MYC Family Matters
MYC Family Matters
Myles Brown (Center for Functional Cancer Epigenetics, Dana-Farber Cancer Institute)
View the Transcript Below:
MYC Family Matters
Himisha Beltran, MD [00:00:13] Hello, everyone. It is my great pleasure to introduce my colleague and friend, Dr. Myles Brown. Dr. Brown is the Emil Frei III Professor of Medicine at Dana Farber Cancer Institute in Harvard Medical School, where he also directs our Center for Functional Cancer Epigenetics. Dr. Brown’s pioneering work has really shaped the field’s understanding of hormone-dependent cancers, including the function and regulation of the AR, ER, and their co regulators in both prostate and breast cancer. Today he’ll be talking about MYC family members and prostate cancer progression based on exciting new data from his lab. Please join me in welcoming Dr. Brown.
Myles Brown, MD [00:01:05] Well, thanks, Misha. It’s really a great pleasure to be back at the PCF retreat again. So, what I thought I’d do today actually is continue the theme that Michael started really trying to understand regulators of some of the heterogeneity and different types of CRPC and CRPC to neuroendocrine to NEPC transition in prostate cancer. So, these are my disclosures. So, several years ago now, our center started to look at a variety of different tumor types and found an epigenetic convergence of diverse neuroendocrine cancers. And by that, I mean we profiled by ATAC-seq and RNA-seq, neuroendocrine models, NEPC, small cell lung cancer models, Merkel cell carcinoma. In fact, this was the title of a new program project that Himisha’s part of as well that we got funded late this year, only actually when we agreed to take the word diverse out of the title. So, but what that shows is that what we found was that NEPC, for example, when we look either by RNA-seq or ATAC-seq resembles neuroendocrine cancers like Merkel cell carcinoma or small cell carcinoma are more than they resemble prostate adenocarcinoma. And if you try to look at what’s driving that from the open chromatin aspect, you can see that there are sets of common open regions of the genome that are in neuroendocrine CRPC that are common with small cell lung cancer, common with Merkel cell carcinoma. And if you explore what are those sites using either motif analysis or using tools, we have developed in our Cistrome toolkit to predict what factors from existing ChIP-seq data or existing ATAC-seq data might be regulating. What you see are both canonical neuroendocrine transcription factors like ASCL1, but you also see changes in MYC. And in fact, the changes in MYC are actually down. That is, you see in NEPC a decrease of overlap with existing MYC ChIP-seq data sets. So, as I’m sure you all know, MYC is an oncogenic transcription factor. In fact, there are three family members c-MYC or MYC, MCYN or N-MIC, and L-MYC. And What we were interested in trying to understand was whether there were changes in MYC members in the progression from prostate adenocarcinoma to NEPC. So prior work for Himisha had already shown us that N-MYC plays an important role in a subset of neuroendocrine cancers, and we wanted to explore the impact of L-MYC. So work for my lab in the high-MYC, high c-MYC model, mouse model, this is work that was done in collaboration with David Labei, who was a former postdoc, but his lab and our center, asking is there some type of functional interplay between c-MYC and androgen receptor in the high-MYC mouse model, sort of a model of AR positive CRPC. And what we found was that c-MYC inhibited AR activity, but not by inhibiting AR occupancy on the genome, but rather at a subsequent step of release of polymerase from proximal pause sites. So MYC, c-MYC actually inhibits in some ways AR activity, but not its ability to define the genome. And what that suggested to us is that high-MYC, so c-MYC is often amplified in prostate adenocarcinoma, that it’s diverting AR’s activity and potentially some of the other co-regulators from more normal AR functions. But if you look across prostate adenocarcinoma and NEPC models, either in the LuCaP series or in Himisha’s clinical data, what you find is that there is upregulation of L-MYC in a subset of NEPCs. In addition, we collaborated with Michael Haffner who examined the UW rapid autopsy series of CRPC, and these are divided into AR low NE negative, AR low NE positive, AR positive NE negative, AR negative NE positive. And we looked particularly at the three MYC family members, at AR itself, and then important transcription factors for both neuroendocrine differentiation and prostate differentiation NKX3-1 and then a series of neuroendocrine transcription regulators. And what Michael found was that particularly in the AR negative neuroendocrine positive subset of cases both MYC-N and L-MYC were upregulated compared to the other types. So, as we heard discussed in this the last talk, the number of models for NEPC are really limiting. And to me, that’s one of the powers and one of the reasons to have organoids and PDXs is if you don’t have cell lines, that’s a mechanism to study these different phenotypes. But there is one well-characterized NEPC cell line, the NCI-H660. And while DepMap wasn’t able to profile it using CRISPR, if you go back and look at the RNAi screens, which actually with the recurrent tools are actually well analyzed and quite reliable, what