Allopregnanolone Promotes Regeneration and Reduces β-Amyloid Burden in a Preclinical Model of Alzheimer's Disease

Previously, we demonstrated that allopregnanolone (APα) promoted proliferation of rodent and human neural progenitor cells in vitro. Further, we demonstrated that APα promoted neurogenesis in the hippocampal subgranular zone (SGZ) and reversed learning and memory deficits in the male triple transgenic mouse model of Alzheimer's (3xTgAD). In the current study, we determined the efficacy of APα to promote the survival of newly generated neural cells while simultaneously reducing Alzheimer's disease (AD) pathology in the 3xTgAD male mouse model. Comparative analyses between three different APα treatment regimens indicated that APα administered 1/week for 6 months was maximally efficacious for simultaneous promotion of neurogenesis and survival of newly generated cells and reduction of AD pathology. We further investigated the efficacy of APα to impact Aβ burden. Treatment was initiated either prior to or post intraneuronal Aβ accumulation. Results indicated that APα administered 1/week for 6 months significantly increased survival of newly generated neurons and simultaneously reduced Aβ pathology with greatest efficacy in the pre-pathology treatment group. APα significantly reduced Aβ generation in hippocampus, cortex, and amygdala, which was paralleled by decreased expression of Aβ-binding-alcohol-dehydrogenase. In addition, APα significantly reduced microglia activation as indicated by reduced expression of OX42 while increasing CNPase, an oligodendrocyte myelin marker. Mechanistic analyses indicated that pre-pathology treatment with APα increased expression of liver-X-receptor, pregnane-X-receptor, and 3-hydroxy-3-methyl-glutaryl-CoA-reductase (HMG-CoA-R), three proteins that regulate cholesterol homeostasis and clearance from brain. Together these findings provide preclinical evidence for the optimal treatment regimen of APα to achieve efficacy as a disease modifying therapeutic to promote regeneration while simultaneously decreasing the pathology associated with Alzheimer's disease.


Introduction
Alzheimer's disease is the result of a multifactorial disease process that ultimately leads to a decline in neural plasticity, neuroregenerative capacity, and development of amyloid-beta (Ab) plaques and neurofibrillary tangles [1]. In addition, Alzheimer's disease is also associated with myelination abnormalities in specific brain regions that are most vulnerable to AD pathology, including the hippocampus and entorhinal cortex [2]. Studies of quantitative volumetric magnetic resonance imaging assessments have revealed white matter atrophy within these regions in brains of incipient stage AD patients [3][4].
The triple transgenic mouse model of Alzheimer's disease (3xTgAD) mouse developed by Oddo, LaFerla and colleagues bears mutations in three genes (human APP SWE , Tau P301L , and PS1 M146V genes) linked to AD and fronto-temporal dementia and exhibits an age-related neuropathological phenotype including both Ab deposition and tau hyperphosphorylation [5]. Further, similar to human AD progression, 3xTgAD mice exhibit significant region-specific alterations in myelination and oligodendrocyte profile prior to the development of Ab and tau pathology [6].
Previously, we demonstrated that allopregnanolone (APa, 3ahydroxy-5a-pregnan-20-one) increased proliferation of neural progenitor cells in vitro [7]. Further, acute single administration of APa reversed both neurogenic and cognitive deficits in vivo in male 3xTgAD mice prior to the appearance of AD pathology [8]. We further demonstrated that APa successfully promoted neurogenesis and reversed cognitive deficits in male 3xTgAD mice following the onset of AD pathology. Humans with AD display reduced levels of cortical APa, which were inversely correlated with Braak and Braak neuropathological disease stage [9,10]. The APOE4 allele is also associated with reduced APa levels [10]. Aside from Alzheimer's, APa has also been shown to increase myelin basic protein in organotypic slice cultures of rat cerebellum [11] and delay demyelination in Niemann-Pick C mice [12]. The mechanism of APa induced protection of myelin integrity is suggested to involve liver X receptor (LXR) and pregnane X receptor (PXR) systems, important regulators of cholesterol, fatty acid, and glucose homeostasis [13,14]. Interestingly, the LXR and PXR synthetic ligand, T0901317 significantly decreased Ab secretion and increased the expression of ABCA1, an enzyme involved in Ab clearance, in 11-week-old APP23 mice [15].
In this study, we first determined the optimal APa treatment regimen to achieve regenerative benefits and reduction of AD pathology in the male 3xTgAD mouse model. Comparative analyses between three different APa treatment regimens indicated that APa administered 1/week for 6 months was maximally efficacious for simultaneous promotion of neurogenesis and survival of newly generated cells and reduction of AD pathology. We further investigated the impact of different therapeutic intervention stages (pre-and post-Ab pathology) of the 1/week/6 months APa treatment to reduce AD pathology and preserve myelination.

