Name: Yael Paran
Position: Staff scientist, department of molecular cell biology, Weizmann Institute of Science, Rehovot, Israel
Background: Postdoc, laboratory of Benjamin Geiger in collaboration with Zvi Kam, department of molecular cell biology, Weizmann Institute of Science, 2003-2006; PhD, biological regulation, Weizmann Institute, 1997-2002; MSc, chemistry, Weizmann Institute, 1994-1997
Scientists from the Weizmann Institute of Science in Rehovot, Israel, have developed a high-resolution and high-throughput automated microscope for cell-based screening using off-the-shelf components.
They began developing the microscope several years ago to automatically analyze and screen specific sub-cellular protein complexes at resolutions that could not be achieved by still relatively young commercial systems. Some commercial systems exist now, but by and large, automated imaging is still done at a fairly low resolution compared with bench-top microscopy.
Contributing to the development of the platform were Weizmann scientists Zvi Kam and Benjamin Geiger, who headed the project; Yuvalal Liron, who helped develop image-analysis and control software; and Irena Lavelin, Sabina Wibograd-Katz, Suha Naffar-Abu-Amara, and Yael Paran, who applied specific biological assays and conducted screens using the platform.
Paran, who recently finished her postdoctoral work on the project, presented a poster on the technology at Cambridge Healthtech Institute’s High-Content Analysis conference, held this week in San Francisco. CBA News caught up with Paran to discuss her role in the project, the impetus behind the development of the imaging system, and some of the microscope’s features.
Tell me about your background, and how it fed into the development of this system.
I did my PhD on magnetic resonance imaging, but then I moved as a postdoc to this group led by professors Zvi Kam and Benjamin Geiger, who collaborate at Weizmann. This was essentially my postdoctoral work. Before I arrived, they had developed this autofocus apparatus for automated microscopy, but when I came, I became involved in integrating it into biological work: adding the cell-based assays, actually performing the experiments, and making the system work.
Now the platform is working, and we have signed a contract with a company called IDEA in Israel to manufacture the system. It has actually only manufactured two copies of our system, which are both at Weizmann. It is not commercially available yet, but now they are considering whether to move to a commercialization business plan. We have used it in our group already for a few different screens, and other groups from the Weizmann have conducted some screens on the system.
Can you describe the autofocus technology that drives this?
The autofocus is based on a laser beam that is reflecting when it moves from the air to the bottom of the plate, and there is another reflection when the beam crosses from the plate to the water of the well. Software actually collects the intensity of these reflections. It detects two peaks, and uses a mathematical algorithm to identify where exactly the bottom of the plate is. Control software then knows how to move the objective to the right plane to give the right focus based on this laser signal.
This is designed for well plates and slides?
We can use both, but it is used more intensively for multi-well plates.
How does the microscope move from well to well?
The software controls everything, and you can define how many images you want per well, and it decides how to move within the well. It moves from field to field, and autofocuses, and it is very accurate for high-magnification imaging, which is the reason we built it.
Is the imaging confocal?
No, it’s brightfield. We can use either fluorescence or phase-contrast. The laser is only used for the focusing. A CCD camera takes the images.
What are some of the applications you are using this for, and why did you develop this microscope for these applications instead of using commercially available automated imaging platforms?
We are interested in the biological issues of the cell cytoskeleton, and adhesion of cells to the extracellular matrix. For that, the cells use focal adhesion points, which are very small complexes of proteins used as points of adhesion. We were interested in looking at these points and learning the morphology in a very accurate manner, and getting as many parameters – morphology and intensity – of these structures, which are very small, a few microns. We needed a high-resolution imager, and we wanted to do it in a high-throughput manner and in a cell-based context. When we started to develop this about five years ago, all of the commercial instruments were in the development process, and just starting to hit the market, and none of them had the resolution we needed. We are not industry, so we took our time to develop this, so we could look at these structures. Additional structures that are interesting to us are, for example, microtubules and actin fibers. It’s important to look at these in high resolution. You can see them in low resolution, but the details that you can gain are not the same as in higher resolution.
So the first screens we conducted were for these focal adhesion points, where we prepared stable cells lines that expressed proteins linked to yellow fluorescent protein and localized to these focal adhesion points. We could then see these objects in the cell, and we screened for a set of chemicals – both natural and small-molecule compounds – for their affect on these focal adhesion points. We developed the image-analysis processes that extract this information automatically. We have the visualization tools in the software, where you can look at the data, and also the analysis tools. These tools are developed for each user in our group, for looking at specific morphological features. We look at the statistics, or the distribution of the parameters that describe these focal adhesion points in control wells, and compare that statistically to the parameters in treated wells.
For that, we don’t need many cells, because in each cell, there are many such complexes. We did this with chemicals, and then moved to an siRNA screen to look for genes that affect these focal adhesion complexes and the related pathways. And we also conducted a screen to look for new genes or proteins that belong to the cytoskeleton. Additional screens conducted by other groups besides are included siRNA screens in Drosophila, and some yeast screens. These are all looking at different sub-cellular structures.
So there are a lot of other potential applications for this instrument?
Yes. Other applications involve, for example, mitochondrial screens, because the mitochondria are very small organelles, and also, Golgi bodies, small vesicles, and more. These things can be seen at lower magnifications, but it doesn’t compare to the amount of detail you get at high resolution. There are also viable applications from scientific literature. For example, there are viruses that use the cytoskeleton to move inside the cell, and need it to replicate. It might be possible to screen and search for chemicals that affect these viruses, and you could not look at these moving through the cells using lower magnifications. Translocation of proteins into the mitochondria as a part of the apoptotic pathway is another example.
Now that high-content imaging has just started to establish itself in industry, do you think that people are ready to start exploring such high magnifications and obtain such fine detail in screens?
Well, walking around this show and looking at websites — there are machines that now go to 60X magnification, such as ours does. But I’m not sure if they have refined the focusing method as well as we have. They know that it is needed. We are certainly not the only ones thinking about this. From talking with people, many think that it is more useful to screen at low magnifications, extract a large amount of data and cells, extract hits, and explore the mechanisms using very high magnifications — but not in the context of screening. We think that in some cases this is true, but we also think that in some cases you lose a lot of hits because you don’t look at high-enough magnifications. Each person should choose what he needs, but there are screens that could use this capability.
The magnification of your instrument is 60X?
It can be used with any objective. The most common use is for the 60X, but with 0.9 NA, which is very important, because that is a very high resolution, and it needs very high accuracy in terms of focusing.
What about analyzing these high-resolution images with software?
It’s difficult to do, and in some cases it needs a different approach than lower magnifications. For example, it is not simple to define the cell, because in many images, you see only [a] part of one cell, or a few parts of [a] few cells. It’s a big question we have been dealing with for a long time — how to extract the cells and how to get the information from them, which is something we haven’t done yet. We introduce information about the area in a well that is covered with cells, but not how many cells there are. We have some people working on this.
Weizmann owns the patents to this?
And you are working with IDEA to commercialize this? Have you started to look into placing these for beta-testing?
They are working more on the business plan for this, and this is part of it. But it is mostly in their hands – we are interacting with them, but they are looking into commercialization.