Neuroprotective Effects of a Combined Therapy With Memantine, Donepezil and Vitamin D in Ovariectomized Female Mice Subjected to Dementia Model

The postmenopausal period is characterized by a decrease in the hormonal supply which is associated with Alzheimer's Disease (AD). Vitamin D is neuroprotective and can be used in combination with pre-existing medications to improve its effects. The objective was to evaluate the effect of vitamin D associated with memantine and donepezil in female mice submitted to ovariectomy (OVX) for 5 months and subjected to an AD-induced dementia model. Animals were divided into 5 groups who received 17 days of treatment and were subjected to behavioral tests. The animals underwent euthanasia at 18 th day. OVX groups exhibit reduced levels of E2 and triple treatment group had high levels of vitamin D. The induction of dementia with OVX induced short- and long-term spatial and habituation memories damage. Also, induced reduction of BDNF and IL-4 levels in hippocampus, and increasing levels of TNFα in hippocampus and of IL-1β in hippocampus and frontal cortex of animals, as well as a signicant increase on GFAP immunoreactivity. Triple-association treatment reversed the effects of long-term spatial and habituation memories damage, as well as reversed changes in TNFα, IL-1β, IL-4 and GFAP immunoreactivity levels in hippocampus of treated animals. Therapeutic association has benecial effects on memory and inammation parameters in female mice subjected to OVX and the AD animal model of dementia.

menstruation (Takahashi et al, 2015;Davey, 2017;Siebert et al, 2017). Depletion of female sex steroid hormones, especially estradiol (E2) (Hansberg-Pastor et al, 2015), seems to increase vulnerability to diseases in tissues sensitive to these hormones, such as bones, cardiovascular system, and brain (Lobo et al, 2014). Therefore, it is necessary to search for effective alternatives to treat deleterious cognitive effects of post-menopausal E2 deprivation, which can lead to the onset of AD (Annweiler, 2011).
Currently, only four drugs are approved for the treatment of AD -memantine, donepezil, galantamine, and rivastigmine (Graham, Bonito-Oliva e Sakmar, 2017; Kim et al, 2017); however, the effectiveness of these drugs is limited. In addition, these drugs do not present clear evidence regarding both long-term safety and e cacy of treatment, since none of the drugs modify primary pathological processes of disease or prevent its progression, despite providing a symptomatic relief ( To minimize symptoms of AD, the development of new therapeutic strategies is essential (Annweiler et al, 2011), and one current alternative is to search for substances with neuroprotective properties, preferably, already approved by FDA (Food and Drug Administration) and with known toxicity, to facilitate access to AD patients (Landel et al, 2016). Currently, links between vitamin D and AD have been highlighted. In fact, de ciency of this vitamin is prevalent both in AD patients and individuals who have dementia in general acts on the central nervous system (CNS) through vitamin D-speci c nuclear receptors (VDR) (Landel et al, 2016), which are present in neurons and glial cells (Annweiler, 2016), especially in cognition-related regions such as the hippocampus, hypothalamus, cortex, and subcortum (Annweiler et al, 2014). Thus, modulation in these receptors by vitamin D exposure could regulate the expression of structural proteins, neurotrophic factors (Annweiler, 2016), and neurotransmitters de cient in neurodegenerative diseases (Mpandzou et al., 2016), attenuates in ammation related to natural senescence and Aβ peptide toxicity (Annweiler, 2016).
Based on this, new therapeutic approaches have been studied aiming at the use of this vitamin alone (Briones e Darwish, 2012) or in association with other substances (Annweiler et al, 2011), contributing to multiple drug treatment and a possible enhancement of favorable effects in the management of AD. In this study, we submitted female mice ovariectomized (OVX) and injected with Aβ peptide to memantine, donepezil, and vitamin D combination treatment.

OVX SURGERY FOR HORMONE DEFICIENCY INDUCTION
Female mice 3-month-old were anesthetized with ketamine/xylazine (ketamine: 60 mg/kg and xylazine: 4 mg/kg intramuscular, i.m). After disinfection with chlorhexidine, a cutaneous incision in the dorsal midline was made caudal to the back of the ribs. Using blunt dissection to the subcutaneous tunnel, lateral to skin incision, the posterior abdominal wall muscles were separated to expose the abdominal cavity. The ovary was located on a fat pad 1 to 2 cm below muscles. The periovarial fat was clamped with forceps to lift and externalize the ovary. The fallopian tube was crushed, and the ovary was removed by cauterization. The skin incision was closed using suture number 6. The control (Sham) animals only underwent surgery without cauterization of ovaries. After surgery, to ensure the prolonged absence of ovarian hormones, animals remained in the vivarium for 5 months (Rigalli and Loreto, 2009;Fatemi et al, 2018 PO 4 , 26 mM NaHCO 3 , 10 mM glucose and 4 mM sucrose). Animals were anesthetized with iso urane (1.5%) using an inhalation anesthesia apparatus. Aβ oligomers were injected through a 3 mm long 28gauge stainless steel needle coupled to a Hamilton syringe. The needle was inserted unilaterally through the skin and skull directly into the left lateral ventricle via i.c.v. Aβ oligomers were injected in a volume of 4µl (400 pmol/site) for 10 seconds, followed by a 10-second delay to allow full diffusion of the solution. After administration, animals were placed in a community cage with ad libitum access to food and water and then recovered from anesthesia.

