Pancreatic Organoids - A Conversation with Senthil Muthuswamy
, by NCI Staff
Editor's note: Senthil Muthuswamy is Director of the Cell Biology Program at Beth Israel Deaconess Medical Center, which is a research partner of the Dana Farber / Harvard Cancer Center in Boston. As a young researcher he investigated cell polarity in 3-dimensional breast acini in the context of wild-type or mutant ErbB receptors. In 2015 Muthuswamy, Ling Huang, and colleagues described methods for deriving pancreatic organoids from patient tumor samples and from human pluripotent stem cells. In a recent telephone conversation Dr. Muthuswamy discussed his research career and his hopes for using pancreatic organoids to optimize patient treatment. The text below has been edited for clarity.
How did you become interested in 3-dimensional cell structures? When I was in grad school I took pathology courses and the pathologists told us that loss of cell and tissue structure within a tissue is what defines cancer. If the structure is normal, then an abnormal increase in cell number is usually considered hyperproliferative disease and not cancer. Then I came into my cancer research lab and the cell culture models used to study oncogenes and tumor suppressors pathways did not model the cell structural and biological changes noted by pathologists. Further, back in those days, fibroblast transformation was considered as a pan-cancer model. This disconnect between bench and clinical science prompted me to think about growing epithelial cells as a three-dimensional structure, which can create some aspects of tissue structure, and use these 3D epithelial structures for asking questions like "How do oncogenes transform epithelial organ-like structures?" Mina Bissell's method of growing breast epithelium in 3 dimensions was the most advanced platform at that time.
Why don't we start even further back. Before Beth Israel and Cold Spring Harbor you were in Toronto, correct?
Yes, I came to North America from Tamil Nadu, in the south of India, where I majored in plant science and then did a Masters in plant genetics. It was 1989 when I came to Canada to get my Ph.D. at McMaster University. In 1995 I contacted Michael Gilman at Cold Spring Harbor Laboratory to do a postdoc, but he had just decided to move to ARIAD Pharmaceuticals, so I had to decide whether to join him at ARIAD or not. Mike had plans to maintain an academic lab within the company, so I decided to take a chance and join him there as a postdoc. I wrote a grant to the Department of Defense to combine a 3D culture platform to grow breast cancer cells with a protein dimerization technology that was present within ARIAD to activate HER2 / ERBBB2 oncogenes within 3D epithelial structures. I got funded, which gave me the freedom to pursue an academic project within the company.
In 1997 Dr. Gilman decided to give up his academic lab in ARIAD, and Joan Brugge, who had also been at ARIAD, was moving on to Harvard Medical School. So I approached Joan and request to join her laboratory to continue my research project with her. Joan was an expert in cSRC biology and cell signaling in general. She must have liked my ideas, because she took a chance of accepting me into her lab. I moved to Harvard and within a year Joan connected me with Mina Bissell. In 1999 I spent some time in Mina's lab learning their 3D cell culture methods. Then I came back and established a 3D culture method to grow MCF-10A human breast epithelial cells in 3D in Joan's lab, and that's what led to the 2001 paper. Before the paper was published I was offered a faculty position at Cold Spring Harbor Laboratory to start my own lab, and I was there until 2008 first as an Assistant and later as an Associate Professor. In 2008 Ben Neel, who was a faculty member at Beth Israel when I was a postdoc in the Brugge lab, had moved to Canada to become the Director of the Ontario Cancer Institute. He approached me at a meeting and explored the possibility of me moving. Since I knew Ben from postdoc days and the position in Canada was within a cancer hospital and part of a multi-hospital network, which meant close ties to clinical research, and Canada was not an alien place to me, I decided to take another chance. The leadership at Cold Spring Harbor Laboratory was very kind and generous and gave me the option to maintain my laboratory at CSHL to allow my lab members who did not want to move to Canada to continue their research. So I left my R01 grant and people at CSHL and for the next 4 years I did a back and forth commute once a month. We had Skype-based joint lab meetings every week and I ran 2 labs. When everybody in CSHL lab had graduated or moved on, I became more strongly established in Toronto. In 2014, I heard about some emerging opportunities at Beth Israel. This was exciting for me because I have always been keen on the possibility of coming back to Harvard Medical School and Boston, because, in my opinion, it's one of the most dynamic places in the world to do research.
Tell us about the origins of your work using 3D culture at Cold Spring Harbor.
