Read the Full Video Transcript

Andrea Miyahira: Hello everyone. Thank you for joining us today. I'm Dr. Andrea Miyahira at the Prostate Cancer Foundation. Today, I'm joined by Dr. Andrew Hsieh, an associate professor at the Fred Hutch. We're going to discuss his group's recent paper, Transcriptional-Translational Conflict is a Barrier to Cellular Transformation and Cancer Progression, that was recently published in Cancer Cell. Andrew, thank you so much for joining me today.

Andrew Hsieh: Really happy to be here and very excited to share our work. So just as an overview, we have uncovered a new tumor suppressor mechanism that we call transcriptional-translational conflict. These are my disclosures. It was really spearheaded by Sujata Jana in the laboratory, and it was very much a labor of love. It took a long time, about six years, for her to figure this one out. But basically, when we think about tumor suppressors and the process of protein synthesis, we usually think that they go against each other. Meaning, when the tumor suppressor is there it keeps down protein synthesis and when it's gone, protein synthesis goes up and that drives cancer. There are multiple groups have shown this for major tumor suppressors. We and others have shown over the years that increased protein synthesis is very important for the genesis of cancer.

When I started in the lab in 2014, very interested in getting into the bladder cancer field, and that was also the same year TCGA came out. And what I noticed looking at TCGA for bladder cancer, was that chromatin remodelers were heavily deregulated there. In particular, a chromatin remodeler component of the SWI/SNF complex called ARID1A, also known as AT-rich interactive domain-containing protein 1A. And hat this complex does is it actually moves nucleosomes around to expose promoter and enhancer regions that enable wide changes in transcription. Others have already shown that it seems to be a tumor suppressor, because when you lose it it leads to the up-regulation of oncogenic gene networks, impaired DNA double strand breaks and decatenation, ultimately tumor formation.

So we were very interested in studying this in the bladder. We want to make sure that it is indeed lost. So we are focusing on the urothelium in the bladder for the models that I'm going to show you. But first we looked at our rapid autopsy bladder cancer program and looked at tissue samples. And what we found was that indeed, about 20% of patients don't have any ARID1A suggesting that perhaps it is actually tumor suppressors. Now, we wanted to study this in vivo. We developed a mouse model in which we could lose ARID1A in an inducible manner by using a floxed allele of ARID1A around exon 8. And this is using a whole body creep.

When we did that, you see ARID1A levels go away. And what's really interesting is when we do that at a transcript level, when we do RNAseq, you get an up-regulation of transcripts that are really important for cell proliferation. And some of these are actually really important in cancer such as aurora kinase B, ETV1, FGFR3, including IGF2.

So others have shown that these genes are sufficient to cause tumors in mice. So got us thinking, okay, it probably is a tumor suppressor and because of that it's probably going to cause tumors. So we went ahead and aged these mice for 400 days and we saw no difference at the end of the 400 days, no tumor formation, no change in the thickness of the urothelium, and that got us thinking that maybe there was a process that's going against this increase in transcript levels that could then be tumor suppressive. And so we did a very simple assay, which is to measure protein synthesis within these tissues using this puromycin incorporation assay. And we saw that when you lose ARID1A, protein levels go down.

Now this is a whole body knockout. We wanted to be more precise, so we made a basal cell specific knockout within the urothelium. And what we found is that when you do that, and if you look at the demarcated yellow region right here, you lose ARID1A and protein synthesis is almost completely, is very significantly decreased. Now we wanted to make absolutely sure it was cell autonomous, so we developed organoids from this and showed that indeed you'd get decreased protein synthesis when you lose ARID1A and when you add it back, it restores protein levels as shown here.

So what we saw was this conflict between transcription and translation, but we wanted to know how this thing was occurring. And so what we ended up doing was taking a closer look at what aspects of translation, either the initiation or the elongation of translation. And what we saw was that there seemed to be an elongation defect because the ribosome seemed to be moving more slowly across RNA. And in fact, if you do this thing called ribosome half transit times, the ribosomes are moving about half as fast as they normally do. So there's clearly an elongation defect. And in fact, those genes that were up-regulated, when we looked at each one of them in our polysome RNAseq data, about 70% of them were actually stalled. And in fact, when we then do mass spec and look at those same proteins that should be up because the transcript levels, we see that there's no change.

