STARD1 and NPC1 expression as pathological markers associated with astrogliosis in post-mortem brains from patients with Alzheimer's disease and Down syndrome

Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder of complex etiology and the most common form of dementia, which comprises early-onset AD (EOAD, ~5% of patients) and late-onset AD (LOAD, 95% of patients). The generation of toxic beta amyloid (Aβ) peptides is considered a central player in AD pathogenesis, although other factors such as tau phosphorylation are also key for the progression of AD [1]. While alterations in cholesterol homeostasis have been linked to AD the role for this association is controversial. For instance, there has been evidence indicating that either increased or decreased total brain cholesterol levels are associated with AD-pathogenesis [2]. Whether intracellular cholesterol pools (e.g. endolysosomes and mitochondria) rather than total cholesterol levels are more relevant for AD pathology remains to be established. Indeed, hippocampus and frontal cortex of patients with AD and AD-Tg mice have been shown to exhibit increased expression of the lysosomal cholesterol transporter NPC1 [3], while increased steroidogenic acute regulatory protein (STARD1) expression, which regulates the mitochondrial cholesterol trafficking [4], has been reported in pyramidal hippocampal neurons of patients with AD [5]. Moreover, mitochondrial cholesterol accumulation in a hypercholesterolemic AD-Tg mouse model has been shown to impair mitochondrial antioxidant defense and stimulate oxidative stress, which results in the acceleration of AD-like neuropathology [6, 7]. However, the intracellular cholesterol homeostasis in human AD has not been explored.

Down syndrome (DS) is considered a genetically determined form of AD and individuals with DS invariably develop AD by their fourth decade [8]. Patients with DS exhibit cognitive markers of preclinical and prodromal AD [9]. The progression of AD pathology assessed by neuroimaging and/or cerebrospinal fluid levels of Aβ and tau (total and phosphorylated) revealed a similar pattern between normal population and individuals with DS, although these events in DS occur 10–20 years earlier than in normal population [10, 11]. Aging, obesity and ApoEε4-associated dyslipidaemia are important risk factors for development of AD dementia in normal and DS population [12]. Interestingly, hypercholesterolemic subjects with DS treated with statins exhibit a lower risk factor for AD development that correlates with slower amyloidogenic burden [13]. Besides the overlap between AD and DS, intracellular brain cholesterol homeostasis in patients with DS has not been reported. We hypothesized that alterations in intracellular cholesterol homeostasis in lysosomes and mitochondria in the brain could contribute to the molecular events involved in the AD continuum, emerging as a pathological event that could contribute to disease progression. Investigating the intracellular cholesterol homeostasis is critical to improve and expand the therapeutic armamentarium for AD, especially at preclinical stages, given that current clinical trials for AD have been disappointing. In contrast to the controversial role of changes in total brain cholesterol levels in AD [2, 14], our findings on the expression of putative intracellular cholesterol carriers (e.g. STARD1 and NPC1) in patients with AD and DS correlate with disease markers, such as Aβ₄₂, and suggest that alterations in intracellular cholesterol homeostasis may be an early molecular event that favours the progression of the AD pathology. Moreover, the accuracy of STARD1 to discriminate controls from AD in the general population and in subjects with DS suggests that StARD1 could be a potential novel marker associated with early molecular events of AD pathology in both populations.

Results

Discussion

We have analyzed the intracellular cholesterol trafficking in post-mortem samples from patients with DS and AD and assessed whether this event correlated with AD markers. An intriguing finding was that Aβ₄₂ but not p-tau immunoreactivity significantly increased in CA1-3 hippocampal regions only in patients with DS, with a trend for increased Aβ₄₂ levels in CA1 from patients with AD. These findings in DS are consistent with the intrinsic Aβ overproduction in these subjects [8]. Moreover, neuropathological analysis of amyloid pathology in DS described the onset of mid-severity changes in hippocampal CA1 region at 35 years of age, in which Aβ₄₂ deposition occurred in the absence of tau pathology [18]. Conversely, tauopathies and the presence of NFTs that commonly coexist with LOAD can occur without amyloid pathology [19, 20]. In addition, it has been suggested that the earliest appearance of tau pathology in AD occurs in the locus caeruleus instead of hippocampus [21], which could account for the lack of p-tau immnunoreactivity in hippocampal samples from our cohort of patients with AD. While the present work is the first to our knowledge to address the expression of putative intracellular cholesterol carriers (e.g. STARD1 and NPC1) in AD and DS and their correlation with pathological markers, the role of changes in total cholesterol levels in AD is controversial [2, 14], consistent with the unsettled impact of statins in the prevention or treatment of the disease, which will require further randomized controlled trials [22, 23].

