Identification of diagnostic DNA methylation markers in the blood of Japanese Alzheimer’s disease patients using methylation capture sequencing

Identification of diagnostic DNA methylation markers in the blood of Japanese Alzheimer’s disease patients using methylation capture sequencing | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Identification of diagnostic DNA methylation markers in the blood of Japanese Alzheimer’s disease patients using methylation capture sequencing Risa Mitsumori, Kayoko Sawamura, Kimi Yamakoshi, Akinori Nakamura, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6200381/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 20 Jun, 2025 Read the published version in Clinical Epigenetics → Version 1 posted 10 You are reading this latest preprint version Abstract Background Methylation capture sequencing (MC-seq), which relies on next-generation sequencing technology, offers advantages over the widely used array-based approach that Illumina Inc. developed regarding both resolution and comprehensiveness for detecting DNA methylation changes across genomes. In the present study, MC-seq was employed for the first time to identify DNA methylation markers for Alzheimer’s disease (AD). Results We compared DNA methylation in the blood of 12 AD patients with brain amyloidosis and 12 cognitively normal elderly Japanese individuals without brain amyloidosis. Candidate methylation differences were validated in the two cohorts using bisulfite amplicon sequencing. Significant differentially methylated regions were identified in the ANKH , MARS , ANKFY1 , LINC00908 , and KLF2 genes and a slight methylation change in CHRNE ( p = 0.061). Furthermore, our AD diagnostic prediction model showed that combining the methylation levels of ANKH and MARS with the APOE genotype provided diagnostic accuracy, achieving AUCs of 0.90 and 0.81 in the discovery and validation datasets, respectively. Conclusions The present results suggest the potential of combining these markers for diagnosing AD and support the validity of our approach for identifying disease-related DNA methylation markers using next-generation sequencing. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background In vertebrate genomes, cytosine bases at cytosine-phosphate-guanine dinucleotide sites (CpGs) are the primary substrates for methylation by DNA methyltransferases. Due to its stability in blood samples [ 1 ], susceptibility to environmental factors, and regulatory roles in gene expression, DNA methylation is a promising biomarker and a potential key to identifying genes associated with complex diseases, including Alzheimer’s disease (AD). Many CpG sites associated with AD have been identified by comprehensive microarray DNA methylation analyses and targeted gene approaches using blood- and brain-derived DNA [ 2 ]. However, few CpGs have been replicated across independent studies [ 2 ]. This lack of reproductivity may be attributed to unresolved limitations in microarray methylation analyses [ 3 ] [ 4 ] [ 5 ]. Furthermore, microarrays capture only 3% of the 28 million CpGs in the human genome, suggesting that many AD-associated CpGs have yet to be discovered. The identification of hidden CpGs in blood samples will contribute to the more accurate prognostication and diagnosis of AD as well as a more detailed understanding of its etiology. To discover novel and reproducible DNA methylation changes associated with AD, we employed methylation capture sequencing (MC-seq), also known as targeted bisulfite sequencing, an alternative method originally named solution hybrid selection bisulfite sequencing [ 6 ]. Using next-generation sequencing (NGS), MC-seq directly quantifies methylation levels at individual CpG sites in DNA fragments captured by biotinylated RNA or DNA probes designed for specific genomic regions. In this study, we selected the TruSeq Methyl Capture EPIC library prep (TruSeq EPIC) from Illumina Inc., which enables methylation quantification at more than 3.3 million CpG sites per individual, representing 12% of all CpGs in the human genome. Capture probes target 107 Mb of genomic DNA (gDNA), covering key regulatory elements, such as promoters, enhancers, CpG islands, and CpG island shores/selves [ 7 ]. MC-seq has already successfully identified a number of disease-related markers [ 8 ] [ 9 ] [ 10 ]; however, to the best of our knowledge, it has not yet been applied to AD research. Although MC-seq is more cost effective than whole-genome bisulfite sequencing [ 11 ], the processing of large sample sets remains expensive. Therefore, in the present study, we used MC-seq to identify differentially methylated regions (DMRs) in the first cohort composed of 12 AD patients (Pittsburgh compound-B positron emission tomography (PiB-PET)-positive) and 12 matched cognitively normal (CN) elderly individuals (PiB-PET-negative). Candidate DMRs detected through MC-seq were validated in two independent cohorts of AD and CN individuals using whole-blood DNA and bisulfite amplicon sequencing (BA-seq), another NGS-based method [ 12 ]. Methods Subjects The cognitive status and dementia severity of subjects were assessed using the Mini-Mental State Examination (MMSE) by trained physicians at the National Center for Geriatrics and Gerontology (NCGG) hospital. The first cohort was composed of 12 AD cases and 12 CN elderly individuals, with and without brain amyloidosis, as detected by PiB-PET scanning at the NCGG hospital. In a replication analysis, we prepared second and third cohorts, each consisting of 48 clinically diagnosed AD cases and 48 CN individuals, all aged 60 years or older. Clinical information, including MMSE scores, sex, age, and the apolipoprotein E ( APOE ) epsilon (e) allele genotype, was obtained from the NCGG biobank. Preparation of gDNA gDNA was extracted from peripheral blood using a Maxwell RSC Instrument (Promega, USA) with the Maxwell RSC Buffy Coat DNA Kit (Promega, USA). The quantity of gDNA was measured using the Quant-iT 1×dsDNA BR Assay (Thermo Fisher Scientific, USA), and the intactness of gDNA was assessed by electrophoresis on 1.0% agarose gels. Sodium bisulfite treatment of DNA In a methylation analysis, 200 ng of DNA samples was treated with sodium bisulfite to convert methylated cytosine to cytosine and demethylated cytosine to uracil, either by the EZ DNA methylation Gold Kit or EZ-96 DNA Methylation-Gold Kit (Zymo Research, USA) according to the manufacturer’s instructions. MC-seq We processed twenty-four samples of gDNA from the first cohort using the target enrichment system, the TruSeq methyl capture EPIC Kit (Illumina, Inc., CA, USA). Captured DNA fragments were sequenced by NGS as follows: gDNA was fragmented to a few hundred base pairs (bps) by a sonicator (M220, Covaris), followed by blunting, phosphorylation, the addition of 3’-dA, and the ligation of indexed adaptors to both ends of DNA. DNA was mixed and hybridized to capture oligos, and streptavidin-conjugated magnetic beads recovered hybridized DNA. Eluted DNA from magnetic beads was treated with sodium bisulfite and amplified by PCR to construct a sequencing library, which was sequenced using the Illumina HiSeq 2500 platform with paired-end reads of 100 bp according to the manufacturer’s instructions. Library construction and sequencing were conducted by Takara Bio, Inc. (Shiga, Japan). Raw sequence data were converted from a bcl file to a fastq file using Illumina bcl2fastq2 Conversion software v2.17. Sequencing data were aligned to the reference sequence (Genome Reference Consortium Human Build 37 (GRCh37)/UCSCgh19) using the application, MethylSeq, through the cloud service named BaseSpace (Illumina, Inc., CA, USA). The numbers of methylated and unmethylated cytosines at each CpG site, namely, the methylation call, were counted by MethylKit, another application available through BaseSpace. The same application used methylation call data to detect differential DNA methylation between AD subjects and CN elderly. We set the conditions of MethylKit for differentially methylated CpG sites to be identified as follows: ≥10 coverage with independent reads and ≥15% methylation differences with a q -value ≤0.01. Definition of DMR Genomic regions where differentially methylated cytosines cluster are referred to as DMRs, although the criteria for defining DMR may be arbitrary. In the first cohort, we identified DMRs using a sliding window approach, as shown in Fig. 1 . We initially searched for genomic regions containing at least five or three cytosines with consistent methylation differences ( D b ≥0.15 or D b ≥0.25, respectively) within a 1-kb window. The window was then slid up to 1 kb in either or both directions as long as the new window contained a combination of differentially methylated cytosines that met the above criteria. A region was defined as a DMR if even a single window met these conditions. PCR primers PCR primers for bisulfite-treated gDNA were designed by MethPrimer [ 13 ] or by Pyrosequencing Assay Design Software ver. 2.0 provided by Qiagen (CA, USA) for pyrosequencing. Preparation of amplicons from the second cohort One or multiple regions of each of the 88 DMR candidates were amplified from each subject’s DNA in the second cohort by using KOD Multi & Epi (Toyobo, inc., Shiga, Japan) in 20 µl of the reaction mixture with the following touchdown PCR conditions: at 94℃ for 2 min; 4 cycles at 98°C for 10 sec, 64°C for 30 sec, and 68°C for 15 sec; 4 cycles at 98°C for 10 sec, 60°C for 30 sec, and 68°C for 15 sec; 4 cycles at 98°C for 10 sec, 58°C for 30 sec, and 68°C for 15 sec; 30 cycles at 98°C for 10 sec, 55°C for 30 sec, and 68°C for 15 sec. The average length of amplicons was 237 bp. All amplicons were electrophoresed on 2% agarose gels with 0.5×TAE buffer, followed by staining with GelRed (Biotium, USA), and PCR was repeated for unsuccessful amplification. We collected 1 µl from each of the 133 amplicons from single subjects and mixed the aliquots in single tubes. Pooled PCR products were then purified by QIAvac 96 (QIAGEN, CA, USA), which removes low-molecular weight DNA (≤80 bp). We electrophoresed 1 µl of each eluate on a 2% agarose gel to check the recovery of pooled PCR products. The concentrations of purified pools of amplicons were measured using a 2200 TapeStation Instrument (Agilent Technologies, Santa Clara, CA, USA). Preparation of amplicons from the third cohort One or multiple regions of each of the six DMRs identified in the present study were amplified from each subject in the third cohort, and all amplicons were assessed using the method described above. Based on the amplification level judged from agarose gel electrophoresis of each amplicon, we took either 1 or 4 µl from each amplicon per subject and pooled the aliquots into single tubes. The final volume of each pooled sample was adjusted to 100 µl with distilled water. Purification of the mixture, followed by agarose gel electrophoresis and quantification of the purified mixture were performed as described above. BA-seq We constructed libraries for BA-seq from 80 ng of purified amplicon mixtures by using TruSeq Nano DNA Library Prep Kits (Illumina, USA) following the manufacturer’s instructions, except for the enrichment process of DNA fragments. We used AMPure XP (Beckman Coulter, USA) instead of the sample purification beads included in the kit. We started from the middle of the instructions because the fragmentation of DNA was unnecessary for PCR amplicons. We started with the reaction of “A-tailing”, which adds adenine to the ends of amplicons. The molarity of the libraries and the degree of the undesirable inclusion of adaptor dimers in the libraries, which appeared around 150 bp, were assessed using the Agilent D1000 screen tapes system on a 2200 TapeStation system (Agilent, USA). A 4 nM pool of libraries and 4 nM PhiX Control v3 (Illumina, USA) were each denatured for 5 min with freshly prepared 0.2 N NaOH and adjusted at concentrations of 6 and 8 pM, respectively. Denatured pools were mixed with denatured Phi X at volume ratios of 85:15 and 70:30 for the second and third cohort DNA amplicon libraries, respectively, and 600 µl of each of the prepared samples was loaded into the reagent cartridge of the MiSeq Reagent Kit v2 (Illumina, USA). The cartridge sets on the Illumina MiSeq platform were used to sequence libraries with 151 paired-end, dual-indexing cycles per read (2 × 151). The densities of clusters generated on a flow cell were 322 ± 8 and 343 ± 10 k/mm 2 for BA-seq of the second and third cohorts, respectively. Data analysis of BA-Seq We trimmed adaptor sequences from read sequences using Trim Galore (version 0.6.6), setting the trimming cut-off at 20 (-q) and using the rrbs option (--rrbs). After trimming, sequencing data were mapped to the human reference genome (GRCh37) using Bismark (version 0.22.3) with the “genome_preparation” function, an aligner optimized for bisulfite sequence data and methylation calling. The BAM files generated were sorted by chromosomes using SAMTools (“sort”). Sorted BAM files were then utilized for methylation calling and the calculation of methylation rates using the methylKit package in R (genomeBismarkAln), with the following settings: human genome assembly ‘hg19’, read context ‘CpG’, and a minimum coverage cut-off of ‘10’. By using the DNA methylation rates (meC/meC + C) of CpGs in DMRs as explanatory valuables, the significance of methylation differences at DMRs was evaluated using the burden test (SKAT in R), adjusting for sex and age (kernel=“liner.weighted”). Selection of best DMR sets We selected CpG sites with methylation differences between CN and AD that were significant ( P bon \(\:\le\:\:\) 0.05) in the second and third cohort analyses and that showed the same direction of methylation change. Second and third cohort subjects were mixed and randomly split: four-fifths were used for a discovery dataset and one-fifth for a validation dataset, using sklearn (“train_test_split”) in Python3. Using discovery data, prediction models were constructed based on clinical information (age, sex, and APOE ε 4 genotypes), and all possible combinations of DMRs using a logistic regression model. The optimal model, incorporating clinical information and DMRs, in the discovery dataset was selected using a 3-fold cross-validation. After finalizing the model on the entire discovery dataset, its performance was evaluated on an independent validation dataset, with the area under the receiver operator characteristic (ROC) curve (AUC) being used to assess discriminative accuracy. Analyses were implemented using the sklearn library in Python (penalty: “elasticnet”, solver: “saga”, 1l_ratio: 0.5). Sensitivity to the values of the regularization parameter (C parameter (0.1, 01, 10, 100)) was examined with the help of simulating by the best model. Results Detection of candidate DMRs in the first cohort The present study focused on DMRs, clusters of neighboring CpGs with significant methylation changes in the same direction. There were two main reasons for this approach. Cytosine-to-thymine transitions at CpG sites, the most frequent de novo mutation occurring in the human genome [ 14 ], are indistinguishable from unmethylated cytosines after a bisulfite treatment, potentially leading to false differentially methylated positions (DMPs). To prevent this, we considered only DMPs aggregated in a short distance, defining them as DMRs for use as diagnostic markers. Furthermore, the utility of DMRs in blood DNA as diagnostic markers for AD has already been demonstrated [ 15 ] [ 16 ]. Our approach to comprehensively identify DMRs in the blood DNA of AD patients is shown in Fig. 2 . We first directly compared sequences from the first cohort, composed of 12 CN