Nat Genet: A study of single nucleotide polymorphisms in Alzheimer's and Parkinson's diseases, revealing variations in the genetic risk points of both diseases

Alzheimer's disease (AD) and Parkinson's disease (PD) are the two most common neurodegenerative diseases.
although the Genome-wide Association Study (GWAS) has identified several potential genetic risk locations.
most risk locations are in non-coding areas, so it is not clear whether these genes are related to disease function or to other genes.
single nucleotide polymorphism (SNPs) mainly refers to DNA sequence polymorphism at the genomic level caused by variation of a single nucleotide (base conversion or conversion, insertion or loss).
is one of the most common human genetic variants.
most functional non-coding SNPs can play their role by interfering with the binding of transcription factors and regulating the function of the components to alter gene expression.
note that these regulatory elements have a high degree of cell type specificity, suggesting that SNPs also have cell type specificity.
Therefore, the importance of classifying and functionally revealing the active regulatory elements in each brain cell type in the correct tissue and regional context helps to clarify the function of gene risk points in the molecular pathogenesis of common neurodegenerative diseases.
, 2020 in the journal Nature Genetics, Thomas J. Montine, Stanford University School of Medicine, U.S.A., on October 26, 2020. The professor and Professor Howard Y. Chang teamed up to identify cell type-specific regulatory components using single-cell chromatin maps and apply mechanical learning to help predict the functional SNPs of AD and PD.
researchers identified target genes and cell types for several non-coded GPAS site in AD and PD, and these data and techniques also provide guidance for the application of other neurological disorders, allowing us to better understand the role of genetic non-coding mutations in disease.
chromatin accessibility is commonly understood as open chromatin, which refers to areas where the transverse regulatory elements and transverse action factors such as promoters, enhancers, insulators, silencer, etc. can be accessed after the structure of the dense nucleosome is destroyed, which is closely related to the transcription regulation of the true nuclear organism.
as early as 2013, William J. Professor Greenleaf and Professor Howard Y. Chang's laboratory have developed a method that can be used to study chromatin accessability, called Assay for Transposase-Accessible Chromatin with high-level sequencing, or transposase accessible chromatin sequencing, or transposase accessible chromatin sequencing, or chromatin-open sequencing technology, or ATAC-seq.
the principle is to easily bind the properties of the open chromatin by trans-seatase Tn5, and then sequence the DNA sequence captured by the Tn5 enzyme.
the main advantage of ATAC-Seq over other technologies, such as FAIRE-Seq or DNase-Seq, which studies similar chromatin characteristics, is that the number of cells required for this determination is small and its two-step approach is relatively simple to operate.
Then, given the powerful advantages of ATAC-Seq, first of all, the authors used ATOC-Seq to map large chromatin in brain anatomy region samples of 39 cognitively normal individuals in the new cortical, hippocose, black, sprite, etc., to study the role of non-coding genomes in neurodegenerative diseases (Figure 1).
these large ATAC-seq data, or peaks, indicate regional differences in chromatin accessability.
1 ATAC-Seq peaks (right) in the brain regions (left) and brain region samples studied in this paper (pictured from Corces, M.R., et al. Nat Genet 2020; 52: 1158-1168) Figure 2 sample of scATAC-seq data (image quoted from: Corces, M.R., et al. Nat Genet 2020; 52: 1158-1168) What is the chromosome accessability of different cells in different brain regions? The authors then analyzed the chromatin accessability of more than 70,000 single cells in the sample, namely single-cell ATAC-Seq (scATAC-Seq) and locked 24 cell groups (clusters), including excitable neurons, inhibitory neurons, small glial cells, less protrusive glial cells, astroid glial cells and less progenitic cells (OPCs) (Figure 2-3).
importantly, the scATAC-Seq analysis showed the abunding of specific far-end/containing sub-peaks and the absence of initiation sub-peaks, which is consistent with the role of remote regulatory elements in cell type-specific gene regulation (Figure 3).
addition, the results also suggest the usefulness of scATAC-Seq, especially considering that scATAC-Seq has obvious advantages when peaking of a particular cell type is identified from large tissues containing many different cell types.
3 chromatin accessability heat map (left) and adjustment element base sequence analysis (right) (pictured from Corces, M.R., et al. Nat Genet 2020; 52: 1158-1168) In order to further study which transcription factors may be responsible for establishing and maintaining these cell type-specific regulatory procedures.
, the authors performed a base sequence affluge analysis, or domain analysis, on the peak of each cell type.
results show that several known cell type recognition drivers, such as the subsectors SOX9 and SOX10 in less protrusion cells, the base sequence of ASCL1 in OPCs, and the riches of transcription factors SPI1 and JUN/FOS in small glial cells and neurons, respectively (Figure 3).
data further demonstrate the specificity of ATAC-seq cell types, in particular the heterogeneity of the brain regions of glial cells such as astrocytes and OPCs.
4 30 "neuron groups" (images quoted from: Corces, M.R., et al.) Nat Genet 2020; 52: 1158-1168) Taking into account the diversity of neuron types and functions, the researchers wanted to further refine the scATAC-seq data.
batch correction analysis showed 30 "neuron groups", each representing a unique type of neuron cell or cell state, and determined the peak, gene, and transcription factor activity of neuron cell specificity (Figure 5).