Determination of the Optimal APa Treatment Regimen to Promote Regenerative Capacity and Reduce b-Amyloid
Our previous studies indicated dose-dependent efficacy of APa on neurogenesis [7,8]. To determine the optimal regimen for therapeutic efficacy of APa treatment, we investigated the efficacy of three APa treatment paradigms depicted in Figure 1A, which were designed to determine the optimal APa treatment regimen to both promote the regenerative capacity of the brain and to reduce Ab pathology. Three different APa treatment regimens were compared. The 1/month single injection of APa treatment regimen replicated our previous approach [8] whereas the chronic treatment 3/week/3 months and 1/week/6 months paradigms were developed to simulate potential clinical treatment regimens. The optimal dose of 10 mg/kg APa was based on previous analyses [8] and was continued in this study for all three treatment regimens. Both APa treatment regimens of a single exposure of 1/ month and 1/week/6 months APa treatment significantly increased the survival of cells that were generated at the first exposure to APa. The 1/week/6 months APa treatment regimen had greater regenerative efficacy ( Fig. 1B, P,0.01). However, the 3/week/3 months regimen significantly reduced regenerative efficacy (Fig. 1B). In addition, we conducted pilot immunofluorescent labeling assays to assess the efficacy of different APa treatment regimens to reduce amyloid beta (Ab) accumulation. Results indicated that both the 1/week/6 months regimen and the 3/week/3 months regimens significantly reduced Ab accumulation in hippocampal CA1 pyramidal neurons, whereas the 1/ month single dose of 10 mg/kg APa failed to reduce Ab accumulation (Fig. 1C). Together these data suggest that the 1/ week/6 months APa treatment regimen is optimal for APa to both achieve regenerative efficacy and reduce Ab pathology.

Window of Therapeutic Efficacy
Upon determination that the 1/week/6 months regimen was the optimal treatment paradigm as it induced the greatest efficacy for promoting regenerative capacity and reducing Ab pathology, we expanded our investigation to explore the impact of initiating APa treatment at pre-versus post-Ab pathology. Specifically, we initiated the 1/week/6 months APa treatment when mice were 3 months of age with no observable amyloid pathology or when mice were 6 months of age with overt intraneuronal accumulation of Ab [5,8]. Both treatment paradigms were for six months. Mice treated at 3 months of age were 9 months of age at the end of treatment when intraneuronal Ab is apparent. Mice that began treatment at 6 months of age had intraneuronal Ab and by 12 months of age at sacrifice would have developed Ab plaques.

Allopregnanolone Promoted Survival of Newly Generated Neural Cells
To further evaluate the short term, mid term, and long term survival of newly generated cells in the 1/week/6 months APa treatment paradigm, three thymidine analogs, (BrdU, IdU and CldU) were injected at the beginning (BrdU), middle (IdU) and end (CldU) of the experiment following injections of APa to distinguish the survival of newly generated cells for the duration of 6 months (BrdU+), 3 months (IdU+) and one week (CldU+) ( Fig. 2A). Results from flow cytometry analysis indicated that APa treatment of 1/week/6 months initiated at 3 months of age significantly increased the long term (6 months; BrdU+, 98.45% 622.92) and mid term (3 months; IdU+, 68.77%618.68) survival of newly generated cells and moderately increased the short term (1 week; CldU+, 48.36%615.42) survival of newly generated cells ( Fig. 2B upper panel). The population of IdU positive cells was also increased by APa treatment initiated at 6 months of age.

Allopregnanolone Reduced Ab Oligomer Accumulation
The 3xTgAD mouse model used in the current study exhibits an age-related neuropathological progression pattern [5]. Analyses of pathological AD phenotype confirmed that mice derived from our colony expressed age-dependent and region specific Ab and tau pathology (data not shown). We first characterized the expression of different forms of Ab in 3xTgAD mice. Western blot analyses of cortex from 3xTgAD mice revealed three major Ab-related bands: full-length APP band at ,100 kD; oligomeric Ab bands at ,27 kD and ,56 kD representing hexamers (6-mer) and dodecamers (12-mer) respectively (Fig. 3A). No immunoreactive bands were detected in samples derived from non-transgenic (nonTg) mice (Fig. 3A). One 27-month 3xTgAD brain sample was included as a positive control and confirmed an age-dependent increase in band intensity (Fig. 3A). Because Ab 56 kD oligomer has been proposed to be responsible for cognitive and memory deficits in AD, our analyses focused on this oligomer and its precursor Ab 27 kD. The 56 kD Ab oligomer is likely derived from dimerization of the 27 kD Ab oligomers, which requires both the availability of the 27 kD Ab oligomer and the catalytic process of dimerization. Compared to vehicle treatment, APa treatment partially reduced Ab*56 in both age groups, with a 2564% reduction in the pre-pathology treatment group (n = 5, P,0.01) and 1564% reduction in the post-pathology treatment group (n = 3-4, P = 0.05). APa treatment also reduced the level of Ab 6mer as indicated by a significant reduction in 27 kD band (35610%, n = 5, P,0.05) intensity in the pre-pathology treatment group. In contrast, in the post-pathology treatment group, no significant reduction of Ab 6-mer occurred (Fig. 3B). These data indicate that pre-pathology APa treatment is reducing the generation of the early oligomer (27 kD) and the late oligomer (56 kD) whereas the post-pathology APa treatment appears to selectively delay the formation of the 56 kD oligomer. In addition, APa had no effect on APP protein expression, suggesting that APa did not directly affect APP generation. Together these data suggest that APa reduced the oligomerization of Ab. Additional immunofluorescent analyses confirmed the findings from Western blot that APa treatment induced an apparent reduction of Ab immunoreactivity in specific brain regions including hippocampus, cortex, and amygdala ( Figure 3C).