and 12 AD individuals, using the commercially available TruSeq Methyl Capture EPIC kit (Illumina Inc.). All donors in this cohort underwent PiB-PET imaging, with 12 CN and 12 AD individuals testing negative and positive for brain amyloidosis, respectively (Table 1). Table 1. Sample characteristics of three cohorts used in this study 1st cohort 2nd cohort 3rd cohort Characteristics CN- AD+ CN AD CN AD Samples 12 12 48 48 48 48 Male 4 (33.3) 2 (16.7) 24 (50) 24 (50) 24 (50) 24 (50) Age, years 73.3 ± 3.2 74.4 ± 6.4 70.3 ± 2.4 71.2 ± 4.0 69.8 ± 3.0 71.8 ± 2.6 MMSE score 29.0 ± 1.6 21 ± 3.4 28.8 ± 1.3 16.2 ± 3.5 29.5 ± 0.7 16.9 ± 4.9 APOE \(\:\epsilon\:\) 4 homozygote 0 (0) 2 (16.7) 1 (2.1) 5 (10.4) 1 (2.1) 6 (12.5) APOE \(\:\epsilon\:\) 4 heterozygote 1 (8.3) 8 (66.7) 5 (10.4) 30 (62.5) 10 (20.8) 19 (39.6) APOE \(\:\epsilon\:\) 4 non-carriers 11 (91.7) 2 (16.7) 42 (87.5) 13 (27.1) 37 (77.1) 23 (47.9) PiB-PET negative positive - - - - Age and MMSE are expressed as the mean ± SD, while the other variables are expressed as numbers, n (%). Abbreviations: SD, standard deviation; MMSE, Mini-Mental State Examination; APOE , apolipoprotein E; -, not executed. A total of 437,792 target DNA fragments were recovered from the first cohort, yielding 107,403,887 bp of sequence data. These fragments were mapped to UCSDhg19 using Bowtie2 [ 17 ], with an average mapping rate of 83.9% (CN: 84.07 ± 0.45%; AD: 83.81% ± 0.37). The methylation rate at each CpG site was calculated by Bismark [ 18 ], which identifies methylated and unmethylated cytosines in sequenced reads. Illumina integrates both programs into the online application, MethylSeq, available through BaseSpace Sequence Hub, which was used in the present study. A total of 3,233,016 CpG sites were analyzed, covering 96.8% of the 3,340,894 CpGs detectable by the TruSeq Methyl Capture EPIC kit. These results indicated the successful construction of an enrichment library and sequencing (Table 2). An aggregated summary and raw data of MC-seq are shown in Table 3 and supplementary Table S1 , respectively. Table 2. Expected and actual recovery of target regions and CpGs from the first cohort Illumina TruSeq Methyl Capture EPIC Template Expected Actual The human genome Total length of targets (Mb) 107 107 3,000 Number of target regions (%) 437,792 (100%) 436,268 (99.7% a ) Number of CpGs 3,340,892 (11.8% c ) 3,233,016 (96.8% b ) 2,8217,009 (100%) The ratio of recovered targets a or CpGs b to the expected number of targets or CpGs to be recovered from the experimental design. The ratio of the expected number of CpGs to be analyzed c to the total number of CpGs in the human genome. Table 3. Statistics of methyl capture sequencing of the first cohort CN- AD+ Total PF reads a 135879157 ± 11735593 135600493 ± 22796635 Aligned reads b , % 84.07 ± 0.45 83.81 ± 0.37 Duplicate aligned reads c , % 12.42 ± 3.90 10.49 ± 1.70 Unique read enrichment d , % 79.84 ± 0.76 79.17 ± 1.87 Target coverage at least 20×, % 93.60 ± 1.96 93.56 ± 3.12 Cytosine methylation in the CpG context, % 52.66 ± 0.69 51.91 ± 0.62 Cytosine methylation in the non-CpG context, % 1.10 ± 0.02 0.98 ± 0.03 CN- and AD + indicate cognitively normal elderly negative for amyloid PET and clinically diagnosed AD cases positive for amyloid PET, respectively. a The total number of pass-filter reads for the sample. b The percentage of pass-filter reads that aligned to the reference genome. c The percentage of paired-end reads flagged as duplicates. d 100×(Targeted unique aligned reads/Unique aligned reads). Variables are expressed as the mean ± SD. We used MethylKit [ 19 ], accessible through BaseSpace Sequence Hub, to identify DMPs that may constitute DMRs under the following stringent settings: (1) CpGs with a minimum coverage of 10 reads (coverage ≥ 10), (2) CpGs showing a methylation difference of at least 15% (| D b| ≥15%) between CN and AD, and (3) CpGs with a q -value < 0.01. We obtained 10,381 DMPs, including 1,077 DMPs with | D b| ≥25%. After excluding 124 and 1 DMPs from the X and Y chromosomes, respectively, we retained 10,256 DMPs of | D b| ≥15%, of which 1,065 had | D b| ≥25%. To detect DMRs, we applied a sliding window analysis (Fig. 1 ), defining DMRs as regions containing at least seven DMPs with | D b| ≥15% or five DMPs with | D b| ≥25% within a genomic region of a few kb. We identified 106 DMRs, including 6 DMRs composed entirely of | D b| ≥25% DMPs (supplementary Table S2). Of these, 86 DMRs (81%) were hypomethylated (CN > AD), while 20 (19%) were hypermethylated (CN < AD) (Fig. 3 ). The predominance of hypomethylation in AD blood samples is consistent with previous findings [ 20 ]. Detailed information on the positions and lengths of DMRs is shown in Supplementary Table S2. Validation of DMRs with the second cohort To validate methylation differences in the 106 AD-related DMRs detected from the first cohort, we analyzed a second cohort of 48 AD cases and 48 CN individuals (none of whom underwent PiB-PET imaging, Table 1) by BA-seq. Primer sets were successfully designed for 91 DMRs, while the remaining 15 DMRs could not be targeted (Supplementary Table S3). Primer specificity was initially examined to establish whether they uniquely PCR amplified DNA of the expected sizes from control bisulfite-treated human gDNA by electrophoresis of PCR products on 2% agarose gels. Alternative primer sets were tested for unsuccessful amplifications; however, the expected PCR products were not obtained for 3 DMRs (Supplementary Table S3). In total, 133 amplicons were ultimately generated against 88 DMRs (Supplementary Table S4), and a bisulfite-amplicon DNA library was constructed for paired-end sequencing. The genomic locations of the DMRs, their relative positions within genes, primer sequences, amplicon lengths, and other details are shown in Supplementary Table S4. Since bisulfite-treated DNA and PCR amplicons significantly reduce library diversity, we spiked the sample with 15% PhiX control and reduced the library concentration to achieve a lower cluster density on a flow cell than recommended for standard DNA sequencing. The MiSeq sequencing run yielded a cluster density of 322 ± 8 (K/mm 2 ), with 82.12 ± 2.96 (%) of the cluster passing filter, and a ≥Q30 score of 91.0%. The average read numbers per sample were similar between the CN and AD groups: 38,977 and 38,656, respectively (Table 4). Table 4. Averaged results of bisulfite amplicon sequencing 2nd cohort 3rd cohort Characteristics CN AD CN AD Total number of pair-end reads 38977 ± 4111.3 38656 ± 4994.5 29702 ± 5871.4 32623 ± 8886.1 Reads uniquely mapped to the reference genome 30747 ± 3454.8 29541 ± 3986.6 28341 ± 5645.9 31103 ± 8486.8 Reads unable to be mapped at any loci 5764 ± 913.4 6501 ± 954.2 1361 ± 285.4 1519 ± 456.6 Reads mapped at multiple loci 2466 ± 377.2 2614 ± 510.4 0.25 ± 0.56 0.27 ± 0.54 Mapping efficiency (%) 78.9 ± 2.03 76.4 ± 1.70 95.4 ± 0.64 95.3 ± 0.78 All continuous variables are expressed as the mean ± SD. The average mapping rates of reads from CN and AD samples were 78.9 and 76.4%, respectively (Table 4). Among the 88 DMRs analyzed, eight amplicons corresponding to eight DMRs were excluded due to insufficient read coverage (< 10 reads) in at least one sample. Additionally, 10 amplicons representing 10 DMRs were excluded because they could not be uniquely mapped to the reference genome, likely due to the short length and low complexity of DMR sequences (Supplementary Table S5). Regarding the remaining 70 DMRs, 114 amplicons, including 1,251 CpG sites, were analyzed. Methylation differences at 511 CpGs in 67 DMRs were significant ( P bon ≤0.05/1251). Of the 67 DMRs, 52 contained 496 CpGs with at least three CpGs per DMR showing significant methylation differences. Using the DNA methylation rates (meC/(meC + C)) of these CpGs as explanatory valuables, the significance of methylation differences at these DMRs was evaluated using the burden test, adjusted for sex and age [ 21 ]. The burden test is a collapsing method used for genetic association analyses of rare variants. The method collapses variants within a gene and analyzes them as a unit to assess whether the gene is associated with a disease. In the present study, the methylation rates of CpGs in a DMR were treated as the allele frequencies of genetic variants in a gene, and the relationship with AD was tested. The p -values of DMRs calculated by the burden test were converted to q -values using the Benjamini-Hochberg procedure [ 22 ], and 6 DMRs were found to be significant with q -values < 0.05 (Table 5). Five of the 6 DMRs were located in the gene body regions of the following genes: ANKilosis Homolog ( ANKH ), MARS , ANKFY1 , Cholinergic Receptor Nicotinic Epsilon Subunit ( CHRNE ), and KLF2 , while the remaining DMR was found in the long non-coding RNA gene, LINC00908 . Of these, five DMRs in ANKH , MARS , ANKFY1 , CHRNE , and LINC00908 showed lower methylation levels in AD than in CN, whereas one DMR, located in and around KLF2 , exhibited a higher methylation level in AD (Fig. 4 ). Table 5. Summary of the validation of 6 DMRs in two cohorts Burden test Methylation difference a Chromosome Gene Amplicon (bp) Location Group p value FDR (%) 5 ANKH 213 intron 2nd cohort 0.0034 0.0447 -4.74 3rd cohort 0.0633 meta-analysis 0.0017 12 MARS 169 exon 2nd cohort 0.0054 0.0467 -5.17 3rd cohort 0.2452 meta-analysis 0.0072 17 ANKFY1 594 intron 2nd cohort 0.0046 0.0467 -5.26 3rd cohort 0.0680 meta-analysis 0.0021 17 CHRNE 615 exon-intron 2nd cohort 0.0013 0.0331 -4.42 3rd cohort 0.6739 meta-analysis 0.0608 18 LINC00908 424 intron 2nd cohort 0.0023 0.0402 -4.53 3rd cohort 0.0645 meta-analysis 0.0011 19 KLF2 536 exon-intron 2nd cohort 0.0006 0.0319 7.37 3rd cohort 0.4455 meta-analysis 0.0015 a For each CpG site in Table S6, the methylation difference in the second cohort was calculated, and the sum of the differences was then averaged. Validation of 6 DMRs with the third cohort Methylation changes in the 6 genes were reexamined by BA-seq with the third cohort, composed of 48 CN and 48 AD subjects (none of whom underwent PiB-PET imaging, Table 1). We generated 12 amplicons that covered 108 CpGs in the 6 DMRs from the third cohort, and a bisulfite DNA library for Illumina MiSeq sequencing was again constructed from the 12 amplicons (Table 4). The MiSeq sequencing run resulted in a cluster density of 343 ± 100 (K/mm 2 ), a cluster passing filter of 87.09 ± 1.85, and Q30 ≥94.2%. The mean read numbers per sample from the CN and AD groups were 29,702 and 32,623, respectively, with average mapping rates of reads from CN and AD being 95.4 and 95.3%, respectively. We did not obtain any reads for an amplicon in the KLF2 DMR from one AD subject. DNA methylation differences were significant at 96 CpGs among the 108 CpGs surveyed ( P bon ≤0.05/108). We also assessed the methylation difference at 6 DMRs in third cohort samples using the Burden test (Table 5). The lowest p -value was 0.063 at ANKH , and the highest was 0.446 at KLF2 . However, combining data from the second and third cohorts for the 6 DMRs showed a significant methylation difference ( p < 0.05) in all genes, except for CHRNE , which showed a slight difference ( p = 0.06). We regarded the five DMRs with significant methylation differences as confirmed DMRs. Diagnostic ability of AD with DNA methylation We hypothesized that the five DMRs may contain CpG sites capable of discriminating between CN and AD. We selected CpG sites with significant methylation differences between CN and AD ( P bon ≤0.05) in the second and third cohort analyses, ensuring that the direction of the methylation change was consistent across both cohorts. The numbers of CpG sites meeting these criteria were as follows: 6 CpGs in ANKH , 7 in MARS , 14 in ANKFY1 , 3 in LINC00908 , and 30 in KLF2 . The locations of these CpGs are shown in Supplementary Table S6. Methylation levels at these CpGs in the five DMRs, together with the following covariates: age, sex, and the APOE e4 genotype, were used as explanatory variables in a logistic regression analysis to identify the model that best discriminated AD from CN (Fig. 5 A). The APOE genotype is a well-established risk factor for AD [ 23 ]. We performed an AUC-ROC analysis incorporating methylation levels from ANKH and MARS adjusting with the APOE e4 allele, which yielded the highest AUC score in the validation set of 0.81 (95% CI: 0.67–0.95, Fig. 5 B and Supplementary Table S7). The base model, which used only the APOE e4 genotype, showed a lower AUC of 0.77 (95% CI: 0.69–0.84, Fig. 5 C and Supplementary Table S7). Therefore, we concluded that combining methylation levels at specific regions in ANKH and MARS with the APOE genotype provided the best model for the diagnosis of AD among the combinations assessed. Discussion In the present study, we examined the validity of MC-seq in identifying AD-related DMRs in blood for the first time. We obtained 106 DMRs in the first cohort using MC-seq, and 5 DMRs remained after validation in two independent AD and CN cohorts using BA-seq. The low survival rate of DMRs, less than 5%, from the first screening was likely due to the small sample size, which may have led to false positives. We expect this survival rate to improve with a larger sample size in the first cohort because marker reliability is dependent on sample size [ 24 ]. In the second cohort, methylation differences between CN and AD in the five DMRs were approximately 5%, whereas the initial screening criteria required a 15% difference. This suggests that the observed methylation differences in the first cohort were exaggerated due to sample-specific variability. Similarly, we previously detected small methylation changes in known AD-associated genes, including CLU , CR1 , and PICALM [ 15 ]. These findings indicate that DMRs in AD blood may not exhibit large differences, highlighting the need for highly sensitive methods to detect small methylation changes in blood-based biomarkers for AD. The combination of methylation levels at two of the five DMRs with APOE genotypes gave a discriminative power (AUC) of 0.81 for the validation cohort. Due to the small sample size for the first cohort, we expect that increasing the sample size of the first cohort analyzed by MC-seq will result in improved DMRs and, consequently, a higher AUC. Additionally, further improvements may be achieved by using capture probes specifically designed by researchers, as demonstrated in other studies. This approach may enhance the sensitivity and specificity of the method, ultimately improving the identification of reliable biomarkers for AD [ 25 ]. We previously showed that AD-related DNA methylation decreased at CpG island shores of CLU , CR1 , and PICALM [ 15 ]. CpG island shores are located 2 ~ 3 kb from CpG islands and are typically intermediately methylated [ 26 ]. In the present study, all 5 DMRs were found in intermediately methylated regions and, except for KLF2 , showed lower methylation levels in AD. However, these DMRs were too distant from CpG islands to be classified as CpG island shores (Fig. 4 ). This suggests that, regardless of their positions related to CpG islands, intermediately methylated regions within genes may be susceptible to AD-related methylation changes. Since aging is the most significant risk factor for AD and age-related methylation changes in the human genome [ 27 ], it will be interesting to examine whether DNA methylation at these DMRs decreases with age. Hypomethylation in gene bodies has been linked to transcriptional integrity, including spurious transcription initiation [ 28 ], mis-splicing [ 29 ], and readthrough beyond transcriptional termination signals [ 30 ], all of which may decrease the activities of hypomethylated genes. Recent advances in targeted