Discovery of a novel mechanism unique to severe COVID leads to a potential treatment


In a large meta-analysis, circulating protein levels of severe COVID patients along with many thousands of genomic datasets leads to an exciting discovery: a mutation in a gene called FAS that turns it into a non-membrane-bound, soluble form, interrupting normal immune function. In addition, an on-market chemotherapeutic drug has been shown to lower soluble FAS, leading to new clinical trials for severe COVID.

A few weeks ago you read about the depth of the UK BioBank which I called a ‘veritable fountain of discovery’ along with five interesting results and hints of what’s to come in the future. Combining many thousands of individuals where both genomic variation data along with circulating protein data will lead to new understandings of the mechanisms of disease, new protein targets to assist in developing new medications, and new markers of disease to develop better and more accurate diagnostics.

And then not stopping there a week later I wrote up a brief history of the Human Genome Project and its subsequent milestones, from the HapMap Project to the Thousand Genomes Project to the Exome Aggregation Consortium (ExAC) then finally the Genome Aggregation Database (GnomAD). As a result of this immense set of genetic variation data across literally hundreds of thousands of individuals are vast numbers of individual genotypes shared across worldwide populations, whose precise location in the genome are known.

What is not known for the vast majority of these polymorphisms is how they affect our health and disease; also not only as risk markers for disease but also in what specific role the gene encoding protein plays in disease. Protein Quantitative Trait Loci, or pQTLs, are the variants in the genome that affect levels of a particular protein in the bloodstream. Though massive computational power, a statistical “protein-level as phenotype” phenome-wide association study can tie the particular SNP to the quantitative trait of a protein concentration level. And to top this off, Mendelian Randomization, by making a few assumptions, can provide evidence of causality for a particular disease.

Naturally the more proteins you can measure, the better, and naturally the more genotypes you have (i.e. individuals who have their exomes or genomes sequenced) the better.

One other thing about pQTLs to mention here is that they come in two forms: a cis-pQTL where the variant is close or within the coding region of the protein in question, and a trans-pQTL where the variant is several hundred kilobases away (or as is primarily the case on a completely different chromosome). These trans-pQTLs are of great interest, as they reveal previously unknown biological interaction of other proteins involved with the expression level of the protein in question.

The current challenge of COVID-19

As someone who has written a book about the molecular biology of COVID, you do not need me to describe what a toll this worldwide pandemic has had. An official death toll as of this writing in Spring of 2022 there are over 6.2M deaths attributed to COVID worldwide, and severe COVID (where external ventilation is needed followed by organ failure) is characterized by exaggerated inflammatory responses and immunopathology. It is critical to understand these immune responses and mechanisms that underlie a wide range of clinical presentations of COVID-19. From this understanding, rational therapeutic strategies can then be attempted; indeed there have been over 2,000 drug clinical trials to tackle COVID-19, with only one antiviral and a few monoclonal antibody treatments approved to-date.

In 2020 an international group of genetic scientists founded the COVID-19 Host Genetics Initiative, whose aim is to collect samples from COVID-19-infected patients and use the tools of genomic science to determine risk alleles, uncover new drug targets, and trace the evolution of this disease. Its fifth data release had no less than 5,582 samples of severe-COVID individuals and their genomic sequence.

Combining proteomics and genomics with Mendelian Randomization

Now the COVID-19 Host Genetics Initiative is an enormous resource, with those 5500 samples of individuals with severe COVID (many of whom have passed away from the disease). This dataset, absent circulating proteome data, has been combined with two smaller studies from the SCALLOP Consortium in a technique called 2-sample Mendelian Randomization (MR).

In the prior two posts (this one about the UK BioBank and this one about the power of combining proteomics with genomics, in case you haven’t seen them) I introduce the concept of MR and how it is used to combine SNP variation data across many individuals along with their circulating protein data to determine proteins that are causal for disease, if certain assumptions are met. Note the word ‘causal’ – it is not an association with disease or a probability of risk: MR determines the causal role of a particular protein for a disease condition.

