iPSC Reprogramming Identifies the Role of Cryptic Drivers in Preleukemic Clonal Evolution

Myeloid neoplasms are clonal disorders driven by acquisition of somatic mutations in hematopoietic stem cells (HSCs). HSCs acquire mutations at each cell division, most of which are thought to be effectively neutral, while only recurrent driver mutations in a relatively small number of cancer-causative genes are thought to promote clonal expansion of HSCs, or clonal hematopoiesis (CH), and leukemia progression. However, nearly 50% of CH cases are not associated with an annotated driver mutation, suggesting that the true number of drivers might be much larger than catalogued to date. Since CH infrequently progresses to myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML), the dynamics of antecedent preleukemic clonal evolution in individuals who develop myeloid neoplasms remain poorly understood. Here, we use induced pluripotent stem cell (iPSC) reprogramming to track genomic and functional evolution of preleukemic clones in 18 patients with MDS or AML. We generated a panel of 42 iPSC lines spanning all stages of premalignant evolution marked by 1, 2, or 3 annotated driver mutations and isogenic wild-type (WT) clones derived from normal HSPCs in the same patient. To measure genome-wide dynamics of mutation acquisition during clonal evolution, we performed whole exome sequencing of MDS/AML patient samples, preleukemic iPSC clones, and matched WT iPSCs. We defined preleukemic mutations as somatic variants present in the primary MDS/AML patient sample and at least one preleukemic clone but absent in the isogenic WT iPSCs. We found that coding mutation burden in 1-, 2-, and 3-driver clones increased linearly. In a neutral process, the ratio of amino acid-changing to neutral mutations is expected to remain constant. By contrast, we found that 2- and 3-drivers clones harbored a significantly higher proportion of amino acid-changing nonsynonymous and indel mutations compared to 1-driver clones. A regression model incorporating driver and neutral mutation counts showed a significant increase in amino acid-changing mutations during progression relative to neutral counts. These findings show that clonal progression is marked by increasing mutation burden with an excess of amino acid-changing variants, supporting a model of intensified positive selection for unannotated “cryptic” drivers. These genes are likely not annotated as drivers because they are infrequently mutated in myeloid malignancies. To understand how cryptic drivers impact HSC fitness, we first sought to better define the impact of recurrent preleukemic DNMT3A R882H mutations. DNMT3A mutations enhanced hematopoietic output of iPSC-derived HSPCs and reduced expression of immune response pathways and HLA genes encoding the major histocompatibility complex class II (MHC-II). MHC-II expression was concordantly reduced on the surface of DNMT3A-mutant HSPCs and dendritic cells. We next carried out single cell transcriptomics on individuals with DNMT3A-mutant CH. Consistent with the iPSC findings, DNMT3A-mutant HSPCs and dendritic cells had decreased expression of MHC-II genes compared to age-matched controls. These data implicate MHC-II as an axis regulated by common preleukemic DNMT3A mutations. We then extended this analysis to candidate cryptic drivers in our dataset. Of these, NSD1 and IRF1 were infrequently mutated (<1%) in patients with MDS and AML and previously implicated as regulators of HLA expression. To investigate how inactivating NSD1 and IRF1 mutations affect hematopoiesis, we performed knockdown (KD) of these genes in human CD34+ cord blood HSPCs. NSD1-KD expanded CD34+CD133+ HSPCs, while IRF1-KD blocked differentiation into myeloid lineages. Notably, both NSD1- and IRF1-KD displayed significant reduction of MHC-II on CD34+CD133+ HSPCs, suggesting that both common and cryptic drivers converge on MHC-II down-regulation. In conclusion, tracking preleukemic hematopoiesis by iPSC reprogramming, we present evidence that unannotated cryptic driver mutations are under positive selection in preleukemic evolution and nominate MHC-II reduction as convergent pathway by which common and cryptic drivers promote clonal fitness. We propose that downregulation of antigen presentation via MHC class II may be a common mechanism by which cryptic drivers contribute to clonal expansion.