Vascular dementia (VaD) is a neurocognitive disorder that also has other characteristics such as behavioral symptoms, locomotor abnormalities, and autonomic dysfunction (O’Brien and Thomas, 2015). Various clinical and population studies have indicated that the prevalence of early-onset VaD (<65 years old) ranges from 3.1 to 44% (Vieira et al., 2013). VaD is considered to be the second most common form of dementia after Alzheimer’s disease (AD), posing a heavy burden on families and societies (Khan et al., 2016). However, no specific drugs have been approved for VaD treatment. Growing evidence indicates that stroke, hypertension, diabetes, and atherosclerosis are risk factors for VaD (Kalaria, 2018). A recent study found that the incidence of post-event dementia at 1 year was 34⋅4% in patients with severe stroke, 8⋅2% in patients with minor stroke, and 5⋅2% in patients with transient ischemic attack (Pendlebury and Rothwell, 2019). The underlying mechanism of VaD is still largely unknown but may be associated with pathological processes including chronic hypoperfusion and hypoxia (Tukacs et al., 2020), vascular endothelial dysfunction and damage (Wang et al., 2018), increased blood–brain barrier permeability (Ueno et al., 2019), and oxidative stress and inflammation (Guo et al., 2020; Wang et al., 2020). Therefore, there is a need to elucidate the detailed molecular mechanisms of VaD and explore potential drugs for VaD treatment.

Bioinformatics analysis of transcriptomic data is used widely to identify biomarkers, therapeutic targets, and the mechanisms of various diseases, including dementia (Ma et al., 2019). Key modules and hub genes related to AD have been identified (Rahman M. R. et al., 2020; Lee and Lee, 2021), yet VaD has received little attention. A vascular basis for neuronal atrophy in both the temporal and frontal lobes in VaD that is entirely independent of any Alzheimer’s pathology has been reported (Kalaria, 2016). Therefore, in this study, we analyzed transcriptome signatures of the temporal cortex and frontal cortex samples from a microarray dataset of VaD to identify overlapping differentially expressed genes (DEGs) between the two brain regions. Gene ontology and pathway enrichment analyses were conducted to predict the biological activities of the overlapping DEGs. Protein–protein interaction (PPI) analysis was performed to detect potential relationships among the proteins encoded by the overlapping DEGs. Hub genes and key modules were identified using Cytoscape software. We also performed hub gene–transcription factor (TF) and hub gene–microRNA (miRNA) interaction analyses to detect potential transcriptional and posttranscriptional regulatory factors, and hub gene–drug interaction analysis to identify candidate drugs. These inclusive systems biology processes allowed us to identify potential biomolecular signatures of VaD, which will give new insights into the underlying mechanisms and provide potential targets for therapeutic intervention.

A schematic diagram of the workflow of the bioinformatics analyses is shown in Figure 1.

Vascular dementia is a progressive disease that affects cognitive abilities, especially executive functioning. The heterogeneity of causes of VaD makes it challenging to elucidate the neuropathological substrates and mechanisms of VaD. In this study, we used multi-stage bioinformatics analyses to identify biomolecular signatures that underlie the pathophysiological mechanisms of VaD by analyzing the gene expression patterns in the temporal cortex and frontal cortex samples from patients with VaD.

A total of 159 overlapping DEGs between these brain regions were identified. We performed GO and pathway enrichment analysis to obtain further insights into the functions and signaling pathways of the overlapping DEGs. We found that these DEGs were enriched mainly in inflammation and innate immunity, synapse pruning, regeneration, positive regulation of angiogenesis, response to nutrient levels, and positive regulation of the digestive system process. Accumulating evidence indicates that inflammation and innate immunity play crucial roles in the progression of VaD. Synaptic pruning is essential for the development and maintenance of healthy brain circuitry by removing less active or “weak” synapses to allow the strengthening and maturation of more active connections in a neural activity-dependent process. Disruption of synaptic pruning has been associated with several neural disorders, such as schizophrenia and AD (Presumey et al., 2017). The synaptic pruning pathway was shown to be activated early in the brain of patients with AD where it mediates synapse loss (Hong et al., 2016), yet its role in VaD is largely unknown and needs future research. Increasing evidence has suggested that aberrant angiogenesis may play a role in the development of VaD, but its precise role remains controversial. A significant association between angiogenesis activity and cerebrovascular disease, which is a significant contributor to VaD, has been reported (Callahan et al., 2020), whereas other studies have indicated that promoting angiogenesis, especially in functional blood vessels, can reduce the extent of ischemia and improve cognition in VaD (Yang et al., 2017; Zhao et al., 2019). Clearly, more studies are needed to better understand the potential of angiogenesis as an intervention target.

