Genome-Wide Methylation of Mild Cognitive Impairment in Mexican Americans Highlights Genes Involved in Synaptic Transport,Alzheimer’s Disease-Precursor Phenotypes,and Metabolic Morbidities

Abstract

The Mexican American population is among the fastest growing aging population and has a younger onset of cognitive decline. This group is also heavily burdened with metabolic conditions such as hypertension, diabetes, and obesity. Unfortunately, limited research has been conducted in this group. Understanding methylation alterations, which are influenced by both genetic and lifestyle factors, is key to identifying and addressing the root cause for mild cognitive impairment, a clinical precursor for dementia. We conducted an epigenome-wide association study on a community-based Mexican American population using the Illumina EPIC array. Following rigorous quality control measures, we identified 10 CpG sites to be differentially methylated between normal controls and individuals with mild cognitive impairment annotated to PKIB, KLHL29, SEPT9, OR2C3, CPLX3, BCL2L2-PABPN1, and CCNY. We found four regions to be differentially methylated in TMEM232, SLC17A8, ALOX12, and SEPT8. Functional gene-set analysis identified four gene-sets, RIN3, SPEG, CTSG, and UBE2L3, as significant. The gene ontology and pathway analyses point to neuronal cell death, metabolic dysfunction, and inflammatory processes. We found 1,450 processes to be enriched using empirical Bayes gene-set enrichment. In conclusion, the functional overlap of differentially methylated genes associated with cognitive impairment in Mexican Americans implies cross-talk between metabolically-instigated systemic inflammation and disruption of synaptic vesicular transport.

Keywords

Alzheimer’s disease cognitive dysfunction epigenetics Mexican Americans SEPT8 protein

INTRODUCTION

Alzheimer’s disease (AD) affects 5.8 million Americans and is currently the sixth most common cause of death [1]. AD is a neurodegenerative disease that manifests as a result of intracellular tau tangles and extracellular amyloid-β plaques within the brain, causing cellular inflammation and neuronal loss, which lead to changes in memory and personality [1]. Mild cognitive impairment (MCI), a common precursor to AD, affects over a quarter of individuals over the age of 65, with a third of these individuals developing AD later in life [2].

Within the United States, the Mexican American elderly population is experiencing rapid growth and is predicted to increase 6-fold by 2050 [3]. This population has an earlier age of AD onset [4] and are typically diagnosed at more advanced stages of disease relative to non-Hispanic whites [3, 4]. Further, Mexican Americans are disproportionately burdened with comorbid conditions such as depression [5, 6], cardiovascular disease [7], and metabolic conditions (e.g., metabolic syndrome, type 2 diabetes, obesity) [4, 7]. Other factors such as education and socio-economic status have also been suggested to play a role, where fewer years of education and lower economic status are known to increase one’s risk for AD [8].

There are several critical differences in the etiology of AD when comparing Mexican Americans and non-Hispanic whites, and these distinctions may be key to understanding the observed disparities in the population. Mexican Americans are less likely to carry the APOE ɛ4 allele which is the one of the highest predisposing genetic factors for AD among non-Hispanic whites [4, 8]. Additionally, while a vascular/inflammatory phenotype predominates risk for cognitive decline among non-Hispanic whites, risk among Mexican Americans is due primarily to metabolic factors [9]. In a study of serum biomarkers from 363 Mexican Americans (AD = 49, NC = 314), the biomarker profile of AD among Mexican Americans was weighted heavily for metabolic markers as opposed to the inflammatory-laden profile that has previously been observed among non-Hispanic whites [10].

Cognitive impairment is highly heterogeneous, with both environmental/lifestyle and genetic factors playing a role. Since AD pathology can initiate up to 20 years before the first visible symptoms [1], changes in methylation patterns may signify relevant early molecular responses to the disease. Methylation at CpG (cytosine-phosphate-guanine) sites throughout the genome is known to affect downstream expression of various genes and has been implicated in normal aging and disease processes [11, 12]. Differential methylation relating to AD and MCI has already been reported in several populations [11 –14]. While metabolic stress detected in the peripheral blood is associated with cognitive impairment, the exact mechanisms are unknown [15]. Wang et al. have identified epigenetic changes in the brain of AD patients that exacerbate synaptic plasticity within functional networks overlapping multiple brain regions [16]. Moreover, changes in DNA methylation patterns have been shown to be highly correlated between peripheral blood and brain [17]. Since metabolic burden affects both the peripheral and central nervous system [18], detecting epigenetic signatures may help identify risk regions that overlap metabolic burden and cognitive dysfunction.

Here we sought to investigate genome-wide differential methylation at site- and region-specific levels to identify epigenetic factors that may confer risk for cognitive impairment in Mexican Americans.

METHODS AND MATERIALS

Samples and cohort design

This study was approved under the North Texas Regional IRB #2012–083. All participants provided written consent. Peripheral blood buffy coat of fasted-state Mexican American participants (n = 90) enrolled in the Health & Aging Brains of Latino Elders (HABLE) study was obtained.

Each participant underwent an interview (i.e., medical history, medications, health behaviors), neuropsychological testing, blood draw, and medical examination. Global cognition was assessed via the Mini-Mental State Examination (MMSE) [19] and disease severity rated according to the Clinical Dementia Rating scale [20] sum of boxes scores (CDR-SB) [21, 22]. An informant interview was conducted for each research participant to obtain information regarding his/her activities of daily living (basic and instrumental). All information was presented at a weekly consensus review conference with diagnoses of AD [23] and MCI [24] assigned according to published criteria. Cognitively normal control (NC) participants performed within normal limits on psychometric assessment [25].

Phenotypic differences between the two groups (Normal Controls & Mild Cognitive Impairment) were tested for two-tailed significance using independent t-test for continuous variables, and chi-square test for categorical variables (Table 1). Participants were matched on sex and age. Peripheral blood samples were handled in accordance to the UNTHSC Institutional Biosafety Committee approved protocol IBC-2018-0078. Forty-five individuals (both male and female) diagnosed with MCI (based on a battery of neuropsychological tests) were selected. Forty-five age and sex-matched healthy individuals were then chosen to complete the study population. Metabolic risk score (0–5) was calculated for each individual following adapted International Diabetes Federation (IDF) guidelines based on obesity, elevated triglycerides, reduced HDL cholesterol, hypertension diagnosis, and high fasting glucose [26]. Waist circumference was used to diagnose obesity, with 35.4 inches being the cut-off value for Mexican American adults [27]. Missing data was handled according to Masconi et al., 2015 [28].

Table 1

Demographic characteristics for the sample cohort

	Normal Controls N = 45	Mild Cognitive Impairment N = 45	p
Mean±SD
Age	63.49±7.20	63.53±7.16	0.9790
MMSE	26.16±2.94	23.87±3.16	0.0006
Education	8.60±4.35	6.11±3.79	0.0048
Metabolic Score	2.87±1.41	3.36±1.13	0.0723
(IDF)
N (%)
Sex
Female	32 (71%)	31(69%)	1.00
Male	13 (29%)	14 (31%)
APOE Status
3/3	30 (67%)	29 (64%)	0.488
3/4	12 (27%)	9 (20%)
2/3	3 (7%)	6 (13%)
4/4	0 (0%)	1 (2%)

DNA extraction and quantification

DNA was extracted from peripheral blood buffy coat using the MagBind^® Blood and Tissue DNA HDQ Kit (Omega Bio-tek, Norcross, GA) on a Microlab STAR liquid handling system. DNA concentrations were quantified using Qubit^®4 dsDNA BR Assay Kits with the Qubit^® Fluorometer (Thermo Fisher Scientific Inc.) and the NanoDrop^® spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA). For downstream applications, DNA extracts with concentrations ≥10 ng/μL were considered sufficient due to the manufacturer’s requirement of 200 ng of input DNA for both methylation and genotyping arrays. Low yield DNA extracts were concentrated using Microcon^® DNA Fast Flow Filters (Sigma Aldrich, St. Louis, MO). Any extracts that remained <10 ng/mL following concentration, were excluded from downstream processing.

