C03: Deciphering virus-membrane interactions with advanced optical microscopy and machine
Projects of the CRC 1768
C03: Deciphering virus-membrane interactions with advanced optical microscopy and machine
A critical step in virus infection is the initial uptake of viruses into host cells. Interestingly, and perhaps counterintuitively, a particular virus may use different modes of crossing the plasma membrane, either in the same cell or depending on cell type. Even for members of the same virus family, different modes of uptake have been observed. The main uptake pathways include the use of membrane fusion or endocytosis. A number of endocytic mechanisms have been discovered and these depend on multiple factors, such as cell type and surface receptors, as well as cell membrane properties. Importantly, the uptake path usually determines the fate and infectivity of a virus. Thus, predicting the mode of entry is important for choosing and optimising preventative or impeding measures of infections.
However, determining an uptake mechanism is challenging due to multiple factors involving assay design and observation technology. An accurate investigation requires an adapted experimental design for each virus, receptor and host cell, analysing binding, uptake, routing, and the infection cycle of cellular and virus proteins. At the same time, direct microscopic imaging of the viruses and other molecules involved in the uptake into living cells should employ as high a temporal and spatial resolution as possible, since every entry mode displays specific spatiotemporal dynamics of virus and molecular movements at the cell and virus surface.
Sketch of different uptake mechanisms, ranging from various endocytosis pathways to membrane fusion to the injection of genetic material (as for bacteriophages).
Finally, data analysis needs to be tailored to correlate a “microscopic signature” of entry events with fusion signatures and particular endocytic mechanisms, which would then make it possible to identify the mode(s) of entry of an emerging virus. Unfortunately, the realisation of high-resolution microscope experiments of virus uptake is usually too complex to accomplish a straightforward and quick prediction of entry modes of newly emerging viruses. This is, however, within the overall aim of this CRC VirusREvolution, which is to set up tools to quickly react to potential pandemic threats. We therefore propose to realise a fast and facile tool to predict the mode(s) of virus entry from standard low-resolution live-cell microscopy data. To achieve this, we will combine high spatiotemporal
resolution microscopy data with methods from Machine Learning (ML) / Artificial Intelligence (AI) for fast, objective, and quantitative analysis to establish a tool that allows correlation with the standard microscopy data of lower spatiotemporal resolution. This shall enable the characterisation and differentiation of the membrane interactions of virtually anyvirus, usingwidelyusedstandardandlesscomplexmicroscopyapproaches,suchaswidefieldorconfocal microscopy. We will employ a machine learning approach that builds on the experience gained from working with a variety of viruses. Thus, the performance of our tool will gradually improve as more and more virus data becomes available. Our experimental and computational tools will contribute to answering the research questions of the CRC VirusREvolution, such as revealing mechanisms of virus entry and tackling the goals of the description and prediction of the infectivity of viruses and their hosts.
Project Overview
Central to our project plan is the realisation of a fast and facile tool to initially predict the mode(s) of virus entry from standard low-resolution live-cell microscopy data. Underlying this approach is the idea that virus entry modes can be differentiated by virus particle kinetics and signal intensity profiles prior to and during fusion or endocytic uptake, and can be linked to mechanistic information. These parameters depend on (1) receptor engagement and its mobility in the membrane, and (2) critically, the mode of crossing the plasma membrane. While (1) may vary, (2) is limited to distinct patterns. We propose to elucidate these patterns and link them to mechanistic information on virus entry. To this end, we will record correlative live-cell virus-entry data from low-resolution standard and high-resolution SRM fluorescence microscopy, and employ computational parameters to enable the extraction of in-depth information from standard techniques. To achieve this, we will combine standard and SRM microscopy data and employ machine learning methods for fast, objective, and quantitative analysis of entry modes of virtually any virus with standard microscopy data only, Fig. C03.4. We will utilise a machine learning approach that will gradually improve as more and more virus data becomes available. Overall, we will realise (a) the direct visualisation of virus membrane interactions, (b) tracking of virus entry events, and (c) prediction of the mode of entry for viruses. The aim is that the improved understanding of the virus entry mechanism will open new therapeutic developments. We combine expertise in employing standard microscopy to virus entry analysis (Schelhaas lab), development of SRM and its application towards membrane characterisation and virus cycling (Eggeling lab), and unique knowledge of data analysis, including ML algorithms to disclose various cell-biological events, including pathogen-cell interactions (Figge lab).