Although the groups studied were not age-matched given the limited number of samples and the life-expectancy of individuals with DS [8], we performed the comparisons with AD and controls considering AD pathogenesis as a function of age in DS, a population that exhibit a high risk to develop AD-type dementia [12, 24]. Moreover, as described previously [18], individuals with DS developed signs of AD-pathology by 50-55 years of age, which coincides with the age range of the subjects with DS analyzed in our study (Table 1). In this context, interestingly we found increased expression of NPC1 and STARD1 in cortical samples from patients with AD and DS, respectively, (Figure 1). A link between lysosomal cholesterol and Aβ₄₂ deposition in neurodegeneration has been described in Niemann–Pick type C (NPC) disease, a rare autosomal recessive disease associated mostly with NPC1 mutations, characterized by lysosomal accumulation of unesterified cholesterol [25]. While the etiology and epidemiology of AD and NPC differ, these neurodegenerative diseases share many disease-related molecular pathways, including cholesterol accumulation, lysosomal abnormalities, tau hyperphosphorylation, APP processing and Aβ deposition [2, 25]. Moreover, while increased mRNA and protein levels of NPC1 have been described within the hippocampus of both patients with AD and mice [3], the hippocampal expression pattern of NPC1 in human DS-brains has not been reported. Our findings suggest that while altered hippocampal NPC1 expression could be a common feature of DS and AD (Figure 2), the NPC1 immunolabeling as a disease marker of AD does not discriminate AD-pathology in patients with DS from the controls (Figure 3), suggesting that the AD continuum in the hippocampus of subjects with DS differs from NPC disease.

Cholesterol is known to promote the amyloidogenic processing of APP, which occurs predominantly in lipid-rafts where most components of the amyloidogenic machinery, such as APP, Aβ, BACE1, presenilins (PSEN1/2), and γ-secretase are present [26, 27]. Moreover, cholesterol is thought to interfere with the α-secretase-dependent non-amyloidogenic APP processing [28], and PSEN1/2 and γ-secretase have been localized at the mitochondrial ER-associated membrane (MAM) [29, 30]. Interestingly, although mitochondrial membranes are low in cholesterol, the pool of cholesterol of the mitochondrial inner membrane (MIM) is crucial for the synthesis of neurosteroids to maintain physiological GABAergic responses that modulate memory function, which is particularly affected in subjects with DS [31, 32]. The STARD1-dependent trafficking of cholesterol to MIM is the rate-limiting step for steroidogenesis and changes in mitochondrial cholesterol levels can impact this process and affect mitochondrial antioxidant defense [2, 5–7, 33]. However, the role of STARD1 in human AD is poorly understood and has not been examined in DS. We show for the first time that the hippocampal overexpression of STARD1 correlates with amyloid deposition suggesting a relationship between the accumulation of mitochondrial cholesterol and the development of AD-pathology (Figure 2). Unexpectedly, our findings in hippocampal samples from patients with DS and AD suggest that STARD1 immunolabeling can act as a pathological mark of AD even better than Aβ₄₂ deposition in post-morten samples (Figure 3), implying that the hippocampal alterations in the expression of STARD1 could occur in a temporal pattern parallel to the pathogenic amyloidogenic burden. Whether these alterations in STARD1 occur before Aβ₄₂ deposition in human AD requires further investigation. However, previous evidence in experimental models pointed that enhanced mitochondrial cholesterol levels in AD mouse models have been shown to sensitize neurons to Aβ-induced inflammation and toxicity by depleting mGSH by an ER stress-dependent mechanism [6, 7, 33]. Moreover, as Aβ-induced ER stress is considered an indirect effector of the Aβ-neurotoxicity in early stages of AD [34], our data suggest that STARD1 overexpression in the hippocampus may be a potential early molecular event associated with AD, as strengthened by the increase seen in DS, an accelerated genetic form of AD. In this context, it is widely recognized that neuroinflammation in AD is an early pathological-associated response, which is exacerbated with ageing and contributes to disease progression [18, 35].