you see is that this NCI-H660 cell line preferentially expresses L-MYC, so that’s actually a Western blot we did, showing that it doesn’t express c-MYC or N-MYC, and then the sort of canonical prostate adenocarcinoma cell lines express c-MYC. But if you look at dependencies using in the RNAi data, you see that NCI-H660 is uniquely dependent on L-MYC among the prostate cancer cell lines. And so, if you look just at what cell lines, what prostate lines were sensitive, that is required, MYCL, you see NCI-H660 as an outlier. And if you compare it to LNCaP, so LNCaP, AR HoxB13. MYCS, c-MYC and MACs turn out to be key dependencies, whereas in NCI-H660, SOX2, FOXA2, MYCL are dependencies. And if you ask what are the top dependencies in NCI-H660, what you find is actually MYC targets. But we know it’s not c-MYC because it doesn’t express c-MYC, it’s L-MYC targets. So, we wanted to ask whether if we express L-MYC in prostate adenocarcinoma models, so either castration-sensitive models like LNCaP and VCaP, or in CRPC models, particularly LNCaP ABL, LNCaP95, and then most of the studies we did were using LNCaP cell line that has p53 and Rb knockout generated from Pete Nelson’s lab. And what we see was that if you upregulate L-MYC in the prostate adenocarcinoma cell lines, nothing much happens to c-MYC or AR expression, it doesn’t really change. But if you express it in the CRPC cell lines or in the double knockout cell line, what you see is that both AR and c-MYC are significantly downregulated. So, there’s some, and this is relatively rapid, we can see this within three days, there’s downregulation of c-MYC and AR. So, we ask, well, does that actually affect the ability of MYCL to induce resistance to enzalutamide? So, what we found was that overexpression of L-MYC in LNCaP didn’t do anything, had no effect on enzalutamide resistance. And that’s in contrast to what would happen if you expressed either N-MYC or c-MYC in LNCaP. But in contrast, if you do this experiment in the double knockout, what you see is that you get resistance to enzalutamide with L-MYC expression only in the double knockout. That’s probably not that surprising since I showed you in the double knockout, AR is downregulated. So, we wanted to understand how this is working, so we looked at the genes that were differentially expressed in either LNCaP or the LNCaP double knockout when we overexpress MYCL. So, in the wild type LNCaP or in LNCaPs, I’m not showing here with either a single Rb or single p53 knockout, nothing happens. You overexpress MYCL, and the only gene you see upregulated is MYCL. Has no effect on the gene expression at all. In contrast, if you do look at the double knockouts or the other CRPC cell lines, forced expression of MYCL upregulates a set of genes, including genes that have terms like neurogenesis and generation of neurons. So, sort of neural and neuroendocrine genes is upregulated by MYCL, but only in the setting of Rb and p53 knockout. And as I mentioned already that if you express overexpressed MYC or MYCN in LNCaPs, that’s sufficient to drive enzalutamide resistance in LNCaPs, but it’s not, but if you express MYCL, it’s not. So, it really suggests really specific functions of these different MYC family members at different stages of prostate cancer progression. And we sort of profile the transcriptomes again of MYC, MYCN, MYCL in LNCaP, p53 Rb double knockouts. And what you see is that while MYC and N-MYC do change gene expression in the double knockout, it’s a different set of genes than are regulated by MYCL. And I think that suggests that, for example, the MYCN positive neuroendocrine cancers will have a different transcriptome than the MYCL cases. And again, this is just a heat map showing that again, if you can compare the MYCN and MYCL gene expression changes in LNCaP Rb p53 double knockouts, again you see upregulation of genes with terms associated with neurogenesis and neural development. So, how does that work? What’s driving that? And so, we performed ATAC-seq to look at the accessible chromatin. And what you can see is that in the double knockout cells, in contrast to its effects in LNCaP, where it really doesn’t change accessibility, you get a big increase in accessible chromatin when you express MYCL in p53 Rb double knockouts. And again, you get genes that are neuroendocrine or neuronal in nature. Not surprisingly, perhaps, that if you look at what happens to the sites that are lost in the MYCL overexpression, what you find are enhancers of the AR gene itself. So presumably, at least part of the way it’s down regulating AR expression is through closing the chromatin at the AR enhancers. So, we wanted to know whether if cells become neuroendocrine or neuroendocrine like through MYCL overexpression, do they continue to require MYCL, become addicted to it? And what we found was that if you over express MYCL and culture these cells for a long time in the presence of MYCL, removing MYCL leads to their death. So, they are unable to grow once you have grown them in the presence of high MYCL expression. Removal of MYCL leads to regression. So that suggests that they’re oncogene addicted to MYCL, and I guess we’re going to hear sometime at this meeting from Arul about MYC inhibitors and might suggest that perhaps in this setting a MYCL selective inhibitor or degrader would be valuable. So, I just want to summarize