IMPaCT Plenary Session: Game-Changing Research Video (Text Version) - Dr. Scott Dehm

AR Intragenic Rearrangement and Altered mRNA Splicing in Castration-Resistant Prostate Cancer

Natasha Kyprianou
Our next speaker, Dr. Scott Dehm, is an Assistant Professor of Laboratory Medicine and Pathology at the University of Minnesota in Twin Cities. Dr. Dehm is a Department of Defense 2009 Prostate Cancer Research Program New Investigator Award Recipient with the goal of developing new androgen receptor-targeted therapies. Today, he will be presenting his exciting work on alterations in androgenic receptor messenger RNA splicing and castration-resistant prostate cancer; Dr. Dehm.

Scott Dehm, Ph.D.; Assistant Professor of Laboratory Medicine and Pathology, University of Minnesota, Twin Cities
Thank you Natasha; thank you Westley for the opportunity to come tell you today about some of my lab’s work which is supported by a New Investigator Award from the Department of Defense Prostate Cancer Research Program. Today I’m going to be talking about some of our work on androgen receptor splicing as it pertains to prostate cancer progression.

So I’m hopeful that this slide will serve as sufficient introduction to this topic on androgen receptor and prostate cancer. This is a functional schematic of the modular androgen receptor protein. And all of the therapies we have targeted to the androgen receptor for systemic prostate cancer inhibit the function of the androgen receptor ligand-binding domain, and this is the part of the protein to which the androgens testosterone and dihydride testosterone bind as well as anti-androgens such as bicalutamide. This ligand-binding domain in the c-terminus of the receptor is encoded by androgen receptor exons four through eight in the androgen receptor gene. The androgen receptor also has a central DNA binding domain encoded by exons two and three and the entire n-terminus of the androgen receptor, which functions as a potent transcriptional activator, is entirely encoded by exon one in the androgen receptor domain. And it’s well-established that one of the challenges in prostate cancer is the fact that the androgen receptor can remain active despite therapies targeted and designed to inhibit this protein. And so that’s the focus of my lab’s work is to try to understand how the androgen receptor remains active in advanced stages of the disease.

So when I was a postdoc in Don Tindall’s lab I made the serendipitous discovery that the androgen receptor could be alternatively spliced giving rise to truncated constitutively active isoforms. The way we stumbled upon this discovery was we were studying 22RV1 cells which we knew from previous studies displayed very high levels of constitutive androgen receptor activity in an androgen-independent fashion. And we were using siRNAs targeted to the androgen receptor and it just so happened that these siRNAs were targeted to different regions of the androgen receptor message. And we were able to identify that an siRNA targeted to exon seven of the androgen receptor abolished this full-length androgen receptor isoform but there was a smaller isoform of the androgen receptor of about 75 or 80 kilodaltons which had previously been thought to be a proteolytic degradation fragment of the receptor. But upon usage of an siRNA targeted to androgen receptor exon one, we were able to ablate all isoforms of the androgen receptor. And this really suggested to us that these two different sized proteins in these cells were encoded by separate messenger RNAs. This was important because we were able to identify the fact that the full-length receptor was modulating all the androgen-sensitive characteristics of this cell line, whereas these smaller isoforms were modulating all the important androgen independent characteristics of the cell line including the ability of the androgen receptor to activate genes in an androgen-independent fashion as well as the ability of these cells to grow in the absence of androgens.

So I’m going to fast forward to present day; this is a schematic now about what we know regarding androgen receptor alternative splicing in prostate cancer progression. So we know that there are cryptic exons highlighted in red here within the androgen receptor gene that can get incorporated into the message ultimately giving rise to truncated receptors that lack the ligand-binding domain. So one commonality of all of these different exons that can get incorporated into the androgen receptor messages that they have premature translation stop codons, and these receptors are missing the ligand-binding domain and are constitutively active, and various flavors of these truncated androgen receptors have been identified to date by various groups but ultimately they share the core structure of containing the entire androgen receptor n-terminal domain, the androgen receptor DNA binding domain, and in short variable length c-terminal extensions. And, of course, these androgen receptors would be completely resistant to therapies we have targeted to the androgen receptor because they’re missing the part of the protein to which drugs bind.

So the clinical relevance for this story I think has really been highlighted by two important studies that were published after ours which identified an isoform of the androgen receptor made by synthesis or splicing of androgen receptors exons one, two, three, and a cryptic exon term CE3. And this isoform of the androgen receptor has been given various names, ARV7 in one study, AR3 in another study but this c-terminal extension on this novel receptor has been exploited to develop a targeted antibody and this antibody has been used to show that these truncated receptors are indeed expressed in clinical prostate cancer. They increase in their expression from progression from androgen dependent to castration resistant disease and their high-level expression can also predict biochemical relapse following surgery.

So we’ve been interested in trying to understand how these increases in expression of these truncated receptors may arise and we’ve turned to the CWR22 model of prostate cancer progression to try to understand this in more detail. 22RV1 cells was a cell line in which truncated receptors were originally identified and recently the CWR22 PC cell line was developed from this CWR22 xenograft tumor, and these cells are androgen dependent. And so what we have here are isogenic matched pairs of androgen-dependent castration-resistant cell lines that we might be able to exploit to understand these differences in phenotype. And you can see here that these 22PC cells are completely androgen dependent; they need androgens to grow. But the 22RV1 cell line can proliferate in an androgen-independent environment.

