G-CSF Disrupts the Stem Cell Niche by Increasing Turnover of Bone Marrow Osteoblasts.

Accumulating evidence suggests that osteoblast-lineage cells play a key role in supporting hematopoiesis, providing signals that maintain the normal function of hematopoietic stem cells (HSC). We and others previously showed that treatment with G-CSF reduces the number and activity of osteoblasts in the bone marrow. Here we confirm these findings using transgenic mice expressing GFP in osteoblast lineage cells under control of a 2.3 kb fragment of the col1a1 promoter. Analysis of these mice confirmed that G-CSF treatment reduced the number of osteoblasts 2.9-fold compared to control mice (n=4, p<.01). Consistent with these findings, expression of several osteoblast genes was sharply reduced during G-CSF treatment; alkaline phosphatase, osteocalcin, stromal-derived factor 1, osteoprotegerin, and bone sialoprotein were all decreased (7.9, 19, 3.8, 14.2, and 4.1-fold decreased; n=5–12 each group, p<.05). These findings are consistent with either a true loss of osteoblasts or induction of osteoblast quiescence. To distinguish between these possibilities, the fate of osteoblasts during G-CSF treatment was assessed by in vivo BrdU labeling. At baseline, 34.9 ± .8% of osteoblasts were labeled after 16 days of BrdU administration. In untreated mice, this percentage decreased to 14.8% ± 0.05% after 9 days. However, in mice harvested 4 days after a 5-day course of G-CSF, the percentage of BrdU positive osteoblasts decreased to 6.5% ± 0.9% (n=2 each group, p<.05). These data suggest that G-CSF treatment increases the turnover of osteoblasts rather than inducing their quiescence. Since apoptosis is one potential mechanism of osteoblast turnover, we measured caspase 3 activation in G-CSF-treated mice. No increased apoptosis was observed [percent activated caspase 3 positive cells + SEM: 8.0% ± 1.5% (untreated),10.5% ± 0.5% (day 3), 5.8% ± 1.2% (day 4); n=2–4 each group, p=ns]. The sharp decrease in osteoblast number suggests that HSC activity might be altered by G-CSF treatment. To address this possibility, competitive repopulation assays were performed with HSC harvested from G-CSF or control mice. In mice competitively reconstituted with control HSC, the percentage of donor cells was 33.7% ± 4.7% at 4 months after transplantation. In contrast, in mice receiving G-CSF treated HSC, the percentage of donor cells was only 10.0% ± 2.6% (n=3–4 each group, p<.05). As this loss in repopulation potential may result from decreased HSC number or function, we estimated HSC number by measuring Kit⁺ lineage⁻ Sca⁺ CD34⁻ cells. We found no significant difference in this fraction between treated and untreated mice [number KLS CD34⁻ per femur + SEM: 966 ± 229 (untreated) versus 945 ± 183 (treated); n=6–7 each group, p=ns]. These results suggest that the diminished repopulation potential of G-CSF treated bone marrow results from impaired HSC function rather than reduced HSC number. Finally, we hypothesized that the loss of long-term repopulation potential in HSC from G-CSF treated mice may result from the loss of factors in the bone marrow niche that maintain HSC quiescence. We therefore measured BrdU uptake in the KLS fraction of bone marrow cells. Preliminary results show a trend toward increased BrdU uptake in KLS cells from G-CSF treated mice (48.5% ± 9.2%) compared with untreated mice (26.5% ± 1.4%; n=2 each group, p=.14). In summary, these results suggest a model in which G-CSF induces increased osteoblast turnover, disruption of the stem cell microenvironment, and loss of HSC quiescence.