Soluble VCAM-1 impairs human brain endothelial barrier integrity via integrin α-4-transduced outside-in signalling
- 2.4k Downloads
Human brain microvascular endothelial cells forming the blood–brain barrier (BBB) release soluble vascular cell adhesion molecule-1 (sVCAM-1) under inflammatory conditions. Furthermore, sVCAM-1 serum levels in untreated patients with multiple sclerosis (MS) correlate with a breakdown of the BBB as measured by gadolinium-enhanced MRI. To date, it is unknown whether sVCAM-1 itself modulates BBB permeability. Here, we provide evidence that human brain endothelium expresses integrin α-4/β-1, the molecular binding partner of sVCAM-1, and that sVCAM-1 directly impairs BBB function by inducing intracellular signalling events through integrin α-4. Primary human brain microvascular endothelial cells showed low to moderate integrin α-4 and strong β-1 but no definite β-7 expression in vitro and in situ. Increased brain endothelial integrin α-4 expression was observed in active MS lesions in situ and after angiogenic stimulation in vitro. Exposure of cultured primary brain endothelial cells to recombinant sVCAM-1 significantly increased their permeability to the soluble tracer dextran, which was paralleled by formation of actin stress fibres and reduced staining of tight junction-associated molecules. Soluble VCAM-1 was also found to activate Rho GTPase and p38 MAP kinase. Chemical inhibition of these signalling pathways partially prevented sVCAM-1-induced changes of tight junction arrangement. Importantly, natalizumab, a neutralising recombinant monoclonal antibody against integrin α-4 approved for the treatment of patients with relapsing–remitting MS, partially antagonised the barrier-disturbing effect of sVCAM-1. In summary, we newly characterised sVCAM-1 as a compromising factor of brain endothelial barrier function that may be partially blocked by the MS therapeutic natalizumab.
KeywordsMultiple sclerosis Blood–brain barrier Endothelial cell Integrin alpha4 Vascular cell adhesion molecule-1 Natalizumab
Human brain microvascular endothelial cells
Human umbilical vein endothelial cells
Peripheral blood mononuclear cells
Soluble vascular cell adhesion molecule-1
Very late antigen-4
Cell-bound vascular cell adhesion molecule-1 (VCAM-1, CD106) allows human brain microvascular endothelium to control immune cell trafficking across the blood–brain barrier (BBB). It is upregulated in inflammatory-active brain lesions of patients with multiple sclerosis (MS), a chronic degenerative autoimmune disease of the CNS [1, 16, 45]. Endothelial VCAM-1 serves as a binding partner for integrin α-4/β-1 (very late antigen-4, VLA-4) on peripheral blood mononuclear cells (PBMC), as it does for α-4/β-7 heterodimers to a lesser extent [5, 17]. This molecular interaction enables PBMC to firmly adhere to the vessel wall after rapid activation of integrin α-4-mediated intracellular signalling cascades, allowing subsequent immune cell extravasation .
A soluble form of VCAM-1 (sVCAM-1) is shedded from the surface of brain endothelial cells upon inflammatory activation . In vitro, sVCAM-1 blocked leukocyte adhesion to activated human brain endothelium. Soluble VCAM-1 was therefore considered as an inflammation-limiting factor at the inflamed BBB . Clinical studies in MS patients treated with recombinant interferon-β (IFN-β), which moderately reduces relapse rate, disability progression and MRI disease activity , seemed to support this hypothesis: IFN-β therapy increased sVCAM-1 serum levels, which correlated with a reduction of gadolinium (Gd)-enhancing MRI brain lesions, indicating less inflammatory disease activity at the BBB [12, 20, 38]. Together, these in vitro and clinical studies suggested a local anti-inflammatory effect of sVCAM-1 at the human BBB due to an inhibition of immune cell extravasation .
