Background
The microenvironment of the brain is tightly regulated by the blood-brain barrier (BBB) the anatomical basis of which is the cerebral endothelium. The BBB endothelium is highly specialized and different morphologically, functionally and immunologically from small and large vessel EC of other organs. Under normal physiological conditions, the presence of interendothelial tight junctions and absence of a vesicular transport system restrict the entry of proteins, ions, lipid insoluble non-electrolytes and circulating haematogenous cells into the brain [
1,
2]. Yet in response to infectious, inflammatory diseases, ischemia, hemorrhage or trauma, there is an influx of leukocytes to sites of brain damage. Interactions between endothelial cells (EC) and circulating leukocytes have been increasingly implicated in the initiation and evolution of inflammatory processes in the central nervous system (CNS). Thus, molecular changes induced on the endothelium by cytokines lead to specific interactions with inflammatory cells that mediate their entry into the brain and accumulation at sites of antigenic challenge [
3].
Chemokines are a family of chemoattractant cytokines characterized by their unique ability to both recruit and activate a variety of cell types. Currently, there are about fifty known chemokine members [
4] which are divided into four sub-families by virtue of highly conserved N-terminal cysteine motifs (disulfide bonds) and the presence or absence of intervening amino acids. There are two major sub-families, which consist of the α- (or CXC) and β- (or CC) chemokines; and two minor sub-families, the γ- (or C) and δ-chemokines (or CX
3C). Chemokines have been associated with a broad range of biological and pathological processes [
5‐
7], which include angiogenesis [
8], CNS development [
9,
10], atherosclerosis [
11], cancer biology [
12‐
14], autoimmune diseases [
15], nervous system inflammation [
16], asthma [
17], and haematopoiesis [
18].
There is considerable evidence that the β-chemokines CCL2 (MCP-1) and CCL3 (MIP-1α) play an important role in CNS inflammation. Several studies have shown the presence of CCL2 and/or CCL3 in multiple sclerosis (MS) lesions. In acute MS lesions, CCL2, CCL3 and CCL4 are selectively expressed by astrocytes, macrophages, and microglia in the lesion centre and in the surrounding white matter, whereas in actively demyelinating plaques CCL5 localizes on EC, perivascular cells and reactive astrocytes [
19]. In acute and chronic active MS lesions, CCL2, along with CCL7 and CCL8, are expressed primarily by hypertrophic astrocytes and variably by inflammatory cells [
20,
21]. Significantly lower CSF levels of CCL2 have been reported in MS patients with active disease compared to controls and patients with stable MS [
22,
23]. In this regard, it has been shown that CCL2 is "consumed" by T lymphocytes and monocytes as they migrate across brain EC monolayers in vitro, leading to downregulation of the CCR2 receptor in response to CCL2, which may account for the decreased CSF CCL2 levels [
24]. The expression of CCL3 has been associated with microglia and macrophages in the white matter lesions of MS patients [
25].
In experimental allergic encephalomyelitis (EAE), the animal model of MS, the onset of the disease coincides with the mRNA expression of CCL2, CCL3 and other chemokines [
26,
27] and the accumulation of CXCL10 and CCL2 [
28]. In the Lewis rat model, CCL2 mRNA increased before the onset of clinical signs, peaked with height of clinical disease, and declined with resolution [
29]. CCL2 expression was localized on lymphocytes, macrophages, astrocytes, and EC and correlated with disease activity [
30]. In chronic relapsing EAE, increased expression of CCL2, CXCL10 and KC was observed in astrocytes, whereas infiltrating leukocytes were the source of CCL3 and CCL5 [
31]. In the SJL/J mouse, the expression of CCL3 correlates with the clinical onset of EAE and administration of anti-CCL3 Abs prevents the development of both acute and relapsing disease and infiltration of mononuclear cells into the CNS initiated by the transfer of neuroantigen peptide-activated T cells [
32]. Furthermore, these chemokines could drive Th1/Th2 lymphocyte differentiation [
33]. Vaccination of Lewis rats with naked DNA encoding CCL2 and CCL3 prevented EAE [
34]. These data suggest that CCL2 and CCL3 have critical and non-redundant roles in the pathogenesis of autoimmune CNS inflammation.
In this study, we investigated the kinetics of expression and release of CCL2 and CCL3 by resting and cytokine or lipopolysaccharide (LPS) treated human brain microvessel EC using a well-characterized in vitro model of the BBB. We show that the constitutive expression and release of CCL2 by HBMEC is significantly upregulated following EC activation in a time-dependent manner. In contrast, CCL3 expression under resting conditions is negligible and its release requires stimulation by cytokines or LPS. Since leukocyte integrin activation by chemokines is required for firm adhesion and transendothelial migration, these findings further support the active participation of the BBB endothelium in neuroinflammation.