Single nucleotide polymorphism (SNP) genotyping

All 90 subjects were genotyped on the Infinium HTS Global Screening Array v.2 (Illumina, San Diego, CA) following manufacturer’s instructions. Intensity data files were imported into Genome Studio^®. All SNPs with call rates <0.99 were excluded from analysis. Remaining SNPs were QC’d in PLINK; SNPs with > 5% missingness were removed (–geno) [29]. The QC’d SNP dataset was used to identify presence of population substructure using ADMIXTURE [30]. The cross-validation error rate was lowest for k of 1, concluding that the population did not have any sub-clusters (Supplementary Figure 1).

DNA methylation array assay

DNA methylation levels were analyzed for all 90 participants. DNA was bisulfite converted using the EZ DNA Methylation^TM kit (Zymo Research, Irvine, CA). Bisulfite converted DNA was then processed on the Infinium HD MethylationEPIC Array (Illumina) following manufacturer’s guidelines. The array has an exhaustive coverage of more than 850,000 CpG sites and is highly correlated with Infinium HumanMethylation450 BeadChip and whole genome bisulfite sequencing.

Intensity data (IDAT) files were imported into Genome Studio^® and technical replicates were compared to determine consistency in array processing. Technical replicates all displayed R² values greater than the 0.98 suggested threshold (Illumina).

Data analysis and visualization

Raw IDAT files were processed using the ‘ChAMP’ Bioconductor package [31] in R, which converts fluorescent intensities into beta values (ratio of methylated:unmethylated) ranging from 0 to 1 for each individual probe. Data was normalized using the BMIQ method, followed by inspection for batch effects using singular value decomposition. Both the lowest (i.e., sample plate) and highest (i.e., array) stages in the experimental workflow were corrected for using ComBat. Due to the consistent placement of technical replicates on the beadchips, one of each replicate pair was removed from analysis allowing for further correction of variance at the array level. The final dataset for analysis after QC included 738,919 CpG sites from 90 individuals. An association test to identify significantly differentially methylated probes was performed using M-values (transformed beta-values) and adjusted for variation in proportions of the cell types lymphocytes, monocytes, and neutrophils, using limma [32]. Differentially methylated probes were considered significant based on multiple testing corrected p-value <0.05. Differentially methylated regions were analyzed using DMRcate [33]. Functionally differentially methylated modules were also analyzed using normalized, batch-corrected beta-values using FEM package [34] built-in ChAMP.

The significant genes identified were analyzed for gene ontology enrichment using empirical Bayes gene set enrichment analysis (GSEA; built-in ChAMP) [35] and the ShinyGO [36] tool set for biological processes, cellular components and molecular functions at an FDR <0.05. Visualizations were created using Gviz [37] and ggplot2 [38] in R. Candidate genes were analyzed for pathway enrichment, and subsequent figure was generated through the use of IPA (QIAGEN Inc., https://www.qiagenbio-informatics.com/products/ingenuity-pathway-analysis software).

RESULTS

Study characteristics

The cohort participants did not differ on demographic characteristics: age, sex, metabolic risk score, and APOE status (Table 1). The metabolic risk score was derived as per IDF guidelines [26] wherein the presence of hypertension, waist circumference, fasting glucose, triglycerides, and HDL levels is evaluated. Both groups had two or more metabolic risk factors.

Our study cohort is representative of the Mexican American population, which often 1) is metabolically burdened (as reflected by high metabolic risk scores), 2) has fewer years of education (on average) [4, 39], and 3) has a lower frequency of the APOE ɛ4 allele. The most prevalent APOE status within our cohort was ɛ3/ɛ3, with a single individual being ɛ4/ɛ4; this was optimal for our study as the frequency and risk effect of the APOE ɛ4 allele is often less in Hispanic populations [40]. This is consistent with other studies reporting differences in APOE ɛ4 frequency (approximately 20–25%) in Mexican American population [41, 42]. MMSE scores were lower for both normal and MCI participants than usually observed in individuals of European ancestry [4]. The lower MMSE scores and number of years of education are correlated and these differences have been reported by other research groups and confirms the disparity in ethnic populations [8, 39].

Differentially methylated sites and regions

Genome-wide methylation profiles were generated using the Illumina MethylationEPIC BeadChip, a microarray-based method which assesses over 850,000 cytosine methylation sites (CpG sites) throughout the genome. These sites are enriched for enhancer regions but also include gene bodies, promoters and exons boundaries. After QC and normalization, the transformed beta values (identifying percentage of methylation) were significant (FDR p-value <0.05) for 10 CpG sites when comparing normal and MCI individuals. The differentially methylated probes (DMPs) identified were annotated to genes using UCSC’s reference gene name (Supplementary File 1). Four of these sites were hypomethylated among participants with MCI compared to normal controls: cg25016219 (KLHL29), cg26479998 (SEPT9), cg02586267 (not mapped to a gene), and cg18978297 (CPLX3). Six CpG sites were found to be hypermethylated in individuals with MCI: cg22360048 (PKIB), cg20904111 (intergenic), cg05917713 (BCL2L2-PABPN1), cg20201669 (OR2C3), cg14179796 (CCNY), and cg22327037 (intergenic).

We further tested the CpG sites using the DMRcate package [33] which maps to hg19 coordinates and clusters significant sites into regions. The genome-wide region analysis yielded four differentially methylated regions (DMRs): two on chromosome 5 located within TMEM232 and SEPT8, one on chromosome 12 within SLC17A8, and one on chromosome 17 within ALOX12 (Supplementary File 2). These four regions comprised of 41 significant CpG sites that were located primarily within islands, with some being located in open sea and shore regions (Fig. 2). CpG islands are 0.5–2 kb sequences with higher GC content located in the promoter region and methylation of these sites is believed to affect gene expression. Surrounding regions within the 2 kb of islands are the CpG shores, and regions which are further away from the island are considered open sea [43, 44].

Fig.1

(A) Manhattan plot highlighting differentially methylated probes. The annotated genes using UCSC RefSeq are shown above the CpG ids. (B) Scatter plots. The distribution of beta intensity for each of the FDR significant CpG probe is shown between the NC (normal control) and MCI (mild cognitive impairment).

Fig.2

A) Differentially methylated regions. An ideogram representation of the location of differentially methylated regions and corresponding genes, number of CpGs in the regions and FDR p-value of the region. B) Classification of CpGs features. An overview of all the CpGs sites in the identified regions based on their feature type-CpG island are sites located in 0.5 kb stretches with high GC content; CpG shores are located with ±2 kb of the island and CpG opensea are intergenic sites in the genome. C) Dendrogram plot. Beta values for the CpG sites within the differentially methylated regions are represented as shades of grey. The x axis is the chromosome and cpg site location (bp) within each region. The y axis is study participants clustered based on similarity of beta values of the cpg sites (See Supplementary File 2 for details).

Differentially methylated modules

We next analyzed differential methylation by clustering CpG sites into known gene modules of protein-protein interactions (PPI) from pathway commons database. The methylated CpG sites are mapped to genes and weighted based on the distance to transcription start sites. These genes, driven by weighted averages of methylation levels of their CpG sites, are evaluated for association with the phenotype. The protein-protein interaction network with significant genes are considered functionally methylated modules [34]. The functional methylation analysis identified four gene-sets (modules) to be significantly hypomethylated in MCI compared to NC: RIN3 (also known as SLC24A4), SPEG, CTSG, and UBE2L3 (Supplementary File 3). The RIN3 and SPEG cluster identified overlapping significant genes: RIN3, SPEG, XDH, and KNDC1; while in the CTSG cluster, we found CTSG, F2RL1, and MMRN1 to be significant. Lastly, in the UBE2L3 gene-set, we found RNF144A, UBE2L3, and NEDD4L to be significant (Fig. 3).

Fig.3

Differential methylated gene modules. The differentially methylated genes within the set are shown in yellow/blue on the legend; yellow signifies hypomethylated and blue signifies hypermethylated. Each of the module was found to be significant based on its overall p < 0.05. Individual significance of each gene (p < 0.05) in the module is highlighted with blue asterisk.

