Functional inhibition of osteoblastic cells in an in vivo mouse model of myeloid leukemia

Abstract

Pancytopenia is a major cause of morbidity in acute myeloid leukemia (AML), yet its cause is unclear. Normal osteoblastic cells have been shown to support hematopoiesis. To define the effects of leukemia on osteoblastic cells, we used an immunocompetent murine model of AML. Leukemic mice had inhibition of osteoblastic cells, with decreased serum levels of the bone formation marker osteocalcin. Osteoprogenitor cells and endosteal-lining osteopontin(+) cells were reduced, and osteocalcin mRNA in CD45(-) marrow cells was diminished. This resulted in severe loss of mineralized bone. Osteoclasts were only transiently increased without significant increases in bone resorption, and their inhibition only partially rescued leukemia-induced bone loss. In vitro data suggested that a leukemia-derived secreted factor inhibited osteoblastic cells. Because the chemokine CCL-3 was recently reported to inhibit osteoblastic function in myeloma, we tested its expression in our model and in AML patients. Consistent with its potential novel role in leukemic-dependent bone loss, CCL-3 mRNA was significantly increased in malignant marrow cells from leukemic mice and from samples from AML patients. Based on these results, we propose that therapeutic mitigation of leukemia-induced uncoupling of osteoblastic and osteoclastic cells may represent a novel approach to promote normal hematopoiesis in patients with myeloid neoplasms.

Figure 1

Murine model of bcCML. (A) Murine stem cell virus construct containing BCR/ABL and GFP. (B) MSCV construct containing Nup98/HoxA9 and YFP. (C) Schematic representation of the transplant strategy used to produce the leukemic mice used. (D) Flow cytometric gating strategy to identify leukemic cells as GFP⁺ and YFP⁺. (E) Representative anti-GFP immunohistochemistry of the femur's marrow space at the metaphysis. GFP is visualized by brown staining, with a hematoxylin counterstain. Top panel: 20× objective; bottom panel: 60× objective. (F) Flow cytometric data represent bcCML cells as a percentage of total marrow mononuclear cells. (G) Flow cytometric data representing bcCML cells as a percentage of total spleen. (H) Total peripheral blood mononuclear cells over the course of 10 days. *P ≤ .05. **P ≤ .01. ***P ≤ .001. n = 5 mice per time point. Bar represents SEM in this and subsequent experiments.

Figure 2

Leukemia decreases osteoblastic number and function. (A-B) Osteopontin immunohistochemistry was performed on paraffin-embedded sections. Representative images are shown of (A) a naive femur and (B) a leukemic femur. Left panels 20×, and right panels 60× objectives. Osteopontin⁺ cells are stained brown, and sections were counterstained with hematoxylin (blue). Arrowheads indicate osteopontin⁺ cells. (C) Quantification of serum osteocalcin measured by ELISA. (D) Real-time RT-PCR quantifying osteocalcin RNA expression in osteoblast-like cells collected from the long bones of normal or leukemic mice at day 6 or 11 and magnetically separated based on CD45 expression. Statistical significance was determined compared with naive mice (day 0). n = 5 samples per experimental group. (E-F) CFU-OBs formed per well from (E) whole marrow after 28 days in culture and (F) cells collected by collagenase digestion of bone fragments after 15 days in culture. **P ≤ .01. ***P ≤ .001.

Figure 3

Leukemic environment induces bone loss. (A-B) Representative images of H&E-stained paraffin sections of the distal femur from a (A) naive and (B) leukemic mouse 10 days after transplantation, 4× objective. (C-D) Representative micro-CT images from the metaphysis of the femur from (C) naive and (D) leukemic mice. (E) Micro-CT analysis of femur trabecular bone volume/total volume. (F) Femur cortical bone volume/total volume. (G) Trabecular number. (H) Trabecular thickness. (I) Trabecular spacing. *P ≤ .05. **P ≤ .01. n = 4 mice per experiment.

Figure 4

Leukemic environment mildly and transiently increases osteoclastic numbers in vivo. (A) Serum levels of carboxy-terminal collagen cross-link CTX, a marker of bone resorption as measured by ELISA. (B) Representative images of paraffin sections stained for the osteoclastic marker TRAP. TRAP⁺ cells are pink and highlighted by arrowheads. Top panels 20×, and bottom panels 60× objectives. (C) Quantification of multinucleated TRAP⁺ cells in a 1-mm² area just proximal of the distal growth plate in the femur from sections represented by the panels in B. (D) Serum levels of TRACP 5b, the osteoclast specific TRAP enzyme, measured by ELISA. *P ≤ .05. ***P ≤ .001. Each point indicates an individual mouse in this and subsequent experiments; n = 5 mice per experiment.

Figure 5

Leukemia cells do not differentiate into osteoclasts and do not resorb bone. Representative light micrographs of (A) normal and (B) leukemic spleen cells under osteoclastogenic conditions in vitro. Pink cells are positive for TRAP activity. (C) Quantification of TRAP⁺ cells in panels A and B. (D) Low and high power scanning electron micrographs of osteoclasts on bovine bone wafers. (E) Low and high power scanning electron micrographs of leukemia cells on bovine bone wafers. (F) ELISA quantification of CTX released into culture media during culture of cells with bovine bone wafers. (G-H) Representative light micrographs of cocultures containing osteoblasts and (G) normal marrow cells and (H) leukemic marrow cells. (I) Quantification of TRAP⁺ cells formed per well from osteoblastic cocultures with normal and leukemic marrows. *P ≤ .05. ***P ≤ .001. n = 3 or 4 mice per treatment group.

Figure 6

ZA rescues trabecular, but not cortical, bone loss. (A) Schematic for the treatment schedule of leukemic and normal mice with ZA. Leukemia was initiated on day 0 after 2 weeks of ZA treatment. Arrows indicate injection of ZA. (B) Serum CTX levels in normal mice after 2 weeks of treatment with ZA. (C) Serum CTX levels in normal and leukemic mice after the ZA treatment schedule. (D) Trabecular bone volume/total volume. (E) Cortical bone volume/total volume. *P < .05 (F) Trabecular number. (G) Trabecular spacing in normal or leukemic mice as quantified by micro-CT analysis after treatment with ZA. **P ≤ .01. ***P ≤ .001. (D-G) n = 4 mice per treatment group.

Figure 7

CCL-3 expression is increased in malignant cells from leukemic mice. (A) Top panel: Representative wells from CFU-OB cultures stained for alkaline phosphatase activity (pink). Bottom panel: CCL3 levels in culture media from CFU-OB cultures. (B) CCL3 protein levels in murine model of AML compared with normal controls. (C) Relative expression of CCL3 in bone marrow mononuclear cells isolated from whole bone marrow, cells sorted for GFP and YFP expression according to Figure 1D, and cells liberated from bone fragments by collagenase digestion and magnetically separated based on CD45 cell surface expression. (D) Relative expression of human CCL3 in primitive CD34⁺CD38⁻CD123⁺ AML cells compared with normal controls. Each bar represents a single AML sample normalized to 3 normal controls. (E) CCL3 protein levels in human AML patient marrow plasma. Each bar represents a single AML marrow sample compared with 7 normal controls. *P ≤ .05. ***P ≤ .001. (A-B,D) Each data point represents an individual mouse. (C) n = 3 mice per treatment group.