The Potential Role of SSX2 and Trametinib in MERS and SARS

Published on February 25, 2020
⏱ 7 min read

Sars Cov2 Virus Model

Introduction

In my previous post on Finding Potential Biomarkers And Therapeutic Drugs With Coronavirus Datasets, I analysed an in vitro infection with a Middle Eastern coronavirus using our special edition of the Omics Playground platform. In my analysis I identified the SSX2 gene as a potential marker of early infection and trametinib as a potential antiviral drug.

In order to extend such findings to other coronavirus species, I looked at publicly available MERS and SARS datasets stored in our platform. Starting with MERS, I chose two datasets. 

The first dataset is from a more recent published study on the role of accessory ORFs in MERS infections. My main reason for picking this study was that Calu-3 cells were used for in vitro infections, the same type of cells used in the samples I analysed in the previous part of this post and that similar time points were also chosen. 

The second dataset is from an unpublished study made public in 2016. It consists of 50 samples of fibroblasts infected with wild type MERS viruses collected at regular intervals between 0h and 48h post-infection and the corresponding control samples. This allows me to verify if my observations apply to later time points and a different type of host cells.

Potential role of SSX2 as an early infection biomarker

Because I already had a potential biomarker, I checked straight away the expression levels of the SSX2 gene in both MERS datasets. Encouragingly, I observed a high expression of the gene in infected cells (as early as 7h post-infection) but very low expression in uninfected cells in both datasets (Figure 1A and B).

Log2 CPM expression in control and infected samples for gene SSX2
Figure 1. Log2 CPM expression in control and infected samples for gene SSX2. A) Expression in individual infected and control Calu-3 cell samples. B) Expression in infected and control fibroblasts samples, combined by infection status and time point.

To confirm this, I also run the “biomarkers analysis” module on the Calu-3 cells dataset to confirm that SSX2 was indeed distinguishing infected and non-infected cells (Figure 2). These results confirm the potential role of SSX2 as an early infection biomarker in both MERS and MERS-like coronaviruses.

Decision tree based on infected and control Calu-3 cell samples. SSX2 expression levels distinguish infected (icMERS) from non-infected (MOCK) samples.
Figure 2. Decision tree based on infected and control Calu-3 cell samples. SSX2 expression levels distinguish infected (icMERS) from non-infected (MOCK) samples.

Trametinib as a potential antiviral drug

With the “Drug connectivity map” module I could verify if the trametinib gene set was still negatively correlated with MERS infection profiles. In the infected Calu-3 cells the correlation was near statistical significance 24h post-infection (Figure 3A). A statistically significant correlation in the infected fibroblasts dataset that spanned multiple time points was also evident (Figure 3B and C). As was the case with the MERS-like infections in my previous post, inhibitors targeting various kinases were heavily represented among the most negatively correlated profiles.

Drug connectivity map results for the MERS datasets. A) GSEA profile for trametinib against the 24h infected Calu-3 cells sample comparison. B) GSEA profile for trametinib against the 24h infected fibroblast sample comparison. C) Drug activation matrix for the MERS-infected fibroblast study. Trametinib is highlighted in yellow and shows significant inhibitory potential against the 24h, 36h and 48h post-infection comparisons.
Figure 3. Drug connectivity map results for the MERS datasets. A) GSEA profile for trametinib against the 24h infected Calu-3 cells sample comparison. B) GSEA profile for trametinib against the 24h infected fibroblast sample comparison. C) Drug activation matrix for the MERS-infected fibroblast study. Trametinib is highlighted in yellow and shows significant inhibitory potential against the 24h, 36h and 48h post-infection comparisons.

Next, I selected an unpublished SARS dataset for further analysis. This dataset is rather comprehensive, consisting of a time series from 0h to 72h post-infection of in vitro infections with a human SARS virus strain and a mutated SARS-like bat coronavirus in Calu-3 cells.

When looking at SSX2 expression across sample groups, I noticed little expression in both the control group and the bat SARS-like virus infected group across all time points. On the other hand, there was stronger SSX2 expression in the SARS-infected samples (Figure 4). However, increased gene expression started later than in MERS and MERS-like viral infections, with no discernible increase before 30h post-infection, and was not as pronounced (Figure 5).