BEHAVIORAL TESTS
On the 14th day of treatment, animals were submitted to the behavioral test of the radial maze, which persisted for 5 days to evaluate spatial memory. Another independent group of animals, on the 11th and 12th days of treatment, were submitted to open eld test for habituation evaluation, and on the 13th day, short-term spatial memory was evaluated with the Y maze test. After behavioral tests, the animals were induced to euthanasia.

RADIAL MAZE TEST
The radial maze test was performed to evaluate the spatial memory. On the 14th day of treatment, the task began. The radial maze apparatus had 8-arms, which were numbered from 1 to 8 (48 × 12 cm) and extended radially from a central area (32 cm diameter). The apparatus was placed 50 cm above the oor, and geometric shapes were positioned in the straight arms where the food was placed (visual cues). On the rst day, each animal was placed in the apparatus for a total of 10 min, allowed only to explore, and then returned to its cage. On the 2nd day (or the true rst day of test), the animals were placed in the apparatus, where food (cereal) had already been deposited in four of the eight arms. The food bearing arms had visual cues at the end of each arm. Over a period of 10 min, the entry into each arm (total errors to nd food) and the time each animal took to nd the four pieces of cereal were recorded (latency to nd food). The same test was held over four consecutive days, with one trial occuring per day. At the end of the test, the animals were euthanized (Foyet et

Y-MAZE TEST
The Y-maze test was performed to evaluate short-term spatial memory. It was performed at the 13th day of treatment, the apparatus consists in a 3 armed labyrinth (50 x 10 x 20) at a 120° angle, being similar to the shape of the letter "Y". The rst arm has a visual clue to ease the spatial location of the animals. This behavioral test is made at a dark room, only lighted by a red bulb, and it occurs in two sessions, divided by a 2-hour break. At the rst session, the animal starts at the rst arm (start) and has free access to explore the second arm (other) during ve minutes, while the third arm (novel) is closed (Dellu et al., 1992). The animal returns to his cage at the end. 2 hours later, the animal returns to the apparatus with all 3 arms open to explore freely during ve minutes. His permanence at all arms is timed separately, and the short-term special memory is evaluated through his permanence at the "novel" arm (Dellu et al., 1992).

HABITUATION TO OPEN FIELD TEST
The habituation to open eld test was performed to evaluate the habituation memory, at the 11th and 12th days of treatment. The apparatus is made with 4 walls (60 x 40 x 50 cm), one of them is made of glass, its oor is divided in 9 equal squares divided by lines drawn. It's done in 2 separate sessions divided by a 24-hour break. At the rst day, the animal is put inside the posterior left square of the open eld apparatus, once inside, the times the animal crosses lines (crossings) and stands on his hind legs (rearings) are counted separately during 5 minutes. (Vianna, 2000). After this period, the animal is removed from the apparatus and returns to it's cage. 24 hours later, the protocol is repeated. His locomotor activity is evaluated through the number of crossings and his habituation memory is evaluated through the number of rearings (Vianna, 2000).

BIOCHEMICAL ANALYZES
Animals were subjected to euthanasia on the 18th day of the experiment by cervical dislocation. The frontal cortex and hippocampus were dissected and homogenized in phosphate buffer to analyze cytokine (TNFα, IL-1β, and IL-4) levels by enzyme immunoassay kits (ELISA). Hippocampus was also used for the analysis of BDNF levels by ELISA. Blood was collected to analyze 25-hydroxyvitamin D and estrogen (E2) levels by chemiluminescent immunoassay in serum. The samples were stored under refrigeration at -80°C until use.

