Read the Full Video Transcript

Andrea Miyahira: Hi, everyone. I'm Andrea Miyahira with the Prostate Cancer Foundation. Joining me is Dr. Mindy Graham of Northwestern University to discuss her paper, Convergent Alterations in the Tumor Microenvironment of MYC-Driven Human and Murine Prostate Cancer. This was recently published in Nature Communications. Dr. Graham, thanks for joining us.

Mindy Graham: Yeah, thanks for having me. I look forward to sharing this recent work with folks. So I should first start off by saying that this body of work really involved a lot of folks. So as you can see through this author list, this wouldn't have been possible without all of these folks helping to support this project.

And I'd also like to highlight my mentors, Vasan Yegnasubramanian, and our collaborator, Angelo De Marzo at Johns Hopkins University. And I thought I would start off by really sort of talking about what motivated us to do this study and how we derived our conclusions. And just to start off with, I'd like to talk about human prostate cancer.

So we know that it's a very heterogeneous disease, meaning that among patients, there are different molecular drivers. And even within the same patient, prostate cancer can be multifocal, meaning that there are different prostate cancers that can emerge in the same patient that have different molecular drivers.

And yet, despite all this heterogeneity, there are certain key molecular features as well as histological features that are consistent across all prostate cancers, meaning there are precursor lesions called prostatic intraepithelial neoplasia, or pin, that we observe that have very notable features that then do progress into invasive carcinoma.

And so these molecular and histological features are the same across patients or, rather, I should say, have some commonalities. And we asked, well, how is this possible? And how can we study this in a way that would make sense with the existing tools?

And so we leveraged single-cell RNA sequencing as well as immunostaining of tissues with our collaborator, Angelo De Marzo, who is an expert on prostate cancer histology, to understand this convergence that we see. And studying the single-cell data, we found that MYC was a very—it was a common denominator across all of the heterogeneity we saw in prostate cancer.

And with that, then we were able to leverage MYC-driven models, transgenic models in mice of prostate cancer, to really sort of understand the changes that were happening as a result of MYC activation in the prostate tumor microenvironment.

And things that were sort of surprising to us, and these are sort of high-level features—which I'll talk a little bit more in detail later on—are that with MYC activation early on in disease progression, it's actually immunogenic. And that's surprising to us, right? Because we know that prostate cancer is actually quite immune-suppressed, meaning that there's not much in the way of T-cell infiltration and other features consistent with a pro-inflammatory response.

And yet, what we saw later on, as prostate cancer transitioned from precursor to more invasive carcinoma, that's where we saw these immune-suppressed features that we sort of know and recognize in prostate cancer. And so we start off with human and then used mouse models and then went back to human to make sure that the observations that we observed in our transgenic models were not just a mouse-specific phenomenon but really could be found in human prostate cancer.

So to get sort of into the details of our findings, we had patient tissue samples. These were patients all diagnosed with prostate cancer, who had prostatectomies. We took these prostates and took tissue punches in each of the zones of the prostate. So as you may already know, the prostate has three zones: the transition, central, and peripheral zone. And prostate cancer primarily happens in the peripheral zone.

So we took tissue punches in these benign-enriched areas, and they also took a tissue punch right where we saw a lot of tumor enrichment. And so for each of these tissue punches, we reserved half for histology and then performed single-cell RNA sequencing with the other half. So the idea was that anything we observed in the single-cell data, could we then sort of add more context to it by looking at the tissues and performing in situ tissue staining?

When we looked at the epithelial population in particular, we were able to capture all of this prostate cancer heterogeneity. So we observed eight sort of distinct clusters among these 10 patients for which we had single-cell data. And broadly, what we observed is that these cancer clusters all had upregulated MYC activity by gene signature analysis.

And we saw this in our own single-cell data, but to really get a sense of, well, is this true for the majority of prostate cancers, we leveraged an existing publicly available dataset, the TCGA primary prostate cancer, which is known to have various molecular subtypes, androgen receptor activity scores, as well as Gleason scores. And what you can see—for the vast majority of prostate cancers—there is an upregulation of MYC. So this was sort of consistent with what we were seeing in our single-cell data.