Allopregnanolone Reduced ABAD Expression in 3xTgAD Mice
The mitochondrial protein ABAD (amyloid-binding alcohol dehydrogenase) is over-expressed in human AD brains [16,17,18] and 3xTgAD mice [19]. Ab binds to ABAD and disrupts mitochondrial function, leading to the generation of reactive oxygen species and cellular oxidative damage [18]. In parallel to APa-induced reduction of Ab generation, APa-treatment significantly decreased cortical expression of ABAD in the prepathology group (3064%, n = 3, P,0.05) and induced a trend of reduction in the post-pathology treatment group (2067%, n = 3-4, P = 0.07) ( Figure 4A). Similarly, additional immunofluorescent analyses revealed a qualitative reduction in ABAD immunofluorescent intensity in APa-treated 3xTgAD mouse brain sections relative to vehicle-treated brains (Fig. 4B). Together, APa treatment induced a parallel reduction of Ab and ABAD expression indicating a potential protective mechanism whereby APa prevents Ab induced mitochondrial dysfunction [18].

Allopregnanolone Modulated Phosphorylated-tau Expression in 3xTgAD mice
In the 3xTgAD mouse model, tau pathology has been demonstrated to follow Ab accumulation [20]. To determine whether the APa-induced reduction of Ab levels would lead to reduction in tau pathology, we investigated the impact of APa on tau phosphorylation by Western blot analysis and immunofluorescent labeling with monoclonal phosphorylated-tau antibody in 3xTgAD mouse model. All three APa treatments were initiated when mice were 3 months old. Upon completion of each treatment paradigm, BrdU stereology (for 3/week/3 months treatment paradigm, n = 7-8/group) or flow cytometry assay (for 1/ week/6 months treatment paradigm, n = 2-5/group and 1/month treatment paradigm, n = 10-12/group) was used to assess neurogenesis. Both the 1/month APa treatment and the 1/week/6 months APa treatment induced a significant increase in neurogenesis, with the latter regimen yielding the greater increase in neurogenesis. However, the 3/week/6 months treatment induced a significant decrease in neurogenesis. The percentage of increase is presented as mean 6 SEM * P,0.01. (C) Impact of APa on Ab accumulation in 3xTgAD mouse model. Brain sections from 3xTgAD mice treated with APa (10 mg/kg) or vehicle were stained. Ab immunoreactivity was detected with 6E10 antibody (green) and nuclei counter-stained with DAPI (blue). Representative images indicated that the 1/week/6 months APa treatment significantly decreased 6E10 immunoreactivity and showed the highest efficacy of Ab reduction; whereas the 3/week/3 months APa treatment had a comparable efficacy of Ab reduction and the 1/month APa treatment showed minimal effect of reducing Ab immunoreactivity. doi:10.1371/journal.pone.0024293.g001 (AT8), which recognizes phosphorylated tau serine 202. There was a trend towards reduction in phospho-tau band intensity in both pre-pathology treatment and post-pathology treatment groups, which did not reach statistical difference between APa-and vehicle-treated samples (Fig. 5A). However, immunofluorescent analyses indicated that APa induced a reduction in phospho-tau immunoreactivity at 12 months in the hippocampal CA1, frontal cortex and amygdalar regions ( Figure 5B). In the pre-pathology treatment group when mice were at 9 months of age, phospho-tau was barely detectable.

