice were housed in plastic cages at 23 ± 2 • C and 55% relative humidity with a 12:12 h light/dark cycle. Water and food were given ad libitum. Body weight was measured weekly.

Ovariectomy and Drug Administration
Ovariectomy All surgeries were performed under ketamine (100 mg/kg)/xylazine (10 mg/kg; i.p.) anesthesia using aseptic conditions following institutional guidelines. Female mice in diestrus were bilaterally ovariectomized (OVX) at the age of 4 months (Figure 1). The dorsal surface was shaved and the vascular supply was ligated, then ovaries dissected out through bilateral incisions. Sham-operated (sham-op) control mice were subjected to surgery and ovaries were manipulated but left intact.

E2 Treatment
A pellet containing 0.18 mg of E2 (Innovative Research of America, Sarasota, FL, USA) was subcutaneously implanted FIGURE 1 | Time chart of the experimental procedure. Horizontal arrow indicates the mouse age (month). ↓, E2/TAPI-2 administration; +, time point of experiment examination; E-E2, early period; M-E2, mid period; L-E2, late period; HE, histological examination; WB, western blotting; BT, behavioral tests. at the following three time points: immediately after OVX, 2 months after OVX (6-month-old), or 4 months after OVX (8-month-old). The pellet continuously releases E2 for 60 days, resulting in serum levels of estrogen in the physiological range (Carroll et al., 2007). At the end of each period of E2 treatment, specimens were collected or behavioral tests were carried out (see time chart of the experimental procedure in Figure 1).

Histological Examination and Quantitative Evaluation
The mice were injected (i.p.) with BrdU (50 mg/kg; Sigma-Aldrich) twice daily at 6 h intervals for three consecutive days as previously described (Taniuchi et al., 2007). At 24 h after the last BrdU injection, the mice were anesthetized and perfused transcardially with cold phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were post-fixed overnight in 4% PFA at 4 • C and coronal section (40 µm) were cut using a vibrating microtome (Microslicer DTK 1500, Dousaka EM Company, Kyoto, Japan).

Quantitative Evaluation of Immuno-Positive Cells
Immuno-positive BrdU (BrdU + ) and DCX (DCX + ) cells were assessed using a conventional light microscope (Olympus DP70, Japan) with a 40× objective. Images of Ki67-immunolabeled sections were observed using a fluorescence microscope (Olympus DP70, ×100). The total number of BrdU + cells or Ki67 + cells was calculated from 12 sections and multiplied by 5 to obtain a total number for each mouse. The number of DCX + cells in the subgranular zone (SGZ) was counted and divided by the length of SGZ to obtain the density of DCX + cells (per micrometer; Sha et al., 2016).

Aβ Immunochemistry and Plaque Counting
The blocked sections were incubated in rabbit polyclonal anti-Aβ (1:300; Invitrogen, Carlsbad, CA, USA). Immuno-reactivities were visualized by the ABC method. Aβ-labeled plaques in the hippocampus with a diameter of more than twice the size of a neuronal cell body (the mean diameter is approximately 15 µm; Gilman et al., 2017;Weisenburger and Vaziri, 2018) were counted, and the number was multiplied by 5 to obtain the total number of plaques in the hippocampus (Carroll et al., 2007).

Enzyme-Linked Immunosorbent Assay (ELISA)
Sandwich Aβ ELISAs were performed as previously described (Heikkinen et al., 2004). Briefly, hippocampal samples were dissected from the mice under deep isoflurane anesthesia and homogenized in 8× cold guanidine buffer (50 mM Tris-HCl, pH 8.0, 5 M guanidine-HCl). The samples and Aβpeptides used as standards were diluted with reaction buffer containing protease inhibitors and AEBSF (Roche, Mannheim, Germany) to prevent degradation of Aβ and centrifuged at 16,000× g for 20 min at 4 • C. The quantity of total protein in the supernatants was quantified by bicinchoninic acid (BCA) protein assay kit (Pierce Biotechnology Inc., Rockford, IL, USA). The total levels of hippocampal Aβ were quantified using human Aβ 42 and Aβ 40 ELISA kits (Invitrogen #KHB3441 and #KHB3481, respectively) following the manufacturer's instructions. Absorbance at 450 nm was measured with an automated microplate reader (ELx800 BioTek Inc., Winooski, VT, USA). Aβ 42 and Aβ 40 concentrations in the samples were standardized to total protein levels and are expressed in nanograms per milligram of total protein.

