Abstract
NPM1 mutations are among the most common mutations in acute myeloid leukemia (AML), being found in about 30% of de novo AML in adults. NPM1 is a multifunctional nucleolar chaperone involved in genomic stability and ribosome biogenesis. All the mutations in NPM1 described so far result in cytoplasmic protein localization (NPM1c) through the acquisition of a nuclear export signal (NES) at the C-terminus. Based on these observations, we hypothesized that cytoplasmic localization of NPM1c is necessary for AML pathogenesis and maintenance. To test our hypothesis, we edited the C-terminus of the NPM1 mutant allele in order to remove the NES and re-localize NPM1c to the nucleus.
Specifically, using our recently optimized CRISPR-Cas9 approach (Gundry, et al. Cell Reports 2016), we sought to introduce indels to disrupt the C-terminal NES in AML cells bearing NPM1 mutation A (heterozygous 4bp insertion at the C-terminus), thereby creating novel edited alleles encoding for a mutant NPM1 with nuclear localization. Cell lines, primary derived xenografts (PDX) and primary AML samples harboring NPM1 mutation A were successfully edited with an sgRNA spanning the 4bp insertion.
While the NPM1 wild-type allele remained intact, the mutant allele was edited with up to 90% efficiency. The novel edited alleles could direct nuclear localization of a GFP-NPM1 fusion construct. Additionally, re-localization of cytoplasmic NPM1 in edited cells was confirmed by immunofluorescence. Return of NPM1 protein to the nucleus resulted in terminal differentiation, and significantly impaired cell growth, colony forming ability and engraftment in xenograft models. Transcriptome analysis on two cell lines with mutated NPM1 revealed that upon nuclear re-localization of NPM1c, only 22 genes were down regulated more than 2 fold in both cell lines. Strikingly, 11 of these 22 genes were HOX genes. Altogether these data indicate that the cytoplasmic localization of NPM1c is necessary to maintain the transcriptional signature and leukemic phenotype of AML cells with mutated NPM1 .
To extend our study further, we used homology-directed repair (HDR) to successfully correct NPM1 mutations to the wild type sequence in AML cell lines (Fig.1). We simultaneously fused GFP to the C-terminus of the protein to verify its appropriate localization (Fig.1). Moreover, to assess whether the phenotype of AML cells depends on the amount of mutant NPM1 in the cytoplasm, we inserted a series of NPM1 mutant alleles (a, b and c) that resulted in different proportions of cytoplasmic versus nuclear localization (i.e. a - no cytoplasmic mutant NPM1, b - ~50% cytoplasmic/50% nuclear mutant NPM1, c - ~100% cytoplasmic mutant NPM1; Fig.1) in two different AML cell lines with mutated NPM1 . Cells with high levels of nuclear NPM1 tended to differentiate rapidly, while cells with a greater proportion of cytoplasmic NPM1 remained less differentiated and continued to proliferate (Fig.1). This established an ongoing competition in which cells with higher nuclear NPM1 were rapidly outcompeted by unedited cells, while cells with higher cytoplasmic NPM1 were not. Thus, these experiments clearly demonstrate the dependence of the leukemic phenotype on the cytoplasmic localization of NPM1.
Recently, compounds able to inhibit the nuclear exporter CRM1/XPO1 have been developed and are currently being tested in clinical trials (e.g. selinexor). Since NPM1 is shuttled to the cytoplasm by CRM1/XPO1, its inhibition re-localizes NPM1c to the nucleus. We therefore treated cell lines, PDX and primary samples with selinexor with the aim to reproduce the nuclear re-localization observed using CRISPR-Cas9. As expected, selinexor treatment resulted in clear growth arrest, differentiation, and NPM1c re-localization with dynamics similar to what was seen with gene editing. Importantly, after only 3 days of treatment selinexor produced a dramatic drop of HOXA, HOXB and MEIS1 expression.
In conclusion, allele-specific editing is a powerful tool to probe the mechanistic aspects of oncogenic dependencies. By achieving nuclear re-localization of mutant NPM1, we demonstrated that cytoplasmic localization of NPM1c is necessary for NPM1-mutant AML cells to maintain their leukemic phenotype. Drugs promoting mutant NPM1 nuclear localization are therefore attractive candidates for clinical treatment of AML with mutated NPM1 .
Figure 1 Figure 1.
Disclosures
No relevant conflicts of interest to declare.
NSC348884 cytotoxicity is not mediated by inhibition of nucleophosmin oligomerization