Allopregnanolone Regulated LXR, PXR and HMG-CoA-R Expression in 3xTgAD Mice
APa has been proposed to regulate cholesterol homeostasis via the LXR and PXR system [21,22] and dysregulation of cholesterol homeostasis is associated with the generation of Ab [15]. In the current study, we investigated APa regulation of inducible liver X receptor (LXR), pregnane X receptor (PXR) and 3-hydroxy-3methylglutaryl coenzyme A reductase (HMG-CoA-R) expression. APa induced a significant increase in LXR, PXR and HMG-CoA-R expression in the pre-pathology treatment group. In contrast, in the post-pathology treatment group, APa treatment resulted in a reversal of cholesterol homeostatic responses, which was particularly evident in the significant reduction of PXR expression (Fig. 6). This suggests that the effect of APa is specific to the synthesis and clearance of cholesterol rather than the proteins involved in cholesterol trafficking.

Allopregnanolone Treatment Inhibited Microglial Activation
Microglial activation is well-documented in the pathogenesis of AD and is potentially related to presence of oligomerized Ab. Further, in a model of cholesterol dysregulation, APa inhibited microglial activation in Niemann-Pick C mice [23]. The reduction in the 56 kD Ab oligomer and ABAD in the post-pathology APa treatment group would be expected to result in a reduction of neuroinflammatory responses but to a lesser extent than the prepathology group. Consistent with these results, we observed a significant age-dependent increase in microglial activation in 3xTgAD mice as indicated by the significant increase activated microglia expression through immunoblotting with anti-CD11b/c (OX42) antibody, from 9 months to 12 months in the vehicle Figure 2. Impact of optimal 1/week/6 m allopregnanolone treatment regimen on neurogenesis and survival in 3xTgAD male mouse. (A) The paradigm of once per week for 6 months of treatment. Three-month-old (pre-detectable pathology) and six-month-old (readily detectable pathology) male 3xTgAD mice were randomized into vehicle or APa treatment groups. APa or vehicle was subcutaneously administered at 10 mg/kg once per week for 6 months (1/week/6 month). To assess the survival capability of newly generated cells, mice were treated with sequential injections of BrdU for the first 5 days of the study (indicating a 6 month survival period), IdU for 5 days at the midpoint (indicating a 3 month survival period), and CldU 5 days before completion of the study (indicating a short term survival period). After 6 months of APa treatment, mice at 9-and 12-months-old were sacrificed. (B) Impact of survival of newly generated cells in hippocampus subgranular zone (SGZ). Flow cytometry analysis showed APa treatment (1/week/6 months) significantly increased IdU positive cells in 9-month-old (upper panel) and 12-month-old mice (lower panel); BrdU positive cells were increased in 9-month-old mice (upper panel). CldU positive cells were not significantly increased in both ages suggesting that this APa treatment regimen sustained survival of newly generated cells, and that neurogenic capacity of APa decreased with age. The percentage increase was presented as mean 6 SEM; * P,0.01; n = 2-5/group. doi:10.1371/journal.pone.0024293.g002 group mice (Fig. 7A, P,0.01). More importantly, compared to the vehicle control, APa treatment in 3xTgAD mice significantly decreased OX42 expression level in both the pre-pathology treatment (26.264.6%, P,0.05) and the post-pathology treatment group (18.067.4%, P = 0.05) (Fig.7A). Further immunofluorescent labeling with rabbit polyclonal anti-Iba1, another activated microglia marker, revealed APa treatment induced reduction of microglia immunoreactivity in hippocampal CA1 region (Fig. 7B).

Allopregnanolone Increased a Marker of Myelination in Brains of 3xTgAD Mice
Decline in myelination is a neuropathological consequence in the 3xTgAD animal model [6]. APa-enhanced myelination has been reported in Niemann-Pick C mice [23]. To investigate the therapeutic potential of APa to delay or reverse myelination deficits that occur in human AD [24,25] and in 3xTgAD mice, we investigated the impact of APa on myelin expression in 3xTgAD mouse brain. APa induced a significant increase in the expression of CNPase, a marker of oligodendrocytes and myelin, relative to vehicle control by 40% in nonTg (P,0.01, n = 3) and 50% in 3xTgAD (P,0.05, n = 3) mice in the pre-pathology treatment group (Fig. 8A). Similarly, in the post-pathology treatment group, APa treatment also induced a 30% increase in CNPase levels in 3xTgAD mice (P,0.05, n = 3-4) (Fig. 8A). In the nonTg group, expression of CNPase increased in the 12-month-old vehicle group compared to 9-month-old vehicle, in which CNPase expression was not increased by APa (Fig. 8A). Immunofluorescent labeling with the same antibody revealed a region-specific increase in CNPase immunoreactivity in the hippocampal CA1 (B), entorhinal cortex (C) and primary somatosensory cortex (D) (Fig. 8B). . Allopregnanolone reduced expression of Ab oligomers in 9-and 12-month-old male 3xTgAD mice. Equal amounts of frontalparietal-temporal cortex samples from 3xTgAD mice treated with APa (10 mg/kg) or vehicle were loaded onto the gel. Ab expression was determined with Ab antibody 6E10 by immunoblot analysis. (A) Three major immunoreactive bands were detected in the samples from 9 and 12 month old 3xTgAD mouse brains. Full-length APP band was detected at ,100 kD. The bands at ,27 kD and 56 kD indicated Ab hexamers (6-mer) and Ab dodecamers (12-mer) respectively. One 27-month old 3xTgAD cortex sample was included as a positive control. (B) For better separation of the Ab 56 and 27 kD bands, a 10-20% gradient gel was used to analyze the effects of APa on Ab oligomers. APa treatment significantly reduced Ab 56 kD oligomer in the pre-pathology treatment (2564%, n = 5, P,0.01) and the post-pathology treatment (1564%, n = 3-4, P = 0.05) groups. Ab 6-mer at ,27 kDa was also significantly reduced by APa treatment in the pre-pathology treatment group (35610%, P,0.05). Bars represent mean relative expression 6 SEM (* P#0.05 and ** P,0.01 compared with vehicle control group). (C) Region specific reduction in Ab IR by APa treatment in 3xTgAD brains. Representative images of Ab immunostaining indicated a significant decrease of Ab IR in the hippocampus and amygdala in APa-treated 3xTgAD mice relative to vehicle. doi:10.1371/journal.pone.0024293.g003