Although not statistical significant, our findings do suggest a trend for lower expression of STARD3/MLN64 in the cortex of AD samples (Figure 1B), whose impact in regulating the mitochondrial pool of cholesterol remains to be understood. In this regard, it is interesting to note that increased expression of MLN64 has been reported in the lysosomal storage disorder Niemann-Pick type C disease, which correlated with increased mitochondrial cholesterol accumulation [36]. However, it has also been reported that targeted mutations of the MLN64 START domain have been shown to cause minor alterations in metabolism and intracellular distribution of cholesterol [37]. Further investigation will be required to decipher the impact of MLN64 in mitochondrial cholesterol regulation and AD pathogenesis.

Reactive astrogliosis has emerged as a relevant pathological response in the early stages of AD [38], which correlates with cognitive decline [16]. Since astrocytes are the most prevalent cell type in the brain and form an intricate system of connections to control a wide range of brain functions, including regulation of blood-brain barrier, the astroglial atrophy could have far-reaching consequences on synaptic transmission that accounts for the spatio-temporal progression of AD-neurodegeneration [17]. We confirm the presence of reactive astrocytes surrounding amyloid plaque in the hippocampus of AD and DS, and observed a hippocampal region-specific increased expression of STARD1 within astocytes (Figure 4) that correlates with the accumulation of mitochondrial cholesterol in both groups (Figure 5). Similarly, we also found astroglial NPC1 overexpression and astroglial accumulation of lysosomal cholesterol in AD and DS (Figures 4 and 5). These results are in line with findings reporting the activation of rat astrocytes induced by cholesterol exposure in vitro, resulting in enhanced APP/BACE-1 interaction within lipid-rafts, increased APP content, and enhanced ROS production [39]. Moreover, recent observations from a single cell brain atlas of human AD indicated the diversity of astrocytes in AD under the control of the transcription factor EB, a master regulator of lysosomal function, which initiated a regulatory cascade containing multiple AD GWAS genes [40]. Taken in consideration that subjects with DS display abnormal lipid metabolism from the embryonic stages [41] and that SOD1, BACE-2, S100β and APP genes are located at HSA21 [42], it is conceivable that subjects with HSA21 trisomy could have a basal cholesterol-mediated astrocyte activation. These findings along with the observed increased NPC1 expression in DS individuals (Figure 2) suggests that an enhanced hippocampal lysosomal cholesterol trafficking could result in increased mitochondrial cholesterol loading in hippocampal areas, possibly coinciding with signs of early AD-pathogenesis. Moreover, elevated astrogial-STARD1 expression in individuals with AD implies accumulation of mitochondrial cholesterol in activated astrocytes within CA1 independently of NPC1. Whether this outcome reflects that post-mortem samples from patients with AD represent a late stage of the disease masking the early intra-organelle cholesterol mobilization remains to be further investigated.

Finally, whether the increase in hippocampal Aβ₄₂ deposits reflects an impaired Aβ clearance by increased lysosomal cholesterol remains to be further confirmed in human neurodegeneration. Although the astrocyte-dependent lipidation of ApoE could contribute to the ability of glia to clear Aβ burden, this event does not seem to occur in ApoE4-carrying subjects [2], and defects in Aβ clearance have been detected in cerebrospinal fluid of >98% of LOAD cases [43], which may reflect defects in the endolysosomal/autophagy system in the early neuropathological stages of AD [44]. In line with this possibility, we have recently reported that elevated intracellular cholesterol levels in a hypercholesterolemic mouse model increased Aβ-induced autophagosome formation, but impaired the fusion with lysosomal vesicles causing deficient autophagy-dependent Aβ degradation [45]. Considering that DS might serve as a model for early AD [12], the observations of increased lysosomal cholesterol loading support the notion that abnormal cholesterol distribution in lysosomes can impact mitochondrial function through sustained mitochondrial cholesterol accumulation and Aβ clearance, emerging as an early event in AD pathogenesis.

In conclusion, this exploratory study in post-mortem brains from patients with AD and DS revealed increased mitochondrial and lysosomal cholesterol levels in hippocampus from patients with AD and DS, which correlate with higher expression of STARD1 and NPC1. Moreover, our findings indicate the ability of STARD1 expression to discriminate controls from AD in the general population and in subjects with DS and suggest that STARD1 could be a potential marker associated with early molecular events of AD pathology. Further studies with a higher number of human samples are needed to validate these findings. In addition, future research using mice with tissue-specific deletion of STARD1 are warranted to define its role in AD pathology.

Materials and Methods

Supplementary Materials

Supplementary Methods

Supplementary Figures

Supplementary Tables