Name: Lucia Sironi
Position: Postdoc, Jan Ellenberg group, Gene Expression Unit, European Molecular Biology Laboratory, Heidelberg, Germany
Earlier this year, a research team led by Jan Ellenberg at the European Molecular Biology Laboratory published a notable paper in Nature Methods describing the use of automated time-lapse imaging technology to conduct more detailed high-throughput RNAi screening than can be accomplished with typical endpoint imaging experiments (see CBA News, 4/21/2006).
Various members of that team are now using the method to devise so-called secondary functional genomics screens to follow up on primary RNAi-based chemical genomics screens of mitotic genes recently conducted by the MitoCheck consortium, in which Ellenberg’s EMBL group is a participant.
Last week, at the American Society for Cell Biology meeting in San Diego, Lucia Sironi, a postdoc in the Ellenberg group, presented a poster describing the automated quantification of microtubule dynamics in living cells using similar time-lapse microscopy techniques and RNAi knockdown. Specifically, Sironi and colleagues are measuring microtubule growth rates in HeLa cells expressing a fluorescently tagged tracking protein, confocal microscopy, and particle-tracking algorithms – their own version of high-content imaging.
Sironi took a few moments at the conference to discuss her work with CBA News.
Why did you develop this assay?
This work follows the primary screening of the MitoCheck group, which identified a series of genes involved in mitosis. The idea now is to develop a series of secondary assays to better characterize the mitotic phenotypes we’ve seen. In the [Nature Methods] paper, the first screen consisted of basically total time lapse microscopy for two days, just looking at the chromosomes, because the cells are expressing 2bGFP. We could very easily identify a mitotic phenotype, because you could look at misalignment, problems in segregation, or problems in nuclear envelope breakdown. In addition, you could see the chromosome not condensing. Basically we identified all the phenotypes just looking at the chromosomes on a low 10X magnification. That means that in a field of view, you got maybe 200 cells, so you couldn’t actually look precisely at their phenotypes.
Now, the idea is to better characterize phenotypes by looking at specific markers, and to better pinpoint where the phenotype is arising. So what we see in mitosis – is it the result of something occurring beforehand, or is it actually a mitotic phenotype? That’s the reason we use a variety of markers.
In my specific case, I am trying to develop an assay that will be able to easily detect whether there are problems in microtubule dynamics. “Easily” meaning that we just run our assays on the cells and you can immediately say what’s happening with microtubule dynamics from images. For example, we are looking at microtubule growth velocity. We are perturbing growth velocity using RNAi knockdown. There are two assays that we thought about. One was basically a multiple particle tracking assay – in other words, following all the microtubule tips within the cell, and then looking at the average behavior of all the tracks we record, and extrapolating their average velocities. That gives us an indication of whether it’s the same or not as wild type.
The other assay is to count nucleation events at the centrosome. We know – and it’s already been published – that the ability of centrosomes to nucleate microtubules varies along different stages of the cell cycle. It increases tremendously in prophase. If you modify or alter the amount of microtubules that are nucleated, you will actually have mitotic defects. Ours is just a very simple assay to count the tips – fluorescently tagged EB3 tips that originate from the centrosome – and to compare that to the wild type.
Both of these assays – tracking a large number of particles, and manually counting fluorescently tagged proteins – that seems like it would involve a lot of standing over a microscope. Have you automated this?
The number of genes we want to look at is very large, so the idea is to automate all of the different steps as much as possible. We would definitely like to automate the imaging. At the moment, my movies – and I’m talking right now just about my assay, the microtubule assay, because the lab is actually trying to come up with other assays for other markers, like the nuclear envelope, chromosomes, et cetera – my movies involve looking at single cells with a 100X objective, with a much higher time and spatial resolution.
The idea here is that in the future we would like to still use the RNAi arrays that were used in the Nature Methods paper. Our thought is that if you manage to do this and automate the imaging, you would first take an image with a low magnification, be able to identify with imaging software the cells you are interested in – for example, the mitotic-based cells. And then you would switch objectives and do a high-resolution movie of the cell of interest. This is where we would like to increase the automation of the microscopy. Right now I’m still working on getting the best conditions for doing the high-resolution movies.
Then, in the analysis, it’s already a workflow of short protocols – image processing and analysis steps that are not manual. We’re not physically counting – it’s a computer – but at the moment, it’s still in a workflow that has to be organized by a person. It’s me on a single cell doing one after the other, a series of programs. In the future, what we would like to do is automate the analysis in such a way where you start with a specific cell, but then it’s a workflow that’s totally automated. You can program all the parameters in the beginning, and then the image-analysis software will do all the processing and feed the processed data into the data-analysis software. We’re not at that point, but all the different steps are automated. There’s a program that does all the counting and tracking.
Is this a program that your lab developed?
Yes, in collaboration with [Damian] Brunner’s group [in the Cell Biology and Biophysics group] at EMBL.
How does the MitoCheck project play into this?
We are using some of the data from that project. All of the data is going to be published soon, and it will be freely available. Basically they did a total genome screen looking for mitotic phenotypes. They screened for 20,000 genes using microscopy and siRNA knockdown, and they isolated a number of genes that could potentially be involved in mitosis, because they could see the resultant mitotic phenotype.
Will your lab continue to develop its own automation for this? This sounds like something that could be accomplished with dedicated high-content imaging systems.
I am not really the person in charge of this area. I’m sure that if some interesting targets come out of this, I will be interested more in the biology as opposed to the automation. If automation or collaboration occurs, it’s probably going to come from the group’s other part of this consortium.
Are you planning to publish the data you presented here?
The idea is to publish a method paper soon, and the specific biology will come later.
Is this something that would be useful in drug discovery?
I think my assay is more useful as a way to better characterize mitotic phenotypes. For that you want to understand where a defect is arising from, and where and when it occurs. If you want to do it by microscopy, you have to look at a number of different proteins within the cell – so different markers: microtubules, chromosomes, nuclear envelope, centrosomes, kinetochore proteins – to have a complete picture of all the components in the cell.