Network medicine links SARS-CoV-2/COVID-19 infection to brain microvascular injury and neuroinflammation in dementia-like cognitive impairment
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (ScreenIT)
Abstract
Background
Dementia-like cognitive impairment is an increasingly reported complication of SARS-CoV-2 infection. However, the underlying mechanisms responsible for this complication remain unclear. A better understanding of causative processes by which COVID-19 may lead to cognitive impairment is essential for developing preventive and therapeutic interventions.
Methods
In this study, we conducted a network-based, multimodal omics comparison of COVID-19 and neurologic complications. We constructed the SARS-CoV-2 virus-host interactome from protein-protein interaction assay and CRISPR-Cas9-based genetic assay results and compared network-based relationships therein with those of known neurological manifestations using network proximity measures. We also investigated the transcriptomic profiles (including single-cell/nuclei RNA-sequencing) of Alzheimer’s disease (AD) marker genes from patients infected with COVID-19, as well as the prevalence of SARS-CoV-2 entry factors in the brains of AD patients not infected with SARS-CoV-2.
Results
We found significant network-based relationships between COVID-19 and neuroinflammation and brain microvascular injury pathways and processes which are implicated in AD. We also detected aberrant expression of AD biomarkers in the cerebrospinal fluid and blood of patients with COVID-19. While transcriptomic analyses showed relatively low expression of SARS-CoV-2 entry factors in human brain, neuroinflammatory changes were pronounced. In addition, single-nucleus transcriptomic analyses showed that expression of SARS-CoV-2 host factors ( BSG and FURIN ) and antiviral defense genes ( LY6E , IFITM2 , IFITM3 , and IFNAR1 ) was elevated in brain endothelial cells of AD patients and healthy controls relative to neurons and other cell types, suggesting a possible role for brain microvascular injury in COVID-19-mediated cognitive impairment. Overall, individuals with the AD risk allele APOE E4/E4 displayed reduced expression of antiviral defense genes compared to APOE E3/E3 individuals.
Conclusion
Our results suggest significant mechanistic overlap between AD and COVID-19, centered on neuroinflammation and microvascular injury. These results help improve our understanding of COVID-19-associated neurological manifestations and provide guidance for future development of preventive or treatment interventions, although causal relationship and mechanistic pathways between COVID-19 and AD need future investigations.
Article activity feed
-
-
SciScore for 10.1101/2021.03.15.435423: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
Experimental Models: Cell Lines Sentences Resources Specifically, six of these datasets were based on CRISPR-Cas9 assay results, including (1-2) CRISPR_A549-H and CRISPR_A549-L, based on high (-H) and low (-L) multiplicity of infection of SARS-CoV-2 in A549 cells [21]; (3-5) CRISPR_HuH7-SARS2, CRISPR_HuH7-229E, CRISPR_HuH7-OC43, based on HuH7 cells infected by SARS-CoV-2, HCoV-229E, and HCoV-OC43, respectively [22]; and (6) CRISPR_VeroE6, based on SARS-CoV-2-infected VeroE6 cells [23]. A549suggested: NoneHuH7suggested: NoneVeroE6suggested: JCRB Cat# JCRB1819, RRID:CVCL_YQ49)Software and Algorithms Sentences Resources Functional … SciScore for 10.1101/2021.03.15.435423: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
NIH rigor criteria are not applicable to paper type.Table 2: Resources
Experimental Models: Cell Lines Sentences Resources Specifically, six of these datasets were based on CRISPR-Cas9 assay results, including (1-2) CRISPR_A549-H and CRISPR_A549-L, based on high (-H) and low (-L) multiplicity of infection of SARS-CoV-2 in A549 cells [21]; (3-5) CRISPR_HuH7-SARS2, CRISPR_HuH7-229E, CRISPR_HuH7-OC43, based on HuH7 cells infected by SARS-CoV-2, HCoV-229E, and HCoV-OC43, respectively [22]; and (6) CRISPR_VeroE6, based on SARS-CoV-2-infected VeroE6 cells [23]. A549suggested: NoneHuH7suggested: NoneVeroE6suggested: JCRB Cat# JCRB1819, RRID:CVCL_YQ49)Software and Algorithms Sentences Resources Functional enrichment analyses, including Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) biological process enrichment analyses, were performed using Enrichr [30] for the CRISPR datasets. KEGGsuggested: (KEGG, RRID:SCR_012773)Enrichrsuggested: (Enrichr, RRID:SCR_001575)Neurological disease gene profiles: We extracted neurologic disease-associated genes/proteins from the Human Gene Mutation Database (HGMD) [31], and defined a gene to be disease-associated, if it had at least one disease-associated mutation from HGMD