Jens Rittscher: Biological imaging
Video microscopy aims to improve target discovery and drug development and to do so generates large volumes of data. Fluorescence labelling helps make intrinsic cellular functions visible and computational tools then enable analysis of these data sets to improve our understanding of cellular functions.
Q: What is high throughput screening?
Jens Rittscher: Unfortunately, the instrument that is used is a little bit too big to bring in to the office. Therefore I have brought a normal petri dish and this well plate to illustrate the principle. Normally people use these petri dishes on a single microscope. Today we use a robotic platform and this platform has, in this example, a 384-well plate in it. The images that we capture are very different from the images that you would capture with your phone or camera. You see now an image that is a bright-field image of cells and they have very little contrast. If we want to look inside the cells then we need to use fluorescent labels. In this example the fluorescent labels make the nuclei of the cells light up. Now that we have so many experiments on one plate we generate a large volume of data. This justifies why we need computational tools to analyse these data sets.
Q: Can you tell us about your research on biological imaging?
JR: The goal of my research in biological imaging is build computational tools that improve our understanding of cellular function. As cells divide they go through a process called the cell cycle, they move, and they die as well, which is important for cancer research. All of these processes are tightly regulated. We are at the stage where we can do large scale video microscopy. Building the tools that analyse the video feeds that come from high-throughput instruments is a significant challenge.
This is in one sense not much different from the work that I have done before where we analysed surveillance videos or news feeds: we looked at how people move, what actions they perform and tried to understand scenes. It is an area that is called computer vison. Now we are essentially translating these tools at the microscopic scale and extending them to extract biologically relevant information.
Q: How does this apply to phenotypic screening?
JR: When you walk along the beach and you find different types of seashells, a scientist would say that the different colours, shapes and sizes means that they are all different phenotypes. In some sense we now want to define the different phenotypes of cells. Even within one cell type, say a liver cell line, these cells will behave differently, they will have different genetic defects or they will react differently to their environment. By the measurements we are taking from these images, by looking at their function, we want to define these different phenotypes. Then in collaboration with experimental scientists we want to understand how the phenotype is either pre-programmed or how it can be affected.
Q: What are the most important lines of research that have developed in the past 5 or 10 years?
JR: First of all the whole idea of fluorescence imaging, of labelling interior parts of the cell, labelling proteins, and making very intrinsic cellular functions visible is one very important area. The second very important area is the progress that we have made in understanding natural images and also medical images, images that are taken by X-rays and ultrasounds. We are more or less translating and extending these techniques into microscopy imaging. Another very important line of research that affects a lot of us today is the work that has been done on machine learning: recognising patterns, understanding patterns and finding these patterns in very noisy data. These three things come together and enable us to do exciting things in biological imaging now.
Q: Why does your line of research matter, why should we put money into it?
JR: We need new tools to improve target discovery and drug development. A lot of drug development fails fairly late in the development process. Therefore it is very expensive. If we have better information upfront then we believe that we can improve our target discovery and drug development overall. This kind of large-scale biology is probably one tool that will help improve this process. Secondly, the use of more relevant model systems will also help us in the sense that we will also have cellular systems that provide us with better and more predictive data.
You can now see a video of what is called an organoid. It is a little model of the intestinal epithelium and it grows from a single cell within a few days. We can already monitor its growth process. Ultimately, we would like to look inside the cell cultures and make very detailed measurements on how it develops, what cell types are active and how the cells move around within that cell culture. Ultimately we believe that these kinds of models will improve biological discoveries.
Q: How does your research fit into Translational Medicine within the Department?
JR: The overall goal of the Target Discovery Institute is to improve the discovery of new targets and also to build better tools that help researchers within the University to translate basic biological insights into potential leads for drug development. The work I am talking about, biological imaging, is part of the computational platform that is necessary to facilitate this process. It is a building block of target discovery and therefore it fits in the department and also within the Ludwig Institute for Cancer Research.