Transcriptomic analysis of lipid metabolism genes in Alzheimer’s disease: highlighting pathological outcomes and compartmentalized immune status

doi:10.21203/rs.3.rs-3978725/v1

Download PDF

Research Article

Transcriptomic analysis of lipid metabolism genes in Alzheimer’s disease: highlighting pathological outcomes and compartmentalized immune status

https://doi.org/10.21203/rs.3.rs-3978725/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 22 May, 2024

Read the published version in Journal of Molecular Neuroscience →

You are reading this latest preprint version

The dysregulation of lipid metabolism has been strongly associated with Alzheimer's Disease (AD); however, the biomedical implications and clinical relevance of these findings have not been systematically examined. Here, we conducted a comprehensive bioinformatic evaluation of AD-derived transcriptome datasets from postnatal brains and peripheral blood. We utilized differential gene expression and hierarchical clustering to identify co-expressed modules of lipid metabolism genes in patients based on their molecular functions in biological enrichment and molecular pathway analysis, association with pathological phenotypes, and molecular network correlation. Additionally, we analyzed the expression patterns of these genes in immune and nonimmune cells as well as cell type enrichments in both brain tissue and peripheral blood. By categorizing patients into distinct transcriptional clusters and stratified groups, we found enrichment in biological pathways for neurodegenerative diseases, oxidative phosphorylation, synaptic transmission, unexpected infections, and molecular functions for cellular translation and energy production in the stratified clusters and groups. Biological network analysis indicates striking differences between lipid-metabolism differential expression genes (DEGs) in the periphery and CNS, with restricted processes being enriched. Notably, neurons, glial cells involved in neuroinflammation, and peripheral blood immune cell infiltration revealed a marked disparity in the clustering subgroups in patients’ hippocampi and peripheral regions. Differentially expressed genes such as PLD3, NDUFAB1, OXCT1, PI4KA, and AACS in the brain and DBI, MBOAT7, and RXRA in the periphery correlate well with disease pathologies and immune cell preferences. These results suggest that lipid metabolism is critical for disease progression and immune cell activation, thus providing an innovative approach to diagnosing and treating AD.

Alzheimer’s disease

lipid metabolism

bioinformatics

immune status

transcriptomics

biomarkers

Alzheimer's disease (AD) is the most prevalent neurodegenerative disorder of age-related cognitive decline (Soria Lopez JA 2019). The spectrum of AD encompasses a continuum of disorders spanning several years, proceeding from mild cognitive impairment (MCI) to dementia. Genetic factors influence AD development, although the majority of individual cases do not exhibit dominant inheritance(Kunkle et al. 2019). The defining features of AD are synaptic malfunction, tau-containing neurofibrillary tangles, amyloid plaques comprising amyloid β-protein (Aβ), and reduced neuron viability (Knopman et al. 2021). The discovery of reliable biomarkers or mechanisms for predicting, diagnosing, and treating AD remains elusive.

Interference with lipid metabolism expedites the progression of pathological brain damage (Akyol et al. 2021, Whiley et al. 2014, Zang et al. 2021). Blood lipid levels partly reflect body lipid metabolism, making them valuable biomarkers for tracking disease progression. In addition, abnormal lipid profiles may be closely related to the marked structural modifications (as gauged by the Braak stage) observed in the brain during the phenotypic transition of mild AD(Akyol et al. 2021). Currently, through the multi-omics methodologies combined with unsupervised clustering, lipid metabolism analysis in AD has identified molecules enriched in specific subtypes of the condition(Rosenthal et al. 2022, Xu et al. 2020). Lipidomics data from peripheral blood and brain sources of AD reveal a dysregulated lipid metabolism associated with specific genetic variations (Liu et al. 2021, Wood et al. 2015). The blood-brain barrier (BBB) separates the central and peripheral milieu, which suggests that different molecular components are involved in lipid metabolism in disease progression.

More recently, genome-wide association studies (GWAS) have consistently identified several genes related to lipid homeostasis, including apolipoprotein E (APOE), as major risk factors for both sporadic and inherited familial AD(Giri et al. 2016). Meta-analyses have confirmed that genes involved in lipid processing and immunoregulation are linked to the development of AD(Kunkle et al. 2019, Xu et al. 2020). Immune cell infiltration, a nonnegligible form of trigger neuroinflammation, extensively harms the central nervous system (CNS) through the interplay between immune cells and neurons(Dai and Shen 2021, Heneka et al. 2015). Studies that characterize molecular pathways and gene networks relevant to lipid metabolism using transcriptomic data with expanded modules and the correlation between the transcriptomes of lipid metabolism and the immune status of AD have not been thoroughly identified.

In the present study, we leveraged the large patient transcriptomic cohort from the NCBI Gene Expression Omnibus (GEO) to inform lipid metabolism in AD. A patient stratification analysis was conducted by analyzing RNA-sequencing (RNA-seq) expression data from AD patients' hippocampus and peripheral circulation. Gene set variation analysis (GSVA) and gene set enrichment analysis (GSEA) suggest a molecular pathway that captures most disease mechanisms previously associated with AD pathophysiology(Rosenthal et al. 2022). Notably, the expression and association of isolated lipid metabolism genes in the central and peripheral regions are distinct, as characterized by the activation of immune cells linked to particular expression profiles and the cellular selectivity of different lipid metabolism genes in the two areas. In addition, our study reveals new associations between AD and region-specific lipid metabolism genes, thus offering novel possibilities for understanding disease pathologies and developing diagnostic or drug-targeting strategies.

Data preparation

Transcriptome expression profiles and corresponding complete clinical annotations were all from the GEO databases (https://www.ncbi.nlm.nih.gov/geo/). The analysis included 104 RNA microarray samples of hippocampal from four datasets (GSE48350, GSE36980, GSE1297, and GSE84422) and 89 samples from two whole-blood datasets (GSE63060 and GSE63061) (Table S1). In addition, the GSE163577 and GSE157827 scRNA-seq datasets were employed to classify cell-specific lipid metabolism gene expression in cell subpopulations of the hippocampus and the prefrontal cortex (PFC). Combat methods were used to modify batch processing effects, eliminating discrepancies between batches and facilitating downstream difference analysis.

