RAS mutations are frequent in extramedullary AML and drive the tissue infiltration of leukemic cells via the induction of jaml

Abstract

Abstract Extramedullary acute myeloid leukemia (EML) is a rare manifestation of acute myeloid leukemia (AML), where myeloid blasts invade extrahematopoietic tissues. Recent data highlight a negative prognostic impact of EML. The knowledge about the molecular pathogenesis behind EML is insufficient and mainly derived from the analysis of corresponding bone marrow (BM) manifestations. We aimed to shed more light on the molecular pathogenesis of EML and performed next-generation sequencing (NGS) of 49 myeloid neoplasm-associated genes in paraffin-embedded EML biopsies of 76 patients. Our analyses validated the high frequency of mutations in NPM1 (26%), DNMT3A (21%), TET2 (19%), and FLT3 (16%). In addition, 43% of patients exhibited one or more mutations in RAS or RAS-modifying genes (RASmut; NRAS, KRAS, PTPN11, CBL, NF1). The frequency of RASmutwas significantly higher in EML biopsies as compared to BM specimens from four unselected AML cohorts comprising more than 1400 patients (cohort Graz, 81/278 [29%], P=0.015; cohort TCGA-LAML, 34/200 [17%], P<0.001; cohort Beat-AML, 217/785 [28%], P=0.023; cohort TARGET-AML, 40/150 [27%], P=0.035). Paired NGS results of EML biopsies and affected BM of the corresponding patients were available in 22 cases. Of these, four patients developed a RASmut in the EML specimen that was not present in the BM. Furthermore, the normalized variant allele frequencies (VAF) of RASmut in the EML specimen were significantly higher compared to those in the corresponding AML BM (70% vs 40%, P=0.002). To clarify the functional role of RASmut in the tissue infiltration of myeloid blasts and EML formation. Therefore, we introduced the NRASG12D mutation in the RAS-wildtype (WT) myeloid leukemia cell line K562 by employing the CRISPR/Cas9 knock-in strategy to establish an isogenic cell line model. This mutation was chosen as it was the most frequent alteration detected in the primary EML specimens. In-vitro, K562-NRASG12D showed higher rates of migration (P=0.034) and invasion (P=0.035) compared to K562-NRASWTcells, with no changes in proliferation (P=0.287). To exclude cell line specific effects, and ensure that the findings are also applicable to other RASmut (beyond NRASG12D), these analyses were validated in another cell line model (HEL-NRASG12D vs HEL-NRASWT; P=0.02 for migration and P=0.005 for invasion) and murine hematopoietic progenitors carrying KrasG12D (Mx1-Cre/KrasG12D vs Mx1-Cre/KrasWt; P=0.01 for migration and P=0.004 for invasion). We then validated the effects of RASmut on EML formation and invasion in ex-ovo chorioallantoic membrane (CAM) assays in chicken embryos. In agreement with the in-vitro assays, K562-NRASG12D invaded significantly better and formed larger tumors in the CAM than K562-NRASWT(P=0.0246). We then aimed to validate these data in-vivo and injected the K562 cells subcutaneously into immunocompromised NRG nude mice. This way of administration was preferred over the intravenous route as we aimed to transfer the cells into an extramedullary environment. Again, K562-NRASG12D formed significantly bigger tumors than K562-NRASWT(P=0.011). To delineate the mechanisms behind RASmut-driven leukemic tissue infiltration, we performed RNA sequencing of the subcutaneous NRASG12D and NRASWT tumors. This analysis revealed 244 differentially expressed genes with the majority (n=185) showing decreased expression in NRASG12D tumors. As we were mainly interested in identifying candidate genes amenable to direct therapeutic targeting, we focused on the 59 genes with increased expression in the NRASG12D-mutated tumors. Candidate selection via comparison with the Beat-AML cohort and literature review, and validation by qPCR and Immunoblot revealed the junctional adhesion molecule-like protein (JAML) as the most promising target. Indeed, the causative role of JAML expression could be corroborated by repeating the migration assays, where siRNA-mediated JAML knockdown in K562-NRASG12D significantly decreased the migration potential of these cells (P=0.007).Taken together, we delineate the molecular landscape of EML and show that RASmut are associated with the development of this AML subform. We further show that RASmut are functionally involved in the tissue infiltration of leukemic blasts and EML formation. Finally, we show that upregulation of JAML is a central step in RASmut-mediated EML formation and represents an interesting therapeutic target for RASmut EML manifestations.