Specific Bone Marrow Niche Components Determine Degree of Protection from Gilteritinib Induced Differentiation Response in FLT3-ITD AML

Background: the FLT3 tyrosine kinase inhibitor (TKI) gilteritinib relieves the differentiation blockade in around 50% of acute myeloid leukaemia (AML) patients carrying FLT3 internal tandem duplication (ITD) mutations. However, differentiation responses are often incomplete and are poorly understood. This limited understanding is partly due to the strong cytotoxic effects seen with gilteritinib in cellular in vitro models and suggests that the in vivo bone marrow microenvironment (BMM) plays roles in modulating the differentiation response. Recent evidence suggests that the BMM induces changes to the signalling and metabolism of FLT3-ITD AML cells (Joshi et al, 2021). However, links between specific bone marrow cells, gilteritinib induced differentiation in FLT3-ITD AML and the underlying mechanisms have not been explored.

Aims: to uncover the degree and mechanisms by which specific BMM cellular components modulate differentiation in response to gilteritinib in FLT3-ITD AML.

Methods: we treated a murine model of NPM1c/FLT3-ITD AML and a patient derived xenograft (PDX) with gilteritinib and determined responses in different bone marrow regions. We designed manipulable in vitro co-culture systems of human AML cell lines and primary cells with several cellular components of BMM derived from both stromal cell lines and healthy donor mesenchymal stromal cells. Following mono- and co-culture experiments, we performed viability, immunophenotypic, transcriptional and metabolic analysis.

Results: treatment of NPM1c/FLT3-ITD murine leukaemia with gilteritinib reduces the primitive pool (Lineage^- cKit⁺ Sca-1⁺) and increases the proportion and number of cells expressing lineage markers, indicative of a differentiation response. This results in a moderate but sub-optimal increase in the survival of gilteritinib treated mice reminiscent of responses seen in patients. Analysis of AML immunophenotypes in different BM regions reveals that in the highly haematopoietic adipocyte sparse thoracic vertebrae gilteritinib induces a strong differentiation response. However, in the highly adipocytic tail vertebrae upon gilteritinib treatment the proportion of quiescent primitive cells remains the same, revealing adipocyte protection of the leukaemia stem cell pool. A similar effect is also observed in a gilteritinib treated PDX. Furthermore, in vitro adipocytes protect FLT3-ITD AML cells from death and block induction of myeloid differentiation to a greater extent than other BMM cellular components. As such, we focused on adipocyte driven protection and show that, while this is partly retained in transwell assays, it is most potent upon direct contact. Adipocyte conditioned media is not protective, suggesting that cellular proximity and/or reprogramming of adipocytes by AML cells is needed for protection. Transcriptional profiling of co-cultured adipocytes and AML cells and cytokine analysis of spent media suggest that metabolic reprogramming contributes to protection. Indeed, metabolic profiling of AML cells reveals that adipocyte co-culture rescues the abundance of tricarboxylic acid (TCA) cycle intermediates following gilteritinib treatment. Uniformly labelled ¹³Carbon₆ glucose (U¹³C₆-Gluc) tracing shows that the increase of TCA cycle intermediates in co-culture is caused both by unlabelled and U¹³C₆-Gluc labelled carbon sources. This is likely due to fatty acids contribution to the unlabelled pool, as previously described (Ye et al, 2016) and indeed in vitro adipocyte labelling confirms transfer from adipocytes to AML cells. Moreover, metabolic analysis of co-culture media and AML cells following adipocyte U¹³C₆-Gluc pre-labelling show presence of adipocyte derived labelled glycolytic and TCA cycle intermediates both in media and AML cells. This suggests that adipocytic derived polar metabolites also provide an important carbon source to AML cells in co-culture.

Conclusion: the degree of gilteritinib induced differentiation in FLT3-ITD AML is dependent on specific components of the BMM. The adipocytic BM niche protects against gilteritinib induced differentiation and this correlates with metabolic reprogramming of AML cells, driven by both fatty acid and glucose derived carbon sources. These findings offer opportunities for therapeutic interventions aimed at overcoming BM stromal protection and promoting positive responses following FLT3-TKI therapy.