Read the Full Video Transcript

Andrea Miyahira: Hi, everyone. I'm Andrea Miyahira, here at the Prostate Cancer Foundation. Please help me to welcome Dr. Jean Tien of the University of Michigan. We will discuss her recent paper, CDK12 loss drives prostate cancer progression, transcription-replication conflicts, and synthetic lethality with paralog CDK13 that was published in Cell Reports Medicine. Dr. Tien, thanks for joining us.

Jean Tien: Thank you for having me. Yeah, it's a great honor to be able to present right here.
So today, I'll be talking about CDK12, a gene inactivated in a subset of prostate cancer. I'll make the case that CDK12 behaves as a tumor suppressor gene and discuss that targeting its paralog, CDK13, is a therapeutic option for treatment of CDK12-mutant tumors.

CDK12 is a transcription associated cyclin dependent kinase, meaning that instead of promoting cell cycle progression, it regulates various aspects of transcription. For instance, when bound to its activating partner, cyclin K, it phosphorylates the C-terminal domain of RNA polymerase to enable its progression through the gene. CDK12 targets include genes related to genomic stability, cell proliferation, and cell survival.

Our group's interest in CDK12 emerged from a large-scale sequencing study of primary and metastatic prostate cancers. Putative CDK12 inactivating mutations occurred in less than 2% of primary tumors, but nearly 7% of metastatic tumors. CDK12-mutant tumors displayed a tandem duplication genomic instability phenotype.

It remains unclear, however, whether CDK12 was a bona fide tumor suppressor gene in prostate cancer. To address this, we conditionally ablated CDK12 in the mouse prostate epithelium using a probasin promoter-driven Cre recombinase. The resultant animals develop hyperplasia and high-grade PIN at one year old. Cross section area occupied by this lesion was about 10% in the anterior and dorsal lobes, and 4% to 5% in the ventral and lateral lobes. Lesions were associated with basal cell accumulation and increased cell proliferation.

Organoids derived from CDK12-null basal cells had an abnormal compact appearance, while also showing increased cell proliferation, [INAUDIBLE] signaling, and enzalutamide resistance in culture.

We used this organoid to screen for positive genetic interaction with CDK12. Specifically, we infected a CDK12-null organoid with a CRISPR knockout library, growing the resulting cells as a subcutaneous allograft in immunocompromised mice. Then, after six months, we sequenced the allograft to identify the most common guide RNAs.

P53 was the top hit, meaning that its loss most enhanced tumor formation, and these findings are consistent with the fact that CDK12 and P53 mutations often co-occur in human prostate cancer. We implanted organoids lacking CDK12, P53, or both into immunocompromised mice. Strikingly, only the double knockout organoid formed viable tumors. Notably, these tumors displayed considerable DNA damage, indicated by gamma H2AX staining.

We next evaluated the interaction between CDK12 and P53 loss in human tumors. Clinical data from U of M showed tumors with co-mutation of CDK12 and P53 (the purple line) conferred reduced survival versus tumors with individual loss of CDK12 (the green line), or P53 (the blue line), as well as tumors in which both genes are intact (the red line). Data from Jonathan Cho at UCSF showed reduced time to metastasis in tumors with CDK12-P53 co-mutations (the black dashed line) versus CDK12 loss alone (the blue line).

After ruling out simple reductions in the expression of DNA repair genes, which are known to be positively regulated by CDK12 in other cancers, we looked for other mechanisms of DNA damage. Transcription-replication conflicts (TRCs) are instances in which DNA and RNA polymerase collide. They are common across multiple cancers and give rise to R-loops, DNA-RNA hybrid structures induced by cancer-related hypertranscription, and without proper resolution, they predispose DNA to double-strand breaks.

So we assayed for RNA loops in our organoids by staining with the S9.6 antibody that recognizes DNA on hybrid structures. As you can see, these increased in the CDK12 knockout organoids, but were not detectable when the samples were treated with RNase H.

We next used proximity ligation assay for PCNA (a DNA polymerase-interacting protein) and RNA Pol II. Labeled PLA foci indicate regions where DNA polymerase and RNA Pol II are in close proximity, and these are indicative of transcription-replication conflicts.

We observed that the loss of CDK12 and P53 independently induced PLA foci, though the increase was more pronounced in the CDK12 knockout group, and the highest in the CDK12 knockout sgP53 group. These data showed that CDK12 loss promotes DNA damage through induction of TRCs.

Since there are no targeted therapies for CDK12-mutant prostate cancer, we looked for synthetic lethal genes that could be targeted when CDK12 function was lost. Our collaborator, Chris Lord, from the University of London, performed a screening experiment using a HeLa cell line developed by the Green Leaf Lab at Duke. In this line, one CDK12 allele is ablated using CRISPR, and the other is modified so CDK12-as can be inactivated by treatment with the chemical 1-NM-PP1. Here is a Western blot validating the system, showing that 1-NM-PP1 treatment reduced phosphorylation of the RNA polymerase II C-terminal domain.

