PDF
Citation
Print
Introduction
The cerebellum carries out multiple important functions including motor control, motor learning [1, 2] and recent studies have revealed its roles in aspects of cognitive processing including language, verbal memory [3, 4], executive functioning [5], working memory [6], and emotional processing [7]. The cerebellum has been noted to express aromatase [8], and estrogen receptors [9] suggesting that estradiol plays a role within the cerebellum.
It has been documented that estradiol plays a role in dendritic growth, spinogenesis, and synaptogenesis in the cerebellum [10]. Previous studies show that estradiol maintains and increases the Purkinje dendritic length, spine density and arborization [11-13]. Estradiol was demonstrated to maintain neuronal numbers and morphology in other regions of the brain such as the inferior olivary nucleus and the hippocampus [14-16]. Furthermore, exogenous estradiol administration reduces Purkinje cell death in ethanol – induced cerebellar excitotoxicity [17]. In addition, postmenopausal female receiving estrogen therapy have greater gray matter volumes in the cerebellum compared to their counterparts not receiving hormonal replacement therapy [18]. This indicates that it may play a neuroprotective role in the cerebellum.
In recent times, aromatase inhibitors (AI) are increasingly being used as first line adjuvant treatment for estrogen receptor positive breast cancer in postmenopausal female [19]. They inhibit the aromatase enzyme hence reduce the availability of estrogens [20]. Letrozole is a commonly used third generation non-steroidal AI of high potency and efficacy [21] that has been established to achieve almost complete inhibition of aromatase [22] and to penetrate the blood brain barrier [23]. Due to this, letrozole is expected to inhibit endogenous estradiol production abolishing its neuroprotective effects on the neurons of the cerebellum. Correspondingly, use of AIs has been linked to cerebellar defects including ataxia [24], poorer balance performance [25], cognitive impairment [26, 27], and diminished verbal memory [28, 29]. These adverse effects suggest cerebellar impairment, but there is paucity of data on the histological changes in the cerebellar cortex as a consequence of estradiol deprivation during AI use. Therefore, this study aimed at describing the histological changes in the cerebellar cortex that may explain the functional deficits associated with letrozole use.
Materials and Methods
This was a randomized experimental study trial that utilized the rat model. Albino rats were used as the preferred animal model due to their close physiological and anatomical similarity to humans, ease of handling [30] and are less likely to be stressed by human contact [31]. Previous studies on the effects of estradiol on the cerebellum have used rat models due to the similar histoarchitecture [32], expression of aromatase and the expression of estrogen receptors in the rat cerebellar cortex [11]. Female rats were excluded as ovarian aromatase is not inhibited by AIs and would act as a confounder in this study that focuses on the effects of brain aromatase. No significant sex differences in the expression and levels of cerebellar aromatase have been demonstrated in rats and humans [8, 11]. Rats with any demonstrable pathology or head injuries were excluded. Absence of pathology was confirmed by observation and assistance from qualified animal house attendants. The 2.5 mg Femara (letrozole) tablets that were utilized were manufactured by Novartis, Switzerland and were obtained from a local pharmacy store in Nairobi, Kenya.
Handling and treatment of animals
Twenty-seven male rats of the Rattus norvegicus species were used in the study. The study was conducted at the Department of Veterinary Anatomy and Physiology animal house and tissue harvesting and processing of specimen done at the Department of Human Anatomy, University of Nairobi Approval to conduct the parent study was sought from the Biosafety, Animal Use and Ethics Committee, Faculty of Veterinary Medicine, University of Nairobi (Reference: FVM BAUEC/2020/249) and conducted according to the committee’s guidelines. The animals were weighed and then kept in the study area for one week prior to the start of the study for acclimatization. They were exposed to a 12-hour light/dark diurnal cycle and were provided with standard rat pellets and water ad libitum.
