G-CSF induces E-selectin ligand expression on human myeloid cells Nilesh M Dagia 1,2 , Samah Z Gadhoum 1,2 , Christine A Knoblauch 1,2 , Joel A Spencer 3 , Parisa Zamiri 3 , Charles P Lin 3 & Robert Sackstein 1,2,4,5 Clinical use of G-CSF can result in vascular and inflammatory complications 1–7 . To investigate the molecular basis of these effects, we analyzed the adherence of G-CSF–mobilized human peripheral blood leukocytes (ML) to inflamed (TNF-a stimulated) vascular endothelium. Studies using parallel plate assays under physiologic flow conditions and intravital microscopy in a mouse inflammation model each showed that ML take part in heightened adhesive interactions with endothelium compared to unmobilized (native) blood leukocytes, mediated by markedly increased E-selectin receptor-ligand interactions. Biochemical studies showed that ML express the potent E-selectin ligand HCELL (ref. 8) and another, previously unrecognized B65-kDa E-selectin ligand, and possess enhanced levels of transcripts encoding glycosyltransferases (ST3GalIV, FucT-IV and FucT-VII) conferring glycan modifications associated with E-selectin ligand activity. Enzymatic treatments and physiologic binding assays showed that HCELL and the B65-kDa E-selectin ligand contribute prominently to the observed G-CSF–induced myeloid cell adhesion to inflamed endothelium. Treatment of normal human bone marrow cells with a pharmacokinetically relevant concentration of G-CSF in vitro 9,10 resulted in increased expression of these two molecules, coincident with increased transcripts encoding pertinent glycosyltransferases and heightened E-selectin binding. These findings provide direct evidence for a role of G-CSF in the induction of E-selectin ligands on myeloid cells, thus providing mechanistic insight into the pathobiology of G-CSF complications. G-CSF is widely used clinically to augment neutrophil recovery after myelosuppressive chemotherapy, radiotherapy or both, and to mobi- lize bone marrow hematopoietic progenitors for use in hematopoietic stem cell transplantation (HSCT) 11 . Though it has generally been considered safe, there are increasing observations that G-CSF admin- istration can promote leukocyte-endothelial adhesive interactions resulting in vascular and inflammatory complications 1–7 . Indeed, G-CSF administration has been shown to (i) recruit neutrophils to lung vasculature, resulting in respiratory distress syndrome 3 ; (ii) compromise cardiac perfusion, resulting in angina pectoris and myocardial infarct 2,4 ; (iii) cause neutrophil infiltration in dermal vessels, leading to development of cutaneous leukocytoclastic vasculitis 5 ; (iv) intensify arthritic symptoms 6 and (v) precipitate sickle-cell vaso-occlusion 7 . A better understanding of the molecular basis of the heightened leukocyte-endothelial interactions accompany- ing clinical G-CSF administration could yield strategies to prevent these complications. Under physiologic blood flow conditions, leukocytes initially make contact on the vessel surface by engagement of counter-receptors for relevant endothelial molecules that mediate shear-resistant interac- tions 12 . One of the principal effectors of these interactions is E-selectin, which is an inducible endothelial molecule expressed at sites of inflammation 12 that binds sialofucosylated carbohydrate ligands expressed on leukocytes 13 . An expanding body of evidence causally links upregulated E-selectin expression to vascular complica- tions of G-CSF administration 14–16 . Notably, the receptor for G-CSF is expressed on endothelium 17 , and G-CSF directly induces E-selectin expression on endothelial cells in culture 18 . However, there is little information on whether G-CSF administration modifies E-selectin ligand expression on mobilized, circulating leukocytes. To address this issue, we analyzed adhesive interactions of ML obtained from donors undergoing pheresis for HSCT and of native human peripheral blood leukocytes (NL) on TNF-a–stimulated human umbilical vein endothelial cells (HUVECs) in a parallel-plate flow chamber assay. Under hemodynamic flow conditions, ML showed markedly greater endothelial adhesive interactions than NL on stimulated HUVECs; these interactions were abrogated by a function-blocking monoclonal antibody (mAb) to E-selectin and by chelation of calcium (Fig. 1a and Supplementary Fig. 1). Moreover, ML rolled distinctly more slowly than NL on stimulated HUVECs (Fig. 1b). Because activated integrins support deceleration of cells in flow, we measured the surface expres- sion of activation-dependent epitopes of integrins LFA-1 (CD11a/ CD18; a L b 2 ) and VLA-4 (CD49d/CD29; a 4 b 1 ), and of chemokine- receptor CXCR4 on ML and NL. Flow cytometry revealed no difference in the expression of these molecules (Supplementary Fig. 1), indicating that the marked decrease in rolling velocity of ML was primarily due to their increased capacity to engage Received 16 June; accepted 25 July; published online 17 September; corrected online 1 October 2006; doi:10.1038/nm1470 1 Department of Dermatology, Brigham & Women’s Hospital. 2 Harvard Skin Disease Research Center, 77 Avenue Louis Pasteur, Room 671, Harvard Medical School, Boston, Massachusetts 02115, USA. 3 Wellman Center for Photomedicine, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts 02114, USA. 4 Department of Medicine, Brigham & Women’s Hospital. 5 Department of Medical Oncology, Dana-Farber Cancer Institute. Correspondence should be addressed to R.S. (rsackstein@rics.bwh.harvard.edu). NATURE MEDICINE VOLUME 12 [ NUMBER 10 [ OCTOBER 2006 1185 LETTERS © 2006 Nature Publishing Group http://www.nature.com/naturemedicine