IMMUNOFLUORESCENCE MICROSCOPY
After the end of the experimental protocol, mice were anesthetized (75 and 10 mg/kg of xylazine mixture, respectively, intraperitoneally), perfused through the left cardiac ventricle with 0.9% saline solution, and by 4% paraformaldehyde in 0.1M phosphate-buffered saline (PBS), pH 7.4. The brains were collected and post-xed in the same xative solution for 24 h at room temperature (RT), and cryoprotected by immersion in a 30% sucrose solution in PBS at 4ºC. Serial coronal sections (40 µm) of hippocampi were obtained with a Vibratome. The free-oating sections were rst blocked using 5% horse serum (HS) diluted in PBS containing 2% Triton X-100 (PBS-Tx) for 2 h at RT. Then, the sections were incubated 24 hours at 4ºC with mouse anti-GFAP (glial brillary acidic protein) (Sigma, 1:400) in 1% HS diluted in 0.5% PBS-Tx. After three washes in PBS, tissue sections were incubated with anti-mouse Alexa 488 (Invitrogen, 1:400) in 1% HS diluted in 0.5% PBS-Tx for 2 h at RT. After incubation in secondary antibody, the sections were washed three times in PBS. Thereafter, the sections were washed several times in PBS, mounted on slides with CC/Mount (Sigma), and covered with coverslips. Finally, images from mouse hippocampi were obtained with a Microscopy EVOS® FL Auto Imaging System (AMAFD1000 -Thermo Fisher Scienti c; MA, USA) (de Oliveira et al., 2014).

STATISTICAL ANALYSIS
The results were analyzed using STATISTICA version 8.0 (StatSoft, Inc., USA). Shapiro-Wilk normality test was performed to con rm that the data had a normal distribution. The results of the radial maze test were con rmed by analysis of variance of one-way repeated measures analysis of variance (ANOVA), followed by Newman-Keuls post-hoc test when appropriate. Open eld, and Y-maze data were analyzed by Student's T-test. Data on levels of estrogen, vitamin D, BDNF, and cytokines were analyzed by one-way ANOVA, followed by the Newman-Keuls post-hoc test when appropriate. Data were expressed as mean ± standard error of the mean (SEM). The values of p are considered signi cant when <0.05.
The radial labyrinth data show a signi cant effect [F(12.126)=2.09, p<0,05] on latency to nd the rewards.
The control group (Sham + ACSF + water) learned the task on the third and fourth test days. The Sham + Aβ1-42 + water group learned only on the last day, indicating long-term spatial memory damage. Animals from OVX + ACSF + water group learned the task on the third and fourth days, and the control group, indicating that OVX did not damage long-lived spatial memory. However, the OVX + Aβ1-42 + water group showed spatial memory damage on all test days, and this damage was even more signi cant when comparing Sham + Aβ1-42 + water group to OVX + Aβ1-42 + water group. This result shows that OVX associated with peptide causes more severe impairment in memory. It was possible to observe that the treatment with the triple association of vitamin D + memantine + donepezil effectively reversed spatial memory damage caused by peptide associated with OVX since the animals in this group learned on the third day of training as control. However, data on the errors that animals committed until they found the reward did not reveal any signi cant difference [F(12.126)=0.53, p=0.05] (Figure 3).
The Y-maze data showed a signi cant difference only in the Sham + ACSF + water group [t(9): -2.78, p<0.05], indicating that only the control learned the task, showing no short-term spatial memory impairment. While other groups, including the treatment group, did not show signi cant differences between the other and the new arm, indicating that animals did not learn the task, i.e., short-term spatial memory damage and treatment could not reverse it (Figure 4). .01] with increased levels in Sham + Aβ1-42 + water, OVX + ACSF + water and OVX + Aβ1-42 + water groups (Figure 8). IL-4 levels in frontal cortex were signi cantly reduced [F(4.20)=6.67, p<0.01] in Sham + Aβ1-42 + water and OVX + ACSF + water groups. In hippocampus, there was a signi cant increase [F(4.17)=7.57, p<0.01] in cytokine levels in OVX + ACSF + water group, and reduction of levels in OVX + Aβ1-42 + water group (Figure 9). The effects on all cytokines were reversed by treatment with vitamin D + memantine + donepezil in the hippocampus resembling the control group results at these dosages.
Finally, to investigate the astrocytes activation in hippocampal sections of experimental groups, we analyzed the immunoreactivity of GFAP. We observed a signi cant increase on GFAP immunoreactivity in the hippocampi of Sham + Aβ1- 42 Figure 10).