Identification of robust differentially expressed genes of lipid metabolism.

Differentially expressed genes (DEGs) of lipid metabolism in the expression series matrix were identified using the “limma” R package in R software(Ritchie et al. 2015). The cutoff of |log2 (fold change) (log2FC)|>=0.3 and p_adj < = 0.05 was used for significance. The R package "VennDiagram" was used to obtain the intersection of upregulated and downregulated genes, and the R package "ggplot2" (Gustavsson et al. 2022)was utilized for volcano plots of DEGs. Subsequently, the R package “RCircos” was employed to visualize gene expression levels and chromosomal locations, while the R package "corrplot" was used to scatterplot the diagram of hub genes.

Unsupervised clustering molecular subtypes and biological pathway analysis

Unsupervised classification was used to cluster 58 hippocampal samples (from the combined data of GSE1297, GSE84422, and GSE48350) into molecular subtypes. After receiving lipid metabolism DEGs from the Combat package, the “ConsensusClusterPlus” executed an unsupervised clustering algorithm. The consensus clustering algorithm was then subjected to k-means clustering and stability assessment. The “WGCNA” package module Eigengenes then scored the first principal components to identify correlations between the score and the grouped modules.

Examining the biological functions of coexpressed genes in clusters and grouped modules

The R package “GSVA” combined with the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis explored different biological functions between the clusters and grouped modules. The 15 most significant pathways were selected and plotted using the R “heatmap” package. The screening criteria were |logFC| > 0.1 and adjusted p < 0.05.

Analysis of immunocyte infiltration in clusters and grouped modules

Immune cell infiltration was evaluated with the ssGSEA (single-sample gene-set enrichment analysis) approach outlined above(Zhang et al. 2020). Gene sets for each immune cell type were derived from Charoentong, which contains a range of biomarkers for 23 different human immune cells, including plasma cells, B cells, T cells, and various myeloid cell subpopulations(Hu et al. 2023). ssGSEA immune cell enrichment scores were utilized to evaluate the relative proportion of immune cells in each cluster and grouped module. Spearman correlation analysis was conducted to evaluate DEGs and immune cell enrichment, and the R package “corrplot” was used to plot the matrix of immune cells and gene expression.

Lipid metabolism DEG profiles in brain cell subtypes

To gain insight into heterogeneous cell populations and evaluate the cell expression variability of lipid metabolism genes in AD brains, signatures of neuronal and non-neuronal cells obtained from scRNA-seq data have been established to evaluate transcriptome profiling at a single-cell level(Lau et al. 2020). Cluster-specific genes annotating cell types were obtained from the GSE163577 and GSE157827 datasets(Lau et al. 2020, Yang et al. 2022) and used to classify cell types. The R package “Seurat” was employed to investigate the single-cell RNA-seq data. The uniform manifold approximation and projection (UMAP) method executed a non-linear dimensional reduction.

Molecular network analysis of lipid metabolism genes between the brain and blood

To create an interaction network diagram of lipid metabolism DEGs from the AD brain and blood, we analyzed the gene network using the GeneMANIA database(Franz et al. 2018) with 21 DEGs from the brain and 13 from the whole blood. Data was visualized in the Cytoscape software.

Identifying the biological functions and signaling pathways of DEGs.

To gain further insight into the biological functions and signaling pathways of the lipid metabolism DEGs, the 21 genes from the AD hippocampus and the 13 genes from whole blood were subjected to GO and KEGG enrichment analyses. The R package "clusterProfiler" was used for GO enrichment analysis, including biological process (BP) and KEGG pathway analysis for DEGs present in the diseased brain or whole blood. Data were visualized using the "GOplot" R software package as a chord diagram or heatmap.

Establishment of molecular-based patient prognosis

Given the substantial expression differences revealed by unsupervised clustering, a molecular-based approach relying on subtype-specific gene expression could inform patient disease status. LASSO regression was performed with the "glmnet" data package to determine significant genes. Logistic regression (LR) with the "glmnet" data package and random forest (RF) regression with the "randomForest" data package was used to assess diagnostic utility. The nomogram was evaluated with the C-index, represented by the area under the curve (AUC) of time-related ROC curves.

Drug prediction for targeting DEGs

Data pertaining to medications that target lipid metabolism genes were retrieved from DrugBank (https://go.drugbank.com), a web database of comprehensive drug properties, mechanisms, interactions, and targets.

Statistical Analysis

Statistical analyses were performed using R software and GraphPad Prism (version 8.0.1). The t test and Mann–Whitney U test was selected according to whether the data conformed to a normal distribution. Quantitative values denote discontinuous data, while continuous data are represented as the means ± SEMs. Statistical significance was defined as p < 0.05.

Identification of Lipid Metabolism-Related Differential Genes in the AD hippocampus

Lipid metabolism DEGs were extracted from GSE48350 and GSE36980 and visualized on a volcano plot (Fig. 1A&B). A total of 128 DEGs were identified, 79 of which were downregulated and 49 of which were upregulated. Venn diagram analysis revealed six upregulated DEGs and 15 downregulated DEGs (Fig. 1C), and a circos plot was used to depict the chromosomal loci for 21 DEGs from the two datasets (Fig. 1D, Table S2). Transcript correlations among those lipid metabolism DEGs were plotted on a color matrix after filtration (Fig. 1E). PLD3, OXCT1, FABP3, PI4KA, and ELOVL4 expression were positively correlated with most other molecules, while AGT, ELOVL6, BDH2, ACSS3, and SLC44A1 were illustrated as having a negative correlation.

Unsupervised clustering and biological functions of differentially expressed genes in the AD brain

We next combined data from GSE1297, GSE84422, and GSE48350, which were classified into subtypes using DEGs identified from the AD hippocampus, and biological functions were analyzed (Fig. 2A-D). Numerical values indicate the likelihood that patients were distributed into two distinct clusters with a stable consensus clustering of K = 2 (Fig. 2A); the advantage of additional collections diminished beyond this point. Furthermore, we designed and validated an alternative subtype clustering scheme that uses the first principal component of the gene expression data. The majority of the high-scoring patients clustered in subcluster B (Fig. 2E); the score of Cluster B (weighted 97%) was consistently much higher than that of Cluster A (Fig. 2F).

