Letrozole Potentiates Mitochondrial and Dendritic Spine Impairments Induced by β Amyloid

Reduced estrogens, either through aging or postsurgery breast cancer treatment with the oral nonsteroidal aromatase inhibitor letrozole, are linked with declined cognitive abilities. However, a direct link between letrozole and neuronal deficits induced by pathogenic insults associated with aging such as beta amyloid (Aβ 1–42) has not been established. The objective of this study was to determine if letrozole aggravates synaptic deficits concurrent with Aβ 1–42 insult. We examined the effects of letrozole and oligomeric Aβ 1–42 treatment in dissociated and organotypic hippocampal slice cultures. Changes in glial cell morphology, neuronal mitochondria, and synaptic structures upon letrozole treatment were monitored by confocal microscopy, as they were shown to be affected by Aβ 1–42 oligomers. Oligomeric Aβ 1–42 or letrozole alone caused decreases in mitochondrial volume, dendritic spine density, synaptophysin (synaptic marker), and the postsynaptic protein, synaptopodin. Here, we demonstrated that mitochondrial and synaptic structural deficits were exacerbated when letrozole therapy was combined with Aβ 1–42 treatment. Our novel findings suggest that letrozole may increase neuronal susceptibility to pathological insults, such as oligomeric Aβ 1–42 in Alzheimer's disease (AD). These changes in dendritic spine number, synaptic protein expression, and mitochondrial morphology may, in part, explain the increased prevalence of cognitive decline associated with aromatase inhibitor use.


Introduction
Currently, aromatase inhibitors, leading to reduction of estradiol synthesis from testosterone, have been favoured in the treatment of breast cancer of postmenopausal women [1]. Letrozole is one such oral nonsteroidal aromatase inhibitor which prevents the aromatase from producing estrogens by competitive, reversible binding to the heme of its cytochrome P450 unit [2,3]. Patients receiving letrozole have shown deficits in learning and memory [4]. However, it is still unclear if these functional impairments occur because of neurosteroid deficits and if they become exacerbated in the presence of an additional insult associated with aging, such as an excess of soluble, conformationally altered  . Such forms of  can interact with synapses and cause pronounced degenerative changes, an observation made in brain tissue from postmortem Alzheimer's disease (AD) patients and animal models [5]. It has been well documented in AD that there are changes in neuronal and glial morphologies such as dendritic spine atrophy and increased glial growth [5,6]. In contrast to these negative effects caused by  , it is well established that the neurosteroid estradiol can increase the number of dendritic spines and proteins in humans, as well as promoting synaptogenesis and cellular models of learning such as long-term potentiation (LTP) in hippocampal cultures [7,8]. Despite these well-established positive effects of neuroestrogens in the CNS, 17--estradiol hormone therapy in postmenopausal women did not exhibit clear cognitive benefits [9][10][11][12][13]. These clinical studies did not provide conclusive evidence whether exogenous estrogen exposure in early menopause affects AD or cognitive aging [14] and to what extent (if any) peripheral estrogen in blood could play a role in patients or animal models of AD [15][16][17].
To investigate whether aromatase suppression in the CNS affects synaptic responses to pathological insults, we used confocal microscopy to examine the effect of letrozole 2 Journal of Aging Research and 1-42 on two model systems, dissociated hippocampal neurons, and organotypic slice cultures. The slice cultures have a preserved neural circuitry, exhibit similar synaptic transmission as in vivo, and permit long-term studies allowing for the assessment of synapse morphology and function following pharmacological aromatase inhibition. The objective of the present study was to investigate the response of hippocampal neurons and glia to soluble  species upon inhibition of CNS aromatase by letrozole. We have found significant decrease in neuronal mitochondrial volume, dendritic spine density, synapse number, and the synaptic protein, synaptopodin, in cultures that were either treated with letrozole or  alone. With letrozole and 1-42 co-treatment, the synaptic and mitochondrial deficits were further enhanced in the hippocampal neurons. These findings could explain why women on letrozole have a higher incidence of cognitive decline and may be more vulnerable to the detrimental effect of 1-42 .

Preparation of -Amyloid ( 1-42 ).
1-42 (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in 5% NH4OH in Tris Buffered Saline and vortexed until fully dissolved (several minutes) at a concentration of 1 mM. Aliquots were stored at −80 ∘ C. Upon reconstitution, aliquots are dissolved directly in media prior to treatment at a final concentration of 1 M [18].

