Background
Alzheimer’s disease (AD) is characterized neuropathologically by the presence of senile plaques with extracellular aggregation of amyloid-β (Aβ) and tau neurofibrillary tangles [
1]. In addition to Aβ and tau, neuroinflammation and oxidative stress reactions are observed in AD pathogenesis [
2]. Mitochondria are responsible for not only energy supply of adenosine triphosphate (ATP) but also main intracellular source of reactive oxygen species (ROS) that cause cellular damage [
2,
3]. Mitochondrial dysfunction was reported to be play a pivotal role in the pathogenesis in AD [
4,
5]. Previous studies showed that Aβ directly affected mitochondrial bioenergetics and produced mitochondrial morphological change [
6]. However, our recent study did not support the theory that Aβ deposition itself accelerates mitochondrial dysfunction in AD [
7]. According to the amyloid cascade hypothesis of AD, Aβ is responsible for triggering downstream tau pathology which is closely related neurodegeneration [
8]. Tau localizes predominantly in axons and contributes to the axonal transport [
9]; the accumulation of pathological tau affects mitochondrial transport and causes mitochondrial dysfunction [
5,
9]. Therefore, depicting mitochondrial dysfunction might be a useful marker to elucidate the effects of tau pathology on neuronal function. In turn, mitochondrial dysfunction leads to neuronal degeneration were reported to be associated with ROS [
3,
10]. Oxidative stress contributes to tau phosphorylation and formation of neurofibrillary tangles [
11]. Our recent work in tau transgenic mice (rTg4510 TauTg) showed a significance of tau pathology for mitochondrial dysfunction [
12]. Thus, an in vivo study in the clinical setting is needed to verify the relationships between tau pathology and mitochondrial dysfunction in the living brains of patients with AD.
[
18F]BCPP-EF is a newly developed PET tracer which binds to mitochondrial complex I (MC-I) [
13]. In the electron transport chain (ETC) in mitochondria, MC-I is the first and rate-limiting enzyme required for ATP production and is a site of ROS production [
14,
15]. [
18F]BCPP-EF also permits the investigation of the topographical distribution of mitochondrial dysfunction, as well as relationships with tau-PET parametric changes. The aim of this study is to examine the relationship between MC-I availability, Aβ and tau deposition, and their influence on cognitive decline in mild AD patients using PET. We predict a pathophysiological association between tau pathology and MC-I availability in the clinical setting.
Discussion
The aim of this study is to investigate the relationship of mitochondrial activity with tau pathology and its influence on clinical symptoms in mild AD dementia. The present study demonstrated decreased levels of MC-I availability in the temporal, parietal and frontal cortices, and significant increased level of tau load in the temporo-parietal and frontal cortices. Within the trans-entorhinal, entorhinal and hippocampal cortices (Braak stage I-II), levels of mitochondrial availability displayed a negative association with levels of tau aggregation, suggesting that intracellular and partly extracellular aggregation of tau is an important detrimental event for neuronal failure caused by MC-I availability in AD. No correlation was observed between levels of MC-I availability and Aβ accumulation. Altered MC-I availability in the trans-entorhinal and entorhinal region reflects a loss of normal homeostasis in neurons as an early biomarker of pathophysiology in AD.
It is increasingly recognized that mitochondrial dysfunction occurs in earlier disease stages [
7,
10,
23] in AD. The present finding of mutual relation between mitochondrial dysfunction and tau pathology in AD is consistent with previous reports [
7,
23], suggesting the coexistence of mitochondrial dysfunction and tau pathology. Because of this study’s cross-sectional design, it was not possible to infer whether mitochondrial dysfunction or these abnormal protein depositions occurs first. Since these analyses were exploratory, further study with multiple comparison will be needed. However, when the MC-I availability as measured with [
18F]BCPP-EF SUVR is plotted against tau load measured with [
11C]PBB3 BP
ND, the results provided the significant negative correlation in the ROI of Braak stage I-II. In addition, we confirmed altered MC-I availability and tau pathology in the Braak stage I-II ROI in AD, although the standard deviation was relatively large. According to the amyloid-β cascade hypothesis [
33,
34], Aβ triggers tau pathology which is accompanied by subsequent neurodegeneration [
8]. According to the Braak’s stage of tau distribution, pathological tau deposition initially occurs in the medial temporal area, followed by the frontal and lateral temporal area, and later spread to the primary motor and sensory areas [
28,
33]. Tau pathology sequentially induces neurodegeneration, and greater tau load is related to longer disease duration [
33].
Medial temporal area covering trans-entorhinal and entorhinal cortex are particularly vulnerable to AD-related pathological changes such as tau accumulation [
28,
35]. Thus, it is likely that tau aggregation in AD constitutes an important detrimental event, imposing energy failure in neurons and neurodegeneration. Negligible associations between [
18F]BCPP-EF with [
11C] PiB suggested that mitochondrial damage in the trans-entorhinal and entorhinal regions is not directly linked directly to Aβ deposition in mild AD. In addition, the absence of association between reduced [
18F]BCPP-EF with increased [
11C] PiB may indicate that MC-I dysfunction is late and not early in the AD disease process. However, it is important to note that [
11C] PiB SUVR is not a strong correlate of Aβ oligomers, toxic Aβ species and it is plausible that no correlation between MC-I uptake and the reactive or “benign” accumulation of Aβ plaques that happen with aging and are accelerated in AD.