Gene ontology and pathway enrichment test

Genes annotated from significant differentially methylated CpG sites, regions, and modules were analyzed for their involvement in pathways and processes. Several biological processes were found to be significantly enriched, which clustered primarily on immunological processes and systemic arterial blood pressure (Fig. 4). Four KEGG pathways—synaptic vesicle cycle, bacterial invasion of epithelial cells, inflammatory mediator regulation of TRP, and ubiquitin mediated proteolysis—were found to be enriched. Using IPA to assess enriched disease pathways, we found neuronal cell death and metabolic disease (insulin-resistance) among the most significant pathways (Fig. 5), in addition to the aforementioned processes.

Fig.4

GO enrichment results for biological processes. Significant processes are shown in the bar plot. The y-axis shows names of the processes, and x-axis shows number of genes, and each observation is labelled with the number of genes present in the respective gene category. The color of bars corresponds to the scale of FDR corrected p-values.

Fig.5

Pathway enrichment. Genes that were identified for neuronal cell death pathway are highlighted in purple and genes identified for metabolic dysfunction/insulin resistance are highlighted in blue.

Bayes gene-set enrichment analysis

Traditional GSEA methods, as reported above, use singular CpG sites/regions which represent genes for enrichment in the network. The empirical Bayes method [35] for gene set enrichment of differential methylation overcomes this singularity by analyzing average methylation of all CpG sites (i.e., untransformed beta-values) mapped to a gene. These genes are ranked based on degree of differential methylation. The ranked list is then tested for biological enrichment using the annotated gene sets in the Molecular Signatures Database (MsigDB, containing 17,810 gene sets).

We found a total of 1,450 gene-sets (8.14% of MsigDB sets) to be significantly enriched (adjusted FDR p-value < 0.05). The gene sets were sorted by area under curve (AUC) and significance.

The gene sets with the highest classification accuracy (AUC ∼72–78%) were attributed to the following gene sets, among others: Regulation of protein, Rho GTPase activity, and Aging (Table 2 in Supplementary File 5).

The gene-sets with the highest statistical significance were from experimentally-derived datasets, previously associated with H3K27 methylation in various experimental models (Table 1 in Supplementary File 5). The top gene sets were: ‘BENPORATH_ES_WITH_H3K27ME3’ (AUC – 60%; p-value: 5.1e-27), ‘MEISSNER_BRAIN_HCP_WITH_H3K4ME3_AND_H3K2 7ME3’ (AUC – 59.3%; p-value: 7.5e-21), and ‘BENPORATH_PRC2_TARGETS’ (AUC – 61%; p-value: 2.76e-19). The experimentally-derived gene sets contained ∼1,100 genes, and the top three gene set members were analyzed using the Panther database [45] for representative pathways in order to understand their biological relevance (Supplementary File 5). We found four pathways to be common in all three gene-sets: 1) Alzheimer disease-presenilin pathway, 2) Ionotropic glutamate receptor pathway, 3) Heterotrimeric G-protein signaling pathway, and 4) Cadherin signaling pathway (Fig. 6).

Fig.6

Visual summary of identified genes and pathways. Visual representation of methodology and an overview of findings converging on phenotypes and pathways. The figure shows known characteristics of the population, followed by levels of analysis performed for identification of differential methylation with cognitive impairment. The Venn diagram highlights the genes involved in key processes based on research studies reported in the discussion section. The gene enrichment was performed on differentially methylated probes, regions and modules. The top three gene sets identified using Bayes gene enrichment are sorted by statistical significance (left) and AUC (right) with corresponding details for each is highlighted in the box.

DISCUSSION

The etiology of MCI/AD in Mexicans-Americans is distinctive from non-Hispanic whites. Much of this disparity has been attributed to the elevated prevalence of metabolic dysfunction within this population, with 43% of individuals with dementia also suffering from diabetes, stroke, or both [8]. While several studies have reported an association between metabolic burden and cognitive impairment, the exact mechanisms linking the two are poorly understood [46]. In the last 20 years, based on PubMed query, we found only 71 articles that were published in the genetics of AD in Mexican-American population. There are no publications listed when we replace the term genomics and its associated terms with ‘methylation’ or ‘epigenetics’ (Supplementary Figure 2). Studying alterations in genome-wide methylation profiles associated with cognitive decline in a metabolically burdened admixed population may help identify changes that occur as a result of both genetic and lifestyle factors. Understanding these factors will be vital for development of effective diagnostic and therapeutic strategies.

Genes implicated in differentially methylated sites and regions

Here we found 10 differentially methylated CpG sites significantly associated with MCI in our Mexican American cohort. The hypomethylated sites were in genes KLHL29, SEPT9, and CPLX3. KLHL29 has been reported to be associated with diabetes [47], while other members of the Kelch-like gene family have been implicated in hypertension [48]. Septin 9 (SEPT9), known to be involved in cell cycle regulation, has primarily been associated with neuralgic amyotrophy, a peripheral nervous system disorder characterized by weakness and atrophy of the upper extremities [49]. SEPT9 variants have also been associated with systolic blood pressure [50]. The hypomethylation of SEPT9 observed in our analysis was also reported by Dayeh and colleagues, who found hypomethylation of SEPT9 on multiple CpG sites and an upregulation of SEPT9 in human pancreatic islets from type 2 diabetes donors [51]. Complexin 3 (CPLX3) is involved in regulation of SNARE-associated synaptic vesicle fusion, thereby impacting neurotransmitter transporter activity and localization [52]. The CpG site within CPLX3 was found to be hypomethylated in our study based on peripheral blood, and Annese et. al have reported CPLX3 to be upregulated in the hippocampal region of individuals with late-onset AD [53].

The hypermethylated CpG sites were in genes PKIB, BCL2L2-PABPN1, OR2C3, and CCNY. PKIB is a kinase inhibitor in PKA signaling known to suppress insulin secretion under chronic hyperglycemia [54]. Interestingly, PKA signaling also regulates mitochondrial function/morphology as well as neuronal survival, and has been suggested as a therapeutic target for treatment of neurodegenerative diseases [55]. BCL2L2 (BCL-w or BCL2-w), encoding an anti-apoptotic protein, was hypermethylated in our study, and has been reported to be downregulated in AD brains afflicted with increased phosphorylation of tau proteins [56]. Furthermore, variants in BCL2L2-PABPN1 have been reported to be associated with basal metabolic rate and body mass index in obese Korean women [57]. Mutations in PABPN1 encoding a poly A binding protein, have been associated with increased protein aggregation, a pathological hallmark of many neurodegenerative diseases [58]. Olfactory deficits have been reported to precede clinical symptoms of AD and MCI [59], and our study identified hypermethylation of OR2C3, encoding a GPCR that binds to the extracellular domain that generates signal to the olfactory sensory neurons in the brain [60]. OR2C3 binding results in additional activation of Akt, MAPK, and Rho-signaling pathways [60]. Therefore, it is plausible that hypermethylation of sites within OR2C3 may cause downregulation of its expression resulting in olfactory deficits and neuronal dysregulation [61]. The CpG site within Cyclin Y (CCNY) was observed to be hypermethylated in our study; downregulation of CCNY has been reported to inhibit the formation of new synapses during synaptic remodeling [62]. CCNY also regulates Wnt-signaling pathways, which have been found to be altered in the prefrontal cortex of individuals with AD [63].

We also identified four regions within TMEM232, SEPT8, SLC17A8, and ALOX12 that were significantly differentially methylated. ALOX12, arachidonate 12-oxidoreductase, is an enzyme involved in the generation of lipid metabolites and loss of function mutants of ALOX12 (mouse model) have demonstrated sensitization of pancreatic beta cells to oxidative stress [64] leading to declines in insulin secretion, an event that precedes onset of type 2 diabetes [65]. Interestingly, ALOX12 has also been reported to modulate glutamate-induced degeneration of neurons [66], and SNPs within this gene have been implicated in loss of prefrontal cortical thickness in individuals with posttraumatic stress disorder [67]. Vesicular glutamate transporter 3 (otherwise known as VGLUT3 or SLC17A8) has been shown to be co-expressed with glutamate receptor GABRG3 in peripheral glutamatergic neurons; together they are suspected to play a compensatory role in response to loss of glycinergic activity which occurs during hearing loss [68], a precursor symptom to MCI and AD [69]. Another member of the Septin gene family, SEPT8, was found here to be associated with MCI. SEPT8 is believed to regulate the transportation of synaptic vesicles through interaction with different SNARE-complex components [70]. Dysfunction of synaptic vesicle transmission in hippocampal and cortical regions has been associated with cognitive decline [71]. Additionally, multiple variants in TMEM232, encoding a transmembrane protein, have been implicated in various neuropathies [72]. SNPs in this gene have also been associated with schizophrenia, depression, and bipolar disorder in a Chinese population [73].

