Identification of hub genes and molecular genetic mechanisms in Alzheimer’s Disease and Schizophrenia

2.2. Data Preprocessing and Differential Expressed Analysis

GEO query (http://www.bioconductor.org/packages/release/bioc/html/GEOquery.html) was used to obtain the normalized expression profiles. Limma (Linear Models for Microarray Analysis) (https://bioconductor.org/packages/release/bioc/html/limma.html) packages were used to screen differentially expressed genes (DEGs). The Benjamini-Hochberg method was used to adjust the original p-values, and the false discovery rate (FDR) procedure was used to calculate the fold-changes (FC). Gene’s expression values of the |log2 FC|>1and adjusted p < 0.05 were used to filter the AD-DEGs. Meanwhile, the |log2 FC|>0.263 and the adjusted p < 0.05 were used to identify the SCZ-DEGs. Given that schizophrenia does not have gross brain pathology, the disorder involves subtle pathological changes within specific neural cell populations and in cell–cell communication [8]. Additionally, Venn diagramswe was calculated and made for co-DEGs for AD-DEGs and SZ-DEGs.

2.3. GO annotation and Pathway Enrichment Analysis

To assign gene ontology terms for their possible roles in the biological process, the Bioconductor R package ClusterProfiler [13] was applied to explore the functions among the genes of interest, with a cut-off criterion of an adjusted p-value (p < 0.05). The Kyoto Encyclopedia for Genes and Genomes (KEGG)-based screening (https://www.kegg.jp/) was also performed to identify the role of the DEGs and hub genes in various metabolic pathways. Gene Set Enrichment Analysis (GSEA) is used to assess whether a predefined gene set shows statistically significant differences between the two groups using an R package cluster profiler for AD and SCZ. Expression datasets that collapsed to the gene symbol and phenotype information were uploaded to the GSEA for enrichment analysis with default parameters. Enrichment results with an adjusted p value < 0.05 as well as an FDR < 0.25 were considered statistically significant.

2.4. PPI Network Construction and Hub Gene analysis

Protein-protein interactions (PPI) networks, including direct physical interaction of proteins and indirect functions, were predicted using the Search Tool for the Retrieval of Interacting Genes (STRING, https://string-db.org/) database. DEGs were mapped to STRING later, with the cut-off criterion of combined score greater than 0.4. The PPI network was visualized with the open-source bioinformatics software Cytoscape (version 3.7.2). CytoHubba, a Cytoscape plugin, was used to obtain the network center nodes (hub genes). The proteins expressed by the central node were usually an important protein/gene with critical physiological function. Ten genes in the maximum correlation criterion (MCC) were chosen by the CytoHubba plugin and sequentially ordered. A redder color represents more forward rankings.

3.1 Identification of DEGs in AD and SZ

After normalization, the mean gene expression values for each sample in AD and SZ were fundamental equal. By limma package (Version 3.26.9),2281 DEGs, including 601 upregulated and 1680 downregulated DEGs, were identified in GSE58793 (SZ)(Figure.1.a). While in GSE5281 (AD) 2769 DEGs, including 1437 upregulated and 1332 downregulated DEGs, were identified(Figure.1.b). Then we mapped these top 50 differentially expressed genes into heatmaps to assess the differences in expression between disease group and control group (Figure.1.c). As shown in (Figure.1.d), 613 overlapping of DEGs were found in SZ and AD, including 222 co-upregulated DEGs and 391 co-downregulated DEGs.

3.2. Functional annotation of DEGs

Then we explore the potentially altered functional characteristics associated with the DEGs in AD and SCZ, and Gene Ontology (GO) analysis was carried out for the differences in the biological processes between the disease group and the control group in AD and SCZ respectively.

In AD, genes upregulated in the disease group were major involved in protein modification, including regulation of chromatin organization, positive regulation of histone deacetylation, regulation of histone modification, and positive regulation of protein deacetylation. In contrast, downregulated genes in the AD disease group were tiedly related to metabolic process, including ATP metabolic process, cellular respiration, RNA catabolic process and regulation of mRNA metabolic/cellular amino acid metabolic process. Moreover, most of the positively related genes were enriched in KEGG terms including long-term depression, long-term potentiation. And the negatively correlated genes were enriched within the citrate cycle (TCA cycle), biosynthesis of amino acids, and amyotrophic lateral sclerosis. Next, validation was carried out by GSEA analyses, showing highly expressed genes that were significantly associated with long-term depression and estrogen signaling pathway, long-term potentiation, dopaminergic synapse, and Amoebiasis. The low expression gene group was significantly associated with oxidative phosphorylation, Huntington disease, proteasome, Parkinson disease, carbon metabolism, metabolic pathways, and Amyotrophic lateral sclerosis.

