That only happens at S and G2. And so, when you think about HR and the signal transduction that happens with BRCA and ATM, that actually can only happen during certain cell cycle phases. Now, the other way that cells repair double-strand breaks is through Non-Homologous End Joining, or NHEJ. That can happen in all phases of the cell cycle, but is preferred during G1. Then of course, single-strand breaks can be repaired during all phases of the cell cycle, primarily mediated by PARP and ATR. Now, what's really, really cool is that all of these DNA-damaged sensing mechanisms are governed by kinases. For the medical oncologists in the room, we love kinases because they are druggable enzymes. Again, thinking again about how checkpoint signaling and cell cycle are linked is really, really key. Again, showing you another cartoon here that certain cell cycle phases prefer different ways to repair DNA. The other really important thing is that DNA repair engagement actually causes the cell cycle to pause. In other words, if you are in S phase and you get a double-strand break and you need to undergo homologous recombination, signaling through Chk1 and Chk2 actually puts on the brakes so that the next CDK that drives entry into the next phase of cell cycle is actually stopped. This is ultimately critical, right? Because if the cell cannot repair its DNA and can't figure out how to deal with the problem, it really has three choices.
One is to further delay the cell cycle and give it enough time to repair. Number two is to engage in a different repair pathway. This is what happens in patients who are BRCA-deficient. Or the cell has to die. And so, this brings me to the concept of radiosensitization, how and why. I think as medical oncologists and radiation oncologists, we've been thinking about this for a long, long time. For those of us who treat bladder cancer, GU oncologists who treat bladder cancer, we know that we can use chemotherapies like gemcitabine and cisplatin to radiosensitize, which we have not yet talked about as a potential combination. The goal is actually to generate catastrophic amounts of double-strand breaks, that the cell has no choice, no ability to repair the DNA, and has no choice but to die, while minimizing, of course, toxicity to normal tissues. We talked about chemotherapies, which we've used. Then what I'd like to talk today about is different DNA repair pathways. Various DNA repair inhibitors have been tried, and Dr. Sandhu gave a wonderful primer on this. I wanted to highlight PARP as one of the things that she already talked about. We already talked a little bit about the heme toxicities, the optimal strategy. Does it need to be primed and then treated? If it needs to be treated, for how long? I think at the end of the day, I think there are potentially other enzymes that are better. How do we go about identifying what might be even better targets than PARP? Well, I'm a molecular biologist. I like to use functional genomics, let the system, let the biology instruct me to tell me what is the right answer here.
Actually, when we were trying to set this up a few years ago, we actually came across this paper from Dr. Rodney Hicks's group at the University of Melbourne, which he actually did a CRISPR-Cas9 screen of all 20,000 genes in the genome, in the setting of lung cancer using lutetium-dotatate. The way that he set up the screen is exactly how we were thinking about setting up the screen, and so I just wanted to walk you through some of his data. Essentially, you can inject or you can transduce your cell line with the Brunello library that is very commonly used now, knock down every single gene in the genome, treat your cells with either your lutetium-dotatate or your control, and then essentially ask which genes will drop out of your pool of cells. That's what's actually shown in Panel B. The black lines are the controls, the light blue and the dark blue are the cells that were you inhibit these genes and you give them the radiation treatment, these cells actually decrease in number. You could actually essentially count the number of cells using next-generation sequencing. If you actually put all the genes that they identified in their study and put them into biological pathways, what is very clear that drops out is that all of these genes are involved in double-strand DNA repair, ligation of DNA, which you need to repair DNA, and also DNA recombination. Here, just showing you one example where they knocked out DNA-PK, which we had just talked about. This is the main enzyme that modulates non-homologous end joining. The lines that are important to look at are actually the light blue line, which is the control cell treated with 10 megabecquerel of radiation and the orange line. You can see that the orange line is underneath the light blue line because of the increased sensitization. They did this in vivo, and again, you can see that the combination gives you the best tumor control and the best Kaplan-Meier curve in mice.
We actually, while we were optimizing the screen, we said, "Well, let's just reanalyze their screen and see if there's any other hits that they didn't focus on in their paper that might be really good hits." We took their data, that's in their supplemental, replotted it, and then put it through our filter system to look for druggable sensitizing targets, and also targets that did not affect the control cells, because you don't want to just kill your control cells, you want to have some degree of synergy. When we did this, what we saw was, of course, DNA-PK came out as we expected. Then the other kind of interesting list of enzymes, a lot of these are really interesting, but I'm just going to focus on one of them today, which is ATM. One of the reasons why this was super interesting is if you actually take a look, and we talked a little bit about this again in the last session about looking at genomics, if you take a look at patients who have ATM mutations treated on Pluvicto, what you can actually see, and this is actually data again from the Australian group that Edmond Kwan published. You can see that these patients, their PSA-50 all comers, about 50 to 60%, it is upwards of 80% in terms of their response. Some of the ATM patients here have extraordinary responses, some of them don't, and we can talk a little bit about that, but that is in contrast to another subtype that I'm really interested in characterized by biallelic mutations in CDK12.
There is a 0% response rate in patients who have CDK12 mutations who are treated on Pluvicto. The other super interesting data is that the people who have been studying GBMs have been thinking about this for a long, long time, much longer than we have. In fact, what they've noticed is that you need a concomitant p53 mutation, which happens in about 40% of our patients with prostate cancer, to actually induce radiosensitization with their ATM inhibitor. So multiple different ATM inhibitors, ATR inhibitors, and DNA-PK inhibitors are out there, many of them actually made by AZ. As a proof of concept, we wanted to just look at, well, what happens if we combine an ATM inhibitor with a radiotherapy? This is done in collaboration with Rob Flavell at UCSF. What I'm showing you is an in vitro clonogenic assay. The top line is the controls, the purple are the cell colonies. You can see when you either treat with a low-dose of actinium radiotherapy or with the ATM inhibitor, nothing really happens. What happens if you take a look at the second from the bottom and the bottom line, if you use combinations of the ATM inhibitor to pretreat the cells and then add the actinium, you see an extraordinary amount of synergy. Summary and future directions here, I think we can think about expanding the therapeutic efficacy and window between tumor and normal by really thinking a little bit more deeply about harnessing different DNA repair pathways, and also using the clinical genomics to guide us. Thinking about, well, if you already have an ATM mutation, maybe you want to then give them a DNA-PK inhibitor. In terms of sequencing and dosing, I think we need to think a little bit about how do we prime? How long should that priming period be?
How long should the co-treatment phase be? Then thinking about what other targets we might want to explore, perturbation of the cell cycle because they're so intimately linked, why aren't we doing a low-dose gemcitabine study? That's pretty cheap to do. Then how do cells actually protect themselves? So, in the setting of resistance. I think we'll hear a little bit more about that later in the session, but I think potentially, do they increase DNA repair? Is this a strategy to combat resistance? Do they downregulate target expression as another example? Just a very brief, we've been really interested in trying to understand how cells regulate these different tumor antigens. We've done CRISPR flow screens, try to identify them. We do see AR as a key mediator of PSMA expression. Of course, now we're trying to figure out how do we increase PSMA on the cell surface in order to make PSMA target therapies work better. Okay. I will stop there right before the music comes on and thank people who did the work.