Impact of estrogen receptor agonists and model of menopause on enzymes involved in brain metabolism, acetyl-CoA production and cholinergic function

Abstract

Our goal is to understand how loss of circulating estrogens and estrogen replacement affect brain physiology and function, particularly in brain regions involved in cognitive processes. We recently conducted a large metabolomics study characterizing the effects of rodent models of menopause and treatment with estrogen receptor (ER) agonists on neurochemical targets in hippocampus, frontal cortex, and striatum. Here we characterize effects on levels of several key enzymes involved in glucose utilization and energy production, specifically phosphofructokinase, glyceraldehyde 3-phosphate dehydrogenase, and pyruvate dehydrogenase. We also evaluated effects on levels of β-actin and α-tubulin, choline acetyltransferase (ChAT) activity, and levels of ATP citrate lyase. All experiments were conducted in young adult rats. Experiment 1 compared the effects of ovariectomy (OVX), a model of surgical menopause, and 4-vinylcyclohexene diepoxide (VCD)-treatments, a model of transitional menopause, with tissues collected at proestrus and at diestrus. Experiment 2 used a separate cohort of rats to evaluate the same targets in OVX and VCD-treated rats treated with estradiol or with selective ER agonists. Differences in the expression of metabolic enzymes between cycling animals and models of surgical and transitional menopause were detected. These differences were model-, region- and time- dependent, and were modulated by selective ER agonists. Collectively, the findings demonstrate that loss of ovarian function and ER agonist treatments have differing effects in OVX vs. VCD-treated rats. Differences may help to explain differences in the effects of estrogen treatments on brain function and cognition in women who have experienced surgical vs. transitional menopause.

1. Introduction

Estrogens exert a profound influence on the brain including effects on brain architecture, synaptic function and plasticity [2, 3], neurotransmitter production and release [4, 5], neuroimmunomodulation [6, 7] and neuronal survival [8]. In many cases these effects have been associated with significant enhancements of learning, memory, and attention [9–11]. This is true both for animals and for humans [12–15]. Loss of estrogens (e.g., following menopause) likewise is thought to contribute to cognitive decline both in normal aging and in association with Alzheimer’s disease and Parkinson’s disease [16, 17]. Estrogen treatments have been shown to reverse some of these effects; however, not all studies agree, and mechanisms are still poorly understood.

Menopause refers to a loss of ovarian function with a corresponding reduction in ovarian estrogens. Ovariectomy or surgical menopause consists of the rapid and complete removal of the ovaries resulting in a complete loss of all ovarian hormones. In contrast, natural or transitional menopause is caused by the selective loss of ovarian follicles. This results in irregular hormonal fluctuations during the transition period, and a substantial decline in ovarian production of estrogens and progesterone [18]. In contrast with surgical menopause, ovarian stromal cells remain and continue to produce androgens, producing an increase in the ratio of circulating androgens:estrogens. A majority of women experience a transitional menopause beginning at approximately 50 years of age. Fewer women (~13%) experience oophorectomy or surgical menopause, many in their 20s and 30s, often to prevent cancers associated with BRCA1 and BRCA2 mutations [19].

Rodents do not undergo a natural menopause. Consequently, most preclinical studies have used ovariectomy to study how loss of ovarian function affects the brain. Recent studies showed that daily injections of the chemical 4-vinylcyclohexene diepoxide (VCD) can both gradually and selectively destroy primordial and primary ovarian follicles in rodents, thus producing a rodent model of natural menopause [20–24]. As in natural menopause, stromal cells remain intact and continue to produce androgens. Using this model, studies are beginning to show that effects of menopause as well as estrogen treatment on brain physiology and function can differ depending on when and how loss of ovarian function occurs (see [25] for recent review).

Another important variable are the specific estrogen receptors that are responsible for the effects of estrogens in different regions of the brain. Three estrogen receptors (ERs) have been cloned and characterized. ERα and ERβ are nuclear receptors that function to activate or suppress the expression of specific genes [26, 27]. These receptors also can be localized to membrane compartments where they participate in the rapid activation of signal transduction pathways [28–30]. A membrane-associated G-protein coupled estrogen receptor 1 (GPER1) also has been identified and participates in the rapid activation of signal transduction pathways [3] as well as in cross-talk with ERα- and ERβ- mediated pathways [31]. There also is evidence for the existence of a variety of estrogen-related receptors which are structurally related to the identified ERs and can participate in mediating estrogen effects [32–34]. These receptors are expressed by neurons and glial cells in different regions of the brain and determine the effects of estrogens on brain physiology and function.

The goal here was to characterize the effects of transitional and surgical menopause, as well as treatment with different estrogen receptor agonists, on metabolic pathways related to glucose utilization and energy production, acetyl-CoA production and cholinergic function. Note that these analyses are part of a larger metabolomics study designed to characterize the effects of surgical and transitional models of menopause, as well as treatment with different estrogen receptor agonists, on neurochemical endpoints in the brain. Here we used modified Western blot methods to measure effects on the levels of three enzymes that play critical roles in glycolysis, i.e., phosphofructokinase (PFK), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and pyruvate dehydrogenase (PDH), in addition to levels of β-actin and α-tubulin, and ATP citrate lyase (ATP-CL), an enzyme involved in the cytosolic production of Acetyl-CoA. We also measured effects on choline acetyltransferase (ChAT) activity, the enzyme responsible for the production of acetylcholine from Acetyl-CoA and choline. Effects were measured in brain regions that are known to contribute to learning and memory processes in task-specific ways.

2. Methods

2.1. Ethics statement

All procedures were carried out in accordance with PHHS policies on the use of animals in research, and with the approval of the University of Pittsburgh’s Institutional Animal Care and Use Committee.

2.2. Animals

Rats used in this study are the same rats that were used for a metabolomics study recently reported [35, 36]. One hemisphere of each brain was used for the prior studies while the other hemisphere was used for the current studies. Young adult (3mo old) female Sprague-Dawley rats were purchased from Harlan Sprague-Dawley Laboratories, Inc. and used for all experiments. Unlike humans, rodents do not experience follicular atresia in mid-life. Rodents experience reproductive senescence, which involves a hypothalamic dysregulation of the HPG axis leading to loss of estrous cycles and infertility. Consequently, young adults rats were used for all experiments to avoid the confounds associated with aging and reproductive senescence and to enable direct comparison of the ovariectomy (OVX)- and 4-vinylcyclohexene diepoxide (VCD)-treated rats. Rats were individually housed on a 12-hr light/dark cycle and had free access to food and water. In addition, all rats were acclimated to the housing conditions for two weeks prior to use.

2.3. Experimental design and treatments

Two experiments were performed. Experiment 1 evaluated the effects of surgical (OVX) vs. transitional (VCD-treated) menopause on protein expression and ChAT activity at 2 time points (1 week and 6 weeks), compared to gonadally intact controls sacrificed at proestrus and at diestrus. Experiment 2 used different rats to characterize the effects of estradiol and of selective ER agonists on these same endpoints in the two model of menopause, and after 1 or 6 weeks of treatment.

2.3.1. Selection of Targets

Targets were selected using several criteria: (1) an identified pivotal role in at least one major metabolic pathway, (2) evidence that gonadal hormones have direct and/or indirect effects on the expression and function of the target, and (3) changes in the expression or activity of each target must be associated with functional effects on glucose metabolism or on a particular neuronal function related to acetylcholine production. More detailed information about the selected targets is provided in Supplemental Table 1.