Western Blotting Analysis
The brains were rapidly removed and the coronal brain slices (1 mm) were cut in ice-cold PBS. The dentate gyrus (DG) of the hippocampus was micro-dissected from the sections on dry ice and homogenized. The protein (50 µg) was separated by SDS-PAGE and transferred to membranes. The antibodies used were rabbit polyclonal anti-p75 NTR (1:200; Santa Cruz) and mouse monoclonal anti-β-actin (1:2,000; Millipore). Horseradish peroxidase-conjugated secondary antibodies were applied and the immunoreactive signals visualized with enhanced chemiluminescence detection system (Millipore). Protein band intensity was quantified using the image analysis software package (ImageJ; NIH Image, Bethesda, MD, USA) and is expressed as the ratio relative to β-actin level.

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)
Total RNA from DG regions was extracted using TRIzol reagent kit (Invitrogen). cDNA synthesis was performed using the Prime Script RT reagent kit (TaKaRa; Japan), according to manufacturer's instructions. Briefly, 500 ng of RNA was used as a template to prepare cDNA. A 1 µl 20-fold diluted cDNA (each sample in triplicate) was then used for each run. The primer sets used for p75 NTR , GAPDH, TERT, and β-actin were designed according to the previous publications (AlMatrouk et al., 2018;Zhang et al., 2019). PCR conditions for TERT and β-actin were 30 cycles of denaturation at 94 • C for 45 s, annealing at 65 • C for 45 s, and extension at 72 • C for 45 s. PCR product of TERT was resolved by 1.5% agarose gel electrophoresis and visualized by ethidium bromide staining. The band intensities were determined using the image analysis software package (ImageJ), and β-actin gene expression was detected as the internal control. The other PCR products were assessed using an ABI Prism 7300 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) in the presence of a fluorescent dye (SYBR Green I; TaKaRa). The relative expression of genes was determined using the 2 −∆∆Ct method, with normalization to GAPDH expression. The values were averaged from four sets of independent experiments and are expressed as a percentage relative to WT mice.

Telomeric Repeat Amplification Protocol (TRAP) Assay
The telomerase activity was measured from fresh hippocampal DG regions using the TRAP EZE XL Telomerase Detection Kit (#S7707; Millipore), following the manufacturer's instruction and previously described  with some modifications. All samples were analyzed in triplicate. Briefly, samples were resuspended in CHAPS lysis buffer and incubated on ice for 30 min. Each reaction contained 10 µl 5× TRAP EZE RT Reaction mix, 0.4 µl (2 U) Taq polymerase, 37.6 µl nuclease-free H 2 O, and 2 µl sample extract (500 ng/µl). The TSR8 control template was used to generate a standard curve. After a 30 min incubation at 30 • C, the samples were amplified over 38 cycles at 94 • C for 30 s 59 • C for 30 s and 72 • C for 1 min. The amplified telomerase products were quantified using a fluorescence plate reader (MQX200 BioTek Inc.). Telomerase activity was calculated from the ratio of telomerase products to an internal standard for each lysate (∆FL/∆R).

Y-Maze Test
Working memory performance was assessed by recording spontaneous alternation behavior (SAB) in a single-session Y-maze as described before . The custom-made maze had wooden arms separated by 120 • angle. Each arm was 30 cm long, 10 cm wide, and restricted by 20 cm high walls. The mice were placed at the center of the Y-maze and were allowed to freely explore for 8 min. The series of arm entries was recorded visually. An arm entry was considered to be complete when the mouse placed all four paws into a given arm. The percentage alternation was calculated as the ratio of actual alternations (the number of triads containing successive entries into the three arms) to the maximum possible alternations (the total number of arms entered minus 2).