And so what we call this process is transcriptional-translational conflict that basically is tumor suppressive. And what we find through much mechanistic work, and I'm not going to go into the details today, is that there's a mechanistic link between chromatin remodeling through ARID1A and the control of this particular signaling pathway linking RASGRP1, MAPK signaling, and a translation elongation factor called eEF2. This causes transcriptional-translational conflict.

Now what's really interesting is we say this is tumor suppressive. So how do we know that? And so we did a series of cell-based assays and basically found that when you de-repress translation elongation using a GENNEK model, it allows the cells to grow, basically it has replicative immortality, which is not what we see with ARID1A alone. And when we actually do this and we look more closely at a clonogenic assay or a cell cycle, we see better clonogenicity when you de-repress translation elongation, we see more cell proliferation, and in fact, when you do this in a mouse, you see now instead of not having any change, you have uncontrolled cell growth. Okay?

And surprisingly, this doesn't seem to be just dependent on eEF2 or translation elongation because if you use a carcinogenesis model that just preferentially increases protein synthesis when tumors form and you now go ahead and cause a tumor which has more protein synthesis and then lose ARID1A, what you see remarkably is now you see the tumor suppressive effects of ARID1A taking effect where the mice live much less, they proliferate much faster their tumors, and those previously conflicted genes where there's no change at all have now been de-repressed and are expressed at very, very high levels. Which leads us to our final conclusion, which is could we actually induce transcriptional.

Translational conflict using drugs? And so to do that, we used a molecule called homoharringtonine, which actually FDA approved for treatment refractory CML. We tested in our bladder cancer of mouse model and we saw that indeed ARID1A deficient tumors were much more sensitive to HHT compared to their wild type controls. We looked at bladder cancer cell lines that had high or low levels of ARID1A and it's the ones that had low levels of ARID1A that were sensitive.

And then we screened a series of patient-derived xenograft models and looked for ones that have no medium to high levels of ARID1A. And what we found was that the ones with high levels were very sensitive to the inhibitory effects of HHT, whereas the medium one was less and the ones with high levels of ARID1A were completely insensitive to translation inhibition. So what I've shown you today is that we've discovered a new tumor suppressive process called transcriptional-translational conflict. We found that if you can de-repress transcriptional-translational conflict genetically or through tumorgenesis, you can actually drive gene expression parody, which is pro-tumorogenic and it unleashes the tumor suppressive properties or the oncogenic properties of ARID1A loss.

We also found that if you use small molecules to now turn back the tide of translation, this has an anti-tumor effect which is specific to ARID1A deficient tumors. We think this is a really important finding because in the context of what we know about cellular stresses and oncogenes and tumor suppressors, it's been shown there's a precedence where oncogenes or tumor suppressors when they're activated can actually lead to cellular stresses that are not good for cancer cells. For instance, MYC can cause apoptosis, PTEN loss can lead to senescence. PIK3CA could lead to cellular differentiation. We posit that ARID1A loss leads to eEF2 media transcriptional-translational conflict, which is new molecular stress. That one we need to think about.
So I want to thank you for your time. This work would not have been done without a vibrant bladder cancer research community here in Seattle, and I'm very indebted to the people that I work with here as well as my own laboratory. These are my lab members and our funding sources. And Andrea, thank you very much for this opportunity. Happy to answer any questions.

Andrea Miyahira: Well, thank you for that presentation, Andrew. I learned a lot and think this is really interesting to see this new mechanism and have us think in a different way about how tumor suppressors work. So some questions. My first is is ARID1A loss or SWI/SNF commonly seen in combination with mutations that alter protein expression in human cancers?