In SCZ, genes upregulated and downregulated in the GO terms and the KEGG pathways are shown as Fig. 3. Our GSEA analysis results indicated that the low expression genes were distinctly enriched in pathways of neurodegeneration disease, including AD, prion disease, Huntington disease(HD), Amyotrophic lateral sclerosis(ALS).

Further, we conducted the KEGG pathway and GO enrichment analysis of the 222 co-upregulated and 391 co-downregulated DEGs to study the functions of the 613 overlapping genes. And 222 overlapping co-upregulated genes within the biological processes (BP) were found closely related to the regulation of chromatin organization, positive regulation of histone deacetylation, regulation of histone modification, positive regulation of protein deacetylation, and regulation of protein-containing complex assembly. (Table 1) (Figure.2a). The BP of the 391 co-downregulated DEGs was primarily related to the regulation and activation of the innate immune response, ATP metabolic process, anaphase-promoting complex-dependent catabolic process, antigen processing and presentation of exogenous peptide antigen and regulation of stem cell differentiation. (Table 2) (Figure.2b). Moreover, in the KEGG pathway enrichment analysis, positively related co-DEGs were enriched within the MAPK signaling,cancer and the mTOR signaling pathways (Figure.2.c), while the negatively correlated genes were enriched within pathways of multiple neurodegeneration disease, AD,HD, ALS included(Table 3) (Figure.2d).

3.3 PPI Network Construction and Hub Genes Identification

Overlapping genes were analyzed by PPI network, 222 co-upregulated DEGs and 391 co-downregulated DEGs were established using the STRING database. In PPI networks, hub genes were defined as genes with stronger interactions with numerous other genes. And hub genes are potential drivers of the pathology of the diseases. In order to screen hub genes among all the DEGs via the MCC scores, cytoHubba plugin for Cytoscape was used. Interestingly, all these 10 hub genes got from screening were downregulated co-DEGs (Figure.3). And proteasome subunit alpha 5 (PSMA5), proteasome subunit beta 7 (PSMB7), proteasome 26S subunit, non-ATPase 12 (PSMD12), proteasome subunit alpha 1 (PSMA1), proteasome 26S subunit, ATPase 3 (PSMC3), proteasome 26S subunit, non-ATPase 4 (PSMD4), proteasome subunit beta 3 (PSMB3), proteasome subunit beta 1 (PSMB1), proteasome 26S subunit, non-ATPase 1 (PSMD1), and proteasome subunit alpha 7 (PSMA7) are the top 10 genes with the highest MCC sores, respectively .

3.4 GO Enrichment Analysis of the hub genes

All ten hub genes are downregulated DEGs both in SCZ and AD. Using the R package clusterprofiler. And among those genes, the top 10 GO terms were primarily related to the regulation of hematopoietic stem cell differentiation, regulation of cellular amino acid metabolic process, TAP-dependent,antigen processing and presentation of exogenous peptide antigen via MHC class I, antigen processing and presentation of exogenous peptide antigen via MHC class I, regulation of cellular amine metabolic process, regulation of transcription from RNA polymerase II promoter in response to hypoxia, anaphase-promoting complex-dependent catabolic process, regulation of hematopoietic progenitor cell or stem cell differentiation, and SCF-dependent proteasomal ubiquitin-dependent protein catabolic process. (Figure.4) (Table 4).

4.1 Biological function of the 613 Overlapping Genes

Some well-known pathways associated with AD and SCHIZ, such as the ATP metabolic and cellular amino acid metabolic processes, have also been observed in our study. These processes will not be discussed in this article [16, 17]. GO enrichment analysis of the co-DEGs also involved positive regulation in chromatin organization, histone deacetylation, and protein deacetylation, which, to some extent, related to epigenetics modifications. Modifications of histone proteins are associated with chromatin structure and play a pivotal role in epigenetic regulation of transcription[18]. Histone acetylation disrupts the structured arrangement of histone proteins and loosens the chromatin structure. This process makes it more accessible to transcription binding, while histone deacetylation removes the acetyl groups and generally links to chromatin inactivation. Experimental evidence suggests that treatment with histone deacetylase inhibitors ameliorates neuronal deficiencies and enhances synaptic plasticity, memory, and learning as a promising new strategy for neurodegenerative and psychiatric disorders[19]. Protein-protein interaction, deacetylation mechanism of non-histone protein, histone post-translational modifications and direct association with disease proteins have also been proved to link to neuronal imbalance, as well as transcriptional regulation[20]. These results are consistent with our analysis, and provides an open mechanism research point in epigenetic modifications of neuronal disease development. Therefore, new insights into the etiology of the pathogenesis of AD and SCHIZ, as well as new avenues for developing therapeutic strategies, remain to be explored[21].