Treating with an siRNA library targeting CDK12-related genes found that the knockdown of CDK13 caused the greatest reduction in cell viability. CDK12 and CDK13 are paralogs with overlapping function in regulation of genes required for cell division and survival. The data therefore reveal a novel paralog-based synthetic lethality.

We then tested whether pharmacologically targeting CDK13 would preferentially kill CDK12 knockout prostate cancer cells. To do this, we used YJ9069, a PROTAC-based CDK13 degrader developed by our collaborator, [INAUDIBLE]. Because CDK12 and 13 are structurally similar, this agent causes degradation of both. That said, it is more specific than previous CDK12/13 inhibitors, which also target other CDKs like CDK9.

When we treated the CDK12 wild-type and knockout organoids with YJ9069, we saw that the drug had a greater effect on the viability of the knockout. The Western blot shows that in the setting of CDK12 knockout, there is a compensatory increase in CDK13 that enables maintenance of polymerase II serine phosphorylation. By contrast, protein phosphorylation declines when CDK12 wild-type cells are treated with YJ9069, and this phosphorylation is essentially absent when CDK12 knockout cells are treated with the drug, corresponding with reduced viability.

We next analyzed the organoid lines generated from patient-derived xenografts (PDX). The green line shows PDX with CDK12 inactivation, and the black line shows PDX with intact CDK12. As you can see, YJ9069 treatment preferentially killed PDX cells without functional CDK12.

Finally, we grew PDX as a subcutaneous xenograft in mice. In this system, YJ9069 blunted the growth of the CDK12-mutant PDX but had relatively little impact on the growth of PDX with intact CDK12.

In summary, CDK12 inactivation, per se, causes cancer-like changes. In mice, they develop hyperplasia and PIN lesions, and the organoids derived from the CDK12 knockout mice show the compact morphology. CDK12 inactivation enhances the tumorigenic potential of P53 inactivation—P53 knockout organoids do not form tumors in vivo allograft, but P53/CDK12 double knockout cells do form tumors. These tumors exhibit transcription-replication conflicts leading to marked DNA damage.

CDK12 inactivation causes vulnerability to paralog synthetic lethality. CDK12-null organoid-derived tumors and human CDK12-mutant PDX are especially sensitive to CDK13/12 degraders.

So thank you for listening. I would like to acknowledge my mentor, Arul Chinnaiyan, for his support and guidance, and I'd also like to thank the MCPT Lab members—it's a great place to work. And I would like to thank the funding sources and collaborators who made this study possible.

Andrea Miyahira: Thank you so much, Dr. Tien, for sharing this with us. So are CDK13 alterations ever observed in prostate cancer?

Jean Tien: No. So actually, to my knowledge, the CDK13 mutation has not really been observed in prostate cancer.

Andrea Miyahira: So you noted that CDK12 loss was seen in a small percentage of primary prostate cancer. Is CDK12 loss considered an earlier truncal alteration in human prostate cancer?

Jean Tien: Yeah. So according to our prior work from the group, CDK12 loss appeared independent of other classical truncal mutations like MMR deficiency, HR deficiency, and ETS fusions. Therefore, it appears to be an early truncal alteration that defines a specific subtype of prostate cancer.

Andrea Miyahira: OK. Thank you. And what does your study imply for guiding the use of PARP inhibitors or immunotherapy, et cetera, in CDK12 loss tumors?

Jean TIen: So I tried immunotherapy in my CDK12/P53 double-knockout allograft models, specifically treating the mice with anti-PD-1 and CTLA4. I did see a response to the therapy, but clinical efforts to treat CDK12-mutant tumors with these agents have yielded decidedly mixed results, so there's likely more to the story in terms of defining potential responders versus non-responders.

Regarding PARP inhibitors, I did try that as well, but I observed that CDK12-null tumors do not show enhanced sensitivity to this agent, though there are publications to the contrary. So, as with the immunotherapy issue, there might be more to the story.

Andrea Miyahira: OK, thanks. And what translational strategy do you think would be best for targeting CDK12 and CDK13 together? For instance, would you apply this specifically in CDK12-mutant prostate cancer, or would you consider combination treatments?

Jean Tien: Well, I think the strongest therapeutic direction indicated by our work is the inhibition of CDK13 in CDK12-mutant cancers. In addition to prostate, CDK12 is also mutated in ovarian cancer, and we have unpublished data indicating the effectiveness of CDK13 inhibition in that setting.

You can find our group also demonstrated synergy between CDK12/13 inhibition and AKT inhibition as an effective treatment in cell lines and PDX lines from various cancers, regardless of CDK12 mutation status.

Andrea Miyahira: Thank you. And then what are your next steps, and do you have any translational plans?

Jean Tien: So I'm finishing up the story showing the impact of CDK12 in a series of ovarian cancer. And the findings are quite similar to those seen in prostate cancer. And so translational applications of CDK13 inhibitors or degraders are on the horizon for multiple CDK12-mutant cancer types.

Andrea Miyahira: OK. Well, thank you so much for sharing this with us today.

Jean Tien: Thank you. Thank you for having me.