At A Glance
Name: James Dennis
Position: Senior Investigator, Samuel Lunenfeld Research Institute; Professor of Molecular Genetics and Microbiology, University of Toronto
Background: PhD, biochemistry, Queen’s University, Kingston, Ontario, Canada; Postdoc, German Cancer Research Center (DKFZ), Heidelberg, Germany, and University of Toronto
High-content cellular assays have made their way into more than just pharmaceutical drug-discovery labs. A prime example is the laboratory of James Dennis at the Mount Sinai Hospital’s Samuel Lunenfeld Research Institute in Toronto, where Dennis and colleagues have been using high-content assays in their studies of receptor glycosylation and cytokine signaling. Dennis, who will be presenting his team’s findings at Cambridge Healthtech Institute’s upcoming High-Content Analysis conference, took a few minutes last week to discuss the research with Inside Bioassays.
How did you first become interested in high-content assays?
My research is probably typical of a lot of cell biologists and geneticists who are looking at different mutant mice and mutant cells to find out how pathways of signal transduction — and other sorts of pathways that regulate both disease and development — are constructed. A lot of the things we don’t understand are how these pathways interact with one another. Often people will be looking at what they think is a linear pathway for signal transduction, and it’s not really linear at all — it has many bifurcating interactions that lead to other molecular interactions and phenotypes. You can design an assay where you’re looking at one thing, and really you should be looking at a lot more. And I think that a lot of these cell-based assays are going to allow us to do that. This is where I think the power of being able to follow these phenotypes together in real time — not only to look at them in a static way, but also in a dynamic way. There’s a huge amount of potential for new information from which one can construct models of how the phenotypes look in real time.
We got into this about three years ago when we purchased a Cellomics instrument, and we’ve been using some of the fairly simple, straightforward assays, such as cytonuclear translocation of proteins like Erk and SMAD, which are part of the basic signal transduction machinery in the cell. I think all of the manufacturers now make machines that will do some things fairly similar, but one of the things that really struck us right away when we started using the instrument was that we got much greater sensitivity with cytonuclear translocation assays than one can get with a Western blot, which is a standard thing that people have used for years. We could measure things with time courses and dose responses on a 96-well plate with very few cells in each well, and get great statistical reproducibility. So this is something with which you can generate a huge amount of data — more than you could with a Western, at least for us. This really gave us a lot of flexibility, and led us to really thinking about the dynamics … and we can titrate in both time and dose, so it led to all sorts of new experiments. When you have more potential for looking at all these things at once, you can do a lot more.
What specific signaling pathways have you been using high-content assays to elucidate?
We had some mouse mutants that we were interested in, and they involved protein glycosylation, which is a very pleiotropic modification in the cell, and it’s been a difficult thing to determine the function of, say, Golgi pathways for N-glycan modification, which is one of my areas. This gave us some other ways to approach the problem — just being able to take cells from our knockout and wild-type mice, and, to deal with it on a larger scale for different pathways of signaling that might be affected by this N-glycan. We came up with some really striking results early on. We probably could have found them with other techniques, but I think the cell-based assays gave us much more flexibility. But we found that we could remove tumor cells from these different backgrounds, and in the case of the mutation we were studying, which was called Mgat5, that this could regulate a number of different imaging pathways. So the cell-imaging system helped us track down what regulation was all about at the molecular level. It turned out that the sugar structures on the cytokine receptors were binding to the family of lectins called galectins, and this seemed to protect them on the cell surface and kept them longer on the cell surface. They actually signaled better when these sugars were on the receptors. Our mouse mutants were deficient in signaling because of the loss of these receptors, and when we went back to look at the mice again, we realized that in vivo, these mice have a lot of problems as they age. It turns out that aging is a problem where receptors are declining on stem cells and a lot of other tissues. A lot of things like insulin resistance and general declines in stem cells might well be due to changes in receptor levels in our sugar structures. We’ve also used smaller organism like C. elegans to do some work.
This research also has implications for cancer and autoimmunity?
Yes — cancer, autoimmunity, and aging, which sounds like everything, and it probably is pretty pleiotropic in that regard. But we’re beginning to understand a little about how it works. One part of our problem was being able to trace back further in the pathway where the sugar structures were being made in the Golgi, and how metabolic intermediates would contribute to that. It turns out that in some of the key metabolic pathways that everybody is interested in, the intermediates are very basic and also feed protein glycosylation. We’ve now traced that back with a computational model where we can reconstruct the inputs from metabolism into the Golgi. The cell imaging techniques will be very powerful for some of the predictions that we’re hoping to be able to test from the model. We hope to be able to use the computational model to make changes in silico to the pathway to predict how this should affect signal transduction, and then go back and test these on the cell-based assay platform.
From the standpoint of an academic lab, do you find these high-content systems to require any special expertise, or are they sometimes prohibitively expensive for a lab to use?
I think you might say the latter was true for most labs. We were fortunate to have a group grant based on equipment from the Canadian Innovation Foundation, which was meant for the whole institute, but we have gotten a lot of use out of the instrument for the last three years. I would say it is pretty expensive, but one of the things that has been happening, in talking with some of my colleagues, is that the software is becoming more easy to use for anyone to pick it up and use it quickly. But it’s also going to drop dramatically in price. There’s a lot of competition for better software, and there are a lot more competitors for providing instruments as well, so the price is liable to become a lot more reasonable sometime soon. But it’s still a bit pricey.
Do you think it has as much potential in academic research as it does in industrial-scale drug discovery?
I think you can do both. There are some really exciting things we’ve been thinking about with some of the computational models you can derive for signaling pathways, or whatever you’re interested in. And then you can test drugs for how they affect the kinetics of these pathways. You can use the computational models to help predict where the drug targets are, and also help you to triage which ones you’re more interested in. So there’s a lot more latent potential in these machines on the computational side.
What’s next for your lab?
We’re hoping to update all the software and we’re really looking forward to being able to do the next generation of assays. As an academic lab, though, we’ve not been able to update it with this most recent stuff. Recently we’ve gotten another one of these CFI grants, and we’re all set to update everything. So the next generation is going to be, shall we say, multidimensional phenotyping where you can apply some sort of artificial intelligence software — basically to cross-reference all of these phenotypes to get patterns that are much more complex than we have now, that will tell us about signaling pathways and drugs are actually functioning when we look at the cells.