So this motivated us to look at transgenic mouse models. And we selected a MYC-driven, Hi-Myc model. So these are androgen response elements upstream of a human MYC transgene that is expressed primarily in the prostate luminal epithelial cells.

And you see early on that these mice developed precursor lesions that do eventually go on to develop invasive carcinoma. And this was a mouse model originally developed in Charles Sawyer's group. And this was in the background of FVB. But we know that different mouse strains can actually have slightly different phenotypes.

And so we used another Hi-Myc model that was backcrossed—this was developed by our co-author, Brian Simons—into the Black 6 mice. And so any observations that we observed in our mouse models, we could sort of account for things that might be sort of unique to a particular strain.

And just to note, some of the differences between mouse and human prostate are that the mouse prostate actually has lobes and not zones. And we performed single-cell RNA sequencing on each of the lobes: the anterior, dorsal, lateral, and ventral lobes of these mouse models with their wild-type counterparts.

And when we looked at the epithelial population, we could see that there were the neoplastic MYC-expressing luminal cells. And you can see them here—we call them MYC 1 and MYC 2. MYC 1 tended to be highly enriched in the dorsal and lateral lobes, and then MYC 2 was mostly found in the ventral lobe of the mouse prostate.

But needless to say—or at least I should highlight—that really these were populations only observed in the Hi-Myc models. But what was fascinating to us is we also saw another unique set of epithelial cells emerge, and we call them reactive basal and reactive luminal. These are these populations right here. And what was striking is that they were highly enriched in the Hi-Myc models—you could see a little bit in the wild type, but really in the Hi-Myc models. And these did not express the transgene.

And so that sort of suggested that these were epithelial populations that were emerging as a consequence of MYC activation, although not expressing MYC directly. And these cells primarily express Ly6d, which you can see here.

Now when we compared the MYC-expressing luminal cells with the normal luminal cells in wild-type animals, what was really sort of striking to us was that—well, first of all, I should say we did see something that we would expect to see, and that is the MYC targets pathway was one of the top upregulated pathways.

But we also saw that the interferon response pathways were upregulated in these MYC-expressing luminal cells. And that was sort of strange to us because we know prostate cancer is, in humans at least, immune-suppressed. But when we went back and reviewed the tissues and thought about the time point—really, this is a time where you see a little bit of invasive carcinoma, but a lot of the MYC-expressing luminal cells resemble more pin or precursor stage at six months. So that was the time point that we had in these studies.

And so with the expectation of something more of a pro-inflammatory phenotype, we did see infiltration of CD3-positive T cells in Hi-Myc, as well as F4/80-positive macrophages. And when we looked at the MYC-expressing cells at a later time point—10 months—where at this stage, most of the dorsolateral lobes are comprised of invasive carcinoma,

we actually did see a suppression of inflammatory response pathways as well as IL-6/JAK/STAT3 signaling, which you can see here. And we highlight some of the genes associated with this pathway. In addition to that, with the immune suppression that we were seeing in the later time point, we also saw some of the pathways associated with acting as the gatekeepers against malignant transformation—those were also suppressed. So that included DNA repair and the p53 pathway.

And in addition to these sort of cell-intrinsic changes that we were observing, we also saw that in the tumor microenvironment, the immune composition was changing. In particular, we saw an enrichment of these TREM2-expressing macrophages, which you see here from six months to 10 months, and then really none of them present in the wild-type animals that were age-matched.

And then in invasive carcinoma, we begin to see the presence of myeloid-derived suppressor cells as well as regulatory T cells. And so again, we saw this very distinct pro-inflammatory to immunosuppressive switch that was happening in the tumor microenvironment.

Now coinciding with these immune changes, we were also seeing sort of significant changes in the stroma. And one population that I really would like to highlight are these TIMP1-expressing fibroblasts. We really only saw them in the Hi-Myc animals. So you can see that represented here on these dimensional reduction plots. And then here, when you look at the cell proportion from six months to 10 months compared to age-matched wild types, you really see this enrichment.