Genes implicated in differential methylated modules

We found four gene-sets in our study, RIN3, SPEG, CTSG, and UBE2L3, to be significantly hypomethylated in individuals with MCI. Four genes, RIN3, SPEG, XDH, and KNDC1, were hypomethylated in both RIN3 and SPEG gene-sets. Hypomethylation of CpG sites in the 3’UTR of RIN3 has also been observed in peripheral blood of individuals with sporadic early onset AD [74]. Interestingly, in our study, the functional module analysis which weights location of CpG sites from transcription start sites in an overall gene-based analysis, also identified hypomethylation of RIN3 derived from peripheral blood of individuals with MCI, suggesting that hypomethylation at this site may serve as an early indicator of cognitive decline. Due to its involvement in regulation of endocytic trafficking and processing of amyloid-β protein precursor [75], hypomethylation of RIN3 may also serve as a compensatory response to amyloid toxicity. This postulation is supported by Boden et al. who observed increased hypomethylation of RIN3 in AD brains in comparison to blood, which they suggest to be due to prolonged exposure of the brain to an amyloid-rich environment [74]. In a study by Stage et al., a single SNP within RIN3 (rs10498633) was found to be associated with gray matter density within the medial temporal lobe, a phenotypic change detectable even in early stages of cognitive impairment [76]. Other variants within RIN3 have also been associated with metabolic syndrome in Japanese individuals [77]. KNDC1, encodes a Ras-guanine nucleotide exchange factor, and Ji et al. have shown KNDC1 upregulation to stimulate p53-oxidative stress responses [78]. Other studies focusing on 5-hydroxymethylation in AD have found KNDC1 to be enriched in GO terms related to neurogenesis and signal transduction [79]. Remarkably, a genome-wide association study of 1 million individuals implicated variants in KNDC1 to be associated with ‘self-reported educational attainment’, a highly correlated risk factor for cognitive impairment [80]. Xanthine dehydrogenase (XDH), a hydroxylase involved in oxidative purine metabolism, is capable of converting to xanthine oxidase, a major producer of reactive oxygen species (ROS) [81]. Elevated ROS is a hallmark of both AD and type 2 diabetes [82]. Upregulation of XDH activity has also been demonstrated in the brain of rats with late-stage diabetes [83]. Hypomethylation of CpG sites within XDH may allow for increased expression permitting potential downstream elevation in ROS.

Cathepsin G (CTSG) is an anti-inflammatory protease found in azurophil granules of blood neutrophils that functions to breakdown pathogens and inflammatory tissues. Specifically, CTSG has been shown to play a role in abating neuroinflammation in AD through cleavage of amyloid-β [84]. The hypomethylation of CTSG observed in our study was reported to be upregulated in the hippocampus of hypertensive aged rats suggesting perhaps that neuroinflammation may bridge hypertension and cognitive dysfunction [85]. The G-protein coupled receptor F2RL1 (otherwise known as PAR2) also accelerates neurodegeneration via inflammatory mediators [86]. Interestingly, upregulation of PAR2 has been reported to be induced by alpha-synuclein [87]. PAR2 has also been shown to be upregulated in adipose tissues and is a contributor to insulin resistance and metabolic dysfunction [88], while the upregulation of MMRN1 has been associated with cognitive deficits in Parkinson’s disease [89].

The UBE2L3 gene-set revealed several ubiquitin-related proteins to be hypomethylated. UBE2L3 (E2F1) encodes a ubiquitin-conjugating enzyme, while NEDD4L and RNF144A encode E3 ubiquitin ligases. The expression of UBE2L3 is upregulated in neurons of AD brains [90] and is also increased in obesity for maintenance of metabolic homeostasis [91]. SNPs within NEDD4L have been associated with essential hypertension, a known comorbidity for AD [92]. Hypomethylation and upregulation of mRNA expression of RNF144 was reported in liver samples of individuals with type 2 diabetes [93]. Further, the interactions between these enzymes are important for maintaining synaptic plasticity via selective degradation of abnormal or misfolded proteins [94].

Shared processes and pathways of differentially methylated genes based on gene-set enrichment

Three genes, F2RL1, ALOX12, and UBE2L3, were identified in both KEGG and IPA’s knowledge base highlighting their role in inflammation, neuronal cell death, and metabolic processes. We also conducted GSEA analysis using empirical Bayes approach which clusters differentially methylated probes into genes followed by overall ranking of genes based on differential methylation and testing for enrichment of gene-sets. Several experimentally derived gene-sets stored in MSigDB were found to be enriched in our genome-wide methylation analysis. The gene members of each of the top three gene sets coincided for four pathways: Alzheimer disease-presenilin, Ionotropic glutamate receptor, Heterotrimeric G-protein signaling, and Cadherin signaling pathway.

The Alzheimer disease-presenilin pathway consists of the presenilin gamma-secretase complex which is involved in the cleavage of the amyloid-β protein which is known to form plaques in brains of AD individuals [45]. The glutamatergic receptors are involved in neurotransmission, and overexcitation of the glutamatergic synapses results in neuronal cell death [95] in AD. The Heterotrimeric G-protein pathway consists of G-protein coupled receptors [45], which are known to interact with BACE1 secretase molecule and are involved in phenotypes of vision and olfaction [96]. Lastly, the molecules in cadherin signaling such as N-cadherin[45] are required for maintaining synaptic plasticity, which is otherwise deregulated by amyloid-β and increases tau and p38 phosphorylation in AD [97].

Conclusion: Reported genes and pathways converge between cognitive dysfunction, metabolic burden, and inflammation

Our results point to modification of genes involved in metabolic, neuroinflammatory, and AD-precursor phenotypes (Fig. 6). The data presented here links peripheral metabolic dysregulation with cognitive decline. Metabolic syndrome which includes obesity, diabetes, and hypertension has been known to provoke poor cerebral blood flow, resulting in vasoconstriction and endothelial dysregulation; this leads to production of ROS and inflammation [98]. Insulin receptors are expressed in multiple brain regions, including the olfactory bulb, hypothalamus, and hippocampus, and are sensitive to changes in the peripheral system due to metabolic conditions [99]. Insulin resistance alters cAMP/PKA signaling pathways which not only induces inflammation from oxidative stress, but also affects synaptic plasticity of hippocampal neurons [100]. In the presence of hypertension, homeostatic dysfunction results in microglia-induced inflammation in the brain [101]. Metabolic syndrome–obesity induces low-grade systemic inflammation which is detected by brain [102] causing disruption of the synaptic transporter activity [103, 104]. While our findings do not incriminate any single pathway for pathogenicity of cognitive dysfunction, they highlight genes and pathways involved in cross-talk between peripheral metabolic burden and neuroinflammation. We postulate that insults sustained from metabolic dysregulation results in low grade inflammation (reported earlier by our group) [10] affecting synaptic vesicle activity which ensues cognitive deficits in Mexican Americans. Therefore, it is possible that cognitive decline in Mexican Americans is a manifestation of genetic/lifestyle factors of metabolic stress as seen here in their epigenetic profile.

There are several limitations to our study. First, our study is limited in sample size; therefore, the genes reported here should be considered as preliminary findings and should be replicated in larger cohorts. We have also tested cognitive-associated methylation changes in peripheral blood, which may not be parallel to brain-specific methylation alterations. Epigenetic variations in conjunction with RNA expression in larger cohorts of Hispanic individuals will provide much needed characterization of the pathogenesis of cognitive impairment.

