Thread by @rosslevinemd

Its double preprint time! After 5+ years of planning and hard work, @bowman_rl and @andrewdunbar_md’s amazing work is ready to share with all of you on @biorxiv! First one biorxiv.org/content/10.110…

And second one here! biorxiv.org/content/10.110…

Boom!

Landscape studies and scDNA work by @lindemilesphd, @ bowman_rl in our lab and @koichitakahashi @scell_papers have allowed the field to delineate of the somatic mutational repertoire of myeloid malignancies #leusm and provided an unprecedented view of clonal evolution->AML.

The challenge, then, is how do we model sequential mutagenesis in vivo?

This is the “@jenga puzzle” challenge @hasbro. How can we build sequential models of transformation, and then can we figure out which Jenga pieces to pull out to see AML “crumble”? Bobby developed models of AML transformation using 3 orthogonal recombinase (Cre/Dre/Flp)

He began with a Flp inducible FLT3-ITD allele, showing that somatic FLT3-ITD induction leads to a biphasic myeloproliferative disorder with dynamic gene expression changes (activation->feedback), but that FLT3-ITD mutant stem cells have no hematopoietic self-renewal.

However, when activated in concert with NPM1c or sequentially after NPM1c->rapid progression to fulminant AML capable of propagating into secondary recipients. By contrast, activating FLT3-ITD after DNMT3A led to myeloid expansion and enhanced self-renewal but not to AML

He then made a 3-mutation AML model, adding one more wrinkle->mutational reversibility!

Meet GOLDI-Lox; Lox sites flank a Rox pair which flank an inverted Flt3-ITD allele. The proximity of Lox sites at baseline prevents Cre recombination. Dre activation induces Flt3ITD & subsequent Cre can now recombine->FLT3-ITD deletion!

He shows that sequential Cre->Flp->Dre activation of DNMT3A->NPM1c->FLT3 is possible with 3 recombinases, and that this gives rise to AML. We now have a model of the most common AML mutational triad with the mutations acquired in the most common order observed in patients!

To remove the bottleneck of transplantation, Bobby then developed orthogonal chemical tools that allow in vivo induction of all 3 recombinases. First, a FlpoDHFR system such that Flp recombinase is induced with Trimethoprim. Works beautifully!

For Dre, he got really creative, and placed a HCV-derived NS3 protease site (@michaelzlin) in the middle of the Dre recombinase, with an ODC degron on the N-terminal fragment. When HCV protease inhibitors are dosed in vivo->transient Dre expression and recombination!

Even better, he showed that all you need is low level recombination to induce FLT3-ITD in a small population of pre-leukemic cells->fulminant AML. We can titrate mutational activation all the way down and model mutational acquisition as a rare event… see the clones expand!

As the “piéce de résistance” he used the reversible feature of the FLT3 Goldilox allele to reverse FLT3-ITD in IDH2/FLT3 and NPM1/FLT3 models, and to show that oncogene reversal has distinct phenotypic/transcriptional effects on different FLT3-ITD mutant models

This work lays out a path to model sequential mutagenesis in vivo, and to use these systems to investigate clonal evolution, interclonal crosstalk/competition, and mutational dependency. As human genomic studies increase->we will build models of increasing clonal complexity.

Thanks to everyone in the @Levine lab, plus our awesome collaborations across both papers including @viny and @Landau on the JAK2 story and @SaraMeyerLab Martin Carroll, @FerrellLabVUMC, and @TrowbridgeLab here!

Thanks to our funders, including @theNCI @DamonRunyon @MSKCancerCenter @Cycle4Survival!