RNA-seq transcriptome analysis of the anti-inflammatory effect of artesunate on cerebral malaria

Re-Analysis

Published on April 25th, 2023
by Axel Martinelli, PhD

⏱ 11 min read

Figure 8. GSEA plots indicate a negative correlation with inflammation-related genesets and a positive correlation with genesets linked to dendritic activity when comparing ATN-treated samples against infected (ECM) samples.

Introduction to the Cerebral Malaria Study

Cerebral malaria is the main cause of death in patients infected with Plasmodium falciparum. Artemisinin and its derivatives, such as artesunate (ATN), are the first-line treatment prescribed for severe malaria and display immunomodulatory properties (Mancuso et al, 2021). 

In order to study the effects of artesunate in the brains of Plasmodium-infected hosts, Wang and colleagues have analysed the transcriptome in the brain of mice infected with P. berghei, a murine malaria species that also induces cerebral inflammation, and subsequently treated with ATN (Wang et al, 2022).

As expected, they observed the differential expression of various cytokines (in particular Tnf-α and Il-1β) upon infection and then treatment with ATN. Gene Set Enrichment Analysis (GSEA) indicated the dysregulation of pathways related to the JAK-STAT signaling pathway, apoptosis, and Toll-like receptor signaling pathway.

As the raw transcriptomic data is currently available on the GEO database (https://www.ncbi.nlm.nih.gov/geo/) under the ID GSE162535, I decided to take a look at it using the Omics Playground Platform and, where possible, expand on the original findings by the authors.

The platform was run by intersecting the default choices of methods for differential gene expression analysis (limma, EdgeR and DEseq2), with cutoffs set at p<0.05 and logFC>0.5. For GSEA, the camera, fgsea and gsva methods were intersected with cut-offs set at p<0.05 and logFC>0.2.

Clustering Analysis

The data shows a strong clustering based on infection and treatment, with the infected samples (“ECM” and “Artesunate”) forming a distinct cluster compared to control samples and being further separated by treatment on a UMAP plot (Fig. 1A).

Functional annotation of the clustered heatmap (Fig. 1B) highlights, among others, the upregulation (cluster S3) of various inflammatory pathways, apoptosis and the IL6-JAK-STAT3 pathway, based on the Hallmark gene set collection (Fig. 2A), and of the Toll-like receptor signaling pathways, based on the GO Terms collection (Fig 2B), following infection with P. berghei.

This replicates the main findings of the publication. 

Figure 1. (A) UMAP plot of the samples separated by groups (Control, ECM=experimental Cerebral Malaria, Artesunate=Artesunate-treated infected samples). (B) Clustered heatmap of the samples labeled by treatment groups. Cluster S3 highlighted in green.
Figure 1. (A) UMAP plot of the samples separated by groups (Control, ECM=experimental Cerebral Malaria, Artesunate=Artesunate-treated infected samples). (B) Clustered heatmap of the samples labeled by treatment groups. Cluster S3 highlighted in green.
Figure 2. (A) Hallmark collection-based functional annotation of cluster S3 indicates the upregulation of inflammation-related gene sets, apoptosis and the IL6-JAK-STAT3 pathway in both infected sample pre- and post-treatment. (B) GOBP term annotation further highlights the upregulation in infected samples of the Toll-like receptor signaling pathway.
Figure 2. (A) Hallmark collection-based functional annotation of cluster S3 indicates the upregulation of inflammation-related gene sets, apoptosis and the IL6-JAK-STAT3 pathway in both infected sample pre- and post-treatment. (B) GOBP term annotation further highlights the upregulation in infected samples of the Toll-like receptor signaling pathway.

Experimental Similarity

A useful information that can be gained from the platform regarding the nature of an experiment is obtained by comparing a selected expression signature against a collection of datasets collected from the GEO public database.

In this case, the gene expression signature generated by malaria infection against uninfected samples indicated a strong correlation with an experimental infection with P. berghei in mice (Fig. 3), which was part of a larger study on the pathogenesis of neuroinflammation (Torre et al, 2017). 

