Acute myelogenous leukemia-induced sympathetic neuropathy promotes malignancy in an altered hematopoietic stem cell niche

Abstract

Perivascular mesenchymal stem and progenitor cells (MSPCs) are critical for forming a healthy hematopoietic stem cell (HSC) niche. However, the interactions and influence of acute myelogenous leukemia (AML) stem cells with the microenvironment remain largely unexplored. We have unexpectedly found that neuropathy of the sympathetic nervous system (SNS) promotes leukemic bone marrow infiltration in an MLL-AF9 AML model. Development of AML disrupts SNS nerves and the quiescence of Nestin(+) niche cells, leading to an expansion of phenotypic MSPCs primed for osteoblastic differentiation at the expense of HSC-maintaining NG2(+) periarteriolar niche cells. Adrenergic signaling promoting leukemogenesis is transduced by the β2, but not β3, adrenergic receptor expressed on stromal cells of leukemic bone marrow. These results indicate that sympathetic neuropathy may represent a mechanism for the malignancy in order to co-opt the microenvironment and suggest separate mesenchymal niche activities for malignant and healthy hematopoietic stem cells in the bone marrow.

Figure 1. Sympathetic neuropathy promotes leukemogenesis

(A) Gating strategy for flow cytometry analysis of LSC/L-GMP. (B) Absolute numbers of L-GMP per femur in control and denervated leukemic mice, 20 days after transplantation (normalized to control, n=19–20). (C) Absolute numbers of L-GMP per ml blood (left) and spleen (right) in control and denervated leukemic mice, 23 days after transplantation (normalized to control, n=4–5). (D) Survival curve of control and denervated leukemic mice (n=5). (E) Left, flow cytometry gating strategy for bone marrow analysis of human hematopoietic engraftment by gating on human (h) CD45⁺ cells, detecting exclusively myeloid hCD33⁺ cells, excluding hCD3⁺ and hCD19⁺ expression (not shown). Representative flow cytometry plots from each experimental condition (BMT=bone marrow transplantation). Right, human myeloid BM engraftment 4 weeks after transplantation of primary human AML cells in control or denervated NSG mice (data are normalized to its paired control, n=4 human AML samples). (F) Top, Z-stack confocal images from bone marrow, spleen and cremaster muscle stained for PECAM1⁺ endothelial cells and TH⁺ nerve fibers. Scale bar: 10 μm. Bottom, assessment of the TH⁺ fiber density per arteriole by quantifying the total length of all TH⁺ branches divided by the area of the corresponding arteriole (bone marrow: n=33–49 arterioles from 6–8 mice per group; spleen n=21–29 arterioles from 5 mice per group; cremaster muscle: n=16–17 arterioles from 6 mice per group). *P<0.05, **P<0.01 determined by Student’s t test. Data are shown as mean ± s.e.m. See also Figure S1.

Figure 2. Bone marrow MSPCs and endothelial cells significantly expand in AML

(A) Left, representative flow cytometry plots, gated on stromal (CD45⁻Ter119⁻CD11b⁻CD31⁻) bone marrow cells, showing Nes-GFP⁺ cells in control and leukemic mice. Middle and right, frequency and absolute numbers of Nes-GFP⁺ cells per femur (n=9–10). (B) Summary of mesenchymal surface marker screening expressed by stromal Nes-GFP⁺ bone marrow cells from leukemic mice (red columns) and healthy controls (grey columns), as detected by flow cytometry analysis (n=3–10). (C) Representative flow cytometry plots and quantification of PDGFRα and CD51 double-positive bone marrow stromal cells (n=9–15). (D) Cell cycle analysis of PDGFRα and CD51 double-positive bone marrow stromal cells, representative plots and quantification by flow cytometry with anti-Ki67 and Hoechst 33342 staining (n=3). (E) Top, Z-stack confocal images of thick bone sections stained with anti-Perilipin antibody and Hoechst 33342. Scale bar: 300 μm. Bottom, frequency of PDGFRα and CD51 double-positive bone marrow stromal cells (left, normalized to control) and quantification of Perilipin⁺ adipocytes in 0.5mm² area under the growth plate (right) 6 days after sublethal irradiation (n=4). (F) Left, frequency of non-hematopoietic (CD45⁻Ter119⁻) CD31⁺ endothelial cells (n=6). Right, Z-stack confocal images of thick bone sections stained in vivo with anti-PECAM1 and VE-cadherin antibodies. The magnified confocal images within the area were defined by the rectangle. Scale bar: 500 μm. (G) Cell cycle analysis of endothelial cells quantified by flow cytometry with anti-Ki67 and Hoechst 33342 staining (n=3). *P<0.05, **P<0.01, ***P<0.001 determined by Student’s t test. Data are shown as mean ± s.e.m.