The experimental group were administered with 0.5 mg/kg of letrozole orally by gavage every day for 50 days. This dosage was derived from the Food and Drug Administration -approved dose of 2.5 mg daily by calculation a la Nair and Jacob [33]. Given that the average duration of letrozole adjuvant therapy is 5 years [19] and the translation of rat days to human years is 10 rat days for every human year [34], an average study intervention period of 50 days was used to study the long-term effects on the cerebellum. The control group was administered with normal saline daily by gavage. On experimental days 20, 35, and 50, rats were randomly selected from the intervention and control group, euthanized, perfused with formal saline and the cerebella harvested.
Tissue harvesting and processing
The animals were weighed then euthanized by placing them in lidded containers having cotton wool soaked in 1% halothane. Euthanasia was confirmed by the absence of heartbeat and ocular reflexes. A midline longitudinal incision was made on the chest and abdomen and the skin reflected. The sternum and the ribs were removed to allow access to the heart for trans-cardiac perfusion. The animals were then adequately perfused with normal saline to flush out blood and subsequently followed by 10% formal saline to fix the body organs including the cerebellum. The rats were then decapitated and the scalp was cleaned to expose the cranium. The cranium was removed via use of sharp stainless-steel fine scissors using the optic foramen through the orbit as an access point. The laminae of the first few vertebrae were broken using the same stainless-steel fine scissors and the brain and the attached brainstem were gently extracted and placed into a container containing 10% formal saline. The cerebellum was separated from the brainstem by cutting through the cerebellar peduncles. The cerebellum was divided into two equal halves with a midsagittal cut. The harvested tissues were dipped in 10% formalin in specimen bottles for at least 24 hours before routine processing and staining for light microscopy. Fifty-four blocks were produced with one hemicerebellum being stained with hematoxylin & eosin stain to demonstrate cellular details of the cerebellar cortical layers, while the other was stained with the modified Patro’s Golgi stain to elucidate the dendrites of the Purkinje neurons. These blocks were sectioned to produce thirty serial sections from which every third section was selected to obtain ten standard sections for histomorphometric analyses.
Modified Patro’s Golgi staining technique
The Golgi staining protocol that was utilized has high specificity for Purkinje cells hence was suitable for this study [35]. This was to increase the number of Purkinje neurons stained and minimize glia staining that may produce artifacts. A block of cerebellar tissue (≈10×5 mm) was immersed into a solution containing 5% potassium dichromate, 5% chloral hydrate, 6% formaldehyde, 2% glutaraldehyde, and 6 drops of dimethylsulphoxide in the dark for 72 hours. 2 ml of 15% sucrose was also added in the chromation step as it has been shown to improve staining quality and reliability [35]. The block was rinsed with 0.75% silver nitrate solution until the brick-red precipitate stopped forming on the surface of the tissue block. The precipitate coat that would have otherwise prevented impregnation was gently brushed off. It was thereafter immersed into a 2% aqueous silver nitrate solution in the dark for another 72 hours. The tissue was subsequently rinsed with the 2% silver nitrate solution to remove any precipitate on the tissue surface and dehydrated in ascending grades of alcohols. It was cleared in toluene for 2 hours and infiltrated overnight in paraffin wax. The mounted sections were cut approximately 30 µm thick to avoid cutting through neurons and their processes. The sections were floated in a warm water bath to enhance spreading. The sections were fished from the water bath, onto a gelatinized glass slide that was prepared using gelatin and chromium potassium sulfate dodecahydrate. The sections were dewaxed in xylene before mounting using DPX Mountant (Sigma-Aldrich) and observing under a light microscope.
Histomorphometric analysis
Out of the 10 serially stained sections, every second serial section was chosen for histomorphometric analysis. Photomicrographs were taken using a ZeissTM digital photomicroscope (Carl Zeiss AG) for histomorphometric analysis. Measures of cell counts from the Purkinje layer from five random fields, preferentially avoiding areas with sulci and gyri, on the slides were done and mean values calculated. The number of Purkinje cell bodies with visibly stained nuclei in each photomicrograph were counted. The Purkinje cell layer length was determined using Fiji ImageJ software by using the Segmented Line function to draw a line connecting the centers of the Purkinje cell bodies (Fig. 1). Purkinje linear density (PLD) was calculated using the formula below [36]:
Five isolated Purkinje neurons per section were identified and randomly selected based on previously established criteria of effective staining that entails homogenous and complete impregnation throughout the extent of the neuronal processes, neurons not obscured by neighboring structures and the terminal dendrites possessing natural terminations [37]. One dendritic segment was chosen per Purkinje cell based on an established protocol ensuring each segment was approximately 10 µm long, unobscured by nearby branches or structures and that had visible dendritic spines [11, 38].
