Read the Full Video Transcript

Andrea Miyahira: Hi everyone. I'm Andrea Miyahira at the Prostate Cancer Foundation. With me today is Dr. Andrew Goldstein, an associate professor at UCLA. His group recently published the paper "Prostate Lineage-Specific Metabolism Governs Luminal Differentiation and Response to Anti-Androgen Treatment" in Nature Cell Biology. Dr. Goldstein, thank you for sharing this work with us today.

Andrew Goldstein: Thank you for having me. Well, I'm very excited to discuss our recent paper led by two outstanding grad students from my lab, Jenna Giafaglione and Preston Crowell. And we got a lot of support from collaborators at UCLA and actually all around the world.

The normal prostate epithelium is made up primarily of basal and luminal cells. During cancer initiation, there's an expansion of malignant luminal-like cells that express the androgen receptor. After AR-targeted therapies, some resistant tumors emerge with a loss of luminal differentiation, and this is often referred to as lineage infidelity or lineage plasticity.

Luminal differentiation occurs in prostate development, in response to tissue repair or inflammation, or during cancer initiation from basal cells. Whereas, the loss of luminal differentiation can occur during disease progression, in response to AR-targeted therapies, or as a result of gaining specific mutations. But two really important questions are: Does the differentiation state really matter for prostate cancer, and, what regulates prostate differentiation?

Results from many studies are starting to show that lineage phenotype really does matter. Here, as an example, a set of metastatic castration-resistant prostate cancers were classified into basal or luminal subtypes based on gene expression. And the basal-like tumors have low AR signaling, and importantly, a poor clinical response to AR-targeted therapies. So, differentiation state in prostate cancer definitely matters because it's associated with response to therapy and patient outcome. And specifically, the more luminal-like tumors are more responsive to AR blockade.

And so it's critical to understand what makes a prostate cell more or less luminal? Prior studies point to epigenetic regulation of prostate lineage identity, and so we wanted to understand if upstream metabolism might contribute to prostate luminal differentiation. We performed metabolic profiling and nutrient tracing in distinct prostate cell types. We found evidence of lineage-specific metabolic features, including terms like pyruvate metabolism and oxidative phosphorylation enriched in luminal cells. So, to summarize a lot of hard work and some really beautiful data, we found evidence that distinct cell types have distinct metabolic preferences in the prostate epithelium.

And so, if the cell type dictates the metabolism, then we wanted to know if metabolism would change during differentiation. So, one of the models we used was a model of basal to luminal differentiation in ex vivo culture. We see clearly the luminal marker K8 goes up and the basal marker P63 goes down. So, we asked, during this process, how do labeled glucose carbons get incorporated into glycolytic metabolites, TCA cycle metabolites, or nucleotide intermediates? And we found very clear evidence that as luminal differentiation increases, we see increased labeling from glucose into TCA cycle intermediates. And this is indicative of a process called pyruvate oxidation.

So, pyruvate oxidation goes up as cells become more luminal. We wondered if this is just happening or if it might functionally be contributing to luminal differentiation. Pyruvate oxidation is driven by a complex called the mitochondrial pyruvate carrier or MPC. And this can be blocked pharmacologically with a small molecule called UK5099 or through genetic approaches to delete MPC1 or MPC2. And we found that MPC inhibition or MPC knockout led to reduced expression of luminal markers like K8 and increased expression of basal markers like P63. We did this in a wide range of organoid, PDX, and cell line models.

So, blockade of pyruvate oxidation impairs luminal differentiation, which we thought was very cool, but we wanted to understand what's the mechanism? How does pyruvate oxidation influence differentiation?

Now, when pyruvate is blocked from entering the mitochondria, more of it becomes available to be converted to lactate, and we found more lactate in cells after MPC inhibition. And surprisingly, just supplementing cells with lactate was sufficient to also reduce K8 and increase P63. When we did nutrient tracing, we found that lactate supplementation and MPC inhibition actually had very different effects on metabolism. And so, we wondered if perhaps the phenotype might be driven by epigenetic effects. Studies have suggested that lactate can act as an inhibitor of histone deacetylase activity, and so we used ATAC sequencing and we found that MPC inhibition or lactate supplementation increased chromatin accessibility here at the promoter of P63, as well as in the vast majority of basal lineage genes. And so, intracellular lactate accumulation seems to regulate lineage-specific gene expression, at least in part, through changes in chromatin accessibility.

And since lactate influences luminal differentiation, we wondered if it also influences the response to AR inhibition. In studies where clinical tumors were sequenced and then patients were classified as exceptional responders or non-responders to AR-targeted therapies, we found that MPC expression was lower in the non-responders. And this was true in studies from distinct stages of the disease, suggesting that there's a strong correlation between poor response to hormone therapy and low pyruvate oxidation. But again, is this just a correlation or is this actually a driver? And so, we wanted to test this functionally. Using a PDX model, we blocked pyruvate oxidation or supplemented with lactate and then subjected cells to AR inhibition. And while control cells were quite sensitive to Enzalutamide, modulating lactate metabolism increased resistance to AR inhibition.

So, in clinical tumors, good responders to AR inhibition are more luminal and have higher MPC expression. Poor responders are less luminal and have low MPC expression. And experimentally, increasing lactate was sufficient to promote resistance to Enzalutamide.

So, to summarize, basal and luminal cells have lineage-rooted metabolic features, pyruvate oxidation increases with luminal differentiation, and reducing pyruvate oxidation experimentally or increasing intracellular lactate abundance is sufficient to impair luminal differentiation, alter chromatin accessibility at lineage-specific genes, and promote resistance to AR inhibition.