2.3.2. Experiment 1 – Comparison of menopausal models

Details of the experimental design have recently been reported [36] and are illustrated in Figure 1. Briefly, rats were assigned to one of four groups, proestrus (P), diestrus (D), VCD treatments and OVX. For P and D rats, daily vaginal smears were conducted to determine estrous cycle stage [37]. Additional rats received daily injections of VCD at 80 mg/Kg i.p. in sesame oil for 30 days (n=16). This dose was selected based on several studies showing that this regimen destroys the majority of primary ovarian follicles with little damage to stromal cells or to other organs [38–40]. One study [41] reported that adult rats (~9 months of age) treated with this dose also showed evidence of neutrophilia, increased blood urea nitrogen (BUN) and creatinine, increased liver weight, and some mortality due to peritonitis, as well as evidence that these effects are age-dependent. Rats in our study were much younger, being less than 4 months of age at the start of treatment. Note that we did not experience any mortality or overt signs of distress (e.g., weight loss, lethargy, hypersensitivity, porphyrin staining) associated with VCD treatment.

Figure 1:

Illustration of experimental designs involving OVX, VCD and cycling rat model. Two separate experiments were conducted. (A) Transitional menopause modeled in young SD rats by daily injections of 4-vinylcyclohexene diepoxide (VCD) at 80 mg/kg for 30-days. This regimen destroys >95% of primary ovarian follicles with little toxicity to stromal cells or to other organs, and no other lasting toxic effects. For surgical menopause model, cycling animals were given daily vehicle injections of sesame oil and after 30-days of treatment underwent bilateral ovariectomy. Rats that completed 3-day VCD treatment underwent sham surgery. Cycle stage of regularly cycling rats were determined by daily vaginal smears. Rats were anesthetized at 1-week and 6-weeks from model establishment and tissues from the hippocampus (HPC), striatum (STR), and frontal cortex (FCX) were collected. This resulted in a total of 6 treatment groups: P, D, OVX-1W, OVX-6W, VCD-1W, and VCD-6W with at least n= 4 or higher per group. (B) Following surgery, VCD and OVX rats were given 1-week recovery before beginning of treatments with estrogen receptor agonists. A mini osmotic pump was planted subcutaneously in the dorsal neck region for continuous administration of 5 μg/day of: 17β-estradiol (E2), PPT (a selective ERα agonist), DPN (a selective ERβ agonist), G-1 (a selective GPER1 agonist), or vehicle. Tissue from the HPC, STR and FCX were collected at 1-week and 6-weeks after beginning of agonist treatments.

Remaining rats received injections of sesame oil. Rats that received vehicle injections for 30 days then received bilateral OVX as described [1]. Likewise, rats that received VCD underwent sham surgery to control for OVX. At two time points following surgery (1 week and 6 weeks), rats were killed with an overdose of ketamine (280 mg/kg) and xylazine (56 mg/kg) and tissues were collected. This resulted in a total of 6 treatment groups: P, D, OVX-1W, OVX-6W, VCD-1W, and VCD-6W with at least n= 4 or higher per group.

2.3.3. Experiment 2 – Effects of estrogen receptor agonists in OVX and VCD-treated rats

Again, details of the experimental design have recently been reported [35]. Rats were assigned to either the OVX or VCD treatments as described above. After 1-week recovery, rats began receiving treatment with different ER agonists. Agonist treatments consisted of continuous subcutaneous (s.c.) administration of 17β-estradiol (E2), 4,4’,4”-(4-Propyl-[1H]-pyrazole-1,3,5-triyl)trisphenol (PPT; a selective ERα agonist), diarylpropionitrile (DPN; a selective ERβ agonist), G-1 (a selective GPER1 agonist), or vehicle. Agonists were dissolved in a vehicle containing 10% DMSO and 20% hydroxypropyl-β-cyclodextran (HPCD) [42]. E2 was administered at a dose of 3 μg/day. All other agonists were administered at a dose of 5 μg/day. Agonists were administered by miniosmotic pump (Alzet model 2002; Durect, Inc.) implanted s.c. in the dorsal neck region. Treatment groups and group sizes in experiment 2 are illustrated in Figure 1 and summarized in Supplemental Table 3.

2.4. Collection of Tissues

Rats were anesthetized with ketamine and xylazine and then decapitated. Brains were rapidly removed and dissected. Tissues from the hippocampus (HPC), striatum (STR), and frontal cortex (FCX) were collected, frozen on dry ice and kept at −80°C. Rats experiencing regular estrous cycles were killed on proestrus or diestrus and brain tissues collected. Trunk blood was collected from all rats and serum was processed and stored at –20°C for quantification of hormone levels.

2.5. Analysis of Hormone Levels by UPLC-MS-MS

Serum levels of E2, testosterone (T) and androstenedione (AD) were quantified by UPLC-MS/MS as recently described [35, 43, 44]. The lower limits of detection were 2.5 pg/mL for E2 and 10.0 pg/mL for T and AD.

2.6. Processing and Analysis of Brain Tissue Samples

Tissues were thawed in cold sonication medium as previously described [35]. Following sonication, half of each sample was immediately aliquoted into a freshly-prepared cold Western-blot lysis buffer [1]. All reagents were purchased from Sigma-Aldrich, Inc. Remaining sample was used for measuring ChAT activity. Bradford assay was used to determine protein concentration in each sample [45]. Samples were stored in −80 °C prior to processing.

2.7. SDS-PAGEs

Brain homogenates were processed for sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot (WB) using procedures recently developed and validated in our laboratory [1]. Briefly, samples were diluted to 2.4 μg/μl in sonication buffer. Samples were then mixed with an equal volume of Laemmli buffer (BIO-RAD, Cat# 1610737)), heated to 98°C for 7-minutes and loaded onto commercially prepared 12% Mini-PROTEAN® TGX™ gels. 12 μg in 10 μl were loaded per lane. A pre-stained protein ladder (Chameleon 700; LI-COR, Inc., Lincoln, NE, USA) was included on all gels. Gels were run for 2-hours at 100 V and were transferred for 2 hr, 100 V, 4°C to Immobilon FL PVDF membranes (Millipore, Inc., Burlington, MA. USA).

2.8. Total Protein Membrane Stain for Normalization

Total protein was detected on PVDF membranes using the LI-COR REVERT™ Kit (LI-COR, Inc.) as recently described [1]. Briefly, membranes were submerged in REVERT for 5 minutes, washed two times for 30 seconds each with wash solution [1] and imaged at 700 nm using the Odyssey® imaging system (LI-COR, Inc.) at a scanning resolution of 169 μm and a setting of ‘normal’ image quality. Membranes were then incubated in reversal solution [1] for 5 min and imaged again to verify removal of the total protein signal.

Membranes were then incubated Odyssey® TBS blocking buffer (LI-COR, Inc, cat#927–50000; 1 hour at room temperature) and incubated overnight at 4°C with a mixture of primary antibodies (Abcam, Inc., Cambridge, UK): ATP-CL (rabbit polyclonal anti-ATP-CL, 1:1500, ab40793), PFK (rabbit polyclonal anti-PFK, 1:1500, ab154804), α-tubulin (rabbit polyclonal anti-α-tubulin, 1:5000, ab52866), PDH (mouse monoclonal anti-PDH-E1α subunit, 1:2000, ab110330), β-Actin (mouse polyclonal anti-β-actin, 1:5000, ab6276; Experiment 1 only), and GAPDH (rat polyclonal anti-GAPDH, 1:5000, ab181602).

After incubating for 1 overnight with antibody, membranes were washed with TBS-T and then placed in TBS-T containing secondary antibodies (goat anti-mouse IgG, LI-COR IR Dye 800CW, cat#32210; or goat anti-rabbit IgG, LI-COR IRDye800CW, cat#32211) diluted 1:5,000 for 1 hr at room temperature as recently described [1].