In addition to their regulation of immune cell trafficking, brain endothelial cells strictly govern the exchange of solute and soluble factors across the BBB. Endothelial molecular control mechanisms include active transendothelial transport systems and tight junctions. The latter are highly dynamic trans-membrane protein complexes, tightly sealing the interendothelial clefts . Extravasation of soluble factors such as albumin or immunoglobulins is a direct correlate of BBB dysfunction as visualised by Gd-enhanced MR imaging [23, 30]. Interestingly, sVCAM-1 serum levels in MS patients not receiving IFN-β treatment were shown to positively correlate with the presence of Gd-enhancing lesions on brain MRI scans and with clinical disease activity in a majority of studies [15, 18, 19, 22, 36, 37]. This seems to contradict the IFN-β studies cited above where an inverse correlation between sVCAM-1 serum levels and MRI disease activity was observed. The mechanisms underlying these divergent findings are unknown. Furthermore, it currently remains unknown whether sVCAM-1 exerts any direct effect on brain endothelial barrier function.
So far, it is unknown whether undiseased human brain endothelial cells or those in MS CNS lesions express the established binding partners of sVCAM-1, i.e. integrin α-4 heterodimers. An expression of integrin α-4 by human brain endothelium in situ was previously described in activated glioma endothelial cells . Furthermore, expression of integrin α-4/β-1 and of α-4/β-7 was reported in human umbilical vein endothelial cells (HUVEC) and in adult human synovial membrane endothelium from patients with rheumatoid arthritis [6, 29]. VLA-4 expression was furthermore documented in adult human dermal microvascular endothelial cells . Endothelial integrin α-4 expression was lower than integrin β-1 expression in these endothelial cell types, but could be upregulated by pro-inflammatory stimulation with TNF-α [6, 32]. HUVEC binding to recombinant VCAM-1 and to the extracellular matrix protein fibronectin was found to be mediated by VLA-4 [26, 29]. Furthermore, sVCAM-1 was shown to activate p38 MAP kinase and focal adhesion kinase in HUVEC via integrin α-4, which promoted HUVEC migration in vitro and angiogenesis in mouse corneas in vivo . Effects of sVCAM-1 on the barrier function of integrin α-4-positive endothelial cells have not been reported to date.
Here, we investigated the expression of sVCAM-1-binding integrins on adult human brain endothelium both under non-inflammatory and inflammatory conditions, and studied whether and how sVCAM-1 affects the barrier function of cultured primary human brain microvascular endothelium.
Materials and methods
Cryopreserved early post-mortem autopsy specimens from normal human brain and spinal cord of three donors who had died from non-neurological diseases and cryopreserved brain biopsy samples from two donors without pathological changes in the biopsy material used for this study were from the brain bank at the Department of Neuropathology in Würzburg and used for research purposes as approved by the local Ethics Committee of the Faculty of Medicine at the University of Würzburg.
Furthermore, brain tissue from seven cases with clinically diagnosed and pathologically confirmed MS (3 female, median disease duration 22 years, median age at death 60 years, range 43–90) and four non-neurological control cases (2 female, median age at death 82 years, range 73–91) was obtained by rapid autopsy (median post-mortem delay 6 h, range 4–18 h) and immediately frozen in liquid nitrogen (The Netherlands Brain Bank, Amsterdam, coordinator Dr. Huitinga). The Netherlands Brain Bank received permission to perform autopsies for the use of tissue and for access to medical records for research purposes from the Ethical Committee of the VU University Medical Centre, Amsterdam, The Netherlands. Supratentorial white matter MS and control tissue samples were selected on the basis of post-mortem MRI. All patients and controls, or their next of kin, had given informed consent for autopsy and use of their brain tissue for research purposes. MS lesion activity was classified based on the expression of myelin and inflammatory cells, defined by staining for proteolipid protein and HLA-DR as described before [13, 27].