Andrew Hsieh: So that's a very important question. What's very interesting is that there's been a series of mouse models that have been generated to actually show ARID1A alone is actually not sufficient to cause cancers. And these are in other models like ovarian cancer for instance, is one that's popularly studied because ARID1A is lost even more frequently in ovarian cancer. And what they've shown is that mutations to PTEN or activating mutations to PIK3CA, so the PTEN ones loss of function, it's the PIK3CA's beta function as well as over-expression of MYC, all three of those actually synergize with ARID1A loss to actually drive a more aggressive tumor than any one of those things alone, which I thought was really, really interesting because PTEN and PIK3CA and MYC are all central regulators of translation or mRNA translation. So I would say there definitely is a precedence for the fact that there are the mutations that seem to be very important in driving translation at the same time as losing ARID1A and gaining the transcriptional benefits of that.

Andrea Miyahira: Are there other examples of how major roles of other tumor suppressor genes in driving cancer involves this conflict versus parity between transcription and translation? I guess other examples of these synthetic lethal interactions?

Andrew Hsieh: Right. So to my knowledge, we're the first group that has shown that transcriptional-translational conflict exists. So my understanding is that there's not very much known at this point in cancer about how pervasive this is because it's essentially a silent process, right? Basically the cells at the end of day won't be there. But my lab's very interested in understanding how pervasive this is. And one thing we are thinking about is, well, you really can't study this one gene at a time. Oh, we could study one gene at a time, but nowadays in the landscape that we live in with screens available, we're very interested in developing tools and we have developed now a tool to actually do in vivo based screens that will allow us to uncover this in normal cells.

So I think that's the other important thing. We can't be studying this in cancer cells. We got to study this in cells that are normal and probably the best normal cell would be a tissue. And so we're very interested in diving into this in bladder tissue, but I'm almost positive and I think there's going to be another question on this one that this is important in other tissues as well.

Andrea Miyahira: Okay, cool. Do you have any translational plans for these findings?

Andrew Hsieh: Yes. So at this point right now, I talked to, actually, I'm working with my colleague Petros Rivas to develop a phase two where we use a translation inhibitor to treat patients with metastatic bladder cancer who are pre-selected for having ARID1A mutations. We actually looked at our sequencing of patient data here, and that turns out to be roughly, it's consistent with what TCGA has reported. It's about 25% of patients. So that's the goal right now we're working on, we have potential, actually we do have drug, the question is of course getting the finances to do an IAT and we're working on that right now. We're very, very excited about it because I mean, it would be a realization of not just a very interesting mechanism, but its application to patients, which I think is the most important thing.

Andrea Miyahira: Okay. That's really exciting. And since many of our viewers are likely prostate cancer researchers, do you know or have you investigated these mechanisms yet in prostate cancer or prostate cancer models?

Andrew Hsieh: So there's been very nice work in the field basically showing that SWI/SNF is important in prostate cancer. And so we wanted to see how pervasive these changes actually are in other tissues. So we did this set of experiments here, which is in the paper where we basically either knocked down or floxed out ARID1A in different cell types. So the experiment on the left, it was actually in PrEC cells, so these are prostate epithelial cells from people. And when we knocked down ARID1A, you see a significant decrease in puromycin incorporation here. And when we do the same thing in bone marrow stromal cells developed from ARID1A floxed mice, we also see a decrease.

So I would say that there is very likely molecular processes that are happening in these directions the same way in other tissues as well. Now, whether or not they are something that a tissue needs to get over in order for oncogenesis to incur, I think that needs to be studied one tissue at a time. One of the complex things about ARID1A that we're realizing, actually not such, it's decades of work by other folks, is that ARID1A seems to have a very significant tissue dependency. So certain tissues seem to be more sensitive lost to it than others. And so I think for a real good study of this, we would need to really kind of consider the context where ARID1A is loss. And it brings up the greater question too, of whether or not other SWI/SNF component members are important as well in the same way.

There's a very nice piece of work by Capucine Van-Rekham at Stanford that actually looked at different cell lines, and in this case was also looking at other SWI/SNF component members. And certain ones also do the same thing to protein synthesis. So I think there's a lot to be addressed here. We're excited to be part of the field and we're hoping to make more discoveries in the future.

Andrea Miyahira: Okay. Well thank you so much for sharing this with us, and congratulations again on this exciting discovery and paper. Thank you.

Andrew Hsieh: Thank you very much.