2.9. Image Acquisition and Data Analysis

The Odyssey application software (Ver 3.0.30) was used to analyze all PVDF-membranes. Integrated intensities of signal bands were obtained as recently described [1]. Total protein was measured by manually by measuring total protein signal in the molecular weight range 50–90KDa. Previous studies determined that this molecular weight range has a wide linear range of detection as a function of protein loading and low signal variability across samples [1]. Target proteins were then normalized to total protein detected on the same lane. These ratios were then normalized to an average of the same target calculated from a set of samples from the control group (run on the same gel). This was done to facilitate between-gel comparisons. At least 3 samples from either the proestrus (experiment 1) or a vehicle-treated (experiment 2) control group were present on each gel and averaged within group to represent an intra-gel normalization factor. Prior studies from our laboratory revealed that this type of normalization enables us to perform between-gel comparisons with excellent between-gel consistency and an intraclass correlation coefficient of 0.89 (i.e., excellent reliability) [1].

2.10. Choline acetyltransferase activity assay

To measure ChAT enzyme activity, a modification of the radioenzymatic method of Fonnum [46] was applied as described by Pongrac and Rylett [47]. Sonicated samples were thawed and diluted to a concentration of 0.8 mg/mL using cold sonication buffer. Accuracy of the working dilution was verified by a second, post-dilution protein assay. Reaction triplicates (5ul) from each sample dilution were then incubated in 37 °C water bath for 30min in the presence of 10X acetyl-CoA (Sigma, cat#A2056) and [³H]-acetyl-CoA (PerkinElmer, Waltham, MA, USA, Cat# NET290050UC) for 30,000 d.p.m/tube in a volume of 10ul with a final concentration of 0.25mM acetyl-coA, physostigmine sulfate (0.2mM, Sigma) and incubation buffer (containing: 20mM choline chloride, 20mM EDTA, 600mM NaCl and 50mM sodium sulfate buffer at pH 7.4). Reaction was timed and terminated by 4mL of reaction rinse at 4°c containing 10mM sodium phosphate buffer (Sigma). Then, 1.6mL of acetonitrile (Sigma, Cat# 360457) containing 5mg/mL tetrephenylboron (TCI America, Portland, OR, USA, Cat# 143–66-8-A5130) was added and 6.4mL of followed by Insta-Fluor Plus (PerkinElmer, Cat# 6013167). Reaction samples were stored in 4°c for 24 hours allowing phase separation. Amount of [³H]-acetylcholine produced was determined by counting total cpm emitted from the organic phase in 8mL EconoFluor scintillation cocktail (PerkinElmer, 6NE9699). Identical tubes to which no sample was added were used to determine background. Background cpm was subtracted from total cmp. Total protein in the reaction tube was determined by Bradford assay. Average from three reaction tubes per sample were calculated and ChAT activity was calculated as pmol of ACh synthesized / hr / μg protein.

2.11. Statistical Analysis

Statistical comparisons were conducted using JMP Pro software (SAS, version 13.0.0). Statistical significance was defined as p < 0.05 and results are presented as mean ± SEM.

Experiment 1

Hormone levels, WB data and ChAT activity were first analyzed for differences between tissues collected at proestrus vs. diestrus by two-tailed t-test. Next, comparisons among all 6 groups were analyzed for each target by one-way analysis of variance (ANOVA). For WB, each gel was designed to include representatives from all 6 groups and WB data were normalized to the proestrus group [1]. ChAT activity and WB data also were analyzed by 2-way ANOVA by using Model and Time as factors and omitting the data from normal cycling rats. Post-hoc comparisons were made using either a Tukey test to compare all groups or a Dunnet test to compare groups with the proestrus controls. These analyses were conducted for each target in each region of the brain. For WB, an additional set of gels were run and included samples of OVX and VCD animals at 1 and 6 weeks (data not reported). These data were normalized to OVX-1W and demonstrate remarkable agreement with data generated by gels containing all 6-groups, demonstrating excellent reproducibility of these methods.

Experiment 2

Hormone levels and ChAT activity were analyzed by 3-way ANOVA with Model, Time, and Agonist Treatment as between factors. Effect of Agonist Treatment on ChAT activity were analyzed by a series of one-way ANOVAs comparing five Agonist Treatment groups per each menopausal model (OVX or VCD) and one of the two time points (1 or 6 weeks). Significant main effects and interaction effects were further explored using Tukey post-hoc comparisons.

Statistical analysis of the WB data required a different approach. In experiment 2, each gel was organized to include representatives of all agonist treatments and the appropriate controls, using samples from one menopausal model (OVX or VCD) and one of the two time points (1 or 6 weeks). Results were normalized to the respective vehicle-treated controls included on the same gel. Thus, the quantified values represent fold-change relative to the control group for a specific model and time point (e.g., agonist effects for the OVX-1W group reflect fold change relative to OVX-1W controls; agonist effects for the VCD-1W group reflect fold change relative to VCD-1W controls, etc…). These data were each analyzed by one-way ANOVA. The lack of an internal control present on all gels prevents a between gel comparison. Hence this portion of the analysis does not capture potential differences between the four vehicle-treated control groups (OVX-1W-Veh, OVX-6W-Veh, VCD-1W-Veh and VCD-6W-Veh).

To mitigate this limitation a separate set of gels were run, each gel containing samples from the four different control groups. These data were normalized to the OVX-1W controls and reflect fold differences between the control groups. Each of these blocks of data were then analyzed by 1-way ANOVA either comparing the controls, or comparing effects of agonists within each model and time-point. Post-hoc comparisons were conducted using the Tukey test.

3. Results

3.1. Experiment 1: Comparison of Surgical and Transitional Models of Menopause Relative to Proestrus and Diestrus

3.1.1. Serum Hormone Levels

Serum levels of E2, T and AD are summarized in Figure 2 and Supplemental Table 2. Hormone levels are very similar to those previously reported [36], but with some small variation due to the fact that a small number of samples were lost to processing. Hence some rats that were included in the previous analysis are not included here and vice-verse. Briefly, levels of all three hormones were significantly higher at proestrus than at diestrus. Mean serum levels of circulating hormones in proestrus rats were E2: 50.6 ± 4.5, T: 186.9 ± 43.2 and AD: 196.5 ± 42.9, and levels in diestrus rats were 13.9 ± 2.8 pg/mL for E2, 92.2 ± 14.5 pg/mL for T, and 67.0 ± 7.2 pg/mL for AD. Two-way ANOVA revealed significant differences in hormone levels as a function of model and time. Also as previously reported [35], levels of E2, T and AD were significantly higher at proestrus than in VCD-treated rats (p < 0.05 for all cases). In contrast, levels of the hormones in VCD-treated rats did not differ significantly from levels detected in diestrus rats. None of the hormones were detectable in OVX rats, whereas VCD-treated rats had significantly higher levels of E2, T and AD than OVX rats (p < 0.05 in each case). This was true both at 1-w and at 6-w after treating with VCD. This is consistent with the selective loss of ovarian follicles and the preservation of interstitial cells after treating with VCD [22, 25, 48–50].

Figure 2:

Serum hormone levels of (A) E2 (17β-estradiol), (B) T (testosterone) and (C) AD (androstenedione) by time (1- or 6- weeks) and rat model of: proestrus, diestrus, surgical (OVX; O) or transitional (VCD; V) menopause. All three hormones were below detection limits in OVX rats. One-way ANOVA: ** p < 0.01 and *** p < 0.001 are significant compared to all other groups; # p < 0.05 compared to VCD-1w; ^$ $ p < 0.01 compared to P and ^¥¥ p < 0.01 compared to P and D. Data shown as Mean ± SEM.