We used GSVA to inform these two clusters' biological pathways from the gene sets' enrichment fraction. A heatmap of the 15 most significant pathways reveals 14 pathways in subtype B with substantial locus-specific gene expression. In contrast, transporter pathway expression was lower in the B cluster (Fig. 2G). The B cluster was unexpectedly enriched in a prominent pathway associated with Vibrio cholerae infection (Fig. 2G). The accuracy of the analysis was confirmed by examination of GSE84422 and GSE1297 (Fig S5A, C), indicating that the combination and identification of subtype-specific phenotypes has not affected reliability.

Analysis of neuroinflammation and inflammatory activation in stratified clusters

Next, we investigated correlations between lipid metabolism DEGs and inflammation activation. Single-cell signatures classified neuron and glial cell subpopulations. Significant enrichment was observed in both inhibitory and excitatory neurons in cluster B and the high-score group (Fig. 3A, B), while astrocytes, microglia, oligodendrocytes, and endothelial cells were reduced in these groups. Neurons and neural progenitor cells were enriched, while endothelial cells and most glial cells, including oligodendrocyte precursor cells (OPCs) in Cluster B, were depleted when subclassified by cell type using the CellMarker database (Fig. S2).

Brain interfaces experience structural and biological alterations during AD that allow the infiltration of peripheral immune cells into the brain(Jorfi et al. 2023). Similarly, we found that B cells, CD4⁺ T cells, mast cells, monocytes, natural killer (NK) cells, and Th 17 cells were enriched in Cluster B and the high-score group (Fig. 3C, D), and we observed immune cell activation in the GSE84422 and GSE1297 datasets (Fig. S3B, D). We examined the association between lipid metabolism DEGs in AD brains and subpopulations of infiltrating immunocytes. Eosinophils and NK T cells were negatively correlated with lipid metabolism DEGs, including PLD3, NDUFAB1, OXCT1, PI4KA, AACS, and ELOVL4. However, they were positively correlated with monocytes (Fig. 3E). Most genes were negatively associated with T and B immune cell subsets, while AGT and SLC44A1 were positively correlated with the activation of these cell subtypes. ELOVL7 and SGPP2 were explicitly expressed in endothelial cells, and SLC44A1 was mainly expressed in oligodendrocytes of the diseased hippocampus (Fig S4). Furthermore, in the AD PFC, AGT and ACSS3 were distinctly expressed in astrocytes (Fig S5), suggesting that neurons, glial cells, and immunocytes in the AD brain process distinct lipid metabolism states with cellular selectivity.

Lipid metabolism gene-predicted pathology stage and drug targeting

We used GSEA enrichment on the six most pertinent pathways in the KEGG analysis to inform the molecular pathways from our clustering. Six tracks related to neurodegenerative diseases, including AD, Huntington's disease (HD), Parkinson's disease (PD), glycolysis, gluconeogenesis, oxidative phosphorylation (OXPHS), and long-term potentiation (LTP), were active in the high-score group (Fig. 4A-F). Lipid metabolism DEGs were active in all six pathways, confirming the GSVA findings (Fig. 2E-G).

The Braak staging classification, a diagnostic tool for assessing AD progression, predicts patient-specific pathological progressive aberrations and amyloid protein regional distribution (Braak and Braak 1991). The expression levels of five specific genes, PLD3 (p = 0.028), NDUFAB1 (p = 0.032), AACS (p = 0.016), OXCT1 (p = 0.016), and PI4KA (p = 0.03), were significantly decreased in Braak stages IV-VI, demonstrating a correlation with disease severity (Fig. 4G-K). These results directly link lipid gene transcription, clinical outcomes, and disease progression. Based on the biological functions and target information from the DrugBank database, predicted small-molecule inhibitors have been identified for these lipid metabolism DEGs (Table S2, 3).

Characterization of DEGs in the peripheral blood of AD patients

To characterize the lipid metabolism gene expression in the periphery of AD, we analyzed the datasets of GSE63060 and GSE63061 from the whole blood. Venn diagram analysis yielded 369 DEGs, with 219 overlapping downregulated and 150 overlapping upregulated genes (Fig. 5A). The biological effects of these DEGs were explored using enrichment analysis, which indicated that most DEGs were involved in biological processes and pathways associated with AD, including cytoplasmic translation, oxidative phosphorylation, ribosomal biosynthesis, and ATP metabolism (Fig. 5B, C). Four most active pathways were related to the metabolism of lipids, including glycerophospholipids, glycosphingolipids (GSLs), sphingolipids, and the transcriptional regulation of white adipocyte differentiation (TRWAD). Interestingly, one pathway is associated with COVID-19 (Fig. 5D, E). We utilized the GSE177477 dataset from COVID-19-infected patients to analyze lipid metabolism DEGs; in contrast to AD, most genes were less expressed in COVID-19 samples and were enriched in two pathways: glycerophospholipid metabolism and reactome transcriptional regulation of white adipocyte differentiation (Fig. S6).

Construction and Blood Validation of a Diagnostic Model

Lipid metabolism DEGs were identified in the GSE63060 and GSE63061 datasets, and Venn plots revealed 12 upregulated and one downregulated gene. A diagnostic LASSO regression model was constructed using the 13 DEGs, with nine genes remaining after in-line filtration (Fig. 6B, C; Table S4). Using the approach of Han et al.(Han et al. 2014), a rigorous machine-learning method was employed to evaluate the potential diagnostic utility of these nine genes (Fig. 6D); the expression profile was accurately classified by algorithm performance with LR (AUC = 0.903) and RF (AUC = 1.00) in the training set (Fig. 6D). Internal and external validation by regression performance with the GSE63060 (Fig. 6E) and GSE63061 (Fig. 6F) datasets yielded discrimination values of LR (internal AUC = 0.806; external AUC = 0.701) and RF (internal AUC = 0.798; external AUC = 0.612). According to a nomogram assessment, DBI, MBOAT7, and RXRA have significant clinical diagnostic potential (Fig. 6G). Gene expression features were filtered to ensure that the transcripts had quantifiable chromosomal mapping reads that distinguished AD patient samples (Fig. 6H). These findings reveal differences in lipid metabolism gene expression in whole blood vs. hippocampal tissue and imply that lipid metabolism is aberrant in AD patients.