Morris Water Maze (MWM) Test
Two days after Y-maze test, spatial learning and memory ability were evaluated with the MWM test . A circular water tank made of white polypropylene (diameter = 120 cm; height = 45 cm) was prepared. The pool was filled to a depth of 33 cm with water (24 ± 1 • C) made opaque by adding a small amount of non-toxic white paint. Swimming paths were analyzed using a computer system with a video camera (AXIS-90 Target/2; Neuroscience Inc., Tokyo, Japan). During the visible-platform training (2 days), a cylindrical dark-colored platform (7 cm in diameter) was placed 0.5 cm above the water surface with the position of the midpoint of one of the four quadrants. The mouse was randomly released from four different quadrants, respectively, and allowed to swim for 90 s. During the hidden-platform training (5 days), the platform was submerged 1 cm below the water surface opposite to the position in the visible-platform training. Four trials were conducted per day with a 30-min intertrial interval. Average swimming speed (m/s) and latency (s) to reach the platform were scored on all trials. On day 8 of training, a probe trial was carried out by removing the platform. The mouse was released from the opposite quadrant in which the platform was located in the hidden-platform training and allowed to swim for 90 s, and the percentage of time spent in the target quadrant was used to evaluate memory retention.

Data Analysis/Statistics
All of the outcome analyses were carried out by the independent investigators blinded to the treatment conditions and mouse types. Data were retrieved and processed with the software Micro cal Origin 9.1 (Origin Lab Corp., Northampton, MA, USA). Group data were expressed as mean ± standard error (SE), and statistical analyses were performed using SPSS software, version 18.0 (SPSS Inc., Chicago, IL, USA). Experimental results were compared among treatment groups by analysis of variance (ANOVA) or Student's t-test, followed by Bonferroni's post hoc analysis for significant effects. P < 0.05 was considered statistically significant.
The results indicate that E2 is beneficial for the hippocampi at the initiation stage of Aβ formation rather than at the more advanced stage of Aβ deposition.

Involvement of p75 NTR in the Effects of E2-Induced Protection of Neurogenesis in APP/PS1 Mice
Activated p75 NTR ligands were shown to be involved in the neurogenic effect of Aβ (Sotthibundhu et al., 2009). To test the role of p75 NTR in the preservation of neurogenesis by E2 in APP/PS1 mice, we examined p75 NTR expression in 6-monthold mice that were OVX or sham-op at 4 months old and treated with E2 or vehicle for 2 months (Figure 1). Compared to 6m-WT mice, levels of p75 NTR protein (P = 0.082, n = 8; Figure 5A) and mRNA (P = 0.056, n = 8; Figure 5B) in 6m-APP/PS1 mice had a slight but insignificant increase. OVX-6m-APP/PS1 mice showed markedly higher levels of p75 NTR protein (P < 0.001, n = 8) and mRNA (P < 0.001, n = 8) than sham-op 6m-APP/PS1 mice. The application of E2 abolished the increases in p75 NTR expression in OVX-6m-APP/PS1 mice (P < 0.001, n = 8 for both protein and mRNA). We further used TAPI-2 to prevent the activation of p75 NTR and examined Aβ-impaired neurogenesis. Compared to vehicle-treated OVX- 6m-APP/PS1 mice, the number of BrdU + cells was significantly reduced in TAPI-2-treated OVX-6m-APP/PS1 mice (P = 0.014, n = 8; Figure 5C); however, the density of DCX + cells was unchanged (P = 0.636, n = 8; Figure 5D). Thus, E2 deficiency results in elevated expression of p75 NTR and consequently, increased proliferation of hippocampal NSCs in APP/PS1 mice, but has no effect on the reduction of immature neurons.

