Chairman of Molecular Cytology and Director of the Center for Advanced Microscopy
Swammerdam Institute for Life Sciences, University of Amsterdam
At A Glance
Name: Theodorus Gadella
Position: Chairman of Molecular Cytology and Director of the Center for Advanced Microscopy, Swammerdam Institute for Life Sciences, University of Amsterdam
Background: Assistant professor of molecular biology, Wageningen University and Research Center, The Netherlands, 1999-2001; Postdoc, Wageningen University, 1994-1999; Postdoc, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany, 1991-1994; PhD, biochemistry, Utrecht University, 1991.
As director and founder of the Center for Advanced Microscopy at the University of Amsterdam, Theodorus Gadella has worked with a variety of advanced fluorescence microscopy techniques to further his research on cell biology — in particular, cell signaling pathways. Gadella is the lead author of a review paper published in the August issue of ChemBioChem in which he discusses what multi-parameter imaging can bring to cell biology, some of its limitations, and what basic building blocks are needed before it can be fully exploited in a drug discovery context. Gadella shared some of these views last week in an interview with CBA News.
Do you study cellular imaging in a basic research context, a drug discovery context, or both?
The main purpose of our research is fundamental research, not so much drug discovery. Of course, drug discovery is a logical application of our work, but we need to do more groundwork to make this feasible and to fully exploit the high-content aspects of this approach.
Tell me a little bit more about your specific research?
Within the division we have several research areas, and they are all related to cell biological work, particularly the imaging of molecules in living cells with microscopes. The research most relevant to the publication in ChemBioChem is our intracellular signaling research. For that we have developed many fluorescent probes based on GFP technology. We transfect these into cells and visualize several signaling events in living cells. The idea is also to apply more advanced fluorescence microscopy techniques to be able to extract more detailed information. So the lab has a strong tradition in advanced microscopy, and high-resolution imaging has a strong base at the University of Amsterdam.
The overall aim is to fully understand the mechanism of a signaling cascade. Of course, we can see certain pathways where A activates B, and B activates C, but how is it logistically managed within a cell, and can we understand that? When we look at the single-cell level, even simple reactions seem difficult, and they open a whole new world.
The concept of multi-parameter imaging has become popular in drug discovery as high-content screening, but in many ways, it's still only one or two parameters measured in high-throughput. Do you think that true multi-parameter imaging as you describe it has a place in drug discovery?
We need to somehow create, as I call it, a way of literally visualizing how signaling cascades work. Usually they're drawn in schemes with arrows between the components, but in the cells, this is different. It looks like, for example, that oscillations may be important; or motors moving from A to B — so localization in time; and also the conformation of a molecule at that position is important. There are many receptors, especially in GPCR-based signaling — and of course, this is a really big target for drug discovery — but very few of the underlying signaling cascades are used. There seem to be much more cellular outputs and inputs than there are signaling pathways connecting them. So that is a basic feature that we want to understand. How can very many processes, for instance, be regulated by calcium, with a different input and a different output? I think the answer is to understand where the calcium is produced in the cell, and what are its dynamics? This also holds true for other molecules. Where is a specific molecule located? So you can use the same molecule, but if you activate it at different locations in a cell, it will have a different response. So it is not just that we have to understand multi-parameter imaging. The way a typical high-throughput screen is done is that you get plus/minus-based readouts. We need to understand not only if a signal is turned on, but where it is turned on, and in what subtle ways is it turned on? This is what I mean when I say we have to understand signaling cascade components logistics. This really needs a very concentrated single-cell analysis to be done first. Then, of course, you can discriminate between different responses and fine-tune high-content and high-throughput screening even further to score for more subtle features. Especially in multi-parameter imaging, you can use several components at a certain location, and even combine two or three signals in a certain spatial and temporal manner. This is difficult to do, though, and a lot of ground work needs to be done. There are many technical challenges, but in the end, we may fully understand what's going on, and it will be not so much like gold digging, but more like an intelligent design.