Despite the limitations, we have strengthened our study design by implementing a very rigorous and conservative QC approach, correcting for multi-level batch effects, adjusting for cell composition and incorporating balanced demographic characteristics, representative of the Mexican American population. Our study has found several novel epigenetic markers associated with metabolic burden and cognitive dysfunction which contribute toward addressing the critical health disparity faced by Mexican Americans.

Footnotes

ACKNOWLEDGMENTS

The research team thanks the local Fort Worth community and participants of the Health & Aging Brain Study. Research reported in this publication was supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R01AG054073 and R56AG054073. We would also like to acknowledge the NIH – Neurobiology of Aging T32 grant AG020494 for supporting this research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Alzheimer’s Association (2019) 2019 Alzheimer’s disease facts and figures. Alzheimers Dement 15, 321–387.

Eshkoor

, Hamid

, Mun

, Ng

(2015) Mild cognitive impairment and its management in older people. Clin Interv Aging 10, 687–693.

Gonyea

, López

, Velásquez

(2016) The effectiveness of a culturally sensitive cognitive behavioral group intervention for Latino Alzheimer’s caregivers. Gerontologist 56, 292–302.

O’Bryant

, Johnson

, Balldin

, Edwards

, Barber

, Williams

, Devous

, Cushings

, Knebl

, Hall

(2013) Characterization of Mexican Americans with mild cognitive impairment and Alzheimer’s disease. J Alzheimers Dis 33, 373–379.

Johnson

, Gamboa

, Vintimilla

, Cheatwood

, Grant

, Trivedi

, Edwards

, Hall

, O’Bryant

(2015) Comorbid depression and diabetes as a risk for mild cognitive impairment and Alzheimer’s disease in elderly Mexican Americans. J Alzheimers Dis 47, 129–136.

Johnson

, Dong

, Vijg

, Suh

(2015) Genetic evidence for common pathways in human age-related diseases. Aging Cell 14, 809–817.

Johnson

, Large

, Izurieta Munoz

, Hall

, O’Bryant

(2019) Vascular depression and cognition in Mexican Americans. Dement Geriatr Cogn Disord 47, 68–78.

Vega

, Cabrera

, Wygant

, Velez-Ortiz

, Counts

(2017) Alzheimer’s disease in the Latino community: Intersection of genetics and social determinants of health. J Alzheimers Dis 58, 979–992.

Edwards

, Hall

, Williams

, Johnson

, O’Bryant

(2016) Molecular markers of amnestic mild cognitive impairment among Mexican Americans. J Alzheimers Dis 49, 221–228.

10.

O’Bryant

, Xiao

, Edwards

, Devous

, Gupta

, Martins

, Zhang

, Barber

(2013) Biomarkers of Alzheimer’s disease among Mexican Americans. J Alzheimers Dis 34, 841–849.

11.

Horvath

(2013) DNA methylation age of human tissues and cell types. Genome Biol 14, R115.

12.

Qazi

, Quan

, Mir

, Qing

(2018) Epigenetics in Alzheimer’s disease: Perspective of DNA methylation. Mol Neurobiol 55, 1026–1044.

13.

Wang

, Oelze

, Schumacher

(2008) Age-specific epigenetic drift in late-onset Alzheimer’s disease. PLoS One 3, e2698.

14.

De Jager

, Srivastava

, Lunnon

, Burgess

, Schalkwyk

, Yu

, Eaton

, Keenan

, Ernst

, McCabe

, Tang

, Raj

, Replogle

, Brodeur

, Gabriel

, Chai

, Younkin

, Zou

, Szyf

, Epstein

, Schneider

, Bernstein

, Meissner

, Ertekin-Taner

, Chibnik

, Kellis

, Mill

, Bennett

(2014) Alzheimer’s disease: Early alterations in brain DNA methylation at ANK1, BIN1, RHBDF2 and other loci. Nat Neurosci 17, 1156–1163.

15.

Pugazhenthi

, Qin

, Reddy

(2017) Common neurodegenerative pathways in obesity, diabetes, and Alzheimer’s disease. Biochim Biophys Acta Mol Basis Dis 1863, 1037–1045.

16.

Wang

, Gong

, Zhao

, Tang

, Varghese

, Nguyen

, Bi

, Bilski

, Begum

, Vempati

, Knable

, Ho

, Pasinetti

(2014) Epigenetic mechanisms linking diabetes and synaptic impairments. Diabetes 63, 645–654.

17.

Braun

, Han

, Hing

, Nagahama

, Gaul

, Heinzman

, Grossbach

, Close

, Dlouhy

, Howard

, Kawasaki

, Potash

, Shinozaki

(2019) Genome-wide DNA methylation comparison between live human brain and peripheral tissues within individuals. Transl Psychiatry 9, 47.

18.

Marioni

, McRae

, Bressler

, Colicino

, Hannon

, Li

, Prada

, Smith

, Trevisi

, Tsai

P-C

, Vojinovic

, Simino

, Levy

, Liu

, Mendelson

, Satizabal

, Yang

, Jhun

, Kardia

SLR

, Zhao

, Bandinelli

, Ferrucci

, Hernandez

, Singleton

, Harris

, Starr

, Kiel

, McLean

, Just

, Schwartz

, Spiro

, Vokonas

, Amin

, Ikram

, Uitterlinden

, van Meurs

JBJ

, Spector

, Steves

, Baccarelli

, Bell

, van Duijn

, Fornage

, Hsu

Y-H

, Mill

, Mosley

, Seshadri

, Deary

(2018) Meta-analysis of epigenome-wide association studies of cognitive abilities. Mol Psychiatry 23, 2133–2144.

19.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

20.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology 43, 2412–2414.

21.

O’Bryant

, Lacritz

, Hall

, Waring

, Chan

, Khodr

, Massman

, Hobson

, Cullum

(2010) Validation of the new interpretive guidelines for the clinical dementia rating scale sum of boxes score in the national Alzheimer’s coordinating center database. Arch Neurol 67, 746–749.

22.

O’Bryant

, Waring

, Cullum

, Hall

, Lacritz

, Massman

, Lupo

, Reisch

, Doody

(2008) Staging dementia using Clinical Dementia Rating Scale Sum of Boxes scores: A Texas Alzheimer’s research consortium study. Arch Neurol 65, 1091–1095.

23.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

24.

Petersen

ed. (2003), Mild Cognitive Impairment: Aging to Alzheimer’s Disease, Oxford University Press, New York.

25.

Ivnik

, Malec

, Smith

, Tangalos

, Petersen

, Kokmen

, Kurkland

(1992) Mayo’s older Americans normative studies: WAIS-R norms for age 56 to 97. Clin Neuropsychol 6, 1–30.

26.

Alberti

, Zimmet

, Shaw

(2005) The metabolic syndrome–a new worldwide definition. Lancet 366, 1059–1062.

27.

Jackson

, Sarafrazi

, Momen

(2011) Appropriate waist circumference cutoff values for the diagnosis of metabolic syndrome in Mexican American adults. FASEB J 25, lb243–lb243.

28.

Masconi

, Matsha

, Erasmus

, Kengne

(2015) Effects of different missing data imputation techniques on the performance of undiagnosed diabetes risk prediction models in a mixed-ancestry population of South Africa. PLoS One 10, e0139210.

29.

Chang

, Chow

, Tellier

, Vattikuti

, Purcell

, Lee

(2015) Second-generation PLINK: Rising to the challenge of larger and richer datasets. GigaScience 4, 7.

30.

Alexander

, Novembre

, Lange

(2009) Fast model-based estimation of ancestry in unrelated individuals. Genome Res 19, 1655–1664.

31.

Yang

, Tian

, Teschendorff

, Webster

, Feber

, Beck

, Morris

(2017) ChAMP: Updated methylation analysis pipeline for Illumina BeadChips. Bioinformatics 33, 3982–3984.

32.

Ritchie

, Smyth

, Phipson

, Wu

, Hu

, Shi

, Law

(2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43, e47–e47.

33.

Peters

, Buckley

, Statham

, Pidsley

, Samaras

, V Lord

, Clark

, Molloy

(2015) De novo identification of differentially methylated regions in the human genome. Epigenetics Chromatin 8, 6.

34.

Jiao

, Widschwendter

, Teschendorff

(2014) A systems-level integrative framework for genome-wide DNA methylation and gene expression data identifies differential gene expression modules under epigenetic control. Bioinformatics 30, 2360–2366.