3.1.2. Comparison of target protein levels and ChAT activity in tissues collected at proestrus vs. diestrus

Representative images showing detection of total protein and multiple protein targets are provided in Figure 3. Figure 5 and Supplemental Table 2 summarize the levels of target proteins detected in tissues collected at proestrus vs. diestrus. Differences varied by brain region. For example, levels of β-actin were significantly lower at diestrus than at proestrus in the (p < 0.02) and FCX (p < 0.01), with a strong trend in the STR (p < 0.06). In the FCX levels of the enzymes ATP-CL, PDH and GAPDH also were significantly lower at diestrus than at proestrus (p < 0.05 in each case), as well as a trend towards lower levels of α-tubulin (p = 0.074). In the STR levels of PDH were significantly higher in tissues collected at diestrus than at proestrus (p = 0.02). In contrast, no significant differences in ChAT activity were detected in tissues collected at diestrus vs. proestrus in any of the three brain regions examined (Figure 4).

Figure 3:

Representative images from a single PVDF membrane with multiplexed fluorescence signals densitometry of total protein stain and five individual proteins. (A) TP stain produced consist signal (red) across lanes loaded with different samples at 12 mg-protein/lane; and (B) densitometry of single-band targets were measured on the same membrane for: ATP-CL (122 kDa), PFK (85 kDa), α-tubulin (50 kDa), PDH-E1α (43 kDa) and GAPDH (36 kDa). Normalization to total protein allowed for between-gel comparisons along with wide linear range of detection and low signal variability as described previously [1].

Figure 5:

Effects of transitional (VCD; V) and surgical (OVX; O) menopause rat models on the expression of cytoskeletal proteins and enzymes involve in glucose metabolism and acetyl-CoA production. Gonadally intact rats at proestrus (P) and diestrus (D) are used as controls. Bars shows Mean ± SEM and normalized to P. Expressions were evaluated in three brain regions: HPC (A), FCX (B) and STR (C) at two post-menopausal time points (1- and 6-weeks following the completion of VCD treatments or OVX surgery). Symbol repetition represents level of significance. One-way ANOVA: * Compared to P with * p < 0.05, ** p < 0.01, *** p < 0.001. Symbols are: * Compared to P, # Compared to D, $ Compared to OVX-1, ± Compared to OVX-6, § Compared to VCD-1, ¥ Compared to VCD-6. T-test between P and D shown by ȼ.

Figure 4:

Choline acetyltransferase (ChAT) activity in HPC, FCX and STR of proestrus (P), diestrus (D), surgical (OVX; O) or transitional (VCD; V) menopause rat models. Data shown as Mean ± SEM. No significant differences in ChAT activity were detected in tissues collected from gonadally intact rats at diestrus vs. proestrus and not in comparison to all other groups by either one-way or two-way ANOVA in any of the three brain regions examined.

3.1.3. Effects of OVX and VCD Treatments on target protein levels and ChAT activity in comparison with tissues collected at proestrus and diestrus

Effects of OVX and VCD-treatments on target proteins were both brain region-specific and time-dependent. These data are illustrated in Figures 4 and 5 and are organized by brain region below. The data also are summarized in Supplemental Table 2 along with statistics.

HPC

One-way ANOVAs revealed significant effects on levels of PFK, α-tubulin, and PDH, with trends for effects on levels of ATP-CL and GAPDH. Higher levels of PFK were detected in VCD-1W than in OVX-1W (p = 0.003), and in OVX-6W than in OVX-1W (p = 0.026). Higher levels of α-tubulin were detected in both VCD-6W and VCD-1W compared to P, D and OVX-1W and were higher in VCD-6W compared to OVX-6W (p < 0.05 for all comparisons). Levels of PDH were higher in OVX-1W than VCD-6W (p = 0.01).

Two-way ANOVAs of Model and Time (omitting the P and D data) revealed significant effects on levels of PFK, α-tubulin and PDH. Across time, levels of α-tubulin were higher in VCD-treated rats than in OVX rats, whereas levels of PDH were higher in OVX rats than in VCD-treated rats. Across models, levels of PDF were higher at 1w than at 6w, whereas levels of GAPDH were higher at 1w than at 6W. There also was a significant Model * Time interaction on levels of PFK. Specifically, VCD-1W had higher levels of PFK than OVX-1W and VCD-6W had lower levels of PFK than OVX-6W. ANOVA also revealed a non-significant trend for an effect on ChAT activity in the HPC with lower activity in VCD-6W compared with diestrus (p = 0.05). No other effects on ChAT activity were detected in the HPC by one-way or by two-way analyses.

FCX

One-way ANOVAs revealed significant effects on levels of ATP-CL, PFK, and PDH. Higher levels of ATP-CL were detected at proestrus than at diestrus (p = 0.007), OVX-1W (p = 0.006) and OVX-6W (p = 0.004). In addition, levels of ATP-CL were higher in VCD-1W vs. OVX-6W (p = 0.04), and higher in VCD-6W than in diestrus (p = 0.004), OVX-1W (p = 0.003) and OVX-6W (p = 0.002). Levels of PFK were significantly higher in tissues collected at proestrus than in OVX-1W (p = 0.035). Levels of PDH likewise were significantly higher at proestrus than at diestrus or in OVX-1W, OVX-6W and VCD-6W (all p < 0.03). Two-way ANOVAs revealed an overall effect on ATP-CL. Across time, overall levels of ATP-CL were higher in VCD-treated rats than in OVX rats. No other significant effects were detected. No significant effects on ChAT activity were detected in the FCX as a function of OVX or VCD treatments either by one-way or by two-way analyses.

STR

One-way ANOVA revealed significant effects on ATP-CL, PFK, and β-actin. Higher levels of ATP-CL were detected in VCD-1W vs. diestrus (p = 0.045). High levels of PFK were detected in VCD-1W vs. proestrus (p = 0.0005), diestrus (p = 0.011), VCD-6W (p = 0.0008) and OVX-1W (p = 0.014). Higher levels of PFK also were detected in OVX-6W vs. proestrus (p = 0.008), diestrus (p = 0.017), and VCD-6W (p = 0.012). Expression of β-actin was higher in OVX-1W rats vs. diestrus (p = 0.034) and in OVX-6W rats vs. diestrus (p = 0.007) and proestrus (p = 0.04).

Two-way ANOVA revealed a significant overall effect on levels of PFK_. A significant Model * Time interaction on the levels of PFK was detected. Higher levels of PFK were detected in in VCD-1W vs. OVX-1W, whereas lower levels of PFK were detected in VCD-6W vs. OVX-6W. In addition, two-way ANOVA for ATP-CL showed a non-significant trend (p=0.095). Across both models, overall levels of ATP-CL were higher at 1-week than at 6-weeks. No significant effects on ChAT activity were detected in the STR by either one-way or two-way analyses.

3.2. Experiment 2: Effects of estrogen receptor agonists

3.2.1. Serum hormone levels

Serum levels of E2, T and AD are illustrated in Figure 6. Statistics are summarized in Supplemental Tables 3 and 4. As with Experiment 1, hormone levels are very similar to those previously reported [35], but with some variation due to the fact that in each analysis a small number of samples were lost to processing. Three-way ANOVAs revealed significant overall effects on E2, T, and AD.