Photomicrographs at ×1,000 magnification using immersion oil were taken from the Golgi-stained sections. The dendritic spines whose heads and stems were in focus were counted. This was determined using Fiji ImageJ software (National Institutes of Health) (Fig. 2). Dendritic segment length was measured by drawing a freehand line along the dendritic segment. The following formula was used [38]:
Statistical analysis and presentation
Histomorphometric data on the PLD and dendritic spine densities (DSD) was entered into IBM SPSS Statistics software (version 25.0; IBM Co.) for statistical analysis. Normality of the data was assessed using the Shapiro–Wilk test and visual inspection of the histograms, box plots and normal Q-Q plots generated from the data. The histograms displayed skewed distributions and, thus, non-parametric tests were employed. Kruskal Wallis H-tests were employed to check for statistically significant differences over time in both the control and experimental group over the study period. A Dunn Bonferroni post-hoc test was carried out in case statistically significant differences were noted in the Kruskal Wallis test. Mann Whitney U-tests were carried out to assess for significant differences in the PLD and DSD between the control and experimental group on each of the perfusion days. A P-value of ≤0.05 was considered significant. Data were presented in tables, photomicrographs and line graphs were plotted to show the trend observed in the measured variables.
Results
All the rats used in the study survived until their respective perfusion timings. The animals gained weight appropriately over the study period. At the time of euthanasia, no discernible gross anatomical differences in cerebella of the intervention groups were noted when compared to the control and the baseline groups.
PLD of the control group rose slightly up to day 20 after which it remained relatively constant over the remainder of the study period (Fig. 3). The Purkinje cells of the control group appeared to be uniform in size and some had prominent nucleoli (Fig. 4A, C, E). The photomicrographs revealed a reduced number of Purkinje cells in the intervention group with some pyknotic Purkinje cells also being observed (Fig. 4B, D, F). There were no statistically significant differences between the groups (Table 1). On the other hand, a decrease in the PLD was noted on administration of letrozole over the 50-day study period. PLD dropped progressively with the lowest cell density being observed on day 50 (P=0.165).
A marked decrease in DSD of the Purkinje neurons was observed in the experimental group (Fig. 5). The terminal dendrites of the Purkinje neurons in the control groups have a relatively high dendritic spine density (Fig. 6A, C, E). This was evidenced by a gradual sparsity of dendritic spines on the terminal dendrite segments with continued letrozole administration from day 20 onwards (Fig. 6B, D, F). In addition, the dendritic segments in the experimental group at day 50 appeared to be more slender (Fig. 6F) as compared to the corresponding control group (Fig. 6E). A slight decrease in DSD of the Purkinje neurons was also noted in the control group (Fig. 5) but was not statistically significant (P=0.09).
The intervention group displayed statistically significant temporal differences in the DSD during the study period (H=12.462; P=0.01). This was noted since the experimental group at day 50 had a significantly lower DSD when compared to the DSD at day 0 (P=0.01). Upon comparing the DSD between groups, statistically significant differences were revealed between the control and the experimental groups with the latter group having consistently lower values (all P-values=0.036) (Table 2).
Discussion
This study has shown structural changes in the Purkinje neurons consistent with the previously noted deficits. Considering that use of letrozole is on the rise [19], structural effects of its administration on the cerebellar cortex should not be taken lightly.
