One of the most common mutations in adult AML is a constitutively activating internal tandem duplication in the juxtamembrane domain of the Flt3 receptor (Flt3-ITD), which portends poor prognosis due to high recurrence rates and defines a distinct subtype of disease with unique features and biology. Although several FLT3 TKIs have been developed for clinical use, responses to these drugs, especially as single agents, are limited and are not sustained. The objective of our study was to determine the contribution of bone marrow stromal populations to LSC resistance to Flt3-targeted TKI in Flt3-ITD AML.

We utilized a newly generated Flt3-ITD TET2flox/flox Mx1-cre mouse model of AML, as well as primary Flt3-ITD TET2 mutant AML patient samples, to identify phenotypic populations with leukemia initiating capacity (LIC) in Flt3-ITD AML. In the animal model, administration of pIpC leads to deletion of TET2 and development of AML characterized by leukocytosis, accumulation of blasts, splenomegaly, anemia, thrombocytopenia, and lethality. Limiting dilution transplantation of FACS-sorted ST-HSC, MPP and GMP populations revealed that LIC were absent from GMPs and almost exclusively limited to the phenotypic ST-HSC population (calculated stem cell frequencies: <1:180,000, 1:63,635, and 1:2,730 for GMP, MPP, and ST-HSC, respectively). We similarly found that in samples from human Flt3-ITD TET2 mutant AML patients LIC capacity was restricted to primitive HSPC populations (Lin-CD34+CD38-), and was not seen in committed GMP (Lin-CD34+CD38+CD123+CD45RA+).

We characterized bone marrow stromal cells in Flt3-ITD AML mice by flow cytometry on collagenase digested bone fragments. We also transplanted murine AML cells into CXCL12-GFP mice to assess alterations in CXCL12-expressing stromal populations in AML bone marrow. We found expansion of several stromal populations in AML vs. WT mice, including a 3.5-fold increase in mesenchymal stem cells (CD45-Ter119-CD31-VECadherin-Sca1+CD51+) and a 1.5-fold increase in osteoprogenitors (CD45-Ter119-CD31-VECadherin-Sca1-CD51+). CXCL12 expression, however, was greater than 2-fold higher in osteoprogenitors and 2-fold lower in mesenchymal stem cells in AML vs. WT mice. We also showed that Flt3-ITD AML HSPCs have nearly 2-fold higher CXCR4 expression than WT HSPCs. These data taken together supported further exploration of the role of a CXCL12-expressing niche in supporting Flt3-ITD AML LSC.

To assess the effect of CXCL12 deletion from the marrow microenvironment on AML TKI response, we transplanted murine AML cells into CXCL12flox/flox UBC-cre mice and control Cre-ve mice. We found that AML developing in Cre-ve control mice was resistant to single agent Flt3 TKI (AC220, Quizartinib) treatment, but that CXCL12 deletion modestly improved response to TKI. We next tested a combination of standard-of-care "7+3" chemotherapy (cytarabine + doxorubicin) and AC220, and found that this approach resulted in more effective and selective, but only partial, reduction of leukemia cells in this model. We found that control AML mice showed an initial response to combination chemo + TKI, but developed disease recurrence by 3 weeks of treatment. In contrast, CXCL12-deleted AML mice maintained peripheral blood response for up to 3 weeks, and showed enhanced suppression of LIC-containing populations compared to control mice. We are now performing secondary transplants using BM cells harvested from these treated mice to assess long-term effects on leukemia stem cell capacity. We are also testing the effect of the combination of chemotherapy and TKI following osteoblast-specific deletion of CXCL12, using CXCL12flox/flox BGLAP-cre mice, to assess whether osteoblastic cells are the source of CXCL12 responsible for this effect.

In conclusion, our results suggest that LSC in Flt3-ITD AML are found within a primitive phenotypic ST-HSC population as opposed to GMP populations as seen in some other types of AML. Furthermore, CXCL12-expressing bone marrow microenvironmental cells contribute to drug resistance in AML LSC and global knockout of CXCL12 enhances drug response in these populations. Our studies support a potential role for a CXCL12-expressing osteoprogenitor niche in supporting Flt3-ITD AML LSC growth and drug resistance, targeting of which could improve responses and outcomes in Flt3-ITD AML.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.