Logo: to the web site of Uppsala University

Proteins are one of the most fundamental building blocks of life and their interactions regulate every cellular process. Historically they have been conceptualized as predominantly folded entities with well-defined secondary and tertiary structures. However, in recent decades, up to 50% of the human proteome has been shown to contain long disordered sequences that are flexible and unstructured in their natural environment. These intrinsically disordered regions exhibit low levels of sequence conservation and are enriched in short linear motifs (SLiMs). SLiMs are typically less then 10 amino acids long and serve as docking sites recognized by various globular domains. They exhibit a high degree of sequence degeneracy and evolutionary plasticity, allowing for rapid de novo emergence. SLiMs play crucial roles in a variety of cellular processes including cellular signalling, trafficking, transcriptional modulation, and protein degradation. Because they are small and degenerate, located in disordered regions, and form relatively weak interactions, they are difficult to identify using conventional high-throughput methods such as mass spectrometry and the yeast two-hybrid system. The same attributes that make them difficult to identify also make them ideal targets for viral SLiM mimicry, of which several examples have been described to date.

To address the elusive nature of SLiMs, we have developed a novel approach for the discovery of motif-mediated interactions at the proteome scale using proteomic peptide phage display. We constructed two separate phage libraries with either human or viral disordered regions displayed on their surface. These libraries were then subjected to phage display selections against over 300 globular domains, resulting in the identification of more than 1,700 potential novel interactions. We validated a subset of these interactions with affinity measurements and GST-pulldown assays, solved the crystal structure of human globular domains in complex with viral linear motifs, and demonstrated that the gained knowledge can be applied to design peptidomimetic inhibitors of viral replication. In addition, we demonstrated that direct binding of viral SLiMs to the N-terminal domain of clathrin disrupts cellular trafficking and identified the C-terminal domain of polyadenylate-binding protein 1 as a novel target for viral SLiM mimicry. Furthermore, we demonstrated that SARS-CoV-2 viral proteins possess both SLiMs that bind to human proteins, and globular domains that recognize human SLiMs, showcasing the versatility of SLiM-mediated interactions. Finally, we examined the evolutionary trajectory of the interaction between the SWIB domain of MDM2 and the SLiM of p53 TAD to describe an example of the extraordinary evolutionary plasticity of SLiM-mediated interactions.

Overall, the research presented in this thesis created the basis for an atlas of human motif-mediated interactions, yielded an extensive dataset of potential and validated cases of viral SLiM mimicry, and expanded our understanding of motif-mediated interactions from an evolutionary perspective.  

Proteins are responsible for countless processes in living creatures, but most often they do not perform these tasks alone. Rather, they engage in interactions with other proteins, creating whole protein-protein interaction (PPI) networks. Some of these interactions are formed between a folded protein domain and a short linear motif (SLiM), which is a small, 3-10 amino acid long stretch usually in the intrinsically disordered regions of proteins. These interactions tend to be low-to-medium affinity and transient, therefore their capture requires special tools. Furthermore, viruses often hijack the human cellular machinery through PPIs as they have limited genomes and are obligate cellular parasites. Therefore, the investigation of viral-host PPIs is of great importance and can lead to the development of novel antivirals.

In my thesis, I used mostly peptide-based and mass spectrometry (MS) techniques to validate and further explore motif-based PPIs. The main objectives were to: i) evaluate and compare synthetic peptide-based pulldown approaches, ii) validate and further explore the interaction between viral peptides and human polyadenylate-binding protein (PABP) using green fluorescent protein (GFP)-tagged peptide repeats, iii) confirm interactions, define and refine human interaction motifs that engage in interactions with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) proteins by employing peptide SPOT (synthetic peptide arrays on membrane support technique) arrays and alanine scanning, iv) investigate the change in the interactome of the nuclear pore complex protein 153 (NUP153) between uninfected and tick-borne encephalitis virus (TBEV)-infected states using GFP-tagged full-length protein for pulldown.

First, we explored the potential of affinity purification-mass spectrometry (AP-MS) and protein interaction screen on peptide matrix (PRISMA) to capture SLiM-based PPIs. The peptide pulldown approach proved to be more applicable over a wide range of affinities and interactions, however, protein concentration and the local concentration of presented motifs were limiting factors in certain cases. We then investigated SLiM-based interactions between RNA-viruses and human proteins. Here, using green fluorescent-peptide pulldowns I confirmed the interaction between viral peptides and the human poly-A binding protein. Next, we uncovered that some human SLiMs interact with SARS-CoV-2 proteins, and I was able to highlight the interaction motif using peptide arrays when only a handful of peptides were available. Lastly, I identified different enriched proteins in NUP153-pulldowns from mock-infected and TBEV-infected cell lysate, that were complementary to the changes observed with other techniques.

In conclusion, I explored a range of techniques that are valuable in the validation of PPIs, which is crucial in combination with high-throughput approaches. As more and more SLiM-based interactions are explored and predicted, the value of these tools continues to increase.

Cell signaling is mediated by signaling proteins that relay the signal in an intricate network of interactions before the signal is translated into a biological response. Short linear motifs (SLiMs) in intrinsically disordered regions of proteins serve as docking sites for protein interaction in all aspects of cell regulation including signal transduction. SLiM-mediated interactions are transient and low affinity and can be hijacked by virus. Only a small fraction of SLiMs have been described, but many more exist. Platelet derived growth factor receptor β (PDGFRβ) belongs to the family of receptor tyrosine kinases (RTKs) and controls cell growth, proliferation and migration. Dysregulation of PDGFRβ mediated signaling pathways is seen in many cancer types. To discover and characterize protein interactions, large scale high through-put methods are needed in concert with low through-put methods, that can characterize the interaction in a cellular context. The aim of this thesis has been to study protein-protein interactions in internalization and signaling of PDGFRβ and motif-mediated host-virus interactions through the use of in situ proximity ligation assay (PLA).  

Signaling via PDGFRβ is compartmentalized and depends on receptor internalization. In paper I we investigated the effects of dynamin inhibition for activation and signaling of PDGFRβ, and found that dynamin inhibition leads to impaired dimerization of the PDGFRβ. The results indicate that membrane localization of PDGFRβ is affected by dynamin. 

In paper II we developed a new method, Molboolean, for localized simultaneous detection of both free protein and protein in complex in cells. Molboolean is based on the principles from PLA with a fluorescent read out detectable with fluorescence microscopy.       

In paper III we mapped SLiM based host-virus interactions and explored their mechanisms. Using proteomic peptide phage display, we identified 1712 potential virus-host interactions by screening a library covering intrinsically disordered regions of the proteome for 229 RNA viruses. Clathrin mediated endocytosis was found to be a common target for viral hijacking, and viral binding of clathrin impaired PDGFRβ internalization.       

Some RTKs are proteolytically cleaved following ligand activation. In paper IV we characterized the Ca2+-dependent proteolytic cleavage of PDGFRβ. The cleavage resulted in two PDGFRβ fragments and was dependent on receptor internalization. Inhibition of the proteasome with bortezomib prevented internalization and cleavage and resulted in increased activation of PLCγ and STAT3. 

This thesis provides insight in the regulation of PDGFRβ signaling and internalization, and highlights contributions of both large-scale screenings and low through-put methods for studying protein-protein interactions.