Effects of E2 on Telomerase Regulation in Aged APP/PS1 Mice
Telomerase, a reverse transcriptase, is required for embryonic and adult neurogenesis (Lobanova et al., 2017). To investigate whether telomerase is implicated in the E2 mediated protection of NSCs in APP/PS1 mice, we analyzed the levels of TERT and telomerase activity in 10-month-old APP/PS1 mice. Compared to 10m-WT mice, 10m-APP/PS1 mice displayed significantly lower TERT mRNA level (P < 0.001, n = 8; Figure 6A) and telomerase activity (P < 0.001, n = 8; Figure 6B). However, TERT level (P < 0.001, n = 8) and telomerase activity (P < 0.001, n = 8) were markedly lower in OVX-10m-APP/PS1 mice than in sham-op 10m-APP/PS1 mice, but were increased by the administration of E-E2 (mRNA: P = 0.011, n = 8; activity: P = 0.037, n = 8; the administration of E2 is shown in Figure 1). However, compared to sham-op 10m-APP/PS1 mice, the TERT level (P = 0.016, n = 8) and telomerase activity (P = 0.032, n = 8) in OVX-10m-APP/PS1 mice that received E-E2 treatment were not FIGURE 5 | p75 NTR participates in E2-induced protection of NSC proliferation in APP/PS1 mice. (A,B) Influence of E2 on the level of p75 NTR in 6-month-old APP/PS1 mice. The upper panels represent representative blots of p75 NTR . Bar graphs show the level of p75 NTR protein or p75 NTR mRNA in APP/PS1 mice or OVX-APP/PS1 mice treated with vehicle or E2 for 2 months. **P < 0.01 vs. sham-op APP/PS1 mice; ## P < 0.01 vs. OVX-APP/PS1 mice (two-way ANOVA). (C,D) Influence of the p75 NTR inhibitor TAPI-2 on the excessive proliferation of NSCs and survival of immature neurons in the DG of 6-month-old APP/PS1 mice. The bar graphs show the number of BrdU + cells and the density of DCX + cells in APP/PS1 mice or OVX-APP/PS1 mice treated with vehicle or TAPI-2 for 7 days. *P < 0.05 vs. sham-op APP/PS1 mice; # P < 0.05 vs. OVX-APP/PS1 mice (two-way ANOVA). completely restored. The administration of M-E2 or L-E2 had no effect on TERT expression (all P > 0.05, n = 8) or telomerase activity (all P > 0.05, n = 8) in OVX-10m-APP/PS1 mice. These results indicate that telomerase activity declines in the Aβimpaired hippocampus, which is exacerbated by E2 deficiency; however, this is reversed by early period E2 treatment.

Effects of E2-Induced Protection of Neurogenesis on Cognitive Deficits in Aged APP/PS1 Mice
Adult hippocampal neurogenesis has been implicated in the cognitive deficits that are a signature abnormality of AD (Gonçalves et al., 2016). Thus, it is critically important to assess whether E2-mediated protection of neurogenesis in APP/PS1 mice is linked to improvements in cognitive function. Hippocampal-dependent learning and memories were evaluated by the Y-maze and MWM at the age of 10 months (Figure 1). The SAB rate in the Y-maze, which is regarded as a measure of short-term working memory, was remarkably reduced in 10m-APP/PS1 mice compared with that in 10m-WT mice (P < 0.001, n = 10; Figure 7A). Moreover, OVX-10m-APP/PS1 mice showed a distinctly decreased rate compared to sham-op 10m-APP/PS1 mice (P < 0.001, n = 10). Central to our analysis was the observation that E-E2 administration partly restored the alternation rate (P = 0.003, n = 10) but did not completely reverse the impairment in working memory in OVX-10m-APP/PS1 mice relative to sham-op 10m-APP/PS1 mice (P = 0.026, n = 10). Additionally, the administration of M-E2 and L-E2 had no effect on the SAB rate in OVX-10m-APP/PS1 mice (all P > 0.05, n = 10). The rate in 10m-WT mice was unaffected by E2 deprivation and E2 treatment (data not shown). To exclude the possibility that the changes in SAB performance were caused by differences in motility, we quantified the number of arms entered by mice in the Y-maze. As shown in Figure 7B, there was no significant effect of genotype (F (1,54) = 0.582, P = 0.449), ovariectomy (F (1,54) = 0.005, P = 0.944) or E2 treatment (F (3,54) = 1.191, P = 0.322) on the number of arm entries.
A probe trial was carried out 24 h after the last day of training, and the percentage of time spent in the target quadrant was measured to evaluate the strength of the memory trace. As shown in Figure 7D, 10m-APP/PS1 mice spent less time in the target quadrant than 10m-WT mice (P < 0.001). OVX-10m-APP/PS1 mice showed much shorter swimming time in the target quadrant than sham-op 10m-APP/PS1 mice (P < 0.001). In addition, the swimming time in the target quadrant was longer for E-E2-treated OVX-10m-APP/PS1 mice than for vehicletreated OVX-10m-APP/PS1 mice (P = 0.021), but shorter than for sham-op 10m-APP/PS1 mice (P = 0.023). The performance in the target quadrant by OVX-10m-APP/PS1 mice was unaltered FIGURE 7 | The administration of E2 in the early period post-OVX alleviates the cognitive deficits in aged APP/PS1 mice. (A,B) Influence of E2 treatment in the early period (E-E2), mid period (M-E2), or late period (L-E2) on working memory as assessed by the Y-maze test in 10-month-old APP/PS1 mice. The bar graphs show the spontaneous alternation ratio (%) and the total number of arm entries in the Y-maze test of APP/PS1 mice. **P < 0.01 vs. WT mice; # P < 0.05 and ## P < 0.01 vs. sham-op APP/PS1 mice; ++ P < 0.01 vs. OVX-APP/PS1 mice (two-way ANOVA). (C) Spatial learning and memory were assessed by the Morris water maze (MWM). Each point represents the mean latency (sec) to reach the platforms (upper panel) and the swimming speeds (m/s; bottom panel). **P < 0.01 vs. WT mice; # P < 0.05 and ## P < 0.01 vs. sham-op APP/PS1 mice; + P < 0.05 vs. OVX-APP/PS1 mice (repeated measure ANOVA). (D) Probe trial test. The bars represent the percentages of swimming time spent in the platform quadrant. **P < 0.01 vs. WT mice; # P < 0.05 and ## P < 0.01 vs. sham-op APP/PS1 mice; + P < 0.05 vs. OVX-APP/PS1 mice (two-way ANOVA).
by the administration of M-E2 or L-E2 (P > 0.05 in each group). Taken together, these results imply that E2 alleviates cognitive deficits in aged APP/PS1 mice, but only when administered during the early period post-OVX.