One striking difference between these two cell lines is the fact that both of them express the full-length androgen receptor at approximately equal levels but the androgen independent sub-line CWR 22RV1 express very high levels at the mRNA level of these truncated androgen receptor mRNAs and at the protein level very high levels of truncated androgen receptor protein. I want to point out though that the 22PC cell line, which is androgen dependent, does express this V7 or AR3 isoform and it is also detectable on western blot with the antibody specific for this domain. And so this really illustrates that the mere presence of truncated isoforms in a cell line does not perfectly correlate with its androgen responsiveness; rather we think it’s this increase in expression which would tip the scale in favor of gene expression modulated by these receptors that’s important.

So another striking feature of these cells is the fact that there appears to be a rearrangement within the androgen receptor gene. We know from our studies that these cells have the ability to make mRNA consisting of AR exons one, two, three, and two-b and the only way a messenger RNA like that could be synthesized is if there was some reconfiguration of the androgen receptor gene allowing exon two-b to be spliced downstream of exon three. In addition, all of the alternative exons that have been identified to date in these cells are tightly clustered around androgen receptor exon three. So we queried gene copy number along the length of the androgen receptor on the x-chromosome and you can see that the androgen-dependent cells contain one normal copy but there was an increase signal up to two copies in this region surrounding AR exon three and alternative exons. We were excited by this finding thinking that this might be a mechanism by which efficient synthesis of these receptors could occur.

So the question at this point is do rearrangements in the androgen receptor gene occur in clinical prostate cancer? This was challenging because most castration-resistant prostate cancers also exhibit high-level amplification of the androgen receptor gene. A recent genome-wide Affymetrix SNP (snip) six profiling study on metastatic prostate cancers isolated from 14 patients published by Steve Bova at Johns Hopkins we thought we could exploit this data to potentially identify whether rearrangements were occurring within the androgen receptor gene in these cohorts. So we developed a breakpoint finding algorithm to take probe level copy number data, so those blue dots represent the copy number of individual probe sets on these Affymetrix SNP (snip) six arrays and the red lines through them represent segments called by a segmentation algorithm that we developed. And what you can see first and foremost, this is log-base two scale on the left, the androgen receptor is amplified in this particular tumor, but you can see that there’s a much higher signal for a segment encompassing androgen receptor exon three compared to neighboring regions, and it appears to be a one to two to one relationship which is similar to what we observed in CWR 22RV1 cells.

We were able to do this segmentation on 44 primary prostate cancers as well as all 58 tumors from 14 patients with castration-resistant disease and as expected in the androgen-dependent tumors, androgen receptor copy number was one and there was no evidence for any alterations within the gene. But in a subset of castration-resistant prostate cancer patients, there appeared to be a higher signal of this region encompassing androgen receptor exon three compared to a segment encompassing androgen receptor exon four. What was really interesting is that multiple tumors within a single patient had the same profile. We were able to do our targeted copy number analysis on two tumors, on DNA from two tumors isolated from patient 16 to confirm that indeed there are rearrangements occurring with the androgen receptor gene in conjunction with amplification in these tumors.

It’s also important to point out that this breakpoint finding algorithm also suggests additional breakpoints within the androgen receptor gene. We have always thought of AR as being amplified in castration-resistant disease; this provides evidence that the ARG might be amplified and rearranged. To get some functional significance for this, we took advantage of the fact that these androgen-dependent cells will eventually assume a castration-resistant growth phenotype with long-term castration, so over 10 days they don’t grow in the absence of androgens, but eventually over a month they will assume an androgen-independent growth phenotype. Importantly, they start to express high levels of truncated androgen receptors and we can proliferative foci emerging at approximately Day 20. We mapped the breakpoints in 22RV1 cells precisely and were able to develop an outward facing PCR assay to specifically identify the breakpoints. In the context of a normal locus, PCR primers are facing away from each other and will not generate a product. But in the context of the rearrangement identified in RV1 cells, these primers are facing each other and will develop—and will generate a PCR product.

We were able to show that the emergence of proliferative foci, the emergence of a signature for this breakpoint in RV1 cells and the emergence of expression of truncated isoforms are all linked in this model of prostate cancer progression. It’s important to note that in this androgen-dependent population we were unable to identify any evidence for this rearrangement. It suggests that there is a rare subset of cells in this otherwise androgen-dependent population that can progress to an androgen-independent phenotype. I am not aware of any other models of progression where a genomic marker of the castration-resistant phenotype is known a priority.

So to summarize, we now have a story where truncated androgen receptors can be synthesized that are missing the ligand-binding domain and are expected to not respond to therapies targeted to that part of the androgen receptor. We also put forth this new concept of an amplified and rearranged androgen receptor gene and I think there’s addition evidence for this concept provided in a recent paper from Charles Sawyers’ lab where they identified a novel androgen receptor mRNA that can arise from splicing of AR exons one, two, three, and four and a novel exon located about 1,000,000 base pairs upstream of the amplified androgen receptor locus in a cell line derived from the MYC transgenic mouse. So this is another situation where you get splicing of exons that are not configured normally in the context of amplification of the androgen receptor.

I would like to acknowledge Yingmin Li in my lab who performed the majority of these studies as well as members of our Bioinformatics Corps and our Electrical and Computer Engineering Department at the University of Minnesota that helped with breakpoint analysis. I’d like to thank collaborators for providing reagents for these studies and I’d also like to thank Don Tindall, who is was my postdoctoral mentor when I made this initial serendipitous discovery that androgen receptor can be alternatively spliced in prostate cancer. Thank you.