A comparative analysis of Marburg virus-infected bat and human models from public high-throughput sequencing data

Abstract

Marburg virus (MARV) disease (MVD) is an uncommon yet serious viral hemorrhagic fever that impacts humans and non-human primates. In humans, infection by the MARV is marked by rapid onset, high transmissibility, and elevated mortality rates, presenting considerable obstacles to the development of vaccines and treatments. Bats, particularly Rousettus aegyptiacus, are suspected to be natural hosts of MARV. Previous research reported asymptomatic MARV infection in bats, in stark contrast to the severe responses observed in humans and other primates. Recent MARV outbreaks highlight significant public health concerns, underscoring the need for gene expression studies during MARV progression. To investigate this, we employed two models from the Gene Expression Omnibus, including kidney cells from Rousettus aegyptiacus and primary proximal tubular cells from Homo sapiens. These models were chosen to identify changes in gene expression profiles and to examine co-regulated genes and pathways involved in MARV disease progression. Our analysis of differentially expressed genes (DEGs) revealed that these genes are mainly associated with pathways related to the complement system, innate immune response via interferons (IFNs), Wnt/β-catenin signaling, and Hedgehog signaling, which played crucial roles in MARV infection across both models. Furthermore, we also identified several potential compounds that may be useful against MARV infection. These findings offer valuable insights into the mechanisms underlying MARV's pathophysiology and suggest potential strategies for preventing transmission, managing post-infection effects, and developing future vaccines.

Keywords: Homo sapiens; Marburg virus (MARV); Rousettus aegyptiacus; bioinformatics; zoonotic disease.

Figure 1

Graphical abstract of the study design. Data were obtained from the Marburg virus (MARV)-infected RoNi/7.1 cell line, derived from Rousettus aegyptiacus, and from MARV-infected primary proximal tubular cells from Homo sapiens in the Gene Expression Omnibus (GEO) database. By crossing fold change > 2.0 upregulated genes (MARV-infected groups vs. a mock-infected control groups) in each MARV-infected model using a Venn diagram analysis, common genes were projected to pathway analyses and functional interpretations using bioinformatics approaches.

Figure 2

Hierarchical clustered heatmaps of gene ontology (GO) enrichment. A GO enrichment analysis was performed on sets of differentially expressed genes (DEGs) within the two Marburg virus (MARV)-infected models compared to corresponding mock-infected groups, including (A) a MARV-infected bat-derived cell line at 24 h post-infection, and (B) a MARV-infected human-derived cell line at 20 h post-infection.

Figure 3

UMAP clustering of top enriched gene ontology (GO) terms represented for the top distinct diferentially expressed genes in the two Marburg virus (MARV)-infected models. The top 10 most enriched GO terms across biological process, molecular function, and cellular component categories in (A) a MARV-infected bat-derived cell line at 24 h post-infection, and (B) a MARV-infected human-derived cell line at 20 h post-infection.

Figure 4

Comparison of differentially expressed gene (DEG)-enriched signaling pathways in Marburg virus (MARV)-infected models. (A) The Venn diagram illustrates the number of distinct and common DEGs (with fold change of > 1.2) when comparing MARV-infected groups to their respective mock-infected control groups within the GSE117367 and GSE226148 datasets. The overlapping region represents genes shared between the two datasets, while the non-overlapping areas indicate genes unique to each dataset. (B) List of the top 15 DEG-enriched signaling pathways regulated by the common DEGs, generated by MetaCore, and sorted in descending order of log(p values). (C) Visualization of the global signal transduction pathway network generated by MetaCore confirmed that the “Immune response_IFN-alpha/beta signaling via JAK/STAT” pathway was highly enriched in both MARV-infected models.

Figure 5

Summary visualization of common and distinct biological process (BP) gene ontology (GO) terms and pathway analysis associated with the top differentially expressed genes (DEGs) in two Marburg virus (MARV)-infected in vitro models. The top enriched GO biological process terms in (A) the MARV-infected Rousettus aegyptiacus in vitro model, and (B) the MARV-infected Homo sapiens in vitro model. (C) Distinctive and common pathways regulated by the top DEGs in the two MARV-infected models at a cutoff log2 fold change of > 1.5 and a p value of < 0.05.

Figure 6

A key protein-protein-interacting (PPI) network and list of potential compounds from the connectivity map (CMAP)-based analysis. (A) The 90 differentially expressed genes (DEGs) shared between two MARV-infected groups, compared to the mock control group of both models, namely the Marburg virus (MARV)-infected Rousettus aegyptiacus (GSE117367) and MARV-infected Homo sapiens (GSE226148), were subjected to a functional protein association network analysis using the STRING database and the k-means clustering algorithm (n=10). (B) The CMap analysis presents a significant connectivity score when querying gene expression changes of 90 genes of interest against compounds from the LINCS L1000 database. Small-molecule compounds that caused similar gene expression signatures resulted in a positive correlation (red), while those causing opposing gene expression signatures resulted in a negative correlation (green). The top 50 potential compounds were sorted in descending order based on normalization of connectivity scores (norm_cs) at a cutoff false discovery rate (FDR)_q_nlog10 = 2.