Brain endothelial cells
Cryopreserved single-donor primary human brain microvascular endothelial cells (HBMEC) isolated from normal human brain and reported to be mycoplasma negative by the provider were purchased from Cell Systems Corp. (Kirkland, WA, USA) at passage 2. Preparations were analysed from nine different donors. Before experimental use, each preparation was extensively characterised for potential contamination by other cell types and for expression of tight junction-associated molecules as previously described . The endothelial cell fraction was >98 % in all samples. A well-characterised immortalised HBMEC line (HBMECKim) was kindly provided by Kim . Both cell types were plated on culture dishes (Nunc, Roskilde, Denmark) coated as indicated and cultured at 37 °C/5 % CO2 in Medium 199 (Lonza, Cologne, Germany) containing 10 % foetal calf serum (FCS; Biochrom, Berlin, Germany), endothelial cell growth supplement (20 μg/mL; Sigma-Aldrich, Schnelldorf, Germany), heparin (100 μg/mL; Sigma-Aldrich), amphotericin B (250 μg/mL), gentamycin (50 μg/mL), penicillin (50 U/mL) and streptomycin (50 μg/mL; all from Invitrogen, Karlsruhe, Germany). All primary cell preparations as well as HBMEC were mycoplasma negative, as revealed by a commercial PCR-based mycoplasma detection kit (PK-CA91; PromoKine, Heidelberg, Germany). Primary endothelial cells were used from passage 4 to 6 for experiments, immortalised endothelial cells from passage 20 to 30.
For immunofluorescence histochemistry, 5-µm cryostat sections were blocked with a donkey serum solution and incubated with primary antibodies (Abs) at 4 °C overnight. Tested monoclonal mouse Abs against integrin α-4 included clones 2B4, 7.2R (both from R&D, Karlsruhe, Germany) and L25 (BD Biosciences, Heidelberg, Germany). For staining of integrin β-1, we used a rat monoclonal Ab (clone mAb13, BD Biosciences), and for integrin β-7, we employed the mouse monoclonal Ab FIB504 (BD Biosciences). Rabbit polyclonal Abs against von Willebrand factor (vWF; DAKO, Hamburg, Germany) or laminin (MP Biomedicals, Eindhoven, Netherlands) were used to stain the brain vasculature. A biotinylated antibody against HLA-DR (clone LN3) was used to stain immune cells. Isotype-matched control Abs were from BD Biosciences. Fluorescence-conjugated secondary Abs were from Dianova (Hamburg, Germany) and incubated for 1 h.
For bright field stainings, sections were deparaffinised and treated with 0.3 % H2O2 in methanol for 20 min to reduce endogenous peroxidase activity. Antigen retrieval was achieved by incubating the sections at 100 °C in Tris–EDTA buffer (10–0.5 mM, pH 9.0) for 10 min. After washing with phosphate-buffered saline (PBS), sections were treated with 0.1 % saponin in PBS, washed, and subsequently incubated with the same primary Abs as used for immunofluorescence histochemistry in PBS overnight at 4 °C. Slides were then washed and incubated with EnVision + Dual Link reagent (DAKO, Glostrup, Denmark) for 30 min followed by visualisation with the peroxidase substrate diaminobenzidine (DAB) (DAKO, Glostrup, Denmark). After a short rinse in tap water, sections were incubated with haematoxylin for 1 min and extensively washed with tap water for 10 min. Finally, sections were dehydrated with ethanol followed by xylene and mounted with Entellan (Merck, Darmstadt, Germany).
Samples were analysed with an Olympus IX-70 inverted system microscope with IX-FLA observation attachment for fluorescence imaging, with a Zeiss LSM 780 confocal microscope or with a Leica DM6000 microscope (Leica Microsystems, Heidelberg, Germany).
HBMEC were grown to subconfluency on 6-well plates and then stimulated as indicated. Vascular endothelial growth factor-165 (VEGF165) was purchased from PeproTech Inc. (Hamburg, Germany) and TNF-α was from R&D Systems. Thereafter, cells were detached with 500 μL Accutase™ (PAA Laboratories, Coelbe, Germany), washed and incubated with uncoupled mouse anti-integrin α-4 (clone L25), PE-coupled mouse anti-integrin β-1 (clone MAR4, BD Biosciences), PE-coupled mouse anti-integrin β-7 (clone FIB504, BD Biosciences) or isotype-matched control Abs (all from BD Biosciences) for 30 min at 4 °C. For integrin α-4 staining, cells were additionally incubated with a FITC-labelled secondary Ab against mouse IgG (R&D, Karlsruhe, Germany) after a washing step. Fluorescence-activated cell sorting analysis was performed on a FACSCalibur (BD Biosciences). For all integrin stainings including integrin β-7, human PBMC, freshly isolated by Ficoll gradient centrifugation according to a standard protocol, were used as a positive control.