We performed PPI analysis to discover the potential relationships among the proteins encoded by the overlapping DEGs and identified 10 hub genes (GNG13, CD163, C1QA, TLR2, SST, C1QB, ITGB2, CCR5, CRH, and TAC). Heterotrimeric G proteins, which consist of alpha, beta, and gamma subunits, function as signal transducers for the seven-transmembrane, G protein-coupled receptors. GNG13, which encodes a gamma subunit, is expressed in taste, retinal, and neuronal tissues, and the GNG13 protein is involved in the gustatory signal transduction pathway by interacting with proteins containing PDZ domains (Li et al., 2006; Liu et al., 2012). Liu et al. (2018) suggested that the GNG13 subunit was a critical signaling component in both the main olfactory epithelium and apical vomeronasal epithelium, by playing an important role in odor-triggered social behaviors including male–male aggression. A recent study found the expression level of GNG13 was significantly reduced in patients with AD compared with its expression in healthy control subjects and GNG13 may act as a potential biomarker in Purkinje cells, indicating the state of health of the cerebellum (Sanfilippo et al., 2021). However, insufficient information is available on its role in VaD. Future experiments are needed for verification.

The protein encoded by CD163 is a cell-surface glycoprotein in the scavenger receptor cysteine-rich superfamily, and it is expressed exclusively in monocytes and macrophages (Ritter et al., 1999). A recent study found that CD163+ macrophages promoted plaque angiogenesis, vascular permeability, inflammation, and progression of atherosclerosis, and deletion of CD163 in mice reduced intraplaque neovascularization and plaque progression (Guo et al., 2018). Besides, CD163+ macrophages and microglia were in the central and peripheral nervous system and may play a role in the inflammatory process (Fabriek et al., 2005; Galea et al., 2008). Atherosclerosis and inflammatory responses are known to be related to VaD (Kalaria, 2018), which is consistent with our results. The proteins encoded by C1QA and C1QB belong to the C1q family, whose members are the first components of the complement pathway and are involved in complement system regulation and play crucial roles in neurological disorders (Lee et al., 2019). A previous study indicated that aged C1qa-knockout mice showed reduced levels of cognitive and memory decline (Stephan et al., 2013), which is in accordance with the results of our study.

The protein encoded by TLR2 is a member of the toll-like receptor (TLR) family, whose members are significant pattern recognition receptors of the innate immune system, initiating inflammatory cascades by recognizing pathogen- and damaged-associated molecular patterns. TLR2 was found to play a pivotal role in inflammation after ischemic brain injury (Wang et al., 2011) and was involved in the development of diabetic microvascular complications, including endothelial dysfunction and cognitive impairment (Hardigan et al., 2017). In this study, we found that TLR2 was upregulated in the VaD samples compared with the controls samples, which supports the previous results.

Somatostatin, which is encoded by SST, is a widely distributed peptide in the central nervous system where it affects the rates of neurotransmission, and it was reported to be decreased in cerebrospinal fluid in patients with AD and VaD (Heilig et al., 1995). CCR5 encodes a seven-membrane, G protein-coupled receptor that may impact learning and memory by acting on CREB signaling, a pathway that is critical for learning and memory (Necula et al., 2021). CRH encodes a member of the corticotropin-releasing factor family that functions as a significant regulator of homeostasis, mediating the autonomic, behavioral, and neuroendocrine responses to stress (Dedic et al., 2019). Previous studies indicated that CRH might be involved in the regulation of cognitive performances (Bernardi et al., 2000).