Preparation of Dissociated Neural Cell Cultures.
All procedures for animal handling were carried out according to the guidelines of the Canadian Council on Animal Care (CCAC) and approved by the Animal Resource Committee of the School of Medicine, McGill University. Hippocampal tissues from mouse pups of 3-5 days old were dissected, trypsinized, dissociated, and incubated as described previously [19]. Briefly, hippocampi of P3-5mice were isolated, mechanically and enzymatically (0.25% trypsin; Gibco, Life Technologies Inc, Burlington, ON, Canada) dissociated, and plated onto coated (poly-l-ornithine, 100 g/mL (Sigma-Aldrich) and laminin, 0.587 g/mL (Invitrogen, Life Technologies Inc, Burlington, ON, Canada) glass coverslips and maintained at 37 ∘ C and 5% CO 2 in a 24-well plate (Corning; Lowell, MA, USA). Culture medium consisted of neurobasal A (Invitrogen) supplemented with 2% (v/v) B-27 supplement (Invitrogen), 1% (v/v) PSN (Invitrogen), and 1 mM Lglutamine (Sigma-Aldrich).

Pharmacological Treatments of Dissociated Neural Cell
Cultures. Dissociated cells were treated on day in vitro (DIV) 8 with 1 M letrozole (Novartis, Dorval, QC, Canada) and/or 1 M 1-42 (American Peptide Company, Sunnyvale, CA, USA) [20] according to the respective design outlined in the results section and corresponding figure legends.

Preparation of Slice Cultures.
Hippocampal slices (400 m) from P6-8 transgenic mice expressing membranetargeted MARCKS-eGFP under the Thy-1 promoter in a subpopulation of CA1 cells [21] were prepared as previously described using the roller-tube method [22]. Slice cultures were maintained in the incubator with a roller-drum for three weeks before experimentation to allow for spine maturation [23].

Pharmacological Treatments of Slice
Cultures. Slice cultures were maintained for at least 3 weeks in vitro to allow for maturation prior to treatment. In addition, all treatments were done on sister cultures (slice cultures that were prepared from the same mice litter and handled the same way at the same time). To prevent the effects of estradiol known to be present in serum, which is normally included in the culture media, we placed our preparations in serum free media prior to testing the effects of  or letrozole [24]. Following preincubation, cultures were then treated with media containing either, vehicle,  (1 M), letrozole (1 M), or 1-42 + letrozole for 24 or 72 hours. No media change was performed during the time course.

Mitochondrial Labeling and Morphological Status.
Morphological status of mitochondria was determined by labelling live cells (treated or untreated) with 500 nM Mitotracker Deep Red FM (Molecular Probes) for 3 minutes and imaged with a confocal microscope at 63× (Zeiss LSM510, Carl Zeiss MicroImaging GmbH, Jena, Germany); 633 nm HeNe laser at 20% max excitation intensity. Z-stacks (10 slices, 0.33 m steps) were acquired of dendritic segments from 4-6 different neurons per coverslip, and 2-3 coverslips per condition were analyzed. 3D deconvolution of the raw stack was performed using Huygen's Deconvolution Software (Scientific Volume Imaging, Hilversum, The Netherlands) with maximum likelihood extrapolation, and images were further analyzed for volumetric measurements using Imaris (Bitplane, Zurich, Switzerland).

Image Acquisition and Quantification of Dendritic
Spines. Following treatments, slice cultures were transferred to a temperature-controlled chamber (30 ∘ C) mounted on an upright confocal microscope equipped with W Plan-APOCHROMAT 63×/1.0 objective (Zeiss LSM710, Carl Zeiss MicroImaging GmbH) and continuously perfused with Tyrode solution containing (in mM): NaCl, 137; KCl, 2.7; CaCl 2 , 2.5; MgCl 2 , 2; NaHCO 3 , 11.6; NaH 2 PO 4 , 0.4; and glucose, 5.6 (pH 7.4). Secondary and tertiary dendritic branches from either apical or basal dendrites of CA1 pyramidal neurons were imaged using 488 nm argon-ion laser line. Spine classification was based on previously established criteria and separated into three main morphological subtypes, stubby, mushroom, and thin-type spines based on the ratio between the diameter and length of spine heads and necks [23].

Immunohistochemical Staining of Synaptic Proteins.
Following treatment, hippocampal slice cultures were removed from the cover glass and fixed in 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (PB) overnight at 4 ∘ C. Primary antibodies were incubated for five days at 4 ∘ C in the permeabilizing solution containing 0.4% Triton X 100. Rabbit monoclonal anti-synaptophysin primary antibody was used at 1 : 400 dilution (Zymed, CA, USA) and mouse monoclonal anti-synaptopodin primary antibody was used at 1 : 400 dilution (Abcam, MA, USA), as well. Secondary antibodies (Texas Red (Invitrogen); Cy-5 (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA)) were prepared at 1 : 250 dilutions in 0.1 M PB containing 1.5% horse serum and incubated overnight at 4 ∘ C. Slices were then mounted with DAKO fluorescent mounting medium (Dako Canada, ON, Canada) onto microscope slides and imaged. Secondary and tertiary dendrites of CA1 pyramidal neurons were imaged using a Leica TCS SP2 scanhead (Leica Microsystems, Concord, ON Canada) on a Leica DM6000 B upright microscope (Leica Microsystems).

