AR EcDNA Discovery in Advanced Prostate Cancer Study - George Zhao
January 28, 2025
George Zhao joins Andrea Miyahira to discuss research examining three-dimensional genome organization in metastatic prostate cancer. The study integrates multi-omic data including RNA-seq, whole-genome sequencing, and Hi-C sequencing. Their key finding reveals that approximately one-third of samples show disrupted genomic organization around the androgen receptor (AR) gene, which they discover is due to extrachromosomal DNA (EcDNA). This AR EcDNA correlates with reduced response to androgen receptor signaling inhibitors like abiraterone or enzalutamide. The findings revise the understanding of AR dysregulation in castration-resistant prostate cancer and suggest potential new therapeutic approaches, including AR degraders and drugs specifically targeting EcDNA. The study's comprehensive dataset is publicly available for further research exploration.
Biographies:
Shuang (George) Zhao, MD, Physician Scientist, Assistant Professor, Department of Human Oncology, University of Wisconsin School of Medicine and Public Health, Madison, WI
Andrea K. Miyahira, PhD, Director of Global Research & Scientific Communications, The Prostate Cancer Foundation
Biographies:
Shuang (George) Zhao, MD, Physician Scientist, Assistant Professor, Department of Human Oncology, University of Wisconsin School of Medicine and Public Health, Madison, WI
Andrea K. Miyahira, PhD, Director of Global Research & Scientific Communications, The Prostate Cancer Foundation
Read the Full Video Transcript
Andrea Miyahira: Hi, everyone. I'm Andrea. I'm here at the Prostate Cancer Foundation. I'd like to welcome Dr. George Zhao of the University of Wisconsin to discuss his paper Integrated Analysis Highlights Interactions Between the Three-Dimensional Genome and DNA, RNA, and Epigenomic Alterations in Metastatic Prostate Cancer. This was published in Nature Genetics. Dr. Zhao, thanks for joining us.
George Zhao: Thanks so much for having me. I'm excited to tell you all about the paper. So our manuscript really focused on integrating multi-omic data from metastatic prostate cancer samples in order to understand not just DNA, RNA, and epigenomic alterations, but the three-dimensional organization of the genome and how that contributes to prostate cancer growth and progression.
And so to do this, as part of the West Coast Dream Team consortium funded by the Prostate Cancer Foundation and Stand Up to Cancer, this consortium has collected RNA-seq, whole-genome sequencing, whole-genome bisulfite sequencing, and 5-hydroxymethylcytosine sequencing from a large cohort of patients with mCRPC. And from this cohort, we've really developed a much deeper understanding of linear genomics.
So things like methylation, structural variants, histone modifications, as well as transcription and mutations in the gene itself. In this study, we paired the results with public RNA-seq data and public ChIA-PET data, as well as Hi-C sequencing, which is sequencing that looks at the three-dimensional organization of the genome. We wanted to use this to really understand the three-dimensional organization of the genome from large-scale organization in alpha and beta compartments, medium-range interactions between promoters and enhancers, as well as the architecture of the topology of the genome.
Andrea Miyahira: OK. Thank you for this summary. So what were some of the most surprising findings that emerge from combining these different multi-omics data sets?
George Zhao: That's a great question. I think the answer to that is the most important finding that we had in prostate cancer was regarding the region around androgen receptor and what was going on in the genome there. And so this figure illustrates the region around androgen receptor. The rows each represent a single sample, and the columns, going from left to right, show the genomic coordinate with the androgen receptor gene in the middle of that coordinate.
I want to describe what we're looking at a little bit more. So areas of green are areas where the genome is contacting itself at very high levels. You can think of a region of the genome that's condensed and has a lot of interactions with neighboring regions, whereas the regions in blue are areas where the genome does not have much contact with the surrounding regions, indicating an isolated region of the genome.
In the top half of the figure—really, the top two-thirds of the figure—those samples have a pretty regularly organized structure around androgen receptor, with alternating regions of blue and green. However, you can see at the bottom, about a third of samples have this organization disrupted right around the androgen receptor. And we call this the regional contact frequency score, so the degree to which any particular region is contacting the DNA around it.