hypomethylation techniques may enable functional analyses of these DMRs in model systems [ 31 ]. Two genes identified in this study, ANKH and CHRNE , have also been genetically related to AD [ 32 ] [ 33 ]. ANKH encodes a transmembrane protein involved in regulating pyrophosphate levels in joints and other tissues. Mutations in ANKH may lead to excessive mineralization, contributing to joint pain, arthritis, atherosclerosis, and diabetes. A minor allele of rs112403360, located in an intron of ANKH , has been associated with a 1.09-fold increased risk of AD [ 32 ]. In contrast, the major allele, considered to be a protective allele against AD, has been associated with cognitively healthy centenarians [ 34 ]. However, the mechanism by which the minor allele of the ANKH gene increases the AD risk remains elusive. A recent study on a Chinese cohort demonstrated that ANKH was epigenetically associated with AD, showing lower methylation levels in an intronic CpG site in individuals with mild cognitive impairment or AD than in cognitively healthy controls [ 35 ]. CHRNE , which encodes a subunit of the cholinergic receptor, is also genetically associated with AD. A variant in the 3’-UTR region of CHRNE has been associated with a 1.09-fold increased risk of AD, possibly by enhancing gene expression [ 33 ]. The potential role of CHRNE up-regulation in AD is further supported by evidence that the AD-associated SNP rs113260531, located 0.3 Mb upstream of CHRNE , has been linked to increased CHRNE expression in all four brain regions [ 36 ]. The cholinergic neurotransmitter pathway is a well-established target for AD treatment. Another gene harboring a DMR in its coding region, KLF2 , has also been implicated in AD. KLF2 belongs to the mammalian KLF family, composed of 17 members ( KFF1 - KFF17 ). All KLFs are moderately to highly expressed in retinal ganglion cells and the brain, playing critical roles in neuronal development, regeneration, and brain homeostasis. Immunohistochemical analyses showed a significant reduction in KLF2 in AD brains [ 37 ]. In mouse models, KLF2 expression was found to be down-regulated by Ab 1 − 42 , leading to cerebrovascular dysfunction [ 38 ] and hippocampal neuron injury [ 39 ], both of which are key pathological features of human AD brains. Although limited information is currently available on the functions of the long intergenic non-coding RNA, LINC00908 , apart from its implications in carcinogenesis, it was recently shown to be up-regulated in the temporal cortex of AD brains (Fig. 6 ) [ 40 ]. Three of the six DMRs identified in blood were located in genes associated with AD. Similarly, we previously detected DNA hypomethylation in three of the six AD-related genes examined in AD blood [ 15 ]. This high incidence of methylation changes in AD-related genes in blood suggests that investigating DNA methylation changes in genes may complement genome-wide association studies, particularly for identifying AD-related genes that lack informative DNA variants in or around them. There are a number of limitations that need to be addressed. The generalizability of the DMRs identified across different ethnicities remains unclear because our analysis was conducted on Japanese blood samples. Furthermore, the timing of DMR emergence in AD patients has yet to be clarified because this study used cross-sectional cohorts. Moreover, we did not perform functional analyses of the genes harboring DMRs. Future research is needed to confirm the methylation changes observed in postmortem human brain tissues because previous studies found only a modest correlation between methylation levels in the brain and blood at most CpG sites [ 24 ] [ 41 ]. In addition, brain amyloidosis may have been heterogeneous in subjects in the second and third cohorts because biomarkers did not detect it. To estimate the diagnostic ability of DMRs more accurately, they need to be validated in cohorts that are similar the first cohort, i.e., the PiB-PET-examined cohort. Conclusions We employed MC-seq to isolate AD-associated DNA methylation markers for the first time in this study. Despite the small sample size, we identified two methylation markers that efficiently distinguished AD from controls ( ANKH and MARS ), achieving a discriminative value AUC of 0.81 when combined with the APOE genotype. In future studies, we expect more reliable markers for diagnosing the onset and progression of AD to be identified by applying or modifying the method established in the present study with larger sample sizes and more diverse populations. Abbreviations AD Alzheimer’s disease ANKH ANKilosis Homolog APOE apolipoprotein E AUC area under the ROC curve BA-seq bisulfite amplicon sequencing bp base pair CHRNE Cholinergic Receptor Nicotinic Epsilon Subunit CN cognitively normal CpGs cytosine-phosphate-guanine dinucleotide sites DMP differentially methylated position DMR differentially methylated region GRCh37 Genome Reference Consortium Human Build 37 MC-seq methylation capture sequencing MMSE Mini-Mental State Examination NCGG National Center for Geriatrics and Gerontology NGS next-generation sequencing PiB-PET Pittsburgh compound-B positron emission tomography ROC receiver operator characteristic SD Standard deviation Declarations Ethics approval and consent to participate The present study was conducted with the approval of the Ethical Committee of NCGG (708-8 and 521-4). The design and performance of the present study involving human subjects were clearly described in a research proposal. All participants were voluntary and completed informed consent in writing before registering at the NCGG Biobank, which collects human biomaterials and data for geriatrics research. All DNA and associated information used in this study were obtained from the NCGG Biobank (H27013(2)). For those with substantial cognitive impairment, written informed consent was obtained from the legal next of kin, which the local Ethics Committee had previously approved. Consent for publication Not applicable. Availability of data and materials MC-seq data were deposited in the GEO repository, https://ncbi.nlm.nih.gov/geo/ (GSE24435). The datasets analyzed during this study are available from the corresponding author upon reasonable request. Competing interests The authors declare that they have no competing interests. Funding This work was partly supported by grants from Research Funding for Longevity Sciences from the National Center for Geriatrics and Gerontology (26-16 and 29-20 to NS). Authors’ contributions RM performed all of the core experiments, analyzed the data, and wrote the manuscript. KS and KY analyzed the amplicons of the 2nd and 3rd cohorts. AN and YA conducted PiB-PET. DS, KO, SN, and NS designed the study. NS supervised the study. All authors read and approved the final manuscript. Acknowledgments We thank Yuko Nishimura and the staff of the Biobank of the National Center for Geriatrics and Gerontology for their contribution to this study. References Groen K, Lea RA, Maltby VE, Scott RJ, Lechner-Scott J. Letter to the editor: blood processing and sample storage have negligible effects on methylation. Clin Epigenetics. 2018;10:22. Milicic L, Porter T, Vacher M, Laws SM. Utility of DNA Methylation as a Biomarker in Aging and Alzheimer's Disease. J Alzheimers Dis Rep. 2023;7(1):475-503. Olstad EW, Nordeng HME, Sandve GK, Lyle R, Gervin K. Low reliability of DNA methylation across Illumina Infinium platforms in cord blood: implications for replication studies and meta-analyses of prenatal exposures. Clin Epigenetics. 2022;14(1):80. Sugden K, Hannon EJ, Arseneault L, Belsky DW, Corcoran DL, Fisher HL, et al. Patterns of Reliability: Assessing the Reproducibility and Integrity of DNA Methylation Measurement. Patterns (N Y). 2020;1(2). Logue MW, Smith AK, Wolf EJ, Maniates H, Stone A, Schichman SA, et al. The correlation of methylation levels measured using Illumina 450K and EPIC BeadChips in blood samples. Epigenomics. 2017;9(11):1363-71. Lee EJ, Luo J, Wilson JM, Shi H. Analyzing the cancer methylome through targeted bisulfite sequencing. Cancer Lett. 2013;340(2):171-8. Illumina. TruSeq methyl capture EPIC library prep kit. 2016. Kim SS, Lee SC, Lim B, Shin SH, Kim MY, Kim SY, et al. DNA methylation biomarkers distinguishing early-stage prostate cancer from benign prostatic hyperplasia. Prostate Int. 2023;11(2):113-21. Ohmomo H, Komaki S, Sutoh Y, Hachiya T, Ono K, Arai E, et al. Potential DNA methylation biomarkers for the detection of clear cell renal cell carcinoma identified by a whole blood-based epigenome-wide association study. Epigenetics Communications. 2022;2(1). Allum F, Shao X, Guenard F, Simon MM, Busche S, Caron M, et al. Characterization of functional methylomes by next-generation capture sequencing identifies novel disease-associated variants. Nat Commun. 2015;6:7211. Kacmarczyk TJ, Fall MP, Zhang X, Xin Y, Li Y, Alonso A, et al. "Same difference": comprehensive evaluation of four DNA methylation measurement platforms. Epigenetics Chromatin. 2018;11(1):21. Bernstein DL, Kameswaran V, Le Lay JE, Sheaffer KL, Kaestner KH. The BisPCR(2) method for targeted bisulfite sequencing. Epigenetics Chromatin. 2015;8:27. Li LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. Bioinformatics. 2002;18(11):1427-31. Jonsson H, Sulem P, Kehr B, Kristmundsdottir S, Zink F, Hjartarson E, et al. Parental influence on human germline de novo mutations in 1,548 trios from Iceland. Nature. 2017;549(7673):519-22. Mitsumori R, Sakaguchi K, Shigemizu D, Mori T, Akiyama S, Ozaki K, et al. Lower DNA methylation levels in CpG island shores of CR1, CLU, and PICALM in the blood of Japanese Alzheimer's disease patients. PLoS One. 2020;15(9):e0239196. Ozaki Y, Yoshino Y, Yamazaki K, Sao T, Mori Y, Ochi S, et al. DNA methylation changes at TREM2 intron 1 and TREM2 mRNA expression in patients with Alzheimer's disease. J Psychiatr Res. 2017;92:74-80. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9(4):357-9. Krueger F, Andrews SR. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics. 2011;27(11):1571-2. Akalin A, Kormaksson M, Li S, Garrett-Bakelman FE, Figueroa ME, Melnick A, et al. methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol. 2012;13(10):R87. Breen C, Papale LA, Clark LR, Bergmann PE, Madrid A, Asthana S, et al. Whole genome methylation sequencing in blood identifies extensive differential DNA methylation in late-onset dementia due to Alzheimer's disease. Alzheimers Dement. 2024;20(2):1050-62. Lee S, Emond MJ, Bamshad MJ, Barnes KC, Rieder MJ, Nickerson DA, et al. Optimal Unified Approach for Rare-Variant Association Testing with Application to Small-Sample Case-Control Whole-Exome Sequencing Studies. The American Journal of Human Genetics. 2012;91(2):224-37. Benjamini YH, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289-300. Strittmatter WJ, Saunders AM, Schmechel D, Pericak-Vance M, Enghild J, Salvesen GS, et al. Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc Natl Acad Sci U S A. 1993;90(5):1977-81. Zhang L, Silva TC, Young JI, Gomez L, Schmidt MA, Hamilton-Nelson KL, et al. Epigenome-wide meta-analysis of DNA methylation differences in prefrontal cortex implicates the immune processes in Alzheimer's disease. Nat Commun. 2020;11(1):6114. Komaki S, Nagata M, Arai E, Otomo R, Ono K, Abe Y, et al. Epigenetic profile of Japanese supercentenarians: a cross-sectional study. Lancet Healthy Longev. 2023;4(2):e83-e90. Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, Onyango P, et al. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet. 2009;41(2):178-86. Seale K, Horvath S, Teschendorff A, Eynon N, Voisin S. Making sense of the ageing methylome. Nat Rev Genet. 2022;23(10):585-605. Neri F, Rapelli S, Krepelova A, Incarnato D, Parlato C, Basile G, et al. Intragenic DNA methylation prevents spurious transcription initiation. Nature. 2017;543(7643):72-7. Yearim A, Gelfman S, Shayevitch R, Melcer S, Glaich O, Mallm JP, et al. HP1 is involved in regulating the global impact of DNA methylation on alternative splicing. Cell Rep. 2015;10(7):1122-34. Shirai M, Nara T, Takahashi H, Takayama K, Chen Y, Hirose Y, et al. Identification of aberrant transcription termination at specific gene loci with DNA hypomethylated transcription termination sites caused by DNA methyltransferase deficiency. Genes Genet Syst. 2022;97(3):139-52. Nuñez JK, Chen J, Pommier GC, Cogan JZ, Replogle JM, Adriaens C, et al. Genome-wide programmable transcriptional memory by CRISPR-based epigenome editing. Cell. 2021;184(9):2503-19.e17. Bellenguez C, Kucukali F, Jansen IE, Kleineidam L, Moreno-Grau S, Amin N, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet. 2022;54(4):412-36. de Rojas I, Moreno-Grau S, Tesi N, Grenier-Boley B, Andrade V, Jansen IE, et al. Common variants in Alzheimer's disease and risk stratification by polygenic risk scores. Nat Commun. 2021;12(1):3417. Tesi N, van der Lee S, Hulsman M, van Schoor NM, Huisman M, Pijnenburg Y, et al. Cognitively healthy centenarians are genetically protected against Alzheimer's disease. Alzheimers Dement. 2024;20(6):3864-75. Wu S, Yang F, Chao S, Wang B, Wang W, Li H, et al. Altered DNA methylome profiles of blood leukocytes in Chinese patients with mild cognitive impairment and Alzheimer's disease. Front Genet. 2023;14:1175864. He L, Loika Y, Kulminski AM. Allele-specific analysis reveals exon- and cell-type-specific regulatory effects of Alzheimer's disease-associated genetic variants. Transl Psychiatry. 2022;12(1):163. Fang X, Zhong X, Yu G, Shao S, Yang Q. Vascular protective effects of KLF2 on Abeta-induced toxicity: Implications for Alzheimer's disease. Brain Res. 2017;1663:174-83. Wu C, Li F, Han G, Liu Z. Abeta(1-42) disrupts the expression and function of KLF2 in Alzheimer's disease mediated by p53. Biochem Biophys Res Commun. 2013;431(2):141-5. Duan Q, Si E. MicroRNA-25 aggravates Abeta1-42-induced hippocampal neuron injury in Alzheimer's disease by downregulating KLF2 via the Nrf2 signaling pathway in a mouse model. J Cell Biochem. 2019;120(9):15891-905. Chen C, Zhang Z, Liu Y, Hong W, Karahan H, Wang J, et al. Comprehensive characterization of the transcriptional landscape in Alzheimer's disease (AD) brains. Sci Adv. 2025;11(1):eadn1927. Yu L, Chibnik LB, Yang J, McCabe C, Xu J, Schneider JA, et al. Methylation profiles in peripheral blood CD4+ lymphocytes versus brain: The relation to Alzheimer's disease pathology. Alzheimers Dement. 