MCI is considered an early predementia stage (prodromal); lipid metabolism is altered in the MCI brain and serum(Wood et al. 2015). An enormous serum inflammatory element is found in the blood of individuals who restricted the MCI stage and in those with AD-related MCI (King et al. 2018). We considered the lipid metabolism transcriptome in MCI-stage subjects, and dementia AD patients would have a comparable profile. We examined the correlations and transcriptional dependence of lipid metabolism DEGs and found that all genes positively correlated with transcription, except for DBI (Fig. 7A, B). (Fig. 7) Additionally, compared to MCI, genes of CEBPB (p = 0.0085) and CEBPD (p = 0.028) in the GSE63060 dataset and CEBPB (p = 0.026) and EP300 (p = 0.028) in the GSE63060 dataset had reduced expression in AD, suggesting that decrease in CEBPB can help to diagnostically distinguish between MCI and AD (Fig. 7C, D).

Prediction of distinct immunocyte activation in AD blood

From whole blood lipid metabolism DEGs, patients were categorized into two subclusters (Fig. 8A-C) that were evaluated for correlations between the microenvironment and gene expression in the AD periphery and infiltrating immune cells. Activated CD4⁺ T, CD8⁺ T cells, dendritic cells, and Th2 cells were elevated in cluster B and the high-score group. However, MDSCs, macrophages, monocytes, NK T cells, neutrophils, and plasmacytoid dendritic cells were depleted in cluster B and the high-score group (Fig. 8D, E). (Fig. 8) We examined correlations between lipid metabolism DEGs and immune cell activation in the AD periphery. Activation of CD4⁺ T cells, CD8⁺ T cells, MDSCs, macrophages, monocytes, neutrophils, and plasmacytoid dendritic cells was correlated with lipid metabolism genes. Intriguingly, apart from DBI, the lipid metabolism DEGs negatively correlated with these cells (Fig. 8F).

Molecular network analysis of differential genes identified for central and peripheral regions

We examined the biological network behavior of lipid metabolism transcriptomes between the central and peripheral regions of AD; GO biological process enrichment analysis revealed striking differences between lipid metabolism DEGs in the periphery and CNS, with two restricted processes being enriched in the blood and 94 in the brain. Genes relevant to phosphatidylcholine metabolism were elevated in blood samples (Fig. 9A), while genes related to fatty acid metabolism and biosynthesis, fatty-acyl-CoA, and long-chain fatty acid metabolism were elevated in the AD brain (Fig. 9B). Although peripheral and central DEGs have distinct lipid metabolism profiles, their protein-protein interaction networks reveal substantial interplay (Fig. 9C).

Here, we cluster lipid transcriptomic profiles and correlate them with immune and nonimmune cell enrichment signatures in a large cohort of AD transcriptomes from the brain and peripheral blood. Our analysis highlights the correlation between the dysregulation of lipid metabolism, immune activity, and cellular pathophysiology, thus providing druggable targets and a diagnostic tool. AD individuals have distinct lipid metabolism gene expression patterns and cell type correlations. Most clustered genes are enriched for immunocytes, providing insights into the dysregulation of lipid metabolism that affects the immune microenvironment.

Previous GWAS have suggested genetic correlations with AD and APOE, NPC1, and LAMTOR5, which are linked to disruptions in lipid transport and endosomal/lysosomal irregularities, implying that lipid metabolism signaling pathways underlie vulnerability to AD neurodegeneration(Giri et al. 2016). In our bioinformatic enrichment analysis, 21 unique genes in the AD hippocampus are involved in lipid metabolism processes, including fatty acid synthesis and elongation, phospholipid hydrolysis, and choline transmembrane transport. Dysregulation of PLD3(Hinz and Geschwind 2017), NDUFAB1(Wu et al. 2023), OXCT1(Zhang et al. 2020), PI4KA(Koran et al. 2014), and AACS(Perkins et al. 2016) has been observed in AD patients, and reduced transcription of these genes have been found at later stages of the disease (Braak IV/VI). However, six of 21 genes were upregulated in the AD hippocampus, including ELOVL6 and SLC44A1, which were not previously reported. Analyzing subpopulation expression has shown that SLC44A1 is expressed explicitly in oligodendrocytes; in contrast, AGT and ACSS3 are expressed exclusively in astrocytes, suggesting that dysregulation of lipid metabolism gene expression occurs heterogeneously involving several cell types.

Analysis of AD peripheral blood identified genes related to energy metabolism and nucleic acid processing that correlated with various neurodegenerative diseases. GSEA revealed nine hub lipid metabolism genes linked to neurodegenerative diseases, particularly AD, which validates this analysis. Three of the nine genes (DBI, RXRA, and MBOAT7) may be useful in clinical diagnostics. DBI(Mills et al. 2013) and RXRA(Kölsch et al. 2009) encode proteins that function in steroidogenesis and cholesterol metabolism in AD, respectively. MBOAT7 encodes the lysophosphatidylinositol acyltransferase that transfers arachidonic acid to lysophosphatidylinositol to produce phosphatidylinositol. Mutations in MBOAT7 lead to intellectual disability with epilepsy and autistic features(Johansen et al. 2016). We initially linked MBOAT7 to AD as a peripheral diagnostic marker, but we later identified CEBPB as a discriminant between MCI and AD that distinguishes the prodromal and clinical stages.