[
18F]BCPP-EF SUVR in Braak stage I-II area covering trans-entorhinal cortex and entorhinal cortex was positively correlated with the WMSR-LM score that reflects logical memory function [
36]. Memory impairments are referred to as the early clinical manifestation of AD [
37]. Correspondingly, in vivo imaging showed that tau pathology appears early in the medial temporal lobe, with medial temporal lobe atrophy and memory impairments continuing in AD [
28]. Our result of reduced mitochondrial activity correlating with memory decline gives an additional important message such as a considerable contribution of an early mitochondrial dysfunction (supporting a mitochondrial cascade hypothesis [
4]) to this tau-based memory impairment. In contrast, we failed to show any relationship between [
18F]BCPP-EF SUVR in Braak stage I-II area and MMSE or FAB score. On the other hand, [
18F]BCPP-EF SUVR in Braak stage III-IV area covering limbic area including temporal cortex and insula was positively correlated with the MMSE or FAB score. The MMSE and FAB scores reflect the levels of general cognition [
22] and frontal lobe-based executive function [
38]. These findings are consistent with the present clinical evaluation that the AD patients examined in this study were rated as those with mild severity of dementia. The lack of correlation between MC-I availability and MMSE in Braak stage I-II area might be due to the fact that this correlation builds up with the spreading of the pathological process to the later stages of tau accumulation in Braak stages III-IV area. In addition, the correlation between MC-I availability and FAB in not Braak stage I-II area but Braak stages III-IV area support the notion that executive dysfunction in AD appears after memory problems [
37]. Thus, the level of [
18F]BCPP-EF binding in the trans-entorhinal cortex and entorhinal cortex can be a biomarker that predicts the memory-dominant cognitive decline in mild AD. Although our previous study indicated the association between tau deposition detected by [
11C]PBB3 BP
ND and MMSE score in the whole brain analysis [
23], our ROI analysis failed to show any relationship between [
11C]PBB3 BP
ND and neuropsychological battery scores in Braak stages area. This discrepancy might be a due to methodological differences. The relatively large size of the ROI (especially at the high Braak stages) used in this study might be an explanation for the lack of correlation.
The correlation between mitochondrial dysfunction and tau deposition was restricted to the early sites of tau accumulation, i.e. medial temporal region. In this study, patients with mild AD (CDR0.5 or 1) were recruited. Therefore, there might be a chance of positive correlation between mitochondrial dysfunction and tau deposition more widely if preclinical or more severe stage of AD were recruited. Other than the genetical reason, mitochondrial dysfunction can be caused by various factors such as neuroinflammation and misfolded proteins like Aβ and tau [
4,
34]. Once mitochondria are impaired, mitochondria-generated ROS causes neuroinflammation [
2,
34,
39], and mitochondria are in turn targeted at by generated ROS [
2,
40,
41]. Our results indicated the presence of MC-I dysfunction in the subcortical areas outside the cortical region. Subcortical area such as thalamus and basal ganglia were reported to be one of the brain areas showing mitochondrial dysfunction, which were associated with wide neuronal loss and synaptic alternations [
42]. Future studies are needed to know how neuroinflammation contributes to the exacerbation of the brain milieu in a vicious cycle by tau aggregation and mitochondrial dysfunction [
34].
There are several limitations that must be considered when interpreting this study. The sample size was small. Although there was no significant difference of age between AD patients and controls subjects, the relatively wide age range and high standard deviation of both groups might be a confounding factor. Hence, it might be more appropriate to consider many factors affecting the brain functions and morphology such as age, education, diet etc. as confounding covariates. Although the setting of ROIs was completely automatic, any automatic manipulation would suffer more or less misplacement of ROIs even in spatially-normalized images (which sometimes accompanies distortion), which might cause an interpretation bias in this study. However, this possible bias is a universal problem to be overcome in in vivo imaging studies even using spatially normalized images. To reduce the criticism, we are referring to the widely-accepted atlas. Previous studies indicated that the first-generation of tau tracers such as [
11C]PBB3 showed off-target binding in the basal ganglia, thalamus, choroid plexus, and venous sinus [
43]. Therefore, we need to be cautious about describing any increase in [
11C]PBB3 BP
ND in these areas. The spill-over radioactivity might affect the signal. The localization of the choroid plexus in the lateral ventricles, in close proximity to the hippocampus could lead to problematic quantification of the tracer retention and asymmetric fixation of [
11C]PBB3 [
44]. Additional studies with other tau PET tracer are needed to confirm our result [
45]. In addition, an association between mitochondrial dysfunction and tau deposition could only be examined in AD patients at relatively mild clinical stage, and the utility of MC-I PET in predementia AD is not known. Despite this preliminary condition, the present information about within subject changes in these pathophysiological biomarkers would help understand the prognostic use of mitochondrial dysfunction as a biomarker in mild AD. To confirm this, further studies with large sample size and several disease stages of AD patients are needed by focusing changes in the associations between mitochondrial dysfunction and misfolded proteins aggregation or neuronal damage.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.