2016;12(9):942-51. Additional Declarations No competing interests reported. Supplementary Files AdditionalFile1.xlsx Supplementaryinformation.docx Cite Share Download PDF Status: Published Journal Publication published 20 Jun, 2025 Read the published version in Clinical Epigenetics → Version 1 posted Editorial decision: Revision requested 10 Apr, 2025 Reviews received at journal 04 Apr, 2025 Reviewers agreed at journal 04 Apr, 2025 Reviews received at journal 01 Apr, 2025 Reviewers agreed at journal 01 Apr, 2025 Reviewers agreed at journal 25 Mar, 2025 Reviewers invited by journal 24 Mar, 2025 Editor assigned by journal 22 Mar, 2025 Submission checks completed at journal 21 Mar, 2025 First submitted to journal 11 Mar, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6200381","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":436896768,"identity":"00daa89f-ff7b-43b6-be49-1fc4f48cdd1b","order_by":0,"name":"Risa Mitsumori","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Risa","middleName":"","lastName":"Mitsumori","suffix":""},{"id":436896769,"identity":"17809e8d-d78a-4785-a5f7-887d2469fef6","order_by":1,"name":"Kayoko Sawamura","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Kayoko","middleName":"","lastName":"Sawamura","suffix":""},{"id":436896770,"identity":"8c6eeada-8067-41a8-a6dd-8323e3775a8e","order_by":2,"name":"Kimi Yamakoshi","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Kimi","middleName":"","lastName":"Yamakoshi","suffix":""},{"id":436896771,"identity":"34b9d699-e8f6-4ed7-beb9-da8eb0e78fb0","order_by":3,"name":"Akinori Nakamura","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Akinori","middleName":"","lastName":"Nakamura","suffix":""},{"id":436896772,"identity":"228851e2-a110-4764-842e-24586795cde3","order_by":4,"name":"Yutaka Arahata","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Yutaka","middleName":"","lastName":"Arahata","suffix":""},{"id":436896773,"identity":"5fef0a4d-0939-40a8-a268-2d5025b3b56b","order_by":5,"name":"Shumpei Niida","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Shumpei","middleName":"","lastName":"Niida","suffix":""},{"id":436896774,"identity":"51920074-f0f0-4654-850b-dd814e1a68ad","order_by":6,"name":"Daichi Shigemizu","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Daichi","middleName":"","lastName":"Shigemizu","suffix":""},{"id":436896775,"identity":"fbdd7124-e8f5-4b1d-90e1-baae4bf112ee","order_by":7,"name":"Kouichi Ozaki","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Kouichi","middleName":"","lastName":"Ozaki","suffix":""},{"id":436896777,"identity":"540fab48-20f1-4227-8c34-dff8330d79b7","order_by":8,"name":"Nobuyoshi Shimoda","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0klEQVRIiWNgGAWjYDACCcYGBgY2Bjm4AIhPlBZjUrSACDaGRIIK4UB+dnPrhg9ldunb2Y9ffsHwy4aBeTYB3QZ3DrbdnHEuOXdnT06ZBWNfGgPjnAMEtEgktt3mbWPO3XCDJ82AsecwA+OMBAIOmwHWUp9uQLQWhhtgLYcTDG6wH37A8IMILQZALUC/HDfccCYHFG5pPAT9Ij8j/dmND2XV8gbHjz/+8OGPjZwhoRBDAjxmwKBg4DGcQbQOBvbHHxj+AO2VIF7LKBgFo2AUjAwAAKQnS7ZZ4z0xAAAAAElFTkSuQmCC","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":true,"prefix":"","firstName":"Nobuyoshi","middleName":"","lastName":"Shimoda","suffix":""}],"badges":[],"createdAt":"2025-03-11 05:53:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6200381/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6200381/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13148-025-01905-0","type":"published","date":"2025-06-20T15:57:33+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":79824512,"identity":"5bdd1994-d427-48c8-a808-5a23b9be4ce1","added_by":"auto","created_at":"2025-04-03 09:21:05","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":63817,"visible":true,"origin":"","legend":"\u003cp\u003eAn image of our sliding-window approach to define DMRs.\u003c/p\u003e\n\u003cp\u003eIn this study, CpGs that meet either of the following two conditions are defined as constituting a DMR: at least five or three CpG sites of |\u003cem\u003eΔ\u003c/em\u003eb| ≥15% (A) or |\u003cem\u003eΔ\u003c/em\u003eb| ≥25% (B) in 1 kb, respectively. We regard consecutive DMRs, as shown above, as one DMR. \u003cem\u003eΔ\u003c/em\u003ebis the average methylation level (%) difference at the same CpG site between the CN and AD genomes.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/475ffba916914e59b53f1989.jpg"},{"id":79824517,"identity":"32ffc6a8-a00f-419c-9254-8cff6995208e","added_by":"auto","created_at":"2025-04-03 09:21:05","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":50608,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart of analysis step\u003c/p\u003e\n\u003cp\u003eSmall-scale discovery analysis of the first cohort followed by replication in two independent cohorts.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/32910cb197d99d2460f91f2e.jpg"},{"id":79826370,"identity":"fb99a0a4-f721-4a93-9fed-b7c9fb0c02a2","added_by":"auto","created_at":"2025-04-03 09:37:05","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":49539,"visible":true,"origin":"","legend":"\u003cp\u003eCharacteristics of DMRs\u003c/p\u003e\n\u003cp\u003eMC-seq of the first cohort showed the unbiased chromosomal distribution of DMRs. (A) Chromosomal\u003cstrong\u003e \u003c/strong\u003edistribution of DMRs showing methylation differences ≥15%. Numbers in parentheses are those ≥25%. (B) Hypo- and hypermethylation ratios of DMRs. (C) Locations of DMRs relative to genes.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/6a0d24057083f9cded331309.jpg"},{"id":79824520,"identity":"6bc4c103-02ec-4151-960b-2e067d398ff4","added_by":"auto","created_at":"2025-04-03 09:21:05","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":59825,"visible":true,"origin":"","legend":"\u003cp\u003eDifferences in DNA methylation levels in 6 DMRs in the second cohort\u003c/p\u003e\n\u003cp\u003eThe schematics depict the distributions of CpG dinucleotides along the five genes and a long intergenic non-coding (\u003cem\u003eLINC\u003c/em\u003e) RNA in which DMRs were found. Vertical lines show the positions of CpG dinucleotides. Open rectangles indicate axons. Underlined were the regions for which the methylation levels were analyzed by bisulfite amplicon sequencing. The five genes and a \u003cem\u003eLINC\u003c/em\u003e harboring a DMR were as follows: \u003cem\u003eANKH\u003c/em\u003e (A), \u003cem\u003eMARS\u003c/em\u003e (B), \u003cem\u003eANKFY1\u003c/em\u003e (C), \u003cem\u003eCHRNE\u003c/em\u003e(D), \u003cem\u003eKLF2\u003c/em\u003e (E), and \u003cem\u003eLINC00908\u003c/em\u003e(F).\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/9b44a98d751ff6bad62aeb49.jpg"},{"id":79825850,"identity":"cf5a408c-17a4-4de5-b0a1-1bebf5bbe8ab","added_by":"auto","created_at":"2025-04-03 09:29:05","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":104021,"visible":true,"origin":"","legend":"\u003cp\u003eThe flowchart for machine learning and receiver operator curves.\u003c/p\u003e\n\u003cp\u003e(A) Flowchart of the machine learning process to discover the best combination of DMRs. ROCs were generated with\u003cem\u003e \u003c/em\u003ethe combination of \u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eMARS\u003c/em\u003e methylation and the \u003cem\u003eAPOE\u003c/em\u003egenotype (B) and with only the \u003cem\u003eAPOE\u003c/em\u003egenotype (C) in the discovery (orange) and validation (blue) cohorts.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/b2aa9d37c819a5183cf3de6d.jpg"},{"id":79825852,"identity":"a0cb3c2a-a14a-424a-b326-5d508b221317","added_by":"auto","created_at":"2025-04-03 09:29:05","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":17579,"visible":true,"origin":"","legend":"\u003cp\u003eLINC00908 expression increased in the temporal cortex of AD brains. The p-value is shown at the top. The image was retrieved from the ADAtlas. https://hanlaboratory.com/AD_atlas/ accessed 18 Jan. 2025. \u0026nbsp;RPM: Reads Per Million.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/39971ca0083aa2b272efb727.jpg"},{"id":85231436,"identity":"b088ec48-b91b-40b1-8fd6-fbd38e05967a","added_by":"auto","created_at":"2025-06-23 16:08:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1436594,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/37043dcb-50c5-488a-b37c-25984ad529d6.pdf"},{"id":79824519,"identity":"23930aca-d9b1-4162-97f5-2612a87c7063","added_by":"auto","created_at":"2025-04-03 09:21:05","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":56117,"visible":true,"origin":"","legend":"","description":"","filename":"AdditionalFile1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/e1a383dab24a50165e92faea.xlsx"},{"id":79825848,"identity":"a22c522b-b559-43bc-90d7-b8d28f183963","added_by":"auto","created_at":"2025-04-03 09:29:05","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":17995,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6200381/v1/3cbf95981febac4859956057.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identification of diagnostic DNA methylation markers in the blood of Japanese Alzheimer’s disease patients using methylation capture sequencing","fulltext":[{"header":"Background","content":"\u003cp\u003eIn vertebrate genomes, cytosine bases at cytosine-phosphate-guanine dinucleotide sites (CpGs) are the primary substrates for methylation by DNA methyltransferases. Due to its stability in blood samples [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], susceptibility to environmental factors, and regulatory roles in gene expression, DNA methylation is a promising biomarker and a potential key to identifying genes associated with complex diseases, including Alzheimer\u0026rsquo;s disease (AD). Many CpG sites associated with AD have been identified by comprehensive microarray DNA methylation analyses and targeted gene approaches using blood- and brain-derived DNA [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. However, few CpGs have been replicated across independent studies [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. This lack of reproductivity may be attributed to unresolved limitations in microarray methylation analyses [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Furthermore, microarrays capture only 3% of the 28\u0026nbsp;million CpGs in the human genome, suggesting that many AD-associated CpGs have yet to be discovered. The identification of hidden CpGs in blood samples will contribute to the more accurate prognostication and diagnosis of AD as well as a more detailed understanding of its etiology.\u003c/p\u003e \u003cp\u003eTo discover novel and reproducible DNA methylation changes associated with AD, we employed methylation capture sequencing (MC-seq), also known as targeted bisulfite sequencing, an alternative method originally named solution hybrid selection bisulfite sequencing [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Using next-generation sequencing (NGS), MC-seq directly quantifies methylation levels at individual CpG sites in DNA fragments captured by biotinylated RNA or DNA probes designed for specific genomic regions. In this study, we selected the TruSeq Methyl Capture EPIC library prep (TruSeq EPIC) from Illumina Inc., which enables methylation quantification at more than 3.3\u0026nbsp;million CpG sites per individual, representing 12% of all CpGs in the human genome. Capture probes target 107 Mb of genomic DNA (gDNA), covering key regulatory elements, such as promoters, enhancers, CpG islands, and CpG island shores/selves [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. MC-seq has already successfully identified a number of disease-related markers [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]; however, to the best of our knowledge, it has not yet been applied to AD research. Although MC-seq is more cost effective than whole-genome bisulfite sequencing [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], the processing of large sample sets remains expensive. Therefore, in the present study, we used MC-seq to identify differentially methylated regions (DMRs) in the first cohort composed of 12 AD patients (Pittsburgh compound-B positron emission tomography (PiB-PET)-positive) and 12 matched cognitively normal (CN) elderly individuals (PiB-PET-negative). Candidate DMRs detected through MC-seq were validated in two independent cohorts of AD and CN individuals using whole-blood DNA and bisulfite amplicon sequencing (BA-seq), another NGS-based method [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSubjects\u003c/h2\u003e \u003cp\u003eThe cognitive status and dementia severity of subjects were assessed using the Mini-Mental State Examination (MMSE) by trained physicians at the National Center for Geriatrics and Gerontology (NCGG) hospital. The first cohort was composed of 12 AD cases and 12 CN elderly individuals, with and without brain amyloidosis, as detected by PiB-PET scanning at the NCGG hospital. In a replication analysis, we prepared second and third cohorts, each consisting of 48 clinically diagnosed AD cases and 48 CN individuals, all aged 60 years or older. Clinical information, including MMSE scores, sex, age, and the apolipoprotein E (\u003cem\u003eAPOE\u003c/em\u003e) epsilon (e) allele genotype, was obtained from the NCGG biobank.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePreparation of gDNA\u003c/h3\u003e\n\u003cp\u003egDNA was extracted from peripheral blood using a Maxwell RSC Instrument (Promega, USA) with the Maxwell RSC Buffy Coat DNA Kit (Promega, USA). The quantity of gDNA was measured using the Quant-iT 1\u0026times;dsDNA BR Assay (Thermo Fisher Scientific, USA), and the intactness of gDNA was assessed by electrophoresis on 1.0% agarose gels.\u003c/p\u003e\n\u003ch3\u003eSodium bisulfite treatment of DNA\u003c/h3\u003e\n\u003cp\u003eIn a methylation analysis, 200 ng of DNA samples was treated with sodium bisulfite to convert methylated cytosine to cytosine and demethylated cytosine to uracil, either by the EZ DNA methylation Gold Kit or EZ-96 DNA Methylation-Gold Kit (Zymo Research, USA) according to the manufacturer\u0026rsquo;s instructions.\u003c/p\u003e\n\u003ch3\u003eMC-seq\u003c/h3\u003e\n\u003cp\u003eWe processed twenty-four samples of gDNA from the first cohort using the target enrichment system, the TruSeq methyl capture EPIC Kit (Illumina, Inc., CA, USA). Captured DNA fragments were sequenced by NGS as follows: gDNA was fragmented to a few hundred base pairs (bps) by a sonicator (M220, Covaris), followed by blunting, phosphorylation, the addition of 3\u0026rsquo;-dA, and the ligation of indexed adaptors to both ends of DNA. DNA was mixed and hybridized to capture oligos, and streptavidin-conjugated magnetic beads recovered hybridized DNA. Eluted DNA from magnetic beads was treated with sodium bisulfite and amplified by PCR to construct a sequencing library, which was sequenced using the Illumina HiSeq 2500 platform with paired-end reads of 100 bp according to the manufacturer\u0026rsquo;s instructions. Library construction and sequencing were conducted by Takara Bio, Inc. (Shiga, Japan). Raw sequence data were converted from a bcl file to a fastq file using Illumina bcl2fastq2 Conversion software v2.17. Sequencing data were aligned to the reference sequence (Genome Reference Consortium Human Build 37 (GRCh37)/UCSCgh19) using the application, MethylSeq, through the cloud service named BaseSpace (Illumina, Inc., CA, USA). The numbers of methylated and unmethylated cytosines at each CpG site, namely, the methylation call, were counted by MethylKit, another application available through BaseSpace. The same application used methylation call data to detect differential DNA methylation between AD subjects and CN elderly. We set the conditions of MethylKit for differentially methylated CpG sites to be identified as follows: \u0026ge;10 coverage with independent reads and \u0026ge;15% methylation differences with a \u003cem\u003eq\u003c/em\u003e-value \u0026le;0.01.\u003c/p\u003e\n\u003ch3\u003eDefinition of DMR\u003c/h3\u003e\n\u003cp\u003eGenomic regions where differentially methylated cytosines cluster are referred to as DMRs, although the criteria for defining DMR may be arbitrary. In the first cohort, we identified DMRs using a sliding window approach, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. We initially searched for genomic regions containing at least five or three cytosines with consistent methylation differences (\u003cem\u003eD\u003c/em\u003eb \u0026ge;0.15 or \u003cem\u003eD\u003c/em\u003eb \u0026ge;0.25, respectively) within a 1-kb window. The window was then slid up to 1 kb in either or both directions as long as the new window contained a combination of differentially methylated cytosines that met the above criteria. A region was defined as a DMR if even a single window met these conditions.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePCR primers\u003c/h2\u003e \u003cp\u003ePCR primers for bisulfite-treated gDNA were designed by MethPrimer [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] or by Pyrosequencing Assay Design Software ver. 2.0 provided by Qiagen (CA, USA) for pyrosequencing.