Neuroinflammation characterizes AD with the activation of immune cells and infiltration of activated peripheral immune cells into the CNS during AD progression. The immune infiltration mediated by naive B cells, endothelial cells, activated/resting NK cells, macrophages, CD4⁺ T cells, memory B cells, and dendritic cells participates in AD development (Gu et al. 2022, McLarnon 2021, Tian et al. 2022, Zlokovic 2011). An essential strength of this study is the exploration of lipid metabolism in inflammation and neuroinflammation. It has been found that distinct distributions and expression profiles between neurons and glial cells, innate immunocytes (e.g., mast cells and monocytes), and adaptive immunocytes (e.g., CD4⁺ T, CD8⁺ T) in the unsupervised clusters derived from brain and peripheral blood. In AD progression, immune T cells, such as CD8⁺ T cells, penetrate the brain parenchyma and bind tightly with neurons and microglia(Jorfi et al. 2023, Unger et al. 2020), as demonstrated by functional transcriptomic analysis of CD8⁺ T cells, which reveals distinctive activation profiles that are functionally involved in AD (Altendorfer et al. 2022). The lipid metabolism genes NDUFAB1, OXCT1, PI4K4, and ELOVL4 were significantly elevated in monocytes but reduced in T cells and NK cells. Although the molecular mechanisms remain incompletely understood, the transcriptional differences in lipid metabolism between the brains and blood cells of AD patients may link the varying contributions of these cells to AD pathology.

Although experimental evidence indicates eosinophil infiltration in amyloid plaque phagocytes, eosinophil can damage myelinated nerve fibers through secreted neurotoxin(Durack et al. 1979), and defects the recruitment of peripheral neutrophils (Fiala et al. 2005) even in early AD. Neutrophils promote disease pathology and cognitive decline in mouse models and AD patients, and depletion or inhibition of neutrophils restores the neuropathology(Gabbita et al. 2015, Zenaro et al. 2015). Conditional deletion of fatty acid synthase results in selective inhibition of neutrophil production, although without disrupting neutrophil differentiation. Consistent with previous reports(Järemo et al. 2013, Wu et al. 2020), we found elevated eosinophils and neutrophils in peripheral circulation in lipid genes clustering subtypes, which may indicate lipid metabolism functional involvement in systemic innate immune response in AD.

Unexpectedly, within the complete blood sample from GSE177477, two pathways were poorly expressed in COVID-19 and opposition to AD. COVID-19 patients exhibit neurological symptoms, including stroke, loss of taste and smell, and cognitive alterations. Transcriptomic analysis of COVID-19 victims revealed marked microglial activation(Fullard et al. 2021) and a significant influx of activated CD8⁺ T cells into the brain parenchyma (Yang et al. 2021). Single-cell sequencing of cerebrospinal fluid from patients with persistent COVID-19 infection revealed changes in CD4⁺ T-cell depletion. The COVID-19 pandemic profoundly impacted dementia patients, with more than two new symptoms occurring (Heming et al. 2021). Cellular lipid synthesis is needed for COVID-19 virus replication(Arthur et al. 2022), and pharmacologic inhibition of FAS blocks COVID-19 replication(Chu et al. 2021).

Treatment strategies based on the amyloid cascade hypothesis have proven ineffective, so identifying and exploring novel therapeutic targets is crucial to improving treatment. We used targeted drug prediction on DEGs within the DrugBank database to identify specific drugs that target central and peripheral protein products of these genes. Several medications target lipid metabolism for both immune and degenerative diseases. Sparsentan (DB12548) targets AGT in adults with primary immunoglobulin A nephropathy, which is characterized by rapid deterioration of galactose-deficient IgA1 antibodies; endothelin and angiotensin II receptor antagonists reduce proteinuria(Komers et al. 2020). By inhibiting IL-15 signaling mediated by angiotensin II or endothelin-1, sparsentan determines renal CD8⁺ TRM cell fate through a mechanistic pathway(Li et al. 2022). Zinc and copper perturbations have been observed in patients and AD models, which may indicate amyloidogenesis and tauopathy(Moynier et al. 2020, Solovyev et al. 2021). The substrates of SLC44A1, choline (DB00122), and choline salicylate (DB14006) are precursors of acetylcholine and are critical in lipid metabolism(Kolykhalov et al. 2022). Lipid transporter ALC44A2 inhibitors are promising candidates for treating immune and degenerative diseases(Traiffort et al. 2013).

In summary, we identified DEGs involved in lipid metabolism and performed unsupervised clustering to classify molecular subtypes of hub genes from both the brain and peripheral blood of AD patients through a comprehensive bioinformatics analysis. Furthermore, we elucidated the strong association between the hub lipid metabolism genes and the enrichment of immune cells, the immune microenvironment, as well as AD pathology. The neuroinflammation and immune infiltration revealed a marked disparity in cell subpopulations among the clustering subtypes in both the brain and blood samples. Five hub lipid metabolism genes (PLD3, NDUFAB1, OXCT1, PI4KA, and AACS) expression levels in the brain significantly decreased in the later stages of AD, and their expression positively correlated with monocytes but negatively with T cells, B cells, and NK cells, although not significantly with PLD3. DBI, MBOAT7, and RXRA, from the 13-lipid metabolism DEGs identified from the AD blood, have significant clinical diagnostic potential. In contrast to the other genes, DBI has an inverse correlation with the enrichment of the immune subpopulation. Simultaneously, an enrichment analysis was carried out on the biological functions and molecular pathways of genes related to lipid metabolism in disease and the prognosis of potential drugs. These findings have significantly contributed to an enhanced comprehension of the pathology underlying AD, thereby offering novel perspectives for investigating potential diagnostic biomarkers and therapeutic strategies.