Boyden chamber assay
For paracellular permeability assays, 1 × 105 per well cells were seeded on rat collagen-coated (100 µg/mL) filters (0.4 µm pore size) of a 24-well Boyden chamber system (Corning Life Sciences, Wiesbaden, Germany) and grown to confluency which usually took 3 days. Confluency was assessed by DAPI staining of filters grown in parallel. Part of the cells was pre-incubated with natalizumab (Biogen Idec, Ismaning, Germany) or a corresponding IgG4κ isotype control (Sigma-Aldrich) for 1 h as indicated. Stimulations were performed with recombinant human VCAM-1 or ICAM-1 (R&D Systems) or TNF-α plus interferon-γ (IFN-γ, R&D Systems) for the indicated durations. After stimulation, medium was removed in the upper and lower chambers and replaced by HEPES buffer. To trace cell permeability, 1 mg/mL FITC-dextran 3000 was added to the upper chambers and relative fluorescence in the lower chambers was measured 90 min later, using a Fluoroskan Ascent® (Thermo Electron Corporation, Dreieich, Germany) microplate fluorometer. Dextran concentrations in the lower chambers were determined using a dextran standard curve. To maximise assay precision, experiments were performed in 12 wells per stimulation condition.
For immunocytochemical stainings of tight junction-associated molecules, cells were grown on 2 % gelatin-coated 24-well plates and stimulated as indicated. Rho-associated kinases inhibitor Y-27632 and p38 inhibitor SB203580 were from Calbiochem—Merck4Biosciences (Darmstadt, Germany). Then cells were fixed with 3.7 % formalin for 10 min, permeabilised with 0.1 % Triton X-100 for 6 min, and washed with Dulbecco’s PBS. Unspecific Ab binding was blocked by 5 % BSA (Sigma-Aldrich) in PBS for 60 min at room temperature. Subsequently, primary Abs to zonula occludens (ZO)-1 (rabbit polyclonal, Invitrogen), occludin (mouse monoclonal, clone OC-3F10, Invitrogen) and junctional adhesion molecule (JAM)-A (mouse monoclonal, clone M.Ab.F11, AbD Serotec, Düsseldorf, Germany) were incubated for 60 min at room temperature. After washing, an appropriate Cy3-coupled secondary Ab was incubated for another hour at room temperature. After a further washing step, nuclei were counterstained with DAPI. After a final washing step, an antifading agent (Dabco, Merck, Germany) was added. Negative controls were performed by omitting the primary Abs and by stainings with isotype-matched control Abs (data not shown). The stainings were analysed by an Olympus IX-70 inverted system microscope with IX-FLA observation attachment for fluorescence imaging. All analyses were performed by blinded observers directly at the microscope and not from electronic images.
For generation of whole cell protein extracts, cells grown to subconfluency in 25 cm2 flasks were stimulated as indicated, washed with icecold PBS and scraped into radioimmunoprecipitation assay (RIPA) buffer composed of 50 mM HEPES, 125 mM NaCl, 1 % Nonidet P40, 1 mM EDTA pH 7.4 and 1 × Roche COMPLETE® protease inhibitor mix (Roche Diagnostics GmbH, Mannheim, Germany). Samples were shaken vigorously for 30 min, and then centrifuged at 18,000g for 15 min at 4 °C. Supernatants were subjected to Western blot analysis. Primary Abs were used against phospho-p38 (cat.-no. 9211, Cell Signaling, Danvers, MA, USA), phospho-ERK (cat.-no. sc-7383, Santa Cruz, Heidelberg, Germany) or phospho-JNK (cat.-no. sc-6254, Santa Cruz). Appropriate peroxidase-coupled secondary Abs were employed with a standard enhanced chemoluminescence system (Amersham, Arlington Heights, IL, USA). After peroxidase inactivation, membranes were reprobed with Abs against total p38 (cat.-no. 9212, Cell Signaling), ERK (cat.-no. sc-94, Santa Cruz) or JNK (cat.-no. sc-474, Santa Cruz).
