This project will build on our recent publication in which we uncovered the consequences of chromosome 8p alterations in engineered iPS cells and characterized its effects on gene expression and neural differentiation: pubmed.ncbi.nlm.nih.gov/40894639/.
I’m excited to share a new postdoctoral opportunity in my lab at Stanford to study the consequences of gene dosage alterations in iPS cells. Check out the posting below and shoot me an email if you’re interested -
This work was led by Kaitlin Long, a phenomenal undergrad/tech in my lab. Many of you reading these tweets received a PhD application from her over the weekend! I cannot emphasize enough what an exceptional scientist she is - I think that any lab would be lucky to have her join.
Additionally, we think that PAC-1 could have significant utility as part of a broader combination-therapy regimen, to delay or reverse MDR1 activation and enhance tumor sensitivity to SOC chemotherapies.
Excitingly, PAC-1 has entered clinical testing, where it has been well-tolerated and resulted in multiple patient responses. However, no biomarker capable of predicting sensitive tumors was reported. We think that MDR1 expression could be the missing biomarker.
How is this possible? We found that PAC-1 is being effluxed by MDR1 - but PAC-1 bound to iron is effluxed much more rapidly than PAC-1 by itself. So, MDR1 activity doesn't protect cells - it accelerates iron starvation in MDR1-high cells by pumping out the drug-iron complex.
We also found that we could use PAC-1 to reverse this process: we took chemo-resistant, MDR1-high cells and cultured them in PAC-1 for a few weeks, and we found that this resulted in a 20-fold decrease in MDR1 expression and a 7-fold increase in chemotherapy sensitivity.
Similarly, evolving cancer cells in chemotherapy causes MDR1 upregulation and resistance to a broad range of anti-cancer. But, these evolved drug-resistant cells exhibited significant collateral sensitivity to PAC-1.
To confirm this association, we generated MDR1-KO clones and we verified that these cells were more sensitive to standard chemotherapies, as we expected. However, we also found that the chemo-resistant MDR1-high cells were much more sensitive to PAC-1 than the MDR1-KO cells!
Surprisingly, using PRISM, we found that the cells most sensitive to PAC-1 were over-expressing MDR1 (also called P-gp). MDR1 stands for MultiDrug Resistance 1: it’s an efflux pump that expels drugs from tumors. But our data suggested that MDR1 conferred PAC-1 sensitivity!
We used the PRISM dataset from @corsellos and found that PAC-1 was behaving like an iron-chelation agent. We confirmed that PAC-1 depleted cellular iron, upregulated an iron-starvation transcriptional response, and PAC-1 lethality could be reversed with iron supplementation.
The drug is called PAC-1. It was initially developed to target cancer cells by activating the executioner caspases. But, we generated CASP3/6/7 triple-knockouts and it still eliminated cancer cells, demonstrating that it must have some other target.
New from my lab on bioRxiv - we found an existing drug that appears to be safe in humans that selectively kills chemotherapy-resistant cancer cells.
Very cool!
1) did you make any attempt to eliminate guides likely to cause false-positives due to chromosome truncations? (PMID: 38811841)
2) are you including guides targeting new genes discovered from T2T sequencing, particularly on the sex chromosomes? (PMID: 37612512)
Next up - we want to improve the tumor-specific accumulation of CDK11 to bypass this toxicity, and we’re looking for other emerging drug targets to create mouse models for. If you’re interested in collaborating, feel free to reach out!
Along the way, we also learned a ton about the biology of CDK11, the 1p36 locus (one of the most frequently-deleted regions across cancer genomes!), and the CDK-dependent control of gene expression.
If you have a mutation that blocks the interaction between your drug and its target, and so long as that mutation is tolerated in mice, then you can do the same thing that we did - make a mouse with the resistance mutation and see what happens after drug treatment.
I think that this approach can substantially improve the drug development process. Nearly all cancer drugs fail during clinical testing, and toxicity is one of the most common reasons why. We urgently need better approaches to predict and study toxicity in a preclinical setting.
We injected the G568S mice with a mouse cancer cell line and then treated them with a high dose of MEL-495R (which was tolerable to the G568S mice but toxic to WT mice). This resulted in a significant anti-cancer effect, verifying that on-target toxicity limits effective dosing.
This suggested that toxicity (which we believed to be CDK11-dependent) was limiting our ability to effectively dose these mice. To verify this, we returned to our CDK11-G568S mouse strain to tease apart the cause.
This gave us confidence to move forward with the drug. We identified a non-toxic dose of MEL-495R and tested it in several xenografts. However, it showed very little anti-cancer activity. A splicing qPCR indicated that this non-toxic dose wasn’t appreciably inhibiting CDK11.
We bred a large cohort of CDK11-mutant (G568S) and CDK11-WT mice, treated them with an ultra-high dose of MEL-495R, and it worked beautifully. The wild-type mice became very sick while the CDK11-G568S mice were totally fine. Our drug is specific for CDK11 - in living mice!
We thought - if the mice expressing this mutation are still affected by our CDK11 inhibitor, then that tells us that it’s causing CDK11-independent toxicity. In contrast, if these mice are resistant to the drug, then any side effects of the drug in WT mice are due to CDK11.
We came up with a way to answer this question. We had discovered a mutation in CDK11 that blocks drug binding to it. We thought - what if we put that mutation into a mouse? So, we found the mouse ortholog of the human mutation, CRISPR’d it into some zygotes, and did exactly that.
You can throw every biochemical assay in existence against a drug, but that won’t do it - we can’t test all ~20,000 human proteins at once, it’s really hard to determine drug concentrations in each tissue, and in vivo drug metabolism can generate dozens of derivative compounds.
This brought us to an issue that is absolutely crucial for cancer drug development. All cancer drugs have at least some toxicity. If you have a drug against a new target (like CDK11), how do you know if that toxicity is due to CDK11 inhibition or due to something else?
Now, we wanted to take the drug in vivo. Unfortunately, it had a terrible ADME profile. We worked with the talented chemists at Meliora to develop an improved CDK11 inhibitor, and we created MEL-495R, which exhibits potent CDK11 inhibition and superior PK properties.
This makes CDK11 a new “CYCLOPS” gene, as described by @RameenBeroukhim and Bill Hahn - a vulnerability established when a gene is deleted so that only one copy of that gene remains: www.cell.com/fulltext/S00...
Next, we figured out why - 1p36 is where CDK11 and its activating cyclin (cyclin L) are encoded. Having a lower dosage of these genes enhances the dependency on the remaining enzyme, creating a synthetic-lethal relationship.
For any cancer therapy, finding a biomarker to predict sensitivity is key. We analyzed screening data with CDK11-targeting CRISPR, CDK11-targeting RNAi, and OTS964 treatment, and they all pointed to the same biomarker: Chr1p36 deletions enhance sensitivity to CDK11 ablation.