Figure 3. Bone marrow Nes-GFP⁺ MSPCs have differentiated toward the osteoblastic lineage

(A-C) Stromal Nes-GFP⁺ cells were sorted from control and leukemic bone marrow and plated at equal numbers at clonal densities under CFU-F and CFU-OB culture conditions. Frequency of CFU-F (n=6–7, in duplicate/triplicate per mouse) (A) and CFU-OB (n=4–5, in triplicate per mouse) (B) from bone marrow Nes-GFP⁺ cells. (C) Representative images of CFU-OB colonies from Nes-GFP⁺ bone marrow cells, stained with alkaline phosphatase and von Kossa and counterstained with hematoxylin. Scale bar: 500 μm. (D) Absolute numbers of stromal CD51⁺/Sca-1⁻ osteolineage cells in the compact bone (normalized to control, n=5–8). (E) Quantification of TRAP⁺ osteoclasts in the metaphyseal area (500 μm under the growth plate area) of the tibia. Number of TRAP⁺ cells in relation to the measured bone surface (n=3). (F) Z-stack confocal images of thick bone sections of Osterix-cre^ERT2/loxp-tdTomato control and leukemic mice. Middle panel shows the magnified confocal images within the area defined by the rectangle. Arrowhead indicates osteoblast precursors. Anti-PECAM1 and VE-cadherin antibodies in vivo. Scale bar: 500 μm. (G and H) Z-stack confocal images of thick bone sections of Osterix-cre^ERT2/loxp-tdTomato control (G) and leukemic (H) mice stained with anti-LepR antibody (arrows denote LepR-expressing Osx-cre/tomato⁺ cells). Scale bar: 20 μm. (I) Z-stack confocal images of thick bone sections of Osterix-cre^ERT2/loxp-tdTomato control (left) and leukemic (right) mice stained with osteocalcin antibody. Scale bar: 50 μm. (J) Micro-CT analysis of femurs from control and leukemic mice (n=3). Analysis of cortical (left) and trabecular (middle) bone volume/total volume (BV/TV) as well as trabecular number (right). Representative micro-CT images of cortical (K) and trabecular (L) bone from control and leukemic mice. *P<0.05, **P<0.01 determined by Student’s t test. Data are shown as mean ± s.e.m. See also Figure S2.

Figure 4. Leukemic bone marrow niche has impaired HSC regulating capacity and regulates LSCs through the β2-adrenergic receptor

(A) Gene expression analysis of key HSC regulatory genes (Vcam1, Cxcl12, Angpt1, Scf and Opn) in sorted bone marrow PDGFRα⁺/CD51⁺ stromal cells by real-time PCR (n=5–6). (B) Frequency of stromal NG2DsRed⁺ cells in the bone marrow (normalized to control, n=6–7). (C) Frequency of phenotypic Lineage⁻Sca-1⁺c-kit⁺Flt3⁻CD34⁻ long-term HSCs in leukemic mice with 86.5% mean bone marrow infiltration (early) and >95% bone marrow infiltration (late) compared to matched control mice (n=10, 5, 8). (D) Quantification of long-term reconstituting HSCs by long-term culture-initiating cell (LTC-IC) assay on sorted GFP⁻ cells isolated from control and leukemic mice. Estimated LTC-IC frequency is given; dashed lines represent 95% confidence interval. (E) LTC-IC numbers per femur, calculated with the frequency of MLL-AF9 GFP⁻ cells in the bone marrow (n=3–5). (F) Absolute numbers of Lineage⁻Sca-1⁺c-kit⁺Flt3⁻ (LSKF) cells in peripheral blood (left) and spleen (right) (n=4–5) in leukemic (mean bone marrow infiltration 86.5%) and matched control mice. (G) Left, colony-forming unit in culture (CFU-C) from 5×10⁴ sorted MLL-AF9 GFP⁻ bone marrow (BM) cells. Right, CFU-C numbers per femur, calculated with the frequency of MLL-AF9 GFP⁻cells (n=5–6). (H) Representative whole-mount images and distribution of HSCs in the sternal bone marrow relative to Nes-GFP^bright arterioles. n= 54, 75 HSCs per control and AML (late stage) group respectively. Arrowheads denote HSCs. Two-sample Kolmogorov-Smirnov test, P=0.04. Scale bar: 10 μm. (I) Absolute numbers of L-GMP per femur in mice treated with the Adrβ3-inhibitor (SR59230A), Adrβ2-inhibitor (ICI118,551) and control leukemic mice, 18 days after transplantation (normalized to control, n=14–15). (J) Survival curve of mice treated with ICI118,551 and control mice (n=5). (K) Frequency of leukemic cells per femur in Adrβ2^−/− and control mice (n=5–7). (L) Absolute numbers of L-GMP per femur in mice treated with the Adrβ2-agonist (Clenbuterol hydrochloride) and control leukemic mice, 19 days after transplantation (normalized to control, n=7–8). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 determined by Student’s t test. Data are shown as mean ± s.e.m. See also Figure S3 and S4.