I'm sure many people in pharma agree with that, but the pressures of time and money in drug discovery have recently precluded such an approach. Is it feasible for pharma to adopt even more of a basic research approach again?
I see a lot of these massively parallel assays and high-throughput screening approaches in drug discovery to be a bit, as I said, like gold digging. You have some way of visualizing something, and then you start to screen and hope you find the gold. This is very appealing and very easy to manage in a business. You have huge amounts of equipment and specialized people, and the question is relatively simple: Does some compound out of some number of compounds activate something? The easy and quick drugs will at some point be there, but the more subtle activities or perhaps even complete new drugs could be designed if we understand the logistics. So I think pharma has to move that way, but it will be more difficult to manage — I'm not only talking about dollars, but also the questions become more complex. Not, 'Which one of these compounds will give us a plus/minus?', but, 'Which ones will give us a certain spatial and temporal reaction that we know turns on certain genes?' It's not a matter of whether pharma can afford it, it's just an evolution, and it will move in that direction. I admit we are at the beginning, and of course, the visualization of molecular events in cells is still young. GFP has not been around for that long. And now we can see these multiple colors, and use multiple fluorescence resonance energy transfer [FRET] dyes. One of the real icons in the field, Roger Tsien, predicted that perhaps for even every biochemical process in a cell, there may be an indicator or biosensor to study the event. We can do things now that we could not dream about a few years ago.
What are some of the major limitations of multi-parameter imaging right now?
There are many limitations, but also many possibilities. The limitations are that we want even more colors that are further extending the red; we want genetically encoded probes, which is why GFP is so popular; and that it takes quite some time to build a good biosensor, and to really characterize it and show that it does not affect the cell function. Once this is done — and it is done for many sensors right now — then it will be there, because it's genetically encoded, you have the cDNA of it, and you don't need to do it again. So as time progresses, more and more specific sensors will be available. We also need smaller probes — GFP is very large, and of course very useful, but it may influence some processes. But with smaller probes, it may prove very difficult to introduce them genetically. In terms of microscopes, there are at some point fundamental limitations on how many colors you can visualize simultaneously, and so usually there is a trade-off between spectral channels and speed, or resolution. You have to choose a compromise somewhere, and for fast signaling cascades, this is problematic. There is still great potential for all of this, it is worth all this effort to make these biosensors and to do this multi-parameter imaging. So I would emphasize more the possibilities than the limitations.
In the paper, you mentioned new hybrid technology that is 'currently only partially available' that might be needed to investigate parameters that can't be monitored with fluorescent probes. What does that refer to?
The main thing there is secretion, or ion flux. That's an application that my colleague Carsten Schulz was thinking about. He studies a signaling cascade that in the end leads to chloride secretion. The best way to monitor that is with patch-clamp techniques. But all the intermediate steps are visible with fluorescent probes, and it's very fast. So you would need to design a microscope that can monitor the multiple signaling parameters, but you would also need a patch clamp on the microscope. This would require some adaptation of advanced microscopes.
Some of the advanced fluorescent techniques that you talk about like FRET and fluorescence lifetime imaging [FLIM] — can those eventually be useful in the context of pharmaceutical research?
Well, I am very fond of FRET. Signaling is basically done in three ways: molecules meet each other, molecules can be cleaved, and they can change conformation. These are the basic options, and all three of these can be detected with FRET, and at a resolution that is below, say, 5 nanometers, which is way below the resolution of an optical microscope. So with FRET, you can bridge molecular resolution and cellular resolution in a living cell. The FRET contrast is not great, so we need some advanced techniques to extract the information quantitatively, and that's where the FLIM and all the other techniques come in. There is a huge revival of FRET since the introduction of the multi-color variations of GFP. This will be extended, soon, into multiple FRET pairs, and that will prove to be very interesting. Most high-content screening using FRET is only done with one FRET pair. We may be able to eventually monitor a molecular interaction, a conformation change, and cleavage, with three different probes. But again, there would need to be a lot of development of probes and advanced instrumentation.