Rho activation assay
Rho activation assays were performed using the Rho Activation Assay Kit from Millipore (Schwalbach/Ts., Germany) according to the instructions of the manufacturer. In brief, active, GTP-bound Rho was isolated from cell extracts using a GST-tagged fusion protein corresponding to residues 7–89 of mouse Rhotekin rho-binding domain and bound to glutathione–agarose, and subsequently detected by immunoblot analysis using anti-Rho.
For statistical analysis of the dextran permeability assays, a Kruskal–Wallis test was followed by Dunn’s post test for multiple comparisons. Calculations were performed with GraphPad PRISM 4 software (GraphPad Software, La Jolla, CA).
Low to moderate normal brain endothelial integrin α-4 expression in situ and in vitro
To further study the subcellular localisation of integrin α-4 in human brain endothelium in situ, we next investigated cryopreserved brain biopsy specimens from two donors without pathological changes in their biopsies, as revealed by extensive neuropathological evaluation. The expression of integrin α-4 was found to be mainly restricted to the luminal membranes and weaker detectable in the abluminal membranes (Fig. 1d).
Increased endothelial integrin α-4 expression after angiogenic stimulation in vitro and in active MS brain lesions
Soluble VCAM-1 increases brain endothelial permeability via integrin α-4 and alters actin and tight junction morphology
Rho GTPase and p38 mediate sVCAM-1-induced alteration of tight junction morphology
Here we report that non-inflamed adult human brain microvascular endothelium expresses integrin α-4 in situ and in vitro. Integrin α-4 was found to be upregulated after angiogenic stimulation in vitro and in active demyelinating MS brain lesions. Integrin α-4/β-1 is an established binding partner of sVCAM-1 which gets released from brain endothelial cells under inflammatory conditions . In our hands, recombinant sVCAM-1 compromised brain endothelial barrier function in vitro, which was mediated by the induction of intracellular signalling events including the activation of Rho GTPase and p38 MAP kinase through binding of sVCAM-1 to integrin α-4. While the MS therapeutic natalizumab is well known to exert a protective effect at the BBB by blocking leukocyte adhesion, our findings suggest a novel additional protective mode of natalizumab action at the BBB by partially inhibiting integrin α-4 on brain endothelial cells.
Our in vitro findings indicate that a positive correlation between sVCAM-1 serum levels and Gd-enhancing MRI brain lesions in untreated MS patients, as observed by many authors [15, 18, 19, 22, 36, 37], at least partially reflect a direct causal relationship between both variables: sVCAM-1, which is released from inflammatory-activated human brain endothelium , could enhance BBB dysfunction in acute MS brain lesions by endothelial autocrine stimulation. Accordingly, increased sVCAM-1 serum levels in untreated MS patients with active disease would particularly be explained by a sVCAM-1 release from endothelium in inflammatory-active brain lesions. In contrast, an elevation of sVCAM-1 serum levels by IFN-β treatment could be due to a sVCAM-1 release particularly from vascular beds exposed to very high IFN-β concentrations, i.e. near the injection sites, but possibly not mainly the brain. However, brain endothelial cells were shown to release sVCAM-1 upon IFN-β exposure in vitro, establishing a direct causal link between exposure of endothelial cells to IFN-β and a release of sVCAM-1 [11, 25]. At IFN-β skin injection sites, strong inflammatory immune reactions, possibly boosting an IFN-β-induced release of sVCAM-1 from local endothelial cells, occur frequently. Such inflammatory reactions are common even if no externally visible skin reactions are present, as demonstrated by a placebo-controlled skin biopsy study . Further following this explanatory model of a spatially differential sVCAM-1 release in untreated active versus IFN-β-treated MS patients, an inverse correlation between sVCAM-1 serum levels and Gd-enhancing MRI brain lesions in IFN-β-treated MS patients could then be explained by a predominant therapeutic effect of sVCAM-1-triggered VLA-4 downregulation on PBMC via ligand–receptor interaction as previously demonstrated . This may render them less responsive to endothelium-bound VCAM-1 expressed in inflammatory-active MS brain lesions and therefore reduce inflammatory MRI disease activity. Importantly, VLA-4 downregulation on PBMC inversely correlated with both sVCAM-1 serum levels and with clinical treatment response . Together, these findings argue for a predominant sVCAM-1 effect on immune as opposed to brain endothelial cells in IFN-β-treated patients. In summary, we suggest that sVCAM-1 may exert either a detrimental or a beneficial net effect on inflammatory MS disease activity, depending on the body region of its primary release. When released at an inflamed BBB where its endothelial-binding partner integrin α-4 is upregulated and integrin β-1 is present according to our study, it may further increase paracellular BBB dysfunction by endothelial autocrine stimulation, reflected by more Gd-enhancing MRI lesions. When released near IFN-β injection sites, sVCAM-1 may primarily render PBMC less responsive to endothelium-bound VCAM-1 in active MS brain lesions by VLA-4 downregulation on PBMC.