Hub genes are considered to be key genes that play vital roles in biological processes and can affect the regulation of other genes in related pathways; thus, hub genes are often important targets and research hotspots. We studied hub gene–TF and hub gene–miRNA interactions to identify potential transcriptional and posttranscriptional regulators of the 10 identified hub genes. We identified four TFs (FOXC1, CREB1, GATA2, and HINFP) and four miRNAs (miR-27a-3p, miR-146a-5p, miR-335-5p, and miR-129-2-3p) as regulators of the hub genes in VaD. FOXC1 is involved in vascular development processes, including arterial specification and angiogenesis regulation, and may play a role in small vessel disease, which is the primary pathology underlying vascular cognitive impairment (Tan and Markus, 2015). Cyclic adenosine monophosphate responsive element-binding protein 1 (CREB1) is a leucine-zipper transcription factor that plays an essential role in long-term memory formation (Sadamoto et al., 2010). Silencing of CREB1 exasperated cognitive dysfunction in vascular dementia (VD) mouse model by inhibiting activation of the PKA-CREB signaling pathway (Han et al., 2018). GATA2, which encodes GATA-binding protein 2, was found to be differentially expressed in AD (Rahman M. R. et al., 2020), but its role in VaD has not been reported so far.

The reduced level of miR-27a-3p found in the cerebrospinal fluid of patients with AD indicated it as a candidate biomarker for AD (Sala et al., 2013). A recent study reported that reduction of miR-27a-3p in vascular smooth muscle cells may result in the development of vascular calcification that develops in association with atherosclerosis, one of the critical risk factors of VaD (Choe et al., 2020). miR-146a-5p was also reported to be associated with dysfunction of the vascular endothelium and may be involved in regulating high glucose-induced endothelial inflammation and atherosclerosis (Lo et al., 2017; La Sala et al., 2019). Capitano et al. (2017) found that overexpression of miR-335-5p impaired spatial memory and long-term potentiation maintenance in mice, indicating that miR-335-5p may play a critical role in spatial learning and synaptic plasticity. The blood level of miR-129-2-3p was significantly lower in ischemic stroke patients and negatively associated with the risk of ischemic stroke (Huang et al., 2019; Lo et al., 2017). These miRNAs can be considered as candidate biomarkers for VaD.

To the best of our knowledge, there are no definitive drugs available for VaD treatment. We performed a drug-hub gene interaction analysis to detect potential target drugs/compounds for VaD treatment and identified a total of 22 drugs that were predicted to have specific types of interactions with the hub genes. We checked the 22 candidate drugs in the ClinicalTrials.gov registry (see footnote 8). Although none of the 22 drugs have been used directly to treat VaD, four of them (maraviroc, cenicriviroc, PF-04634817, and efalizumab) could be potential drugs for VaD treatment. Maraviroc, cenicriviroc, and PF-04634817 are all CCR5 antagonists and efalizumab is an IGTB2 antagonist. Maraviroc and cenicriviroc were reported to improve cognition in HIV-infected individuals with cognitive impairment through their antiretroviral and anti-inflammatory effects (Mora-Peris et al., 2018; Alagaratnam et al., 2019). Moreover, maraviroc was found to reduce cardiovascular risk by modulation of atherosclerotic progression in vivo and in vitro (Afonso et al., 2017; Francisci et al., 2019). PF-04634817 and efalizumab are being developed for the treatment of diabetic nephropathy (Gale et al., 2018) and type 1 diabetes mellitus (Posselt et al., 2010), respectively. Diabetes mellitus and its complications are known risk factors for VaD. These four drugs need to be evaluated as potential drugs in VaD treatment in future studies.

Our study has some limitations. First, our results are based on publicly available microarray data, and no clinical or experimental confirmation of the roles of the proteins encoded by the identified genes of interest was attempted in the present study. Second, the expression profiles from only one GEO dataset were analyzed because of the limited number of available microarray datasets of VaD. In future studies, if available, multiple datasets should be analyzed simultaneously to increase the reliability of the results. Notwithstanding these limitations, this study offers some insights into the underlying mechanisms involved in the progression of VaD.

We identified overlapping DEGs between the frontal cortex and temporal cortex of VaD. These overlapping DEGs were enriched mainly in GO terms and pathways associated with inflammation and innate immunity, synapse pruning, regeneration, positive regulation of angiogenesis, and response to nutrient levels, all of which play crucial roles in the progression of VaD. We also identified hub genes, TFs, and miRNAs that were predicted to regulate the expression of the hub genes, as well as candidate drugs that target the hub genes. Our results may contribute to understanding the underlying mechanisms in VaD and provide potential targets and drugs for therapeutic intervention.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

JS, WW, and LZ conceived and designed the study, analyzed the data, wrote the manuscript, and contributed to the designs of the methods used in this study. All authors have read and agreed to the publication of this version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer YZ has declared a shared parent affiliation with the authors at the time of review.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.