Resistance to Lenalidomide in Del(5q) MDS Is Mediated By Inhibition of Drug-Induced Megakaryocytic Differentiation

The immunomodulatory drug lenalidomide (LEN) is the treatment of choice for del(5q) MDS patients. LEN has been shown to trigger the specific degradation of CSNK1A1 and IKZF1 proteins after binding the E3-ligase substrate adaptor CRBN. When brought below a certain expression threshold, CSNK1A1 deficiency activates a p53-dependent apoptotic response. Thus, the unique sensitivity of del(5q) cells to LEN is explained by CSNK1A1 haploinsufficiency in del(5q) MDS patients. Despite its efficacy, 50% of LEN-treated patients eventually relapse within an interval of 2-3 years after treatment. Treatment failure is associated to low platelet counts and occurrence of additional mutations, such as TP53. To identify novel genetic determinants of LEN resistance, we have compared whole genome sequencing data of paired samples from six del(5q) patients who have been treated with LEN and eventually became resistant to the treatment. We identified 2 patients with mutations in TP53. The remaining four presented RUNX1 alterations: two patients had protein coding mutations in RUNX1 and two had a significant reduction in RUNX1, but not TP53, transcript levels.

As a model of sensitivity, we studied the response to LEN in two human del(5q) cell lines, MDS-L and KG-1a. RUNX1 protein levels are postranscriptionally upregulated upon exposure to LEN, accompanied by increased levels of RUNX1 activity. Deletion of CRBN expression cancelled these effects. RUNX1 overexpression inhibited clonogenic growth and induced apoptosis. We then generated RUNX1 knock-out (KO) clones derived from MDS-L cells using CRISPR/Cas9 system. RUNX1 KO cells presented increased proliferation, increased colony growth and reduced apoptosis in the presence of LEN compared to wild-type (WT) control clones. These results were validated with different shRNAs against RUNX1. Genetic rescue experiments showed that RUNX1 mutants were unable to restore sensitivity to the drug compared to RUNX1 WT. Finally, modeling RUNX1 loss-of-function (LOF) in CSNK1A-depleted human CD34⁺ cells abrogated the effects of LEN on colony forming cell assays. Thus, RUNX1 function is required for the elimination of del(5q) cells by LEN.

To understand the molecular mechanisms underlying the resistant phenotype, we performed RNA-seq on MDS-L cells treated with LEN for 24h. We observed a significant upregulation of Platelet specific genes (ITGB3, ITGA2B, VWF, THBD, SELP, TREML1, GATA1) coupled to downregulation of Cell Cycle genes (E2F2, E2F1, MCM5, CDKN1A), suggesting that LEN induces differentiation in to the Megakaryocytic (Meg) lineage. We found a significant upregulation of CD41⁺/CD61⁺ double positive cells after LEN exposure in vitro and in vivo, associated to the appearance of multinucleated cells. Importantly, the apoptotic response was associated to the emergence of the differentiating population. At the molecular level, CRBN is required for LEN-induced differentiation. Further downstream we identified IKZF1 degradation as key trigger, as IKZF1 overexpression restrained Meg differentiation and a IKZF1 dominant negative isoform enhanced it. In contrast, CSNK1A overexpression did not alter differentiation after LEN, but did reduce apoptotic induction. Moreover, we identified GATA2 targets enriched in LEN-regulated genes and showed that GATA2 overexpression or downregulation using shRNAs significantly increased or reduced LEN induced differentiation respectively.

Finally, gene expression analysis after LEN exposure showed that Meg signatures were not enriched in resistant RUNX1 KO cells compared to WT control. Accordingly, RUNX1 KO cells did not undergo differentiation upon LEN exposure. RUNX1 LOF in CSNK1A-depleted primary human CD34⁺ cells blocked CFU-Mk growth in LEN treated cells. GATA2 overexpression was unable to restore LEN effects in RUNX1 deficient cells, suggesting a cooperative mechanism between both transcription factors. Luciferase assays using the human CD41 promoter showed that RUNX1 mutants reduced promoter transactivation compared to RUNX1 WT. Remarkably, we observed a similar phenotype for LEN-resistant TP53 KO cells.

As a conclusion, our results suggest that GATA2, RUNX1 and TP53 cooperate to drive Meg differentiation after LEN-mediated degradation of IKZF1 protein. Loss of function mutations affecting RUNX1 or TP53 alter the activity of GATA2 transcriptional complex, rendering del(5q) cells unresponsive to LEN.