My research program was started using breast epithelial cells, cancer or normal, grown in three-dimensional culture, to ask questions related to epithelial cell polarity and cancer. I'm not sure if you are familiar with cell polarization, it is a fundamental cell biological property by which cells organize themselves, between each other and within a tissue as a whole. I'll give you an example that is applicable to both breast and pancreas. In pancreas, epithelial cells in the acinus secrete digestive enzymes that go into the luminal space and not into the tissue. The enzymes are never allowed to escape the lumen because if they get to the tissue, they'll digest everything and create problems. So the question is, how do the cells know to secrete vectorally into the luminal space and how do the cells form and maintain a tight seal to keep the enzymes inside so they don't leak into the tissue? Now you take the breast tissue, it's exactly the same. The milk that is secreted by the breast acinus goes into the lumen and collects in ducts and flows out through to the nipple, it never goes into the tissue. These properties are fundamental, essential for life and evolutionarily conserved. Cell polarization, the property by which cells establish and maintain directionality ensures these biological and physiological processes are carried out. This beautiful and essential property of epithelial cells to maintain a directional and highly compartmentalized cell organization is appreciated well by cell biologists and pathologists. Even cells that are dividing and moving around maintain this organization without fail. The cell biological mechanisms that tell a cell where to place what and how does it go wrong in cancer, is a question my laboratory has been interested in since 2001.
That's very clear, thank you. Now tell us about the origins of your work in pancreatic organoids.
It's an interesting story. When we moved to Toronto I met Gordon Keller, who is a world renowned stem cell biologist interested in making beta cells to make insulin and help patients with diabetes. We were sitting down in a student's thesis committee meeting and his student was presenting her work and I asked Gordon if they can differentiate stem cells into beta cells why can't they generate ductal or acinar cells? These are exocrine cells where most of the pancreatic disease originate, and if you can take human stem cells and make them into one lineage, why not make them into the other? And his response was we haven't tried it, I don't think anybody has.
What year was this?
This was 2009. Then I said, well we should try it. Why don't you give us some lineage-committed stem cells and we will see if we can get them to differentiate into duct and acinar cells. So that's how we started. And then I came to the lab and I talked to my post-doc, Ling Huang, who is the first author on the paper and now a junior investigator in my group. We started digging in the literature for what happens in the normal development of pancreas, and what factors are known to allow differentiation and morphogenesis of exocrine pancreas, these are the ducts and acinus. We made a checkerboard of different factors in different concentrations, and it took us about three, four years. It was a long process and Ling didn't give up and I always give him credit for his perseverance. After several permutations we succeeded in inducing human embryonic stem cells to differentiate into exocrine ductal structures showing the expected cell biology and the expression of cell-state-specific markers. So this is how we started working on the pancreas.
There was also a practical motivation, there has been a huge knowledge gap in the pancreas cancer modeling space. Almost everybody doing breast cancer research now grows cells in 3D and our 2001 our paper helped show that 3D culture is powerful and relatively easy method for studying breast cancer. In 2008 there were only a few human pancreas cancer cell lines and they were not being investigated in a three dimensional context. So I thought maybe it's worthwhile getting into this and develop culture models that can help the pancreas research community as a whole.
Is your lab largely devoted to the pancreatic organoids now?
No, we are an epithelial organoid lab that uses organoids to study cell polarity and cancer biology. Most of our work in cell polarity is breast -based with some early stage efforts in pancreas. In collaboration with other labs we are now generating organoid platforms for prostate and maybe for one or two other organs.
How did you get to growing pancreas tumor cells as organoids?
As we developed methods to induce differentiation of embryonic stem cells into pancreas exocrine cells, ducts and acinus, we reasoned that the conditions should or could support the growth of PDAC [pancreatic ductal adenocarcinoma] cells since they too are exocrine derived. We were right, the media conditions were highly effective in supporting growth of primary human PDAC tumors cells. This opened a lot of possibilities to answer a number of questions for pancreatic cancer.
Can you briefly explain the process from getting a patient tumor tissue and how you generate these tumor organoids and what you get in the end?
Sure, we digest the tumor tissue and gently separate the tumor cells from other cells. The tumors are plated on a bed of basement membrane matrix, commercial name Matrigel, in a media cocktail that we have developed. Over the period of the next 2 to 3 weeks the cells grow and organize themselves into mini-tumor-like structures, which we refer to as patient-derived organoid [PDO] cultures. These structures can be serially passaged by expanding the culture at the ratio of 1:2 or 1:3 every 2-3 weeks. After expanding the culture for 2-3 passages, we can to begin to use the culture for drug screening, genomics, histopathology and molecular biology studies aimed at discovery research. These PDO cultures can be cryopreserved and can also be used to generate xenograft tumors in mice for in vivo studies.
There has been a great deal of attention on organoid models for pancreas and multiple methods reported. Is this excitement justified?
