It's the proteases, stupid!

Despite its clinical importance, no intellectually satisfying mechanistic explanations have been provided for G-CSF–induced stem cell mobilization. Insights, however, from studies addressing the mechanism of IL-8–induced mobilization in normal or G-CSF–receptor deficient mice have been obtained. A pool of normal neutrophils and their liberated proteases are found to be critical both in G-CSF– and IL-8–induced mobilization, bolstering the concept that diffusible factors are responsible for mobilization. As a result of neutrophil activation by G-CSF, increases in circulating elastase (NE), cathepsin G (CG), and matrix metalloproteinase 9 (MMP9) have been documented after G-CSF mobilization (Van Os et al, Blood. 1999;94(suppl):637a). But the mode of action of these proteases in vivo, their specific substrates, and their interplay with other molecules was not previously addressed.

Lévesque and colleagues (page 1289) present compelling evidence that integrates the function of proteases and of adhesion molecules in G-CSF–induced mobilization. NE and CG can cleave vascular cell adhesion molecule-1 (VCAM-1) within the confines of bone marrow (BM). The sequestered BM microenvironment with its tight cell-cell and cell-matrix interactions and membrane presentation of proteases provides a favorable environment for effective proteolytic activity evading the action of inhibitors. A progressive decrease in bone marrow VCAM-1 levels and concomitant increases in sVCAM-1 in circulation, together with increases in serine proteases is corroborating evidence. The specificity of NE and CG on BM VCAM-1 cleavage, but not of other proteases tested, is of interest, especially when VCAM-1 in other tissues is cleaved by activated MMPs instead (Hummel et al, J Neuropathol Exp Neurol. 2001;60:320-327). Despite this enlightening scenario, lingering questions remain. To what extent do mice deficient in NE or CG or dipeptidyl peptidase I (DPPI) (Pham et al, Proc Natl Acad Sci U S A. 1999;96: 8627-8632) mobilize in response to G-CSF? Are other targets besides VCAM-1 involved? Why is soluble VCAM-1 increased after SCF or IL-3 treatment? If VCAM-1 shedding occurs fairly quickly, why does progenitor mobilization lag for 2 or 3 days after G-CSF administration? Despite these caveats, the present data open a new window of observation and may eventually suggest new strategies to improve mobilization.