Discussion
The entry of inflammatory cells into the brain is a critical event in the pathogenesis of inflammatory and infectious diseases, as well as non-inflammatory conditions of the CNS, such as stroke and trauma. As an anatomical and immunological barrier, the BBB plays a central role in the recruitment of leukocytes in acute and chronic CNS inflammation. It is now well established that the transendothelial movement of leukocytes is a multi-step process, each step being mediated by specific interactions between EC adhesion molecules and their ligands on leukocytes [
36]. Some of the molecular mechanisms involved in the trafficking of leukocytes across the BBB have been recently elucidated and involve membrane interactions between adhesion molecules induced on cerebral EC by inflammatory mediators and integrins or glycosylated ligands on leukocytes. Recent studies from this laboratory have shown that the adhesion and transendothelial migration of resting and activated T lymphocytes across the BBB depend upon the activation status of the endothelium and the T cells and are mediated by receptor-ligand interactions that are specific for each step and for each class of leukocytes [
37,
38].
In vivo and in vitro studies have established a critical role for chemokines in the leukocyte adhesion cascade. Binding of chemokines to their G-protein coupled receptors on leukocytes triggers inside-out signaling leading to rapid integrin activation and firm adhesion to the endothelium [
39]. Although the expression of chemokines by glial cells in the CNS has been relatively well documented [
21,
40], human cerebrovascular chemokine expression has not been fully characterized. In the present study we show that CCL2 is constitutively expressed by HBMEC with marked increase of the intracellular protein and a 3-fold increase in the release of CCL2 into the media following treatment with TNF-α, IL-1β and LPS. Although IFN-γ alone had no effect on CCL2 release by HBMEC, combination of the highest concentrations of TNF-α and IFN-γ resulted in a 3-fold increase of the CCL2 levels in the supernatants.
Constitutive expression of CCL2 mRNA has been previously reported in porcine brain EC with upregulation upon stimulation with TNF-α [
41]. In vitro studies using rat brain and retinal EC lines showed constitutive expression of CCL2 and increased release into the media following activation with TNF-α, IL-1β and IFN-γ [
42]. Both the constitutive release at 2,000 - 2,700 pg/ml and the stimulated release at 5,500 pg/ml were much lower compared to HBMEC (up to 20,000 pg/ml and 60,000 pg/ml, respectively), which may reflect species differences in CCL2 expression and release. In extracerebral endothelium models, CCL2 RNA transcripts have been demonstrated in human aortic, pulmonary artery and umbilical vein endothelial cell (HUVEC) cultures, as well as in freshly removed human arteries and veins [
43]. Expression of CCL2 mRNA has been detected in resting and cytokine activated HUVEC and human brain EC after IFN-γ stimulation [
44]. Exposure of human cerebrovascular EC to hypoxic astrocyte-conditioned media for 4 - 8 hrs increased the release of CCL2 from a low constitutive level of 100 pg/ml to a modest 600 - 700 pg/ml, both significantly lower than the levels obtained in the present study [
45]. This may be related to the short time of exposure to hypoxic media. Similarly, expression of CCL2 was induced in human brain-derived EC by endothelin-1 and ischemia [
46]. A modest increase in CCL2 release, up to 1,800 pg/ml, has been reported in a human brain EC line after incubation with heat-killed Streptococcus suis [
47]. The same strain had no effect on CCL2 expression by HUVEC. In addition to cytokines, injection of the HIV Tat1-72 protein into the mouse hippocampus was shown to increase the expression of CCL2 on brain vascular endothelium [
48]. Substantial evidence indicates the importance of CCL2 in the induction and propagation of the inflammatory cascade. Thus, CCL2 was shown to stimulate T cell migration across microvascular endothelium [
49] and to mediate the firm adhesion of monocytes under flow conditions [
50]. In the CNS, knockout models for CCL2 and CCR2 provide strong evidence for the importance of the CCL2-CCR2 interaction [
51,
52]. Recent observations indicate that glia-derived CCL2 regulates the development of EAE by attracting TNF-α and iNOS-producing dendritic cells and macrophages to the CNS [
53]. Furthermore, a recent study investigating potential mechanisms for HIV entry into the CNS indicates that CCL2 enhances the transmigration of HIV-infected leukocytes across the BBB via the upregulated expression of CCR2 [
54].