This post is about a very interesting pre-print, titled “Mendelian randomisation identifies alternative splicing of the FAS death receptor as a mediator of severe COVID-19” (Klaric L and Peters JE et al. Mendelian randomisation identifies alternative splicing of the FAS death receptor as a mediator of severe COVID-19. medRxiv. (2021) 2021.04.01.21254789. doi:10.1101/2021.04.01.21254789)

The SCALLOP Consortium is a group of over a dozen institutions and research groups who have made their sample collections publicly available, with the combination of genetic data, circulating proteome data as measured by Olink technology, and extensive phenotype / disease data on their sample collection. Two papers were published as part of the SCALLOP Consortium: the first, Gisby et al. measured 436 proteins from several Olink Target 96 protein panels across 55 severe COVID-infected individuals who also had end-stage kidney disease (and at high risk for severe COVID), and made this dataset available through the consortium. A second paper, Filbin et al., measured 1472 proteins from an Olink Explore 1536 panel across 306 individuals with severe COVID.

In the preprint Klaric et al. they bring these datasets all together and come up with surprising results. This figure illustrates the intersection of the three datasets: two from the aforementioned SCALLOP papers with 140 protein measurements, and the Host Genetics Initiative genomic data.

Portion of Figure 1 from Klaric et al.

After two-sample MR, there were five proteins identified: EPHB4, FAS, LGAL9, TNFRSF10A, and CCL2. A host of sensitivity tests were then applied: tests for heterogeneity (consistency of effects across variants), tests for horizontal pleiotropy, check for common causal variants (called colocalization), and a test for cis-pQTLs (comparing this cis- signal against the trans-pQTL signal strength).

Alternative splicing of the FAS death receptor is a mediator of severe COVID-19

One protein passed all tests, and this protein the FAS death receptor plays a role in apoptosis, which is programmed cell death. FAS is normally membrane-bound, and when this receptor docks with the FAS ligand it sets off a self-destruct mechanism. For specific cellular immune response effector cells, they need to be ‘trimmed’ by these apoptotic signals as they perform their ‘normal duties’ as immune cells.

The primary SNP identified in the cis-pQTL analysis in the FAS gene is called rs982764. Looking up this SNP, it has a prevalence of 31% across worldwide populations. Interestingly, the MAF in European populations is also 31%, while African populations it is lower at 27% and in Asian populations higher at 50%.

Portion of Figure 3 from Klaric et al.

The authors of the preprint Klaric et al. went a step further and looked for the expression of this FAS gene in expression QTL databases GTEx (in tissues) and  eQTLGen (in whole blood) and the expression QTL signal is not there.  

This SNP is not a coding SNP, from this work the scientists determined it actually codes for an intron before exon 6 that causes that exon to be skipped over in its messenger RNA. This was confirmed in RNA-Seq analysis across 4,778 individuals. And exon 6 codes for the transmembrane domain of this death receptor, making this receptor a soluble one, thus acting as a ‘decoy’ in soaking up the FAS-ligand signal, interrupting the apoptotic signal from being effectively communicated to the target cells.

Repurposing an on-market cancer drug ibrutinib to treat severe COVID

It has been observed that a Bruton’s Tryosine Kinase Inhibitor (BTKi) called ibrutinib reduces soluble FAS receptor in in-vitro experiments. Ibrutinib is currently approved to treat two lymphocytic leukemias, two lymphomas, and a rare cancer. Thus its safety profile is known, and Phase II trials have already been initiated to use this cancer drug against cases of severe COVID-19. These can be found online at the NIH Clinical Trials database under NCT04375397 and NCT04439006.

The power of genomics with proteomics

This is a remarkable example of the power of high-throughput sequencing to generate genomic data; and then combining it with high-multiplex immunoassays (in the several hundreds to over fourteen hundred); and then using Mendelian Randomization to determine causality. These results are dependent upon both the genomic and proteomic data being available for the disease, and the statistical technique to determine a particular protein’s role in disease. RNA transcripts were studied to confirm the findings, however these results were not generated from analysis of RNA data.

And another key component of this study by Klaric et al. is the public availability of these datasets and the collaboration between different research groups.

The fact there is an on-market therapy directly affecting the protein of interest is a wonderful result with practical and direct application: instead of waiting years for drugs to be developed against this target, an on-market existing drug is repurposed and a clinical trial assembled going directly to Phase II.

The genomic revolution is just beginning, and proteomics is a vital part of it.


About Dale Yuzuki

A sales and marketing professional in the life sciences research-tools area, Dale currently is employed by Olink as their Americas Field Marketing Director. https://olink.com For additional biographical information, please see my LinkedIn profile here: http://www.linkedin.com/in/daleyuzuki and also find me on Twitter @DaleYuzuki.

Leave a comment

Your email address will not be published. Required fields are marked *