Discussion
Regenerative therapeutics that target AD pathology and the disease process Previously we demonstrated that APa reversed neurogenic and cognitive deficits in the male 3xTgAD mouse model of Alzheimer's [8]. APa was efficacious in the 6-and 9-month old 3xTgAD mice on both neurogenesis and associative learning and memory and was comparable to their respective age-matched nonTg controls which corresponded to a 100% or more increase in response induced by APa [8,26].
In vehicle-treated mice, there was a substantial age-related decline in neurogenesis that was exacerbated in 3xTgAD mice. Between 3 and 6 months there was a 78% decline and between 6 and 9 months there was another 58% decline, similarly an 18% decline was observed between 9 and 12 months [8,26]. At 12 months of age in the 3xTgAD, APa no longer increased neurogenesis nor did it promote survival of neural progenitors. When extraneuronal Ab plaques were present in the 12-month-old 3xTgAD mice, APa was ineffective in reversing the profound neurogenic and cognitive deficits of these mice. This is specific to the transgenic genotype because APa promoted neurogenesis and neuroprogenitor survival in the 15 month nonTg mice [26]. Treatment with APa increased neural progenitor survival and restored learning and memory to that of 12-month-old nonTg mice [26]. By the time 3xTgAD mice reached 12 months of age either the progenitor population was sufficiently depleted or the mechanisms by which APa promotes neuroproliferation and/or survival were no longer present and/or functional.
In the current study, we determined the efficacy of APa to promote the survival of newly generated neural cells while simultaneously reducing Alzheimer's disease (AD) pathology in the same mouse model with different APa treatment regimens. APa treatment paradigms included 1/month, 3/week/3 months, or 1/week/6 months, and were developed to simulate potential clinical regimens. The 1/month APa treatment regimen was efficacious in promoting neurogenesis and reversing cognitive deficits whereas it had no effect on AD pathology. In contrast, the most frequent treatment paradigm, 3/week/3 months APa was most efficacious for reducing AD pathology whereas it was least effective at promoting neurogenesis. Subsequent analyses indicated that APa (10 mg/kg) administered 1/week/6 months achieved an optimal balance between promoting neurogenesis and reducing intracellular b-amyloid accumulation.
Oligomeric Ab species in the brain are considered the major toxic form of Ab, and activate a detrimental cascade that leads to neuronal loss, cognitive decline, and eventually AD diagnosis [1,27,28,29,30,31,32]. Increasing evidence indicates that treatments initiated late in the disease process, i.e. dense and wide distribution of Ab plaques, have a low probability of clinical success [33] suggesting that targeting the early phase of oligomeric Ab generation may have greater clinical efficacy. Results demonstrated that APa administered 1/week/6 months significantly increased survival of newly-generated neurons and simultaneously reduced Ab pathology in both age groups. While AP significantly reduced the 56 kD Ab oligomer in both the prepathology and post-pathology groups, greatest efficacy was observed when APa was administered in the pre-pathology phase where APa reduced both the 27 kD and 56 kD Ab oligomers. Consistent with the biochemical results, Ab immunoreactivity was reduced in hippocampal CA1 and moderately reduced in frontal cortex, amygdala and subiculum. Ab binding alcohol dehydrogenase, a mitochondrial Ab binding enzyme, was reduced in parallel with Ab by APa treatment. ABAD has been well-demonstrated to interact with Ab within mitochondria and cause mitochondrial dysfunction [16]. In AD patients and the 3xTgAD mice, ABAD expression parallels severity of Ab load [19]. Based on these findings, reduction of both Ab and ABAD load with APa treatment would be predicted to prevent the exacerbation of mitochondrial deficits associated with AD and therefore delay or prevent disease progression.
In this mouse model, phosphorylated-tau follows severity of Ab accumulation [34]. The APa-induced reduction in intraneuronal b-amyloid was accompanied by a moderate, although not statistically significant, brain-wide reduction in phosphorylatedtau with the greatest reduction of phosphorylated-tau in ADvulnerable brain regions. The reduction of tau pathology by APa can likely be attributed to the reduction in Ab load. Microglial activation is a well-documented response to AD pathology [35,36,37]. In parallel to the reduction of Ab and phospho-tau, APa treatment significantly reduced microglial activation as indicated by reduced expression of OX42. While the reduction in microglia activation by APa is likely a consequence of APainduced reduction of AD pathology, suppression of microglial activation would relieve the inflammatory burden associated with AD pathology.
White matter abnormalities have been widely reported in AD [2,3,4,38] and observed in the 3xTgAD mouse brain as well [6]. In the 3xTgAD mouse brain, abnormal myelination and loss of axonal integrity occur in the same brain regions vulnerable to AD pathology in humans [6]. In the present study, we found that in parallel to reduction of AD pathology, APa treatment also increased a marker of myelination in these same brain regions. Collectively, results of these analyses demonstrate the concomitant reduction by APa of four major pathological markers of AD, Ab, phospho-tau, microglial activation and white matter loss, and provide pre-clinical evidence in support of the efficacy of APa to decrease and delay development of AD pathology.
Increasing evidence indicates that altered cholesterol metabolism is linked to AD pathology [39,40]. Mechanistic analyses indicated that 1/week APa begun at 3 months and continued for 6 months increased expression of LXR, PXR, and HMG-CoA reductase, three proteins that regulate cholesterol homeostasis. LXR, a nuclear hormone receptor abundant in the brain [41], acts as a molecular sensor of cholesterol levels and initiates cholesterol clearance [15]. LXR activation increases cholesterol efflux through up-regulating ABCA1 and ApoE expression, and prevents the hyper-activation of c-secretase and over-production of Ab [15,42,43]. LXR activation has been demonstrated to improve cognitive function in multiple mouse models of amyloidogenesis [15,44,45,46,47].
LXR ligands frequently activate PXR [48]. Results from our analyses indicated that in parallel with an APa-induced increase in LXR expression in the pre-pathology condition, APa also increased PXR expression in the pre-pathology 3xTgAD mouse brain. PXR activation induces CYP3A enzymes including CYP3A4 and CYP3A13 and leads to cholesterol hydroxylation and activation of organic anion transporters for cholesterol extrusion [49]. The APa induced increase in brain LXR and PXR leads to increased cholesterol efflux, thereby reducing csecretase activation by cholesterol-laden lipid rafts. Increased cholesterol efflux provides a plausible mechanism to explain how APa decreased the generation of both 27 kD and 56 kD intraneuronal Ab oligomers.
To determine whether APa primarily effected cholesterol homeostatic mechanisms primarily in brain, we investigated LXR and PXR expression in liver. Further, these analyses would also address whether the vehicle (2-hydroxypropyl-b-cyclodextrin), which can reduce cholesterol in other tissues, would also affect expression of HMG Co-A reductase, LXR and PXR in liver. To address these issues, we conducted Western blot analysis for LXR and PXR expression in liver. Results of these analyses indicated no significant effect of APa on LXR or PXR expression in liver (data not shown). These data support our findings indicating that APa is primarily affecting cholesterol homeostasis in brain without affecting peripheral cholesterol.
These findings in the 3xTgAD mouse brain are consistent with APa induced activation of a PXR pathway in cholesterol trafficking in the Niemann-Pick C Disease mouse model [22]. APa-induced reduction of AD pathology burden is also consistent with the findings of Mellon and colleagues who reported that APa delayed progression of Niemann-Pick C Disease in a mouse model [23,52,53]. In young animals, either single or multiple injections of APa protected cerebellar Purkinje cells from degeneration and increased animal life span [52]. Less improvement was observed at older ages of Niemann-Pick C12/2 mouse that had disrupted neurosteroidogenesis [52]. APa induced a delay in progression of pathology and enhanced survival of Niemann-Pick C mice through a PXR-mediated mechanism [22].
Our study also revealed that APa induces an increase in HMG-CoA reductase protein expression. The increase in HMG-CoA reductase is at first paradoxical as it is the rate-limiting enzyme in cholesterol synthesis. HMG-CoA reductase is also required for oxysterol generation which is well documented to activate LXR and PXR-mediated gene transcription for cholesterol and lipid transport proteins [40]. If this hypothesis is correct, it would predict decreased activation of c-secretase by cholesterol and lipid laden lipid rafts.
In this study, an interesting and consistent finding was that the pre-pathology treatment exhibited a greater magnitude of efficacy in terms of promoting survival of neural progenitors, reducing Ab pathology, suppressing inflammatory response, and activating LXR/PXR pathways involved in cholesterol homeostasis and Ab clearance. These findings indicate that early development of pathology serves as a critical stage for APa therapeutic efficacy that coincides with intraneuronal Ab accumulation. The appearance of Ab plaques coincides with cessation of APa efficacy. One potential reason for this cessation is that Ab has been transported out of the cell and our data suggest that intraneuronal Ab accumulation is a determining factor in efficacy of APa which is paralleled by the lack of efficacy once intraneuronal Ab is extracellularly localized. The well-established relationship between cholesterol homeostasis and Ab generation coupled with our findings of APa induced pathways of cholesterol homeostasis coinciding with intracellular Ab, suggest that these two systems, i.e. cholesterol homeostasis and intraneuronal Ab and APa efficacy are coupled.
The deposition of Ab in the extracellular compartment disconnects this coupled pathway, leading to a loss of efficacy of APa. Our data indicate that the presence of intraneuronal Ab is the key regulatory factor in determining therapeutic efficacy. Although the molecular content of extracellular and intraneuronal Ab may be similar, their localization is a key indicator of cellular adaptation and therapeutic regulation.