35.

Dong

, Tian

, Zheng

, Teschendorff

(2019) ebGSEA: An improved gene set enrichment analysis method for epigenome-wide-association studies. Bioinformatics 35, 3514–3516.

36.

Marioni

, Harris

, McRae

, Zhang

, Hagenaars

, Hill

, Davies

, Ritchie

, Gale

, Starr

, Goate

, Porteous

, Yang

, Evans

, Deary

, Wray

, Visscher

(2018) Meta-analysis of epigenome-wide association studies of cognitive abilities. Mol Psychiatry 10, 16–16.

37.

Hahne

, Ivanek

(2016) Visualizing genomic data using gviz and bioconductor. In Statistical Genomics: Methods and Protocols, Mathé

Davis

, eds. Springer New York, New York, pp. 335–351.

38.

Wickham

(2009) ggplot2: Elegant Graphics for Data Analysis, Springer Publishing Company, Incorporated.

39.

Espino

, Lichtenstein

, Palmer

, Hazuda

(2001) Ethnic differences in mini-mental state examination (MMSE) scores: Where you live makes a difference. J Am Geriatr Soc 49, 538–548.

40.

Yamazaki

, Painter

, Bu

, Kanekiyo

(2016) Apolipoprotein E as a therapeutic target in Alzheimer’s disease: A review of basic research and clinical evidence. CNS Drugs 30, 773–789.

41.

Campos

, Edland

, Peavy

(2013) Exploratory study of apolipoprotein E epsilon4 genotype and risk of Alzheimer’s disease in Mexican Hispanics. J Am Geriatr Soc 61, 1038–1040.

42.

Gonzalez

, Tarraf

, Jian

, Vasquez

, Kaplan

, Thyagarajan

, Daviglus

, Lamar

, Gallo

, Zeng

, Fornage

(2018) Apolipoprotein E genotypes among diverse middle-aged and older Latinos: Study of Latinos-Investigation of Neurocognitive Aging results (HCHS/SOL). Sci Rep 8, 17578.

43.

, Wang

, Zhang

, Gazdar

(2013) DNA methylation data analysis and its application to cancer research. Epigenomics 5, 301–316.

44.

Schneider

, Dittrich

, Bock

, Nanda

, Muller

, Seidmann

, Tralau

, Galetzka

, El Hajj

, Haaf

(2016) CpG sites with continuously increasing or decreasing methylation from early to late human fetal brain development. Gene 592, 110–118.

45.

Thomas

, Campbell

, Kejariwal

, Mi

, Karlak

, Daverman

, Diemer

, Muruganujan

, Narechania

(2003) PANTHER: A library of protein families and subfamilies indexed by function. Genome Res 13, 2129–2141.

46.

Yates Kathy

, Sweat

, Yau Po

, Turchiano Michael

, Convit

(2012) Impact of metabolic syndrome on cognition and brain. Arterioscler Thromb Vasc Biol 32, 2060–2067.

47.

Nilsson

, Jansson

, Perfilyev

, Volkov

, Pedersen

, Svensson

, Poulsen

, Ribel-Madsen

, Pedersen

, Almgren

, Fadista

, Ronn

, Klarlund Pedersen

, Scheele

, Vaag

, Ling

(2014) Altered DNA methylation and differential expression of genes influencing metabolism and inflammation in adipose tissue from subjects with type 2 diabetes. Diabetes 63, 2962–2976.

48.

Lettre

, Palmer

, Young

, Ejebe

, Allayee

, Benjamin

, Bennett

, Bowden

, Chakravarti

, Dreisbach

, Farlow

, Folsom

, Fornage

, Forrester

, Fox

, Haiman

, Hartiala

, Harris

, Hazen

, Heckbert

, Henderson

, Hirschhorn

, Keating

, Kritchevsky

, Larkin

, Li

, Rudock

, McKenzie

, Meigs

, Meng

, Mosley

, Newman

, Newton-Cheh

, Paltoo

, Papanicolaou

, Patterson

, Post

, Psaty

, Qasim

, Qu

, Rader

, Redline

, Reilly

, Reiner

, Rich

, Rotter

, Liu

, Shrader

, Siscovick

, Tang

, Taylor

, Tracy

, Vasan

, Waters

, Wilks

, Wilson

, Fabsitz

, Gabriel

, Kathiresan

, Boerwinkle

(2011) Genome-wide association study of coronary heart disease and its risk factors in 8,090 African Americans: The NHLBI CARe Project. PLoS Genet 7, e1001300.

49.

Hannibal

, Ruzzo

, Miller

, Betz

, Buchan

, Knutzen

, Barnett

, Landsverk

, Brice

, LeGuern

, Bedford

, Worrall

, Lovitt

, Appel

, Andermann

, Bird

, Chance

(2009) SEPT9 gene sequencing analysis reveals recurrent mutations in hereditary neuralgic amyotrophy. Neurology 72, 1755–1759.

50.

Wain

, Vaez

, Jansen

, Joehanes

, van der Most

, Erzurumluoglu

, O’Reilly

, Cabrera

, Warren

, Rose

, Verwoert

, Hottenga

, Strawbridge

, Esko

, Arking

, Hwang

, Guo

, Kutalik

, Trompet

, Shrine

, Teumer

, Ried

, Bis

, Smith

, Amin

, Nolte

, Lyytikainen

, Mahajan

, Wareham

, Hofer

, Joshi

, Kristiansson

, Traglia

, Havulinna

, Goel

, Nalls

, Sober

, Vuckovic

, Luan

, Del Greco

, Ayers

, Marrugat

, Ruggiero

, Lopez

, Niiranen

, Enroth

, Jackson

, Nelson

, Huffman

, Zhang

, Marten

, Gandin

, Harris

, Zemunik

, Lu

, Evangelou

, Shah

, de Borst

, Mangino

, Prins

, Campbell

, Li-Gao

, Chauhan

, Oldmeadow

, Abecasis

, Abedi

, Barbieri

, Barnes

, Batini

, Beilby

, Blake

, Boehnke

, Bottinger

, Braund

, Brown

, Brumat

, Campbell

, Chambers

, Cocca

, Collins

, Connell

, Cordell

, Damman

, Davies

, de Geus

, de Mutsert

, Deelen

, Demirkale

, Doney

ASF

, Dorr

, Farrall

, Ferreira

, Franberg

, Gao

, Giedraitis

, Gieger

, Giulianini

, Gow

, Hamsten

, Harris

, Hofman

, Holliday

, Hui

, Jarvelin

, Johansson

, Johnson

, Jousilahti

, Jula

, Kahonen

, Kathiresan

, Khaw

, Kolcic

, Koskinen

, Langenberg

, Larson

, Launer

, Lehne

, Liewald

DCM

, Lin

, Lind

, Mach

, Mamasoula

, Menni

, Mifsud

, Milaneschi

, Morgan

, Morris

, Morrison

, Munson

, Nandakumar

, Nguyen

, Nutile

, Oldehinkel

, Oostra

, Org

, Padmanabhan

, Palotie

, Pare

, Pattie

, Penninx

, Poulter

, Pramstaller

, Raitakari

, Ren

, Rice

, Ridker

, Riese

, Ripatti

, Robino

, Rotter

, Rudan

, Saba

, Saint Pierre

, Sala

, Sarin

, Schmidt

, Scott

, Seelen

, Shields

, Siscovick

, Sorice

, Stanton

, Stott

, Sundstrom

, Swertz

, Taylor

, Thom

, Tzoulaki

, Tzourio

, Uitterlinden

, Volker

, Vollenweider

, Wild

, Willemsen

, Wright

, Yao

, Theriault

, Conen

, Attia

, Sever

, Debette

, Mook-Kanamori

, Zeggini

, Spector

, van der Harst

, Palmer

CNA

, Vergnaud

, Loos

RJF

, Polasek

, Starr

, Girotto

, Hayward

, Kooner

, Lindgren

, Vitart

, Samani

, Tuomilehto

, Gyllensten

, Knekt

, Deary

, Ciullo

, Elosua

, Keavney

, Hicks

, Scott

, Gasparini

, Laan

, Liu

, Watkins

, Hartman

, Salomaa

, Toniolo

, Perola

, Wilson

, Schmidt

, Zhao

, Lehtimaki

, van Duijn

, Gudnason

, Psaty

, Peters

, Rettig

, James

, Jukema

, Strachan

, Palmas

, Metspalu

, Ingelsson

, Boomsma

, Franco

, Bochud

, Newton-Cheh

, Munroe

, Elliott

, Chasman

, Chakravarti

, Knight

, Morris

, Levy

, Tobin

, Snieder

, Caulfield

, Ehret

(2017) Novel blood pressure locus and gene discovery using genome-wide association study and expression data sets from blood and the kidney. Hypertension 70, e4–e19.