- Tool to be developed: Optical microscopy-based tools to provide fast and facile predictions of the mode of entry of emerging pathogens.
- Tool to be developed: Optical microscopy protocols for simultaneous recording and correlation of standard and high resolution data of virus-membrane interactions.
- Tool to be developed: Machine learning-supported workflow in JIPipe for fast, objective, and quantitative image analysis and classification of virus-cell interactions.
Hypothesis enabled by the proposed tool: The prediction of the mode of virus uptake is important for choosing preventive measures of infections by emerging pathogens, and adaptation of optical microscopy and data analysis will provide the required tools.
Overarching CRC goals: Our project C03 couples high-spatiotemporal single-virus imaging with a machine-learning pipeline that infers entry modalities from standard widefield/confocal time-lapse data, enabling rapid, scalable annotation of membrane-interaction signatures across diverse viruses (G1, G3). Trained on growing consortium datasets, the classifier provides calibrated predictions of fusion vs. endocytic uptake routes for emerging pathogens, supporting early triage of infectivity-modifying interventions (G4).
Work Packages (WP):
- WP 1: Optimised visualisation of virus-membrane interactions (Eggeling/Schelhaas)
- WP 2: Fast prediction of virus entry modes through correlation of techniques (Eggeling/Figge/Schelhaas)
- WP 3: Machine-Learning-supported image analysis of virus-membrane interaction (Figge/Eggeling/SchelhaaS)
Team Members
Dr. Ziliang Zhao
Postdoc
Dr. Ruman Gerst
Postdoc
Ivan Avilov
Postdoc
Dr. Dhruv Khatri
Postdoc
2025
The actin nucleation promoting factor WASH facilitates clathrin-independent endocytosis of human papillomaviruses Journal Article
In: EMBO reports, vol. 26, no. 22, pp. 5533, 2025.
2024
Effects and avoidance of photoconversion-induced artifacts in confocal and STED microscopy Journal Article
In: Nature Methods, vol. 21, no. 7, pp. 1171–1174, 2024.
2023
Master mitotic kinases regulate viral genome delivery during papillomavirus cell entry Journal Article
In: Nature Communications, vol. 14, no. 1, pp. 355, 2023.
Quantitative analysis of peroxisome tracks using a Hidden Markov Model Journal Article
In: Scientific reports, vol. 13, no. 1, pp. 19694, 2023.
JIPipe: visual batch processing for ImageJ Journal Article
In: nature methods, vol. 20, no. 2, pp. 168–169, 2023.
2021
A Ran-binding protein facilitates nuclear import of human papillomavirus type 16 Journal Article
In: PLoS pathogens, vol. 17, no. 5, pp. e1009580, 2021.
2019
Molecular recognition of the native HIV-1 MPER revealed by STED microscopy of single virions Journal Article
In: Nature communications, vol. 10, no. 1, pp. 78, 2019.
2018
Super-resolution fluorescence microscopy studies of human immunodeficiency virus Journal Article
In: Retrovirology, vol. 15, no. 1, pp. 41, 2018.
2013
Automated characterization and parameter-free classification of cell tracks based on local migration behavior Journal Article
In: PloS one, vol. 8, no. 12, pp. e80808, 2013.
2012
Analysis of virus entry and cellular membrane dynamics by single particle tracking Journal Article
In: Methods in enzymology, vol. 506, pp. 63–80, 2012.