AD: Alzheimer’s Disease; DEGs: differentially expressed genes; Aβ:amyloid β-protein; BBB: blood‒brain barrier; GWAS: genome-wide association studies; APOE: Apolipoprotein E; CNS: central nervous system; GEO: Gene Expression Omnibus; RNA-seq: RNA-sequencing; GSEA: Gene Set Enrichment Analysis; scRNA-seq: single-cell sequencing; PFC: prefrontal cortex; CDF: cumulative distribution function; GSVA: Gene Set Variation Analysis; KEGG: Kyoto Encyclopedia of Genes and Genomes; SVR: support vector regression; NLDR: nonlinear dimensional reduction; ssGSEA: single-sample gene-set enrichment analysis; UMAP: Uniform Manifold Approximation and Projection; GO: Gene Ontology; BP: biological process; LR: logistic regression; RF: random forest; AUC: area under the curve; NPC: neural progenitor cells; OPCs: oligodendrocyte precursor cells; HD: Huntington’s disease; PD: Parkinson’s disease; OXPHOS: oxidative phosphorylation; LTP: long-term potentiation; GSL: glycosphingolipid; PDC: plasmacytoid dendritic cell; WGCNA: Weighted Gene Correlation Network Analysis; CLU: clusterin; SORL1: sortilin-related receptor 1; ABCA7: ATP-binding cassette, subfamily A, member 7; EDN: eosinophil-derived neurotoxin; LPIAT1: lysophosphatidylinositol acyltransferase; FAS: fatty acid synthase; CSF: cerebrospinal fluid; ROS: reactive oxygen species; Gd-IgA1: galactose-deficient IgA1; ET-1: endothelin-1; COX: cytochrome c oxidase; MAO: monoamine oxidase; SOD: superoxide dismutase; MCI: mild cognitive impairment.

Acknowledgments

The findings presented here are, in whole or partly, sourced from the GEO databases (https://www.ncbi.nlm.nih.gov/geo/). We would like to thank Charlesworth (https://www.cwauthors.com/) for English language editing.

Funding

This work was supported by the special launch fund from Soochow University, Science and Technology Program of Suzhou (2022SS02), the National Natural Science Foundation of China (NSFC) (82171414), Suzhou Science and Technology Plan Project (2022SS02), the Science and Technology Development Fund of Shanghai Pudong New Area (PKJ2021-Y07), and the Clinical Plateau Discipline Construction Project of Shanghai Pudong New Area Health Committee (PWYgy2021-05).

Author contributions

M.H.R., Z.Y., and S.Y. conceived and designed the experiments. S.Y., J.M.N., L.X., and M.X.J. performed bioinformatics analyses and prepared graphs. M.H.R., Z. Y., S.Y, and J.M.N drafted the manuscript. M. Y. Z, D. H. H., and Z. T. validated the data and revised the manuscript. M.H.R. and Z.Y. acquired funding and provided supervision. All authors contributed to the edited, reviewed, and approved the manuscript.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

No authors had conflicts of interest relevant to this study.