In KEGG enrichment, the upregulated co-DEGs were enriched within the interleukin-17 (IL-17) signaling pathway (Table.6). Interleukin-17a (IL-17a) was produced by Th17 cells. For pregnant mice suffering immune system activation due to infections or autoinflammatory syndromes, IL-17a may increase the risk for neurodevelopmental disorders in offspring[22]. At the interface of neurological and psychiatric disorders, immune processes play an important role in central nervous system health and disease; the immune system contributes to overall CNS environmental stability, brain reserves and resilience[23]. Genome-wide association studies (GWAS) in patients with neurodegenerative disease are preferentially enriched within the enhancer sequences implicating innate immune processes [24–28]. Preclinical, experimental, and bioinformatic analysis showed that activation of the immune system is accompanied by AD pathology, which contributes to the pathogenesis of AD[29]. Over the past three decades, the causal relationship between pathogens in Alzheimer's disease has been repeatedly postulated[30, 31]. Associations between Alzheimer's disease and infectious burden was investigated in a A cross-sectional study. Results from this study showed that patients with Alzheimer’s disease had higher prevalence in prior infection with CMV, HSV-1, B. burgdorferi, C. pneumoniae, and H. pylori than age-matched healthy controls[25, 32]. Pathological assumption was then raised that pathogens may cause neurological damage by eliciting neuroinflammation in CNS by directly cross the blood-brain barrier[30]. Moreover, elders with a higher IB would develope worse cognition and higher serum Aβ levels, not only AD patients but also healthy controls[32]. As a result, more than one possibility can be proposed for the association between suspected pathogens and Alzheimer's disease. One is that Alzheimer's patients are more susceptible to microbial infections. Bacterial lipopolysaccharide and viral surface proteins shared same receptors that can sense pathogen-associated molecular patterns can both be triggered by Aβ aggregates [33]. The other is that microbial infections have a contributory role in progression of Alzheimer’s disease[34]. A genome-wide association study study examined the association between schizophrenia and markers near the region of the major histocompatibility complex (MHC) on chromosome 6. Many immune-related genes was involved in this association study, including antigen presentation and inflammatory mediators[35].

4.2. Hub Genes from Gene Co Expression Network

Among the co-DEGs, 10 hub genes were selected according to the maximum correlation criterion (MCC) scores (Table 2). Here, we focus on the analysis of the following genes: PSMA5, PSMA7, PSMA1, PSMD4, PSMD1, PSMD12, PSMB7, PSMC3, PSMB1, and PSMB3

PSMA5, PSMA7, PSMA1, PSMB7, PSMB1, PSMB3, and PSMC3 are the subunit of the 20S proteasome, while PSMD4, PSMD1, and PSMD12 are the subunits of 26S proteasome. Apart from mitochondria, other important aging-related transcriptomic changes were the downregulation of genes related to proteasomal functions [36]. The proteasome is responsible for a large number of protein turnovers, especially degradation of oxidized proteins and short life, which mediates the degradation of proteins. The alpha-20S subunits and beta-20S proteasome subunits are responsible for regulating proteasome activity and mediating the different proteolytic specificities of the proteasome respectively. There is evidence that proteasome inhibition contributes to increased oxidative damage, elevated intracellular levels of protein oxidation, and the induction of oxidative stress. Therefor psychiatric disorders and neurodegenerative diseases both had oxidative damage and neuronal loss[37–40]. Reactive oxygen species (ROS) cause nucleic acid breakage, lipid peroxidation, polysaccharide depolymerization, enzyme inactivation,and a host of other destructive processes [41, 42]. Furthermore, elevated ROS concentrations hinders mitochondrial activities, thus triggering the generation of Aβ [43, 44]. Yili Wu et al. used western blots and quantitative PCR to validate that PSMA5 and PSMB7 were downregulated via the over-expression of amyloid precursor proteins in HEK 293 cells [45]. Improving ubiquitin proteasome system activities recovers proteasome activities and ameliorates cell survival of HD-patient derived neurons, which is also a neurodegenerative disease[46]. These facts are consistent with our results of the 10 hub genes involved in the AD and SCZ.

In summary, we have provided information on the DEGs and hub genes involved in both AD and SCZ by using bioinformatics. AD patients have a more remarkable hippocampus gene expression than SCZ patients. Traditional boundaries between neurological and psychiatric disorders become blurred because of the common emerged pathways. However, the limitation is that expression of these genes in the different AD and SCZ subtypes were not involved in this analysis. Another limitation is that, we obtained the criteria of |FC|>1.2 in reducing the number of false negative DEGs in SCZ. And consequently, we may involve many insignificance genes in this study. While our work showed the usefulness and benefit of microarray analysis in extracting DEGs and hub genes for possible new targets of AD and SCZ. It is imperative to improve experimental analysis and prospective clinical studies. And the new insight of the underlying molecular mechanisms shared by AD and SCZ can guild subsequent experimental studies.