Integrin α-4, which was demonstrated to be expressed by endothelium in MS brain lesions in this study, is the molecular target of natalizumab. This humanised monoclonal IgG4κ antibody was approved for the treatment of severe relapsing–remitting MS in 2006. Therapy of MS patients with natalizumab reduced the number of Gd-enhancing MRI lesions by 92 % over 2 years in the AFFIRM trial . This strong therapeutic effect of natalizumab most likely reflects its main mechanism of action in patients with MS, i.e. blockade of the molecular interaction between VLA-4 on T cells and VCAM-1 on the surface of brain endothelial cells, thereby strongly reducing inflammatory brain infiltrates . Our findings, however, suggest that in addition natalizumab may beneficially modulate a potential detrimental interplay between integrin α-4/β-1 on brain endothelium and sVCAM-1 as suggested by this study.
In summary, we demonstrated that sVCAM-1 directly compromises the barrier function of human brain endothelium by integrin α-4/β-1-mediated induction of intracellular signalling events. Based on these findings, we suggested a model of how to explain apparently contradictory findings on the role of sVCAM-1 in untreated versus IFN-β-treated MS patients. Furthermore, our results argue for a novel mode of action of natalizumab at the BBB, where it may partially protect brain endothelial cells from a sVCAM-1-mediated barrier breakdown.
We thank Nadine Kehl and Svetlana Hilz for excellent technical assistance and Kwang S. Kim for kindly providing the human brain endothelial cell line. This work was supported by the Interdisciplinary Center for Clinical Research (IZKF) at the University of Würzburg (A-57 to M. B. and F. B.-S.) and by University research funds from the State of Bavaria.
Conflict of interest
The authors declare no competing interests.
- 23.Hawkins CP, Munro PM, MacKenzie F, Kesselring J, Tofts PS, du Boulay EP, Landon DN, McDonald WI (1990) Duration and selectivity of blood–brain barrier breakdown in chronic relapsing experimental allergic encephalomyelitis studied by gadolinium-DTPA and protein markers. Brain 113(Pt 2):365–378CrossRefPubMedGoogle Scholar
- 27.Kooij G, Mizee MR, van Horssen J, Reijerkerk A, Witte ME, Drexhage JA, van der Pol SM, van Het Hof B, Scheffer G, Scheper R et al (2011) Adenosine triphosphate-binding cassette transporters mediate chemokine (C-C motif) ligand 2 secretion from reactive astrocytes: relevance to multiple sclerosis pathogenesis. Brain 134:555–570CrossRefPubMedGoogle Scholar
- 30.Morrissey SP, Stodal H, Zettl U, Simonis C, Jung S, Kiefer R, Lassmann H, Hartung HP, Haase A, Toyka KV (1996) In vivo MRI and its histological correlates in acute adoptive transfer experimental allergic encephalomyelitis. Quantification of inflammation and oedema. Brain 119(Pt 1):239–248CrossRefPubMedGoogle Scholar
- 38.Rieckmann P, Kruse N, Nagelkerken L, Beckmann K, Miller D, Polman C, Dahlke F, Toyka KV, Hartung HP, Sturzebecher S (2005) Soluble vascular cell adhesion molecule (VCAM) is associated with treatment effects of interferon beta-1b in patients with secondary progressive multiple sclerosis. J Neurol 252:526–533CrossRefPubMedGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.