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePreparation of amplicons from the second cohort\u003c/h3\u003e\n\u003cp\u003eOne or multiple regions of each of the 88 DMR candidates were amplified from each subject\u0026rsquo;s DNA in the second cohort by using KOD Multi \u0026amp; Epi (Toyobo, inc., Shiga, Japan) in 20 \u0026micro;l of the reaction mixture with the following touchdown PCR conditions: at 94℃ for 2 min; 4 cycles at 98\u0026deg;C for 10 sec, 64\u0026deg;C for 30 sec, and 68\u0026deg;C for 15 sec; 4 cycles at 98\u0026deg;C for 10 sec, 60\u0026deg;C for 30 sec, and 68\u0026deg;C for 15 sec; 4 cycles at 98\u0026deg;C for 10 sec, 58\u0026deg;C for 30 sec, and 68\u0026deg;C for 15 sec; 30 cycles at 98\u0026deg;C for 10 sec, 55\u0026deg;C for 30 sec, and 68\u0026deg;C for 15 sec. The average length of amplicons was 237 bp. All amplicons were electrophoresed on 2% agarose gels with 0.5\u0026times;TAE buffer, followed by staining with GelRed (Biotium, USA), and PCR was repeated for unsuccessful amplification. We collected 1 \u0026micro;l from each of the 133 amplicons from single subjects and mixed the aliquots in single tubes. Pooled PCR products were then purified by QIAvac 96 (QIAGEN, CA, USA), which removes low-molecular weight DNA (\u0026le;80 bp). We electrophoresed 1 \u0026micro;l of each eluate on a 2% agarose gel to check the recovery of pooled PCR products. The concentrations of purified pools of amplicons were measured using a 2200 TapeStation Instrument (Agilent Technologies, Santa Clara, CA, USA).\u003c/p\u003e\n\u003ch3\u003ePreparation of amplicons from the third cohort\u003c/h3\u003e\n\u003cp\u003eOne or multiple regions of each of the six DMRs identified in the present study were amplified from each subject in the third cohort, and all amplicons were assessed using the method described above. Based on the amplification level judged from agarose gel electrophoresis of each amplicon, we took either 1 or 4 \u0026micro;l from each amplicon per subject and pooled the aliquots into single tubes. The final volume of each pooled sample was adjusted to 100 \u0026micro;l with distilled water. Purification of the mixture, followed by agarose gel electrophoresis and quantification of the purified mixture were performed as described above.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eBA-seq\u003c/h2\u003e \u003cp\u003eWe constructed libraries for BA-seq from 80 ng of purified amplicon mixtures by using TruSeq Nano DNA Library Prep Kits (Illumina, USA) following the manufacturer\u0026rsquo;s instructions, except for the enrichment process of DNA fragments. We used AMPure XP (Beckman Coulter, USA) instead of the sample purification beads included in the kit. We started from the middle of the instructions because the fragmentation of DNA was unnecessary for PCR amplicons. We started with the reaction of \u0026ldquo;A-tailing\u0026rdquo;, which adds adenine to the ends of amplicons. The molarity of the libraries and the degree of the undesirable inclusion of adaptor dimers in the libraries, which appeared around 150 bp, were assessed using the Agilent D1000 screen tapes system on a 2200 TapeStation system (Agilent, USA). A 4 nM pool of libraries and 4 nM PhiX Control v3 (Illumina, USA) were each denatured for 5 min with freshly prepared 0.2 N NaOH and adjusted at concentrations of 6 and 8 pM, respectively. Denatured pools were mixed with denatured Phi X at volume ratios of 85:15 and 70:30 for the second and third cohort DNA amplicon libraries, respectively, and 600 \u0026micro;l of each of the prepared samples was loaded into the reagent cartridge of the MiSeq Reagent Kit v2 (Illumina, USA). The cartridge sets on the Illumina MiSeq platform were used to sequence libraries with 151 paired-end, dual-indexing cycles per read (2 \u0026times; 151). The densities of clusters generated on a flow cell were 322\u0026thinsp;\u0026plusmn;\u0026thinsp;8 and 343\u0026thinsp;\u0026plusmn;\u0026thinsp;10 k/mm\u003csup\u003e2\u003c/sup\u003e for BA-seq of the second and third cohorts, respectively.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eData analysis of BA-Seq\u003c/h2\u003e \u003cp\u003eWe trimmed adaptor sequences from read sequences using Trim Galore (version 0.6.6), setting the trimming cut-off at 20 (-q) and using the rrbs option (--rrbs). After trimming, sequencing data were mapped to the human reference genome (GRCh37) using Bismark (version 0.22.3) with the \u0026ldquo;genome_preparation\u0026rdquo; function, an aligner optimized for bisulfite sequence data and methylation calling. The BAM files generated were sorted by chromosomes using SAMTools (\u0026ldquo;sort\u0026rdquo;). Sorted BAM files were then utilized for methylation calling and the calculation of methylation rates using the methylKit package in R (genomeBismarkAln), with the following settings: human genome assembly \u0026lsquo;hg19\u0026rsquo;, read context \u0026lsquo;CpG\u0026rsquo;, and a minimum coverage cut-off of \u0026lsquo;10\u0026rsquo;. By using the DNA methylation rates (meC/meC\u0026thinsp;+\u0026thinsp;C) of CpGs in DMRs as explanatory valuables, the significance of methylation differences at DMRs was evaluated using the burden test (SKAT in R), adjusting for sex and age (kernel=\u0026ldquo;liner.weighted\u0026rdquo;).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eSelection of best DMR sets\u003c/h2\u003e \u003cp\u003eWe selected CpG sites with methylation differences between CN and AD that were significant (\u003cem\u003eP\u003c/em\u003e\u003csub\u003e\u003cem\u003ebon\u003c/em\u003e\u003c/sub\u003e \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\le\\:\\:\\)\u003c/span\u003e\u003c/span\u003e0.05) in the second and third cohort analyses and that showed the same direction of methylation change. Second and third cohort subjects were mixed and randomly split: four-fifths were used for a discovery dataset and one-fifth for a validation dataset, using sklearn (\u0026ldquo;train_test_split\u0026rdquo;) in Python3. Using discovery data, prediction models were constructed based on clinical information (age, sex, and \u003cem\u003eAPOE\u003c/em\u003e ε\u003csub\u003e4\u003c/sub\u003e genotypes), and all possible combinations of DMRs using a logistic regression model. The optimal model, incorporating clinical information and DMRs, in the discovery dataset was selected using a 3-fold cross-validation. After finalizing the model on the entire discovery dataset, its performance was evaluated on an independent validation dataset, with the area under the receiver operator characteristic (ROC) curve (AUC) being used to assess discriminative accuracy. Analyses were implemented using the sklearn library in Python (penalty: \u0026ldquo;elasticnet\u0026rdquo;, solver: \u0026ldquo;saga\u0026rdquo;, 1l_ratio: 0.5). Sensitivity to the values of the regularization parameter (C parameter (0.1, 01, 10, 100)) was examined with the help of simulating by the best model.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eDetection of candidate DMRs in the first cohort\u003c/h2\u003e \u003cp\u003eThe present study focused on DMRs, clusters of neighboring CpGs with significant methylation changes in the same direction. There were two main reasons for this approach. Cytosine-to-thymine transitions at CpG sites, the most frequent \u003cem\u003ede novo\u003c/em\u003e mutation occurring in the human genome [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], are indistinguishable from unmethylated cytosines after a bisulfite treatment, potentially leading to false differentially methylated positions (DMPs). To prevent this, we considered only DMPs aggregated in a short distance, defining them as DMRs for use as diagnostic markers. Furthermore, the utility of DMRs in blood DNA as diagnostic markers for AD has already been demonstrated [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur approach to comprehensively identify DMRs in the blood DNA of AD patients is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. We first directly compared sequences from the first cohort, composed of 12 CN and 12 AD individuals, using the commercially available TruSeq Methyl Capture EPIC kit (Illumina Inc.). All donors in this cohort underwent PiB-PET imaging, with 12 CN and 12 AD individuals testing negative and positive for brain amyloidosis, respectively (Table\u0026nbsp;1).\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;1. Sample characteristics of three cohorts used in this study\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e1st cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCN-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAD+\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eCN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eAD\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSamples\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4 (33.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e24 (50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e24 (50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e24 (50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e24 (50)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e73.3\u0026thinsp;\u0026plusmn;\u0026thinsp;3.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e74.4\u0026thinsp;\u0026plusmn;\u0026thinsp;6.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e70.3\u0026thinsp;\u0026plusmn;\u0026thinsp;2.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e71.2\u0026thinsp;\u0026plusmn;\u0026thinsp;4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e69.8\u0026thinsp;\u0026plusmn;\u0026thinsp;3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e71.8\u0026thinsp;\u0026plusmn;\u0026thinsp;2.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMMSE score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29.0\u0026thinsp;\u0026plusmn;\u0026thinsp;1.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21\u0026thinsp;\u0026plusmn;\u0026thinsp;3.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e28.8\u0026thinsp;\u0026plusmn;\u0026thinsp;1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16.2\u0026thinsp;\u0026plusmn;\u0026thinsp;3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e29.5\u0026thinsp;\u0026plusmn;\u0026thinsp;0.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e16.9\u0026thinsp;\u0026plusmn;\u0026thinsp;4.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\epsilon\\:\\)\u003c/span\u003e\u003c/span\u003e4 homozygote\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5 (10.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1 (2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e6 (12.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\epsilon\\:\\)\u003c/span\u003e\u003c/span\u003e4 heterozygote\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (8.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8 (66.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5 (10.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e30 (62.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e10 (20.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e19 (39.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\epsilon\\:\\)\u003c/span\u003e\u003c/span\u003e4 non-carriers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11 (91.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e42 (87.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e13 (27.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e37 (77.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e23 (47.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePiB-PET\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003enegative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003epositive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAge and MMSE are expressed as the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD, while the other variables are expressed as numbers, n (%). Abbreviations: SD, standard deviation; MMSE, Mini-Mental State Examination; \u003cem\u003eAPOE\u003c/em\u003e, apolipoprotein E; -, not executed.\u003c/p\u003e \u003cp\u003eA total of 437,792 target DNA fragments were recovered from the first cohort, yielding 107,403,887 bp of sequence data. These fragments were mapped to UCSDhg19 using Bowtie2 [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e], with an average mapping rate of 83.9% (CN: 84.07\u0026thinsp;\u0026plusmn;\u0026thinsp;0.45%; AD: 83.81% \u0026plusmn; 0.37). The methylation rate at each CpG site was calculated by Bismark [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], which identifies methylated and unmethylated cytosines in sequenced reads. Illumina integrates both programs into the online application, MethylSeq, available through BaseSpace Sequence Hub, which was used in the present study. A total of 3,233,016 CpG sites were analyzed, covering 96.8% of the 3,340,894 CpGs detectable by the TruSeq Methyl Capture EPIC kit. These results indicated the successful construction of an enrichment library and sequencing (Table\u0026nbsp;2). An aggregated summary and raw data of MC-seq are shown in Table\u0026nbsp;3 and supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, respectively.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003eTable\u0026nbsp;2. Expected and actual recovery of target regions and CpGs from the first cohort\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eIllumina TruSeq Methyl Capture EPIC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTemplate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eExpected\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eActual\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eThe human genome\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal length of targets (Mb)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e107\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e107\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3,000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of target regions (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e437,792 (100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e436,268 (99.7%\u003csup\u003ea\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of CpGs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,340,892 (11.8%\u003csup\u003ec\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3,233,016 (96.8%\u003csup\u003eb\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,8217,009 (100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c6\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003eThe ratio of recovered targets\u003csup\u003ea\u003c/sup\u003e or CpGs\u003csup\u003eb\u003c/sup\u003e to the expected number of targets or CpGs to be recovered from the experimental design. The ratio of the expected number of CpGs to be analyzed\u003csup\u003ec\u003c/sup\u003e to the total number of CpGs in the human genome.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;3. Statistics of methyl capture sequencing of the first cohort\u003c/p\u003e\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabc\" border=\"1\"\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCN-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAD+\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal PF reads\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e135879157\u0026thinsp;\u0026plusmn;\u0026thinsp;11735593\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e135600493\u0026thinsp;\u0026plusmn;\u0026thinsp;22796635\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAligned reads\u003csup\u003eb\u003c/sup\u003e, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e84.07\u0026thinsp;\u0026plusmn;\u0026thinsp;0.