reported in the literature. Human Gene Mutation Databasesuggested: (Human Gene Mutation Database, RRID:SCR_001621)All single-cell analyses were performed using Seurat v3.1.5 [39] following the processing steps from the original publication of each dataset. Seuratsuggested: (SEURAT, RRID:SCR_007322)For the bulk RNA-sequencing dataset, differential expression analysis was performed using edgeR v3.12 [40] edgeRsuggested: (edgeR, RRID:SCR_012802)Each PPI edge has one or more source information of five categories of evidence from publicly available databases and datasets: protein complexes identified by robust affinity purification-mass spectrometry from BioPlex V2.016 [49]; binary PPIs discovered by high-throughput yeast two-hybrid systems in three datasets [24, 50, 51]; signaling networks revealed by low-throughput experiments from SignaLink2.0 [52]; low-throughput or high-throughput experiments uncovered kinase-substrate interactions from KinomeNetworkX [53], Human Protein Resource Database (HPRD) [54], PhosphoNetworks [55], PhosphositePlus [56], DbPTM 3.0 [57], and Phospho. BioPlexsuggested: (BioPlex, RRID:SCR_016144)HPRDsuggested: NoneDbPTMsuggested: (dbPTM: An informational repository of proteins and post-translational modifications, RRID:SCR_007619)ELM [58]; and PPIs curated from literatures identified by yeast two-hybrid studies, affinity purification-mass spectrometry, low-throughput experiments, or protein three-dimensional structures from BioGRID [59] BioGRIDsuggested: (BioGrid Australia, RRID:SCR_006334)PINA [60], Instruct [61], MINT [62], IntAct [63], and InnateDB [64]. MINTsuggested: (MINT, RRID:SCR_001523)IntActsuggested: (IntAct, RRID:SCR_006944)InnateDBsuggested: (InnateDB, RRID:SCR_006714)Eigenvector centrality [66] of the nodes in the networks were computed using Gephi 0.9.2 [67] to evaluate the influence of the nodes considering the importance of their neighbors. Gephisuggested: (Gephi, RRID:SCR_004293)Statistical analysis and network visualization: Python package SciPy v1.3.0 [69] was used for the statistical tests unless specified otherwise. Pythonsuggested: (IPython, RRID:SCR_001658)SciPysuggested: (SciPy, RRID:SCR_008058)Networks were visualized with Gephi 0.9.2 [67] and Cytoscape 3.8.0 [70]. Cytoscapesuggested: (Cytoscape, RRID:SCR_003032)The transcriptomic datasets used in this study (GSE147528, GSE157827, GSE138852, GSE157103, GSE149689, and GSE163005) were downloaded from the NCBI GEO database (https://www.ncbi.nlm.nih.gov/geo/). NCBI GEOsuggested: Nonehttps://www.ncbi.nlm.nih.gov/geo/suggested: (Gene Expression Omnibus (GEO, RRID:SCR_005012)Results from OddPub: Thank you for sharing your data.
Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:Limitations: We acknowledge several limitations. First, our human protein-protein interactome was built using high-quality data from multiple sources; yet it is still incomplete. The PPIs in our interactome is undirected. However, it has been shown that incorporating directionality of the human PPI does not change network proximity results [114]. Therefore, the network associations could be either positive or negative, and require further investigation. In addition, as our network proximity analysis relies on disease-associated genes, literature bias could affect the results because more highly-studied genes are more likely to appear in the dataset. Second, we analyzed expression levels of the key SARS-CoV-2 entry factors and found low expression levels for ACE2 and TMPRSS2. However, we cannot rule out the possibility of SARS-CoV-2 directly targeting the brain via as-yet unidentified mechanisms. Third, possible pathways of neuroinflammation and microvascular injury were tested using data of either individuals with AD or COVID-19, but not both. Future studies using genetics and multi-omics data from individuals with both AD and COVID-19 will be needed to confirm and extend these network-based findings. The significance of our findings in the context of the general population of COVID-19 frequently suffering from “brain fog” without a formal diagnosis of AD needs further investigation.
Results from TrialIdentifier: We found the following clinical trial numbers in your paper:
Identifier Status Title NCT04320277 Not yet recruiting Baricitinib in Symptomatic Patients Infected by COVID-19: an… NCT04321993 Recruiting Treatment of Moderate to Severe Coronavirus Disease (COVID-1… Results from Barzooka: We did not find any issues relating to the usage of bar graphs.
Results from JetFighter: We did not find any issues relating to colormaps.
Results from rtransparent:- Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
- Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
- No protocol registration statement was detected.
-