ProLint: a web-based framework for the automated data analysis and visualization of lipid-protein interactions

Abstract

The functional activity of membrane proteins is carried out in a complex lipid environment. Increasingly, it is becoming clear that lipids are an important player in regulating or generally modulating their activity. A routinely used method to gain insight into this interplay between lipids and proteins are Molecular Dynamics (MD) simulations, since they allow us to study interactions at atomic or near-atomic detail as a function of time. A major bottleneck, however, is analyzing and visualizing lipid-protein interactions, which, in practice, is a time-demanding task. Here, we present ProLint (www.prolint.ca), a webserver that completely automates analysis of MD generated files and visualization of lipid-protein interactions. Analysis is modular allowing users to select their preferred method, and visualization is entirely interactive through custom built applications that enable a detailed qualitative and quantitative exploration of lipid-protein interactions. ProLint also includes a database of published MD results that have been processed through the ProLint workflow and can be visualized by anyone regardless of their level of experience with MD. The automated analysis, feature-rich visualization, database integration, and open-source distribution with an easy to install process, will allow ProLint to become a routine workflow in lipid-protein interaction studies.

Graphical Abstract

We present ProLint (www.prolint.ca), a webserver that completely automates analysis of MD generated files and visualization of lipid–protein interactions.

Figure 1.

The ProLint webserver. (A) The webserver along with the different tools released alongside it: a multi-image Docker build that includes the entire source code of ProLint and can be easily installed locally, a stand-alone python library that also includes the visualization applications, and g_surf. (B) The workflow implemented by ProLint: submissions are put into a task queueing system and once the required computer resources become available, they are preprocessed and analyzed through an automated protocol. Results are stored for 24 hours and can only be accessed through a unique ID provided to the user upon submission. The server is hosted on an AWS EC2 instance with an Elastic IP address and secure HTTPS protocol enabled using Certbot. Database management is done using PostgreSQL hosted on AWS RDS and files are stored using AWS S3. Task queueing is done using the Celery software. The backend is build using Django and Node.js, whereas the frontend uses HTML, CSS and JavaScript. Docker is used to provide an easy to install instruction set to run everything in a closed-system locally or host it on a private network.

Figure 2.

Cholesterol interactions with Smoothened. (A) Cholesterol interacts specifically with Smoothened by tightly binding between the helices 2 and 3 of the receptor. This is clearly observable from the distribution of contact points. Hovering over the data points reveals its labeling information. (B) Contact heatmap of the same data points but projected on the surface of the receptor, thus accurately localizing the binding site. Supplementary Figure S2 provides a more detailed overview of the contact heatmaps supported by ProLint.

Figure 3.

Cholesterol and PIP lipids interact through different mechanisms. (A, B) The interaction network of cholesterol and PIP lipids with the serotonin (5HT1B) receptor, respectively. (C) An expanded view of serotonin-PIP lipid interaction profile showing all arginine (ARG) and lysine (LYS) residues. The size of lipid nodes represents their total fraction of all lipids in the system, whereas residue node sizes are based on their relative fraction that make up the serotonin receptor. Edge width visualizes the average number of contacts with each lipid type. It is easy to focus over each node and to switch between different metrics.

Figure 4.

Integrating datasets and gaining novel insight. (A) The enrichment of different lipids around GPCRs. Showing how some lipids are highly enriched while other are depleted. (B) Sequence heatmap of GPCR–lipid interactions across all the GPCR data showing a likely conserved binding site at the extracellular site of helices 6 and 7.