Other pairwise comparisons with ATN-treated samples were still dominated by the P. berghei-induced signature, indicating the same dataset as the most correlated one (data not shown).

Figure 3. The plot displays a strong positive correlation between the gene expression profiles (expressed as Fold Changes, FC) generated by the “ECM vs Control” comparison and the comparison between the brains of P. berghei infected mice against control mice from a previous study (GSE79702).
Figure 3. The plot displays a strong positive correlation between the gene expression profiles (expressed as Fold Changes, FC) generated by the “ECM vs Control” comparison and the comparison between the brains of P. berghei infected mice against control mice from a previous study (GSE79702).

Differential Gene Expression Analysis

Differential gene expression analysis highlights, as expected,  the upregulation of several cytokines and other immune-related genes following infection (Fig. 4). Additionally there was also a marked downregulation of three hemoglobin genes.

Figure 4. Most dysregulated genes when comparing ECM vs control samples. The plots highlight the upregulation of various immune-related genes (e.g. Cxcl10, Gzmb, Iigp1) and the downregulation of two hemoglobin genes (Hbb-bs and Hba-a1) upon infection.
Figure 4. Most dysregulated genes when comparing ECM vs control samples. The plots highlight the upregulation of various immune-related genes (e.g. Cxcl10, Gzmb, Iigp1) and the downregulation of two hemoglobin genes (Hbb-bs and Hba-a1) upon infection.

Treatment of infected cells with ATN resulted in the dampening of immune responses, as highlighted by the downregulation of several immunity-related genes, and the recovery of hemoglobin gene expression (Fig. 5).

Figure 5. Most dysregulated genes when comparing ATN-treated vs ECM samples. The data shows the downregulation of immune related genes (e.g. Il1r2, Ccl2, Ctla2a) and recovery of hemoglobin gene expression (Hbb-bs and Hba-a1).
Figure 5. Most dysregulated genes when comparing ATN-treated vs ECM samples. The data shows the downregulation of immune related genes (e.g. Il1r2, Ccl2, Ctla2a) and recovery of hemoglobin gene expression (Hbb-bs and Hba-a1).

Gene Set Enrichment Analysis

GSEA was performed against all the collections available on the platform. Consistently with the gene expression analysis, infection resulted in the upregulation of various gene sets associated with inflammatory responses, particularly interferon activation (Fig. 6).

Apoptosis, JAK-STAT signaling and Toll-like receptor signaling pathways were also upregulated. This was accompanied by a loss of neuronal activity, as highlighted by the downregulation of glutamate receptor activity (Fig. 7).

Treatment with ATN effectively reversed these trends, dampening immune responses and restoring neuronal activity (Fig. 8).

Figure 6. Top most positively correlated GSEA plots based on the 50,000+ gene sets contained in the platform for the ECM vs. control pairwise comparison. The majority of the upregulated gene sets are related to inflammatory responses (e.g. interferon activation).
Figure 7. GSEA plots indicate the upregulation of apoptosis, JAK-STAT signaling and Toll-like receptor signaling pathways upon infection. Conversely, the negative correlation with several gene sets associated with neuronal function, as exemplified here by the downregulation of glutamate receptor activity upon infection, was also observed.
Figure 7. GSEA plots indicate the upregulation of apoptosis, JAK-STAT signaling and Toll-like receptor signaling pathways upon infection. Conversely, the negative correlation with several gene sets associated with neuronal function, as exemplified here by the downregulation of glutamate receptor activity upon infection, was also observed.
Figure 8. GSEA plots indicate a negative correlation with inflammation-related genesets and a positive correlation with genesets linked to dendritic activity when comparing ATN-treated samples against infected (ECM) samples.
Figure 8. GSEA plots indicate a negative correlation with inflammation-related genesets and a positive correlation with genesets linked to dendritic activity when comparing ATN-treated samples against infected (ECM) samples.

GSEA: L1000 Database

The platform offers access to the L1000 drug connectivity map, a collection of more than a million expression profiles of 1000 genes from cell lines exposed to various drug concentrations, as well as individual gene silencing and overexpression experiments.