DISCUSSION
Consistent with the evidence supporting the existence of a critical period for the benefit of ERT on cognitive function in postmenopausal women with AD (Vedder et al., 2014), here, we showed in AD model mice that E2 treatment in the early, but not in the late period post-OVX, prevented cognitive decline. This involved inhibition the excessive proliferation of hippocampal NSCs by suppression of Aβ-induced overexpression of p75 NTR in young APP/PS1 mice, and preservation of telomerase activity and proliferation capability of NSCs in aged APP/PS1 mice. The positive effects on cognition were evidenced by better scores in the Y-maze and MWM tests. APP/PS1 mouse models develop AD pathology at young age, enabling the study of exogenous E2 efficiency on cognitive performance under varying conditions of ovarian function. Whilst OVX can be used to study the consequences of menopause in women, the procedure induces a rapid gonadal hormone decline in contrast to the progressive decline associated with natural menopause in women. Nevertheless, a recent review has suggested that OVX results in a ''blank ovarian hormone state,'' which is an ideal model to evaluate the effects of ovarian hormone deprivation and the outcomes of subsequent exogenous hormone treatments on the brain (Koebele and Bimonte-Nelson, 2016). Indeed, epidemiological data regarding women who underwent OVX before the onset of natural menopause indicate a long-term increased risk of AD (Rocca et al., 2010) and furthermore ERT efficacy is similar in both natural menopause and oophorectomy in women, depending on the timing of treatment initiation (Mosconi et al., 2004;Maki et al., 2019). Given that female mice continue to experience hormonal cycles with age (Hung et al., 2003), the use of OVX can be considered an effective model to mimic human/primate menopause in mice.
The results presented here are in line with previous studies demonstrating that the proliferation of hippocampal NSCs is increased at the Aβ plaque initiation stage while subsequently decreased at the Aβ plaque progressive stage in the same mouse model as ours (Taniuchi et al., 2007;Xu et al., 2012), indicating a similar ''bell-shaped'' proliferation curve. However, the survival of immature neurons declined faster with age in APP/PS1 mice, showing a steeper curve than that in WT mice. The turning point seems to be at the age of occurrence of massive Aβ deposits. Meanwhile, in agreement with earlier findings, our results showed progressive exacerbation of Aβ pathology in APP/PS1 mice from the age of 6-10 months (Garcia-Alloza et al., 2006). Our previous report demonstrated that Aβ impairs the survival of immature neurons by downregulating the PI3K-Akt-mTOR signaling pathway (Li et al., 2010). The sensitivity of the adult brain to Aβ was found to be reduced after 10 months of age in mice, indicating an age-dependent decrease in the neurogenic effect of Aβ (Sotthibundhu et al., 2009). Therefore, the bell-shaped curve may be attributed to the gradual desensitization of the nervous system to Aβ and depletion of the NSC pool; the decline in NSC proliferation could contribute to the progressive reduction in new neuron production in aged APP/PS1 mice. However, another AD mouse model, APP/PS1 mice on a C57BL/6J background exhibit seizure activity at the age of 3-4 months, and show different hippocampal neurogenesis from our model (Unger et al., 2016). These heterogeneities of phenotypes may be attributable to genetic background and interanimal variability. Additionally, although APP/PS1 transgenic mouse models have been commonly used in AD research, the pathogenesis in these models has distinctions from sporadic AD patients such as lacking of widespread neurodegeneration and none developed neurofibrillary tangles (NFTs; Drummond and Wisniewski, 2017). Further study is needed to determine the critical period of E2 on an ideal model contains both plaques and NFTs.