Figure 6:

Serum hormone levels of (A) E2 (17β-estradiol), (B) T (testosterone) and (C) AD (androstenedione) in rat models of surgical (OVX; O) or transitional (VCD; V) menopause following treatment with estrogen receptor agonists. Data shown at two time points, following 1- and 6- weeks of continuous treatment with selective agonist: Vehicle, E2, PPT (a selective ERα agonist), DPN (a selective ERβ agonist) and G-1 (a selective GPER1 agonist). One-way ANOVA: * p < 0.0001. Three-way ANOVA: ** p < 0.0001 for the main effect of model, where VCD animals had significantly higher levels of T and AD regardless of treatments and time. Treatment and time had no significant effect on levels T. For AD in VCD rats, E2-treated rats had significantly lower levels of AD than DPN-treated rats, regardless of time. Data shown as Mean ± SEM.

E2: As expected, E2 levels were significantly elevated in all rats that received E2 treatment. In all E2-treated OVX and VCD animals, 1 week of E2 treatment produced levels in the high physiological range (86.5 ± 16.45 pg/mL) whereas 6-weeks of treatment produced levels well above the physiological range (225.7 ± 16.45 pg/mL). Significant main effects of Agonist, Time, and an Agonist * Time interaction were detected. Post Hoc analysis revealed that E2 levels were significantly higher in OVX and VCD rats treated with E2 than in any other groups. As expected, the levels of E2 in OVX rats treated with vehicle, PPT, DPN or G-1, were extremely low or below the level of detection at both time points.

T: As expected, levels of T were consistently higher in VCD-treated rats than in OVX rats. Three-way ANOVA revealed main effects of Model, Agonist, and a significant Model * Agonist interaction. Post-hoc analysis showed that levels of T were significantly higher in VCD than in OVX rats. In addition, T levels in rats treated with E2 were significantly lower than in rats treated with DPN regardless of Time and Model.

AD: As with T, three-way ANOVA of AD levels revealed significant main effects of Model, Agonist and Time, as well as a significant Model * Agonist interaction and a significant Model * Time interaction. Post-hoc analyses show that AD levels were significantly higher in VCD vs OVX rats regardless of Agonist Treatment and Time, and higher at 1 week than 6 weeks regardless of Model and Agonist Treatment. Compared to Vehicle, levels of AD were higher in DPN-treated rats and lower in E2-treated rats.

In VCD animals, AD levels were higher after 1-week than after 6-weeks of continuous treatment. In addition, when collapsed across time, AD levels were significantly lower in E2-treated rats and higher in DPN-treated rats compared to vehicle-treated rats.

3.2.2. Effects of ER Agonists on ChAT activity in each brain region as a function of model and time

Overall effects on ChAT activity were detected in the HPC and FCX, but not the STR. Effects differed among brain regions as well as a function of model and time. The majority of effects on ChAT were detected in the HPC. These data are shown in Figure 7 and are organized by brain region below. Multi-way ANOVAs are summarized in Supplemental Table 5 and 1-way ANOVAs in Supplemental Table 6.

Figure 7:

Activity of choline acetyltransferase (ChAT) in HPC, FCX and STR following contentious treatments with selective estrogen agonists for 1- and 6- weeks in rat models of surgical (OVX) and transitional (VCD) menopause. In the HPC, following 6-weeks of treatment in OVX rats, all four treatments resulted in higher ChAT activity than vehicle (OVX-6w-HPC) whereas in VCD only E2 and G1, but not PPT and DPN, had higher ChAT activity than vehicle. Also notable is the menopause model difference in the FCX where following 6-weeks of E2 treatment ChAT activity was higher than vehicles only in VCD animals (VCD-6W-E2) but not OVX (OVX-6W-E2). One-way ANOVA: * p < 0.05 and ** p < 0.001 compared to Veh; $ p < 0.05 compared to PPT. Data shown as Mean ± SEM.

HPC

ANOVA revealed a significant main effect of Model on ChAT activity in the HPC. Overall, ChAT activity was slightly higher in VCD-treated rats than in OVX rats (p=0.03). A significant main effect of Agonist also was detected, as well as significant interactions of Agonist * Time and Model * Agonist * Time. Tukey analysis revealed that levels of ChAT activity were significantly higher in rats treated with, E2, G1 and PPT relative to DPN or vehicle when collapsed across Model and Time (p<0.05A in each case). ChAT activity in E2-treated rats was higher after 6-weeks vs. 1-week of treatment when collapsed across Model (p<0.05). ChAT activity in rats treated with PPT was highest after 1-week of treatment (p<0.05).

Tukey analysis also revealed that effects of Agonist Treatment and Time on ChAT activity differed as a function of Model. For example, in VCD-treated rats ChAT activity in the 1-week controls was significantly higher than at 6-weeks and significantly higher than in OVX controls at both 1-week and 6-weeks. ChAT activity in VCD rats treated with PPT for 1 week were significantly higher than all other agonist treatment groups in both OVX and VCD-treated rats at both the 1- and 6- week time points. In addition, VCD-treated rats treated with G1 for 6-weeks had significantly higher levels of ChAT activity than OVX-1W-Veh, OVX-6W-Veh, OVX-1W-E2, VCD-6W-Veh and VCD-6W-PPT (p<0.05 in all cases).

FCX

ANOVA revealed a significant main effect of Time on ChAT activity in the FCX. Overall, ChAT activity in the FCX was significantly higher at the 1W time point than at the 6W time point. There also was a trend for levels to be higher in VCD-treated vs. OVX rats (p=0.08). A significant Treatment * Time interaction also was detected where ChAT activity was lower at 6-weeks than at 1-week in vehicle-, PPT- and G-1-treated rats, but not in E2- or DPN-treated rats. No other significant main effects or interactions were detected.

STR

No main effects were detected on ChAT activity; however, the interaction of Agonist Treatment * Time showed a strong trend where rats treated with PPT for 1-week had higher ChAT activity than vehicle- and DPN-treated rats at 1-week and E2-treated rats at 6-weeks.

3.2.3. Effects of agonist treatments on target protein levels in each brain region as a function of model and time point

Treatment with ER agonists produced significant effects on the levels of target proteins in all three brain regions. Effects differed as a function of both model and time-point. These data are summarized in Figures 8 – 10 and are organized by brain region below. Within each region results are organized by analysis of OVX-1W, OVX-6W, VCD-1W, and VCD-6W. Note that analysis of these data are limited to one-way ANOVAs due to limitations in the design and collection of the WB data as described under Methods. Statistics are summarized in Supplemental Table 6.

Figure 8:

Effects of 1- and 6- weeks of continuous treatment with selective estrogen agonists on the expression of cytoskeletal proteins and enzymes involve in glucose metabolism and acetyl-CoA production in the HPC of transitional (V) and surgical (O) menopause rat models. Changes in expression were evaluated at two time points representing 1- and 6- weeks of continuous treatments with either Vehicle, E2, PPT (a selective ERα agonist), DPN (a selective ERβ agonist) and G-1 (a selective GPER1 agonist). Bars represents Mean ± SEM and normalized to vehicle treated group. Symbol repetition represents level of significance. One-way ANOVA: * Compared to Vehicle with * p < 0.05, ** p < 0.005, *** p < 0.0001. Symbols are: * Compared to Vehicle; # Compared to E2; $ Compared to PPT; ± Compared to DPN; § Compared to G1; Dunnett’s test: ¥ Compared to Veh.

Figure 10:

Effects of 1- and 6- weeks of continuous treatment with selective estrogen agonists on the expression of cytoskeletal proteins and enzymes involve in glucose metabolism and acetyl-CoA production in the STR of transitional (V) and surgical (O) menopause rat models. Changes in expression were evaluated at two time points representing 1- and 6- weeks of continuous treatments with either Vehicle, E2, PPT (a selective ERα agonist), DPN (a selective ERβ agonist) and G-1 (a selective GPER1 agonist). Bars represents Mean ± SEM and normalized to vehicle treated group. Symbol repetition represents level of significance. One-way ANOVA: * Compared to Veh with * p < 0.05, ** p < 0.005, *** p < 0.0001. Symbols are: * Compared to Veh; # Compared to E2; $ Compared to PPT; ± Compared to DPN; § Compared to G1; Dunnett’s test: ¥ Compared to Veh.