I know, more recently there is a great deal of focus on organoid models for cancer research. I will say that growing normal and tumor cell lines in 3D is not new and there is no reason to be excited about that in particular. However, the recent advances in growing primary patient tumor-derived cell in culture are very exciting. Because, for the first time, we can keep a patient tumor cell alive in culture for extended periods of time, this is strikingly different from establishing cell lines from a patient because the process of establishing cell lines inevitably introduces a selection step that results in cells that need not represent the heterogeneity of the patient tumor it was derived from. Furthermore, one cannot envisage the possibility of establishing a cell line from every patient to understand that patient’s tumor. In contrast, we can generate PDO cultures from a specific patient and use it to experiment with treatment options and test if there are drugs that can control that patient’s tumor cells better than the oncologist’s choice drug. So the patient’s own PDO cultures become the guinea pigs for finding the best possible treatment option. As you can imagine, this is very exciting. In addition, these PDO cultures become a great discovery tool, for finding new drug targets and also possibly in the immuno- oncology space. I like to compare this period of the tumor organoid field to the early days of patient-derived xenograft model research. There is a lot of promise and hope, but lot more work needs to be done to validate the utility of organoids. It is worth noting however that compared to PDX models, PDO cultures can be ~10X faster and cheaper, so the excitement on organoid technology is well justified, in my opinion.
Would you say your method is better that the other methods?
No, every model has its own pros and cons and no model is perfect. We all know that they all are models, and no model can recreate the cancer in a patient. Every model has its benefits, so one may chose a model on the basis of their own interest and reasoning. Having said this, I do think that organoid models bring us several steps closer to modeling a patient tumor in culture, and this is what will help us discover things that we have never done before.
What are some the unique features of your organoid method?
What we especially like about our organoid method is that the organoids can recreate the histopathology seen in the matched primary patient tumors. Every patient derived organoid tends to show patient-relevant histological features, giving us a unique opportunity to model patient tumors in culture. In addition to modeling patient-to-patient variations in histomorphology, we notice intra-tumor differences in tumor histology modeled in our PDO cultures, thus we have an opportunity to recreate both inter-tumor and intra-tumor histological heterogeneity in culture. As you can see, I have come a full circle, from the pathology course about tumor histology during graduate school – to, now, taking patient tumors, growing them in cell culture and getting them to look like the tumor in the patient it came from. In particular, if we want to use patient tumor material ex vivo to discover new biology and design better treatments – it would be important to maintain the tumor cells in a state that is as close as possible to that present in the patient. I should emphasize that our culture media conditions do not activate stem cell pathways such as Wnt. Lack of stem cell pathway regulators helps us maintain tumor cell in their native differentiation state and thus allows them to organize as mini-tumor like growth in 3D culture. We've posted our media and methods on our lab web site [http://muthuswamylab.org/].
What do you hope your patient tumor organoids will allow you to do?
I think organoids can be a powerful clinical and translational discovery tool. The power to keep the patient tumor alive opens door for developing a deeper understanding of patient—patient variations in tumor biology. While genomics can be performed effectively using fixed or frozen tumor material, we need cells to be alive for understanding cell biological phenotypes such as response to drugs, and expression of surface antigens. Organoids become a powerful companion to genomics efforts by providing a platform for correlating phenotypic and genotypic differences in patient tumors. And because organoids are so much more efficient than the PDX approach, you can generate organoids from much larger cohorts of patients and cryopreserve them. When needed, they can be revived and transduced with genes of interest to do small to medium scale screens, or perform a screen with an experimental drug, for example with a RAS inhibitor, and study patient responses.
There is also potential for use in a clinical setting by performing phenotype analysis to match patients with drugs. There is no evidence at this point supporting this use, but efforts are underway in multiple labs, including ours, to determine whether PDO cultures have the predictive power for patient response. I believe that this approach will be a compliment to the large scale genomics efforts that have been underway for many years and yielded limited results. Assays using PDO cultures are focused on phenotypic responses and do not rely on the predictive power of the a genomic signature. Thus, it becomes a direct assay that may be more relevant to predicting patient responses.
You've been talking about the patient-derived organoids. Your Nature Medicine article also described how to induce hPSCs [human pluripotent stem cells] to become exocrine organoids. How do you hope use these?
There are practical advantages and biological advantages in the organoids made from hPSCs. Since we use a cell line to generate these, we can grow them in large numbers in a predictable manner. By contrast, some PDOs grow quickly and some are slow. Using hPSCs, we can induce large batches of cells towards exocrine ducts and acini and use them for understanding normal development and differentiation, and to express pancreas cancer relevant genes and models early lesions. It's important to bear in mind that almost everything we know about early lesions in pancreas comes from using mouse models. So the major advantage for us is to model early human panIN [pancreatic intraepithelial neoplasia] lesions in a culture. In addition to cancer, we hope to use the progenitor organoids to model other pancreatic diseases, such as pancreatitis.
Thank you, Senthil, and much success in your exciting research.