Akyol S, Ugur Z, Yilmaz A et al. (2021) Lipid profiling of Alzheimer’s disease brain highlights enrichment in Glycerol(phospho)lipid, and sphingolipid metabolism. Cells 10. https://doi.org/10.3390/cells10102591
Altendorfer B, Unger MS, Poupardin R et al. (2022) Transcriptomic Profiling Identifies CD8(+) T Cells in the Brain of Aged and Alzheimer’s Disease Transgenic Mice as Tissue-Resident Memory T Cells. J Immunol 209:1272-1285. https://doi.org/10.4049/jimmunol.2100737
Arthur P, Li CY, Southern Indiana Dementia W (2022) Living with dementia during the COVID-19 pandemic: A nationwide survey informing the american experience. J Alzheimers Dis Rep 6:733-737. https://doi.org/10.3233/ADR-220085
Braak H, Braak E (1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82:239-259. https://doi.org/10.1007/BF00308809
Chu J, Xing C, Du Y et al. (2021) Pharmacological inhibition of fatty acid synthesis blocks SARS-CoV-2 replication. Nat Metab 3:1466-1475. https://doi.org/10.1038/s42255-021-00479-4
Dai L, Shen Y (2021) Insights into T-cell dysfunction in Alzheimer’s disease. Aging Cell 20:e13511. https://doi.org/10.1111/acel.13511
Durack DT, Sumi SM, Klebanoff SJ (1979) Neurotoxicity of human eosinophils. Proc Natl Acad Sci U S A 76:1443-1447. https://doi.org/10.1073/pnas.76.3.1443
Fiala M, Lin J, Ringman J et al. (2005) Ineffective phagocytosis of amyloid-beta by macrophages of Alzheimer’s disease patients. J Alzheimers Dis 7:221-232; discussion 255-262. https://doi.org/10.3233/jad-2005-7304
Franz M, Rodriguez H, Lopes C et al. (2018) GeneMANIA update 2018. Nucleic Acids Res 46:W60-W64. https://doi.org/10.1093/nar/gky311
Fullard JF, Lee HC, Voloudakis G et al. (2021) Single-nucleus transcriptome analysis of human brain immune response in patients with severe COVID-19. Genome Med 13:118. https://doi.org/10.1186/s13073-021-00933-8
Gabbita SP, Johnson MF, Kobritz N et al. (2015) Oral TNFalpha modulation alters neutrophil infiltration, improves cognition and diminishes tau and amyloid pathology in the 3xTgAD mouse model. PLOS One 10:e0137305. https://doi.org/10.1371/journal.pone.0137305
Giri M, Zhang M, Lü Y (2016) Genes associated with Alzheimer’s disease: an overview and current status. Clin Interv Aging 11:665-681. https://doi.org/10.2147/CIA.S105769
Gu X, Lai D, Liu S et al. (2022) Hub genes, diagnostic model, and predicted drugs related to iron metabolism in Alzheimer’s disease. Frontiers in Aging Neurosci 14. https://doi.org/10.3389/fnagi.2022.949083
Gustavsson EK, Zhang D, Reynolds RH et al. (2022) ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinformatics 38:3844-3846. https://doi.org/10.1093/bioinformatics/btac409
Han L, Yuan Y, Zheng S et al. (2014) The Pan-Cancer analysis of pseudogene expression reveals biologically and clinically relevant tumour subtypes. Nat Commun 5:3963. https://doi.org/10.1038/ncomms4963
Heming M, Li X, Räuber S et al. (2021) Neurological manifestations of COVID-19 feature T Cell exhaustion and dedifferentiated monocytes in cerebrospinal fluid. Immunity 54:164–175.e166 e166. https://doi.org/10.1016/j.immuni.2020.12.011
Heneka MT, Carson MJ, El Khoury J et al. (2015) Neuroinflammation in Alzheimer’s disease. Lancet Neurol 14:388-405. https://doi.org/10.1016/S1474-4422(15)70016-5
Hinz FI, Geschwind DH (2017) Molecular genetics of neurodegenerative dementias. Cold Spring Harb Perspect Biol 9. https://doi.org/10.1101/cshperspect.a023705
Hu C, Li T, Xu Y et al. (2023) CellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data. Nucleic Acids Res 51:D870-D876. https://doi.org/10.1093/nar/gkac947
Järemo P, Milovanovic M, Buller C et al. (2013) Alzheimer’s disease and granulocyte density diversity. Eur J Clin Investig 43:545-548. https://doi.org/10.1111/eci.12072
Johansen A, Rosti RO, Musaev D et al. (2016) Mutations in MBOAT7, encoding lysophosphatidylinositol acyltransferase I, lead to intellectual disability accompanied by epilepsy and autistic features. Am J Hum Genet 99:912-916. https://doi.org/10.1016/j.ajhg.2016.07.019
Jorfi M, Maaser-Hecker A, Tanzi RE (2023) The neuroimmune axis of Alzheimer’s disease. Genome Med 15:6. https://doi.org/10.1186/s13073-023-01155-w
King E, O’Brien JT, Donaghy P et al. (2018) Peripheral inflammation in prodromal Alzheimer’s and Lewy body dementias. J Neurol Neurosurg Psychiatry 89:339-345. https://doi.org/10.1136/jnnp-2017-317134
Knopman DS, Amieva H, Petersen RC et al. (2021) Alzheimer disease. Nat Rev Dis Primers 7:33. https://doi.org/10.1038/s41572-021-00269-y
Kölsch H, Lütjohann D, Jessen F et al. (2009) RXRA gene variations influence Alzheimer’s disease risk and cholesterol metabolism. J Cell Mol Med 13:589-598. https://doi.org/10.1111/j.1582-4934.2009.00383.x
Kolykhalov IV, Androsova LV, Gavrilova SI (2022) Clinical and immunological effects of choline alfoscerate in the treatment of amnestic type Mild Cognitive Impairment. Zh Nevrol Psikhiatr Im S S Korsakova 122:59-66. https://doi.org/10.17116/jnevro202212211259
Komers R, Diva U, Inrig JK et al. (2020) Study design of the phase 3 sparsentan versus irbesartan (Duplex) study in patients with focal segmental glomerulosclerosis. Kidney Int Rep 5:494-502. https://doi.org/10.1016/j.ekir.2019.12.017
Koran ME, Hohman TJ, Meda SA et al. (2014) Genetic interactions within inositol-related pathways are associated with longitudinal changes in ventricle size. J Alzheimers Dis 38:145-154. https://doi.org/10.3233/JAD-130989
Kunkle BW, Grenier-Boley B, Sims R et al. (2019) Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Abeta, tau, immunity and lipid processing. Nat Genet 51:414–430. https://doi.org/10.1038/s41588-019-0358-2
Lau SF, Cao H, Fu AKY et al. (2020) Single-nucleus transcriptome analysis reveals dysregulation of angiogenic endothelial cells and neuroprotective glia in Alzheimer’s disease. Proc Natl Acad Sci U S A 117:25800-25809. https://doi.org/10.1073/pnas.2008762117
Li L, Tang W, Zhang Y et al. (2022) Targeting tissue-resident memory CD8(+) T cells in the kidney is a potential therapeutic strategy to ameliorate podocyte injury and glomerulosclerosis. Mol Ther 30:2746-2759. https://doi.org/10.1016/j.ymthe.2022.04.024
Liu Y, Thalamuthu A, Mather KA et al. (2021) Plasma lipidome is dysregulated in Alzheimer’s disease and is associated with disease risk genes. Transl Psychiatry 11:344. https://doi.org/10.1038/s41398-021-01362-2
McLarnon JG (2021) A leaky blood-brain barrier to fibrinogen contributes to oxidative damage in Alzheimer’s disease. Antioxidants (Basel) 11. https://doi.org/10.3390/antiox11010102
Mills JD, Nalpathamkalam T, Jacobs HI et al. (2013) RNA-Seq analysis of the parietal cortex in Alzheimer’s disease reveals alternatively spliced isoforms related to lipid metabolism. Neurosci Lett 536:90-95. https://doi.org/10.1016/j.neulet.2012.12.042
Moynier F, Borgne ML, Lahoud E et al. (2020) Copper and zinc isotopic excursions in the human brain affected by Alzheimer’s disease. Alzheimers Dement (Amst) 12:e12112. https://doi.org/10.1002/dad2.12112
Perkins M, Wolf AB, Chavira B et al. (2016) Altered energy metabolism pathways in the posterior cingulate in young adult apolipoprotein E varepsilon4 carriers. J Alzheimers Dis 53:95-106. https://doi.org/10.3233/JAD-151205
Ritchie ME, Phipson B, Wu D et al. (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43:e47. https://doi.org/10.1093/nar/gkv007
Rosenthal SB, Wang H, Shi D et al. (2022) Mapping the gene network landscape of Alzheimer’s disease through integrating genomics and transcriptomics. PLOS Comput Biol 18:e1009903. https://doi.org/10.1371/journal.pcbi.1009903
Solovyev N, El-Khatib AH, Costas-Rodríguez M et al. (2021) Cu, Fe, and Zn isotope ratios in murine Alzheimer’s disease models suggest specific signatures of amyloidogenesis and tauopathy. J Biol Chem 296:100292. https://doi.org/10.1016/j.jbc.2021.100292
Soria Lopez JA GH, Léger GC (2019) Alzheimer's disease. Handb Clin Neurol 167:231–255. https://doi.org/10.1016/B978-0-12-804766-8.00013-3
Tian Y, Lu Y, Cao Y et al. (2022) Identification of diagnostic signatures associated with immune infiltration in Alzheimer’s disease by integrating bioinformatic analysis and machine-learning strategies. Front Aging Neurosci 14:919614. https://doi.org/10.3389/fnagi.2022.919614
Traiffort E, O’Regan S, Ruat M (2013) The choline transporter-like family SLC44: properties and roles in human diseases. Mol Aspects Med 34:646-654. https://doi.org/10.1016/j.mam.2012.10.011
Unger MS, Li E, Scharnagl L et al. (2020) CD8(+) T-cells infiltrate Alzheimer’s disease brains and regulate neuronal- and synapse-related gene expression in APP-PS1 transgenic mice. Brain Behav Immun 89:67-86. https://doi.org/10.1016/j.bbi.2020.05.070
Whiley L, Sen A, Heaton J et al. (2014) Evidence of altered phosphatidylcholine metabolism in Alzheimer's disease. Neurobiol Aging 35:271-278. https://doi.org/10.1016/j.neurobiolaging.2013.08.001
Wood PL, Barnette BL, Kaye JA et al. (2015) Non-targeted lipidomics of CSF and frontal cortex grey and white matter in control, mild cognitive impairment, and Alzheimer’s disease subjects. Acta Neuropsychiatr 27:270-278. https://doi.org/10.1017/neu.2015.18
Wood PL, Medicherla S, Sheikh N et al. (2015) Targeted lipidomics of fontal cortex and plasma diacylglycerols (DAG) in mild cognitive impairment and Alzheimer’s disease: validation of DAG accumulation early in the pathophysiology of Alzheimer’s disease. J Alzheimers Dis 48:537-546. https://doi.org/10.3233/JAD-150336
Wu CY, Bawa KK, Ouk M et al. (2020) Neutrophil activation in Alzheimer’s disease and mild cognitive impairment: A systematic review and meta-analysis of protein markers in blood and cerebrospinal fluid. Ageing Res Rev 62:101130. https://doi.org/10.1016/j.arr.2020.101130
Wu Z, Liu P, Huang B et al. (2023) A novel Alzheimer’s disease prognostic signature: identification and analysis of glutamine metabolism genes in immunogenicity and immunotherapy efficacy. Sci Rep 13:6895. https://doi.org/10.1038/s41598-023-33277-x
Xu J, Bankov G, Kim M et al. (2020) Integrated lipidomics and proteomics network analysis highlights lipid and immunity pathways associated with Alzheimer’s disease. Transl Neurodegener 9:36. https://doi.org/10.1186/s40035-020-00215-0
Yang AC, Kern F, Losada PM et al. (2021) Dysregulation of brain and choroid plexus cell types in severe COVID-19. Nature 595:565-571. https://doi.org/10.1038/s41586-021-03710-0
Yang AC, Vest RT, Kern F et al. (2022) A human brain vascular atlas reveals diverse mediators of Alzheimer’s risk. Nature 603:885-892. https://doi.org/10.1038/s41586-021-04369-3
Zang F, Zhu Y, Liu X et al. (2021) Polygenic effects of the lipid metabolic pathway accelerated pathological changes and disrupted default mode network trajectory across the Alzheimer’s disease spectrum. J Clin Psychiatry 82. https://doi.org/10.4088/JCP.20m13739
Zenaro E, Pietronigro E, Della Bianca V et al. (2015) Neutrophils promote Alzheimer’s disease-like pathology and cognitive decline via LFA-1 integrin. Nat Med 21:880-886. https://doi.org/10.1038/nm.3913
Zhang B, Wu Q, Li B et al. (2020) m(6)A regulator-mediated methylation modification patterns and tumor microenvironment infiltration characterization in gastric cancer. Mol Cancer 19:53. https://doi.org/10.1186/s12943-020-01170-0
Zhang F, Zhong SR, Yang SM et al. (2020) Identification of potential therapeutic targets of Alzheimer’s disease by weighted gene co-expression network analysis. Chin Med Sci J 35:330-341. https://doi.org/10.24920/003695
Zlokovic BV (2011) Neurovascular pathways to neurodegeneration in Alzheimer’s disease and other disorders. Nat Rev Neurosci 12:723-738. https://doi.org/10.1038/nrn3114