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e83.81\u0026thinsp;\u0026plusmn;\u0026thinsp;0.37\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDuplicate aligned reads\u003csup\u003ec\u003c/sup\u003e, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.42\u0026thinsp;\u0026plusmn;\u0026thinsp;3.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10.49\u0026thinsp;\u0026plusmn;\u0026thinsp;1.70\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnique read enrichment\u003csup\u003ed\u003c/sup\u003e, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e79.84\u0026thinsp;\u0026plusmn;\u0026thinsp;0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e79.17\u0026thinsp;\u0026plusmn;\u0026thinsp;1.87\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTarget coverage at least 20\u0026times;, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e93.60\u0026thinsp;\u0026plusmn;\u0026thinsp;1.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93.56\u0026thinsp;\u0026plusmn;\u0026thinsp;3.12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCytosine methylation in the CpG context, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52.66\u0026thinsp;\u0026plusmn;\u0026thinsp;0.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e51.91\u0026thinsp;\u0026plusmn;\u0026thinsp;0.62\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCytosine methylation in the non-CpG context, %\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.10\u0026thinsp;\u0026plusmn;\u0026thinsp;0.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.98\u0026thinsp;\u0026plusmn;\u0026thinsp;0.03\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eCN- and AD\u0026thinsp;+\u0026thinsp;indicate cognitively normal elderly negative for amyloid PET and clinically diagnosed AD cases positive for amyloid PET, respectively. \u003csup\u003ea\u003c/sup\u003eThe total number of pass-filter reads for the sample. \u003csup\u003eb\u003c/sup\u003eThe percentage of pass-filter reads that aligned to the reference genome. \u003csup\u003ec\u003c/sup\u003eThe percentage of paired-end reads flagged as duplicates. \u003csup\u003ed\u003c/sup\u003e100\u0026times;(Targeted unique aligned reads/Unique aligned reads). Variables are expressed as the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD.\u003c/p\u003e \u003cp\u003eWe used MethylKit [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], accessible through BaseSpace Sequence Hub, to identify DMPs that may constitute DMRs under the following stringent settings: (1) CpGs with a minimum coverage of 10 reads (coverage\u0026thinsp;\u0026ge;\u0026thinsp;10), (2) CpGs showing a methylation difference of at least 15% (|\u003cem\u003eD\u003c/em\u003eb| \u0026ge;15%) between CN and AD, and (3) CpGs with a \u003cem\u003eq\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.01. We obtained 10,381 DMPs, including 1,077 DMPs with |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;25%. After excluding 124 and 1 DMPs from the X and Y chromosomes, respectively, we retained 10,256 DMPs of |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;15%, of which 1,065 had |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;25%.\u003c/p\u003e \u003cp\u003eTo detect DMRs, we applied a sliding window analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), defining DMRs as regions containing at least seven DMPs with |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;15% or five DMPs with |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;25% within a genomic region of a few kb. We identified 106 DMRs, including 6 DMRs composed entirely of |\u003cem\u003eD\u003c/em\u003eb| \u0026ge;25% DMPs (supplementary Table S2). Of these, 86 DMRs (81%) were hypomethylated (CN\u0026thinsp;\u0026gt;\u0026thinsp;AD), while 20 (19%) were hypermethylated (CN\u0026thinsp;\u0026lt;\u0026thinsp;AD) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The predominance of hypomethylation in AD blood samples is consistent with previous findings [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Detailed information on the positions and lengths of DMRs is shown in Supplementary Table S2.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eValidation of DMRs with the second cohort\u003c/h2\u003e \u003cp\u003eTo validate methylation differences in the 106 AD-related DMRs detected from the first cohort, we analyzed a second cohort of 48 AD cases and 48 CN individuals (none of whom underwent PiB-PET imaging, Table\u0026nbsp;1) by BA-seq.\u0026nbsp;Primer sets were successfully designed for 91 DMRs, while the remaining 15 DMRs could not be targeted (Supplementary Table S3).\u003c/p\u003e \u003cp\u003ePrimer specificity was initially examined to establish whether they uniquely PCR amplified DNA of the expected sizes from control bisulfite-treated human gDNA by electrophoresis of PCR products on 2% agarose gels. Alternative primer sets were tested for unsuccessful amplifications; however, the expected PCR products were not obtained for 3 DMRs (Supplementary Table S3). In total, 133 amplicons were ultimately generated against 88 DMRs (Supplementary Table S4), and a bisulfite-amplicon DNA library was constructed for paired-end sequencing.\u003c/p\u003e \u003cp\u003eThe genomic locations of the DMRs, their relative positions within genes, primer sequences, amplicon lengths, and other details are shown in Supplementary Table S4. Since bisulfite-treated DNA and PCR amplicons significantly reduce library diversity, we spiked the sample with 15% PhiX control and reduced the library concentration to achieve a lower cluster density on a flow cell than recommended for standard DNA sequencing. The MiSeq sequencing run yielded a cluster density of 322\u0026thinsp;\u0026plusmn;\u0026thinsp;8 (K/mm\u003csup\u003e2\u003c/sup\u003e), with 82.12\u0026thinsp;\u0026plusmn;\u0026thinsp;2.96 (%) of the cluster passing filter, and a \u0026ge;Q30 score of 91.0%. The average read numbers per sample were similar between the CN and AD groups: 38,977 and 38,656, respectively (Table\u0026nbsp;4).\u003c/p\u003e\u003cp\u003eTable\u0026nbsp;4. Averaged results of bisulfite amplicon sequencing\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabd\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAD\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal number of pair-end reads\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38977\u0026thinsp;\u0026plusmn;\u0026thinsp;4111.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e38656\u0026thinsp;\u0026plusmn;\u0026thinsp;4994.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e29702\u0026thinsp;\u0026plusmn;\u0026thinsp;5871.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e32623\u0026thinsp;\u0026plusmn;\u0026thinsp;8886.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReads uniquely mapped to the reference genome\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30747\u0026thinsp;\u0026plusmn;\u0026thinsp;3454.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e29541\u0026thinsp;\u0026plusmn;\u0026thinsp;3986.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e28341\u0026thinsp;\u0026plusmn;\u0026thinsp;5645.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e31103\u0026thinsp;\u0026plusmn;\u0026thinsp;8486.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReads unable to be mapped at any loci\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5764\u0026thinsp;\u0026plusmn;\u0026thinsp;913.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6501\u0026thinsp;\u0026plusmn;\u0026thinsp;954.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1361\u0026thinsp;\u0026plusmn;\u0026thinsp;285.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1519\u0026thinsp;\u0026plusmn;\u0026thinsp;456.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReads mapped at multiple loci\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2466\u0026thinsp;\u0026plusmn;\u0026thinsp;377.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2614\u0026thinsp;\u0026plusmn;\u0026thinsp;510.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.25\u0026thinsp;\u0026plusmn;\u0026thinsp;0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.27\u0026thinsp;\u0026plusmn;\u0026thinsp;0.54\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMapping efficiency (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e78.9\u0026thinsp;\u0026plusmn;\u0026thinsp;2.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e76.4\u0026thinsp;\u0026plusmn;\u0026thinsp;1.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95.4\u0026thinsp;\u0026plusmn;\u0026thinsp;0.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e95.3\u0026thinsp;\u0026plusmn;\u0026thinsp;0.78\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAll continuous variables are expressed as the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD.\u003c/p\u003e \u003cp\u003eThe average mapping rates of reads from CN and AD samples were 78.9 and 76.4%, respectively (Table\u0026nbsp;4). Among the 88 DMRs analyzed, eight amplicons corresponding to eight DMRs were excluded due to insufficient read coverage (\u0026lt;\u0026thinsp;10 reads) in at least one sample. Additionally, 10 amplicons representing 10 DMRs were excluded because they could not be uniquely mapped to the reference genome, likely due to the short length and low complexity of DMR sequences (Supplementary Table S5).\u003c/p\u003e \u003cp\u003eRegarding the remaining 70 DMRs, 114 amplicons, including 1,251 CpG sites, were analyzed. Methylation differences at 511 CpGs in 67 DMRs were significant (\u003cem\u003eP\u003c/em\u003e\u003csub\u003e\u003cem\u003ebon\u003c/em\u003e\u003c/sub\u003e \u0026le;0.05/1251). Of the 67 DMRs, 52 contained 496 CpGs with at least three CpGs per DMR showing significant methylation differences. Using the DNA methylation rates (meC/(meC\u0026thinsp;+\u0026thinsp;C)) of these CpGs as explanatory valuables, the significance of methylation differences at these DMRs was evaluated using the burden test, adjusted for sex and age [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. The burden test is a collapsing method used for genetic association analyses of rare variants. The method collapses variants within a gene and analyzes them as a unit to assess whether the gene is associated with a disease. In the present study, the methylation rates of CpGs in a DMR were treated as the allele frequencies of genetic variants in a gene, and the relationship with AD was tested. The \u003cem\u003ep\u003c/em\u003e-values of DMRs calculated by the burden test were converted to \u003cem\u003eq\u003c/em\u003e-values using the Benjamini-Hochberg procedure [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], and 6 DMRs were found to be significant with \u003cem\u003eq\u003c/em\u003e-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 (Table\u0026nbsp;5). Five of the 6 DMRs were located in the gene body regions of the following genes: \u003cem\u003eANKilosis Homolog\u003c/em\u003e (\u003cem\u003eANKH\u003c/em\u003e), \u003cem\u003eMARS\u003c/em\u003e, \u003cem\u003eANKFY1\u003c/em\u003e, \u003cem\u003eCholinergic Receptor Nicotinic Epsilon Subunit\u003c/em\u003e (\u003cem\u003eCHRNE\u003c/em\u003e), and \u003cem\u003eKLF2\u003c/em\u003e, while the remaining DMR was found in the long non-coding RNA gene, \u003cem\u003eLINC00908\u003c/em\u003e. Of these, five DMRs in \u003cem\u003eANKH\u003c/em\u003e, \u003cem\u003eMARS\u003c/em\u003e, \u003cem\u003eANKFY1\u003c/em\u003e, \u003cem\u003eCHRNE\u003c/em\u003e, and \u003cem\u003eLINC00908\u003c/em\u003e showed lower methylation levels in AD than in CN, whereas one DMR, located in and around \u003cem\u003eKLF2\u003c/em\u003e, exhibited a higher methylation level in AD (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabe\" border=\"1\"\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003eTable\u0026nbsp;5. Summary of the validation of 6 DMRs in two cohorts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eBurden test\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eMethylation difference\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChromosome\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGene\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAmplicon (bp)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLocation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eGroup\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFDR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e(%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eANKH\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e213\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eintron\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0034\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0447\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-4.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0633\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eMARS\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e169\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eexon\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0054\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-5.17\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.2452\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0072\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eANKFY1\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e594\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eintron\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0046\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-5.26\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0680\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eCHRNE\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e615\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eexon-intron\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0013\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0331\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-4.42\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.6739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0608\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eLINC00908\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e424\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eintron\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0023\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0402\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-4.53\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0645\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0011\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cem\u003eKLF2\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e536\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eexon-intron\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2nd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.0319\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7.37\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3rd cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.4455\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003emeta-analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.0015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003csup\u003ea\u003c/sup\u003eFor each CpG site in Table S6, the methylation difference in the second cohort was calculated, and the sum of the differences was then averaged.