Statistical Analysis.
Two-tailed, one-way ANOVA followed by Tukey's or Dunnet's multiple comparison tests or Student's -tests is performed, and Bonferroni's correction is applied when appropriate. Results are expressed as mean ± SEM. All data analyses were performed blind.

Results
It has been previously shown that abnormalities in spine structure and composition are observed in response to aromatase inhibition; the present study considers the effects of soluble oligomeric 1-42 on characterized dendritic spine subtypes and mitochondrial structures in hippocampal neurons exposed to letrozole. Estrogens mediate neuroprotective effects in part through maintenance of mitochondrial structure and metabolic function [25,26], and, thus far, letrozole's effect on mitochondrial network plasticity has not been considered. Live dissociated mixed neural cultures treated for 24 hrs with 1 M letrozole were labelled with the mitochondrial-specific probe, MitoTracker Deep Red FM (see methods and materials). Sectioning of isolated hippocampal neuronal dendrites followed by 3D reconstruction of the resultant image stacks using Imaris software allowed for mitochondrial volume measurement acquisition and thus a readout of mitochondrial integrity, following aromatase inhibition (Figure 1(a)). We found that letrozole treatment resulted in mitochondrial fragmentation, measured as a significant reduction in volume ( Figure 1(b), control, 4.03 ± 0.56 m 3 ; letrozole, 1.72 ± 0.49 m 3 ; = 0.015), implying mitochondrial dysfunction. Next, we thought to address the question of how does blockade of estrogen synthesis affect neuronal circuitry.
To this end, we examined the effects of estrogen synthesis inhibition on the morphological characteristics of neurons and glial cells using letrozole ( Figure 2). Immunohistochemical analyses of colabelled neurons and glia from these mice using -III tubulin as a neuronal marker and anti-glial fibrillary acidic protein (GFAP) as an astrocytic marker revealed profound changes in those cultures that were treated with letrozole (1 M) for 72 hours: primary cultures that were treated with control media have extensive neuronal arbors and normal astrocytic morphology while striking changes were revealed in the letrozole-treated cultures where hippocampal neurons were sparse, with thin dendritic extensions, coupled with massive astrocytic hypertrophy (Figure 2(a)). When the letrozole treatments were repeated using organotypic hippocampal slice cultures, a culture system with more intact neuronal circuitry, we observed a significant loss of dendritic spines following 24 and 72 hours of letrozole treatment. These findings suggest that the aromatase-mediated blockade of estradiol synthesis with letrozole adversely affects neuronal circuitry in the hippocampus.
Journal of Aging Research As the densities of dendritic spines are known to be modulated by estrogens [8], we investigated the effect of letrozole on spine density and shape (Figure 2(b) and Supplementary Table 1 Table 1, stubby, P = 2.1 × 10 −4 ; mushroom, P = 1.0 × 10 −7 ; thin, P = 2.1 × 10 −4 ; total, P = 1.1 × 10 −6 ). Given the observed decrease in spine density, we subsequently investigated whether the significant decrease in spine density was subtype dependent (thin, stubby, and mushroom), which could serve as a correlate of the number of glutamatergic -amino-3-hydroxy-5-methyl-4-isoxazolepropionic (AMPA)-type receptors and synapse strength [27]. All subtypes of the spines were affected with the different treatments after 24 hours. Next, we determined if the dendritic spine loss following the treatment is time dependent, and we examined spine densities after 72 hours of letrozole treatment (Figure 2(c) and Supplementary Table  1, stubby, P = 7.6 × 10 −8 ; mushroom, P = 3.3 × 10 −8 ; thin, P = 1.8 × 10 −10 ; total, P = 1.8 × 10 −10 ). We found, at 72 hours, a continued loss of dendritic spines compared to treatments after 24 hours. These findings show that reduction of estradiol concentrations due to the aromatase inhibition affects all spine subclasses with the most marked reduction in mushroom and thin spines.
Given that letrozole alone caused mitochondrial impairments and synaptic deficits, we next determined whether it could exacerbate changes in response to soluble oligomeric  , an established mediator of neuronal deficits associated with AD pathology. First, we examined the effect of oligomeric 1-42 in conjunction with letrozole treatment on neuronal mitochondrial integrity (Figure 3). Mixed dissociated neural cultures were treated with 1 M letrozole every 24 hours from DIV 6 to 8. On DIV 8, letrozole treatment was combined with oligomeric 1-42 species (1 M) for a subsequent 24 hours, following which mitochondria were labelled and imaged as described above. As shown in Figure 3(b), we observed a trend towards a decrease ( = 0.26) in mitochondrial volume when the neuronal cultures were treated with 1-42 alone. We found a further reduction of mitochondrial volume when the cultures were pretreated with letrozole prior to oligomeric 1-42 species exposure compared to either letrozole ( = 0.17) or 1-42 alone (P = 3.37 × 10 −4 ). Our data suggest that letrozole treatment sensitizes neuronal mitochondria to alterations induced by  oligomers. Next, we examined whether these treatments can also lead to changes in synaptic structures.

