The Stroma of Pancreatic Cancer Supports Proliferation of Tumor Cells

April 30, 2019, by Jim Hartley

Hartley: Your recent paper used biology and chemistry to discover that stromal cells in the pancreas contribute lipid signals to support pancreatic cancer. How did you get together to work on this topic?

Sherman: My area of expertise is the pancreatic tumor microenvironment, and my postdoctoral training was trying to understand transcriptional regulation of the tumor microenvironment, and also how the tumor microenvironment can function in a paracrine manner to regulate transcription in the epithelial compartment. Jurre and I met when we were both postdocs, and our postdoc mentors [Sherman with Ron Evans at the Salk Institute, Kamphorst with Josh Rabinowitz at Princeton] were on a Stand Up to Cancer dream team focused on pancreatic cancer together. Jurre has amazing expertise in measurements of lipids, and so he and I got to talking about these unique cells that are found in the pancreas, these stellate cells, that really are not terribly well characterized, but their perhaps best known feature is an abundance of cytoplasmic lipid droplets.

Kamphorst: I think the initial discussions about this, and the initial observations were back in 2012, that’s quite a while ago. I remember doing a little bit of work on this, but we were both postdocs and we also had our other papers we were working on. When I had my own lab I remember Mara sending an email, something like “should we work on this again?” So Mara gave the push to get it going again and we started doing some experiments and they looked really interesting, and that got us increasingly excited.

Sherman: Based on in vitro modeling as well as extrapolation from studies in the liver, people think these stellate cells are the predominant source of fibroblasts in the pancreatic tumor microenvironment. But when you look at the fibroblasts in these tumors they don’t have these lipid droplets, and so the changes in lipid metabolism that accompany stellate cell activation and what happens to the lipids produced by these cells within the tumor microenvironment really weren’t studied. That was sort of our initial thinking, that maybe we could work together and start studying the lipid metabolism of these cells, and of course the project really evolved from Jurre’s expertise. He really helped guide the project and the detection of the specific lipids, these lysophosphatidlycholines [LPCs] which can then be hydrolyzed to mitogenic LPA [lysophosphatidic acid].

Kamphorst: You could rescue growth of tumor cells in serum-free medium if you give some medium conditioned by these stellate cells, and we started to look at maybe there are signaling lipids in the conditioned medium and that’s when we discovered the presence of LPA lipids. From there we back tracked, and we started looking for the secreted enzyme [autotaxin] that makes LPA, to see if that was present in the medium. And that turned out to be the case. Then we asked if inhibiting autotaxin produced an effect? And it very clearly did. So things came together quite nicely, and it turned out to be a really nice story.

Sherman: We had some early mass spec results showing that the intracellular lipids are dramatically different in stellate cells from a healthy pancreas compared to a stellate cell differentiated to a fibroblastic state, and so we thought there may be features of activated stellate cells with respect to lipid metabolism that would warrant study. But we really didn’t have a story until we started our own labs, and really when Jurre started his lab we did a trans-Atlantic collaboration to really dig into the lipid metabolism of these activated cells.

Hartley: What is the current thinking about the role of the stellate cells in the healthy pancreas?

Sherman: In a healthy pancreas they are small, round cells, they contain lipid droplets, they have low levels of expression of, for example, extracellular matrix components, but you can model their activation in vitro, which has been shown to happen in vivo in the liver. Extrapolating from what we’ve learned in the liver, and from what happens to pancreatic stellate cells in a dish, the assumption is that these cells can also contribute to the fibroblast population in a tumor microenvironment. The thinking about these cells is that they evolved as part of a wound-healing reaction. For example in hepatic fibrosis you see that the majority of the fibroblastic cells in the liver come from hepatic stellate cells, and pancreatic stellate cells seem likely to play a similar role in inflammation and fibrosis. So our thinking is that maybe these cells have evolved to secrete lipids to support proliferative capacity of a regenerating epithelium in a wound healing response, and cancer cells have figured out how to activate these cells in the tissue during tumor progression and take advantage of the same metabolic axis.

Kamphorst: We did have the observation that started everything, that the stellate cells have lipid droplets and after they become activated those start to disappear, and so we were thinking about what might happen to that. We had the idea that maybe some lipids, either directly or after being modified, are being released into the medium. That would be in line with what we know about these cells, after they become activated they start to release all kinds of proteins and even small nutrients. So fairly soon we profiled the medium that had been conditioned by these cells, to see if they were releasing any lipids. We used lipidomics, which is a mass spectrometry-based approach to analyze lipids, to enable us to see what was being released, and based on their mass, their retention time, and their fragmentation patterns we started to see that they were releasing phospholipids, and specifically high levels of these LPCs. And we also saw low levels of these LPA mediators, and since they’re known to be very potent signaling molecules that are implicated in cancer, that’s where we started to dig more.

Hartley: Are the fibroblasts derived from the stellate cells also releasing lipids?