51.

Dayeh

, Volkov

, Salo

, Hall

, Nilsson

, Olsson

, Kirkpatrick

, Wollheim

, Eliasson

, Ronn

, Bacos

, Ling

(2014) Genome-wide DNA methylation analysis of human pancreatic islets from type 2 diabetic and non-diabetic donors identifies candidate genes that influence insulin secretion. PLoS Genet 10, e1004160.

52.

Vaithianathan

, Henry

, Akmentin

, Matthews

(2015) Functional roles of complexin in neurotransmitter release at ribbon synapses of mouse retinal bipolar neurons. J Neurosci 35, 4065–4070.

53.

Annese

, Manzari

, Lionetti

, Picardi

, Horner

, Chiara

, Caratozzolo

, Tullo

, Fosso

, Pesole

, D’Erchia

(2018) Whole transcriptome profiling of late-onset Alzheimer’s disease patients provides insights into the molecular changes involved in the disease. Sci Rep 8, 4282–4282.

54.

Haihua

, Linghai

(2016) Targeting cAMP/PKA pathway for glycemic control and type 2 diabetes therapy. J Mol Endocrinol 57, R93–R108.

55.

Dagda

, Das Banerjee

(2015) Role of protein kinase A in regulating mitochondrial function and neuronal development: Implications to neurodegenerative diseases. Rev Neurosci 26, 359–370.

56.

Banzhaf-Strathmann

, Benito

, May

, Arzberger

, Tahirovic

, Kretzschmar

, Fischer

, Edbauer

(2014) MicroRNA-125b induces tau hyperphosphorylation and cognitive deficits in Alzheimer’s disease. EMBO J 33, 1667–1680.

57.

Lee

, Kwon

, Kim

, Choi

, Park

, Kim

(2016) Genome-wide association study for the interaction between BMR and BMI in obese Korean women including overweight. Nutr Res Pract 10, 115–124.

58.

Tavanez

, Calado

, Braga

, Lafarga

, Carmo-Fonseca

(2005) In vivo aggregation properties of the nuclear poly(A)-binding protein PABPN1. RNA 11, 752–762.

59.

Vasavada

, Wang

, Eslinger

, Gill

, Sun

, Karunanayaka

, Yang

(2015) Olfactory cortex degeneration in Alzheimer’s disease and mild cognitive impairment. J Alzheimers Dis 45, 947–958.

60.

Ranzani

, Iyer

, Ibarra-Soria

, Del Castillo Velasco-Herrera

, Garnett

, Logan

, Adams

(2017) Revisiting olfactory receptors as putative drivers of cancer. Wellcome Open Res 2, 9.

61.

Zeng

, Cai

, Li

, Wang

, Li

, Chen

, Zhu

, Zeng

(2016) Essential roles of cyclin Y-like 1 and cyclin Y in dividing Wnt-responsive mammary stem/progenitor cells. PLoS Genet 12, e1006055–e1006055.

62.

Park

, Watanabe

, Poon Vivian Yi

, Ou

C-Y

, Jorgensen Erik

, Shen

(2011) CYY-1/Cyclin Y and CDK-5 differentially regulate synapse elimination and formation for rewiring neural circuits. Neuron 70, 742–757.

63.

Folke

, Pakkenberg

, Brudek

(2019) Impaired Wnt signaling in the prefrontal cortex of Alzheimer’s disease. Mol Neurobiol 56, 873–891.

64.

Conteh

, Reissaus

, Hernandez-Perez

, Nakshatri

, Anderson

, Mirmira

, Tersey

, Linnemann

(2019) Platelet-type 12-lipoxygenase deletion provokes a compensatory 12/15-lipoxygenase increase that exacerbates oxidative stress in mouse islet beta cells. J Biol Chem 294, 6612–6620.

65.

Leibowitz

, Kaiser

, Cerasi

(2011) β-Cell failure in type 2 diabetes. J Diabetes Investig 2, 82–91.

66.

Khanna

, Roy

, Ryu

, Bahadduri

, Swaan

, Ratan

, Sen

(2003) Molecular basis of vitamin E action: Tocotrienol modulates 12-lipoxygenase, a key mediator of glutamate-induced neurodegeneration. J Biol Chem 278, 43508–43515.

67.

Miller

, Wolf

, Sadeh

, Logue

, Spielberg

, Hayes

, Sperbeck

, Schichman

, Stone

, Carter

, Humphries

, Milberg

, McGlinchey

(2015) A novel locus in the oxidative stress-related gene ALOX12 moderates the association between PTSD and thickness of the prefrontal cortex. Psychoneuroendocrinology 62, 359–365.

68.

Manohar

, Dahar

, Adler

, Dalian

, Salvi

(2016) Noise-induced hearing loss: Neuropathic pain via Ntrk1 signaling. Mol Cell Neurosci 75, 101–112.

69.

Haggstrom

, Rosenhall

, Hederstierna

, Ostberg

, Idrizbegovic

(2018) A longitudinal study of peripheral and central auditory function in Alzheimer’s disease and in mild cognitive impairment. Dement Geriatr Cogn Dis Extra 8, 393–401.

70.

Marttinen

, Kurkinen

, Soininen

, Haapasalo

, Hiltunen

(2015) Synaptic dysfunction and septin protein family members in neurodegenerative diseases. Mol Neurodegener 10, 16–16.

71.

Yang

, Kim

, Ryoo

, Lee

, Kim

, Rhim

, Shin

Y-K

(2015) Amyloid-β oligomers may impair SNARE-mediated exocytosis by direct binding to Syntaxin 1a. Cell Rep 12, 1244–1251.

72.

Rappaport

, Nativ

, Stelzer

, Twik

, Guan-Golan

, Stein

, Bahir

, Belinky

, Morrey

, Safran

, Lancet

(2013) MalaCards: An integrated compendium for diseases and their annotation. Database (Oxford) 2013, bat018–bat018.

73.

Chen

, Long

, Cai

, Chen

(2017) A novel relationship for schizophrenia, bipolar and major depressive disorder Part 5: A hint from chromosome 5 high density association screen. Am J Transl Res 9, 2473–2491.

74.

Boden

, Barber

, Clement

, Patel

, Guetta-Baranes

, Brookes

, Chappell

, Craigon

, Chapman

, ARUK Consortium, Morgan

, Seymour

, Bottley

(2017) Methylation profiling RIN3 and MEF2C identifies epigenetic marks associated with sporadic early onset Alzheimer’s disease. J Alzheimers Dis Rep 1, 97–108.

75.

, Fang

, Ding

, Wu

(2018) Dysregulation of Rab5-mediated endocytic pathways in Alzheimer’s disease. Traffic 19, 253–262.

76.

Stage

, Duran

, Risacher

, Goukasian

, Do

, West

, Wilhalme

, Nho

, Phillips

, Elashoff

, Saykin

, Apostolova

(2016) The effect of the top 20 Alzheimer disease risk genes on gray-matter density and FDG PET brain metabolism. Alzheimers Dement (Amst) 5, 53–66.

77.

Yamada

, Sakuma

, Takeuchi

, Yasukochi

, Kato

, Oguri

, Fujimaki

, Horibe

, Muramatsu

, Sawabe

, Fujiwara

, Taniguchi

, Obuchi

, Kawai

, Shinkai

, Mori

, Arai

, Tanaka

(2017) Identification of rs7350481 at chromosome 11q23.3 as a novel susceptibility locus for metabolic syndrome in Japanese individuals by an exome-wide association study. Oncotarget 8, 39296–39308.