Our results showed that estrogen deficiency caused an increase in NSC proliferation at the Aβ initiation stage and a decrease at the Aβ progressive stage, showing a leftward shift of the proliferation curve. The survival of immature neurons showed a sharper decline in APP/PS1 mice than in WT mice. E2 treatment in the early period post-OVX prevented the over-proliferation of NSCs and increased the immature neuron. The relationship between E2 and hippocampal neurogenesis has been controversial. Exogenous E2 administration enhances hippocampal cell proliferation by activating estrogen receptor, whereas long-term E2 deficiency has no effect on cell proliferation in the DG in female rats or mice (Tanapat et al., 2005;Galea et al., 2006;Lagace et al., 2007). Importantly, E2 treatment at the early period post-OVX decreased the levels of Aβ. According to previous studies, E2 treatment for 3 months reduces the enhanced Aβ accumulation in OVX-3xTg-AD mice (Carroll et al., 2007). E2 treatment can increase Aβ clearance via regulation of β-secretase (BACE1) activity and expression in APP23 mice (Li et al., 2013). A 2-month administration of docosahexaenoic acid-enriched phospholipids inhibits the generation and accumulation of Aβ by suppressing BACE1 expression in senescence-accelerated prone eight mice . These evidences may explain the effect of E2 on Aβ accumulation. Investigations into the relationship between Aβ and hippocampal neurogenesis have proven that exogenous and endogenous Aβ promote the proliferation of NSCs in the subventricular zone via p75 NTR (Sotthibundhu et al., 2009;Zheng J. Y. et al., 2017), and Aβ inhibits the endocytosis of p75 NTR in cholinergic basal forebrain neurons (Ovsepian et al., 2014). p75 NTR is shown to stimulate the proliferation of cultured neurospheres by interacting with several cell-cycle regulatory proteins (Provenzano et al., 2011). Our findings showed that the levels of Aβ and p75 NTR were elevated by OVX, both prevented by E2 treatment in the early period post-OVX; blocking p75 NTR activation reduced NSC proliferation without affecting the survival of immature neurons. Thus, the increased survival of immature neurons by early-period E2 treatment might be related to the reduction in Aβ levels. One study has demonstrated that the deletion of the p75 NTR gene suppresses Aβ production in APP/PS1 mice (Wang et al., 2011). Lower p75 NTR mRNA and protein levels have been observed following chronic E2 treatment (Hasan et al., 2005). Thus, there are multiple levels of interaction between E2, Aβ, and p75 NTR . Combined with these observations, our results indicate that early period E2 treatment blocked the hyperproliferation of NSCs by inhibiting Aβ-induced p75 NTR expression, and alleviated the depletion of the NSC pool in aged APP/PS1 mice, thereby abolishing the leftward overproliferation curve. In accordance with the E2-induced preservation of NSC proliferation in aged OVX-APP/PS1 mice, we found that telomerase activity was reduced in aged APP/PS1 mice, which was exacerbated by OVX and rescued by early-period E2 treatment post-OVX. These observations imply that changes in telomerase activity are accompanied by altered NSC number in the DG. Telomerase is expressed at high levels in NSCs and neural progenitor cells in the adult brain (Liu et al., 2018), and Aβ aggregates are shown to inhibit telomerase activity in senescent human cells (Wang et al., 2015). This implies that the decrease in telomerase activity in aged APP/PS1 mice may be related to the Aβ accumulation. However, we observed that there were no significant differences in Aβ levels between aged APP/PS1 mice, OVX-APP/PS1 mice and E2 treated OVX-APP/PS1 mice, raising the possibility that the NSC pool provides a supply of telomerase. Additionally, pharmacologic induction of TERT is found to increase the expression of nerve growth factor and brain-derived neurotrophic factor (BDNF) while protecting neurons from oxidative stress and the cytotoxic effects of Aβ (Baruch-Eliyahu et al., 2019), implying an interaction between the NSC pool and telomerase.