Interestingly, the data analysis identified a key cell type lost in a PD, the black dopamine-energy neuron group, or rather the tyrosine hydroxyme-positive dopamine-positive neuron group (Figure 5).
analysis of the chain imbalance of neuron-specific GTAS SNPs (image quoted from: Corces, M.R., et al.) Nat Genet 2020; 52: 1158-1168) The above data fully demonstrate the cell type specificity of scATAC-seq data, so is SNPs associated with neurodegenerative diseases rich in specific cell type regions where chromatin access is available? Cell-specific chain imbalance analysis showed that the peak concentration of small glial cells in AD showed a significant increase in the genetic probability of each SNP, while in any cell type of PD, there was a significant concentration of no SNP genetic probability, possibly because PD cells were more complex than AD (Figure 5).
other words, no significant abundance of SNPs was found in the peak regions of any group of AD or PD neurons.
: Chain imbalance (linkage disequilibrium, or LD) is the chance that alleles belonging to two or more genes will appear on one chromosome at the same time, higher than the frequency of random appearance.
of HLA's different gene seats appear at a certain frequency in the population.
, as long as the two genes are not completely inherited independently, there will be some degree of chaining.
this situation is called chain imbalance.
chain imbalances can be in different regions of the same chromosome or on different chromosomes.
results, the authors want to further study the target genes at each GTAS site.
to this end, the authors used HiChIP to label the acetylase hicysteine H3 lysine 27 (i.e. H3K27ac) to mark active enhancers and promoters to map a three-dimensional chromatin 3D composition centered on the enhancer.
data analysis identified a total of 833,975 chromatin 3D interactions in each brain region, 67.4% of the interactions had ATAC-seq peaks in two signal anchor sequences, 29.2% had ATAC-seq peaks in one signal anchor sequence, and the remaining 3.4% did not overlap with ATTIC-seq peaks.
data show that scATAC-seq can pinpoint GBAS polymorphic cell targets, allowing GBAS SNPs to be associated with downstream target genes.
: Signal anchoring sequence refers to a unique signal sequence in membrane proteins, which is used to anchor these proteins to lipid double-layer membranes.
Chain imbalance refers to the chance that alleles belonging to two or more gene seats will appear on one chromosome at the same time, higher than the frequency of random appearance: Figure 6 using multi-layered multigroup methods (left) and mechanical learning (right) to study functional non-coding GTAS polymorphisms (pictured: Corces, M.R., et al.). Nat Genet 2020; 52: 1158-1168) Figure 7 applies multi-layered multi-group methods and mechanical learning to study functional non-coding GTAS polymorphisms in AD and PD (images quoted from Corces, M.R., et al.). Nat Genet 2020; 52: 1158-1168) To further explain the functional effects of GPAS polymorphism, the authors first constructed an AD and PD disease-related SNPs library of 9707 S NPs, of which 3,245 SNPs are distributed on 44 AD-related gene constellations and 6,496 are distributed on 86 PD-related gene constellations, of which 34 SNPs appear on the same gene base for both diseases.
Then the authors developed a multi-layered, multi-group approach to predicting functional non-coding GBAS polymorphisms: first, these SNPs overlapped with the chromatin accessibility peaks of ATAC-Seq (Tier3) and then identified SNPs (Tier2) that might affect regulatory interactions. ), SNPs (Tier1) (Figure 6-7), which may directly affect the binding of transcription factors, were identified, and support vector machine and allegen imbalance analysis was applied to determine the genetic and molecular processes that may be involved in AD and PD, as well as those GWAS site involved in non-coding regulation.
these also show that multi-layered, multi-group approaches can predict functional non-coded SNPs.
: Support Vector Machine (SVM) is a class of generalized linear classifiers that binaryly classify data by supervised learning, and its decision boundary is the maximum margin hyperplane (maximum-margin hyperplane) for learning sample solving.
SVM is a classifier with sparseness and robustness by using hinge loss to calculate empirical risk and adding a normalization item to the solution system to optimize structural risk.
SVM can be classified nonlinearly by nuclear method, which is one of the common nuclear learning methods.
: Figure 8 AD and PD PICALM (left), SLC24A (right) functional non-coding SNPs analysis (photo quoted from: Corces, M.R., et al. Nat Genet 2020; 52: 1158-1168) was followed by the authors focusing on disease-related genes that are still unknown to SNPs, as well as genes in previous studies that are not related to disease pathogens.
although GBAS indicates that genes such as PICALM, SLC24A, BIN1, and KCNIP3 are associated with AD, it is not clear which polymorphism drives this relationship.
First of all, using PICALM as an example, the authors found that there was a potential functional variation that disturbed the FOS/AP1 factor binding site, and that the site was located in the upper part of PICALM in a sequence of protrusive glial cell-specific regulatory elements;
results show the new function of the PICALM gene and its special role in less protrusive glial cells.
similar, the authors found that the SLC24A4 gene base also has a small chain imbalance region, containing 46 SNPs, and SNPs are located in the SLC24A4 containing subs.
one of the SNPs is special, with small glial cell specificity, which interferes with the SPI1 substate and is carried out with the initiaters of the RIN3 gene.