In contrast to CCL2, the constitutive RNA and protein expression of CCL3 by HBMEC is negligible. Furthermore, resting HBMEC do not constitutively release CCL3 into the media. Treatment with TNF-α and IFN-γ increased RNA expression and incubation with individual cytokines or LPS upregulated protein expression and induced CCL3 release. However, under identical experimental conditions, the stimulated CCL2 release was typically two orders of magnitude greater than CCL3 release. IFN-γ alone has no effect on cerebral endothelial CCL3 expression, however, when combined with TNF-α, a synergistic effect resulted in higher protein levels compared to TNF-α alone. Expression of CCL3 by EC has been addressed in a limited number of studies. In animals, CCL3 has been reported in murine bone marrow EC [
55] and in the endothelium of epineurial and endoneurial vessels following transection of the rat sciatic nerve [
56]. Furthermore, expression of this chemokine was induced in a murine endothelial cell line by alloantigen-primed T cells [
57]. In humans, endothelial expression of CCL3 has been documented in HUVEC incubated with activated platelets [
58] or exposed to diamide [
59] and LPS [
60]. In addition, CCL3 has been localized to EC of blood vessels and splenic sinusoids in the hemophagocytic syndrome [
61]. According to a recent report, transmigration of bone marrow-derived dendritic cells across mouse brain EC monolayers was increased in the presence of CCL3 concentration gradients [
62]. The expression of CCL3 by cerebrovascular endothelium under resting and inflammatory conditions has not been previously addressed.
It is now well established that immobilization of chemokines by binding to glycosaminoglycans on the luminal EC surface enhances leukocyte adhesion, while binding to the abluminal surface and subendothelial matrix promotes their directional migration to sites of inflammation [
63]. The present study shows that, following their release into the culture media, CCL2 and CCL3 bind to the surface of HBMEC and to the discontinuous basal lamina-like material under the basal cell surface in a polarized manner, which is distinct for each chemokine. Thus, under resting conditions, CCL2 binds preferentially to the apical (luminal) surface of HBMEC, whereas CCL3 shows minimal binding only to the basal (abluminal) EC surface, which is consistent with our protein synthesis and release results. In cytokine activated HBMEC, binding of CCL2 is redistributed towards the basal cell surface and that of CCL3 preferentially on the apical surface. These findings suggest that CCL3 may be primarily responsible for the initial recruitment and activation of CCR1 and/or CCR5 expressing cells, whereas CCL2 plays a greater role in establishing the chemotactic gradients necessary for the directional cell migration into the brain parenchyma. In accordance with these observations, previous studies have demonstrated the presence of specific and separate binding sites for CCL2 and CCL3 along the abluminal surface of human brain microvessels [
64]. The differential binding of CCL2 and CCL3 to cerebral EC in an inflammatory milieu lends support to previous studies which, taking into account the CCL2 and CCL3 expression patterns and Ab therapy studies, suggest that CCL3 controls mononuclear cell accumulation during acute EAE, whereas CCL2 controls cellular infiltration during relapsing disease indicating that acute and relapsing EAE are regulated by the differential expression of CCL2 and CCL3 [
65].
Previous in vitro studies from this laboratory have documented the expression and cytokine upregulation of the β-chemokines CCL4 and CCL5 by HBMEC [
66] and their role in enhancing adhesion of memory and activated CD4+ T lymphocytes to cytokine treated HBMEC [
67]. The present study points towards substantial differences in cytokine regulation of protein release among the four β-chemokines. The release of CCL2 and CCL3 is upregulated by TNF-α, IL-1β, LPS and TNF-α + IFN-γ, but not IFN-γ alone, in contrast to CCL5 which responds to TNF-α, IFN-γ and LPS, but not IL1-β. The release of CCL4 is augmented by LPS and combinations of TNF-α with IFN-γ or IL-1β, but not by single cytokine treatments. Overall, the release of CCL2 under resting and stimulated conditions is much greater than that of CCL3, CCL4 and CCL5. Additional differences exist in the distribution of bound chemokines to HBMEC. In resting monolayers, CCL2 and CCL5 are bound preferentially to the apical EC surface, CCL4 to both apical and basal surfaces and CCL3 only to the basal surface. Upon cytokine stimulation, this polarized expression is reversed for all four chemokines, with CCL2, CCL4 and CCL5 now present preferentially along the basal surface and subendothelial matrix, and CCL3 distributed mostly along the apical surface. These differences strongly suggest differential and possibly temporal roles of these chemokines in the regulation of leukocyte transendothelial migration in CNS inflammation.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RC carried out the experiments and statistical analysis and contributed to the preparation of the manuscript. KD-Z conceived and designed the study and prepared the manuscript. Both authors have read and approved the final version of the manuscript.