Individual gene manipulation comparison indicated that infection was positively correlated with the overexpression of various cytokines (Fig. 9). 

Figure 9. ECM vs control differentially expressed genes compared to the LINCS L1000 single gene alteration profiles. The results show a strong positive correlation with the overexpression (indicated by the “lig” and “oe” suffixes) of various cytokines, consistent with the inflammatory responses generated by ECM.
Figure 9. ECM vs control differentially expressed genes compared to the LINCS L1000 single gene alteration profiles. The results show a strong positive correlation with the overexpression (indicated by the “lig” and “oe” suffixes) of various cytokines, consistent with the inflammatory responses generated by ECM.

When comparing the infection signature against those generated by various drugs, a strong positive correlation with topoisomerase inhibitors was observed (Fig. 10). As these result in cell apoptosis, it could reflect the pro-apoptotic properties of parasite hemozoin against neuronal cells (Eugenin et al, 2019) or a consequence of the expression of NF-κB induced by pro-inflammatory cytokines in the brain (Punsawad et al, 2013).

Furthermore, the platform indicated that dopamine receptor antagonists could be used to reverse the effects induced by ECM (Fig. 10). Since aberrant dopamine expression plays a critical role in cerebral malaria (Kumar and Babu, 2020), this could represent a plausible novel approach for treating cerebral malaria.

Figure 10. THE ECM gene expression profile was compared against the drug gene expression profiles from the LINCS L1000 database and the most positively and negatively correlated hits were summarised at the mechanisms of actions (MOAs) level. The platform suggests dopamine receptor antagonists (highlighted in red) as the most prominent MOA to counteract (negative correlation) the effects of ECM on the brain and topoisomerase inhibitors (highlighted in blue) as the most positively correlated MOA.
Figure 10. THE ECM gene expression profile was compared against the drug gene expression profiles from the LINCS L1000 database and the most positively and negatively correlated hits were summarised at the mechanisms of actions (MOAs) level. The platform suggests dopamine receptor antagonists (highlighted in red) as the most prominent MOA to counteract (negative correlation) the effects of ECM on the brain and topoisomerase inhibitors (highlighted in blue) as the most positively correlated MOA.

Comparing ATN-treated infected samples vs the control samples highlighted a weak positive correlation with the silencing of gene GABARAPL1 (Fig. 11). This gene is involved in autophagosome maturation (Chakrama et al, 2010) and since artemisinin and its derivatives have been known to induce apoptosis in cancer cells (Kiani et al, 2020), this was a counter-intuitive correlation.

However, a publication from 2018 suggested that artemisinin prevents neuronal cell apoptosis by activating the AKT pathway (Lin et al, 2018). Since activation of the AKT pathway inhibits expression of GABARAPL1 (Su et al, 2017), this could explain the correlation detected in the dataset.

Nonetheless, there is no indication of a downregulation in the expression of GABARAPL1 following ATN treatment, nor any negative correlation of ATN with known AKT inhibitors, which could have been used to imply an activating role (data not shown). It must be stressed that the available data is not suited to address this question, due to the presence of malaria parasites in ATN-treated samples.

Figure 11. The GSEA-like plot shows the positive correlation of the ATN-treated vs. ECM comparison gene expression profile with those of various ranked datasets (black bars) where GABARAPL1 was silenced. The positive correlation is not statistically significant at q>0.05
Figure 11. The GSEA-like plot shows the positive correlation of the ATN-treated vs. ECM comparison gene expression profile with those of various ranked datasets (black bars) where GABARAPL1 was silenced. The positive correlation is not statistically significant at q>0.05.

Conclusion

Through the platform, the main findings of the original publication could be replicated. Thus I observed the upregulation of inflammatory responses and a loss of neuronal function in infected samples and a reversal of these trends upon treatment with ATN.

Thanks to access to the drug connectivity map database, further insights could also be gained from the dataset.

ATN treatment was positively correlated with gene expression profiles associated with topoisomerase inhibitors, known to induce apoptosis, one of the main modes of action of ATN. Meanwhile, the aberrant dopamine expression associated with cerebral malaria indicated the therapeutic potential of dopamine receptor antagonists, which has hitherto remained unexplored in this context.