These samples had lower regional contact frequency scores compared to the typical sample. And so we wanted to investigate a little bit further what was going on in these samples. We first looked at copy number of the androgen receptor itself. It turns out that these samples that had low AR RCFS also had higher AR copy numbers, indicating that amplification was going on.
These samples also have some of the highest gene expression levels of androgen receptor as well, indicated in blue. And we measured this using TPM. We looked at the alpha/beta compartment, and for a lot of these samples, because of the disruption of the normal architecture, it wasn't able to really say which compartment it was in.
What really keyed us in on what was going on was that we ran a tool that could look for extrachromosomal circularized DNA based on the whole-genome sequencing. And many of these samples had AR EcDNA based on this tool. And so that would explain why there is less contact with the surrounding areas of the genome in these samples and androgen receptor: it's because androgen receptor is not in the genome anymore—it's now separated from the linear genome as a circularized region of EcDNA.
And so this figure illustrates it a little bit better. It's comparing two samples. On the top left is a sample that has EcDNA, and on the bottom right is a sample without EcDNA. This is showing the interaction between any two points on the genome. The diagonal represents the genome, with AR and its upstream enhancer indicated by the red and yellow dots, respectively.
What you can see on the bottom is that there's a lot of contact with the different regions of the genome and androgen receptor, as you would expect. But on the top, that is completely absent or very much reduced, especially between the area in blue and the surrounding regions, with a little bit more contact within that blue triangle. And right at the corner, there's an area of increased contact.
This is really what EcDNA looks like in a Hi-C contact map, which is what this plot is showing. And so to confirm this, we took some samples, and we did FISH on it. And what you can see is that in the AR EcDNA-negative samples above, there is some AR but the expected amount of AR. In the EcDNA-positive samples, there is much, much more AR, and it is not just confined in one place on the genome—it is spread throughout the nucleus, indicating that this is, in fact, extrachromosomal DNA.
And so this really allows us to revise our model of AR dysregulation and how it leads to AR overexpression and resistance to therapy. So before all of these West Coast Dream Team sequencing studies, AR was thought to be sitting on chromosome X and driven by the same things that most other genes are driven by.
But through a series of studies, we first found a distant upstream enhancer using whole-genome sequencing, followed by a more complex array of upstream enhancers that we identified using methylation sequencing. And now we've identified that the whole locus is amplified via EcDNA, really driving AR overexpression and treatment resistance in mCRPC. And so this gives us a more complete picture of the biology of AR dysregulation in castration-resistant disease.
It can hopefully lead to new treatment strategies. And so this work was really a large effort by an enormous number of co-authors from all over the world, as well as the West Coast Dream Team consortium itself. And again, all of this work was funded primarily by the Prostate Cancer Foundation and Stand Up to Cancer through the West Coast Dream Team consortium, as well as a number of other funders. I'm happy to take any questions now.
Andrea Miyahira: Thank you so much, Dr. Zhao, for sharing this. So did you see any correlations between the presence of AR EcDNA and any clinical outcomes?
George Zhao: Yes. It turns out that the patients with AR EcDNA had a diminished benefit to androgen receptor signaling inhibitors like abiraterone or enzalutamide compared to patients that did not have AR EcDNA, suggesting that the EcDNA may be playing a role in resistance to not just standard androgen deprivation therapy, but also second-line agents.
Andrea Miyahira: Thank you. And did you see any other oncogenes that we know or we could suspect may be driving prostate cancer that were present on EcDNA?
George Zhao: That is an excellent question and is actually the topic of a manuscript that is coming out of the West Coast Dream Team from Dr. David Quigley's group. That will have a lot more information on that. But the short answer is that there are no other genes that have the level or frequency of EcDNA as androgen receptor, which is perhaps not surprising given the prominence of the gene in prostate cancer.
Andrea Miyahira: OK. Thank you. I'm looking forward to that paper. So can studies like these reveal novel therapeutic strategies or targets?
George Zhao: So that's also a fantastic question. I think the androgen receptor dependence is clearly illustrated by these results—some tumors are so unbelievably dependent on androgen receptor that they need not just 10 copies, but 1,000 copies. The only mechanism by which to achieve that is through EcDNA. And that suggests that further therapies targeting the androgen receptor will also be potentially efficacious.