Discussion
The hormonal senescence of menopause is a natural part of the aging process. Sometimes this decrease in E2 levels results in physiological changes in the brain (Koebele and Bimonte-Nelson, 2016), especially in the hippocampal region, resulting in loss of synaptic spines and neurodegeneration (Nebel et al, 2018), as well as the conversion of a metabolically active and healthy state to a state of low metabolic and oxidative activity, with an imbalance in amyloid clearance capacity (Zhao, Woody and Chhibber, 2015). This suggests that the once bene cial effects of estrogen are due to a greater predisposition to dementia, such as AD (Bove et al., 2014, Pike, 2016. These ndings corroborate with the present study where animals submitted to OVX and, consequently, with lower levels of E2 presented impairment in short-term memory, an increase in proin ammatory cytokines, a reduction of neurotrophin and antiin ammatory cytokine. In addition, when OVX was associated with administration of the Aβ1-42 peptide, animals had a noticeable increase in IL-1β cytokine (proin ammatory) in the cortex and potentiation of long-term memory damage.
In this study, the evaluation of animals' behavioral parameters and learning capacity demonstrated that Aβ1-42 peptide administration could induce long-term and short-term spatial memory damage and memory of habituation in the open eld. Considering that cognitive impairment and spatial memory decline are clinical symptoms of AD and that learning and spatial memory damages have been reported in previous studies with this model , Garcez et al, 2018, the validity of the dementia model in inducing pathologic-like damage is con rmed.
In general, associative memory tasks may be bene cial in the perception of impairment due to AD, since they are dependent on the integrity of the hippocampus, one of the central regions affected in this pathology (Sperling et al, 2010), and which plays a critical role in spatial orientation and navigation (Lee et al, 2014). Since short-term memory is related more to areas of the prefrontal cortex, whereas long-term memory is more closely related to the hippocampal region (Lee et al, 2014;Djiogue et al, 2018), the results related to memory and behavioral tests can be justi ed, at least partially, by the fact that vitamin D has a more remarkable performance in the hippocampus (Brown et al., 2003; Dursun, Gezen-Ak and Yilmazer, 2011; Annweiler and Beauchet, 2012), since the associated treatment of memantine + donepezil + vitamin D was effective in the radial maze (long-term memory), but had no bene cial effects on the Ymaze (short-term memory).
The hippocampus is a critical brain region related to memory and learning functions (Şahin et al, 2019), as well as contextual and spatial awareness (Barrientos et al, 2016;Deuker et al, 2016), and is responsible for forming new associations between elements of previously unrelated information, contributing to episodic coding (Sperling et al, 2010, Knierim, 2015, Barrientos et al, 2016, Deuker et al, 2016. The hippocampal complex is one of the rst regions affected during AD pathology and its connectivity is altered and is an essential parameter in preclinical AD (Harrison et al, 2015). As for menopause, E2 has a remarkable performance in the hippocampus since it exerts its cerebral effects through ERα and ERβ receptors located in the dendritic spines, dendrites, axons, and nucleus of hippocampal neurons (Djiogue et al, 2018). Its relation with vitamin D stands out for the bene cial effects that this substance exerts in this region, such as: reduction of oxidative stress in hippocampal neurons (Dursun, Gezen-Ak and Yilmazer, 2011; Annweiler and Beauchet, 2012), reconstruction of functional neural networks, increased synthesis of neurotrophic agents and acceleration of neuron growth (Brown et al, 2003), attenuation of Aβ1-42 peptide accumulation and reversal of local in ammatory conditions (Nissou et al 2014) and adjustment in the expression of in ammatory cytokines (Landel et al, 2016). Therefore, an impact on the in ammatory process related to the use of vitamin D should not be considered unexpected (Landel et al, 2016).
In this study, it was observed that concerning the expression of in ammatory cytokines, there was an increase in the levels of TNFα in the hippocampus and IL-1β in the hippocampus and in the frontal cortex of animals submitted to OVX and/or administration of Aβ1-42 peptide, which reinforces the thesis that both, E2 deprivation and AD, present a predisposing pro le to an in ammatory environment. Previous studies have demonstrated the relationship between TNFα levels and AD by using antiin ammatory strategies in transgenic mice, resulting in a reduction of pathology related to Aβ1-42 peptide and tau protein and restoring memory de cits after regulating TNFα synthesis (Gabbita et al, 2015) or the performance of its receptor (Detrait et al, 2014). This correlation can be reinforced by observing data from this study about memory damage and proin ammatory cytokine dosage, where the groups of animals that showed an increase of TNFα and IL-1β also showed short and long memory damage duration and habituation in the open eld. In addition, the ampli ed response of cytokines in the hippocampus can be justi ed by the high expression of IL-1β receptors in neurons and glia in the granular cells of the gyrus and the pyramidal cell layer of the hippocampus and by the particularly dense microglial population in this region (Barrientos et al, 2016).
