Genetic reports abstractHippocampus neuronal metabolic gene expression outperforms whole tissue data in accurately predicting Alzheimer's disease progression
Introduction
Alzheimer's disease (AD) is a neurodegenerative disease that is characterized by cognitive decline and is the most common cause of dementia (www.alz.org). Because the incidence and prevalence of AD and other dementias increase with age, the number of patients is expected to grow rapidly as the population ages. Pathological hallmark abnormalities of AD are the extracellular deposits of β-amyloid peptide (plaques) and intracellular twisted strands of tau protein (tangles) in the brain (Johnson and Bailey, 2002, Masters et al., 1985). Current treatment is mainly symptomatic, and there is no treatment available to stop the deterioration of brain cells in AD. Definitive diagnosis of AD requires postmortem examination of the brain, which must contain sufficient numbers of plaques and tangles to qualify as affected by AD (Mattson, 2004). Plaques and tangles are present mainly in brain regions involved in learning, memory, and emotional behaviors, such as the entorhinal cortex, hippocampus, basal forebrain, and amygdala. Unraveling the mechanisms underlying AD and impaired brain function has been difficult because of the complexity of the cellular networks that drive these changes. At present, the only gene that has been consistently associated with sporadic cases of AD is apolipoprotein E (APOE) (Mihaescu et al., 2010). However, APOE genotyping is not considered clinically useful for screening, testing, or diagnosis of AD.
Until a definite diagnosis is confirmed neuropathologically, the diagnosis of AD is based on clinical examination and neuropsychological testing. The cognitive performance in AD subjects is assessed via the Mini-Mental State Examination (MMSE) (Folstein et al., 1975). In addition to its value as a screening test for dementia, the MMSE is often used to document cognitive changes over time in individual patients. This is an important clinical measurement because progressive cognitive loss is a characteristic of neurodegenerative dementing illnesses (Clark et al., 1999). Information on the rate of change over time is also valuable for assessing the results of therapeutic interventions, predicting the severity of cognitive decline, and planning for long-term health care.
Several gene expression studies of AD individuals have been reported so far. Blalock et al. (2004) analyzed hippocampal gene expression microarrays of control subjects and AD patients of varying severity. They tested the correlation of each gene's expression with MMSE and neurofibrillary tangle (NFT) scores. Their work revealed upregulation of many transcription factor signaling genes regulating proliferation and differentiation during AD progression, including tumor suppressors, oligodendrocyte growth factors, and protein kinase A modulators. In addition, upregulation of adhesion, apoptosis, lipid metabolism, and initial inflammation processes was reported along with downregulation of protein folding/metabolism transport as well as several energy metabolism and signaling pathways. Ray and Zhang (2009) used the aforementioned microarrays of AD patients to develop a multiple linear regression (MLR) method to find the strength of the association of each subject's NFT score (dependent variable) with the gene expression profile and the MMSE. Using MLR, they selected 500 genes that can distinguish subjects with incipient AD from healthy control subjects in two different brain regions—the hippocampus and the entorhinal cortex. Liang et al., 2008, Liang et al., 2010) profiled the gene expression in non-tangle-bearing neurons in six postmortem brain regions that are differentially affected in the brains of healthy elderly control subjects, nondemented individuals with intermediate AD neuropathology (NDAD), and AD patients. They compared the expression of 80 nuclear genes encoding mitochondrial electron transport chain subunits in the different brain regions. In a second study, they focused on genes that participate in mechanisms that have been previously implicated as being associated with AD, to assess whether these pathogenic pathways may be enacted in NDAD brains (Liang et al., 2010). These mechanisms include pathways leading to the formation of NFTs and amyloid plaques, ubiquitin–proteasomal pathways, and pathways surrounding synaptic degeneration. Indeed, significant overlapping expression changes were identified in the brains of both NDAD and AD patients compared with those of control subjects.
Following on these important analyses of gene expression in AD, we present here a microarray-based study that focuses specifically on the role of metabolic genes in the cognitive decline of this disease. Our focus has been strongly motivated by cumulative evidence demonstrating that numerous metabolic alterations may cause the impairment of brain cells' function and viability in AD. Decreases in cerebral metabolic rate (CMR) characteristically occur in AD and other dementias (Blass, 2001). Reduced CMRs for glucose and O2 are reported in many studies (Blass, 2001). Decreased activities of key enzymes in energy metabolism in brains of AD patients have also been reported in many studies. Examples for such enzymes are the cytochrome c oxidase, pyruvate dehydrogenase complex, and α-ketoglutarate dehydrogenase complex (Chandrasekaran et al., 1994, Blass, 2001). Mitochondrial function is specifically altered in AD (Wang et al., 2009). Electron microscope studies have demonstrated the accumulation of abnormal mitochondria in senile plaques in AD (Terry et al., 1964). Damage to both the components and the structure of mitochondria, as well as increased oxidative stress, has been extensively reported in AD (Zhu et al., 2006). The mitochondrial respiratory chain is one of the main sources of reactive oxygen species (ROS) (Gibson et al., 2008), resulting in oxidative damage to varied molecules (Ferrer, 2009). The overall effect assumed is a positive feedback cycle where ROS produce oxidative stress that eventually produces more ROS (Bonda et al., 2010). The nervous system is particularly susceptible to oxidative stress (Barnham et al., 2004), as neurons are extremely energy dependent and therefore particularly sensitive to changes in mitochondrial function (Su et al., 2010). Additionally, several proteins linked with metabolic reactions have been shown to be targets of oxidative damage in AD (Butterfield et al., 2010, Reed et al., 2008, Sultana et al., 2006). Finally, other neurodegenerative diseases such as Parkinson's disease (PD) and amyotrophic lateral sclerosis (ALS) also share these common features of extensive oxidative stress (Su et al., 2010), mitochondrial damage, and apoptosis processes (Reed et al., 2008). The prevalence of neurological diseases associated with mutations in mitochondrial genes underscores the important functional role of mitochondria in neuronal metabolism (Barnham et al., 2004, Butterfield et al., 2010). Finally, our work has been further inspired by the recent work of Khaitovich et al. (2008), who studied the role of metabolic genes in another major brain disease—schizophrenia. They found that genes associated with schizophrenia are heavily involved in energy metabolism of the brain. Remarkably, the expression of many of these genes has also changed rapidly during recent human evolution, leading them to suggest that the evolution of human cognitive abilities was accompanied by important adaptive changes in brain metabolism.