Translational implications for the optimal therapeutic regimen
APa induction of neurogenesis [7,8], recovery of learning and memory function [8], and reduction of AD pathology burden provides pre-clinical evidence for APa as a multifaceted regenerative therapeutic. Moreover, APa induction of cell cycle gene expression [7] and key regulators of cholesterol homeostasis provides mechanistic plausibility for its therapeutic efficacy to promote neurogenesis and cognitive function while reducing AD pathology. Two factors are critical to the regimen of APa to achieve therapeutic efficacy. The first of these factors is the temporal regimen of administration. Our data show that regeneration is achieved with either 1/month or 1/week regimen of APa. Reduction of AD pathology can be achieved with 1/week or 3/week regimen. The combination of regeneration and reduction of pathology was only achievable with the 1/week APa treatment regimen. The second regulating factor is the type of pathology. Administration of APa prior to and during the early stages of AD pathology (intraneuronal Ab) was efficacious in reducing the burden of pathology. APa treatment initiated at the stage of Ab plaque development was associated with reduction in efficacy. These findings predict that APa therapeutic benefit in humans would be most efficacious to delay progression of AD when brains still have neurogenic and myelination capacity. Target populations could include those with very early stage genetically mediated familial AD and those diagnosed with mild cognitive impairment (MCI).