The lower PLD noted on administration of letrozole was similar to what was described previously by Hill et al. [39] who used aromatase-knockout mice to mimic a state of estrogen deficiency. These mice displayed a reduction in number of the cortical neurons even in the absence of pathological conditions and external toxic insults. Contrary to this, some studies have concluded that administration of an AI alone does not lead to significant effects on the neuronal number in both the adult inferior olivary nucleus and the neonatal cerebellum [12, 15]. Estradiol has been noted to have anti-inflammatory activity in the brain that is mediated via estrogen receptor-beta (ERβ) that is present on Purkinje cells [40]. It has also been reported to protect Purkinje cells from glutamate excitotoxicity by promoting Gamma-aminobutyric acid production [41]. A review by Amantea et al. [42] additionally indicated that it is responsible for vasodilation and increased cerebral blood flow, upregulation of anti-apoptotic factors such as Bcl-2 and downregulation of pro-apoptotic factors. The reduction in PLD may be explained due to letrozole effectively lowering estradiol levels and the ensuing nullification of its neuroprotective effects. Literature has also shown that Purkinje neuronal survival may be linked to antioxidant properties and the phenolic structure of estradiol that scavenges reactive oxygen species [43].
This reduction in Purkinje cell number may lead to functional deficits as this reduction is commonly seen in various cerebellar pathologies such as cerebellar ataxia [44]. Moreover, Yang et al. [45] proposed that decreased Purkinje cell density is a common histological finding in ataxic patients. Thus, the reduction in PLD possibly explains the ataxia seen in AI use.
The marked reduction in DSD of the Purkinje cells with exposure to letrozole seen in our study closely reflects those seen in estradiol deprivation [15]. In the current study, lowering of estradiol levels in the cerebellum may be attributed to the inhibition of cerebellar aromatase by letrozole. These findings are in concordance with those of Sakamoto et al. [11] and Sasahara et al. [12] who reported reduced DSD in developing murine Purkinje cells secondary to estradiol deprivation. The role of estradiol in dendritic spinogenesis has also been elaborated previously in adult rats treated with exogenous estradiol. These rats exhibited increased DSD of neurons in other regions of the central nervous system such as the ventromedial nucleus of the hypothalamus, the basal forebrain, and the CA1 subfield of the hippocampus [46-48].
Estradiol acts via the nuclear ERβ receptors to upregulate levels of brain derived neurotrophic factor (BDNF) [49]. BDNF then increases the levels of tropomyosin receptor kinase B (TrkB) and this pathway leads to increased spinogenesis in the Purkinje cells [50, 51] and hippocampal neurons [52]. Luine and Frankfurt [51] also hypothesized that rapid changes in the DSD are seen due to the direct actions of estradiol on the synthesis of F-actin, an important component of the cytoskeleton in the dendritic spine [53]. In addition, it has been noted that the long-term potentiation leads to phosphorylation of cofilin is associated with the stabilization of the actin cytoskeleton in the dendritic spines. AI have been noted to impairment of long-term potentiation with subsequent dephosphorylation of cofilin and the downregulation of the dendritic spines on neurons [54].
Decline in the DSD of Purkinje cells has been outlined in rat models of ataxia and essential tremors [38, 55]. This is not surprising as a study by Dickstein et al. [56] has shown that reduced DSD has a major impact on neuronal function since the spines act as a major locus for excitatory neuronal signals. Therefore, the reduction in the DSD noted in the current study may provide a structural basis for the functional deficits noted in chronic AI users due to a reduction in these synapse loci.
The limitations of the current study included the erratic and unpredictable nature of Golgi method of staining and the fact that 3D reconstruction of the Purkinje cells was not carried out leading to underestimation of the dendritic spines. These were delimited by following a specific Golgi staining protocol for Purkinje cells to increase the probability of staining more Purkinje cells and the underestimation was presumed to be carried through all the dendritic measurements. In addition, the study focused on temporal changes in DSD upon exposure to the AI rather than absolute numbers of dendritic spines.
Letrozole administration was associated with the reduction of the PLD and the DSD of the Purkinje cells. These structural differences in the cerebellar cortex might provide the basis for the functional deficits that may be observed in chronic AI use. Future studies could be carried out to obtain accurate neuronal densities and DSD using stereological fractionators and 3D reconstruction of neurons respectively as these would provide a more accurate quantification of the Purkinje cells and immunohistochemical studies of the levels of markers of apoptosis, BDNF and TrkB in the adult cerebellum that may indicate the neuromolecular mechanisms underlying chronic administration of letrozole.
XML Download