78.

, Hao

, Liu

, Lin

, Ma

, Lin

(2018) Effect of KNDC1 overexpression on the senescence of human umbilical vein endothelial cells. Mol Med Rep 17, 7037–7044.

79.

Bernstein

, Lin

, Street

, Lin

, Dai

, Yu

, Bao

, Gearing

, Lah

, Nelson

, He

, Levey

, Mullé

, Duan

, Jin

(2016) 5-Hydroxymethylation-associated epigenetic modifiers of Alzheimer’s disease modulate Tau-induced neurotoxicity. Hum Mol Genet 25, 2437–2450.

80.

Lee

, Wedow

, Okbay

, Kong

, Maghzian

, Zacher

, Nguyen-Viet

, Bowers

, Sidorenko

, Karlsson Linnér

, Fontana

, Kundu

, Lee

, Li

, Royer

, Timshel

, Walters

, Willoughby

, Yengo

; 23andMe Research Team; COGENT (Cognitive Genomics Consortium); Social Science Genetic Association Consortium, Alver

, Bao

, Clark

, Day

, Furlotte

, Joshi

, Kemper

, Kleinman

, Langenberg

, Mägi

, Trampush

, Verma

, Wu

, Lam

, Zhao

, Zheng

, Boardman

, Campbell

, Freese

, Harris

, Hayward

, Herd

, Kumari

, Lencz

, Luan

, Malhotra

, Metspalu

, Milani

, Ong

, Perry

JRB

, Porteous

, Ritchie

, Smart

, Smith

, Tung

, Wareham

, Wilson

, Beauchamp

, Conley

, Esko

, Lehrer

, Magnusson

PKE

, Oskarsson

, Pers

, Robinson

, Thom

, Watson

, Chabris

, Meyer

, Laibson

, Yang

, Johannesson

, Koellinger

, Turley

, Visscher

, Benjamin

, Cesarini

(2018) Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat Genet 50, 1112–1121.

81.

Kim

, Kim

, Rhie

, Yoon

(2015) The role of oxidative stress in neurodegenerative diseases. Exp Neurobiol 24, 325–340.

82.

Ahmad

, Ijaz

, Shabbiri

, Ahmed

, Rehman

(2017) Oxidative toxicity in diabetes and Alzheimer’s disease: Mechanisms behind ROS/ RNS generation. J Biomed Sci 24, 76.

83.

Aliciguzel

, Ozen

, Aslan

, Karayalcin

(2003) Activities of xanthine oxidoreductase and antioxidant enzymes in different tissues of diabetic rats. J Lab Clin Med 142, 172–177.

84.

Stock

, Kasus-Jacobi

, Pereira

(2018) The role of neutrophil granule proteins in neuroinflammation and Alzheimer’s disease. J Neuroinflammation 15, 240–240.

85.

Csiszar

, Tucsek

, Toth

, Sosnowska

, Gautam

, Koller

, Deak

, Sonntag

, Ungvari

(2013) Synergistic effects of hypertension and aging on cognitive function and hippocampal expression of genes involved in β-amyloid generation and Alzheimer’s disease. Am J Physiol Heart Circ Physiol 305, H1120–H1130.

86.

Kempuraj

, Thangavel

, Natteru

, Selvakumar

, Saeed

, Zahoor

, Zaheer

, Iyer

, Zaheer

(2016) Neuroinflammation induces neurodegeneration. J Neurol Neurosurg Spine 1, 1003.

87.

Kempuraj

, Selvakumar

, Thangavel

, Ahmed

, Zaheer

, Kumar

, Yelam

, Kaur

, Dubova

, Raikwar

, Iyer

, Zaheer

(2018) Glia maturation factor and mast cell-dependent expression of inflammatory mediators and proteinase activated receptor-2 in neuroinflammation. J Alzheimers Dis 66, 1117–1129.

88.

Lim

, Iyer

, Liu

, Suen

, Lohman

, Seow

, Yau

, Brown

, Fairlie

(2013) Diet-induced obesity, adipose inflammation, and metabolic dysfunction correlating with PAR2 expression are attenuated by PAR2 antagonism. FASEB J 27, 4757–4767.

89.

Nishioka

, Hayashi

, Farrer

, Singleton

, Yoshino

, Imai

, Kitami

, Sato

, Kuroda

, Tomiyama

, Mizoguchi

, Murata

, Toda

, Imoto

, Inazawa

, Mizuno

, Hattori

(2006) Clinical heterogeneity of alpha-synuclein gene duplication in Parkinson’s disease. Ann Neurol 59, 298–309.

90.

Zyskind

, Wang

, Cho

, Ting

, Kolson

, Lynch

, Jordan-Sciutto

(2015) E2F1 in neurons is cleaved by calpain in an NMDA receptor-dependent manner in a model of HIV-induced neurotoxicity. J Neurochem 132, 742–755.

91.

Denechaud

, Fajas

, Giralt

(2017) E2F1, a novel regulator of metabolism. Front Endocrinol (Lausanne) 8, 311.

92.

Russo

, Melista

, Cui

, DeStefano

, Bakris

, Manolis

, Gavras

, Baldwin

(2005) Association of NEDD4L ubiquitin ligase with essential hypertension. Hypertension 46, 488–491.

93.

Nilsson

, Matte

, Perfilyev

, de Mello

, Käkelä

, Pihlajamäki

, Ling

(2015) Epigenetic alterations in human liver from subjects with type 2 diabetes in parallel with reduced folate levels. J Clin Endocrinol Metab 100, E1491–E1501.

94.

Atkin

, Paulson

(2014) Ubiquitin pathways in neurodegenerative disease. Front Mol Neurosci 7, 63.

95.

Wang

, Reddy

(2017) Role of glutamate and NMDA receptors in Alzheimer’s disease. J Alzheimers Dis 57, 1041–1048.

96.

Zhao

, Deng

, Jiang

, Qing

(2016) G protein-coupled receptors (GPCRs) in Alzheimer’s disease: A focus on BACE1 related GPCRs. Front Aging Neurosci 8, 58–58.

97.

Ando

, Uemura

, Kuzuya

, Maesako

, Asada-Utsugi

, Kubota

, Aoyagi

, Yoshioka

, Okawa

, Inoue

, Kawamata

, Shimohama

, Arai

, Takahashi

, Kinoshita

(2011) N-cadherin regulates p38 MAPK signaling via association with JNK-associated leucine zipper protein: Implications for neurodegeneration in Alzheimer disease. J Biol Chem 286, 7619–7628.

98.

Mellendijk

, Wiesmann

, Kiliaan

(2015) Impact of nutrition on cerebral circulation and cognition in the metabolic syndrome. Nutrients 7, 9416–9439.

99.

, Quan

, Toth

, Campbell

, MacKintosh

, Wang

, Chen

(2015) Fasting and systemic insulin signaling regulate phosphorylation of brain proteins that modulate cell morphology and link to neurological disorders. J Biol Chem 290, 30030–30041.

100.

Massaad

, Klann

(2011) Reactive oxygen species in the regulation of synaptic plasticity and memory. Antioxid Redox Signal 14, 2013–2054.

101.

Ferreira

LSS

, Fernandes

, Vieira

MNN

, De Felice

(2018) Insulin resistance in Alzheimer’s disease. Front Neurosci 12, 830–830.

102.

Pruimboom

, Raison

, Muskiet

(2015) Physical activity protects the human brain against metabolic stress induced by a postprandial and chronic inflammation. Behav Neurol 2015, 569869.

103.

Mottahedin

, Ardalan

, Chumak

, Riebe

, Ek

, Mallard

(2017) Effect of neuroinflammation on synaptic organization and function in the developing brain: Implications for neurodevelopmental and neurodegenerative disorders. Front Cell Neurosci 11, 190.

104.

, Marshall

, Oh

, Jakubowski

, Groot

, He

, Wang

, Petronis

, Labrie

(2019) Epigenetic dysregulation of enhancers in neurons is associated with Alzheimer’s disease pathology and cognitive symptoms. Nat Commun 10, 2246.