HPC

OVX rats

Agonist treatments significantly affected levels of ATP-CL, PFK, PDH, and GAPDH at the 1W time point. Post-hoc analyses revealed significantly lower levels of both ATP-CL and PFK, and significantly higher levels of GAPDH in E2-treated rats vs. controls. DPN and G-1 treated rats also had significantly lower levels of ATP-CL and PFK and higher levels of GAPDH than controls. GAPDH levels also were significantly higher in PPT-treated rats vs. controls. Strong trends for reductions in PDH levels in E2- (p = 0.08) and G-1-treated rats (p = 0.054) vs. controls also were detected. At the 6W time point, agonist treatments significantly affected levels of α-tubulin, PDH, and GAPDH. Post-hoc analyses detected no significant effects of E2, however, there was a trend for lower levels of GAPDH in E2-treated rats than in controls (p = 0.08). Similarly, G-1 treatment showed a trend for lower GAPDH levels in E2-treated rats than in controls (p = 0.055). Rats treated with DPN had significantly higher levels of GAPDH and α-tubulin compared to all other agonist treatments (E2, PPT and G-1). Rats treated with DPN also had higher levels of PDH than rats treated with E2- or G-1.

VCD-treated rats

Agonist treatments significantly affected α-tubulin levels, but not the other targets. Levels of α-tubulin in E2 treated rats were significantly lower than in controls or in rats treated with PPT or G-1. PPT-treated rats had significantly lower levels of α-tubulin than controls and rats treated with DPN. At the 6W time-point, agonist treatments significantly affected levels of both α-tubulin and PDH. Specifically, E2-treated rats had significantly lower levels of α-tubulin than controls and lower levels of PDH than PPT- and DPN- treated rats. Rats treated with PPT had significantly lower levels of α-tubulin than controls and DPN-treated rats.

FCX

OVX rats

Agonist treatments significantly affected levels of ATP-CL at the 1W time point, but not the other targets. Post-hoc analysis revealed significantly lower levels of ATP-CL in E2-treated rats vs. controls. DPN and G-1 treated rats also had significantly lower levels of ATP-CL than controls. Trends for overall effects of Agonists on PFK (p = 0.08) and α-tubulin (p = 0.06) also were detected. DPN-treated rats had significantly lower levels of PFK vs. controls (p = 0.04). At the 6W time-point, DPN treated rats had significantly lower levels of ATP-CL and PFK than G-1-treated rats. α-tubulin levels were significantly higher in DPN-treated rats than in E2-, PPT- or G-1 treated rats.

VCD-treated rats

Agonist treatments significantly affected levels of α-tubulin, PDH, and GAPDH at the 1W time-point. No significant differences between E2-treated rats and controls were detected; however, E2-treated rats had lower levels of PDH and GAPDH than PPT-, DPN- and G-1 treated rats. G-1 treated rats had significantly lower levels of α-tubulin than controls. DPN-treated rats had significantly higher levels of α-tubulin than E2-, PPT- and G-1-treated rats. At the 6W time-point, G-1 treated rats had significantly higher levels of PDH and GAPDH than DPN-treated rats.

STR

OVX rats

Agonist treatments significantly affected levels of GAPDH, but not the other targets, at the 1W time-point. E2-treated rats did not differ significantly from controls. G-1 treated rats had higher levels of GAPDH than vehicle- and E2- treated rats. GAPDH levels also were higher in DPN-treated vs. E2-treated rats. At the 6W time-point, PPT treated rats had significantly lower levels of PFK than E2-, DPN- and G-1 treated rats. DPN-treated rats had significantly higher levels of PDH than all other groups. PDH levels also were higher in E2- vs. PPT- treated rats.

VCD-treated rats

Agonist treatments significantly affected levels of ATP-CL and α-tubulin at the 1W time-point. E2-treated rats did not differ significantly from controls. DPN-treated rats had significantly lower levels of ATP-CL relative to PPT- and G-1 treated rats. PPT- and G-1 treated rats had significantly lower levels of PDH than vehicle- and DPN- treated rats. PDH levels also were lower in G-1- vs. E2-treated rats. At the 6W time-point, DPN treated rats had significantly lower levels of ATP-CL than vehicle-treated controls.

3.2.4. Comparison of vehicle controls as a function of model and time point

As mentioned under Methods, a separate set of gels were needed to capture potential differences among the four vehicle-treated control groups in Experiment 2. These data are summarized in Supplemental Figure 1 and Supplemental Table 7. Several effects of model and time are worth noting. For example, in the HPC levels of GAPDH and ATP-CL were significantly higher at 6 weeks than at 1 week in both models. Levels of PDH at 1 week were lower in VCD-treated rats than in OVX rats. Levels of α-tubulin were lower at 6 weeks than at 1 week in OVX rats, but not in VCD-treated rats, and levels of PFK were higher at 6 weeks than at 1 week in OVX rats, but were higher at 1 week than at 6 weeks in VCD-treated rats. Fewer differences were detected in the FCX and STR. In the FCX levels of ATP-CL were higher in VCD-treated rats than in OVX rats, and were significantly higher in VCD-treated rats at the 6-week time point than in all other groups. In the STR, levels of PFK were substantially higher in VCD-treated rats at the 1-week time point than in all other groups. Levels had declined substantially by 6 weeks. Levels of PFK at the 6-week time point were significantly lower than levels at the 1-week time point for both models. These data demonstrate that the levels of glycolytic enzymes and other target proteins can change significantly over time following the loss of ovarian function, and that in some cases the changes differ between surgical and transitional models of menopause. These differences need to be considered when comparing effects of agonists across models and time-points.

4. Discussion

The primary goal of this study was to characterize and compare the effects of two clinically relevant models of surgical and transitional menopause, as well as treatment with different estrogen receptor agonists, on targets related to glucose utilization and energy production (GAPDH, PDH, PFK), cytoskeletal function and synaptic remodeling (β-actin, α-tubulin), cytoplasmic acetyl-CoA production (ATP-CL), and choline acetyltransferase (ChAT) activity. Three regions of the brain were examined. The hippocampus because of its critical role in learning and memory consolidation. The frontal cortex because of its role in learning, attention, and cognitive performance. And the striatum because it is a major component of the extrapyramidal motor system and plays an important role in associative and motor learning.

4.1. Experiment 1

Experiment 1 focused on differences between OVX and VCD models in comparison with gonadally intact rats collected at proestrus (P) and diestrus (D). In the FCX, significantly lower levels of GAPDH, PDH, β-actin, and ATP-CL were detected at D vs. P, suggesting that in this region of the brain, fluctuations in these endpoints occur in association with the estrous cycle, and are lower at D when gonadal hormones also are low. In contrast, in the HPC only β-actin was significantly reduced at D vs. P, and in the STR no significant differences between P and D were detected. These data demonstrate that physiological fluctuations in these endpoints occur across the estrous cycle but are brain region specific.

Loss of ovarian function likewise produced brain region-specific effects which varied with time and model. Serum hormone levels confirmed successful achievement of each of the two models of menopause. As expected, all three hormones were undetectable following OVX. In contrast, rats treated with VCD had low but detectable levels of E2, T and AD comparable to diestrus rats. This is consistent with prior reports [1, 35, 36, 50] and with the preservation of androgen-producing stromal cells in the ovaries of VCD-treated rats [22, 23, 48].