Discussion
Chronic estrogen deficit, similar to that experienced by menopausal women, is a recognized risk factor for AD [31]. The neurological deficits known to result from estrogen deficiency are accompanied by morphological deterioration of postsynaptic spines and functional impairments of mitochondria. We have now shown similar pathological features in an animal model with estrogen deficiency [32] and in neural cells of normal animals with estrogen deficiency induced by letrozole, a therapeutic agent for breast cancer. However, in both instances, the more marked dendritic spine degeneration was produced only when letrozole treatment was together with oligomeric 1-42 .
Dendritic atrophy and astrocytic hypertrophy are common observations in many neurological disorders [6], including AD [33]. Although it was very clear that both circulating and brain estrogens were reduced in animal models and can lead to learning deficits, the contribution of neuroestrogens versus circulating estradiol could not be determined [15,34]. To circumvent this, we used hippocampal organotypic cultures treated with the aromatase inhibitor, letrozole. This experimental paradigm allowed us to control the concentration and duration of aromatase inhibition and estrogen deficit. Moreover, this model is optimal for studying neuronal connections and glial input in a spatiotemporal (longitudinal) There is a significant decrease in synaptophysin puncta after 24 and 72 hours of 1-42 , letrozole, or 1-42 + letrozole treatments compared to control conditions. After 72 hours, the number of synaptophysin puncta was significantly lower in letrozole and 1-42 + letrozole-treated cultures compared to  alone When synaptophysin puncta densities were compared between 24-and 72-hour treated cultures there was a significant decrease in manner, in parallel with studies of mitochondrial responses to local estrogen deficits. Functional and morphological impairment of synaptic mitochondria was associated with accumulation of 1-42 in AD animal models [35,36]. How does this relate to synaptic functions? Neuronal activity was shown to increase mitochondrial fission and entry of mitochondria into dendritic spines [37]; however fission is also initiated to sequester dysfunctional mitochondrial components and to facilitate apoptosis associated with neurodegenerative changes [38]. Recently, it has been suggested that mitochondria associated with dendritic spines may play an important role in regulating AMPA receptor trafficking, which is altered in the presence of 1-42 oligomers [5,39]. We were interested in whether changes in mitochondrial morphology under conditions of reduced estrogen synthesis in the presence of  oligomers, were associated with synaptic deficits. We noted a significant decrease in mitochondrial volume, suggesting fission after 1-42 , letrozole, or combined treatments with both. These observations are therefore in line with our present understanding of how synaptic degeneration relates to mitochondrial impairment.
Our results also suggest that estradiol may play a role in regulating the balance between the two opposing processes of mitochondrial fission and fusion. Dynamics between these two processes maintains mitochondrial homeostasis in healthy cells but is disrupted in AD [40,41]. Findings reported in this study are in accordance with recent findings indicating a role for  in the regulation of mitochondrial fission in AD [40,42]. Further emphasizing the importance of mitochondria in spine maintenance, impairment of mitochondrial transport to synapses by  has been reported to contribute to AMPAR removal and trafficking defects leading to synaptic inhibition [39].
In order to further understand how these disturbances in mitochondrial dynamics affect neuronal plasticity and function, we next investigated dendritic spine morphology, density, and synaptic proteins, following similar treatments. We examined changes in synaptophysin, a presynaptic marker, and synaptopodin, an actin binding protein, mainly found in a subgroup of mature spines and is regulated by estradiol [29]. The precise role of synaptopodin in hippocampal neurons is still unclear but has been shown to be important in learning and memory and in glutamate functions [30]. Our data show a significant reduction of preand postsynaptic components and synaptopodin with 1-42 and letrozole treatment.
Taken together, present results suggest that the neurological deficits seen in women on letrozole therapy for breast cancer could result, at least in part, from the therapyinduced hippocampal estrogen deficiency ( Figure 6). Herein, we demonstrate that dendritic spine and mitochondrial deteriorations are particularly prominent in states of central hypoestrogenicity ( Figure 6, lower half) and can further increase hippocampal susceptibility to spine/synaptic loss or mitochondrial functional impairment induced by a pathological insult. An important outcome of the current studies is the recognition that caution must be exercised in the use of letrozole as a therapy for breast cancer, and that account must be taken especially of the age of the female patient and/or presence of other recognised risk factors for AD. Given that letrozole is highly lipophilic and can cross the blood brain barrier, its incorporation into a nanodelivery system to prevent its access to the brain and at the same time enhance accumulation in the tumor warrants consideration.