Interestingly, E2 had beneficial effects on neurogenesis and Aβ levels in APP/PS1 mice only when administered during the early period post-OVX. Our findings in this regard can be explained by previous studies showing that the neuroprotective actions of E2 are dependent on time since OVX. The administration of E2 within 3 months, but not delayed 10 months post-OVX, has efficacy in rats (Walf et al., 2009). It has been proposed that the responsiveness of the brain to hormones declines after an extended period of hormone depletion (Hamilton et al., 2011). E2 deficiency or administration after the progressive stage did not affect Aβ levels or the plaque numbers in APP/PS1 mice. Although some studies suggest that estrogen deprivation for 3 months could enhance the deposition of Aβ in APPswe transgenic mice that underwent OVX at 4 weeks old , estrogen treatment after a similar interval following OVX has no effect on Aβ levels in mice that underwent OVX at an older age (Heikkinen et al., 2004). This is in general agreement with earlier work demonstrating that long-term E2 deficiency increased Aβ production at the initiation stage of amyloid pathology but has no effect on Aβ aggregation and plaque formation (Heikkinen et al., 2004;. E2 had no effect on hippocampal neurogenesis in WT mice, although 10-monthold OVX-WT mice showed fewer immature neurons, which is supported by reports that E2 does not affect neurogenesis in normal mice (Lagace et al., 2007).
In addition to E2-mediated effects on hippocampal NSC proliferation, we also observed that E2 treatment in the early period post-OVX partially alleviated cognitive deficits in aged APP/PS1 mice. Adult neurogenesis depends on the function and stable numbers of NSCs. Continuous neurogenesis in the DG is essential for maintaining normal hippocampaldependent cognitive behavior. The hippocampal mechanism of context encoding is considered to depend on a subset of immature adult-born neurons in the DG (Adlaf et al., 2017). NSC transplantation restores spatial learning ability in 18month-old 3xTg-AD mice by elevating hippocampal BDNF level (Blurton-Jones et al., 2009), and stimulating NSC proliferation in 7-month-old 3xTg-AD mice by activating the PI3K-AKT-GSK3β-Wnt pathway results in increased cognitive function (Zheng R. et al., 2017). The relationship between NSC and synaptic resistance to Aβ toxicity has been confirmed by in vivo and in vitro experiments: a larger pool of NSC is associated with reduced Aβ binding to the synapse, while transgenic ablation of endogenous NSC increases synaptic Aβ binding (Micci et al., 2019). Factors secreted during the proliferation of hippocampal NSCs create a favorable microenvironment for the proper functioning of the nervous system (Lee et al., 2007). Moreover, E2 treatment in young OVX mice can lead to a long-lasting enhancement of classical estrogen response element-dependent gene activation in the hippocampus, the effects of which can persist even after cessation of E2 treatment (Pollard et al., 2018).
A recent multimodality brain imaging study has shown that declines in circulating estrogens are a major risk factor for female-specific brain AD abnormalities, and suggests that the therapeutic window of ERT for AD preventive interventions in women might be early in menopause (Rahman et al., 2020). The present study provides evidence that the early period following OVX is the critical window for E2 treatment to alleviate cognitive deficits in aged APP/PS1 mice. E2 likely acts by preventing depletion of the hippocampal NSC pool and preserving basal neurogenesis. Our findings reveal the mechanisms underlying the cognitive efficacy of E2 treatment, especially as related to a critical window after hormone declines, and provide experimental evidence that can guide clinical decision-making for the initiation of ERT in postmenopausal women, particularly those with or at risk of developing AD.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Ethical Committee of Nanjing Medical University.