In the FCX, lower levels of protein targets were consistently detected at both the 1W and 6W time points following OVX and VCD treatments relative to proestrus rats. Several of these reached statistical significance including effects on PDH, PFK (OVX-1W), and ATP-CL (OVX-1W & OVX-6W). Notably, ATP-CL levels were not significantly lower in VCD-treated rats compared to rats at proestrus which may indicate that cytoplasmic production of acetyl-CoA from citrate is less effected in transitional vs. surgical menopause. Lower levels of PDH in both VCD-treated and OVX rats are consistent with the lower levels detected in the FCX at D vs. P and are consistent with the hypothesis that glycolysis in this region is reduced when gonadal hormone levels are low [51–53].

Very different results were detected in the HPC and STR. At the 1W time point, VCD-treated rats had higher levels of PFK in the HPC, and higher levels of both PFK and ATP-CL in the STR. In contrast, OVX rats had significantly higher levels of PFK in both the HPC and STR at the 6W time point. This suggests a fundamental difference between the two models in glucose utilization/response, i.e., that in the HPC and STR of VCD-treated rats the capacity to commit glucose to glycolysis is greater at the 1W time-point than in OVX rats. This could translate into less of a reduction in glucose utilization and energy production at early time points following transitional vs surgical menopause.

Notably, neither of the menopausal models were associated with significant reductions in ChAT activity. This was unexpected given studies showing changes in ChAT activity in the FCX across the estrous cycle [54], fluctuations in ChAT mRNA in the MS and NBM across the estrous cycle [55–57], as well as higher levels of ChAT mRNA in the HPC and FCX of OVX rats following short term estrogen treatment [58, 59]. The reason for this discrepancy is unknown. In the case of VCD-treated rats, sustained low levels of E2, T and AD may contribute to the maintenance of normal levels of ChAT activity.

4.2. Experiment 2

Experiment 2 focused on characterizing the effects of E2 and of selective ER agonists on neurochemical targets, at two time points in OVX and VCD-treated rats. Again, hormone levels in the vehicle-treated controls demonstrated successful representation of the two menopausal models. As expected, E2-treated rats treated had significantly higher levels of E2 than other groups. As previously reported [35], VCD-treated rats that received DPN for 6 weeks had significantly higher levels of T and a strong trend for elevations in AD. The exact mechanism of this currently is unknown; however, both thecal cells and granulosa cells have been shown to express ERβ [60]. Hence it is possible that DPN acting at ERβ in thecal cells increase androgen biosynthesis, or possibly inhibits the conversion of androgens to estrogens in the granulosa cells. This needs to be investigated.

Effects of Agonist Treatments

Effects on multiple targets were detected in response to agonist treatments. These effects were both model and time dependent as well as brain region specific. For clarity, we have organized the discussion of these effects by brain region.

HPC

Effects of agonists in the HPC were both model- and time-dependent. Several effects are particularly notable. In OVX rats, E2 treatment initiated immediately following OVX resulted in significantly lower levels of glycolytic enzymes and ATP-CL at the 1W time point (Figure 8A). Similar effects were produced by DPN and G-1. PPT treatment resulted in reductions in GAPDH, but not the other targets. These data suggest that most of the effects of E2 on these endpoints are mediated by activation of ERβ and/or GPER1, with effects on GAPDH also mediated by ERα. The effects are consistent with significantly reduced metabolic activity in the HPC at this early time point, corresponding with reductions in glycolysis, ATP production, and cytosolic acetyl-CoA. The data also may indicate decreased demands for the synthesis of cholesterol and triglycerides, as well as decreases in glucose-induce insulin secretion, protein isoprenoid-based modifications, and histone acetylation [61, 62]. Such decreases in ATP-CL can lead to a condition of energy deficit by affecting intermediates of the TCA cycle and glucose metabolism [63]. These findings are consistent with our recent results showing significant reductions in the levels of tryptophan and tyrosine in the HPC 1W after OVX [44], and further reductions in these amino acids after 1W treatment with ER agonists [35].

Many of the effects observed after 1W of treatment were reduced or even reversed after 6W of treatment (Figure 8B). For example, the lower levels of PFK and ATP-CL were no longer statistically significant, and the negative effects of DPN on levels of GAPDH and PDH were no observed. In contrast, significantly higher levels of α-tubulin were detected after 6W treatment with DPN relative to all other groups. These data demonstrate that some of the negative effects of ER agonist treatment on glycolytic enzymes and ATP-CL lessen with time despite continued treatment, whereas effects on tubulin emerge and appear to be ERβ-specific. Hence, in this brain region, sustained treatment with ER agonists following OVX has negative effects on metabolic activity which recover over time and may be followed by higher levels of specific cytoskeletal proteins.

ChAT activity also was significantly higher at the 6W time point in response to treatment with each of the four ER agonists. This suggests that continued activation of any of the three ERs is sufficient to produce higher ChAT activity in the HPC of OVX rats. These data are consistent with the increases in high affinity choline uptake and acetylcholine release that have been reported in OVX rats following sustained or repeated treatment with E2 or G-1 [42, 64, 65].

Effects of agonists in the HPC of VCD-treated rats were quite different from those in OVX rats. For example, Figure 8C shows that in VCD-treated rats, 1W of agonist treatment produced no statistically significant changes in the levels of glycolytic enzymes or ATP-CL relative to controls. In contrast, significant reductions in α-tubulin were detected in rats treated with E2, PPT, or G-1. After 6W of treatment reductions in tubulin continued to be detected. Significantly higher levels of ChAT were detected in VCD-treated rats after 6W treatment with E2 or G1, but not PPT or DPN. Collectively, these findings demonstrate that the effects of ER agonists on the levels of glycolytic enzymes, ATP-CL, and tubulin in the hippocampus are fundamentally different in VCD-treated vs. OVX rats. In addition, unlike OVX rats, the data suggest that effects of E2 on ChAT activity in VCD-treated rats requires activation of GPER1.

FCX

Effects of agonists in the FCX also were model- and time-dependent, and were different from those detected in the HPC. Figure 9A shows that in OVX rats, 1W of E2 treatment had no significant effects on glycolytic enzymes, but did produce significantly lower levels of ATP-CL. Similar effects were produced by PPT, DPN and G-1, suggesting that activation of any of the three ERs is sufficient to lower the levels of ATP-CL. Lower levels of PFK were detected in rats treated with DPN, but not in response to E2, PPT, or G-1. Hence short-term agonist treatment following OVX appears to have relatively little effect on glycolytic enzymes and related metabolic activity in the FCX. This is consistent with results showing no significant reductions in TRP or TYR in the FCX in response to OVX or agonist treatments [44]. The lower levels of ATP-CL were similar to effects in the HPC and suggest that this is one effect that is consistent between the two brain regions. Note that by 6W no statistically significant effects of agonist treatments were detected relative to controls. Lower levels of α-tubulin were observed in response to E2, PPT and G-1 as in the HPC, but were not statistically significant. This suggests some consistency in the effects of ER agonists administered immediately following OVX on α-tubulin in these two brain regions.

Figure 9:

Effects of 1- and 6- weeks of continuous treatment with selective estrogen agonists on the expression of cytoskeletal proteins and enzymes involve in glucose metabolism and acetyl-CoA production in the FCX of transitional (V) and surgical (O) menopause rat models. Changes in expression were evaluated at two time points representing 1- and 6- weeks of continuous treatments with either Vehicle, E2, PPT (a selective ERα agonist), DPN (a selective ERβ agonist) and G-1 (a selective GPER1 agonist). Bars represents Mean ± SEM and normalized to vehicle treated group. Symbol repetition represents level of significance. One-way ANOVA: * Compared to Veh with * p < 0.05, ** p < 0.005, *** p < 0.0001. Symbols are: * Compared to Veh; # Compared to E2; $ Compared to PPT; ± Compared to DPN; § Compared to G1; Dunnett’s test: ¥ Compared to Veh.