Further analysis with a SARS dataset

Next, I selected an unpublished SARS dataset for further analysis. This dataset is rather comprehensive, consisting of a time series from 0h to 72h post-infection of in vitro infections with a human SARS virus strain and a mutated SARS-like bat coronavirus in Calu-3 cells.

When looking at SSX2 expression across sample groups, I noticed little expression in both the control group and the bat SARS-like virus infected group across all time points. On the other hand, there was stronger SSX2 expression in the SARS-infected samples (Figure 4). However, increased gene expression started later than in MERS and MERS-like viral infections, with no discernible increase before 30h post-infection, and was not as pronounced (Figure 5).

Overall SSX2 gene expression (in Log2 CPM) by treatment group across all time points in control, SARS-infected and mutant SARS-like virus infected Calu-3 cells.
Figure 4. Overall SSX2 gene expression (in Log2 CPM) by treatment group across all time points in control, SARS-infected and mutant SARS-like virus infected Calu-3 cells.
SSX2 gene expression (in Log2 CPM) by time point for the SARS-infected samples, showing a significant increase in expression from time point 30h.
Figure 5. SSX2 gene expression (in Log2 CPM) by time point for the SARS-infected samples, showing a significant increase in expression from time point 30h.

In order to identify potential antiviral drugs, I selected the 72h post-infection contrast between SARS infected and control samples in the Enrichment tab to look for drugs with negatively correlated expression profiles. Trametinib was among the top 20 hits (Figure 6).

GSEA plot showing a statistically significant negative correlation between the 72h post SARS infection gene expression signature and the trametinib signature.
Figure 6. GSEA plot showing a statistically significant negative correlation between the 72h post SARS infection gene expression signature and the trametinib signature.

Furthermore, a statistically significant negative correlation was present in all the time points pairwise comparisons after 24 h (namely 30, 36, 48, 54, 60 and 72h) post-infection (Figure 7). Interestingly, there was little evidence of inhibitory potential against the mutated bat coronavirus.

Activation Matrix of enriched drug signatures against all the available time point contrasts for SARS and mutated bat SARS-like virus. Trametinib is highlighted in yellow and displays a significant negative correlation (blue dots) with the post-24h SARS infected samples.
Figure 7. Activation Matrix of enriched drug signatures against all the available time point contrasts for SARS and mutated bat SARS-like virus. Trametinib is highlighted in yellow and displays a significant negative correlation (blue dots) with the post-24h SARS infected samples.

Additionally, the mode of actions of the most negatively correlated drug profile against SARS mirrored the observations in my previous post, with various kinase inhibitors (such as MEK, CDK and mTOR inhibitors) prominently represented (Figure 8).

Most enriched mode of actions of the drugs displaying positively and negatively correlated signatures with the 72h SARS infection phenotype.
Figure 8. Most enriched mode of actions of the drugs displaying positively and negatively correlated signatures with the 72h SARS infection phenotype.

Conclusion

First of all, we can conclude that trametinib is a potential broad spectrum inhibitor of coronavirus-induced gene expression profiles in infected cells. The negative relationship was statistically significant in all but one of the datasets (where it was very near the p<0.05 threshold) I analysed in this and my previous post and was independent on type of cells (fibroblasts or Calu-3 lung cancer cells). This was consistent with in vitro studies showing the efficacy of trametinib in inhibiting MERS infections. Additionally, various kinase inhibitors showed inhibitory profiles against MERS/MERS-like and SARS infections alike, indicating that kinase inhibition may be an approach to curtail coronavirus infections in general.

The second conclusion is that SSX2 gene expression appears to be a good predictor of early infection (roughly 7h post-infection) in MERS and MERS-like coronaviruses. SSX2 is also upregulated during SARS coronavirus infections. However, the upregulation starts later in the infection (after 24h post-infection) and is generally weaker than is the case of MERS and MERS-like viruses. Thus, for SARS there may be better potential biomarkers, but SSX2 could still provide some signal, at least for later infection stages.

Are these conclusions applicable to the current COVID-2019 epidemic?

There are no publicly available datasets I could use to answer this question. But I hope that researchers may make use of BigOmics Analytics’ platform to address important questions such as the identification of biomarkers and potential antiviral drugs for COVID-2019. 

Scientist with lens exploring the drug connectivity analysis tab in Omics Playground

Get the most out of your omics data! Identify potential biomarkers and therapeutic drugs with Omics Playground.

Get started now