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eValidation of 6 DMRs with the third cohort\u003c/h2\u003e \u003cp\u003eMethylation changes in the 6 genes were reexamined by BA-seq with the third cohort, composed of 48 CN and 48 AD subjects (none of whom underwent PiB-PET imaging, Table\u0026nbsp;1). We generated 12 amplicons that covered 108 CpGs in the 6 DMRs from the third cohort, and a bisulfite DNA library for Illumina MiSeq sequencing was again constructed from the 12 amplicons (Table\u0026nbsp;4). The MiSeq sequencing run resulted in a cluster density of 343\u0026thinsp;\u0026plusmn;\u0026thinsp;100 (K/mm\u003csup\u003e2\u003c/sup\u003e), a cluster passing filter of 87.09\u0026thinsp;\u0026plusmn;\u0026thinsp;1.85, and Q30 \u0026ge;94.2%. The mean read numbers per sample from the CN and AD groups were 29,702 and 32,623, respectively, with average mapping rates of reads from CN and AD being 95.4 and 95.3%, respectively. We did not obtain any reads for an amplicon in the \u003cem\u003eKLF2\u003c/em\u003e DMR from one AD subject. DNA methylation differences were significant at 96 CpGs among the 108 CpGs surveyed (\u003cem\u003eP\u003c/em\u003e\u003csub\u003e\u003cem\u003ebon\u003c/em\u003e\u003c/sub\u003e \u0026le;0.05/108). We also assessed the methylation difference at 6 DMRs in third cohort samples using the Burden test (Table\u0026nbsp;5). The lowest \u003cem\u003ep\u003c/em\u003e-value was 0.063 at \u003cem\u003eANKH\u003c/em\u003e, and the highest was 0.446 at \u003cem\u003eKLF2\u003c/em\u003e. However, combining data from the second and third cohorts for the 6 DMRs showed a significant methylation difference (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in all genes, except for \u003cem\u003eCHRNE\u003c/em\u003e, which showed a slight difference (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.06). We regarded the five DMRs with significant methylation differences as confirmed DMRs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDiagnostic ability of AD with DNA methylation\u003c/h2\u003e \u003cp\u003eWe hypothesized that the five DMRs may contain CpG sites capable of discriminating between CN and AD. We selected CpG sites with significant methylation differences between CN and AD (\u003cem\u003eP\u003c/em\u003e\u003csub\u003e\u003cem\u003ebon\u003c/em\u003e\u003c/sub\u003e \u0026le;0.05) in the second and third cohort analyses, ensuring that the direction of the methylation change was consistent across both cohorts. The numbers of CpG sites meeting these criteria were as follows: 6 CpGs in \u003cem\u003eANKH\u003c/em\u003e, 7 in \u003cem\u003eMARS\u003c/em\u003e, 14 in \u003cem\u003eANKFY1\u003c/em\u003e, 3 in \u003cem\u003eLINC00908\u003c/em\u003e, and 30 in \u003cem\u003eKLF2\u003c/em\u003e. The locations of these CpGs are shown in Supplementary Table S6. Methylation levels at these CpGs in the five DMRs, together with the following covariates: age, sex, and the \u003cem\u003eAPOE\u003c/em\u003e e4 genotype, were used as explanatory variables in a logistic regression analysis to identify the model that best discriminated AD from CN (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). The \u003cem\u003eAPOE\u003c/em\u003e genotype is a well-established risk factor for AD [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. We performed an AUC-ROC analysis incorporating methylation levels from \u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eMARS\u003c/em\u003e adjusting with the \u003cem\u003eAPOE\u003c/em\u003e e4 allele, which yielded the highest AUC score in the validation set of 0.81 (95% CI: 0.67\u0026ndash;0.95, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB and Supplementary Table S7). The base model, which used only the \u003cem\u003eAPOE\u003c/em\u003e e4 genotype, showed a lower AUC of 0.77 (95% CI: 0.69\u0026ndash;0.84, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC and Supplementary Table S7). Therefore, we concluded that combining methylation levels at specific regions in \u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eMARS\u003c/em\u003e with the \u003cem\u003eAPOE\u003c/em\u003e genotype provided the best model for the diagnosis of AD among the combinations assessed.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn the present study, we examined the validity of MC-seq in identifying AD-related DMRs in blood for the first time. We obtained 106 DMRs in the first cohort using MC-seq, and 5 DMRs remained after validation in two independent AD and CN cohorts using BA-seq.\u0026nbsp;The low survival rate of DMRs, less than 5%, from the first screening was likely due to the small sample size, which may have led to false positives. We expect this survival rate to improve with a larger sample size in the first cohort because marker reliability is dependent on sample size [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. In the second cohort, methylation differences between CN and AD in the five DMRs were approximately 5%, whereas the initial screening criteria required a 15% difference. This suggests that the observed methylation differences in the first cohort were exaggerated due to sample-specific variability. Similarly, we previously detected small methylation changes in known AD-associated genes, including \u003cem\u003eCLU\u003c/em\u003e, \u003cem\u003eCR1\u003c/em\u003e, and \u003cem\u003ePICALM\u003c/em\u003e [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. These findings indicate that DMRs in AD blood may not exhibit large differences, highlighting the need for highly sensitive methods to detect small methylation changes in blood-based biomarkers for AD.\u003c/p\u003e \u003cp\u003eThe combination of methylation levels at two of the five DMRs with \u003cem\u003eAPOE\u003c/em\u003e genotypes gave a discriminative power (AUC) of 0.81 for the validation cohort. Due to the small sample size for the first cohort, we expect that increasing the sample size of the first cohort analyzed by MC-seq will result in improved DMRs and, consequently, a higher AUC. Additionally, further improvements may be achieved by using capture probes specifically designed by researchers, as demonstrated in other studies. This approach may enhance the sensitivity and specificity of the method, ultimately improving the identification of reliable biomarkers for AD [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe previously showed that AD-related DNA methylation decreased at CpG island shores of \u003cem\u003eCLU\u003c/em\u003e, \u003cem\u003eCR1\u003c/em\u003e, and \u003cem\u003ePICALM\u003c/em\u003e [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. CpG island shores are located 2\u0026thinsp;~\u0026thinsp;3 kb from CpG islands and are typically intermediately methylated [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. In the present study, all 5 DMRs were found in intermediately methylated regions and, except for \u003cem\u003eKLF2\u003c/em\u003e, showed lower methylation levels in AD. However, these DMRs were too distant from CpG islands to be classified as CpG island shores (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This suggests that, regardless of their positions related to CpG islands, intermediately methylated regions within genes may be susceptible to AD-related methylation changes. Since aging is the most significant risk factor for AD and age-related methylation changes in the human genome [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], it will be interesting to examine whether DNA methylation at these DMRs decreases with age. Hypomethylation in gene bodies has been linked to transcriptional integrity, including spurious transcription initiation [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], mis-splicing [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e], and readthrough beyond transcriptional termination signals [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], all of which may decrease the activities of hypomethylated genes. Recent advances in targeted hypomethylation techniques may enable functional analyses of these DMRs in model systems [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Two genes identified in this study, \u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eCHRNE\u003c/em\u003e, have also been genetically related to AD [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e] [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. \u003cem\u003eANKH\u003c/em\u003e encodes a transmembrane protein involved in regulating pyrophosphate levels in joints and other tissues. Mutations in \u003cem\u003eANKH\u003c/em\u003e may lead to excessive mineralization, contributing to joint pain, arthritis, atherosclerosis, and diabetes. A minor allele of rs112403360, located in an intron of \u003cem\u003eANKH\u003c/em\u003e, has been associated with a 1.09-fold increased risk of AD [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. In contrast, the major allele, considered to be a protective allele against AD, has been associated with cognitively healthy centenarians [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. However, the mechanism by which the minor allele of the \u003cem\u003eANKH\u003c/em\u003e gene increases the AD risk remains elusive. A recent study on a Chinese cohort demonstrated that \u003cem\u003eANKH\u003c/em\u003e was epigenetically associated with AD, showing lower methylation levels in an intronic CpG site in individuals with mild cognitive impairment or AD than in cognitively healthy controls [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003cem\u003eCHRNE\u003c/em\u003e, which encodes a subunit of the cholinergic receptor, is also genetically associated with AD. A variant in the 3\u0026rsquo;-UTR region of \u003cem\u003eCHRNE\u003c/em\u003e has been associated with a 1.09-fold increased risk of AD, possibly by enhancing gene expression [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. The potential role of \u003cem\u003eCHRNE\u003c/em\u003e up-regulation in AD is further supported by evidence that the AD-associated SNP rs113260531, located 0.3 Mb upstream of \u003cem\u003eCHRNE\u003c/em\u003e, has been linked to increased \u003cem\u003eCHRNE\u003c/em\u003e expression in all four brain regions [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. The cholinergic neurotransmitter pathway is a well-established target for AD treatment.\u003c/p\u003e \u003cp\u003eAnother gene harboring a DMR in its coding region, \u003cem\u003eKLF2\u003c/em\u003e, has also been implicated in AD. \u003cem\u003eKLF2\u003c/em\u003e belongs to the mammalian KLF family, composed of 17 members (\u003cem\u003eKFF1\u003c/em\u003e-\u003cem\u003eKFF17\u003c/em\u003e). All \u003cem\u003eKLFs\u003c/em\u003e are moderately to highly expressed in retinal ganglion cells and the brain, playing critical roles in neuronal development, regeneration, and brain homeostasis. Immunohistochemical analyses showed a significant reduction in \u003cem\u003eKLF2\u003c/em\u003e in AD brains [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. In mouse models, \u003cem\u003eKLF2\u003c/em\u003e expression was found to be down-regulated by Ab\u003csub\u003e1\u0026thinsp;\u0026minus;\u0026thinsp;42\u003c/sub\u003e, leading to cerebrovascular dysfunction [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e] and hippocampal neuron injury [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e], both of which are key pathological features of human AD brains.\u003c/p\u003e \u003cp\u003eAlthough limited information is currently available on the functions of the long intergenic non-coding RNA, \u003cem\u003eLINC00908\u003c/em\u003e, apart from its implications in carcinogenesis, it was recently shown to be up-regulated in the temporal cortex of AD brains (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e) [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThree of the six DMRs identified in blood were located in genes associated with AD. Similarly, we previously detected DNA hypomethylation in three of the six AD-related genes examined in AD blood [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. This high incidence of methylation changes in AD-related genes in blood suggests that investigating DNA methylation changes in genes may complement genome-wide association studies, particularly for identifying AD-related genes that lack informative DNA variants in or around them.\u003c/p\u003e \u003cp\u003eThere are a number of limitations that need to be addressed. The generalizability of the DMRs identified across different ethnicities remains unclear because our analysis was conducted on Japanese blood samples. Furthermore, the timing of DMR emergence in AD patients has yet to be clarified because this study used cross-sectional cohorts. Moreover, we did not perform functional analyses of the genes harboring DMRs. Future research is needed to confirm the methylation changes observed in postmortem human brain tissues because previous studies found only a modest correlation between methylation levels in the brain and blood at most CpG sites [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. In addition, brain amyloidosis may have been heterogeneous in subjects in the second and third cohorts because biomarkers did not detect it. To estimate the diagnostic ability of DMRs more accurately, they need to be validated in cohorts that are similar the first cohort, i.e., the PiB-PET-examined cohort.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eWe employed MC-seq to isolate AD-associated DNA methylation markers for the first time in this study. Despite the small sample size, we identified two methylation markers that efficiently distinguished AD from controls (\u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eMARS\u003c/em\u003e), achieving a discriminative value AUC of 0.81 when combined with the \u003cem\u003eAPOE\u003c/em\u003e genotype. In future studies, we expect more reliable markers for diagnosing the onset and progression of AD to be identified by applying or modifying the method established in the present study with larger sample sizes and more diverse populations.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAlzheimer\u0026rsquo;s disease\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e\u003cem\u003eANKH\u003c/em\u003e\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eANKilosis Homolog\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAPOE\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eapolipoprotein E\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003earea under the ROC curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBA-seq\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ebisulfite amplicon sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ebp\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ebase pair\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e\u003cem\u003eCHRNE\u003c/em\u003e\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCholinergic Receptor Nicotinic Epsilon Subunit\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCN\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ecognitively normal\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCpGs\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ecytosine-phosphate-guanine dinucleotide sites\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDMP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edifferentially methylated position\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDMR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edifferentially methylated region\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGRCh37\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGenome Reference Consortium Human Build 37\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMC-seq\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003emethylation capture sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMMSE\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMini-Mental State Examination\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNCGG\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eNational Center for Geriatrics and Gerontology\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNGS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003enext-generation sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePiB-PET\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePittsburgh compound-B positron emission tomography\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eROC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ereceiver operator characteristic\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eStandard deviation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe present study was conducted with the approval of the Ethical Committee of NCGG (708-8 and 521-4). The design and performance of the present study involving human subjects were clearly described in a research proposal. All participants were voluntary and completed informed consent in writing before registering at the NCGG Biobank, which collects human biomaterials and data for geriatrics research.