No competing interests reported.

Download PDF

Journal Publication

published 22 May, 2024

Read the published version in Journal of Molecular Neuroscience →

Editorial decision: Revision requested
19 Mar, 2024
Reviewers agreed at journal
03 Mar, 2024
Reviews received at journal
02 Mar, 2024
Reviewers agreed at journal
02 Mar, 2024
Reviewers invited by journal
02 Mar, 2024
Editor assigned by journal
22 Feb, 2024
Submission checks completed at journal
22 Feb, 2024
First submitted to journal
22 Feb, 2024

You are reading this latest preprint version

Transcriptomic analysis of lipid metabolism genes in Alzheimer’s disease: highlighting pathological outcomes and compartmentalized immune status

Status:

Journal Publication

Version 1

Abstract

Figures

INTRODUCTION

METHODS

Data preparation

Unsupervised clustering molecular subtypes and biological pathway analysis

Examining the biological functions of coexpressed genes in clusters and grouped modules

Analysis of immunocyte infiltration in clusters and grouped modules

Lipid metabolism DEG profiles in brain cell subtypes

Molecular network analysis of lipid metabolism genes between the brain and blood

Establishment of molecular-based patient prognosis

Drug prediction for targeting DEGs

Statistical Analysis

RESULTS

Identification of Lipid Metabolism-Related Differential Genes in the AD hippocampus

Unsupervised clustering and biological functions of differentially expressed genes in the AD brain

Analysis of neuroinflammation and inflammatory activation in stratified clusters

Lipid metabolism gene-predicted pathology stage and drug targeting

Characterization of DEGs in the peripheral blood of AD patients

Construction and Blood Validation of a Diagnostic Model

Prediction of distinct immunocyte activation in AD blood

Molecular network analysis of differential genes identified for central and peripheral regions

DISCUSSION

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1