The initial objective of this study was to develop an in vitro
inflammatory BBB model to visualize sequential leukocyte-brain endothelial interactions under physiological flow conditions. The leukocyte extravasation process, including apical chemokine deposits, integrin engagement, firm adhesion, locomotion, and transendothelial migration, are all dependent on hemodynamic shear, a fundamental physiological feature of all leukocyte-endothelial interactions. The present dynamic in vitro
BBB model using parallel plate flow chamber and THBMEC allows us to visualize leukocyte-brain endothelial interactions under a flow rate simulating the velocity in brain capillaries (Hudetz et al., 1996
). THBMEC maintained their morphology and the distribution of immunoreactivity for tight junction-associated proteins under flow conditions. These findings indicate that the leukocyte-endothelial interaction was not a passive process caused by mechanical damage to THBMEC.
To simulate inflammatory BBB, we stimulated THBMEC with TNF-α and IFN-γ at concentrations seen in patients with sepsis or systemic inflammatory response syndrome (Brunner et al., 2004
; Collighan et al., 2004
; Kabir et al., 2003
; Watanabe et al., 2005
). Cytokines did not cause observable changes in the distribution of immunoreactivity for tight junction-associated proteins. These observations are consistent with our prior report that aTHBMEC maintained a high transendothelial electrical resistance (TEER) and low solute permeability after cytokine exposure at these concentrations (Ubogu et al., 2006
). As a positive control, we showed that these inflammatory cytokines increased expression of the chemokine GRO peptides CXCL1-3, indicating a biological response. Compared with resting THBMEC, activation of THBMEC with cytokines was required for robust adhesion of leukocytes under flow conditions (). The relationship between the enhancement of adhesion and up-regulation of chemokine and adhesion molecules is under investigation. When G (i/o) signaling was blocked with PTX, PBMC adhesion was almost completely abrogated. This indicates that PBMC adhesion to inflammatory brain endothelial cells is GPCR-dependent.
The α4 integrins are constitutively expressed on lymphocytes, monocytes, and eosinophils (Hemler, 1990
). The interaction between α4β1 integrin on leukocytes and its counter-receptors on the vascular endothelium plays a key role in MS (Rice et al., 2005
; Von Andrian and Engelhardt, 2003
). It is considered likely that α4β1 integrin mediates activated effector memory T cell and monocyte extravasation across the BBB into inflamed MS lesions (Ransohoff, 2007
). NTZ demonstrated efficacy for treatment of MS in two large clinical trials (Polman et al., 2006
; Rudick et al., 2006
). Using the dynamic activated BBB model, we report that NTZ administration to MS patients significantly reduced PBMC adhesion to aTHBMEC under flow conditions. It is worth mentioning that both natalizumab and IFNβ-1a treatment affect leukocyte composition in MS. Natalizumab releases leukocytes from the bone marrow and leads to an approximate doubling of T cells, B cells, pre-B cells, and monocytes in the peripheral blood (Krumbholz et al., 2008
). In MS patients receiving IFN β-1a treatment, the numbers of CD3+, CD8+ T cells, natural killer (NK) T cells per mL blood were slightly lower than in healthy control subjects (Sellebjerg et al., 2005
). The changes of leukocyte compositions are not likely to account for effect of these medications on PBMC adhesion under flow, as PBMC from patients receiving IFN β-1a showed adhesive properties equal to those of healthy controls, while NTZ changes the numbers of PBMC but does not drastically alter their relative proportions. Because standardized numbers of PBMC were used for each experiment, we consider it likely that the direct effects of NTZ on cells, rather than flux in cell populations, led to the results reported here.
The brain endothelial ligand for leukocyte α4 integrin has been unclear (Ransohoff, 2007
). α4 integrin has at least two potential vascular ligands, VCAM- 1 and FN-CS1 (Elices et al., 1990
; Guan and Hynes, 1990
). VCAM-1 is expressed by human aortic endothelial cells (Ganji et al., 2008) and HUVEC, forming transmigratory cups around lymphocytes during diapedesis across HUVEC monolayers (Carman and Springer, 2004
; Miyake et al., 2008
; Engelhardt and Wolburg, 2004
). VCAM-1 was involved in leukocyte homing into lung, as well as melanoma and other tissue (Hallgren et al., 2007
; Deem et al., 2007
). Fibronectin is a high-molecular-weight glycoprotein found in plasma, at the cell surface, and in the extracellular matrix. It is involved in a variety of biological functions, such as cell attachment, spreading, and migration. Fibronectin can be alternatively spliced in at least three regions, generating multiple molecular moieties (Gutman and Kornblihtt, 1987
; Ting et al., 2000
; Wagner et al., 2000
). In autopsy MS brain tissues, FN CS-1 was detected on astrocyte endfeet and astrocytes at the lesion edge (Van et al., 2005
). In EAE mice, blocking α4β1 integrin with antibodies or peptides harboring the CS-1 sequence prevented cellular infiltration in the CNS parenchyma (Kent et al., 1995
; Van der Laan et al., 2002
). FN-CS1 is expressed on THBMEC at rest, with a significant increase in surface expression following cytokine activation (Ubogu et al., 2006
). We confirmed that FN-CS1 mRNA was constitutively expressed by THBMEC. The absence of VCAM-1 expression by THBMEC is consistent with previous reports of lack of VCAM-1 immunoreactivity on brain endothelium in autopsy MS tissue sections (Kivisakk et al., 2003
; Peterson et al., 2002
). Our present and previous results demonstrated that among the two α4 integrin receptors, FN-CS1 is selectively found on THBMEC. When stimulated with higher concentrations of cytokines, THBMEC also exhibited VCAM-1 expression, consistent with previous reports by other groups, using 50–100 fold greater concentrations than those used here (Kallmann et al., 2000
; Wong et al., 1999
). Cytokine concentration may play a key role in inducing different ligands for α4 integrins.
In order to test the function of FN-CS1, FN-CS1 peptide was introduced to occupy the ligand binding site of leukocyte α4 integrin before the adhesion assay. The result showed that PBMC adhesion to aTHBMEC was abrogated. Function-neutralizing antibodies to FN-CS1 also blocked PBMC adhesion to aTHBMEC. It has been reported that Jurkat T cell adhesion to RA synovium was mediated by the interaction between VLA-4 and FN-CS1, but not VCAM-1 (Elices et al., 1994
). Monocyte and T cell migration across in vitro
BBB toward CCL5 was mediated by α4β1 integrin and FN-CS1, not VCAM-1 (Ubogu et al., 2006
). Collectively, these data suggest that FN-CS1, is the ligand for leukocyte α4 integrin on activated brain endothelial cells in vitro.
In conclusion, we describe a dynamic activated BBB model, combining inflammatory human cytokine exposed THBMEC and parallel plate flow chamber. We found that anti-α4 integrin treatment of MS patients inhibited PBMC adhesion to inflammatory THBMEC, as compared with MS controls that received IFN β1a. We found that the ligand for leukocyte α4 integrin on brain endothelial cells is FN-CS1. In addition to α4 integrin, GPCR signaling is essential for leukocyte adhesion on aTHBMEC. This study provides a tool to study leukocyte-brain endothelial interactions, and contributes to our understanding of the effect of anti-α4 integrin antibodyin MS treatment.