In light of the accumulating evidence briefly reviewed earlier in the article, the aim of the current study is to focus on metabolic dimension to identify the metabolic genes and pathways that strongly correlate with AD progression and cognitive decline in the hippocampus, specifically in hippocampal neurons. Although some prime metabolic determinants of these observables may have previously surfaced during whole-genome mRNA analysis, many important findings would have most likely been lost because of the masking effects of nonmetabolic genes and the large feature space, hence leading us to a study focused on metabolism per se. For this purpose, we trained random forest models (Breiman, 2001) for the classification of AD severity using metabolic genes solely. Additionally, we generated regression models for the prediction of the MMSE score from metabolic genes' expression. We compared the selected genes obtained from the models of the two data sets (i.e., whole hippocampus tissue vs. hippocampal neurons) to study the unique processes that occur in different cell types. Finally, we analyzed the selection forces acting on these genes in human and primate evolution to highlight mechanisms involved in the evolution of cognitive abilities.
Section snippets
Data sets
The microarray data used in this study were obtained from the Gene Expression Omnibus (GEO) site (www.ncbi.nlm.nih.gov). The first data set, containing expression data of hippocampus field CA1 neuronal genes, was taken from studies of six regions from postmortem brains (GSE5281) (Liang et al., 2008, Liang et al., 2010). The data set contains expression profiles of laser-capture microdissected non-tangle-bearing neurons from 29 subjects categorized into three groups termed: “Control,” “NDAD,”
Neuronal metabolic genes in the hippocampus predict AD severity more accurately than the whole tissue metabolic genes
To compare the influence of the disease-altered metabolism of neurons with that of the whole tissue, we decided to build classification models of neuronal gene expression from the hippocampus and of gene expression of a whole hippocampus tissue, for the prediction of AD progression. For this purpose, we first used the data set of Liang et al., 2008, Liang et al., 2010). It includes gene expression from hippocampal neurons of patients with three levels of disease severity: control, NDAD, and AD.
Conclusions
The role of metabolism in the progression of AD is studied by inspecting the gene expression of metabolic genes. These genes are shown to be as effective in predicting the severity of AD as the entire gene list. Furthermore, the higher prediction accuracy obtained with neuronal expression per se (vs. whole tissue expression) points to the importance of metabolic processes in this specific cell type. In addition, metabolic whole tissue gene expression can predict the MMSE score better than
Disclosure statement
All authors state that there are no conflicts of interest of any kind. There are no contracts relating to this research. There are no any agreements of the authors or their institutions that could be seen as involving a financial interest in this work.
The data contained in the manuscript being submitted have not been previously published, have not been submitted elsewhere, and will not be submitted elsewhere while under consideration at Neurobiology of Aging.
All authors have reviewed the
Acknowledgements
The authors thank the members of the Ruppin laboratory and especially Ori Folger for helpful discussions. E.R. acknowledges the generous support provided by grants from the Israeli Science Foundation (ISF) to this research. S.S. gratefully acknowledges the support of the Josef Sagol Fellowship for brain research at Tel Aviv University. Y.Y.W. was supported in part by Eshkol (the Israeli Ministry of Science and Technology) and Dan David fellowships.
References (46)
- et al.
Oxidative stress in Alzheimer disease: a possibility for prevention
Neuropharmacology
(2010) - et al.
Impairment in mitochondrial cytochrome oxidase gene expression in Alzheimer disease
Brain Res. Mol. Brain Res
(1994) A theory of the evolutionary origins of psychosis
Eur. Neuropsychopharmacol
(1995)- et al.
The evolution of Alzheimer disease, the reproductive schedule, and apoE isoforms
Neurobiol. Aging
(1999) - et al.
“Mini-mental state”A practical method for grading the cognitive state of patients for the clinician
J. Psychiatr. Res
(1975) - et al.
Gene expression levels assessed by CA1 pyramidal neuron and regional hippocampal dissections in Alzheimer's disease
Neurobiol. Dis
(2012) - et al.
Single-cell molecular biology: implications for diagnosis and treatment of neurologic disease
Biol. Psychiatry
(2003) - et al.
Altered p59Fyn kinase expression accompanies disease progression in Alzheimer's disease: implications for its functional role
Neurobiol. Aging
(2005) - et al.
Positive selection on gene expression in the human brain
Curr. Biol.
(2006) - et al.
Neuronal gene expression in non-demented individuals with intermediate Alzheimer's Disease neuropathology
Neurobiol. Aging
(2010)