So androgen receptor degraders are coming into clinical trials and are in clinical trials. I look forward to seeing if those have a preferential effect on patients with AR EcDNA. There are other companies as well that are making drugs specifically targeting EcDNA. And so those will also be an exciting opportunity to see if they provide a clinical benefit in patients with AR EcDNA.
Andrea Miyahira: OK. Thank you. And what advice do you have for researchers that are interested in exploring this rich data set?
George Zhao: This is a multifaceted, enormous resource for the field of prostate cancer research, and all of the data are deposited into various public repositories. I would encourage people to look at it for any molecular questions they have in prostate cancer or questions that we haven't fully explored in our paper, since there was only so much time and space.
Andrea Miyahira: OK, well, thank you so much for leading such incredible studies and sharing this with us today.
George Zhao: Thanks for having me.
Andrea Miyahira: Hi, everyone. I'm Andrea. I'm here at the Prostate Cancer Foundation. I'd like to welcome Dr. George Zhao of the University of Wisconsin to discuss his paper Integrated Analysis Highlights Interactions Between the Three-Dimensional Genome and DNA, RNA, and Epigenomic Alterations in Metastatic Prostate Cancer. This was published in Nature Genetics. Dr. Zhao, thanks for joining us.
George Zhao: Thanks so much for having me. I'm excited to tell you all about the paper. So our manuscript really focused on integrating multi-omic data from metastatic prostate cancer samples in order to understand not just DNA, RNA, and epigenomic alterations, but the three-dimensional organization of the genome and how that contributes to prostate cancer growth and progression.
And so to do this, as part of the West Coast Dream Team consortium funded by the Prostate Cancer Foundation and Stand Up to Cancer, this consortium has collected RNA-seq, whole-genome sequencing, whole-genome bisulfite sequencing, and 5-hydroxymethylcytosine sequencing from a large cohort of patients with mCRPC. And from this cohort, we've really developed a much deeper understanding of linear genomics.
So things like methylation, structural variants, histone modifications, as well as transcription and mutations in the gene itself. In this study, we paired the results with public RNA-seq data and public ChIA-PET data, as well as Hi-C sequencing, which is sequencing that looks at the three-dimensional organization of the genome. We wanted to use this to really understand the three-dimensional organization of the genome from large-scale organization in alpha and beta compartments, medium-range interactions between promoters and enhancers, as well as the architecture of the topology of the genome.
Andrea Miyahira: OK. Thank you for this summary. So what were some of the most surprising findings that emerge from combining these different multi-omics data sets?
George Zhao: That's a great question. I think the answer to that is the most important finding that we had in prostate cancer was regarding the region around androgen receptor and what was going on in the genome there. And so this figure illustrates the region around androgen receptor. The rows each represent a single sample, and the columns, going from left to right, show the genomic coordinate with the androgen receptor gene in the middle of that coordinate.
I want to describe what we're looking at a little bit more. So areas of green are areas where the genome is contacting itself at very high levels. You can think of a region of the genome that's condensed and has a lot of interactions with neighboring regions, whereas the regions in blue are areas where the genome does not have much contact with the surrounding regions, indicating an isolated region of the genome.
In the top half of the figure—really, the top two-thirds of the figure—those samples have a pretty regularly organized structure around androgen receptor, with alternating regions of blue and green. However, you can see at the bottom, about a third of samples have this organization disrupted right around the androgen receptor. And we call this the regional contact frequency score, so the degree to which any particular region is contacting the DNA around it.
These samples had lower regional contact frequency scores compared to the typical sample. And so we wanted to investigate a little bit further what was going on in these samples. We first looked at copy number of the androgen receptor itself. It turns out that these samples that had low AR RCFS also had higher AR copy numbers, indicating that amplification was going on.
These samples also have some of the highest gene expression levels of androgen receptor as well, indicated in blue. And we measured this using TPM. We looked at the alpha/beta compartment, and for a lot of these samples, because of the disruption of the normal architecture, it wasn't able to really say which compartment it was in.
What really keyed us in on what was going on was that we ran a tool that could look for extrachromosomal circularized DNA based on the whole-genome sequencing. And many of these samples had AR EcDNA based on this tool. And so that would explain why there is less contact with the surrounding areas of the genome in these samples and androgen receptor: it's because androgen receptor is not in the genome anymore—it's now separated from the linear genome as a circularized region of EcDNA.
