The aim of the project to detect SLiMs in Leishmania mediating host-pathogen PPIs. For this task we first need to identify virulence factors (proteins that invade the host), so then we can explore their details and detect tthose motifs that bind to the host proteins. While SMILE is focused on Leishmania, the developed computational pipeline can later be adapted for other eukaryotic pathogens as well.
We soon realized that Leishmania is an extremely complicated organism for computational biology, as data quality from high throughput experiments were lower quality and we also found a lot of sequence errors. We realized that we don't have reliable data about secreted and surface proteins that helps invading the host, so we started manual annotation.
At the same time AlphaFold2 just came out and it revolutionized computational structural biology. We also started to use AlphaFold2 and realized that it can help us to identify surface proteins.
Traditional methods to predict the structure of transmembrane proteins were complex and prone to errors. The introduction of AlphaFold2 (AF2) has been a tremendous help to the scientific community. It complements the limited experimental data by providing 3D predictions for thousands of new alpha-helical membrane proteins. However, AF2 faces challenges due to the lack of reliable structural templates and not being specifically trained to handle certain structural boundaries. To address this, we developed a new database called Transmembrane AlphaFold database (TmAlphaFold database) and used a simple geometry-based method to visualize the most likely position of the membrane plane.
Having TmAlphaFold helped us to expedite manual annotation to our dataset. Beyond manual annotation we included a lot of experimental and computational to develop a database of Leishmania proteins.
LeishMANIAdb is a database specifically created to study how Leishmania's virulence factors might affect host proteins. Since we don't have complete information about the secretomes of various Leishmania species, we gathered experimental evidence and made computational predictions to identify secreted proteins. This resulted in a user-friendly website where people can access all the available information about these experimental and predicted secretomes. Additionally, we manually analyzed interactions between 211 proteins involved in the host-pathogen relationship, as well as the characteristics and functions of 3764 transmembrane (TM) proteins from different Leishmania species. To further understand how the infection works, we added automatic predictions about the structures and functions of these proteins, providing new insights into the molecular mechanisms of the disease.
Developing LeishMANIAdb enabled us to look into more details about Leishmania proteins. Although by now we had a smaller dataset of surface proteins, we still only have limited information about Leishmania secreted proteins. Therefore we first looked for protein trafficking machineries to have a better overall picture of what proteins get secreted.
We studied three ancient protein targeting systems and their receptors to see how they compare to other living organisms, including their hosts. These systems involve important components of cellular life, like secretory signal peptides, endoplasmic reticulum (ER) retention motifs (KDEL motifs), and autophagy signals (motifs interacting with ATG8 family members). While these systems were expected to be similar across different organisms, we found that they actually vary to some extent from what we see in animals, plants, or fungi. We not only describe how they work but also create predictive models that can tell us where these proteins are located or what functions they serve in Leishmania species. This is the first time such predictions have been made for these proteins in Leishmania.
Finally, using all these results we identified potential proteins that help the entry to the host cells. We selected the most promising one and performed experiments to validate the protein-protein interaction. Although we couldn't collect enough reliable data from the experiments to confirm the presence of the identified linear motif, further experiments may help us to verify and characterize the interaction.
All results are available on preprint servers and online. The results will be published in Q1 quality journals. Early results were presented as a conference talk and a poster presentation.
