The age-related clonal expansion of human hematopoietic stem and progenitor cells (HSPCs) is termed age-related clonal hematopoiesis (ARCH) and is associated with the accumulation of recurrent somatic mutations. As the acquisition of some of these mutations by HSPCs precede myeloid malignancies, they are considered to be pre-leukemic mutations (pLMs). While the study of ARCH has been mainly focused on the phenotypic consequences of pLMs, the mutational mechanisms driving pLMs in leukemia’s cells of origin are less characterized. As some of the pLMs were shown to be the first events in leukemia evolution, understanding the biological scenarios in which they occur may shed an important light on the initial steps of human leukemogenesis. In the current work, we aimed to identify recurrent deletion signatures in myeloid malignancies and to study the mutagenic processes promoting them.
To this end, we used publicly available large sequencing datasets of myeloid malignancies. Inspecting the genomic sequences around the most common somatic deletions, we identified a common signature for the three most common deletions in myeloid malignancies which occur in CALR, ASXL1 and SRSF2 genes. This signature involves the joining of two pre-existing identical sequences flanking the deletions. As these identical sequences are referred to as microhomologies (MHs), we termed the deletions carrying this signature ‘MH-based deletions’.
As we were aiming to characterize the biological processes driving these deletions, we first aimed to identify the cell of origin in which they occur. To this end we used primary AML cells harboring MH-based deletions in ASXL1 and SRSF2 genes, together with myelofibrosis cells containing the CALR MH-based deletion. We detected the MH-based deletions in purified T-cells derived from all of these samples, suggesting that these deletions originally occurred is multipotent hematopoietic stem cells (HSCs) capable of differentiating into both myeloid and lymphoid lineages. We used additional methods to validate these results, and concluded that recurrent MH-based deletions originate in early HSCs and therefore are pre-leukemic and part of clonal hematopoiesis.
As we suspected that MH-based deletions might be the result of mutagenic double strand break (DSB) repair in early HSCs, we aimed to model DSB repair around CALR, ASXL1 and SRSF2 hotspot by using the CRISPR/Cas9 system. To study the DSB repair around these hotspots, we analyzed the indel landscape following targeted DSBs in K562 cells. This revealed that DSBs at specific genomic positions in ASXL1 and SRSF2 led to the formation of the MH-based deletions in these genes in K562 cells, recapitulating the mutational mechanisms naturally occurring in pre-leukemic HSCs. Introduction of DSBs in multiple AML cell lines and primary HSPCs successfully validated the results in many different cell types, suggesting an underlining DSB repair machinery which is probably evolutionary conserved.
We therefore used our K562 based model system to study the repair machinery involved in the formation of the recurrent MH-based deletions. DSB induction in LIG4 knockout K562 cells and in cells that were pre-treated with PARP1 inhibitors, provided evidence that the recurrent MH-based deletions are the result of PARP1 mediated and LIG4 independent repair. This DSB repair, known as microhomology-mediated end joining (MMEJ), led us to refer to the recurrent MH-based deletions originating in pre-leukemic HSCs as ‘preL-MMEJ deletions’. As strand synthesis during MMEJ repair was previously shown to be carried out by the DNA polymerase POLQ, we aimed to study POLQ involvement in the generation of preL-MMEJ deletions.
Unexpectedly, POLQ knockout K562 cells successfully generated preL-MMEJ deletions in ASXL1 and SRSF2 in our CRISPR/Cas9 based system. We therefore concluded that the DSB repair responsible for preL-MMEJ deletions is POLQ independent and hypothesized that other DNA polymerases should be involved. In order to identify such a polymerase, we analyzed single bone-marrow CD34+ gene expression profiles from the Human Cell Atlas Consortium’s immune census dataset. Assessment for a possible correlation between the expression levels of PARP1 and a list of human polymerases in early HSCs exposed a correlation between PARP1 and POLQ but also POLD1 and POLE gene expression levels. These genes encode for the major replicative DNA polymerases in eukaryotes. In order to experimentally model the inhibition of these polymerases, we used aphidicolin, a potent replicative DNA polymerase inhibitor. Aphidicolin treated K562 cells prior to DSB induction had significantly reduced fractions of preL-MMEJ deletions in both ASXL1 and SRSF2 loci compared to control cells. This highlights the replication associated nature of the MMEJ sub-pathway promoting preL-MMEJ deletions.
Altogether, in this study we aimed at understanding the biological processes driving early mutations in myeloid malignancies. We identified a shared deletion signatures among the most common deletions in myeloid malignancies and provided evidence that these deletions originate in early multipotent HSCs. Moreover, by applying a CRISPR/Cas9 based model system, we successfully recapitulated the preL-MMEJ deletions and characterized the repair machinery involved in their generation. We demonstrate that this DSB repair sub-pathway is dependent on PARP1 and replication associated DNA polymerases, while it is LIG4 and POLQ independent. Collectively, our findings provide novel insights into the mutational mechanisms involved in early stages of clonal hematopoiesis. A further characterization of this MMEJ sub-pathway, which drives somatic mutagenesis in early HSCs, is required.