Animal Treatments and Ethics
All rodent experiments were performed following National Institutes of Health guidelines on use of laboratory animals and an approved protocol by the University of Southern California Institutional Animal Care and Use Committee (Protocol Number: 11156). The presented study has been approved by the University of Southern California Institutional Animal Care and Use Committee (Ethics Committee).

Transgenic Mice
Colonies of 3xTgAD (3xTgAD, homozygous mutant of human APPswe and tauP301L and PS1M146V) and nonTg mouse strain (C57BL6/129S; Gift from Dr. Frank LaFerla, University of California, Irvine) [5] were bred and maintained at the University of Southern California (Los Angeles, CA) following National Institutes of Health guidelines on use of laboratory animals and an approved protocol by the University of Southern California Institutional Animal Care and Use Committee. In addition, the minimal number of required animals was used for these experiments and pain was minimized. Mice were housed on 12 h light/dark cycles and provided with ad libitum access to food and water. The characterization of amyloid and tau pathologies, as well as synaptic dysfunction in this line of mice has been described previously [20,54] and confirmed in our laboratory. The mice were genotyped regularly to confirm the purity of the colony. To ensure the stability of AD-like phenotype in the 3xTgAD mouse colony, we performed routine immunohistological assays every three to four generations. Only offspring from breeders that exhibited stable AD pathology were randomized into the study. The number of mice per condition is indicated within the results section. Experiments were performed using three-and six-monthold male 3xTgAD and age-matched nonTg mice.