However, the associated treatment was able to reverse the increase of cytokines TNFα and IL-1β signi cantly in animals submitted to OVX associated with AD. Thus, reinforcing the antiin ammatory competence of vitamin D, already reported in previous studies where the effects of this vitamin were able to reduce in ammatory parameters in the hypothalamus (Farhangi et al, 2017), and to reverse the levels of in ammatory cytokines due to the accumulation of Aβ1-42 peptide (Mizwicki et al, 2013).
Cytokines are considered essential mediators in the neuroin ammation process and, although classi ed as pro and antiin ammatory, their balance is that it stands out as fundamental in the effective response to injuries in the central nervous system (Uchoa, Moser and Pike, 2016). Thus, analysis of IL-4 levels, an antiin ammatory cytokine, was also performed in this study, and it was observed that IL-4 levels were decreased in groups submitted to AD or OVX (isolated) in the frontal cortex and groups exposed to OVX and administration of the Aβ1-42 peptide concomitantly in the hippocampus. Previous studies have shown similar results by reducing IL-4 levels after estrogen deprivation (Yasui et al, 2007;Ma et al, 2007) and AD (Tai et al, 2015;Budni et al, 2017;Garcez et al, 2017). However, the triple treatment of association with vitamin D restored levels of this cytokine in the hippocampus in animals submitted to OVX plus AD, corroborating with previously highlighted antiin ammatory activity (Annweiler, 2016). However, in this study, it cannot be said that this effect was exclusively due to vitamin D administration since no group with exclusive administration of vitamin D or memantine and donepezil were performed.
Furthermore, in this study, the levels of BDNF in the hippocampus were evaluated, and it was demonstrated that both OVX and Aβ1-42 administrations were effective in reducing the levels of this neurotrophin. These results support the theory that decreasing E2 levels, as well as AD, lead to a reduction It is worth highlighting that the associated treatment of vitamin D, memantine, and donepezil demonstrated positive results since it was able to reverse damages of long-term spatial memory and memory of habituation in the open eld and reestablishing the levels of antiin ammatory and proin ammatory cytokines. Indeed, neuroin ammation is an important feature of AD, closely related to neuronal damage and cognitive impairments development in the disease. In this process, activated astrocytes release mediators that contribute and amplify the immune system recruitment and activation (Guzman-Martinez et al., 2019). An essential point of this study is that we observed that combined therapy attenuate astrogliosis in hippocampi of female mice exposed to both Aβ and OVX.
This fact is considerably important since there is currently no cure for AD, and current classical treatments, in addition to having transient effects, are limited in relation to cognition enhancements (Annweiler, 2011;Mielke et al, 2012). Therefore, the association of classical drugs with a previously known and approved substance, such as vitamin D, would facilitate the insertion of a combined treatment as part of a "multiple drug" regime aimed at different targets, with the capacity to improve already known damages caused by DA.
In addition, it has previously been shown that levels of the VDR receptor in the brain in the presence of the Aβ1-42 peptide have been decreased, even in patients with su cient levels of vitamin D, indicating that supplementation may be necessary even in patients without hypovitaminosis seeking to stimulate and regulate the mechanisms mediated by vitamin D in the brain, impaired by the pathophysiology of AD (Dursun, Gezen-Ak and Yilmazer, 2011). Thus, the relevance of associating vitamin D with classic drugs present in the therapy of these patients is exposed. Effects of treatment with vitamin D (420 U/kg) + memantine (5 mg/kg) + donezepil (1 mg/kg) on serum E2 levels of female mice exposed to Aβ and OVX. Data are expressed as mean ± S.E.M. (6 animals per group). *p <0.05 and **p <0.01 compared to Sham + ACSF + Water group, ## p <0.01 compared to Sham + Aβ1-42 + water (p <0.001), OVX + ACSF + water (p <0.001) and OVX + Aβ1-42 + water (p <0.01) groups (One-way ANOVA followed by Duncan's post-hoc test).

Figure 10
Effects of treatment with vitamin D (420 U/kg) + memantine (5 mg/kg) + donezepil (1 mg/kg) on astrogliosis in hippocampi of female mice exposed to Aβ and OVX. The images of immuno uorescence staining were obtained with a Microscopy EVOS ®FL Auto Imaging System. (A) Representative images of immuno uorescence assay for GFAP (astrocytes marker) in hippocampi slices of female mice exposed to Aβ and OVX and treated with vitamin D + memantine + donezepil. Scale bars, 400µm. (B) Quanti cation in arbitrary units of uorescence for GFAP in total hippocampus in the different experimental groups.