Reciprocal CRISPR gene editing in pediatric glioma: defining mechanisms & testing drugs

My lab’s new paper in Communications Biology focused on high-grade pediatric glioma that have mutant histone variant H3.3 and we did something fairly novel that we are calling reciprocal CRISPR. Kids with these tumors have a near zero survival rate within a few years of diagnosis so we as a field desperately need something new to give them and their families real hope.

Our paper is entitled, “Reciprocal H3.3 gene editing identifies K27M and G34R mechanisms in pediatric glioma including NOTCH signaling.”

What we did in this paper with CRISPR

What our team did here, led by my postdoc Mike Chen, is to precisely introduce K27M and G34R mutations by gene-editing in the H3.3-coding gene H3F3A into 2 previously H3.3 WT cells (human astrocytes and an H3.3 WT childhood glioma) and in parallel we precisely gene-edited pre-existing K27M mutations in 2 childhood glioma lines (DIPG tumors) back to WT. This gave us a panel of isogenic sets of cells with the only difference being whether H3.3 was mutant or WT.

We then analyzed these cellular “twins” to pinpoint mechanisms by which mutant H3.3 contributes to the glioma. We began looking at their drug sensitivity too and did tumor assays (xenografts) as well.

A key finding: K27 and G34R are functionally similar mutations

One of the most interesting findings was that the K27M and G34R mutations appear to behave so similarly. While there were some differences in potential mechanisms for the two mutations that emerged (and these are interesting too), both mutations impacted NOTCH signaling and some of the same neurogenesis pathways and neural genes too including ASCL1. The two mutations also had similar impacts on histone marks at candidate target genes. We are currently thinking that part of what the G34R mutation does is to impact K27M post-translational modification and then it does unique things as well.

Another finding that struck us was that K27M was connected to ectopic super-enhancer formation (likely due to reduced K27me3) with the super-enhancers linked to oncogenic pathways. While we need to study G34R more, given the genes it impacts it is likely that it has a role in super-enhancer biology as well.

Gene-editing switches tumor phenotype

It was also striking that gene-editing of single codons in H3.3 changed the tumor phenotypes in xenografts in vivo and also altered drug sensitivity in cultured cells as well. See image above of immunostaining data from brains with the SF cells, which in our xenograft assays only rarely formed tumors. However, SF cells when gene-edited to have K27M formed aggressive tumors most of the time.

Again, note that these tumors are nearly 100% fatal within 2 years so new treatments are urgently needed. On the drug screening front, we started looking at candidate drugs that we tested on our panel of cells. We reasoned that drugs that showed greater activity impairing cell viability of H3.3 mutant cells vs. their isogenic “twin” cells that are H3.3 WT are most likely to be truly mutant H3.3-specific and perhaps clinically promising. More to come on this front in the future.

Looking ahead

We found a lot more that’s quite interesting so please read our paper and let us know if you have any feedback or questions.

More broadly, we believe this kind of reciprocal CRISPR gene-editing (where in parallel you precisely introduce disease-associated mutations in previously WT, disease-relevant cells and revert the same mutations back to WT in primary patient cells) is going to be extremely useful for human genetics and genetic disease research including cancer.

I’m very proud of my team for the great work they did here.