And as I mentioned before, we saw the TREM2 macrophages that were also enriched. When we did a correlation analysis, we actually saw a really significant positive correlation between TREM2 macrophages and this TIMP1-expressing fibroblast population. This sort of suggested that maybe these two cell populations were somehow interacting with each other in some way.

And so this motivated us to do a cell communication network analysis, where you could really leverage your single-cell data to look at the ligands and receptors that are expressed in these cell populations and infer what type of interactions might be occurring between them.

And as you can see here, the TREM2-expressing macrophages’ ligands were indeed targeting these TIMP1-expressing fibroblasts. And when we looked at the top ligands that were targeting the TIMP1 fibroblasts, TREM2 macrophages were included in that. We saw several different ligands expressed targeting the TIMP1 fibroblasts, one of which included TGF-beta 1.

And the reason why I highlight this is because TGF-beta 1, when it's activated in fibroblasts, is known to cause a cell state change. They become activated, and they express fibrosis-associated genes. And so when we looked at the gene expression profile of our TIMP1 fibroblasts and we look across the wild-type and Hi-Myc animals, what we see is that in the fibroblast population, with the increased expression of TIMP1, we saw all of these collagens associated with fibrosis—so these extracellular matrix proteins—were also highly enriched, sort of consistent with some of what we were observing with TGF-beta 1 expression by TREM2 macrophages.

Now, in addition to looking from precursor to invasive carcinoma, we used another mouse model that's MYC-driven but also has PTEN loss. And this one's a much more aggressive model. It's called BMPC. It's a HOXB13-driven model of prostate cancer developed by our co-authors Chuck Bieberich and Angelo De Marzo.

And we evaluated the single-cell transcriptomes of this population at a precursor stage, as well as metastatic disease. We collected some lymph node metastatic lesions. And when we looked at the epithelial MYC-expressing population, what you can see—if we did a gene signature analysis—is that as these cells, these luminal cells, progress into more aggressive disease,

there's actually an enrichment of MYC target genes, suggesting that MYC activity is much more enriched in these more aggressive models. And all of these immune response pathways that we begin to see suppressed in Hi-Myc at six months—in the BMPC model, in the metastatic lesions, we saw again that there was significant suppression of these inflammatory pathways, such as the inflammatory response, NF-kappa B signaling, as well as the interferon response pathways.

And with these MYC-driven changes that we saw in Hi-Myc, those were recapitulated in the BMPC model. So we saw these TREM2 macrophages in the metastatic lesions, as well as these TIMP1-expressing fibroblasts.

So just to summarize the key findings, now all of these changes that we observed in our transgenic models, we also could see them in our human single-cell data. So in the model, MYC activation is actually initially very pro-inflammatory, and this results in an infiltration of immune cell populations.

And then as the precursor then sort of switches to invasive carcinoma, we see the increased presence of TREM2 macrophages, as well as immunosuppressive myeloid-derived suppressor cells and regulatory T cells, as well as an increased proportion of these TIMP1 fibroblasts.

And we also saw in the precursor stage, which we could also see in our human data, these reactive epithelial cells, these reactive luminal and basal cells that express LY6D and keratin 6A. And so this particular summary you could also find in our recently published paper.

So as I mentioned, there's a lot of folks that have helped make this paper, this study, possible. But I'd like to highlight when I was a postdoctoral fellow, where a lot of this study sort of initiated with my mentors, Vasan Yegnasubramanian, Ted DeWeese, and our collaborator, Angelo De Marzo, and some of the folks, as well as Rullin Wang, who is a co-first author on this particular study.

And then I'd also like to recognize the funding sources, which included the Prostate Cancer Foundation, as well as the Northwestern Prostate Cancer SPORE Career Enhancement Program. And I'd like to also highlight the folks at Northwestern University, which have allowed me to help complete these studies that I've presented here, and that includes Sarki Abdulkadir and Ted Schaeffer. And finally, I'd like to also highlight that patients have been really important in helping make this research possible. So thanks again for allowing me to highlight the major findings of a recently published study.