And so this figure illustrates it a little bit better. It's comparing two samples. On the top left is a sample that has EcDNA, and on the bottom right is a sample without EcDNA. This is showing the interaction between any two points on the genome. The diagonal represents the genome, with AR and its upstream enhancer indicated by the red and yellow dots, respectively.
What you can see on the bottom is that there's a lot of contact with the different regions of the genome and androgen receptor, as you would expect. But on the top, that is completely absent or very much reduced, especially between the area in blue and the surrounding regions, with a little bit more contact within that blue triangle. And right at the corner, there's an area of increased contact.
This is really what EcDNA looks like in a Hi-C contact map, which is what this plot is showing. And so to confirm this, we took some samples, and we did FISH on it. And what you can see is that in the AR EcDNA-negative samples above, there is some AR but the expected amount of AR. In the EcDNA-positive samples, there is much, much more AR, and it is not just confined in one place on the genome—it is spread throughout the nucleus, indicating that this is, in fact, extrachromosomal DNA.
And so this really allows us to revise our model of AR dysregulation and how it leads to AR overexpression and resistance to therapy. So before all of these West Coast Dream Team sequencing studies, AR was thought to be sitting on chromosome X and driven by the same things that most other genes are driven by.
But through a series of studies, we first found a distant upstream enhancer using whole-genome sequencing, followed by a more complex array of upstream enhancers that we identified using methylation sequencing. And now we've identified that the whole locus is amplified via EcDNA, really driving AR overexpression and treatment resistance in mCRPC. And so this gives us a more complete picture of the biology of AR dysregulation in castration-resistant disease.
It can hopefully lead to new treatment strategies. And so this work was really a large effort by an enormous number of co-authors from all over the world, as well as the West Coast Dream Team consortium itself. And again, all of this work was funded primarily by the Prostate Cancer Foundation and Stand Up to Cancer through the West Coast Dream Team consortium, as well as a number of other funders. I'm happy to take any questions now.
Andrea Miyahira: Thank you so much, Dr. Zhao, for sharing this. So did you see any correlations between the presence of AR EcDNA and any clinical outcomes?
George Zhao: Yes. It turns out that the patients with AR EcDNA had a diminished benefit to androgen receptor signaling inhibitors like abiraterone or enzalutamide compared to patients that did not have AR EcDNA, suggesting that the EcDNA may be playing a role in resistance to not just standard androgen deprivation therapy, but also second-line agents.
Andrea Miyahira: Thank you. And did you see any other oncogenes that we know or we could suspect may be driving prostate cancer that were present on EcDNA?
George Zhao: That is an excellent question and is actually the topic of a manuscript that is coming out of the West Coast Dream Team from Dr. David Quigley's group. That will have a lot more information on that. But the short answer is that there are no other genes that have the level or frequency of EcDNA as androgen receptor, which is perhaps not surprising given the prominence of the gene in prostate cancer.
Andrea Miyahira: OK. Thank you. I'm looking forward to that paper. So can studies like these reveal novel therapeutic strategies or targets?
George Zhao: So that's also a fantastic question. I think the androgen receptor dependence is clearly illustrated by these results—some tumors are so unbelievably dependent on androgen receptor that they need not just 10 copies, but 1,000 copies. The only mechanism by which to achieve that is through EcDNA. And that suggests that further therapies targeting the androgen receptor will also be potentially efficacious.
So androgen receptor degraders are coming into clinical trials and are in clinical trials. I look forward to seeing if those have a preferential effect on patients with AR EcDNA. There are other companies as well that are making drugs specifically targeting EcDNA. And so those will also be an exciting opportunity to see if they provide a clinical benefit in patients with AR EcDNA.
Andrea Miyahira: OK. Thank you. And what advice do you have for researchers that are interested in exploring this rich data set?
George Zhao: This is a multifaceted, enormous resource for the field of prostate cancer research, and all of the data are deposited into various public repositories. I would encourage people to look at it for any molecular questions they have in prostate cancer or questions that we haven't fully explored in our paper, since there was only so much time and space.
Andrea Miyahira: OK, well, thank you so much for leading such incredible studies and sharing this with us today.
George Zhao: Thanks for having me.