what I showed you. I didn’t show you that, but MYCL is known to be amplified and over expressed in a subset of NEPC. MCYL suppresses MYC, c-MYC, and AR expression only in CRPS upon the loss of p53 and Rb in castration-sensitive prostate cancer. And in contrast to MYC or MYCN, MYCL only drives proliferation in the setting of either CRPC or loss of p53 and Rb. And then that genes it’s regulating actually are part of a neuroendocrine program. And that MYCL expression in CRPC is sufficient to lead to enzalutamide resistance, again, partly, I guess, through down regulation of AR. And that the NEPC cell line expressing MYCL are oncogene addicted. And in similar to what’s seen in small cell lung cancer, you know, specific subtypes of NEPC may be driven by different groups of transcription factors, including different MYC family members, and they’re likely to have distinct therapeutic vulnerabilities. We’ve heard about some of them already at this meeting. But I think it’s gonna be important for us to consider you know these different subtypes and whether the MYC family members are a major determinant of subtype. So most importantly, I want to thank the people who did this work. So, Nicole Traphagan Ho is the postdoc in my lab who led this project together with junior faculty member Alok Tewari, and she had a lot of help from her technician Esmé Wheeler, who’s now gone off to medical school. This is a collaboration with Himsha’s lab and Balaji. I didn’t show you, but we also have data looking at the interplay between MYCL and EZH2 and EZH2 inhibitors in this model. And then our Center for Functional Cancer Epigenetics, Paloma Cejas and Henry Long, Xintao Qiu, Rong Li, played important roles in analysis, and we had great collaborators at the Fred Hutch, Michael Haffner and Pete Nelson. And as I mentioned, this is supported now by NCI program project on convergence of neuroendocrine cancers but not diverse ones. So, I’d be happy to take questions. Thanks.
Unknown [00:17:39] Hi, over here. Over here. Hi, nice to see you. Beautiful talk. Can you talk a little bit about the differences in the targets of N-MYC and L-MYC? Because I’m wondering how much we’ve been confused over the years by seeing that MYC targets go up and maybe their L…
Myles Brown, MD [00:17:57] Yeah, so that’s a good question. I can think we haven’t really analyzed that because we haven’t done as much with N-MYC yet, but we’re doing that. So, I think that’s a very good question, is whether I was saying that, you know, mostly what’s in the database, if you look at ChIP-seq data, it’s mostly c-MYC data. Yeah. So, you know, that’s one sort of downside of our Cistrome database is that it’s biased by what people have done. So, you’re often gonna get ER, because ER has been done more than probably any other transcription factor, but so it’s a little bit biased. So that’s something we’re trying to solve.
Unknown [00:18:27] And what about the, you know, I know the cells you showed have the LNCaP cells have no p53 and no Rb. Did you look at one or the other?
Myles Brown, MD [00:18:38] Yeah, so yeah, so I mentioned we did do that, so I didn’t show all the controls, but the single knockouts were like the control. So, you had to have both Rb and p53 knockout to get this phenotype.
Unknown [00:18:51] Hey Miles, so it’s fascinating that L-MYC knocks down c-MYC only in the Rb, p53. So, the adenocarcinomas have a prostate specific enhancer for MYC. I’m wondering if you’ve looked in the Rb p53 LnCaP, is that kind of turned off?
Myles Brown, MD [00:19:12] Yeah, that’s a good question. Yeah, because again, work from Matt had identified that process was selective yeah, we should definitely look at that. Yeah.
Unknown [00:19:21] Yeah, I guess relevant to that is why and how MYC goes away.
Myles Brown, MD [00:19:27] Yeah, so we think it’s transcriptional, but again we’re still working on trying to understand, you know, whether it’s transcriptional and you know we have to be a little bit careful because as you know, we know from David’s work, we saw that the way that MYC downrated AR activity is not actually through binding of AR, you know, that the AR system is actually expanded, though the AR program is reduced.
Unknown [00:19:54] Do you also see other MYC-related transcription factors like Mad-Max?
Myles Brown, MD [00:19:59] Yeah, so again, that’s in my backup slides, but I guess they’re gone. So yeah. We looked at that. We didn’t see any differences in changes of the known Max, Mad-Max, you know, all these large numbers of MYC collaborators, but no obvious candidates.
Unknown [00:20:16] Thank you.
Unknown [00:20:19] So, given the very strong similarities between the different MYC family members, where would the selectivity now come from?
Myles Brown, MD [00:20:28] Yeah, so again that we still have this question. So, this is a question I can ask. Is it is this any MYC will do? And it has to do with having the MYC expressed. But that doesn’t seem to be exactly right because MYCL definitely does different things and doesn’t do things in the prostate adenocarcinoma cell lines. So, there must be some difference, and they are different. And certainly, they’re you know, the major way they’re different is their pattern of expression, but they seem to have subtle but important activity differences.
Himisha Beltran, MD [00:21:04] Great. Thank you, Doctor Brown, for an amazing talk.