Andrea Miyahira: Thank you so much, Dr. Graham, for sharing this super interesting study. So a key finding is that there's a convergence on MYC signaling as a critical driver. So how should we be thinking about the role of MYC versus AR in biology and also therapeutic targeting?

Mindy Graham: Right. Great question. So most of prostate cancer, as we know, is an AR-driven disease, right? So that's why we have things like hormonal targeted therapies—because they do work. Although eventually, many patients who have aggressive disease develop castration-resistant disease.

Now some recent work from David Labbe's group had sort of talked about this interaction between MYC and AR, where MYC is sort of causing a pausing of AR-targeted genes. And there's also some studies to suggest that in some ways they're sort of antagonistic, MYC and AR.

But in addition to that, some of the AR programs behave differently in the cancer setting as opposed to the normal differentiated luminal epithelial setting. And when we think about treatment, right—so patients get hormone-targeted therapies. As their cancers become nonresponsive to these sort of gold-standard treatments, thinking about things like MYC-targeted pathways may be important and sort of the next step as we try to address the issue of resistance to treatment.

Andrea Miyahira: OK. Thank you. And another key finding was that it appears that MYC is driving first a tumor immunogenic stage and then an immunosuppressive state following. So there's some kind of switch going on. So do you think that MYC is required for both of these immune-based states, and how should we think about MYC targeting in the context of immunotherapy?

Mindy Graham: Yeah. That's a really great question because I think there's a lot of enthusiasm in thinking about how can we apply immunotherapies for patients with prostate cancer? And some of these initial attempts, as you know, have been disappointing, right? Patients, unfortunately, their cancers are not responding to immunotherapy, and a lot of that has to do with just the immunosuppressed sort of environment of prostate cancer.

Now, the way we think about MYC activation is actually that it's important throughout the disease progression—from the earliest precursor stage all the way to invasive and even into the metastatic setting. And what we think about this initial MYC activation is that, yes, it's immunogenic.

And so we have all of this immune infiltration—that's what we really want for immune-targeted therapies. But that sort of initial immune infiltration is what sets the stage for eventually what prostate cancer becomes—sort of immune-suppressed in the tumor microenvironment.

Cancer at this precursor stage is sort of starting to figure out, how can I get used to dealing with this immune infiltration? And by the time it becomes invasive carcinoma and it's gone through the malignant transformation, all of that sort of pro-inflammatory response has since been dealt with. And now, you no longer have this immunogenic response as a consequence of MYC activation.

So of course, this is at the precursor stage, and we don't treat patients with precursor disease, right? And so thinking of ways in which patients who have invasive carcinoma—how can we then make it immunogenic again? Perhaps targeting things like MYC, which have sort of driven this immune-suppressed state, may be a way in which we can revert prostate cancer to something that's immunogenic and potentially respond to immunotherapies. And I think that would be very exciting for patients.

Andrea Miyahira: Yeah. Thank you. And then what are the next steps for your studies?

Mindy Graham: Yeah. So in the last part of my presentation, I talked about metastatic disease. I think this is really where there's a lot of unmet need for developing targeted therapies. Now, I did show you that the same sort of MYC-driven alterations that we see in the invasive carcinoma setting exist in the metastatic setting.

So some of these key changes—thinking about the molecular sort of interactions—there might be targetable vulnerabilities in that that we can then use for therapy. But something that I'm also interested in delving into is what makes invasive carcinoma different from metastatic disease, and how can we key into that then to develop new targeted therapies.

And with technology as it's quickly progressing—so there's single cell, there's spatial transcriptomics—there's a lot of opportunity to really investigate that. And I think what we've built here in this study is a really nice foundation of, like, hey, we know what the common changes are for the vast majority of prostate cancers. We know some of these changes in the tumor microenvironment. And now, with this foundation, how can we leverage this to find the targetable vulnerabilities?

Andrea Miyahira: OK. Well, thank you so much for sharing this study with us today.

Mindy Graham: Yeah. Thank you for the opportunity.