Cytocentric Visionaries: Shannon Mumenthaler, University of Southern California
Part Two: Oxygen, Cell Morphology, and Time
In Part One of this two-part interview, we talked with Dr. Shannon Mumenthaler about her latest study published in Nature’s Open Access Journal, Scientific Reports  and her unique combination of high throughput image analysis, heterogeneous cell culture, and full-time control of conditions. Today we talk about adding new dimensions into highly dimensional cell parameter space, including cell shape and time.
AH: I was fascinated by your use of cell and nuclear shape characteristics as cell parameters. What does adding the morphological dimension add to our understanding?
SM: Cancer biology is extremely complex. Cell and nuclear morphology are parameters that tell you a lot about the underlying biology without having to understand the detailed signaling networks that are changing.
We also need to consider time as an important factor in cancer biology. Some of the environmental stresses in vivo will impact cells on a short-term basis, while others may take longer to have a measurable effect. Live cell imaging is essential to the investigation of these temporal dynamics, allowing us to hone in on the appropriate time frames that biological changes are occurring. You might see the morphological changes happening quicker than downstream readouts, such as viability stains. The integrity of the cell membrane is disrupted much later than the minor changes in morphological properties.
In our case, we coupled the hypoxia work station with a competent screening platform because we needed real-time tracking of cellular behavior. More and more we’re seeing these papers saying, “What’s the time frame that you should really be looking at cells under drug treatment?” or “Can we capture those short-term dynamics as well as the long-term dynamics?”
Some cell assays are only read at 24, 48, and 72 hours. Should we consider that a comprehensive understanding of cell behavior? We propose a way to look at an unperturbed system and see the dynamics of how cells are changing.
With this imaging platform, we can start to adjust lots of different environmental stimuli to study how cells respond to oxygen, or drug, or even changes in nutrient availability and use morphology as a readout of stress response. It’s a way to classify the cells into buckets of cell behavior.
AH: Right now I’m struggling with what to do with findings published on cells placed in physiologically relevant oxygen for short periods of time, like 4-8 hours. Isn’t there a history to those cells that goes beyond the few hours in those conditions that makes it difficult to generalize findings to the in vivo state?
SM: I agree that it’s tremendously important to understand the past history. Unfortunately there isn’t enough information to know what cell populations you were selecting for or against when you put them under atmospheric oxygen, then hypoxia for 8 hours, and then put them back again.
We know that certain cell types, depending on their molecular alterations, may have an increased fitness advantage under low oxygen conditions. If you grow them for a long period of time in atmospheric oxygen, that clonal population may be at a very low frequency. You might see an outgrowth of those cells upon switching them into hypoxia, while placing them back in atmospheric oxygen may result in another reduction of that cell population. The clonal dynamics are constantly changing in that heterogeneous population.
Consequently, I believe environmentally controlled experiments are going to be crucial going forward. We need to improve the in vitro early drug screening approaches and control for these important environmental conditions in pre-clinical models. We need evidence from these experiments to convince the field that oxygen is something that needs to be controlled in all labs. We’re starting to see papers come out that are saying this, but I think there is more we can do as a field to emphasize the importance of physiologically relevant environmental and temporal dynamics on our preclinical research.
I think that’s why having a well-controlled system that will allow you to observe changes caused by variations in oxygen over long periods of time is critical. We need to be able to control that environment from the start, so that the cells never hit atmospheric oxygen. The stress of cycling between low and high oxygen, even for short periods of time, can change the underlying biology. That’s something we didn’t appreciate when we first started out and were using a different system. We are now positive that a controlled and well-maintained system is essential.
AH: How comfortable are you with letting go of human image interpretation over time and turning it over to algorithms?
SM: As a biologist, it’s hard to let the algorithms do the interpretation. For me it helps to also interact with computational scientists to ensure that what we’re analyzing makes sense.
The use of high-throughput imaging is a huge change. How can we sift through all the data, especially from a high-content screening platform? The amount of data that is generated in the image files alone is massive, not to mention the spreadsheets that come out with the various parameters. This is probably our biggest bottleneck- how to deal with all this data. On the other hand, we now have a wealth of additional data we can analyze to uncover what’s really happening with the cells under different environmental contexts or treatment responses.
Machine learning is becoming more important in this research space. We can uncover new insights from data trends that we may not have realized existed before.
AH: It helps take some bias out of the system, we’re not coming in with a story or narrative already in our head. It’s the computer.
SM: I agree. We’re not baking the outcomes we want into the data. We’re handing over the data and seeing how the results come out. I think that’s also important to do.
AH: Thank you for your time and your insights. Dr. Mumenthaler. We will be watching for more exciting publications from you and your students as you work with your interdisciplinary team and a well-controlled cell environment!
This concludes our interview series with Dr. Shannon Mumenthaler.
1. Garvey, C.M., et al., A high-content image-based method for quantitatively studying context-dependent cell population dynamics. Scientific Reports, 2016. 6: p. 29752.