How does SARS-CoV-2 hijack the human cell?

Short linear binding motif are used by virus to hijack the machinery of the human cell. We present a phage based approach to identify and validate those interactions in the context of SARS-CoV-2 and other coronaviruses. The information may potentially be used to find leads for inhibitor development.

Like Comment
Read the paper

How it started
During the spring of 2020 when the COVID-19 pandemic hit the world, we in the Ivarsson lab in the Department of Chemistry-BMC, Uppsala University Sweden were still allowed to go daily to the lab. Feeling privileged, we wanted to do what we could to gain insight into how the SARS-CoV-2 virus takes over the human cell, and turn it into a virus producing factory. We decided to use proteomic peptide phage display to screen on large-scale for interactions between SARS-CoV-2 proteins and human proteins. To do this, we first needed a novel phage library and a large collection of bait proteins.  

Making a novel phage peptidome

Our close collaborator Dr. Norman Davey (ICR, UK) designed a novel phage library that tiles the intrinsically disordered regions of SARS-CoV-2 proteins, as well as of other coronavirus proteins.  These intrinsically disordered regions are flexible regions that lack stable secondary structure. We focused on these regions as they often contain short linear binding motifs that can be bound by folded proteins. Being obligate intracellular parasites, many viruses have found ways to hijack the motif-based interactions of the host cell. Having decided on the library design, we obtained the oligonucleotide pool needed to generate a new phage library in no time.

Generating a collection of 139 bait proteins in no time

We also needed a large number of bait human proteins to screen against. We went through our freezers and stocks of expression constructs to find peptide-binding domains that we could produce for the screen. We were inspired by the Gordon et al. manuscript that was released in bioRxiv in March 2020 (1),  and scanned their data for host proteins that we could include in our screen. The final number of proteins landed on 139. Producing these proteins was a sizeable task, so we asked people in our lab, and in the neighbouring lab of Per Jemth to help out with producing and purifying proteins. Everyone was happy to contribute, so in short time we had a large collection of pure proteins that we could use for our screen.

Screening for host-pathogen interactions

Once the phage library was made and the proteins produced we ran the phage selections. This happened over a long weekend, where we would normally have celebrated the arrival of the spring. Instead, my group leader (Ylva Ivarsson) and I performed phage display selections. This included several hours in the cold room, plenty of pipette tips (which was a privilege to have as we had delivery problems of supply) and a massive number of 96-well plates (Figure 1).

Figure 1 On-going proteomic peptide phage display experiments conducted during a long weekend.

After the selections were done, we needed to analyze the results by next generation sequencing (NGS).  As it was a COVID-19 related project, the SNP & SEQ Technology Platform (part of the National Genomic Infrastructure Sweden) prioritised our sample and we had the results back in record-breaking time. Once the NGS data came back from the facility we extracted information of which peptides that were displayed on the surface of the binding enriched phages. We have over the last couple of years built up an efficient pipeline for the bioinformatic analysis. From our analysis pipeline, we get information on the enriched peptides, the binding motifs and which viral proteins the peptides come from. Using the information, we built a pan-viral network of which human proteins coronavirus proteins interact with (Figure 2a).  We also validated most of the found SARS-CoV-2 interactions through affinity measurements (Figure 2b), and pinpointed similarities and differences between different coronavirus strains.

Figure 2 a) Network visualisation of the proteomic peptide phage display results, showing which human bait proteins (blue circles) are hijacked by which coronavirus proteins (as indicated). b) FP results of G3BP1 interacting with peptides from N protein, red: SARS-CoV2, purple: SARS-CoV and yellow: MERS. (Figures adapted from Kruse et al., (2021) (2))

But, what we were really looking for were interactions that could be targeted by existing drugs, or that at least were promising targets for inhibition.  To do so, we needed to turn to virology. Luckily, we already had an established collaboration with the lab of Anna Överby (University of Umeå). The Överby lab was very enthusiastic to test different molecules that would supposedly block some of the interactions we found, including a green tea component that was supposed to inhibit the G3BP1/2 proteins. We got some very promising data on inhibiting G3BP1/2 using the commercially available compound, we thought, but then realized that the effect on viral release was basically from killing the cells, not from inhibiting the virus. Failing to find good compounds we set up a  genetic screening assay where the viral peptides we found were presented intracellularly fused to a green fluorescent protein. This assay confirmed that inhibition of G3BP1/2 reduced viral infection. The result was striking and could potentially be used as a starting point for peptide-based inhibitor in the fight against SARS-CoV-2.

But what could the biological relevance of the interaction between the viral N protein and the human G3BP1/2 proteins be? The interaction is also found for SARS-CoV N protein, but not for other coronaviruses. G3BP1/2 are involved in stress granules formation, large protein-RNA assemblies that are formed in response to various stresses conditions and viral infections. Why is it advantageous for SARS-CoV-2 that its N protein binds to G3BP1/2? Serendipitously the group of Jakob Nilsson at the University of Copenhagen, had found the same interaction through affinity purification coupled to mass spectrometry. We therefore joined forces to find out more about the interaction in a cellular setting. The Danish team brought in expertise in cell biological assays and mass spectrometry, which complemented the skillset of our growing consortium. They elegantly showed that the N protein outcompetes endogenous G3BP1/2 ligands, and that the N protein and G3BP1/2 co-localize in stress granules. Exactly why this is important for viral infection is yet to be understood.

Following this project from the start to the end during this special time was an interesting ride. It was only possible through the collaborations between several labs with complementary skills, that joined forces along the way. It included a lot of email, zoom calls and coordination between different groups which was an experience by itself.

We will continue to investigate the motif-based interactions that different viruses use to hijack host cells and hope over the years to come to contribute to a better understanding of how to counteract viral infections.

(1) D. E. Gordon et al., Science 10.1126/science.abe9403 (2020)

(2) T. Kruse et al.  Nat Commun 12, 6761 (2021) 

Caroline Benz

PhD student, Department of Chemistry - BMC, Uppsala University