Easily analyze your RNA-Seq and proteomics dataset with Omics Playground!

Analyze your dataset

About the Author

Axel Martinelli

Axel Martinelli’s academic background is in molecular biology and parasitology. He earned a Ph.D. on the genetics of strain-specific immunity against malaria infections and a master’s degree in bioinformatics with specialization in the analysis of omics data. During his postdoctoral career, he worked on genomics and transcriptomics studies and is currently the head of biology at Bigomics Analytics.

References

    • Mancuso RI, Azambuja JH, Olalla Saad ST. Artesunate strongly modulates myeloid and regulatory T cells to prevent LPS-induced systemic inflammation. Biomed Pharmacother. 2021 Nov;143:112211. doi: 10.1016/j.biopha.2021.112211. Epub 2021 Oct 5. PMID: 34649344.
    • Wang Q, Tang Y, Pan Z, Yuan Y, Zou Y, Zhang H, Guo X, Guo W, Huang X, Wu Z, Li C, Xu Q, Song J, Deng C. RNA-seq-based transcriptome analysis of the anti-inflammatory effect of artesunate in the early treatment of the mouse cerebral malaria model. Mol Omics. 2022 Sep 26;18(8):716-730. doi: 10.1039/d1mo00491c. PMID: 35960011.
    • Torre S, Polyak MJ, Langlais D, Fodil N, Kennedy JM, Radovanovic I, Berghout J, Leiva-Torres GA, Krawczyk CM, Ilangumaran S, Mossman K, Liang C, Knobeloch KP, Healy LM, Antel J, Arbour N, Prat A, Majewski J, Lathrop M, Vidal SM, Gros P. USP15 regulates type I interferon response and is required for pathogenesis of neuroinflammation. Nat Immunol. 2017 Jan;18(1):54-63. doi: 10.1038/ni.3581.
    • Kumar SP, Babu PP. Aberrant Dopamine Receptor Signaling Plays Critical Role in the Impairment of Striatal Neurons in Experimental Cerebral Malaria. Mol Neurobiol. 2020 Dec;57(12):5069-5083. doi: 10.1007/s12035-020-02076-0. Epub 2020 Aug 24. PMID: 32833186.
    • Eugenin EA, Martiney JA, Berman JW. The malaria toxin hemozoin induces apoptosis in human neurons and astrocytes: Potential role in the pathogenesis of cerebral malaria. Brain Res. 2019 Oct 1;1720:146317. doi: 10.1016/j.brainres.2019.146317. Epub 2019 Jul 2. PMID: 31276637; PMCID: PMC6702100.
    • Punsawad C, Maneerat Y, Chaisri U, Nantavisai K, Viriyavejakul P. Nuclear factor kappa B modulates apoptosis in the brain endothelial cells and intravascular leukocytes of fatal cerebral malaria. Malar J. 2013 Jul 26;12:260. doi: 10.1186/1475-2875-12-260. PMID: 23890318; PMCID: PMC3728032.
    • Lin SP, Li W, Winters A, Liu R, Yang SH. Artemisinin Prevents Glutamate-Induced Neuronal Cell Death Via Akt Pathway Activation. Front Cell Neurosci. 2018 Apr 20;12:108. doi: 10.3389/fncel.2018.00108. PMID: 29731711; PMCID: PMC5919952.
    • Chakrama FZ, Seguin-Py S, Le Grand JN, Fraichard A, Delage-Mourroux R, Despouy G, Perez V, Jouvenot M, Boyer-Guittaut M. GABARAPL1 (GEC1) associates with autophagic vesicles. Autophagy. 2010 May;6(4):495-505. doi: 10.4161/auto.6.4.11819. Epub 2010 May 16. PMID: 20404487.
    • Kiani BH, Kayani WK, Khayam AU, Dilshad E, Ismail H, Mirza B. Artemisinin and its derivatives: a promising cancer therapy. Mol Biol Rep. 2020 Aug;47(8):6321-6336. doi: 10.1007/s11033-020-05669-z. Epub 2020 Jul 24. PMID: 32710388.