Effects in VCD-treated rats again were different from those in OVX rats. In VCD-treated rats, 1W of E2 treatment resulted in lower levels of PDH relative to controls, but did not significantly affect other targets. In contrast, treatment with PPT, DPN or G-1 resulted in significant higher levels of GAPDH and PDH, suggesting that selective activation of ERα, ERβ, or GPER1 alone may increase glycolytic enzymes in the FCX of VCD-treated rats at this early time-point. Since E2 did not have the same effect, this suggests that under specific circumstances, activation of multiple ERs can diminish or reverse regional effects associated with activation of any one ER. After 6W of treatment, many of the effects on GAPDH, PFK and α-tubulin were no longer observed. Exceptions were the significantly higher levels of GAPDH and PDH in rats treated with G-1, suggesting that continuous activation of GPER1 can elevate the expression of these glycolytic enzymes in the FCX of VCD-treated rats.

No significant effects of agonist treatments on ChAT activity were detected in the FCX of OVX rats. In VCD-treated rats agonist treatments had no significant effect on ChAT activity at the 1W time point. E2 produced slightly higher levels of ChAT activity at the 6W time point, but overall there was very little evidence of ER agonist effects on ChAT in the FCX. Again, this was surprising given prior studies showing effect of E2 on ChAT mRNA in the NBM () and ChAT activity in the FCX (Gibbs 2000). Work from our laboratory, however, has shown that effects are dose-dependent, can vary considerably according to the manner and regimen of hormone replacement, and do not persist with prolonged treatment [66]. These factors may help to explain the relative lack of effects on ChAT activity detected in the FCX.

STR

Effects of agonists in the STR were diverse and, as in the other brain regions, were both model- and time dependent (Figure 10). Notably, treatment with E2 produced no statistically significant effects relative to controls in either OVX or VCD-treated rats at either time-point. Hence in contrast to HPC and FCX, treatment with E2 has comparatively little effect on the levels of glycolytic enzymes, α-tubulin, and ATP-CL in the STR. In OVX rats, 1W treatment with G1 resulted in significantly higher levels of GAPDH relative to controls and other treatment groups. No other significant differences were detected at the 1W time point, suggesting the effect was specific to activation of GPER1. Several effects were detected at the 6w time point, including higher levels of PDH in rats treated with DPN, and lower levels of PFK in rats treated with PPT. This suggests that activation of ERα and ERβ can selectively affect different stages of glucose metabolism. Effects were different in VCD-treated rats, again indicating that the effects of ER agonists differ between the two models and are region-specific.

Collectively, these results indicate that ER agonists have much more limited effects on these targets in the STR than in the HPC and FCX. This would suggest more limited effects on metabolic activity, ATP production, and the demands for cytosolic acetyl-CoA in this brain region. Effects of selective ER agonists were detected; however different agonists often had different or opposite effects, which is consistent with the fact there were no significant effects of E2. This also is consistent with the low levels of ERα and ERβ in the STR in comparison with HPC and FCX, and with our prior results showing very limited effects on amino acid levels and neurotransmitter endpoints in the STR of OVX and VCD-treated rats [35].

GPER1GPER1GPER1

Agonist treatments also had no significant effect on ChAT activity in the striatum. Again, this is consistent with the fact that cholinergic neurons in the striatum do not contain ERα or ERβ receptors. The neurons do, however, contain GPER1 receptors comparable to cholinergic neurons in the septum and diagonal [42]. Treatment with a selective GPER1 agonist has been shown to increase potassium-stimulated acetylcholine release in the HPC [42], suggesting that activation of GPER1 is sufficient to increase the functionality of cholinergic projections to this region of the brain; however, similar effects on cholinergic neurons in the striatum have not been demonstrated.

4.3. Limitations

While these studies provide a large amount of new and valuable data, several limitations need to be acknowledged. (1) VCD studies were conducted in young adult rats. Aging itself may interact with loss of ovarian function in ways that are not captured in the present studies. (2) Changes in relative levels of glycolytic proteins and ATP-CL may not reflect corresponding changes in enzyme activity. Many of the glycolytic enzymes are regulated by phosphorylation and by levels of ATP and other substrates. Additional studies will need to determine the degree to which changes in protein levels correspond to changes in activity. (3) All the data are obtained from regional brain homogenates. It is not possible to differentiate effects within specific cell populations within those homogenates. Hence data could reflect effects in neurons, glial cells, or a combination of the two. (4) In experiment 2, methodological limitations prevented us from directly comparing all 4 control groups (OVX-1W-Veh, OVX-6W-Veh, VCD-1W-Veh, VCD-6W-Veh) with all treatment groups for a given brain region. This was mitigated in part by comparing all vehicle-treated controls in a separate set of gels, and by comparing effects of agonists within each model and time-point relative to the corresponding vehicle control. (5) ChAT activity was used as an indicator of the integrity of cholinergic projections, but is not a direct measure of cholinergic activity. Studies that measure acetylcholine release will be needed to evaluate effects on cholinergic activity. (6) Potential changes in the levels of each ER were not assessed. It is possible that OVX and VCD treatments resulted in changes in the levels of ERs over time which, in turn, contributed to effects of ER agonists on each of the targets. Effects of OVX vs VCD treatments on ER expression and activity in different regions of the brain still need to be determined.

4.4. Summary and Conclusions

Collectively these studies provide detailed comparisons of differences in metabolic, cytoskeletal and cholinergic endpoints between cycling, OVX and VCD-treated animals as well as their response to treatment with selective estrogen receptor agonists. Primary conclusions are that OVX and VCD models of menopause have differing effects on these endpoints that are both time-dependent and brain-region specific. One notable observation are the decreases in glycolytic enzymes and ATP-CL detected in the FCX at the 1W time point in both OVX and VCD-treated rats relative to gonadally intact controls. This suggests that the FCX may be particularly vulnerable to loss of ovarian function at early time points with respect to glucose utilization, ATP production, and metabolic activity. Another notable observation is the significant decrease in these endpoints in the HPC of OVX rats (but not VCD-treated rats) after 1W treatment with ER agonists. This suggests that at early time points following OVX, ER agonists may have significant negative effects on HPC function with respect to glucose utilization, ATP production, and metabolic activity. The functional consequences of this need to be determined.

In general, existing literature supports the notion that loss of ovarian function, and particularly ovarian estrogens, leads to lower glucose utilization and higher ketone consumption in the brain [51–53]. The ability of estrogen treatments to restore or protect glycolytic function has been demonstrated in OVX rats [67], in ovariectomized non-human primates [68] and in postemenopausal women [69] where all E2-treated groups showed higher glycolytic metabolism compared to non-treated controls. Our current studies show that the effects are much more complex and can differ depending on type of menopause and the timing at which endpoints are assessed. It is likely that these differences contribute, at least in part, to the differing effects of estrogen treatments on cognitive performance that have been described in women who have experienced transitional vs. surgical menopause.

Highlights.

• Effects on metabolic and cytoskeletal proteins in cycling, OVX and VCD-treated rats

• Effects are time-dependent and brain-region specific

• ER agonists produce differing effects depending on model, brain region, and time

• Decreases in glycolytic enzymes and ATP-CL in frontal cortex following OVX or VCD

• Evidence for negative effects of ER agonists in hippocampus at early time-points