Treatment paradigms
Three different APa treatment paradigms were adopted to determine the optimal APa treatment regimen to promote regenerative capacity and simultaneous reduce AD pathology. Details of these treatment paradigms are listed below ( Figure 1A): Paradigm 1. Three-month-old male 3xTgAD mice were injected with either APa (10 mg/kg) or vehicle once and analyzed 1 month later (1/month). One hour after APa injection, mice were injected with BrdU (100 mg/kg). Mice were sacrificed 27 days after APa administration for cell survival assessment.
Paradigm 2. APa was injected to mice once a week for 6 months (1/week/6 months). In order to investigate whether APa treatment could prevent or reverse AD-related pathologies, we chose 3-and 6-month-old 3xTgAD and age-matched nonTg mice for the experiments on our previous observation that detectable intraneuronal Ab pathology starts at about 6 months of age in the male 3xTgAD model, whereas tau pathology follows Ab pathology development and is detectable at older age [5]. During APa (10 mg/kg) treatment, mice were treated with sequential injections of BrdU (100 mg/kg) at the first 5 days of the study, IdU (100 mg/ kg) for 5 days at the midpoint, and CldU (100 mg/kg) 5 days before completion of the study. After 6 months of treatment, mice at 9 and 12 months of age were sacrificed.
Paradigm 3. Three-month-old male 3xTgAD mice were injected with either APa (10 mg/kg) or vehicle once every other day for 3 months (3/week/3 months). In the first five weeks, BrdU (100 mg/kg) was injected 1 h after each APa given. One week before sacrifice, BrdU was injected again 1 h after APa injection.

Animal Dissection and Tissue Collection
Upon completion of the treatments, animals were sacrificed. Prior to sacrifice, mice were anesthetized with 100 mg/kg ketamine and 10 mg/kg xylazine and perfused with pre-chilled PBS. Brains were immediately dissected along the sagittal line into two hemispheres; the left hemisphere was frozen on dry ice and then stored at 280uC for biochemical analysis, and the right hemisphere was post-fixed in 4% paraformaldehyde for immunohistochemical analysis. Fixed brains were embedded into blocks (40 brain hemispheres in the same block) for cryostat sectioning by NeuroScience Associates (NSA, Inc., Knoxville, TN). The 40brain hemisphere block was serially sliced into 35 mm coronal sections and the free-floating multibrain sections were kept in antigen preservation solution (1% Polyvinyl pyrrolidone and 50% Ethylene glycol in PBS) at 220uC until use.

Unbiased Stereology
Number of BrdU-labeled cells was determined in every sixth section in a series of 40 mm coronal sections using unbiased stereology (optical dissector). The first section of each hemisphere was randomly started at the beginning of olfactory, and serial sections were collected to the end of the cerebellum. Systematic samplings of unbiased counting frames of 50 mm on a side with a 200 mm matrix spacing were produced using a semiautomatic stereology system (Zeiss Axiovert 200 M fluorescent microscope as part of the 3iMarianas digital microscopy and a 606 SPlan apochromat oil objective (1.4 numerical aperture)). Positive cells that intersected the uppermost focal (exclusion) plane and those that intersected the exclusion boundaries of the unbiased sampling frame were excluded from analysis. Cells that met analysis criteria through a 20 mm axial distance were counted according to the optical dissector principle. The granule cell layer reference volume was determined by summing the traced SGZ, granule cell areas for each section multiplied by the distance between sections sampled. The mean granule cell number per dissector volume was multiplied by the reference volume to estimate the total granule cell number. The stereologically determined number of BrdUpositive cells was related to the granule cell layer sectional volume and multiplied by the reference volume to estimate the total number of BrdU-positive cells.

Flow-cytometry Analysis of BrdU, IdU, and CldU Incorporation
Along with the study, our group developed a flow cytometry counting method for analysis of BrdU, IdU, and CldU incorporation. This flow cytometry analysis method has been previously validated by direct comparison between unbiased stereology and flow cytometry analysis [8,55]. Briefly, hippocampi were dissected from the fixed hemispheres using anatomical landmarks as described [56]. Extracted hippocampi were homogenized and nuclei sample collected into a 1.5 ml microcentrifuge tube, washed four times using 200 ml of PBS, and then centrifuged for 10 min at 10,000 rpm. The pellet was then resuspended in 600 ml of PBS plus 0.5% Triton X-100, heated for 1 h at 75 Cu for epitope retrieval, and incubated for 24 h at 4 Cu with primary mouse monoclonal anti-BrdU antibody (1:100, Ab12219, Abcam, Cambridge, MA) for BrdU+ cell count; anti-CldU (1:40, ab6326, Abcam) for CldU+ cell count; anti-IdU (1:20, BD, 340649), for IdU+ cell count. The number of nuclei was estimated by counting the propidium iodide, and the number of BrdU-, IdU-and CldU-labeled cells was detected using Beckman Flow Cytometry System (FC 500) with CXP Software.

Statistical Analysis
Statistical significance for group comparison was performed by a Student's t-Test or one way ANOVA followed by a Newman-Keuls post-hoc analysis. The difference between groups was considered significant when the P value was ,0.05.