\u0026nbsp;\u0026nbsp;All DNA and associated information used in this study were obtained from the NCGG Biobank (H27013(2)). \u0026nbsp; For those with substantial cognitive impairment, written informed consent was obtained from the legal next of kin, which the local Ethics Committee had previously approved.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMC-seq data were deposited in the GEO repository, https://ncbi.nlm.nih.gov/geo/ (GSE24435). The datasets analyzed during this study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was partly supported by grants from Research Funding for Longevity Sciences from the National Center for Geriatrics and Gerontology (26-16 and 29-20 to NS).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRM performed all of the core experiments, analyzed the data, and wrote the manuscript. \u0026nbsp;KS and KY analyzed the amplicons of the 2nd and 3rd cohorts. AN and YA conducted PiB-PET. DS, KO, SN, and NS designed the study. NS supervised the study. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank Yuko Nishimura and the staff of the Biobank of the National Center for Geriatrics and Gerontology for their contribution to this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eGroen K, Lea RA, Maltby VE, Scott RJ, Lechner-Scott J. Letter to the editor: blood processing and sample storage have negligible effects on methylation. Clin Epigenetics. 2018;10:22.\u003c/li\u003e\n\u003cli\u003eMilicic L, Porter T, Vacher M, Laws SM. Utility of DNA Methylation as a Biomarker in Aging and Alzheimer\u0026apos;s Disease. J Alzheimers Dis Rep. 2023;7(1):475-503.\u003c/li\u003e\n\u003cli\u003eOlstad EW, Nordeng HME, Sandve GK, Lyle R, Gervin K. Low reliability of DNA methylation across Illumina Infinium platforms in cord blood: implications for replication studies and meta-analyses of prenatal exposures. Clin Epigenetics. 2022;14(1):80.\u003c/li\u003e\n\u003cli\u003eSugden K, Hannon EJ, Arseneault L, Belsky DW, Corcoran DL, Fisher HL, et al. Patterns of Reliability: Assessing the Reproducibility and Integrity of DNA Methylation Measurement. Patterns (N Y). 2020;1(2).\u003c/li\u003e\n\u003cli\u003eLogue MW, Smith AK, Wolf EJ, Maniates H, Stone A, Schichman SA, et al. The correlation of methylation levels measured using Illumina 450K and EPIC BeadChips in blood samples. Epigenomics. 2017;9(11):1363-71.\u003c/li\u003e\n\u003cli\u003eLee EJ, Luo J, Wilson JM, Shi H. Analyzing the cancer methylome through targeted bisulfite sequencing. Cancer Lett. 2013;340(2):171-8.\u003c/li\u003e\n\u003cli\u003eIllumina. TruSeq methyl capture EPIC library prep kit. 2016.\u003c/li\u003e\n\u003cli\u003eKim SS, Lee SC, Lim B, Shin SH, Kim MY, Kim SY, et al. DNA methylation biomarkers distinguishing early-stage prostate cancer from benign prostatic hyperplasia. Prostate Int. 2023;11(2):113-21.\u003c/li\u003e\n\u003cli\u003eOhmomo H, Komaki S, Sutoh Y, Hachiya T, Ono K, Arai E, et al. Potential DNA methylation biomarkers for the detection of clear cell renal cell carcinoma identified by a whole blood-based epigenome-wide association study. Epigenetics Communications. 2022;2(1).\u003c/li\u003e\n\u003cli\u003eAllum F, Shao X, Guenard F, Simon MM, Busche S, Caron M, et al. Characterization of functional methylomes by next-generation capture sequencing identifies novel disease-associated variants. Nat Commun. 2015;6:7211.\u003c/li\u003e\n\u003cli\u003eKacmarczyk TJ, Fall MP, Zhang X, Xin Y, Li Y, Alonso A, et al. \u0026quot;Same difference\u0026quot;: comprehensive evaluation of four DNA methylation measurement platforms. Epigenetics Chromatin. 2018;11(1):21.\u003c/li\u003e\n\u003cli\u003eBernstein DL, Kameswaran V, Le Lay JE, Sheaffer KL, Kaestner KH. The BisPCR(2) method for targeted bisulfite sequencing. Epigenetics Chromatin. 2015;8:27.\u003c/li\u003e\n\u003cli\u003eLi LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. Bioinformatics. 2002;18(11):1427-31.\u003c/li\u003e\n\u003cli\u003eJonsson H, Sulem P, Kehr B, Kristmundsdottir S, Zink F, Hjartarson E, et al. Parental influence on human germline de novo mutations in 1,548 trios from Iceland. Nature. 2017;549(7673):519-22.\u003c/li\u003e\n\u003cli\u003eMitsumori R, Sakaguchi K, Shigemizu D, Mori T, Akiyama S, Ozaki K, et al. Lower DNA methylation levels in CpG island shores of CR1, CLU, and PICALM in the blood of Japanese Alzheimer\u0026apos;s disease patients. PLoS One. 2020;15(9):e0239196.\u003c/li\u003e\n\u003cli\u003eOzaki Y, Yoshino Y, Yamazaki K, Sao T, Mori Y, Ochi S, et al. DNA methylation changes at TREM2 intron 1 and TREM2 mRNA expression in patients with Alzheimer\u0026apos;s disease. J Psychiatr Res. 2017;92:74-80.\u003c/li\u003e\n\u003cli\u003eLangmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9(4):357-9.\u003c/li\u003e\n\u003cli\u003eKrueger F, Andrews SR. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics. 2011;27(11):1571-2.\u003c/li\u003e\n\u003cli\u003eAkalin A, Kormaksson M, Li S, Garrett-Bakelman FE, Figueroa ME, Melnick A, et al. methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol. 2012;13(10):R87.\u003c/li\u003e\n\u003cli\u003eBreen C, Papale LA, Clark LR, Bergmann PE, Madrid A, Asthana S, et al. Whole genome methylation sequencing in blood identifies extensive differential DNA methylation in late-onset dementia due to Alzheimer\u0026apos;s disease. Alzheimers Dement. 2024;20(2):1050-62.\u003c/li\u003e\n\u003cli\u003eLee S, Emond MJ, Bamshad MJ, Barnes KC, Rieder MJ, Nickerson DA, et al. Optimal Unified Approach for Rare-Variant Association Testing with Application to Small-Sample Case-Control Whole-Exome Sequencing Studies. The American Journal of Human Genetics. 2012;91(2):224-37.\u003c/li\u003e\n\u003cli\u003eBenjamini YH, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289-300.\u003c/li\u003e\n\u003cli\u003eStrittmatter WJ, Saunders AM, Schmechel D, Pericak-Vance M, Enghild J, Salvesen GS, et al. Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc Natl Acad Sci U S A. 1993;90(5):1977-81.\u003c/li\u003e\n\u003cli\u003eZhang L, Silva TC, Young JI, Gomez L, Schmidt MA, Hamilton-Nelson KL, et al. Epigenome-wide meta-analysis of DNA methylation differences in prefrontal cortex implicates the immune processes in Alzheimer\u0026apos;s disease. Nat Commun. 2020;11(1):6114.\u003c/li\u003e\n\u003cli\u003eKomaki S, Nagata M, Arai E, Otomo R, Ono K, Abe Y, et al. Epigenetic profile of Japanese supercentenarians: a cross-sectional study. Lancet Healthy Longev. 2023;4(2):e83-e90.\u003c/li\u003e\n\u003cli\u003eIrizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, Onyango P, et al. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet. 2009;41(2):178-86.\u003c/li\u003e\n\u003cli\u003eSeale K, Horvath S, Teschendorff A, Eynon N, Voisin S. Making sense of the ageing methylome. Nat Rev Genet. 2022;23(10):585-605.\u003c/li\u003e\n\u003cli\u003eNeri F, Rapelli S, Krepelova A, Incarnato D, Parlato C, Basile G, et al. Intragenic DNA methylation prevents spurious transcription initiation. Nature. 2017;543(7643):72-7.\u003c/li\u003e\n\u003cli\u003eYearim A, Gelfman S, Shayevitch R, Melcer S, Glaich O, Mallm JP, et al. HP1 is involved in regulating the global impact of DNA methylation on alternative splicing. Cell Rep. 2015;10(7):1122-34.\u003c/li\u003e\n\u003cli\u003eShirai M, Nara T, Takahashi H, Takayama K, Chen Y, Hirose Y, et al. Identification of aberrant transcription termination at specific gene loci with DNA hypomethylated transcription termination sites caused by DNA methyltransferase deficiency. Genes Genet Syst. 2022;97(3):139-52.\u003c/li\u003e\n\u003cli\u003eNu\u0026ntilde;ez JK, Chen J, Pommier GC, Cogan JZ, Replogle JM, Adriaens C, et al. Genome-wide programmable transcriptional memory by CRISPR-based epigenome editing. Cell. 2021;184(9):2503-19.e17.\u003c/li\u003e\n\u003cli\u003eBellenguez C, Kucukali F, Jansen IE, Kleineidam L, Moreno-Grau S, Amin N, et al. New insights into the genetic etiology of Alzheimer\u0026apos;s disease and related dementias. Nat Genet. 2022;54(4):412-36.\u003c/li\u003e\n\u003cli\u003ede Rojas I, Moreno-Grau S, Tesi N, Grenier-Boley B, Andrade V, Jansen IE, et al. Common variants in Alzheimer\u0026apos;s disease and risk stratification by polygenic risk scores. Nat Commun. 2021;12(1):3417.\u003c/li\u003e\n\u003cli\u003eTesi N, van der Lee S, Hulsman M, van Schoor NM, Huisman M, Pijnenburg Y, et al. Cognitively healthy centenarians are genetically protected against Alzheimer\u0026apos;s disease. Alzheimers Dement. 2024;20(6):3864-75.\u003c/li\u003e\n\u003cli\u003eWu S, Yang F, Chao S, Wang B, Wang W, Li H, et al. Altered DNA methylome profiles of blood leukocytes in Chinese patients with mild cognitive impairment and Alzheimer\u0026apos;s disease. Front Genet. 2023;14:1175864.\u003c/li\u003e\n\u003cli\u003eHe L, Loika Y, Kulminski AM. Allele-specific analysis reveals exon- and cell-type-specific regulatory effects of Alzheimer\u0026apos;s disease-associated genetic variants. Transl Psychiatry. 2022;12(1):163.\u003c/li\u003e\n\u003cli\u003eFang X, Zhong X, Yu G, Shao S, Yang Q. Vascular protective effects of KLF2 on Abeta-induced toxicity: Implications for Alzheimer\u0026apos;s disease. Brain Res. 2017;1663:174-83.\u003c/li\u003e\n\u003cli\u003eWu C, Li F, Han G, Liu Z. Abeta(1-42) disrupts the expression and function of KLF2 in Alzheimer\u0026apos;s disease mediated by p53. Biochem Biophys Res Commun. 2013;431(2):141-5.\u003c/li\u003e\n\u003cli\u003eDuan Q, Si E. MicroRNA-25 aggravates Abeta1-42-induced hippocampal neuron injury in Alzheimer\u0026apos;s disease by downregulating KLF2 via the Nrf2 signaling pathway in a mouse model. J Cell Biochem. 2019;120(9):15891-905.\u003c/li\u003e\n\u003cli\u003eChen C, Zhang Z, Liu Y, Hong W, Karahan H, Wang J, et al. Comprehensive characterization of the transcriptional landscape in Alzheimer\u0026apos;s disease (AD) brains. Sci Adv. 2025;11(1):eadn1927.\u003c/li\u003e\n\u003cli\u003eYu L, Chibnik LB, Yang J, McCabe C, Xu J, Schneider JA, et al. Methylation profiles in peripheral blood CD4+ lymphocytes versus brain: The relation to Alzheimer\u0026apos;s disease pathology. Alzheimers Dement. 2016;12(9):942-51.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"clinical-epigenetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"clep","sideBox":"Learn more about [Clinical Epigenetics](http://clinicalepigeneticsjournal.biomedcentral.com/)","snPcode":"13148","submissionUrl":"https://submission.nature.com/new-submission/13148/3","title":"Clinical Epigenetics","twitterHandle":"@OAgenetics","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6200381/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6200381/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eMethylation capture sequencing (MC-seq), which relies on next-generation sequencing technology, offers advantages over the widely used array-based approach that Illumina Inc. developed regarding both resolution and comprehensiveness for detecting DNA methylation changes across genomes. In the present study, MC-seq was employed for the first time to identify DNA methylation markers for Alzheimer\u0026rsquo;s disease (AD).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eWe compared DNA methylation in the blood of 12 AD patients with brain amyloidosis and 12 cognitively normal elderly Japanese individuals without brain amyloidosis. Candidate methylation differences were validated in the two cohorts using bisulfite amplicon sequencing. Significant differentially methylated regions were identified in the \u003cem\u003eANKH\u003c/em\u003e, \u003cem\u003eMARS\u003c/em\u003e, \u003cem\u003eANKFY1\u003c/em\u003e, \u003cem\u003eLINC00908\u003c/em\u003e, and \u003cem\u003eKLF2\u003c/em\u003e genes and a slight methylation change in \u003cem\u003eCHRNE\u003c/em\u003e (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.061). Furthermore, our AD diagnostic prediction model showed that combining the methylation levels of \u003cem\u003eANKH\u003c/em\u003e and \u003cem\u003eMARS\u003c/em\u003e with the \u003cem\u003eAPOE\u003c/em\u003e genotype provided diagnostic accuracy, achieving AUCs of 0.90 and 0.81 in the discovery and validation datasets, respectively.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThe present results suggest the potential of combining these markers for diagnosing AD and support the validity of our approach for identifying disease-related DNA methylation markers using next-generation sequencing.\u003c/p\u003e","manuscriptTitle":"Identification of diagnostic DNA methylation markers in the blood of Japanese Alzheimer’s disease patients using methylation capture sequencing","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-03 09:21:00","doi":"10.21203/rs.3.rs-6200381/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-04-10T04:53:31+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-04T12:23:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"55173615599356969165646404798881960064","date":"2025-04-04T07:48:02+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-02T01:25:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"295147533234315056067019408410476455288","date":"2025-04-01T06:51:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"157213258220138408109433376140297631203","date":"2025-03-25T12:50:36+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-03-24T10:08:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-03-22T08:03:29+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-03-21T04:42:31+00:00","index":"","fulltext":""},{"type":"submitted","content":"Clinical Epigenetics","date":"2025-03-11T05:51:24+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"clinical-epigenetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"clep","sideBox":"Learn more about [Clinical Epigenetics](http://clinicalepigeneticsjournal.biomedcentral.com/)","snPcode":"13148","submissionUrl":"https://submission.nature.com/new-submission/13148/3","title":"Clinical Epigenetics","twitterHandle":"@OAgenetics","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"35d4235f-3587-4e36-8515-17d367a006f8","owner":[],"postedDate":"April 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-06-23T16:04:14+00:00","versionOfRecord":{"articleIdentity":"rs-6200381","link":"https://doi.org/10.1186/s13148-025-01905-0","journal":{"identity":"clinical-epigenetics","isVorOnly":false,"title":"Clinical Epigenetics"},"publishedOn":"2025-06-20 15:57:33","publishedOnDateReadable":"June 20th, 2025"},"versionCreatedAt":"2025-04-03 09:21:00","video":"","vorDoi":"10.1186/s13148-025-01905-0","vorDoiUrl":"https://doi.org/10.1186/s13148-025-01905-0","workflowStages":[]},"version":"v1","identity":"rs-6200381","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6200381","identity":"rs-6200381","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}