Taking a big-picture view, I think this really argues that metabolism is an important modulator of response to therapy in prostate cancer, and so we really need to consider the consequences of our treatments as well as the effects of systemic or local signals on metabolism.

I want to finish with this great analogy from an accompanying News & Views written by Martin Bakht and Misha Beltran. We, of course, care about the clinical response to treatment, and we know that lineage can influence the response. Underneath the surface is, of course, epigenetics, but even less understood, we think, just as important, is the metabolism. And so, I will finish just thanking the authors once again, as well as the funding support. It was really fun to get to do this work. It took many years and a lot of hard work, but we're really proud of what we've now been able to share, and we hope it's impactful. Thank you.

Andrea Miyahira: Well, thank you so much for sharing this really wonderful study. So, do you have a model for how prostate basal and luminal cells, they're proximal to each other, but how do they maintain different metabolic identities?

Andrew Goldstein: I think it's a really important question. It kind of fascinates us that these two cells sit next to each other and yet can have completely different metabolic features. I didn't show in this short presentation, but we found some really interesting ways that, for example, carbons are shuttled differently within the TCA cycle in basal cells or luminal cells. So, it suggests that what you are does really help you determine how you break down nutrients, which I think is fascinating. How that's maintained in vivo, that's a really tough thing to answer. We had to take the cells out of the prostate in order to do these assays. And so, I can't really comment on things like the polarity or the spatial distribution, but I do think that the lineage-specific transcription factors have a major role in dictating these metabolic features.

Of course, we know the androgen receptor is a major regulator of all things prostate, but particularly in the luminal cells, we do think it's influencing the metabolism as well as other metabolic regulators, perhaps regulators of lipid metabolism in the luminal cells. In the basal cells, we think there's more of a mixed signature going on, which is more of a glycolytic driver, but there's plenty more to be figured out, and I think some of it needs to be done more in a structurally intact context, which is a tough thing with metabolomics at this point in time. But the technology just keeps getting better.

Andrea Miyahira: Do you have any more insights into the mechanisms by which AR interacts with a tumor metabolism axis? And I guess I have a chicken versus egg question. Do you think AR inhibition upregulates basal metabolism genes, or is it the vice versa where metabolic alterations drive AR therapy resistance?

Andrew Goldstein: Yeah, I think this is a good question, and it interacts with the other paper that we recently published, which we've also spoken about. Clearly, AR is majorly regulating metabolism in prostate cells. I think normal luminal cells and prostate cancer cells are kind of in a different scenario. So, normal luminal cells are definitely driven by AR. We think that's likely affecting the metabolism. And as I just mentioned, I think MIC is more active in the basal cells. In prostate cancer, AR and MIC are both quite active, and that certainly regulates metabolism.

But in terms of the effect of AR inhibition, does it recapitulate a basal lineage metabolism? The initial response to AR inhibition is not pushing them toward a basal lineage. But, if you drive it toward a truly resistant phenotype where AR is suppressed, but there's, for example, drivers of a more plasticity phenotype, we do see that more of an acquisition of a basal-like metabolic program. So, presumably there's some relationship between if you have low AR but you're highly proliferative, that's more of a basal metabolism phenotype. If you have low AR, but you're poorly proliferative, that's sort of a less metabolically active state generally.

And then maybe just to hint on the chicken or egg thing, I do think that the major driver of the phenotype is going to be these transcription factors, these programs. But, I do think that our study really shows that modulating metabolism can dramatically influence the phenotype. And so, I think this probably explains a subset, but certainly not all, forms of resistance. And when we see different niches, I think in vivo having these different responses, almost certainly the local metabolism is going to be influencing this.

Andrea Miyahira: Okay. And does this relate to hypoxia, and if so, how?

Andrew Goldstein: I think there's evidence in prostate cancer in the field that hypoxia is associated with plasticity. And hypoxia certainly would be a scenario where you'd increase intracellular lactate abundance. So, it's very likely that this, to some degree, is influencing that plasticity phenotype in vivo, but it's really hard to measure in culture. We've done some organoid experiments to see how hypoxia influences the lineage and the metabolism, but what we see is that hypoxia generally slows down the growth, and when you slow down the growth, you're influencing the metabolism, so that makes it kind of tricky to study in culture. But I do think that in addition to likely a role for HIFs in plasticity, I think probably the lactate abundance is contributing.

Andrea Miyahira: Okay. And what are your next steps? Do you have any translational plans for these studies?

Andrew Goldstein: Yeah, we're kind of thinking about this in a couple of ways. One result of this study is showing that modulating lactate can influence differentiation. So, we're thinking about ways to kind of reverse the step. If we can deprive the cells of lactate, might that be a driver of differentiation? Technically, that's been extremely difficult to do. We've tried a lot of different genetic and pharmacological approaches and haven't quite gotten there yet. So, to be determined, but that could be an interesting approach.

And then we're also trying to apply the principles of this study to other stages of the disease. So, like I mentioned, we're really in this study showing the effect of basal to luminal differentiation, what we've learned about that, and how we can disrupt that. If we can go to a different stage where the cells are losing luminal differentiation, how does the metabolism contribute to the loss of luminal differentiation? Can we block that and keep them more luminal? So, that's kind of how we've been thinking also. And we don't have any functional data I can talk about yet, but there's some promising leads.

Andrea Miyahira: Okay. Well, thanks again for coming on and sharing this really wonderful study with everybody.

Andrew Goldstein: Thanks again for having me. It's an exciting time to get to share the work and to get to share with you. Thank you, Andrea.