Sherman: One of the challenges right now is that once the stellate cells become fibroblastic the hallmarks of stellate cells are gone. Within an established tumor we can’t distinguish fibroblasts that came from stellate cells versus fibroblasts that came from other sources. For our study we made sure that for all of our key experiments we used both. We have experiments that were done with murine bona fide stellate cells, and then we have experiments that were done with human pancreatic CAFs [cancer-associated fibroblasts] that were grown out of surgically resected primary tumor specimens, and these findings seem to hold across both. The secreted lipid levels certainly are higher from a pure population of stellate cells, and we assume that’s because the bulk fibroblasts that grow all derived from stellate cells, and we suspect that the stellate cell-derived fibroblasts are the ones that are mostly responsible for these secreted lipids. But we don’t know that for sure because haven’t been able to separate out yet which ones are stellate cell-derived and which ones are not.

Hartley: Is there a biochemical reason that the lysophospholipids are secreted? Or are they preferentially secreted, compared to the fully acylated phospholipids?

Kamphorst: As a postdoc I had a paper where we actually studied lysophospholipids, and there I actually looked at them in relation to cancer cells. I found that certain cancer cells, specifically those that carry mutant RAS, actually readily take up these lysophospholipids and essentially eat them quite rapidly, to support their growth. In the current stromal cell study we again were able to implicate lysophospholipids, but we found that the stellate cells actually seemed to be releasing them. So it’s a very interesting observation, to see the tumor cells eating those lipids and the stellate cells actually producing and releasing them. But beyond that, why stellate cells are releasing specifically these lipids, that’s not known.

Sherman: And the specific avidity of RAS-transformed cells for these lysophospholipids, and lysophatidylcholines in particular is unclear, but as Jurre said, when his previous paper showing that either RAS mutations or hypoxia, two features that are almost always relevant in pancreatic cancer, cause cancer cells to take up lipids from the extracellular space to support their proliferation and of all the lipids that Jurre looked at, these LPCs really were the preferred substrate for uptake and fatty acid scavenging. And so I think we’re both interested in understanding both the specific avidity of RAS-transformed cells for those species, and also why these stellate cells in particular have evolved to secrete such high levels of them. But we don’t really understand the basis for that yet.

Hartley: What happened when you inhibited autotaxin?

Sherman: The drug we used perhaps not surprisingly doesn’t completely inhibit autotaxin but still we saw a suppression of tumor growth. But when we used a genetic knockdown of autotaxin, so the tumor cells themselves have lower autotaxin expression from the start of the experiment, we find that they can hardly grow in the pancreas in vivo. And we think that beyond what we’ve studied for this paper there may be additional implications of this extracellular LPC-autotaxin-LPA axis in regulating other aspects of the tumor microenvironment, for example the immune microenvironment. Those are things we’re certainly interested in studying but didn’t look at for this particular study.

Hartley: Do you have any thoughts about autotaxin as a therapeutic target?

Sherman: I feel excited about it, but certainly it’s early days. You know we saw nice effects when we knocked down autotaxin in the tumor cells and then implanted them, but of course as a therapeutic strategy we want something that’s really going to induce regressions in established tumors. As a single agent an inhibitor of autotaxin did not give us regressions or stasis, it just suppressed tumor growth compared to controls. This could be in part because the drug was not completely effective in inhibiting autotaxin activity, but I also suspect this is going to have to be used in combination with other agents. That said, it’s known that if you delete autotaxin in an adult mouse in an inducible manner the mouse is viable, which suggests that it’s potentially a therapeutic target that would be tolerated.

Kamphorst: I agree with Mara that autotaxin is likely to be most beneficial in a combinatorial regimen. For instance, it would interesting to study potential synergistic effects with immune checkpoint inhibitors. It’s still early days, however, and more research is needed to figure this out.

Hartley: Would either of you like to reflect on the state of the cancer research enterprise?

Kamphorst: I think it’s good to highlight the importance of funding agencies like Stand Up to Cancer. If they hadn’t funded the pancreatic cancer Dream Team back then, then Mara and I wouldn’t have met and we wouldn’t have been doing this work.

Sherman: And also the importance of having an interdisciplinary team. I see a lot of this when I hear talks that are part of the RAS Initiative, it seems like it’s a very clear goal to bring together people from very diverse backgrounds. This was certainly an example where bringing together some diverse strengths really moved the project forward. I think we’re seeing more and more examples of this as we as a community start to understand the complexity of the tumor microenvironment. I think it’s really important to go to experts when you’re reaching into, for example, lipidomics, or even though that’s not relevant here, the immune microenvironment. I think it’s really helpful, game-changing, to work with people who truly have expertise, perhaps in different biological systems, but to come together so you can really work with experts and get some solid data.

Kamphorst: Research is relying more and more on sophisticated technologies, and we are starting to appreciate how complex something like tumor development really is. I wholeheartedly agree that collaborating in interdisciplinary teams will be the ‘secret sauce’ going forward. It takes extra effort, but it is fun to learn from each other’s expertise and perspective, and it is really what is needed to come to new ideas and new findings.